<%BANNER%>

Socio-economic impacts of community forest management in rural India

University of Florida Institutional Repository

PAGE 1

1 SOCIO-ECONOMIC IMPACTS OF COMMUNI TY FOREST MANAGEMENT IN RURAL INDIA By FREDERICK JOHN ROSSI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007

PAGE 2

2 Copyright 2007 by Frederick John Rossi

PAGE 3

3 I dedicate this to my parents, Joseph and Ire ne Rossi, for the unconditi onal love and enduring support they have always given me. I also wish to dedicate this work to my uncle and godfather, David Rossi. His words once challenged me to ac tion, which ultimately lead to this dissertation. He will be missed always.

PAGE 4

4 ACKNOWLEDGMENTS I give thanks to God for granting me th e ability, the opportunity, and the means for pursuing and completing this degree, and for safe ly conducting the fieldwork in India. I also thank my wonderful wife, Rosa Cossio, for her patience, support, and encouragement when I needed it most. Her presence in my life has given me great joy, and has kept my spirit aloft even during times of mental and physical stress. I thank my parents for always being there to help me improve myself ; without their support during the years my education would not have been possible. More importantly, however, I thank them for the positive influence that they have had on my life and on the lives of my brothers and sister. I also thank my siblings for their encouragement and for the examples that all three of them have set for me in their own different ways. I would also like to take this opportunity to thank the members of my committee: Ridwan Ali, Robert Emerson, and Stephen Perz. Their guidance, valuable suggestions, and general support were crucial to the improvement and co mpletion of this study. I also wholeheartedly thank the co-chairs of my committee: Janaki Alav alapati and Sherry Lark in. Janaki provided me with much knowledge about Indi a and its people and culture, as well as im portant contacts in Andhra Pradesh. More importantly, he was always there with both guidance and enthusiasm for this work. A special note of appreciati on is reserved for Sherry. She has been tremendous in her support and her willingness to ac commodate my all of my ques tions, concerns, and doubts. I acknowledge her significant contributi on to the final form of this di ssertation; her ab ilities to edit and re-structure my work will never be under-appr eciated, and this dissertation is a testament to that fact. In addition, her good nature and optimis m always buoyed my spirits whenever I felt as

PAGE 5

5 if I hit a dead-end. I am truly grateful for ha ving her support and guidance as both my advisor and friend. A special word of thanks is reserved for Vi shnu Reddy for all of the effort and dedication he put into the fieldwork. I also thank Vishnu a nd his family for their friendship, generosity, and hospitality; words cannot begin to convey how much I appreciated and enjoyed being a guest in their home: I thank them from the bottom of my heart. Namaste. I would like to thank the many other people in India that I befriended and were important to the furtherance of this research in many differen t ways. In particular, I wa nt to thank all of my interviewers; especially Ashok, Kiran, Suresh, a nd Sugunakar: this work would not have been possible without their dedication and effort. Fo r their hospitality, conve rsation, and willingness to help me conduct this research, I also want to thank members of the Andhra Pradesh Forest Department; especially Ramesh Kalaghatgi, P. Malikarrjuna Rao, P.V. Padmanabham, Siddhanand Kukrety, M. Waheed, Vi nod Kumar, A.S. Rao, Raj Rao, and Prabakar. I would also like to thank Gopinath Reddy at the Center for Economic and Social Studies (CESS) in Hyderabad for both his insights and valuable as sistance in helping me to hire my staff. Finally, I would like to thank Doris Capistrano and the Center for International Forestry Research (CIFOR). Their financia l contribution helped to support the fieldwork that is the basis of this work; for this I am deeply appreciative of th eir generosity.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........9 LIST OF FIGURES................................................................................................................ .......10 ABSTRACT....................................................................................................................... ............11 CHAPTER 1 INTRODUCTION: PARTICIPAT ORY FOREST MANAGEMENT...................................13 Challenges of Resource Management....................................................................................13 The Role of Institutions....................................................................................................... ...16 A Case Study in India.......................................................................................................... ...20 Forest Management.........................................................................................................20 Research Objectives........................................................................................................23 Methods........................................................................................................................ ...24 Summary and Preview............................................................................................................26 2 BACKGROUND INFORMATI ON ON THE STUDY AREA.............................................28 Overview....................................................................................................................... ..........28 History of Participatory Fore st Management in India............................................................29 Pre-Colonial Era..............................................................................................................29 Social Forestry Programs................................................................................................31 The Arabari Experiment..................................................................................................33 Development of Joint Forest Management (JFM)...........................................................34 Advent of Community Forest Management (CFM)........................................................38 Comparison to Participatory Forestry in Other Nations.........................................................40 Nepal.......................................................................................................................... ......40 China and Other Asian Nations.......................................................................................40 Africa......................................................................................................................... ......41 Mexico......................................................................................................................... ....42 Description of Andhra Pradesh...............................................................................................44 Location....................................................................................................................... ....44 Languages and Castes......................................................................................................46 Geographic/Political Regions..........................................................................................47 The Forest Department....................................................................................................50 Summary........................................................................................................................ .........52 3 QUESTIONNAIRE DEVELOPMENT AN D SURVEY ADMINISTRATION...................55 Introduction................................................................................................................... ..........55

PAGE 7

7 The VSS Questionnaire..........................................................................................................55 The Household (HH) Questionnaire.......................................................................................56 The LSMS Modules........................................................................................................56 Design and Structural Guidelines....................................................................................61 Formatting Guidelines.....................................................................................................63 Survey Structure and Implementation....................................................................................64 HH Questionnaire Testing and Revision.........................................................................64 Sample Selection.............................................................................................................65 Interviewer Training and Impl ementation of the HH Survey.........................................68 Implementation of the VSS Survey.................................................................................69 Data Compilation............................................................................................................... .....73 4 MODELS, HYPOTHESES, AND DATA..............................................................................76 Overview....................................................................................................................... ..........76 Empirical Models............................................................................................................... .....77 Social Welfare Function..................................................................................................79 Average Consumption Model..........................................................................................80 Consumption Inequality Model.......................................................................................81 Consumption-based Poverty Model................................................................................83 Hypotheses for the Explanatory Variables.............................................................................85 Demographic Variables...................................................................................................85 Economic Variables.........................................................................................................87 Bio-physical Variables....................................................................................................88 Institutional Variables.....................................................................................................90 Description of the Data........................................................................................................ ...92 The Adilabad Sample......................................................................................................94 The Visakhapatnam Sample..........................................................................................100 The Chittoor West Sample............................................................................................104 Summary........................................................................................................................ .......109 5 RESULTS AND DISCUSSION...........................................................................................110 Introduction................................................................................................................... ........110 Economic Models................................................................................................................ .112 Average Consumption Model........................................................................................112 Consumption Inequality Model.....................................................................................117 Poverty-gap Ratio Model..............................................................................................118 Forest Quality Change Model...............................................................................................121 Discussion of the Effect of Time under JFM/CFM..............................................................127 Average Consumption Model........................................................................................128 Forest Quality Change Model.......................................................................................132 Summary........................................................................................................................ .......137 6 ANALYSIS OF CFM INSTITUTIONS...............................................................................139 Introduction................................................................................................................... ........139

PAGE 8

8 DP 1: Clearly Defined Boundaries......................................................................................141 Spatial Boundary Awareness.........................................................................................142 Awareness of Access Regulations.................................................................................144 General Awareness of Institutions.................................................................................147 DP 2: Congruence between Rules and Local Conditions....................................................149 DP 3: Collective-choice Arrangements...............................................................................149 DP 4: Monitoring.............................................................................................................. ...153 DP 5: Graduated Sanctions..................................................................................................155 DP 6: Conflict-resolution Mechanisms...............................................................................156 DP 7: Minimal Recognition of Rights to Organize.............................................................157 Summary........................................................................................................................ .......159 7 CONCLUSIONS, POLICY IMPLIC ATIONS, AND FUTURE WORK............................160 Overview....................................................................................................................... ........160 Summarized Conclusions.....................................................................................................160 Regression Analysis......................................................................................................160 Institutional Analysis.....................................................................................................163 Policy Implications............................................................................................................ ...167 Recommendations from the Economic Analysis..........................................................167 Recommendations for Training and Extension.............................................................170 Recommendations from th e Institutional Analysis.......................................................173 Caveats of the Study........................................................................................................... ..177 Survey......................................................................................................................... ...177 Data........................................................................................................................... .....178 Future Work.................................................................................................................... ......179 APPENDIX A THE VSS QUESTIONNAIRE.............................................................................................183 B THE HOUSEHOLD QUESTIONNAIRE............................................................................190 LIST OF REFERENCES.............................................................................................................217 BIOGRAPHICAL SKETCH.......................................................................................................222

PAGE 9

9 LIST OF TABLES Table page 3-1 Composition of the Household Questionnaire...................................................................57 4-1 Explanatory variables utilized in the VSS-level regressions.............................................77 4-2 Mean and standard deviation for VSS-le vel data: comparison by the region sampled.....93 4-3 Selected demographic and economic variables for the Adilabad sample..........................95 4-4 Selected bio-physical and institutio nal variables for the Adilabad sample.......................98 4-5 Selected demographic and economic va riables of the Visakhapatnam sample...............101 4-6 Selected bio-physical and institutional variables for the Visakhapatnam sample...........103 4-7 Selected demographic and economic vari ables for the Chittoor West sample................105 4-8 Selected bio-physical and institutional variables for the Chittoor West sample.............108 5-1 Basic statistics for VSS-level re gression analyses (58 observations)..............................111 5-2 Estimated regression coefficien ts and associated p-values..............................................112 5-3 Actual versus predicted Forest Quality Change (FQC) values........................................123 5-4 Estimated marginal effects and pvalues on categoric al probabilities.............................125 5-5 Probabilities of selection and marginal effects of LT......................................................133 6-1 VSS boundary awareness.................................................................................................143 6-2 Access awareness of VSS-member households by caste.................................................145 6-3 HGAI for VSS-member households................................................................................148 6-4 Response frequency indicating that the V SS Micro-plan reflects the interests of:..........151 6-5 Response frequency regarding elec tions for the VSS Managing Committee..................153 6-6 Relationship between the VSS Management Committee and their APFD field officer..159

PAGE 10

10 LIST OF FIGURES Figure page 2-1 Map of India with Andhra Pradesh (shaded).....................................................................45 2-2 District map of Andhra Pradesh.........................................................................................48 5-1 Actual (dots) and predicted (line) in Rupees, versus LT...........................................129 5-2 Actual (dots) and predicted (line) in Rupees, versus LT: using a subset of VSS where PFC > 4.5 (n = 26)................................................................................................131 5-3 Predicted probabilities of FQ C values as a function of LT.............................................134

PAGE 11

11 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SOCIO-ECONOMIC IMPACTS OF COMMUNI TY FOREST MANAGEMENT IN RURAL INDIA By Frederick John Rossi May 2007 Chair: Sherry L. Larkin Cochair: Janaki Alavalapati Major Department: Food and Resource Economics Department This work provides an economic and institutio nal analysis of Joint Forest Management (JFM) as applied in Andhra Prades h, India. JFM is a natural res ource management paradigm that actively involves local stakeholders in the protec tion and management of local forest resources. In essence, JFM is an institutional framework in which state forest departments work in partnership with local communities to restore de graded forests to a productive and sustainable capacity. Concomitant goals of this program are to increase the incomes of rural participants, and provide an equitable distribution of program benefits throughout local communities. In 1992, Andhra Pradesh enacted legislative orders adopting JFM; these orders were subsequently revised in 2002 to authorize Community Forest Manage ment (CFM), a more devolved version that further involves local communities in the decision-making process of forest management. Primary data collected from three regions of Andhra Pradesh ar e used to analyze community-level impacts that this forest ma nagement program has on indicators of economic well-being, inequality, and poverty. Results suggest there is an a ppreciable opportunity cost of adopting the program when all 58 of the sampled villages are analyzed. When restricting the analytical focus to more forest-dependent villa ges, results indicate the JFM/CFM program has a

PAGE 12

12 beneficial impact on economic we ll-being. This implies the progr am is successful in achieving one of its objectives: increas ing economic benefits for the forest-dependent poor. The CFM institutions of Andhra Pradesh are eval uated using design principles that Ostrom identified as key institutions common to successful, long enduring common-pool resource management regimes. Taken as a group, all of the design principles are found to exist in the state legislative orders and at the community-level (t o varying degrees). However, some of the CFM institutions defined by the state only reflect these desi gn principles in general terms, while the institutions implemented by communities are of ten incomplete or ot herwise deficient. Policy implications and recommendations drawn fr om this study are directly applicable to the CFM program of Andhra Pradesh. This study also provides useful information to those interested in the performance of CFM in Andhra Pradesh, or othe rs interested in the general application and performa nce of JFM in India.

PAGE 13

13 CHAPTER 1 INTRODUCTION: PARTICIPATORY FOREST MANAGEMENT Challenges of Resource Management Most natural resource systems are consider ed to be common-pool resources (CPRs). A CPR is typically defined by some type of natural resource stock (e.g., a fore st or a fishery) from which individual units (e.g., trees or fish) are ex tracted. The resource unit s are rival goods, which means that use is mutually exclusive (i.e., use by one individual precludes use by another). This theoretically allows the possibility of excluding specific individuals or groups from benefiting from the CPR. The ownership of a given natu ral resource is often held by the government, although resource users themselves sometimes hold collective ownership. CPRs that lack an owner or clearly defined ownership (e.g., internati onal fisheries) are often characterized as being an open access resource which essentially means there ar e no regulations governing access, extraction, etc. From the use of both renewable resource a nd game theoretic models of harvesting, economists are aware of the dynamics and incentiv es that drive people to over-exploit natural resources. These include, for example, the lack of regulation, a lack of enforcement, and zero user-costs borne by each individual. When a govern ment cannot effectively exclude local people from accessing a natural resource, and if ther e are no communal rights and responsibilities governing use, a CPR effectively becomes an op en access resource. The f undamental problem of open access resources will then follow: the benefits gained from resource extraction are realized individually, while the costs of extraction (in terms of decreas ed reserves, or environmental degradation) are shared by all who use the resource. As there is no guarantee that reserves will remain in the future, or that productivity will be sustained, extraction w ill proceed as fast as possible. Therefore, there is no incentive for any indivi dual to practice cons ervation, because the

PAGE 14

14 resource has no imputed value. The irony is that th ere is a collective social value to conservation: if the community of resource users can effectiv ely organize and cooperate, they can establish stakeholder-devised institutions for the produc tive management of the shared resource. The phrase tragedy of the commons is of ten used to refer to open access overexploitation resulting from the (p erverse) incentives faced by each individual, which themselves are a function of misguided or non-existent in stitutions governing the resource (Hardin, 1968; Ostrom, 1990). In India, for ex ample, government ownership of forest land, coupled with the historical policy of commercial timber extraction, allowed little consideration for the needs of forest-dependent people (Khare et al., 2000). As the government could not effectively exclude rural inhabitants from accessing public fore stland (and with no communal rights and responsibilities for them to adhere to), the forests became a de facto open access resource. The resulting local pressures that a community exerts on a natural resource, such as a forest, are compounded over time by high population growth rates. In India, the impact of deforestation and forest degradation caused by local inhab itants is significant (K hare et al., 2000). India is not the only nation faci ng threats to forest sustainabi lity and the related poverty of forest-dependent people. Around the world, many developing count ries are involved in forest conservation efforts due to concerns for both the l ong-term sustainability of forest resources and the wider socio-economic benefits that healthy fo rest provide to local communities. Beginning in the 1970s, the international development comm unity has been aware of the important interdependence between forests and rural pe oples (Odera, 2004). As traditional forest management policies became increasingly powerle ss to stem deforestation, different agencies and donors began trying more holistic approaches to forest management that included collaboration with local communities (Odera 2004). As such, community-based forest

PAGE 15

15 management eventually emerged as a paradigm fo r forest conservation and rural development in many developing nations; although wh at actually constitutes community forest management is in fact quite fluid and varies widely. Schmink (2004) stresses that the diversity in ecology, resource endowments, socio-cultural systems, and political and economic histories of each community (let alone nation) requi re approaches that are adapte d to particular situations. Interest in forest conserva tion in particular is now wi despread around the world among both developed and developing countries. This is largely due to visually perceptible environmental degradation caused by deforestati on, such as clear-cuts an d their after effects (e.g., soil erosion, etc.). In a ddition, the controversia l issue of anthropogenic global warming has recently sparked a renewed interest in forest conservation, including research addressing carbonsequestration programs. Concurrent with the interest in forest c onservation is the attention that the development community is giving to decentral ized and participatory processes that embrace local human resources. Empowering local people to be at the forefront of their own development, in the context of forest conservation and mana gement, has become a paradigm for sustainable socio-economic and environmental development (Schmink, 2004). Forest degradation has also been a continui ng problem for India, adversely impacting both the environmental and economic stability of ru ral communities that depend upon forests (World Bank, 2000; Balaji, 2002; Tewari, 2006). India is a country of approximately 1.1 billion people (2005 estimate) that is facing the challenge of forest resource sustainability for approximately 200 to 275 million people that depend wholly, or in part, on income derived from forests (Saxena, 1997; Khare et al., 2000). The future need for timber, fodder, non-timber forest products (NTFPs), and other environmental amen ities and services will require a resource management paradigm that ensures productive and sustainable forests. Thus, the challenge of

PAGE 16

16 reversing forest degradation, improving forest resource productivity and management, and improving rural communities is a daunting task, given that India has 678,333 square kilometers (km) of forest land, of which an estimated 42% is classified as open forest, meaning that it is degraded (Ministry of Environm ent and Forests [MoEF], 2006). The Role of Institutions Institutions are the formal and informal rules and regulations that serve as humanly devised constraints. They shape human in teraction and help structure the incentives of human exchange, thereby decreasing uncer tainty in transactions (North, 1991); and their und erlying social organizations are becoming increasingly rele vant for the attainment of socio-economic development and natural resource su stainability goals. This is partic ularly so in rural areas where a lack of cooperation, cohesiveness, and coordi nation frequently hampers economic progress. For this reason, strong local organizations are of ten touted as a key de terminant of successful rural economic development. They are especially important in mitigating the deleterious effects caused by the over-exploita tion of CPRs such as public fore sts. Institutions often originate spontaneously in response to the particular need s of a group of individuals. Healthy institutions (e.g., those that are functioning a nd appropriate) also grow and evolve to fit the needs and demands of the society they serve (North, 1991). Although population pressure has induced the unsustainable us e of key natural resources such as forests in countries such as Nepal and India (Joshi, 1997; Mo EF, 1999), the threats to natural resource sustainability are not limited to population growth al one. The absence of effective governmental and local institutions th at manage natural resource use is just as detrimental to environmental sustainability, if not more so, than population pressures. Indeed, a viable and appropriate institutional framework that is responsible for natural resource

PAGE 17

17 stewardship is a necessary condition for the ac hievement of socio-economic development and natural resource sust ainability goals. Hill and Shields (1998) and Rangachari and Mukherji (2000) provide empirical evidence for the reversal of severe land degradation a nd deforestation following the adoption of Joint Forest Management (JFM) in different states of India. Given the prope r legal and institutional environment, communities have an incentive to or ganize and manage their resource base in a sustainable manner. The work of Elinor Ostrom has been instrumental in explaining how the strength and resiliency of community-level reso urce management organizations is derived from institutions that have evolved over time. In her book Governing the Commons (1990) she uses several case studies from around the world to anal yze successful institut ional arrangements that govern a diverse array of CPRs, including forest s. Some of these case studies include CPR management regimes that have persisted for several hundred years. Delineated from the case studies that Ostrom ( 1990) presents are a particular set of seven institutions referred to as des ign principles that she suggests all successful, long-enduring CPR management regimes are likely to exhibit.1 Most of the design prin ciples are self-reinforcing, and thus function as an integrated system fo r equitable governance of access to, and extraction from, CPRs. Presented below is a brief description of each design principle (DP) and how it provides an incentive to collectiv e action, or otherwise affects th e successful functioning of CPR management. Clearly defined boundaries (DP 1) are basic to the problem and consist of two components: the spatial delineation of a given resource as common property, and the definition of rights of 1 An eighth design principle (nested enterprises) is aux iliary and only relevant for larger and more complex CPR management systems that have an ordered hierarchy (Ostrom, 1990). For example, irrigation associations that must temporally co-ordinate allocation of water amongst several groups of users, and at several spatial scales in a region, often require the other seven design principles to be organi zed in multiple layers of nested enterprises. This design principle has no applicability to the scope of the present study and will not be discussed further.

PAGE 18

18 access and use to a specific set of stakeholders. When taken together, these components lead to a marked reduction in uncertainty regarding who has legitimate access to the CPR, and clearly establishes the exclusion of non-stakeholders fr om access. These boundaries can be thought of as necessary conditions (though not sufficient) for conversion of an open-access resource to a CPR through the imposition of resource management institutions. Congruence between appropriation and provision rules and local conditions is the second design principle (DP 2). The heterogeneity of re source endowments, even amongst separate usergroups (i.e., villages or communities) of the sa me geographic region emph asizes the importance of designing institutions that re flect the specific characteristics of a given CPR. For example, no two physical environments are exactly the same, even for the same type of CPR such as forests. Therefore, human interactions with CPRs w ill display considerable spatial and temporal variability, and will require institutions tailored to the specific parameters and nuances of a particular CPR. Effective and equitable governing structures will give a voice to all stakeholders and requires that most members are given the opportunity to particip ate in the modification of the institutions. These collective-choice arrangemen ts (DP 3) ensure that the institutions are responsive to the very individuals that they serve. In addition, th is responsiveness also helps to ensure that DP 2 (congruence) is functioning well: by being responsive to its members, the institutions are able to respond and adapt to cha nging conditions. This is because the users of the CPR are in the best position to assess the lo cal environment (political and physical) and any changes therein. Monitoring of the CPR (DP 4) is necessary to keep all stakeholders in compliance with the appropriation rules, and to enforce the exclus ion of non-stakeholders. This design principle

PAGE 19

19 addresses the fundamental problem of open access by eliminating the free-rider and forcing each user to pay the cost of their re source usage. Monitors are accountab le to the stakeholders, and are frequently the resource appropriators themselves. Effective CPR institutions do not rely on external enforcement. The commun ity itself is responsible for the compliance of each individual through both formal and informal (i.e., cultural) institutions. Circumstances will arise where temptations to cheat lead to infractions; thus, effective compliance is a function of graduated sanctions (DP 5). There is a certain amount of fairness that is implied with graduated sanctions. They are important so as not to discourage future compliance when minor infractions occur, or wh en first-time offenders are discovered cheating. However, sanctions should be heavy for repeat offenders in order to ensure increased future compliance by those who follow the rules. Mechanisms for conflict resolution (DP 6) that are characterized by low transaction costs are required. To ensure both the fairness and the continuity of the institutions, each stakeholder must have recourse to a forum established for the purpose of dispute re solution. As any set of rules and regulations is subject to different interpretation by di fferent individuals, and as the design principles are attempting to manage an op en access resource, a forum is necessary for the dispensation of justice or punishment to those accused of non-compliance. Again, an external source of governance is not re quired for this function. Minimal recognition of rights to organize (DP 7) is the seventh and last design principle discussed in this study. This design principle relates directly to an external presence. It refers to the need for an over-arching govern mental policy framework that faci litates, or even encourages, community-level institutions that are capable of natural resource manage ment. The most basic

PAGE 20

20 requirement is for the government to recognize th e rights of local people to devise their own institutions regarding CPRs. These design principles provide a context within which part icipatory forestry programs, including those in India, can be evaluated. Altho ugh the particular aspects of JFM institutions in India do not allow perfect adhere nce to all of the design principles advanced by Ostrom, these principles nevertheless pr ovide a framework for late r discussion in this study. A Case Study in India Forest Management The response to the problem of widespread forest degradation in India has been the development and implementation of Joint Forest Management (JFM). The Government of India introduced legislation in 1988 that was specifically designed to promote forest regeneration while addressing the needs of forest-dependent peopl e living in rural areas. This change in forest policy subsequently led to the adoption of JF M in the individual states of India. JFM attempts to provide for the sustainabi lity of public forests by incorporating local stakeholders into the planning, management, and protection of public forests. Economic incentives provide a foundation for community involvement, and help to improve the socioeconomic outlook of the participants. The cornerst one of the JFM paradigm is the participatory involvement of village stakeholders working (in co llaboration with the state forest department) to reforest degraded local public forest lands. Th is was not always the case: forest departments historically followed a reactive approach to fo rest conservation and management that often alienated local communities. In contrast, JFM is a much more proactive approach. In order to participate in the program, a community must establish a forest protection committee and register this or ganization with the forest department. Membership (by households) in this body is voluntar y. Once approved and registered with the forest department,

PAGE 21

21 the forest protection committee is given a tract of degraded forestland to rehabilitate, manage, and protect. Silvicultural tec hniques and other works (e.g., so il and water conservation) are employed to promote the natural regeneration of the area protected by the forest protection committee. Although ownership of the land is re tained by the government, economic benefits conferred to the forest protect ion committee act as incentives for sustainable management and continued participation. As such, the forest pr otection committee is gran ted usufruct rights for most of the NTFPs that the designated forest area yields. Moreover, a critical aspect of JFM is their right to a portion of timbe r revenues (from future plantati on harvesting), which serves as one of the main incentives for community participation in the program. The basic objectives of JFM are to increase the livelihoods of local forest-dependent inhabitants by restoring degraded public forest lands, and to equitably distribute the related benefits throughout local communities. Since im plementation of this forest management paradigm began in the early 1990s, adoption of JFM by local communities has been widespread. Recent data indicate more than 99,000 register ed JFM committees are managing an estimated 214,300 square km of forest land (32% of total forest land) in all of the 28 states that comprise India (MoEF, 2006). The JFM program has ch anged throughout its tenure (MoEF, 2005), however, and continues to evolve over time. On balance, the basic theory behind JFM is sound: addressing the restoration and sustainability of forests through in centive-driven commun ity participation. The result is a move away from open access over-exploitation and toward collective conservation efforts that will realize future dividends to the community. So me empirical evidence ex ists to support this outcome, both in terms of forest resource restor ation and income generation. For example, Khare et al. (2000) states that JFM is successful when measured in term s of the spatial spread of the

PAGE 22

22 program, the number of hectares of forest ma naged under the program, and the regeneration of degraded forests that has occurr ed. Other researchers have documented increased village welfare derived from timber revenues, NTFPs, and envir onmental services (e.g ., increased groundwater recharge) following implementation of JFM pr ograms (Hill and Shields, 1998; Rangachari and Mukherji, 2000; Asia Forest Network [AFN], 2 002). Rangachari and Mukherj i recount clear, if preliminary, local success followi ng the initial implementation of JFM in the state of Andhra Pradesh. Although they caution against premature clai ms of achieving sustainability after only a few years of JFM, their discussion specifically re ferences improvements in forest protection and regeneration, NTFP yields, water ta ble levels, agricultural produc tion, and forest employment. In this lies the strength of JFM: demonstrating that community-lev el participatory involvement in resource management has the potential to reve rse environmental degradation, even when the institutional arrangements are sub-optimal. Nevertheless, some of the same research raises legitimate concerns of the distributional aspects of JFM by discussing how the most disa dvantaged people appear to be those most adversely affected by restricted access to fore sts (Hill and Shields, 1998; Khare et al., 2000). Reddy et al. (2004) discuss how the livelihood im pact of participatory forest (i.e., JFM programs) falls short of expectations in select ed villages of Andhra Pradesh that were studied. Hildyard et al. (1998) raise conc erns that the rhet oric of participation is not matched by realities on the ground. They are highly critical of projects that are par ticipatory in name only. Overall it appears that the broad success of JFM has been mixed. Considering the multiple objectives that JFM is intended to address, it is perhaps not surprising that success has been variable; especially if manage ment of the resource takes de facto priority over socio-economic concerns, as some authors have suggested (S axena, 1997; Rangachari and Mukherji, 2000).

PAGE 23

23 Despite these shortcomings, JFM is the preemine nt forest management paradigm in India, and one of many participatory forest programs currently practiced ar ound the world. As the generic label community forest management has become ubiquitous in describing most (if not all) of these programs, it is in creasingly apparent that there is no universal spec ification of a community forest management model (Odera, 2004) This is largely be cause the programs and institutions being implemented in different countries around the world reflect the specific situations unique to a given country or region. This is as it should be, thou gh; as each country (or greater/lesser region) will be defined by its own natural resource endowments (e.g., geology, climate, etc.), cultural practices, and sociocultural and political history (Schmink, 2004). Hill and Shields (1998) provide an example from th eir case study of two villages, one each in two different states of India. They found that, in general, site-specific factor s (e.g., tree species, locally important NTFPs, etc.) helped to determine the differential economic success they observed between the two cases. Research Objectives Even though its popular appeal is greater in so me states than others, and may vary locally, there is little doubt that the JFM paradigm has spread throughout India in impressive fashion. This is advantageous for research, as continued efforts to study JFM in different locations can help to shed light on how spatial heterogeneitie s (whether environmental, institutional, and/or socio-cultural, etc.) affect th e socio-economic objectives of th e JFM program. Given that the main overarching goals of JFM are to restore degrad ed forest lands and to increase and equitably share benefits derived from forests, it is im portant that contemporary research analyze the impacts of JFM. Although much has been written about commun ity forestry in general, and JFM in particular, relatively few studies (e.g., Hill and Shields, 1998; Reddy et al., 2004) investigate the

PAGE 24

24 economic efficiency and equity outcomes of JFM in India. It is generally believed that JFM has been successful in rehabilitating degraded fore st lands (Khare et al ., 2000; Rangachari and Mukherji, 2000); yet many questi ons still remain: Does JFM ha ve an identifiable economic impact on the villages participating in the progr am in Andhra Pradesh? Is there any impact on inequality or poverty that can be measured? Are the design principl es of Ostrom reflected in the institutions that currently define Community Forest Management (CFM)2 in Andhra Pradesh; and, if so, are they being implemented a nd adhered to in the local communities? In order to answer these questions, this diss ertation has four specific research objectives: 1. To calculate estimates of several economic indicators in JFM/CFM villages of Andhra Pradesh; including the mean pe r-capita household consumption ( ), the Gini coefficient ( ), and the poverty-gap ratio (PGR); 2. To analyze the influence of demographic, ec onomic, bio-physical, and institutional factors on the economic indicators in Objective 1; 3. To analyze the institutions of Community Fo rest Management (CFM ) as implemented in Andhra Pradesh; and 4. To identify the structural st rengths and weaknesses of the Community Forest Management (CFM) institutions based on the findings of Objectives 2 and 3. Methods Two separate questionnaires were utilized in the field survey of this research study. A village-level questionnaire was designed mainly to collect information about the key institutions of each JFM/CFM village selected for the survey, such as the way in which they monitor their protected forest area, for example. Other questio ns inquire about the basic functioning of their 2 Note that in the state of Andhra Pradesh, JFM was modified in 2002 to further devolve the program to allow local stakeholders more control of forest management. As such, the JFM program there is currently referred to as Community Forest Management (CFM). The compound acronym JFM/CFM is used in the remainder of the text to denote when the continuity of the program is being refe renced (e.g., for communities that began JFM in the early 1990s, and continue their activities under CFM in the pr esent). Use of the acronym JFM will henceforth denote the program as applied throughout India, while use of the CFM acronym will generally represent the current set of institutions that define participatory forestry in Andhra Pradesh.

PAGE 25

25 forest protection committee, the historical quality and uses of their prot ected forest area, and about the forest improvement works they have undertaken. A household-level questionnaire was develope d specifically to collect socio-economic information from the households that were sampled in each of the villages surveyed. It is based upon the modular format of the Living Standa rds Measurement Study (LSMS) questionnaire, which was initially designed and tested by th e World Bank in the 1980s (Grosh and Glewwe, 2000). LSMS surveys have been administered aro und the world; and, in fact, the development of the module structure was mindful of the standa rdization of questionnaire formatting and the flexibility needed to accommodate surveys with different research goals. LSMS surveys are composed of numerous independent modules, each one corresponding to particular topics covered in surveys administered in developing countries that seek to gather information related to various development issues. One of the centr al objectives of LSMS surveys is to measure household consumption in order to document living standards and poverty. Deaton and Grosh (2000) explain that consumption directly gene rates a state of well-being (i.e., an actual condition), whereas income and w ealth connote a power dimension (or potential). Thus, the most important concern of LSMS in estimating consump tion is for measuring the distribution of living standards, including poverty (mainly) and inequality as well. The compilation and analysis of the socio-economic data allows for the empirical examination of the interrelated relationship among CFM, its institutions, and the economic welfare and equity of the local communities that participate in the progr am. Dependent variables for regression analysis are economic indicators th at represent components of social welfare and the level of poverty in the co mmunities surveyed; they are analyzed because the JFM/CFM program is largely concerned with poverty alle viation and improving the livelihoods of rural,

PAGE 26

26 forest dependent communities. In this st udy, social welfare is represented as W = (1 ), where is mean per-capita household consum ption (for a given community), and is inequality (as measured by the Gini coefficient). Because both and are two separate components of social welfare, they are analyzed as distinct de pendent variables in differe nt regression equations. A third equation utilizes the poverty-gap ratio (PGR) as the dependent variable. PGR is a measure that quantifies the incidence of pove rty in a given community; it is based on an objective consumption-based poverty line. Demographic, socio-economic, bio-physical, a nd institutional variables represent the four types of explanatory variables drawn from a survey of individual households and a survey of the forest protection committee leaders. Examples of each variable type include: number of households (demographic); income from forest pr oducts (socio-economic); size of the protected forest area (bio-physical); and the presence or ab sence of a formal patrol (institutional). In particular, the JFM/CFM program is represented by a variable that controls for the length of time that a community has participated in the program ; as such, any impact on the dependent variables can be discerned. Summary and Preview Chapter 2 begins with the history of forest ma nagement in India, including a discussion of the revolutionary experiment that JFM was later modeled on. The next section describes in detail the institutional framework of JFM (and later CFM) as it has been implemented in Andhra Pradesh. Then, a general overview of participator y forestry in other countries is offered to provide a basic comparison with JFM in India. The final section of Chapter 2 provides an introduction to the southern Indian state of Andhra Pradesh. A general description of the linguistic diversity, the social cas te composition, and geographic regions of this state is presented

PAGE 27

27 in order to supply the reader with some basic knowledge of the study location, as well as illustrate some of the challenges posed by the scope of this research study. Included is a brief overview of the role of the Andhra Pradesh Fo rest Department (APFD) in administering the CFM program throughout the state. The third chapter describes both the development and implementation of the communitylevel and the household-level questionnaires, and the overall structure and administration of the survey. The first section describe s the questionnaire used to interv iew the leader(s) of the forest protection committees in each village of the samp le. The second section presents the composition and layout of the household-level questionnaire, discussing the design, structure, and formatting in detail. The final section of this chapter examines some important aspects of how the survey was implemented, including the spatial struct ure of the survey and field methods, etc. Chapter 4 begins by presenting details of the specification of the empirical models. The second section discusses the hypot heses on the independent variab les, while the final section presents and describes the data collected by the field survey. Chapter 5 presents the regression results and discusses the general implications. Ch apter 6 returns to the design principles outlined by Ostrom, discussing them in the context of CFM in Andhra Pradesh and the empirical data that were collected and analyzed. The final chapter summarizes the study resu lts and offers overall conclusions and implications of the research. Policy recommenda tions and opportunities for future work in this area are also provided.

PAGE 28

28 CHAPTER 2 BACKGROUND INFORMATI ON ON THE STUDY AREA Overview This chapter is divided into three main secti ons. The first section de scribes the historical management of natural resources and institutions in India, beginning with the pre-colonial era. Then, the immediate predecessor of JFM in India is discussed: the so-c alled social forestry programs of the mid 1970s to the late 1980s. Due in part to its lack of long-term success, social forestry in India was succeeded by JFM around 1990, although the origin of JFM can be traced back to a pilot program undertaken in the state of West Bengal in 1972 (whi ch is also discussed). Thus, the initial foray into JFM actually predates social forestry in time, but eventually the success of this pilot program in regenerating de graded forest lands became widely publicized (Mitra, 1995). This led to JFM ultimately supplanti ng social forestry as it began being replicated, first in West Bengal, and later th roughout the rest of India. This first section ends with a detailed description of the development of JFM in Andhra Pradesh, and its later transformation into the CFM program. The second section of this ch apter presents an overview of community-based forest management in other areas of the world. Nepal, China, and a few other countries in Asia are discussed; this is followed by a general desc ription of participatory forestry in Africa. Community forest management in Mexico is also presented because, although there are some similarities to JFM in India, there are al so some important institutional and production differences. The third section provides an overview of Andhra Pradesh, which includes a general description of the states linguistic diversity, so cial caste composition, and its three geo-political

PAGE 29

29 regions. The role of the forest department in administering JFM/CFM throughout Andhra Pradesh is discussed as well. The final secti on summarizes the discussion of this chapter. History of Participatory Fo rest Management in India Pre-Colonial Era The historical antecedents of JFM lie in the distant past of India itself. Two examples illustrate that community-level control of natural resources is not a foreign concept to the people of India. Indeed, it becomes clear that institutio nal structures can provide the incentives, or the disincentives, which govern the behavior of people. Rangachari and Mukherji (2000) state that, like many other i ndigenous peoples around the world, the tribal peoples of Indi a practiced shifting cu ltivation (also known as swidden or slash and burn agriculture). Forest land is typically cleared, burned, and planted with crops until soil productivity declines. The process then begins ag ain on a new parcel, or an area left fallow is reutilized. Agricultural output is relatively low compared to high-input, intensive farming techniques used today. However, swidden cultiv ation requires no external inputs (other than labor) and, for that reason, is sustainable in the long-run (given relatively low population densities and an abundance of land). Besides raising crops under this system, the forest is utilized for other subsistence activities like hunting and gathering. Forest s are an integral part of the cultural and economic life of such a community, a nd the institutions that govern access and use to the resource will reflect the values of the pe ople who depend on it for survival. As land is held in common (i.e., there are no priv ate property rights), resource sustainability and community welfare become the implicit object ives of resource management. The southern kingdoms of India offer historical evidence of agricultural surplus and wealth in the post-Sangam era, according to Rangachari and Mukherji (2000), who discuss how villagelevel autonomy and community-lev el social organization was th e norm for medieval southern

PAGE 30

30 India. The institutional structure of this er a (A.D. 300 to colonization) emphasized the decentralized administration of pl anning, investment, and manageme nt of agriculture and natural resources. As an example of an appropriate and functioning institution, Rangachari and Mukherji (2000) discuss the structure of in stitutions that govern ed local water resources. Apart from tax collection, the King would preclude himself from any local administration, but would provide such public goods as large-scale irrigation infrastructure. The lo cal organization and control of smaller-scale irrigation infrastructure was facilitated by the appropriate set of incentives and enforcement that bind together the stak eholders of common property resources. The traditional institutions of forest management became incompatible with the centralized and bureaucratic governance structure of the Britis h following the advent of their political and economic hegemony. The imposition of western ideas (such as the nationalization of forests) naturally supported and furthered their control, enabling them to consolidate their power and increase their wealth. For instance, the Britis h promoted the distribut ion of private property rights for continuously cultivated land, and activel y encouraged the clearing of forest land for agricultural production in order to generate farm incomes and subsequent tax revenues (Khare et al., 2000). Another consequence of the changing instit utional structure was the adoption of the prevailing forest management paradigm of the time which viewed timber primarily as a factor of production. Before the introduction of steel and plastic, wood was u tilized in the production of a wide variety of consumption goods. In particular, aside from use as a fuel source, large quantities of timber were used to produce railway ties, ships, and houses. Following Independence in 1947, the government of India essentially carried over this industrial forestry para digm in the interest of its own national development objectives. Th ey maintained exclusive control over the

PAGE 31

31 production and management of forest s, with concessions to industr y providing a steady flow of income to the states. Social Forestry Programs After India gained independence, the forest po licies of the government continued both the land tenure regime and the management practices of the British. In partic ular, the colonial-era paradigm of scientific forest management proceeded well into the 1970s; there was little regard for the natural diversity of tree speci es as forests were c onverted to monoculture plantations (Khare et al., 2000). The needs of local people continued to be subordinate to national economic imperatives. However, burgeoni ng pressure on the state forests from rural people led to a policy change designed to stem fo rest degradation and ensu re future productivity. Although official forest policy rhet oric had always paid only toke n respect to forest-dependent people, the 1976 National Commission on Agriculture (NCA) explicitly stated that management of government forests was to further the producti on of timber for industrial purposes (Khare et al., 2000). But in order to do so productively, the government realized the necessity of relieving the local pressures that were increasingly being placed on the pub lic forest lands. What became known as social forestry was the resulting ve hicle designed to accomplish this objective. The idea behind social forestry was that th e rural demand for small timber, fuelwood, and other forest resources would be met through production on village co mmunal lands, private farms, and degraded or unproductive governme nt land. Government subsidies and technical assistance were provided as incentives to mobiliz e the local population to adopt this paradigm. More importantly (from the governments perspe ctive), the industrial demand for timber would be met by plantation forestry on government fore sts, which would now be free from rural predation. Thus, the continuation of a national fore st policy that served in dustry still left the welfare of rural inhabitants as a secondary consideration. In fact Khare et al. (2000) state that

PAGE 32

32 the major benefits [of social forestry] had been to the central and state governments in the form of cheap raw material and unbridle d access to forests. Both types of benefits were at the expense of forest-dependent people (p. 59). Despite this reality, social forestry was impor tant because it foreshadowed the concept of JFM by allowing tree cultivation on degraded govern ment land, with at least some management responsibilities in the hands of local people. For example, the community plantations component of social forestry established government-sponsored plantations on community and/or government lands. Communal grazing areas, degraded gove rnment forests, and roadsides are examples of lands that were planted to m eet the small timber and fuelwood needs of local inhabitants. This program could not be su stained beyond the first harvest, however; due in part to a lack of institutions defining both the rights to the trees and the dist ribution of benefits (Khare et al., 2000). In addition, other factors that contributed to the demise of the program were that the forest departments exhib ited a poor u nderstanding a priori of the actual perceptions and practices of local people vis--vis trees, tim ber, and related forest issues. For example, the assumption that the community plantations would be utilized fo r fuelwood was proven false when it became clear that local authorities view ed these plantations as a source of communal income. Farm forestry was the name applied to th e other component of social forestry. The defining feature of farm forestry was tree produ ction on private lands. Farmers were given free or subsidized seedlings, and no restrictions we re placed on what to do with the output (e.g., use as fuelwood or timber). Initially this program wa s very successful, thriving beyond the original expectations of planners, who ha d anticipated social forestry as a scheme for locals to produce their own fuelwood (mainly) and small timber for th eir own use. In realit y, however, participants

PAGE 33

33 were cultivating commercially valuable species in response to the economic incentives provided by the program. As they did not regard the producti on of trees for fuel as a valuable use of the resource, they instead viewed trees as an inve stment. Thus, farmers preferred to focus on the production of poles, small timber, and/ or pulpwood for commercial gain. The subsequent initial success of farm fore stry was so great that the government was unable to assist with the marke ting of surplus timber. Thus, timber prices fell drastically in many states, which negatively impacted further progr am participation. Pulpwood subsidies to paper mills, and legal restrictions on the sale and transpor t of timber, also contributed to the low prices that eventually signaled the end of farm fore stry in the late 1980s. Although farm forestry neither significantly met local n eeds nor improved private wastela nds, an important realization was that farmers responding to the right incenti ves and prices could meet most of the raw material needs of industries based on wood (Khare et al., 2000, p. 59). The Arabari Experiment The origin of JFM can be traced to what Saxe na (1997) refers to as the Arabari experiment. In 1972, an enlightened Divisional Forest Officer in the state of West Bengal took control of 1,272 hectares of deforested land. This area had once been very productive as a good source of fuelwood, food, and livestock fodder. By 1972, however, it no longer produced commercial timber, and the soil erosion associated with th e loss of tree cover was adversely affecting the local agriculture. In order to ensure successful fo rest regeneration, this of ficer realized the need for an integrated development plan to eliminat e the cycle of degradati on. Because the collection and selling of fuelwood in local markets would hinder any afforestation efforts, the officer banned such activity. Income lost from this s ource was replaced by the employment of local people for the forest restoration work.

PAGE 34

34 The majority (55%) of this land was mana ged for the re-growth of coppiced sal ( Shorea robusta ), an important tree species that provides NTFPs in add ition to timber. Most of the remaining land was cultivated as plantation of acacia ( Acacia spp. ), eucalyptus ( Eucalyptus spp. ), cashew nut, etc. In addition to the wages r eceived for their labor, the participants were promised 25% of the final timber harvest (subje ct to the success of the project). Moreover, to ensure that the needs of the participants were met during the maturation of the forest, rice and fuelwood were grown on the land. Rice was subseque ntly sold at cost, while fuelwood was sold for a token price. Rotational grazing was also allowed. By creating incentives to regenerate, manage, and protect this plot of land, the Arabari experiment placed local people at the center of e fforts to rehabilitate a portion of the states forest. The program incentives provided for thei r intermediate needs through wage income and low cost subsistence goods, while the share of timber revenues gave them a stake in the longterm health and productivity of the forest. Saxena states that this arrangement made it clear to the people concerned that they had a right to enjoy the enhanced bene fits of forests, but this right was accompanied by their duty to nurture and protect the forests (1997, p. 99). In 1987, 618 families were awarded a share of the timber revenues following the harvest of 97 hectares of sal and eucalyptus. In addition, each family receive d wages paid for harvesting the timber. The success of this model is evident in that ove r 2,000 villages had adopted the program by 1993, and that approximately 60% of the forest area in the region is managed under this model (Saxena 1997). Development of Joint Forest Management (JFM) Recognizing the value of grassroots participa tion to successful afforestation projects, the Government of India (GOI) enacted the 1988 Nati onal Forest Policy that subsequently enabled the adoption of JFM programs throughout India (W orld Bank, 2000). This policy change shifted

PAGE 35

35 the use of public forests away from commercial exploitation and towards the subsistence needs of forest-dependent people (Saxena, 1997; Khare et al., 2000). This was a radical departure from past forest policies and reflected a new understanding that the t ragedy of the commons is not necessarily an inevitable outcome of resource us e by locals, but rather a function of institutions which are either non-existent or ineffectiv e. This new understanding came in part from demonstrated successes such as the Arabari Experiment. Policy makers were also influenced by the changing political winds. Although environmental activism is not a new phenomenon in I ndia, but is rooted in the past, the plight of forest dwellers was not promoted by intellect uals and activists unti l the early 1970s (Saxena 1997, pp. 40-41). Widespread attention to the diss ent and unrest of forest communities began with the non-violent Chipko movement in northe rn India in 1973, and c ontinues to the presentday with the armed rebellion by so-called Naxal ite groups in Andhra Pradesh and other states. Although the 1988 National Forest Policy reoriented forest policy to the needs of the local stakeholders, the June 1, 1990 Circular actually specified the basic rights that people have in relation to forests under their prot ection (Khare et al., 2000). It ex plicitly enabled the state forest departments to engage locals in the management of forests. Thus, this resolution directly facilitated the implementation of JFM without ever mentioning the words joint forest management. Clearly JFM is the reference poin t, as the text mentions how Village Forest Protection in West Bengal receive s a 25% share of timber revenues. In addition, the text states that similar institutions may be adopted by the other states (Government of India [GOI], 1988), and also encourages the fore st departments to work with non-governmental organizations (NGOs) as intermediaries between the governme nt and the local communities (Saxena, 1997).

PAGE 36

36 With this legal order, state gove rnments now had the authority to draft their own legislation that specified the particular institutions that w ould define JFM in their respective states. JFM in Andhra Pradesh was initiated unde r G.O. MS. No. 218 in August 1992. With reference to the 1990 Circular of the GOI, this order commands the Andhra Pradesh Forest Department (APFD) to introduce and implement the JF M concept to all districts of the state. It provides some general instructions for implementa tion, most notably calling for frequent reviews so that analysis can lead to beneficial amen dments. The order also directs the local village community to be organized into a Vana Samrakshana Samithi (Sanskrit for forest protection committee) if the community collectively desires to participate in JFM. VSS is the acronym for Vana Samrakshana Samithi, and it is commonly used in Andhra Pradesh to denote the forest protection committee general body. The Annexure to G.O. MS. 218 contains the specific institutions of JFM that spell out the composition, functions, respons ibilities, and rights of the VSS and its managing committee; thes e details are summarized next. Given a quorum of 50% of village households, a Forest Officer will explain the concept of JFM to the assembled community. A VSS will be fo rmed if sufficient interest exists, and every household in the village has the opportunity to join. Any two members from a given household are allowed to join the VSS; however, one must be a woman. Each member of the VSS general body will, individually and collectivel y, protect the forest area agai nst grazing, fire, and theft of forest products. In addition, members will assist the forest department in implementing a jointlydeveloped forest management plan (known as the micro-plan). The VSS general body will meet every six months to review the plan. Every VSS shall have a Management Comm ittee (MC) that is charged with the responsibility of carrying out the approved JFM micro-plan. The MC will convene monthly, and

PAGE 37

37 the term of service is one year. It will be co mposed of six to ten members elected from the general body of the VSS, the pr esident of the local government council, and two members of the forest department (who do not have voting privileges). The micro-plan is to be developed by the APFD in consultation with the MC, and will apply to a specifically designated tract of degraded public forest land selected by the MC and the APFD. It will focus on supporting the demand for tr aditional forest products local to the area, and include measures designed to aid the regeneration of the fo restsoil and water conservation measures, in particular, are to be an integral component. Planting of lowvalued fruit trees such as tamarind is allowed, but horticultural spec ies like mango and guava are not permitted. The conservation and development works of the microplan are to be coordinated by the MC, as are the paid and un-paid labor inputs. First preference for paid labor is to go to VSS members. The micro-plan will be in effect for 10 years and is subject to revision by the forest department. Given adherence to its duties and responsibilitie s, the VSS is granted usufruct rights to the forest. The VSS, acting through the MC, is resp onsible for the equitabl e distribution of the usufruct benefits entitled to V SS members. Discretion is given to withhold or lessen the share of benefits according to the contri butions, or lack thereof, of i ndividual members. Each household is considered as one member for the dispensation of the usufruct benefits Rights are divided into two classes. Non-reserved rights ar e granted to leaf and grass fodde r, thatching and other grasses, thorny fencing material, and deadwood. Reserved rights apply to certain NTFPs under contract to a parastatal, and to timber and poles. Afte r three years, access fo r timber and poles are afforded to the community subject to the JFM micro-plan: harvest is shared between the VSS and the APFD, with each receiving 50%. Usufruct rights shall only apply to VSS members, and

PAGE 38

38 any disputes are to be adjudica ted by the MC. The Conservator of Forests (a high-ranking APFD official) has the authority to re lax these rules and regulations. The first revision of the initial legal orde rs came in December 1996 with G.O. MS. No. 173. As well as changing some elements of the orig inal order, it also provides more detail to the composition, functions, responsibilities, and rights of the VSS and its management committee. Important changes included: Ensuring participation of all tribal households and households from the lowest castes by making their membership automatic. Such househol ds are also allotted a certain percentage of membership in the MC. Increasing the number of elected MC members to 10 15 (of which 30% are to be women), and increasing the term of the MC to two years. Allowing the VSS to apprehend offenders and turn them in to the concerned authorities. Devolving more power to the VSS to prepare th e micro-plan, especia lly with the goal of including the input of women a nd more disadvantaged groups. Allowing the VSS to select the species for pl antations, and relaxing other restrictions on what type of trees can be planted. Specifying that all labor contributions be paid. Relaxing the restriction on most NTFPs, a nd specifying that 50% of net income from collection of Beedi leaf (for locally produced cigarettes) be paid to VSS members. Increasing the share of timber and bam boo harvest received by the VSS to 100%. Advent of Community Forest Management (CFM) In February 2002, the legal orders for JFM we re re-written to place greater emphasis on community participation and autonomy. As suc h, JFM was superseded by Community Forest Management (CFM) under G.O. MS. No. 13. According to the APFD, CFM aims at decentralizing the entire process of planning a nd implementation with APFD and Government of Andhra Pradesh (GOAP) acting more as facilitators and providers of technical and infrastructure support to local stakeholders (APFD, undated, p. 4). This is in contrast to JFM, where the

PAGE 39

39 VSS/government relationship is characterized as being more like a partnership. The main changes included revising the membership and functions of the MC, and creating other support councils for the VSS. Some minor revisions to NTFP, timber, and bamboo rights were made also. The important changes were: The VSS can now collect fines (less than 100 Rupees) for minor forest offences. VSS are entitled to all intermediate yields obtained from silvic ultural operations. The MC is now comprised of 15 elected members from the VSS, and at least eight must be women. The MC tenure is in creased to three years. The MC will elect a Vice-Chairperson. Either the Chairperson or Vice-Chairperson must be a woman. The Vice-Chairperson is to work closely and assist the Chairperson with the dispensation of his/her duties. The Chairperson will maintain VSS account books, micro-plans, minutes books, etc. The MC will account for and manage the VSS funds and other resources. There are two accounts. The Government Acc ount contains funds received from the government, and is jointly operated by the Chairperson, Vice-Chairperson, and APFD representative. The VSS Account contains internally genera ted funds, and those derived from non-governmental sources. It is opera ted jointly by the Chairperson and ViceChairperson. APFD staff, NGO representatives, etc. are no longer part of the MC, but will form an Advisory Council to review micro-plans, and advise the VSS on strategies and resources. Other councils will be created to review th e implementation of CFM and provide direction to the APFD regarding the ru ral development. Representatives from the APFD, other government agencies, NGOs, and selected VSS will be included. These councils will be created at the district, forest division, and state levels. G.O. MS. No. 13 represents the current set of institutions authorizing and defining CFM in Andhra Pradesh. This document is a significant imp rovement over the original order authorizing JFM in 1992, both in terms of the actual instituti ons and the clarity of presentation. Although the general program parameters ar e specified by the government, as opposed to being organic institutions self-derived by locals, the clear inte ntion of CFM is to recreate a similar institutional

PAGE 40

40 environment. The continued involvement of th e government has helped the program evolve towards a framework within which local stakeholders can fine-tune certain parameters to fit their specific situation. Thus, from initiation in 1992 to the advent of CFM in 2002, the various updates have ensured and/or strengthened, in pr inciple, the institutions of JFM/CFM, which characterizes participatory forest ma nagement in Andhra Pradesh today. Comparison to Participatory Forestry in Other Nations Nepal The causes of deforestation in Nepal, and s ubsequent adoption of co mmunity forestry, are similar to the experiences of its neighbor Indi a. Population pressures had already begun to impact forest resources in Nepal by 1957, at wh ich time the nationalization of the countrys forests took placeultimately leading to a degraded ( de facto ) open access resource (Joshi, 1997). The legal orders for participatory forest management in Nepal date to 1978. According to Joshi, however, orientation of the policy at th is time was more towards management by local leaders or local political units (i.e., panchayats) than collective management. Later policy revisions ensured that local stakeholders were empowered to manage their forest resources through the formation of Forest User Groups (J oshi, 1997). These legally recognized groups are given rights to timber and NTFPs from the forest area under their management, and there is also an emphasis placed on the participation of women. Unlike JFM in many states of India, however, even well-established forests in Nepal are eligib le for the program in addition to degraded lands. China and Other Asian Nations The 1981 Forest Policy enacted by China be gan to loosen the government monopoly over forest resources. Essentially, policy changes pr omoted afforestation by leasing forestland to individual households, and forest farmers were given greater fl exibility and control of their management (MoEF, 2006). In general, the mode l became less strictly collective and more a

PAGE 41

41 combination of collective and individual ma nagement. In 2002, new laws were passed to strengthen the security of collective forests, which account for approximately 60% of Chinese forests (MoEF, 2006). In addition, forestry income is shared with the farmer based on labor input and other factors. Other Asian nations are also pr omoting various degrees of co mmunity forestry. In Bhutan, small groups of at least five people can obtain use ri ghts to forestland if th ey agree to regenerate it according to a management plan. Changes to forest policy in Myanmar have enabled reforestation cooperatives at the village level, and community forest management. The program allows all benefits derived from the community-m anaged lands to remain with the stakeholders. Vietnam is decentralizing the management right s and responsibilities to provincial and local authorities (MoEF, 2006). Africa Odera states that community forest management in sub-Saharan Africa dates from the late 1980s and early 1990s, but that these early effo rts were generally focused on a narrow band of linkages between people and trees (2004, p. 16 ). Such linkages included exchanging forest access for labor, and joint management scheme s that borrowed principles from wildlife management services. In the inte rvening years, community forest management has continued to develop and evolve as the ope n-ended definition of this paradigm has allowed different countries to devise and/or adap t institutions that are specific to their own experiences and circumstances. However, Odera also notes that different management regimes have emerged that vary from full-fledged particip atory management to only token representation by local people. Nevertheless, experience has shown that commun ity forest management has been successful in reducing deforestation and improving forest co ver and benefitsbut only when people have been empowered with responsibility and have been given secure tenure rights (Odera, 2004).

PAGE 42

42 Most sub-Saharan countries have taken st eps toward implementing community forest management, although most of these programs are clearly in the developing stages with many less than five years old. Like JFM in India, the general rule seems to be that the local stakeholders must register with a government ag ency. This allows them official recognition and formal use and management rights that, combin ed with a management plan, usually cover a period of five to fifteen years (Odera, 2004). Unlike India, however, some countries have granted permanent land rights or ownership titles to local communities; while others have granted permanent titles to forest resources. Management plans are often required for such tenure as well. However, most benefit flows in Africa have been confined to NTFPs because restrictions on tree felling have been imposed in orde r to rehabilitate the forest, ev en when rights have otherwise been transferred to the local people. There appears to be wide variation in terms of the types of forest to which community forest management in Africa is applied. For exam ple, only unclassified forests in Cameroon are eligible (under a 10-y ear agreement), with a maximum si ze of 5,000 hectares. Other countries allow community forest management in reserved fo rest areas, even those that are classified as high-priority in terms of conservation (Odera, 2004). Despite the advent of community forest ma nagement in Africa, there are significant problems. Support for community forest manage ment in many countries is lacking due to internal problems in national forest department s. A lack of funding a nd training are also key constraints. Thus, the role of non-governmental or ganizations (NGOs) is cr itical in helping to serve these needs, and to better implement this management paradigm in general. Mexico Agrarian land reforms in Mexico throughout the twentieth century created substantial amounts of common property in the rural sector, mu ch of it forested (Bra y et al., 2006). Despite

PAGE 43

43 legal rights granting control over common pr operty to local communities, the government maintained significant control over forest resources (and also reta ined ownership of the land until 1992). From 1940 to 1970, government control was exer ted in the form of ineffective bans on logging and the granting of concessions to harv est timber from community lands (Bray et al., 2003). During this period, modest attempts to involve locals ar ose through government efforts that trained local communities to manage commun ity forest enterprises (CFEs). However, these actions were undertaken mainly in support of industries dependent upon timber and do not constitute a serious attempt to empower locals stakeholders by involving them in the direction, planning, and management of forests for their own benefit. In the 1970s, grassroots activism and legal reform stimulated in terest in true participatory forest management at the community level, and led to the development of new CFEs. Subsequent legal reforms lent additional s upport by devolving more power to locals, and by funding various projects that directly focused on community fore st management and development. Thus, in many respects the precursor conditions leading to community forestry in Mexico were similar to those in India. The particular manifestation of community forest ry in Mexico is quite different from that of JFM, however. The main diff erence is that many of the CFEs in Mexico are engaged in commercial production of timber and timber produc ts. Bray et al. (2006) cite an unpublished study that reports over 2,400 communities in Me xico were engaged in commercial logging in 2002. Furthermore, in a previous pa per Bray et al. (2003) state: What is unique about the Mexican case is the large number of communities that are managing common-property forests for the commer cial production of timber, as well as finished timber products in some cases, in industrial processes th at are thought to be beyond the reach of most poor, rural communities.

PAGE 44

44 Moreover, Bray et al. (2003) discuss how comm unities are managing their forests for multiple benefits and with due considera tion for the increasingly robust environmental laws. They also say that CFEs often voluntarily harvest timber at rates less than their management plans allow. Why such a focus on timber production? Mexican forestry policies, ru ral activism, and the traditional system of rural organization have combined in such a way that allows some CFEs to respond competitively to market forces (Bray et al., 2003). In additi on, a history of past experience with commercial logging (i.e., concessi ons), and a legacy of resource use, may have left a residual attitudinal imprint in rural areas The size of communal fore st areas in Mexico is also relatively large (a t least by Indian standards), averaging 3,074 hectares, according to calculations from the unpublished data cited in Bray et al. (2006). Theref ore, large tracts can better accommodate competitive commercial lo gging without being destructive of the environment, because a proportionately larger volume can be harvested. Description of Andhra Pradesh Location Andhra Pradesh is situated along the Bay of Beng al in the southeastern part of the Indian subcontinent (Figure 2-1). The fifth largest st ate in terms of area (276,754 sq. km), Andhra Pradesh comprises 8.4% of the tota l territory of India; and, at 972 km, has the longest coastline of any Indian state (Tata, 2002). The 2001 popul ation of Andhra Pradesh is 75.7 million, of which nearly 73% (55 million) are rural inhabi tants (Tata, 2002). With such a large rural population, agriculture is a key se ctor. Not only is Andhra Pradesh a net producer of cereals, but it also leads all Indian states in the production of rice (Reddy et al., 1992). Cash crops important to Andhra Pradesh include sugarcane, tob acco, cotton, and groundnuts (peanuts). Commercial enterprise and industrial producti on can also be found in the two largest cities: Hyderabad and Visakhapatnam.

PAGE 45

45 The capital city, Hyderabad, is lo cated in the northwest part of the state and is a sprawling metropolis with a 2005 population of 6.1 million ( UN, 2006). Hyderabad rivals Bangalore as a leading center of information technology in Indi a, and is second in film production only to Bollywood (in Bombay). Hyderabad is also a major center for pharmaceuticals and biotechnology. The coastal city of Visakhapatn am (2005 population of 1.5 million) has such economic activities as steel production, petroc hemicals, and shipping (Chagari, 2005; UN, 2006). Figure 2-1. Map of India w ith Andhra Pradesh (shaded).

PAGE 46

46 Languages and Castes The principal language of Andhra Pradesh is Te lugu, which is a Dravidian language that is very similar to the languages of the other states of southern India (e.g., Tamil). Urdu is widely spoken in Hyderabad, owing to the large Muslim popul ation (nearly 50%) in this city. English is spoken by many as a second language, usually by people who have been educated in English medium schools where it is the focal language. In the rural countryside, however, English is likely to be known by very few individuals. The social division of people into castes is a well known aspect of the Hindu religion. There are literally hundreds of castes, each defi ned by specific names and often indicative, at least historically, of an occupation. The GOI ha s delineated four main categories from the many castes and tribes: Scheduled Tribes (ST), Schedul ed Castes (SC), Backwa rd Castes (BC), and Other Castes (OC). This classification is for legal and bureaucratic purposes, as affirmative action-type programs in India are based on this division. These categories are commonly referred to by their initials. Scheduled Tribes include the various tribal populations in India that have their own languages, which are essentially verbal only. Th ey have their own cultural customs as well, although many elements of Hinduism have been ab sorbed. In addition to their own language, most tribal people in Andhra Pradesh speak Telugu. Near the border of an adjacent state, they may speak the official languages of that state in stead of Telugu. Scheduled Tribes are generally poorly educated and engage in subsistence farmi ng and/or casual labor activities. Most of the more remote villages in the mountainous regi ons of Andhra Pradesh are Scheduled Tribes villages, and their dependence upon NTFPs is usually higher than other social groups. Scheduled Castes is the term given to the castes that were formerly identified as Untouchables by the rest of Indi an society. Scheduled Castes t ypically occupy the most menial

PAGE 47

47 jobs (which are often stigmati zed), they usually own no land, a nd are generally poorly educated (as compared to the higher castes). In order to improve the long-term we lfare of the Scheduled Tribes and Scheduled Castes, the Indian gove rnment has written into the constitution an affirmative action-type program that guarantee s access to civil service jobs for members of these two social designations. Re ddy et al. (1992) estimate that Sc heduled Castes and Scheduled Tribes account for 5.9% and 14.9% respectively, of the populati on of Andhra Pradesh. Thus, nearly 80% of the population of Andhra Pradesh is comprised of Backward Castes and Other Castes. Backward Castes include the gr ouping of castes that socially lie between the lower level of Indian society (i.e., Scheduled Tribes and Sc heduled Castes) and the upper castes. Backward Castes are largely analogous to middle class society in the United States; a typical occupation might be a merchant or a military officer. Other Castes comprise the top level of Indi an society. They are generally better off, economically and educationally, than the Backwa rd Castes, and certainly more so than the Scheduled Tribes and the Scheduled Castes. Geographic/Political Regions Figure 2-2 is a map of Andhra Pradesh that is divided into 23 distri cts. The basic sociopolitical unit of a given Indian state is the dist rict, which is roughly anal ogous to counties in the United States. More broadly, Andhr a Pradesh is comprised of th ree different socio-political regions known as Telangana, Rayalaseema, and Coastal Andhra. Andhra Pradesh was formed as a state in 1956, along the li nguistic bounds of Telugu, fr om these three regions.3 3 As such, it is quite possible that one day Andhra Prades h will split into three separate states: Jayashanker (2004) discusses how the recent demands for a separate state of Telangana predate the formation of Andhra Pradesh by a decade.

PAGE 48

48 Most of Andhra Pradesh reminds one of the southwestern United States in terms of physical features and climate, although the coastal z one is much more humid and lush in terms of tropical vegetation. In general, Andhra Pradesh can be characterized as follows: forested mountains in the north, north-central, and north east areas, dry upland plateau with smaller mountain ranges in the central-west and s outh, and humid plains along the coast. Figure 2-2. District map of Andhra Pradesh. The physical-climatic description above broadly conforms to the Telangana, Rayalaseema, and Coastal Andhra regions. Telangana is comprised of nine districts in th e north-northwest parts of the state (the triangular shaped area fr om MahbubnagarKhammamAdilabad in Figure 2-2).

PAGE 49

49 Of these, the four northernmost districts (K hammam to Adilabad) contain forested mountain slopes where teak ( Tectona grandis ) is the naturally predominating tree species, as well as the main tree species managed by most VSS villages in the region. Bamboo grows well locally and is also cultivated in such ar eas for its multiple uses. The other districts of Telangana are somewhat drier and steppe-like, having more exposed bedrock a nd less forest cover (as opposed to the more mountainous and forested terrain in the north). Overall, Telangana receives approximately 890 mm (35 in.) of rain annually, with most of this occurring in the summer monsoon of June to Septem ber (Reddy et al., 1992). The four southern in-land districts of th e state (Kurnool to Chittoor) comprise the Rayalaseema region, and are similar to the southern districts of Telangana in that rock outcrops are ubiquitous. Much of this re gion lies on a rocky plateau inte rspersed with higher mountainous areas and valleys; the climate on the plateau is somewhat more agreeable than surrounding lowlands due to the elevation. Nevertheless, Rayalaseema receives only 670 mm (26 in.) of precipitation per year and is prone to drought co nditions, like much of Telangana (Reddy et al., 1992). The density of forest cover on the hillsides is generally quite low (extremely low where degradation is advanced), and these slopes main ly contain low quality tree species (which are often called scrub forest). Native tree spec ies of great value, such as sandalwood ( Santalum album ) and Red Sanders ( Pterocarpus santalinus ), are largely gone except for individual villages that are attempting to cultivate Red Sanders. Al though non-native to India, eucalyptus is widely cultivated there as part of the CFM program, larg ely because it is a fast growing species used for pulp, and because cows do not like to graze on it. Coastal Andhra is comprised of the district s adjacent to the Bay of Bengal. The low mountain ranges in the west ern part of the southern districts (Nellore to Guntur), and the high

PAGE 50

50 mountains in the northwestern parts of East Godavari, Visakhapatnam, Vizianagaram, and Srikakulam districts form the physical boundary to the west. In between these mountains and the ocean are humid, tropical plains (with some steep peaks locally) that receive ample moisture to ensure a lush and productive agriculture. Co astal Andhra receives approximately 1,000 mm (39 in.) of rain annually (Reddy et al., 1992), and this region is gene rally regarded as being more prosperous than Telangana and Rayalaseema. Much timber is grown commercially here owing to the beneficial growing conditions. Speci es include eucalyptus and casuarina ( Casuarina spp. ), and sal in the district of Srik akulam. Fruit plantations are ubi quitous as well, and bananas, coconuts, cashews, papaya, etc. are all cultivated there. The Forest Department As the agency commissioned with facilitati ng the VSS in their implementation of CFM, the APFD is the most important entity other than the local stakeholders themselves. As such, it is important to summarize their main role under CFM. The job of the field level staff is to work closely with the VSS, providing technical guidance when appropriate. For example, the forest department is tasked with preparing estimat es for VSS project works until the VSS has the capacity to do so. However, the VSS which are able to take up this responsibility are encouraged to do so at the earliest (Government of Andhra Pradesh [GOAP], 2002). APFD field officers also ensure that VSS are in complian ce with the CFM rules and regulations, and that they conform to all other pertinent GOI and GOAP laws. In addition, the forest department is res ponsible for providing training for forest management and planning, includ ing specialized technical trai ning for aspects such as the grafting of high-yielding clonal va rieties, raising and management of tree nurseries, etc. This is an important component of the program because ex tension work is crucial for creating a viable program that can endure in th e long-run. The transfer of skill s and knowledge, especially in

PAGE 51

51 support of the development and enhancement of livelihood opportunities, is necessary for improving the incomes of the program participants. Th is will help to sustain participation in the CFM program when the external funding ends. Concurrent with the transition to CFM is the development and implementation of the Andhra Pradesh Community Forest Management project (APCFM ), supported by major funding provided by the World Bank. This pr oject leaves no doubt that the fo cal objectives of CFM are to alleviate rural poverty through the improvement of livelihoods, by improving forest management and productivity. The APCFM Project Implementa tion Plan also acknowledges the need for developing alternate livelihoods, an d the need for providing coordi nated technical and financial inputs to the VSS (APFD, undated) Therefore it is important to note that this project is essentially a rural development scheme, set in th e context of participatory forest management. In many respects the APFD is put in an une nviable position by being the lead agency implementing the APCFM project. Outside of th eir tradition role, they are now charged with running a rural development programsomething that is not the forte of this organization. It is worth noting that the APFD is aware of this fact at least rhetorically, and is making efforts to adapt: One of the prerequisites for successful CFM is attitudinal change in forest department from one of command and control to that of r ecognizing communities as equal partners. With the introduction of CFM th ey will be required to don th e role of Facilitators and Extension workers. Foresters will be mostly performing regulatory role and they will be facilitating community participation and provi ding technical and in frastructure support. This also warrants greater shift in mindset which can be ensured only through massive trainings. (APFD, undated, p. 9) These observations indicate that the APFD ma nagement structure is committed to CFM, and takes its implementation and f acilitation responsibilities serious ly. Indeed, others agree that the APFD has made substantial progress in adap ting from the traditional command and control

PAGE 52

52 mode of operation, to the role of Facilitators and Extension worker s that participatory forestry requires (AFN, 2002). Summary The legacy of the institutional changes ushered in by the British is still being felt today. A near complete erosion of the traditional institutio ns of forest management, coupled with dramatic increases in population in the twen tieth century, has resulted in th e deforestation and degradation that has left millions in poverty. The poor and other marginalized people of rural communities dependent upon forests for their livelihoods have been the principal losers from government policies that traditionally favored timber producti on (Hill and Shields, 1998; Hildyard, et al., 1998; Khare et al., 2000). Often these people ha ve no other choice but to illegally harvest firewood for both income generation and domestic consumption (Khare et al., 2000). Since implementation of JFM in India offici ally began in the early 1990s, progress has been evident in terms of the adoption of the program by local communities. Citing data from MoEF and others, Khare et al. (2000) report that by the end of the decade there were nearly 35,000 registered forest protecti on committees in 16 states of India; the area under JFM was estimated to be over 7 million hectares, and perhaps as high as 9 million hectares. Some of the forest protection committees recorded are not du e strictly to JFM, however; as many of these committees were established, in some states, under other participatory management regimes (e.g., van panchayats in Uttar Pr adesh) or were self-organized as in Bihar and Orissa. Recent data shows that the JFM program is still expanding, as over 99,000 registered JFM committees now manage an estimated 21.4 milli on hectares in all 28 states (MoEF, 2006). Several states in particular have shown growth of JFM in terms of numb ers of forest protection committees and total area under joint manage ment, including Andhra Pradesh, Madhya Pradesh, West Bengal, and Rajasthan. Andhra Pradesh not only possesses one of the largest JFM

PAGE 53

53 programs in India, but the APFD has made significant progress in adapting to the changing priorities in forestry that JFM represents (AFN 2002, p. 1). Recent figures indicate that there are 8,343 VSS in Andhra Pradesh; the coverage ar ea is about 2.3 milli on hectares (APFD, 2006)4. Some of the initial JFM institutions implement ed in Andhra Pradesh are robust. One strong point is that participation by commu nities is clearly volunt ary. It is inclusive in that everyone in a participating community has an opportunity to join the forest protection committee. Another strength is that the benefits are clearly spelled out. In addition, several of Ostroms design principles are represented to va rying degrees: clearly defined bounda ries [mainly in terms of who has resource access] (DP 1); congruence between rules and local conditions (DP 2); collectivechoice arrangements (DP 3); monitoring (DP 4); and dispute resolution (DP 6). A main weakness of the program, however, is that the JFM institutions are exogenous at the village level as they are originated with government legislative orders. Although the state governments provide for minimal rights to organi ze (DP 7), their influence on the whole process is much greater than Ostroms DP has it in mind. This will be discussed in more detail later. There are several other important weaknesses, as well. For example, women are not accorded a significant role in VSS management; they are only mentioned in reference to household membership in the VSS general body. The govern ment role has not devolved enough power to the local stakeholders: sufficient VSS autonomy is lacking, as evidenced by the fact that the APFD is represented on the MC. In addition, th ere are too many restric tions on the usufruct rights given the VSS as incentives for program participation. That changes were made to the JFM program in Andhra Pradesh is important not only for the improvement of the institutions themselves but because they demonstrate the responsive 4 These data were obtained from the APFD website on Sept. 7, 2006.

PAGE 54

54 involvement of the government in terms of its willingness to make amendments. As a result, equity concerns were promoted by attempting to better facilitate the participation of the disadvantaged (e.g., tribals and other poor) a nd women. More control over enforcement and planning was given to the VSS, and improvement s to the economic incentives were made. For example, the share of timber revenues was increa sed to 100%. In addition, further revisions now allow the VSS to receive a percen tage of fees collected (by th e authorities) from smugglers apprehended by the VSS. The share of fees (or the forest produce in question) to be paid to the VSS was initially set at 25%; it was later increas ed to 50%. With the evolution of the policy, graduated sanctions (DP 5) are seen to be represented implicitly. In conclusion, India is not alone in promoting community-based forest management programs in order to redress forest degradation an d ensure future sustainability of forests. Many countries around the world, both developed and de veloping, are involved in various forms of participatory forest conserva tion efforts. Although the defi nition of community forest management can vary widely depending on the location, the international development community has looked to community-based fore st management as a paradigm for forest conservation and rural development in many de veloping nations. Thus, it is important that different approaches under the umbrella of community forest management reflect the particular context within which they opera te by considering the local ecology, resource endowments, socio-cultural systems, and politic al and economic histor y of an area (Schmink, 2004). As such, a brief examination of some of th e different models of participatory forestry being implemented in different countries is useful for comparison with JFM.

PAGE 55

55 CHAPTER 3 QUESTIONNAIRE DEVELOPMENT AND SURVEY ADMINISTRATION Introduction A significant element of this research st udy was the collection of primary data for empirical analyses. Given the lack of comprehe nsive data on the economic aspects of JFM/CFM in Andhra Pradesh, it was necessary to design que stionnaires and administer a survey of VSS villages and households (HH) within them in rural areas. This chapter begins with brief description of the VSS-level que stionnaire, and is followed by a detailed descrip tion of how the HH-level questionnaire was develo ped, structured, and formatted. The following section of this chapter examines some important aspects of the implementation of the survey, including the spatial structure of the survey and field met hods. Finally, the chapter concludes with a brief description of how the data were compiled a nd reviewed for accuracy and completeness. The VSS Questionnaire Analysis of VSS institutions is one of th e main points of this study. Therefore, development of this questionnair e was initially derived from th e design principles identified by Ostrom (1990) as defining the instituti onal foundations of stab le, well-functioning CPR management regimes. The comparative discussion of CPR management institutions, as presented by Agrawal (2001), was also helpful. Moreove r, the APCFM Project Implementation Plan (APFD, undated) was important to the actual understanding of VSS functioning in Andhra Pradesh (i.e., in applied sense), and guided the overall construction of th e questionnaire. Finally, field interviews with VSS Management Committee me mbers in villages of Medak district in late February 2005 were invaluable in ensuring that the questionnaire was grounded in reality. The VSS questionnaire was designed to elic it information about the institutions and functions of each VSS selected for the survey. For example, key institutional questions inquired

PAGE 56

56 about monitoring of their VSS area and about sanc tions imposed for rules violations, while an important question regarding the functioning of the VSS included queries about the forest improvement works undertaken. Other questions aske d about the historical quality and uses of their VSS forest area, number a nd types of APFD training, any a ssistance from NGOs, etc. The VSS questionnaire is 7 pages in le ngth and is written entirely in English (with a few Anglicized Telugu words); it is reprinted in Appendix A. The Household (HH) Questionnaire The LSMS Modules The LSMS modular format was used as the foundation for constructing the HH questionnaire. Based upon actual experiences wi th the LSMS, Grosh and Glewwe (2000) and other researchers provide deta iled advice on how to design multi -topic household surveys for developing country research; in cluded are discussions by modul e (i.e., information category), examples, and even Microsoft Excel templates on an accompanying compact disc. Several independent modules comprise an LSMS survey with each module corresponding to particular topics of interest that a give n research study is either direct ly or indirectly addressing. For example, household consumption is one of th eir main modules; others include income, employment, education, health, etc. The HH questionnaire contains eight modules and is 27 pages in length (24 pages of questions). Table 3-1 presents the contents of th is questionnaire with a brief description of each section. The ordering of the modules reflects not only a logical sequen ce (e.g., household roster near the beginning), but also the relative impor tance of each module in the study. For example, the two most crucial modules of the survey, in terms of the research objectives, are the Consumption and Forest Resources modules. A copy of the HH questionnaire used in this research study is presented in Appendix B.

PAGE 57

57 Table 3-1. Composition of the Household Questionnaire. Module Item/Module Description Pages -Cover Page 1 1 Metadata Location identifiers date; informed consent 1 2 Household Roster Key demographic information 1 3 Consumption Detailed collec tion of expenditure data 8 4 Forest Resources Qualitative / quantitative forest-related data 6 5 Agriculture Farm production, e xpenditure, & livestock data 3 6 Employment Employment type, time employed, income 1 7 HH Enterprise Enterprise management, income, & cost data 3 8 Remittance Remittances to and from the household 1 -Informed Consent (English) Signature page 1 -Informed Consent (Telugu) Telugu copy given to respondent 1 Total 27 Metadata (Module 1) are the key identifyi ng information about how the survey was conducted. This module contains some crucial pieces of information, such as the unique household identification number, the VSS identifi cation number, and the id entification code of the interviewer. Certain qualitative aspects of the interview that are recorded include the name of respondent, the date of the interview, and th e starting and ending time of the interview. According to Grosh and Muoz (2000), there ar e three main reasons why the collection of metadata is important. First, for substantive an alysis: metadata is often crucial for certain purposes, such as the calculation of sample wei ghts. Second, for survey management: metadata helps to assess the time needed to complete implementation, to anticipate replacement households, etc. Third, methodological research can be assisted w ith use of metadata, and this module can even incorporate re search questions to help improve future surveys.

PAGE 58

58 The Household Roster (Module 2) records th e name and sex of each member of the household, as well as the relationship with th e head of the household (e.g., wife, daughter, grandson, etc.). Key demographic data collected here for each individual include their age, education, and time spent living in the household. The latter information is important to determine if the person is counted as an offici al member of the household; the LSMS format guidelines requires nine m onths per year of residence to o fficially be considered a household member. Other associated details that are impor tant are the VSS member ship status of each individual, and their occupation or main daily activity (e.g., school, chores, etc.). The Consumption Module (Module 3) is de signed around items familiar to respondents: flows of goods and cash flows (e.g., taxes or remittances). Deaton and Zaidi (1999) define household consumption as all reported expend itures on individual goods and services, and all non-market consumption (e.g., own production and/or in-kind transfers). Consumption must be comprehensive for accuracy in measuring welfar e; thus, reliance on one or a few items as a proxy is invalid. The level of disaggregation of the consumption list will vary based on the requirements of a particular study. For instance, the comprehensive Indian National Sample Survey (NSS) has used long lists with good results (i.e., little respondent fatigue, good accuracy). On the other hand, LSMS survey lists are typically much shorte r: examples include a Pa kistani survey with 33 food items and 20 non-food items; and a Vietname se survey with 45 food and 46 non-food items. The trade-offs in time efficiency versus accuracy are implied and will be site-specific; research from LSMS surveys have not settl ed the debate as to whether shorter lists provide sufficient accuracy, although some surveys have indicated th is is so. The LSMS draft module, which is typical of past LSMS surveys, included 70 to 100 total items. Th is study incorporated

PAGE 59

59 approximately 50 food items and 35 non-food ite ms, which falls within the recommended guidelines (Deaton and Grosh, 2000). Recall periods are one of the most difficu lt, yet important, design issues of the consumption module. The objective is to obtain a reasonably accurate estimate of the rate of household total consumption expenditure over the pr evious year. The proto-typical LSMS format asks about consumption expenditures for the past year, in addition to the recall period chosen. In general, however, there are different recall pe riods for different items. High-frequency non-food items such as tobacco, newspapers, etc. use recall periods of 1 to 2 weeks. Low-frequency items may have recall periods of 1, 3, 6, or 12 months; or different recall periods for individual items (e.g., soap in a month, vacations in a year). Deaton and Grosh (2000, pp. 112-13) discuss this in greater detail. Basic options for a LSMS survey are single-vis it or multiple-visit; and the choice will help determine the design of the recall period. Typica l LSMS surveys are large undertakings that incorporate multiple visits; thus, the standard LSMS format is to use two recall periodsthe time since the last visit (typical ly two weeks), and the usual month period. According to Deaton and Grosh, including multiple visits is probably not the highest priority for improving the typical LSMS survey. The following guidelines are given: if one is to compare results with other surveys, then use the prev iously established r ecall period; if not, us e the general LSMS design format with modifications if necessary. If using a single visit, the within the past two weeks recall period can be substituted for s ince last-visit recall period of multiple-visit surveys. In general, a singlevisit strategy is acceptable because consumption is smoothed throughout the year.

PAGE 60

60 For logistical reasons, each household sa mpled in this study was only visited once. Therefore, the food purchases component utilized a recall format that allowed the respondent to self-identify the purchase freque ncy of each item, prior to the interviewer inquiring about the purchase quantity and units, a nd the total purchase price. Imputing the values of non-market transactions is difficult in countri es that are not highly monetized (Deaton and Grosh, pp. 116-17). For LSMS surveys, food is the most important imputed item in household budgets, and generally comes from home production or sometimes as gifts. The recall period can be a year or a usual month. The ma in problem is with valuation because respondents are asked to hypothetically a ssign a value to items that are often rarely purchased or traded. The recomm endation is to collect quantitie s of such goods, ask respondents about prices, and cross-check with other data. For the this survey, the HH questionnaire asked the respondent to recall if there was any home production during the previous year for the food goods listed. For positive responses, they were asked the total quant ity and the units of measure; they were also asked to estimate th e total value of this home production. The Forest Resources Module (M odule 4) consists of two parts. Part A contains 29 questions, loosely divided into four parts, which focus mainly on CFM awareness and participation. Because specific information rela ted to VSS details were sought, Part A was only administered to VSS-member households; and only if the survey respondent(s) was an actual member of the VSS. The number of questions were about evenly sp lit between questions requiring yes/no responses, Likert scale or other categorical re sponses, and magnitude responses (e.g., how many hectares of VSS forest area are there?). Part B quantifies the sales value and home consumption value of NTFPs, fuelwood, and sm all timber; the basic format is styled on the part of the LSMS agriculture module that quan tifies crop production and sales. In addition to

PAGE 61

61 fuelwood and small timber, Part B lists approxim ately 30 NTFP items that might be collected by respondent households, and (if any are sold) inquires about wher e and to whom these products were sold. Farm size and income derived from agriculture are the two main pieces of information that the Agriculture Module (Module 5) collects. The Employment Module (Module 6) tabulates income from work performed outside the house hold. The Household Enterprise Module (Module 7) is included mainly to gather information on household income derive d from home production activities. The Remittance Module (Module 8) concludes the HH questionnaire by inquiring about monetary flows into and out of the househ old that involve cash ex changes with household members, usually in the form of gifts to or from relatives (although pension income is occasionally listed here also). Design and Structural Guidelines The ordering of the modules within th e questionnaire will depend upon the size and implementation of the survey. Respondents themselves may i ndicate the best ordering during field testing, especially if return visits are sche duled for later in the day or week. However, all of the interviews are completed in a single trip to a primary sample unit (i.e., VSS villages) in smaller, scaled-down LSMS survey s such as this study. In genera l though, different factors such as best recall period, the natural or logical location (e.g., end of survey for se nsitive information), etc. will help determine where certain data are to be collected in the interview. The essential point is to ensure that data im portant to the study is collected in at least one of the modules. All surveys begin, however, with the metada ta and household roster modules. Following these modules, the primary respondents are usually determined; the other modules are collected as applicable. Education, housing, and migration are good topics to open with, once the metadata

PAGE 62

62 and household roster are obtained. Employment and other sensitive topics (savings, credit, transfers) should be covered at the end of the intervie w (Grosh et al., 2000). The issue of survey length is extremely important: a general goal is to keep actual survey interviews for any particular respondent to no more than an hour per day. This can vary given the tolerance of the people in a given country, and may depe nd on local conditions. LSMS experience shows that tolerance for a long interview is less in urban versus rural areas, lower for wealthy versus poor households and lower in we althier countries. Grosh et al. (2000) offer guidelines that can be used to pare down a survey to the specific needs of a particular research study. For example, many LSMS prototype modules come in both a long and short version, and the particular needs and resources of a given research study will determine which version is preferred. Although choosing the shortest version of modules will allow th e analysis of more objectives, depth should not be sacrificed for t hose objectives deemed most important to the particular study. For instance, the Consumption Module was the most extensive module in terms of length (in time required to complete) and de pth (number of items/que stions)reflecting its central importance to this study. In addition, th e Agriculture Module was based on the short version of the LSMS agriculture module, and it was modified even further to simplify and shorten it. Draft modules for field testing are recommended to help determ ine and judge the trade-offs being made with respect to data collection. Thei r acceptability will either be confirmed, or will suggest that corrections are nece ssary, or that a suitable alterna tive must be found. Draft modules also allow for the review of the coding and nom enclature to ensure consistency throughout the questionnaire, especially w ith regard to similar ques tions (Grosh et al., 2000). A draft questionnaire is also important to recognize gaps and overlaps between modules, and testing

PAGE 63

63 allows them to be modified or corrected as nece ssary. A draft questionnaire also helps to ensure that at least one of the modules is collecting data important to the study. Formatting Guidelines Below are some important issues pertaining to the formatting of selected questionnaire components as related by Grosh et al. (2000) per the LSMS format: Pre-coding and code boxes should be used exte nsively to increase e fficiency and reduce data entry error. Response codes should be located next to the question. A code key can be placed somewhere on the page. Response codes corresponding to answers must be clear: simple, mutually exclusive, and exhaustive. They should be designed so they are not likely to provoke the same response. Lists of items (e.g., food goods in the consum ption module) that re spondents are asked questions about help the efficiency/accuracy of the survey. By first enumerating all items purchased before collecting details on each it em, the temptation for the respondent to not list something is avoided when they realize th ere are several questions about each item. Use uppercase letters for instruct ion to the interviewers, while lowercase letters are for the actual questions asked of the respondents (Fowler, 2002). The questionnaire should be designed so that interviewers always ask verbatim questions to ensure uniformity among interv iewers and between respondents. Two or three simple questions should be as ked instead of one long, complicated question. Qualifiers are important (e.g., What was th e main reason) to help obtain mutually exclusive answers where more than one an swer could apply. When appropriate, the following convention is also us ed: other ______ (specify). Probe questions are common in consumption, agricultural, and similar modules that attempt to get at how much of something. In terviewers need to know what to probe for, and how to do it. This technique should re duce the number of I dont know responses, for which the DK abbreviation should be the proper response code. Code tables for different units of quantity allow the respondent to choose the unit for which they are most comfortable. This will also tend to reflect the unit in which the action/item discussed occurredwhich may differ from household to household. Code tables are of key importance for us e in quantities produced questions.

PAGE 64

64 Survey Structure and Implementation HH Questionnaire Testing and Revision Grosh et al. (2000) emphasize the importance of field testing the survey instrument and list, by component, some of the key considerations that testin g ought to cover. For the overall questionnaire, field tests should en sure that all the necessary info rmation is being collected, and that there is internal consistency to the inst rument without any need less double-counting. Testing will also reveal if individual modules collect th e intended information, cover all major activities, and avoid any redundancy. For individual questions, field tests will help to know if the wording is clear, if the coding works well or not, and if there are any ambiguous responses due to either. In addition, Fowler (2002) states that questionnaire testing also helps to determine how long it takes to complete a survey instrument. For the present study, the HH quest ionnaire was tested in the Me dak district of Telangana region in late February 2005. The main finding of this test was that th e questionnaire was too long, and redactions were necessary because respondents were showing signs of interview fatigue. Changes included shortening and si mplifying the Consumption Module. For example, the extensive list of food items in Part B wa s reduced by eliminating less common items. In conjunction with this, more use was made of blank spaces for other items not specified. Further reductions were made by consolidating se parate items into like categories in Part C (Non-Food Goods). A more significant change was to revise Part B to allow the respondent to self-identify the purchase frequency of each item. The initial de sign of the Consumption Module relied on the LSMS frequency format: a 7-day recall period fo r daily expenditures (e.g., tobacco); both a 2week and a typical-month recall period for food goods; and both a 30-day and a 1-year, recall period for non-food goods. Changing this avoided the time consuming (and potentially error-

PAGE 65

65 prone) process of inquiring about purchases in the previous two weeksfollowed by asking about the purchases in a typical month. For exam ple, the question became: How often does your family purchase []? with given responses pre-c oded (e.g., weekly = 1, fortnightly = 2, etc.). Changing to this configuration for food items (Par t B) also allowed greater flexibility for the respondent, by allowing them to bett er answer the query in a format closer to how they mentally relate their actual usag e of a given item. Further technical efficiency gained could only have come from extensive field trials that were not logistically possible at the time. Therefore, a second round of formal field testing was foregone, due to time constraints. This mainly had to do with the fact that the survey needed to be completed by June, when the monsoon season begins. Moreover, interviewers had yet to be hired and 1,200 questionnaires had to be printed. In addition, as se veral different individuals in India contributed to the development of the hous ehold questionnaire, it was decided to proceed with the instrument as revised. Sample Selection This research study was designed to obtain a sample of VSS vill ages from each of the three regions of Andhra Pradesh. Thus, the field surv ey is based on the random selection of 20 VSS villages from a district chosen to represent each one of these regions. Selection of villages was made using the random number generator in Micr osoft Excel. The total number of 60 VSS was arrived at in order to ensure sufficient degrees of freedom when conducting the subsequent econometric analyses. For the household survey, each of the 60 VSS villages had 20 households randomly selected for interviews. Thus, the total possible number of observations is 1,186 households (which is less than 1,200 because th ree VSS villages had less than 20 households). Random selection of households was made usi ng either Microsoft Excel or a random number sheet in the field.

PAGE 66

66 The APFD apportions the state into admini strative units (or so -called Circles); occasionally these circles conform to the district s observed in Figure 2-2, but in general they are aggregated somewhat into larger units. Each circle is comprised of smaller divisions: the Visakhapatnam Circle contains fi ve divisions, for example. As such, the following APFD forest divisions were selected for sampling from within each of three circles: 1) Nirmal and Jannaram Divisions (Adilabad Circle); 2) Visakhapatnam Division (Vis akhapatnam Circle); and 3) Chittoor West Division (Anantapur Circle). Random probability sampling was employed to select 20 VSS villages within each circle. The tw o divisions in Adilabad had their sample split based upon the relative number of VSS villages present in each division. Thus, 14 VSS were sampled in the Nirmal Division and 6 VSS were sampled in the Jannaram Division. Within each sampled VSS village, 20 households were also selected using random probability sampling techniques in order to ensure statistical validity when drawing inferences from the results. It is important to point out, however, that the selection of the study areas (circles) within each region (i.e., Telangana, Rayalaseema, a nd Coastal Andhra) was not random. The Adilabad Circle was ostensibly chosen by the APFD on the basis of security concerns, and the Nirmal and Jannaram divisions of this circle were selected to represent two different forest regimes within the Telangana region. The forested area of the Ni rmal Division is dominated by teak, while the Jannaram Division has more bamboo present fo r economic and consumptive activities. The Chittoor West Division of Anantapur Circle was also chosen by the APFD for inclusion in the surveyas was the Visakhapatnam Circle, although the principal investigat or (PI) had originally selected this area to represent the Coastal Andhra region. (The re presentative division of this circle that was actually sampled [i.e., the Visakhapatnam Division] was randomly selected, however.) The implication of the selection process is that each regional part of the overall survey

PAGE 67

67 is essentially representative only of that speci fic location within Andhra Pradesh, and aggregation of all three parts re duces the validity of inferring re sults to the state as a whole. Nevertheless, each village within a given region was still selected at random, and the evaluation of the specific hypotheses of the study takes precedence over the ability to a pply the results on a wider scale. For comparison, the Centre for Economic and Social Studies (CESS) in Hyderabad conducted a survey of Andhra Pradesh in 2004 to investigate the impact of JFM/CFM on livelihoods (Gopinath Reddy, personal communicatio n). Much like the present research, the objectives of the CESS study included examinati on of the institutional and economic dynamics of CFM at the micro level (Reddy et al., 2004). A lthough much smaller, the CESS survey is also structurally very similar to the present studysel ection of one district from each of the three regions of Andhra Pradesh (Adilabad, Visakhapa tnam, Kadapa) with six villages from each district (3 VSS villages and 3 non-VSS villages) At the final sampling stage, 25 households were interviewed in the VSS villages (225 total), while 15 households were interviewed in the non-VSS villages (135 total). Reddy and Chakravarty (1999) used a much smaller survey to investigate forest dependence and income distribution in villages of northern India. Their survey was based upon four development blocks (out of 15 total blocks in a single district) selected because they were contiguous to forest areas. Twelve villages we re randomly selected from the four blocks, followed by individual households serving as the final sampling unit. Households were selected by simple random sampling with replacement, and the total number of usable household questionnaires equaled 233.

PAGE 68

68 Interviewer Training and Impl ementation of the HH Survey During the first week of March 2005, four enumerators were hired in Hyderabad to conduct the household interviews. Staff members of CESS were instrumental in identifying qualified interviewers. Three men that were hire d each held masters degrees and had extensive experience with conducting rural household interviews, for bo th CESS and other rural welfare agencies. In particular, each had done similar work in the district of Ad ilabad, where the first leg of the survey was to commence. Although the four th man had limited survey experience, he had served as an agricultural extension agen t in his home district of Adilabad. Training took place in the town of Nirmal, wh ere the purpose of the survey and the format of the questionnaire were explained in more de tail. In order to become familiar with the instrument and procedures, mock interviews were conducted where each member interviewed anothertaking turns as both interviewer and re spondent. This follows one recommendation of Fowler (2002), who also emphasi zes that interviewer training s hould cover from two to five days. Following this practice session, Vishnu Reddy (the research associate of the PI) had an extensive discussion, in Telugu, w ith the interview t eam about their questions, problems, and other concerns. The PI was also there to answ er questions and to co mment about procedures. Subsequently the next phase of training consisted of actual data collection in the fieldan extra village had previously been selected for this purpose, so each person interviewed a total of 20 households. In addition, the first two VSS village s surveyed are dropped from the analyses and effectively become additional training villages. Therefore, the total number of observations for analysis is 58 VSS villages. The second round of sampling took place in the Visakhapatnam Division of the Visakhapatnam Circle in early April 2005. It was determined that two teams were required to reduce the amount of time needed to complete the research effort. Thus, prior to the beginning of

PAGE 69

69 the second round of sampling, an additional four e numerators were hired. These men were also college graduates and most possessed previous su rvey experience as well. The pre-field training followed the procedures discussed above, with the original sample team assisting in the mock interviews and post-exercise discussion. For this leg of the survey, however, the field training of the new enumerators was increased and consisted of household interviews in two villages (for a total of 40 households per interviewer). Wh en the actual sampling commenced, each new interviewer was paired with a member of the original sampling team. The overall survey proper officially be gan on March 13, 2005; data collection was completed in the Adilabad Circle on April, 1 2005. Following the hiring and training of additional personnel, fieldwork resumed on April 10, 2005; data collection was completed in the Visakhapatnam Circle on April 19, 2005. Fieldw ork for the final round of sampling commenced on May 4, 2005, in the Chittoor West Division of th e Anantapur Circle. Like the previous round, the household surveys were again conducted by tw o interview teams. However, one team was now comprised of three enumerators, due to an unrelated injury su stained by one of the interviewers during the hiatus between the Visakhapatnam portion of the survey and the resumption of fieldwork in the Chittoor Distri ct. Data collection fiel dwork for the study was completed on May 16, 2005. Implementation of the VSS Survey Prior to visiting villages sele cted for sampling, attempts were made to obtain secondary information from local APFD offices, local Velu gu (a statewide rural poverty reduction project) offices or, on occasion, local magistrates. APFD di vision and field offices were visited to obtain VSS micro-plans for relevant information (e.g., V SS size or date of VSS inception), to observe topographic maps, and/or to gain useful anecdot al information. Local Velugu offices were often able to provide access to a lis t of households, and/or hand-drawn social maps of dwelling

PAGE 70

70 units. This information was usually photocopied, or occasionally borrowed. If unable to do either of these, then this information was used to ra ndomly select households at that moment in the Velugu office by recording the names of the househol d heads into a laptop or a notebook, for use in implementing the random selection techniques previously mentioned. Obtaining secondary information from these s ources (especially from Velugu) was often a time consuming and frustrating job, but prove d extremely beneficial when successful. For instance, the household lists offered an objectiv e estimate of the number of households for a given village, and meant that the random selection of households could be made prior to the field visit. This was the main benefit, as it allowed the interview team to avoid this process in the field, saving time and minimizing potential errors: they simply ha d to contact the pre-selected households. Once the secondary information was in hand, the PI and his research associate would visit each village in order to conduct th e VSS interview. The first task was to locate either the VSS Chairman and/or Vice-Chairman for introductions and to give an explanation of the research study. For cultural and practical reasons the PI a nd his research associat e usually interviewed whichever person was male. If the VSS Chairman was a woman, she may not feel comfortable being interviewed: in such cases the interview would be conducted with the Vice-Chairman, with or without the presence of the Chairman.5 In one village, the husband of the VSS Chairman was the former Chairman himself. Thus, this man was interviewed as he was obviously the de facto VSS Chairman of his village. Often during th e interviews, a few other members of the 5 When the women did participate in the interview, whet her they were the Chairman or the Vice-Chairman, they almost always provided little input. Often it seemed th at the women who were VSS Chairman or Vice-Chairman were mere figure-heads serving only to comply with the CFM institutions.

PAGE 71

71 community (whether or not they were VSS Ma nagement Committee members) would usually be present and offer input to the discussion. The interview was quite informal in terms of adhering to a rigid survey format of verbatim questions. The respondent (s) would be engaged in a qua si-structured conversation (in Telugu) based on the VSS questionnaire, which a llowed for a certain amount of flexibility in terms of question order and overall flow of the interview.6 Following the interview, the researchers would ask to take a walk through the VSS forest area (schedule permitting). Often the conversations would continue during the inspection of the prot ected VSS forest areas and/or plantation areas, as the villagers were usually enthusiastic to s how off their forest area and the works they completed. Of the 62 VSS villages surveyed (including training villages), the researchers were able to inspect at least two-th irds of the VSS forests and/or plantations. The visits to each village were intended to be unannounced in order to obtain an independent and unbiased assessment of each V SSthe goal being to conduct the interviews without any positive or negative bias that could po tentially result from the outside influence of the APFD. Occasionally, field offi cers of the APFD would alrea dy be present upon arrival in a village; this was because they knew of the research effort when VSS micro-plans had been borrowed. In such cases, the PI would ask the fore st officers to show him the VSS forest area in order to draw them away from the on-going interv iew: for this reason the PI was not present for approximately five or six of the interviews proper. In addition, for visits to the VSS villages of the sample, the PI declined transportation provi ded by the APFD so as to similarly avoid any direct or official associati on with the government. This is an important consideration for 6 Although the PI could not understand the words, he could follow along with the general mood and tenor of the interview. Upon request, important passages were translated and the PI would pose additional questions to expand on the explanations.

PAGE 72

72 obtaining unbiased interviews in the rural villages of Andhra Pradesh, according to CESS Research Fellow Gopinath Reddy (p ersonal communication). Therefor e, the PI generally rented a vehicle or an auto-rickshaw, rode th e bus if the location was close enough. Before leaving a village, the PI (through his research associate) w ould inform the people interviewed that a sample team would be comi ng within a few days to conduct the household interviews. The research associate would ask one of the villagers to assist the sample team in locating the households that were selected fo r interviews. Sometimes a sample team would accompany the PI to a village (usually in remote areas, or at the beginning or end of a sample leg) and administer the HH que stionnaire while the VSS interview was being conducted; but most of the time the sample team(s) worked inde pendently of the PI and his research associate. For this reason, the PI made every attempt to secure a list of households in advance, but in the event that they were unobtainabl e it was necessary to do the ra ndom selection of households in the field. The desire of the PI to avoid this s ituation by using pre-selec ting households was based on two observable facts: 1) many villages lack an or derly layout, and 2) it is often difficult to distinguish distinct dw elling units amongst a coll ection of structures. Thus, the likelihood of deviating from a completely random sample increases in such a scenario, especially the larger the village and the le ss orderly its lay-out. In the first few days of the survey, the PI wa s afforded the opportunity to demonstrate to the interview team how to randomly select house holds without a list. For example, in the VSS village of Rampur (Adilabad), the 340 households were divided by 20 (i.e., the intended sample) to calculate a selection interv al of 17. For each interviewer, the PI selected a number ranging from 1 to 17 that was taken from the random numb er sheet. The village was divided into four sections and each enumerator was to intervie w every seventeenth household, beginning with

PAGE 73

73 whichever household corresponded to the random number that was selected for them. This procedure was subsequently followed (to the be st of the PIs knowledge) by the sample team leader when the PI was not present for the situ ations where households were randomly selected in the field. Each household interview generally took between 45 and 60 minutes to complete, depending upon the size of the household, their le vel of consumption, and sources of income. Each team would usually complete one village pe r day (i.e., five households per interviewer), although at times it was feasible to finish two villages if they were located near each other. Noncooperation (respondent refusal) was not a problem whatsoever, although occasionally a household would be unavailable (nobody home), said to have migrated, or non-existent for some reason. Such households were ini tially replaced by interviewi ng the next household, but later alternate random selections were made for this scenario. Data Compilation Following completion of the interviews, a template file was created using Microsoft Excel that essentially mirrored the HH questionnaire. This allowed data from each household to be easily transcribed into this format for later co mpilation into master files for each village, and eventually to an overall aggregate file. Each house hold file contains intern al calculations that compute, for example, total household consump tion expenditures and total household income. In addition, there is summary page where important data for the household are compiled into a single row of cells, in order that this line of data can easily be c opied and transferred to a master file for the corresponding village. For the job of transcribing the data, the PI re tained two of the interviewers that were available and also hired a woman specifically to do this work. These three people did the majority of the transcription, although others (including the PI) occasionally assisted also.

PAGE 74

74 Initially, the PI was able to supervise the transc riptions for quality assurance (QA) by instructing the data entry personnel on how to handle specific ambiguities in questionnaire responses and/or information as recoded in the field. By checking completed files to identify and correct entry errors, instructive feedback to the transcripti on crew was also provided. After the PI departed from India, the research associate continued the supervision activities. Prior to sending the completed files to the PI, the re search associate conduc ted initial quality co ntrol (QC) checks of completed household files (approximately 5 per village). The PI completed the final QC check for a ll household data files. When assembling the village master files, it was necessary to open each file to copy and transfer the summary line of data mentioned above. Thus, this opportunity wa s used to compare the HH questionnaire with its associated data file to ensure correct transcription and to rectify errors, which was a very time consuming process. However, it was necessary to perform this function because of persisting ambiguities with certain sections of the HH questionnaire with which the transcription team had problems. For example, NTFP consumption in Part B of the Forest Resources Module is prone to errors, in terms of the frequency and amounts coll ected, that need to be corrected. The confusion stems from the rigidity of the format: res pondents were asked about their consumption on a monthly basis only. As most items collected from the forest ar e done so seasonally, or on an otherwise irregular basis, the self-identification of frequenc y (adopted for the Consumption Module following pre-testing) shoul d have been the standard here also. Thus, it is often difficult to discern the total yearly consumption amount s from the recorded answers; interviewer heterogeneity could also compound the problem. Other QC activities were also conducted by the PI As they were basically transferring data verbatim, the transcription team was necessarily gi ven little responsibility to interpret the data.

PAGE 75

75 Thus, QC consisted of checking for the correct placement of information (e.g., income from a home enterprise in the HH Enterprise m odule instead of the Employment Module), and monitoring the internal consistency of calculatio ns. The correction of all types of errors was made as objectively as possible in an effort to clean the data for subsequent analyses.

PAGE 76

76 CHAPTER 4 MODELS, HYPOTHESES, AND DATA Overview The analytical focus of this study is con cerned with factors influencing the economic impacts of participatory forest management in Andhra Pradesh, India (i.e., JFM/CFM). This study will capture such influences to the extent that the three measures representing different aspects of social welfare vary among the cross-se ction of VSS villages that comprise the sample. The three indicators include the mean per-capita household consump tion value of a given village, the inequality in consumption among households in the village, and the level of poverty within the village. The first two are used to define soci al welfare within an economic context, while the third can be considered an altern ative measure of social welfare since it focuses on the portion of a village (i.e., individuals in households) below a specified poverty leve l. Evaluating whether, and to what extent, various explanatory variab les affect each individua l economic indicator will help to empirically assess if there ha s been any economic impact of JFM/CFM. To accomplish this task, each of the economic indicators will be used as dependent variables in separate models that will largely rely on a common set of explanatory variables. The explanatory variables used in the econometric models are summarized in Table 4-1. The variables are categorized into the following four groups: Demographic, Economic, Bio-physical, and Institutional. There are five to six variables in each group and 21 in total. The following section begins with a brief discussion of the selection of the basic welfare measure and its application. Social welfare is defi ned in an economic context and then each model is described in more detail, including the implicit assumptions be hind each of the indicators chosen to represent these aspects of social welfare. Next, a secti on discussing the hypotheses of each variable shown in Table 4-1 is presented. Lastly, a detailed de scription of the data concludes this chapter.

PAGE 77

77 Table 4-1. Explanatory variables utilized in the VSS-level regressions. Variable Variable Description Unit Type Demographic: N Number of households (HH) in a VSS village Number Integer ST VSS predominately comprised of Scheduled Tribes (1 if yes) 0, 1 Dummy EDU Mean HH Education level Years Continuous NGO Number of NGOs working in the community Number Integer DGI Index: number of DWCRA Groups divided by N Percent Ratio Economic: YFP Mean HH Income from Forest Products Rupees Continuous YAL Mean HH Income from Agriculture and Livestock Rupees Continuous YOS Mean HH Income, Other Sources (employmen t, enterprises, etc.) Rupees Continuous IHW External Investment per HH (avg. wages for VSS works/HH) Rupees Continuous TRN Number of Training events/field tr ips by VSS MC members Events Number Bio-physical: LT Length of Time under JFM/CFM Years Integer LT2 Squared value of LT Years Integer FA VSS Forest Area size in hect ares (ha) Hectares Integer RRE Relative Resource Endowment (VSS forest area per VSS member) ha/mem. Continuous PFC Percent Forest Product Consumption Percent Ratio DTW Depth to groundwater (proxy for spatia l environ. heterogeneities) Feet Continuous Institutional: VBA Percent of VSS HHs with general Boundary Awareness Percent Ratio GAI Mean HH General Awareness of Institutions 0 to 7 Continuous CCA Mean HH proxy for Collective-Choice Arrangements Percent Ratio FP Presence of Formal Patrol of VSS forest area (1 if yes) 0, 1 Dummy GS Graduated Sanctions for rules violations (1 if yes) 0, 1 Dummy Empirical Models The empirical modeling of this study necessarily begins with the cons truction of the three economic indicator variables. First, a suitable measure to represent th e living standards of individuals or households must be selected. The standard choi ces to measure economic wellbeing are either income or consumption. Deaton (1992, 1997) and Deaton and Grosh (2000) explain that while income is a superior welf are proxy for households in developed countries,

PAGE 78

78 consumption serves as a better proxy for social we lfare in less developed countries. Income data are generally inferior to cons umption data for the measurement of living standards of rural people: data on consumption over a period shorter than a year gi ves a more accurate estimate than income. This is because consumption is aff ected less by seasonality than is income, which can be highly seasonal (especially in agriculture). Moreover, it is more cost-effective to gather consumption data and it is more reliable, as peop le are more likely to inaccurately report income and assets (for instance to lower tax liabilities) The Indian National Sample Survey (NSS) also focuses on consumption rather than income. For these reasons, consumption is used to measure social welfare in this study. Household welfare is representative of utility as attained by the consumption expenditures of the household budget constraint. Therefore, the main purpose of the HH questionnaire described in Chapter 3 was to collect household-level consumpti on data so that the economic measures of social welfare could be construc ted for each VSS village. Deaton (1997) emphasizes using individuals as the basis for a welfare measur e because it is hard to think of households as repositories for well-being (p. 150). For exampl e, he discusses how to transform household consumption data into individual welfare measur es, including the complexities and practical constraints involved in assigning di fferent consumption values to different members of the same household. In the end, however, Deaton (1997) r ecommends a simpler method which he deems to be the best practice: assigni ng the per-capita household consum ption value to each individual in a given household. This is the procedure foll owed in this study, wher e the total consumption value of the household ( tcv )7 is divided by the total num ber of household members ( hhs ) to obtain x the per-capita househol d consumption value. 7 The household total consumption value equals all consumption expenditures, plus the imputed consumption of food items produced or collected, and the own consumption value of goods produced by a household enterprise.

PAGE 79

79 Social Welfare Function As explained by Deaton (1997), a measure of so cial welfare (W) transforms individual or household consumption data of a po pulation into a single summary va lue that is useful for policy analysis. The general form of W that Deaton (199 7) presents is decomposable into two parts (Equation 4-1), so that the social welfare of a given population is represented by the average level of consumption ( ) and its distribution ( I ): W = ( 1 I ) (4-1) Equation 4-1 implies that if each household in the population has the mean level of welfare (i.e., the case of perf ect equality), then I = 0 and social welfare it self is equivalent to .8 This expresses a societal preference that more equal di stributions of social welfare, for a given level of are superior to less equal dist ributions. Thus, any deviation fr om a totally equal distribution of welfare will necessarily result in social welfare (W) being less than the mean value ( ). In other words, is the highest level of welfare that is attainable, ceteris paribus This identifies as a purely economic indicator of social welfar e that remains uncorrected for distributional inequality. An important aspect of Equation 4-1 is the explicit illustration that inequality is not synonymous with social welfare. Indeed, it is possible for W to increase while the inequality measure is also increasing, but only if average consumption ( ) increases enough to offset the decrease caused by I (i.e., the rich gain more than the poor, although everyone gained). Such a situation is still a Pareto improvement from the initial scenario. Neverthe less, given that social welfare consists of these two elements, it is necessary to decompose W into two separate 8 Note that because perfect equality does not exist, empirically I will always be non-zero so that W = is merely a theoretical point of abstraction.

PAGE 80

80 variables in order to analyze th em individually. Thus, the presen t study utilizes each component of W as a dependent variable for regression analysis, where is measured as the mean per-capita household consumption value fo r a given VSS village; and I is measured by the Gini coefficient ( ), which is typically used to represent inequality. Average Consumption Model The first measure, which is the focus of this subsection, is It is calculated for each VSS village as the mean per-capita value of household consumption. As data on consumption were collected at the household level, the per-capita household consumption value x was first applied to each individual in a household following Deat on (1997) as discussed previously. In order to derive as based on individual consumption, Equation 4-2 below was used: = J j j J j j ihhs hhs x ) ( ) ( ( j = 1, J ) (4-2) where xi denotes that the per-capita household consumption value x is applied to an individual i ; hhs is the household size (i.e., number of household members) of the jth household sampled in the VSS village; and J is the total number of households sampled in the VSS village. By using random sampling techniques to select the households in each village, Equation 4-2 is used to calculate an estimate of for each of the 58 villages surveyed. The model used to estimate is specified as: = Xe (4-3) and, by taking the natural log of Equation 4-3 is transfor med into a semi-log model: ln = X + (4-4)

PAGE 81

81 that can be estimated using the ordinary leas t squares regression procedure. Equation 4-4 is comprised of the following components: X is a matrix of i ndependent variables, is a vector of coefficients associated with X, and is the disturbance term. The matrix X is composed of the following demographic, economic, bio-physical, and institutional va riables listed in Table 4-1. Demographic: N, ST, EDU, NGO, DGI Economic: YFP, YAL, YOS, IHW, TRN Bio-physical: LT, LT2, FA, PFC, DTW Institutional: GAI, CCA Two location dummies (ADIL and CHIT) are also included in the matrix X in order to control for the different spatial regions of Andhra Prad esh from which the data were collected. Consumption Inequality Model As described in the previous section, an important aspect of Equation 4-1 is that inequality I is a separate component of W, and is therefor e not valid as an independent measure of social welfare. As a dependent variable, I evaluates the degree to which welfare is distributed in an inequitable manner throughout a given community. As mentioned, the typical measure used to represent inequality is the Gini coefficient ( ). This is a ratio measure most commonly associated with the Lorenz curve, which is a graphical depi ction of percentage soci al welfare distribution (e.g., by income or consumption) in terms of population quintiles. Following Deaton (1997), the equation below is used to calculate an estimat e of the Gini coefficient for each VSS village: = 11 2 1 1i i ix M M M M (4-5) where M is the total number of individuals i that belong to househol ds sampled within a VSS village, and is the relative rank of e ach individual, starting with = 1 for the richest person

PAGE 82

82 sampled, and ending with the poorest person ( = M ) sampled.9 As defined previously for Equation 4-2, xi is the per-capita household consumpti on value (which is identical for all members of the same household) while is the mean of all xi (i.e., mean per-capita household consumption value). The expression i xi is summed over all persons sampled in a given VSS village. The use of random sampling techniques to select the households in each village allowed the calculation of an estimate of for each of the 58 villages surveyed. According to Vanhoudt (1998), a measure of distributional in equality can be econometrically modeled as a function of nonconventional factors (e.g., institutional parameters) in addition to typi cal neo-classical factors such as labor and both physical and human capital. Odedokun and Round (2001) discuss how recent studies of inequality have investigated a broad range of factors affecti ng inequality, specifically mentioning institutional factors. Regressing on a collection of explanatory variab les is undertaken to evaluate the contribution of these variables to inequality; the mode l used to estimate inequality is specified as: = u Ze (4-6) which is transformed into a semi-log model specification: ln = Z + u (4-7) where Z is a matrix of independent variables, is the vector of coefficients associated with Z and u is the disturbance term. The matrix Z is composed of the following demographic, economic, bio-physical, and institutiona l variables listed in Table 4-1; Z also contains the two location dummies introduced earlier: 9 For instance, if the household with the highest per-capita consumption value x has four members, then those four individuals would have values of 1, 2, 3, and 4. The Gini coefficient is an example of where it is necessary to use individual consumption values, following the procedure that Deaton (1997) recommends (as just described), in order to calculate a socio-economic indicator that is representative of inequality in a given community.

PAGE 83

83 Demographic: N, ST, EDU, DGI Economic: YFP, YAL, YOS, IHW, TRN Bio-physical: LT, LT2, FA, PFC, DTW Institutional: CCA, FP Location dummy: ADIL, CHIT Consumption-based Poverty Model As mentioned previously, measur es of poverty can be thought of as a special class of social welfare measures that specifica lly address the proportion of the population below some given poverty line. To the extent that there is a significant differe nce between poverty and social welfare in the villages sampled (i.e., poverty is not a uniform condi tion), then it would be useful to estimate a model that tries to explain a ny observed variations in poverty across villages. Deaton (1997) discusses the de rivation and characteristics of several different poverty measures. One of these measures is the headco unt ratio, which is simply the number of people below a predetermined poverty line (e.g., a threshol d level of consumption). The headcount ratio is a poor social welfare metric, however, becaus e it has the disadvantage of violating the socalled principle of transfers (i .e., a transfer from a poor person to one who is less poor could conceivably lift the latter above the poverty line). The poverty-ga p ratio (PGR) rectifies this deficiency by calculating the difference in an individuals welfare (w hich can be based upon either a measure of consumption or income) with the given poverty line, and normalizing to this line. Summing this over all individuals belo w the poverty line, and dividing by the total population, results in a ratio th at (when multiplied by ) makes a suitable measure of social welfare. The PGR is a poverty measure that is commonly used by developm ent researchers (e.g., Reddy and Chakravarty, 1999) and is the measure that will be used in this study. Following Deaton (1997), the poverty-gap ratio is esti mated for each VSS village by calculating:

PAGE 84

84 PGR = i iz x M 1 1 for all xi z ; or (4-8a) PGR = 0 if all xi z (4-8b) where z is the predetermined poverty threshold. It should be noted that considerable debate surrounds the construction and use of poverty th resholds and researchers generally rely on existing estimates (Deaton, 1997); this study uti lizes a consumption-ba sed poverty line already established for rural Andhra Pradesh. Ceteris paribus the PGR will increase the greater the difference between xi and z or if more individuals fall below z Thus, as an economic measure of welfare for the population below th e poverty line, higher PGR values indicate a greater level of inequality in the distribution of c onsumption within a VSS village. Allanson and Hubbard (1998) discuss how to empirically estimate the income-gap ratio from a random sample of different income classe s, and clearly this tec hnique also applies to consumption data. Additionally, th ey relate the income-gap ratio to second-degree stochastic dominance, a concept whereby integration of cu mulative distribution functions allows for the ranking of welfare distributions. Moreover, second -degree stochastic dominance is equivalent to generalized Lorenz dominance (D eaton, 1997), which itself permits the ranking of standard Lorenz curves by scaling them up using the mean of the distribution. Thus, randomly sampled household consumption data can be aggregated into a PGR value (estimated for each VSS village), which serves as the depende nt variable in the equation below: PGR = Y + (4-9) where Y is a matrix of independent variables, is the vector of coefficients associated with Y and is the disturbance term. The matrix Y is composed of the following variables: Demographic: ST, EDU, NGO, DGI

PAGE 85

85 Economic: YFP, YAL, YOS, IHW Bio-physical: LT, LT2, RRE, PFC, DTW Institutional: PBA, CCA, FP, GS Location dummy: ADIL, CHIT Hypotheses for the Explanatory Variables Demographic Variables For the empirical analysis, N is the tota l number of households belonging to the VSS village regardless of whether th e household is a VSS member or a non-member. This variable is expected to be positive with respect to meaning that economic well-being ought to be greater the larger the VSS. As such, an inverse relati onship with PGR is expected, as small communities are more likely to be poor.10 This is because small communities were generally observed to be more isolated and remote, and with employment opportunities that were generally more limited as compared to larger villages. The relationship with is also anticipated to be positive, which would reflect greater in equality due to a larger, and pres umably more socially diverse, community. ST is a binary dummy variable equal to 1 if a given VSS village consists entirely or predominately (as based on the sample mode) of households of the Scheduled Tribes castedesignation. As one of the main stakeholder groups identified in the JFM legislative orders, it is important to control for the Scheduled Tribes VSS villages in the sample. A negative relationship with may be predicted because Scheduled Tribes villages are often more remote than other villages and thus less-integrated with the la rger local economy. (A positive relationship with PGR might be predicted for the same r eason). The relationship between ST and ought to be 10 For variables that are included in both the ln equation and the PGR equations, the relationships with the dependent variables will usually be opposite of each other as the prediction for N illustrates.

PAGE 86

86 negative, as Scheduled Tribes communities are usually very homogeneous both ethnically and economically. However, as Scheduled Tribes have generally benefited from other tribal development projects sponsored by the government or external donors, it is possible that the opposite relationships (i.e., ST having a positive affect on and a negative effect on PGR and ) will be observed (Janaki Alavalapati, personal communication). EDU represents the educational level of a gi ven VSS. It is com puted by calculating the weighted mean11 household education level for each VSS v illage; where the education level of a household is the sum of school years complete d, divided by the total number of household members greater than three years of age. A positive relationship with is anticipated, and an inverse relationship with PGR is expected. The relationship with is likely to be positive, meaning that more education is associ ated with greater inequality. NGO is a binary dummy that equals 1 if a non-governmental organization has provided forestry-related assistance to a given VSS. Misra and Kant (2004) point out that the particular focus of a given NGO must be taken into consid eration; for example, the way in which an NGO affects a given village will depend upon whether th eir orientation is more towards conservation than economic development. The NGO variable as defined is, thus, more conservation oriented; therefore, one might expect a negative relationship with according to Misra and Kant (2004). However, it is possible that NGO c ould have a positive relationship on through improvements in human and social capital. For these reasons a negative relationship with PGR is posited. The recent CFM legislation spec ifically intends to empower women within the political economy of the VSS. In addition, the World Ba nk (2002) places much emphasis upon the critical 11 VSS-level data derived from the household interviews is we ighted to take into account the differential probability of selection of households for diff erent VSS villages (i.e., large versus small villages in terms of N).

PAGE 87

87 importance that womens representation has in th e VSS. The extent to which women participate in (and influence) the collective action decisi on-making is extremely important because women likely know more about forest produce than men in many communities. The DWCRA group index variable (DGI) is the number of womens self-help groups (DWCRA) divided by N, the total number of households in the VSS. DGI repr esents the degree of se lf-reliance and selforganization of women, and proxi es for the general empowerment of women in a given VSS. Empirical evidence from Misra and Kant (2004) i ndicate that there is a positive relationship between womens participation and the economic output of JFM. Thus, the present study anticipates a positive relationship between DGI and meaning a greater amount of selforganized women have a benefi cial influence on average cons umption (i.e., mean per-capita household consumption). In addition, a larger DGI is likely to be inversel y related to poverty (as measured by the PGR variable). A negative relationship with is anticipated, suggesting that a larger DGI results in lower levels of consumption inequality. Economic Variables Annual income is measured with three variables in this study. YFP is the weighted mean household income derived from all forest pr oducts, YAL is the weighted mean household income from agriculture and livestock, and YOS is the weighted mean household income derived from all other sources. YOS is mainly compri sed of income from employment but it also includes (to a much lesser degree) income from household enterprises and remittances. Considering these measures independently differe ntiates low-income households from those that are relatively more affluent, and also categorizes income by its source; this is important because evidence suggests that the poores t households in rural commun ities are the most reliant upon forests for their livelihoods, and for supplementing their household consumption (Kumar, 2002).

PAGE 88

88 As such, YFP is anticipated to be inversely related to while YAL and YOS are predicted to be positively related to Financial support for the APCFM project is de rived from the World Bank, and funds are distributed to the VSS by the APFD. IHW is an important variable because it proxies outside investment to the community in the form of the household average of total payments to individuals participating in V SS works projects, such as silvicultural treatments. IHW is calculated as the man-days of labor employed fo r VSS works in the three previous years as estimated by the VSS Chairman, divided by N (tot al number of households). This value is then multiplied by the wage rate paid by the VSS to th e laborers. IHW is expected to have a positive effect on ; a negative effect on both and PGR is anticipated because the VSS works are more likely conducted by the poorer households. TRN is a count of the number of training even ts and field trips that VSS members (mainly Management Committee members) have participated in through the auspices of the APFD. It is a proxy for the amount of human capit al investments made to a given VSS. This measure was also estimated by the VSS Chairman. Anticipated re lationships for TRN are a positive sign for the average consumption and poverty-gap equations and a negative sign for the consumption inequality equation. Bio-physical Variables The length of time (LT) that a community has pa rticipated in JFM as a registered VSS is extremely important because it is the key variable in terms of evaluating the success, or lack thereof, of the economic impact of JFM. This is because the NTFP benefits of JFM are expected to increase over time from an in itial state of low returns characteristic of the degraded nature of the landscape being protected; hi gh values of LT theoretically imply income is being derived

PAGE 89

89 from timber harvests. Therefore, in relative terms among the cross-section of VSS sampled, a positive relationship between LT and (and/or a negative relati onship between LT and PGR) would likely indicate that JFM has had a be neficial economic impact on the communities engaged in this program. Misr a and Kant (2004) found empirical evidence to support such a relationship. However, a negative sign on LT (vis--vis ) could be indicative of weakened social cohesion through time. Kumar (2002) relates that some forester s fear that interest in JFM may wane when outside support inevitably declin es, affecting project sustainabilitythough this is unlikely to be important here, as funding unde r the APCFM project was still active at the time of the survey. Community malaise may have more to do with local factors, poor or inappropriate management plans, and/or poor engagement by the certain field officers of the forest department. A negative sign on LT for would support an interpretation that JFM has a beneficial impact on distributional equity, again relative to the cross-section of VSS villages sampled. FA is the size of VSS forest area under CFM protection. Forest s are potentially an important form of natural capital in rural areas, and Misra and Kant (2004) found that total forest area has a direct and positive effect on the economi c output of JFM. As such, FA is expected to have a positive statistical influence on the ec onomic measure of welfare in this study (i.e., average consumption, ), and the anticipated relationship betw een FA and inequality is negative. RRE represents the relative forest resour ce endowment managed under CFM, for the VSS sampled in the survey. RRE is calculated as th e VSS forest area (in he ctares) divided by the estimated total number of VSS members for the co rresponding VSS village. This variable can be thought of as a population-normalized measure of th e relative forest asset base of each of the villages selected for the survey This variable is utilized in the poverty-gap equation because poorer VSS villages are more likely to have a gr eater dependence on their forest resources. Thus,

PAGE 90

90 a negative relationship with PGR is expect ed, as larger proportionate areas under CFM protection should mean that more forest resource s are available to help mitigate poverty in these communities. Damodaran and Engel (2003) state that a minimum per capita land allotment of 1 to 2 hectares per VSS member can be consider ed a must for the success of JFM (p. 30). PFC is a proxy of the forest dependency of a community, as based on consumption at the household level. It is calculated as the consumption of forest products as a percent of total household consumption and is averaged (and wei ghted) over the households sampled in a given VSS village. Reddy and Chakravarty (1999) provide data indicating forest dependency is greater among poorer households. Thus, the following rela tionships are predicte d for PFC: a negative relationship in the average consumption and consumption inequality models, and a positive relationship in the PGR model. It is important to control for spatial differences in environmental and/or physical qualities and processes that may impact the components of social welfare that communities possess. One variable is included to represen t a control of this type. DTW is the average depth to groundwater (in feet) for a given VSS, as estimated by th e VSS Chairman or the Vice-Chairman. As agricultural production is a key com ponent of local economies in rural India, this is an important variable. Depths of several hundred feet are common. Thus, smaller DTW should equate with higher and lower PGR. A positive relationship with is likely as greater depths would tend to preclude the poor from accessing the resource. Institutional Variables The institutional variables are based on the Design Principles established by Ostrom (1990) as described in Chapter 1. These vari ables represent a subset of the unconventional factors that may influence the welfare measures ; Misra and Kant (2004) describe the use and

PAGE 91

91 importance of such non-neoclassical factors, in addition to conventional economic factors, in equations defining a joint pr oduction model of JFM. VBA is a variable that ranks the relative boundary awareness of the VSS villages. It is based on data gathered in the HH questionnaire; specifically, it is the pe rcent of VSS households (within a given VSS) that were recorded as being aware of the spatial boundaries of their VSS forest area. This variable is used only in the poverty-gap equati on, because the poorest households tend to utilize forests more than households that are ec onomically better-off. A negative relationship is therefore anticipated, i ndicating that VSS villages with higher levels of boundary awareness are likely to have a lower PG R value because the households of these VSS theoretically have better control over their CPR. The variable GAI measures the average genera l awareness (that households have) of four quantifiable VSS institutions and parameters; two each, respectively. It is based on an additive index (with a scale of 0 to 7) derived from sc oring HH questionnaire data against objective data that describe the VSS (e.g., hectares of forest managed). GAI measures the average index score for each VSS village and, thus, has the same scale (i.e., 0 to 7). It is onl y included in the average consumption equation because of the possibility that it may have a statistical influence on If so, a positive relationship is anticipated as hi gher levels of GAI are likely due to greater engagement of VSS-member households in CFM activities. As such, GAI would act as a proxy for level of participation, and theoretically w ould exert a positive influence on the average level of consumption in a village. CCA is a proxy for collective-choi ce arrangements and is measur ed in terms of whether or not the micro-plan for each VSS is oriented to wards the needs of forest-dependent households. This variable is also based on data collected in the HH questionnaire. CCA is a ratio that

PAGE 92

92 measures the percent of households indicating th at the VSS micro-plan for their VSS village reflects the interests of forest dependent househ olds. Because conflicting interests are always present in a community, and the APFD has a larg e hand in developing the VSS micro-plan, this document may potentially favor ru ral elites, other interest groups, or perhaps even the APFD, instead of the poorer households th at are the focus of CFM. To th e extent that higher values of CCA signify greater adherence to the theoretic al principle of respons ive and representative institutions, a positive relationship with is expected; conversely, a negative relationship is anticipated with respect to both and PGR. Monitoring of the resource is one of the key desi gn principles identifi ed by Ostrom (1990). The FP variable indicates whether or not the VSS has a formal patrol to monitor the VSS forest area. This variable is defined as a 0/1 dummy, where FP takes the value of 1 if the VSS has a regularly scheduled patrol, or employs a watchm an to monitor the VSS forest area, and 0 if neither. Negative relationships for FP are expected with both and PGR. GS represents graduated sanctions applied to locals who break the rules and regulations established by the VSS. This is captured as a dummy; GS equals 1 if the VSS has a two-tiered structure for punishing offenders, and 0 if not. The efficacy of GS as an explanatory variable is likely dependent upon the presence of graduated sanctions being widely known among members of the VSS. Although this is unlikely to be th e case, the poorest househol ds should be affected the most; thus a negative relationship with PGR is expected if GS is statistically significant. Description of the Data Table 4-2 displays the mean and standard devia tion of the variables that will be used in the empirical analysis of the thr ee economic measures examined in this study. These descriptive statistics are presented for each of the three regional samples collected and are intended to

PAGE 93

93 Table 4-2. Mean and standard deviation for V SS-level data: comparison by the region sampled. Adilabad (n = 18) Visakhapatnam (n = 20) Chittoor (n = 20) Variable Mean Std. Dev. Mean Std. Dev. Maximum Minimum Dependent: 9,604 422 8,390 938 5,708 324 ln 9.032 0.036 8.864 0.111 8.538 0.051 0.2396 0.0137 0.2704 0.0411 0.2320 0.0601 ln -1.457 0.059 -1.319 0.152 -1.495 0.273 PGR 0.0048 0.0023 0.0826 0.0702 0.0353 0.0319 Explanatory Demographic: N 174 206 137 297 104 59 ST* 9 -14 -5 -EDU 2.67 0.30 3.36 0.34 3.20 0.13 NGO 0.333 0.485 1.0 0.0 0.800 0.410 DGI 5.1 2.5 5.4 2.4 4.6 2.5 Explanatory Economic: YFP 300 119 1,869 366 764 373 YAL 8,290 2,750 2,621 544 9,044 1,487 YOS 20,848 2,568 19,690 3,828 12,714 745 IHW 8,662 7,618 6,024 6,397 3,922 5,657 TRN 4.9 1.7 4.5 1.9 5.8 2.6 Explanatory Bio-physical: LT 8.0 1.7 7.2 2.1 7.4 2.1 FA 275 138 159 98 312 111 RRE 3.9 3.0 3.2 4.3 3.3 2.1 PFC 1.9 0.3 4.0 0.6 4.3 0.4 DTW 120 97 79 55 359 121 Explanatory Institutional: VBA 0.176 0.125 0.279 0.169 0.234 0.105 GAI 2.12 0.72 2.53 0.83 2.24 0.45 CCA 0.663 0.131 0.658 0.140 0.640 0.164 FP 0.444 0.511 0.400 0.501 0.300 0.470 GS 0.722 0.461 0.700 0.470 0.450 0.510 The value listed for the ST variable is the mode. Denotes a 0/1 dummy variable.

PAGE 94

94 provide the reader with a broad comparison of the regions, which justify the inclusion of the regional dummy variables. Weighted means and cluster-adjusted standard deviations are calculated for those variables for which these me asures are directly derived from the householdlevel data (i.e., ln EDU, YFP, YAL, YOS, PFC).12 In addition, the mode of the ST variable is shown in place of a mean. The remainder of this chapter presents and desc ribes the data of some of these variables in further detail. An additional six variables of inte rest to the synoptic description of the data are also presented. These data are presented by region and variable type such that each of the three regions of Andhra Pradesh is represented by two tables, with each region discussed individually. The first table contains demographic and economic variables, while the second table contains bio-physical and institutional variables. The Adilabad Sample Table 4-3 summarizes the data for the demographic and economic variables collected from 18 VSS villages sampled in the Nirmal and Jannaram Divisions of the Adilabad Circle. The variables in this table are brie fly described as follows. ID is the unique field identification number given to each VSS village sampled. N is the total number of households in each VSS village sampled. CST is the predominant caste group of a VSS village, as determined by the mode of the households sampled. EDU is the av erage household education level, which is measured in years and can range from 0 (no educa tion) to 20 (a Ph.D.). ELC is the educational level of the VSS Chairman. DGI is the averag e number of womens self-help groups across households. MU is the weighted mean per-cap ita household consumpti on value of the VSS 12 For the region-level descriptive statistics in Table 4-2 that are derived from household data, it is necessary to correct the standard deviations in order to account for the effects that the clustered sampling of households has on the variance of data collected in this manner.

PAGE 95

95 village. YFP is the weighted mean household in come derived from forest products. YAL is the weighted mean household income derived from ag riculture and livestock. YOS is the weighted mean household income derived from other sour ces (e.g., employment, household enterprises). Table 4-3. Selected demographic and ec onomic variables for the Adilabad sample. ID VSS NAME N CST EDU ELC DGI MU YFP YAL YOS 101 Dounelli Thanda 77 ST 1. 34 12 2.6 6,435 731 4,327 15,405 103 Chincholi (B) 641 BC 3.93 10 4.7 9,133 0 3,368 31,539 104 Dongir Gama 80 BC 2.36 10 6.3 9,708 0 9,101 13,616 105 Kalva 643 BC 2.56 4 2.3 8,786 5 5,840 17,719 106 Rampur 340 BC 3.15 12 6.8 12,232 0 5,532 27,165 107 Arepalli 168 ST/BC 2.09 0 6.5 10,365 885 6,164 17,585 108 Mamada Thanda 34 ST 1.22 10 8.8 7,776 1,270 16,461 16,250 109 Rachchakota 80 ST 1.62 0 1.3 9,034 1,674 6,229 11,571 110 Ankena 80 ST 2.07 15 7.5 11,579 1,508 11,149 12,196 111 Badhankurthi 475 BC 2.19 10 3.4 10,506 100 23,669 20,326 113 Dildarnagar 150 SC 2.65 12 5.3 8,495 247 4,274 12,264 114 Kothagudem 78 BC 1.94 5 1.3 10,326 690 3,231 23,988 115 Danthanpalle East 80 BC/OC 3.48 12 7.5 10,800 325 11,753 10,706 116 Janguguda 41 ST 0.91 10 7.3 6,271 991 6,987 4,108 117 Gandi Gopalpur East 34 ST 0.63 5 2.9 6,843 1,397 798 10,602 118 Islampur (K) 22 ST 0.65 0 9.1 6,033 4,317 5,403 9,122 119 Puttiguda (Kotha) 35 ST 1.49 10 5.7 7, 583 573 2,735 5,540 120 Kancherabai 72 ST 1.27 7 2.8 8,523 1,799 5,354 10,271 Mean 174 2.67 8.0 5.1 9,604 300 8,290 20,848 Standard Deviation 206 0.30 4.6 2.5 422 119 2,750 2,568 Coeff. of Variation 1.18 0.11 0.58 0.50 0.04 0.40 0.33 0.12 The total number of households for each VSS village in the Adilabad sample (N) ranges from a low of 22 in Islampur (K ) to a high of 643 in Kalva. In general, the sample can be characterized as being comprised mainly of very-small to small VSS villages: 75% of the sample (12 VSS) is comprised of fewer than 80 househol ds. The other six VSS, however, all have more than 150 households. The Adilabad sample can generally be described as being tribal as well. Fully half (nine VSS) of the VSS villages listed in Table 4-3 are comprised of a majority of Scheduled Tribes (ST) households. If a given VSS is listed as majo rity Scheduled Tribes, it will usually be the case

PAGE 96

96 that 100% of the households are of the Scheduled Tribes designation. An exception is Arepalli VSS, where the number of Scheduled Tribes an d Backward Castes households is equal. The other caste group that appears with some frequency (six VSS) as a majority is the Backward Castes; it also shares the majority in two othe r VSS. The majority-Backward Castes VSS villages appear to be associated with larger villages (i.e., greater values of N). The educational variable EDU is a measur e of the average number of school years completed in each household that is then averaged over all of the households sampled in a given VSS village. While the potential maximum for this variable is 20 years, the observed maximum in this sample was less than 4 years. The lower education levels appear to be correlated with Scheduled Tribes. In addition, the four lowest EDU values are among the five smallest VSS villages in terms of N, while the five highest EDU values are in the three largest VSS (i.e., Kalva, Chincholi (B), and Rampur). In general, the educational level of the VSS Chairman is high, with ELC equal to or gr eater than 10 years for eleven of the VSS. Three VSS have a Chairman with no education. MU (average per-capita consumption) range s from a low of 6,033 Rupees (Rs.) in Islampur (K), to a high of Rs. 12,232 in Rampur This range is equiva lent to roughly US$131 to US$266 based on the September 19, 2006 exchange rate The weighted mean is Rs. 9,604, with a very low cluster-adjusted standard deviation (422 ) and coefficient of variation (4%). This means that the variation of the valu e of per-capita consumption am ong households in the Adilabad sample is low. The six lowest values of MU are all in VSS villages with a majority of Scheduled Tribes. The income variables are derived by averaging at the household level, which is why they appear relatively large compared to the mean levels of consump tion at the individual level (i.e.,

PAGE 97

97 MU). As mentioned previously, income data in developing countries ar e subject to inaccuracy due to various factors. The use of YFP, YAL and YOS (measures of income from forest products, agriculture, and other sources, respectively) must, therefore, be cautioned. The weighted mean YFP is greater than Rs. 500 in 11 of the 18 VSS and reached a maximum of Rs. 4,317 in Islampur (K). Only three VSS report ze ro income from forest products. Income from forest products (YFP) is associated with VSS vill ages that are generally both tribal and remote (similar to the VSS villages with the lowest values of MU); and those VSS villages with little or no YFP appear to be larger in size (i.e., larg er N). The exception is Dongir Gama, a small VSS village with commercially worthless trees. They were given poor quality land in the 1980s when the government relocated th eir village due to the construction of a dam. Comparison of the weighted income means also illustrates that agricultural income (YAL) and income from other sources (YOS), which is mainly derived from employment, are the main economic drivers of the VSS in the Adilabad sa mple. This actually applies to rural Andhra Pradesh as a whole. Mean YAL equals Rs. 8,290; only one VSS village averaged less than Rs. 1,000 from agriculture a nd livestock (Gandi Gopalpur). YOS appears to be very important in Adilabad: the mean is Rs. 20,848 and only three V SS have mean YOS values that are less than Rs. 10,000. Table 4-4 summarizes data on the bio-physical and institutional vari ables obtained from the Adilabad sample. LT is the length of time that the VSS has been practicing JFM/CFM. LT was calculated by subtracting the year of VSS es tablishment from the year 2005; as such, the maximum possible age is 12 years, as field impl ementation of JFM in Andhra Pradesh began in 1993. FA is the total forest area (in hectares ) that the VSS manages. Data on both LT and FA were obtained from the VSS micro-plans. DTF is the walking distance (in minutes) to the VSS

PAGE 98

98 forest area, averaged over th e households sampled (w ithin a given VSS). PQF and PMU are the quality and main use, respectively, of the fore st area before protection was established under the VSS. The data for these two variables were de rived from subjective ratings made by the VSS Chairman or Vice-Chairman on Likerd scales. PFC is the percentage of household consumption that is derived from forest produc ts, averaged over the households sampled. EP is an index that measures the equity in particip ation of the VSS; it is the percentage of VSS member households that believe the process of selecting the VSS Management Committee was free and fair, as opposed to being dominated by local elites or ot her groups. PG measures the perception of the government, as based on the relationship betw een the VSS Management Committee and the APFD, according to a Likerd sc ale measure made by the VSS Chairman or Vice-Chairman. The FP variable indicates whether or not the VSS has a formal patrol to monitor the VSS forest area. Table 4-4. Selected bio-physical and inst itutional variables for the Adilabad sample. ID VSS NAME LT FA DTF PFC PQF PMU EP PG FP 101 Dounelli Thanda 6 275 40 4.4 3 4 0.875 2 1 103 Chincholi (B) 5 180 40 0.6 2 3 1.000 2 0 104 Dongir Gama 7 40 33 1.4 4 0 0.583 2 0 105 Kalva 11 621 33 2.0 3 2 0.778 1 0 106 Rampur 9 164 36 1.2 3 2 0.813 2 1 107 Arepalli 9 280 30 1.8 2 4 1.000 2 1 108 Mamada Thanda 8 343 30 3.9 2 4 0.882 1 0 109 Rachchakota 7 227 54 2.7 3 4 0.941 1 0 110 Ankena 6 150 23 2.0 1 4 1.000 1 0 111 Badhankurthi 9 154 60 1.2 3 1 0.750 1 0 113 Dildarnagar 9 385 47 3.1 3 4 1.000 1 1 114 Kothagudem 7 150 20 2.9 3 3 0.625 2 0 115 Danthanpalle East 8 294 71 4.1 2 4 0.714 2 0 116 Janguguda 7 385 54 6.5 3 3 0.944 2 1 117 Gandi Gopalpur East 7 294 42 6.7 1 4 0.900 2 0 118 Islampur (K) 8 257 52 10.1 1 4 1.000 4 1 119 Puttiguda (Kotha) 9 380 39 4.5 3 4 0.900 4 1 120 Kancherabai 12 500 53 5.7 3 4 0.895 1 1 Mean 8.0 275 40 1.9 0.867 0.44 Standard Deviation 1.7 138 3.1 0.3 0.13 0.51 Coeff. of Variation 0.22 0.50 0.08 0.17 0.15 1.15 Mode 3 4 1.000 2 Median 3.0 4.0 0.898 2.0

PAGE 99

99 As shown in Table 4-4, LT (length of time under JFM/CFM) ranges from 5 to 12 years for the Adilabad sample; the mean is 8 years. The size of the VSS forest area (FA) ranges from 40 to 621 hectares, with a mean of 275 hectares. The minimum (40 hectares) is unusually low for the Adilabad sample, as the next largest FA value is nearly four times as la rge. In addition, threequarters of the sample (12 VSS) is greater than 200 hectares. The distance to the VSS forest area (DTF) ra nges from 20 to 71 minutes, with a weighted mean of 40 minutes. PFC is a proxy of the fo rest dependency of a community, based on consumption at the household level. PFC indicates that the consumption of forest products as a percentage of total household c onsumption ranges from 0.6% to 10.1%; the weighted mean is 1.9%. Comparison with the CST variable in Tabl e 4-3 indicates that each of the six highest values of PFC correspond to VSS that are whol ly or predominately composed of Scheduled Tribes households. PQF represents a subjective quality rating of th e forest area prior to protection by the VSS, where 1 = good, 2 = slightly degraded, 3 = degraded, and 4 = very degraded. The mode and median both equal (degraded). This result is not surprising since the legal basis for JFM rests upon rehabilitation of degraded land. However, some VSS clearly were given charge of nondegraded areas in order to avoid potential future degradation, and to s upport the development of communities such as Islampur (K), for example. PMU identifies the main uses of the forest area prior to protection by the VSS, where 0 = no previous use, 1 = fuelwood, 2 = construction material, 3 = agricultural implements, and 4 = NTFPs. The majority of the Adilabad sample (11 of 18 total VSS) indicated that NTFP collection was the main use of their forest prior to JFM/CFM.

PAGE 100

100 EP is an index created to proxy for the equity in participation within a given VSS. The potential range is 0 (no equity) to 1 (perfectly equitable participation), but the range for EP in the sample is 0.583 to 1.000; only five VSS villages have EP values below 0.800. The magnitude of the mean (0.867) reflects a high level of equity with regard to the internal political and institutional process. The perception of government (PG) was meas ured with an ordinal subjective ranking where 1 = very good, 2 = good, 3 = fair, and 4 = bad. The vast majority of the sample (16 VSS) indicated that the relationship with the APFD officer was e ither good or very good. The other two VSS, Islampur (K) a nd Puttiguda (Kotha), both indicat ed a bad relationship. The VSS Chairman of Puttiguda described in detail th e corruption of the local APFD Beat Officer: this person took bribes from two other villages to allow them to cut wood in the VSS area. To underscore this point, the VSS Chairman and his as sociate also intercepted some timber thieves and confiscated their axe during the inspecti on of the VSS forest area by the principal investigator. FP is a 0/1 dummy variable that takes the value of 1 if the VSS either undertakes a formal patrol of their VSS area, or employs a watchm an to monitor it. Surprisingly, only eight VSS villages (44% of the sample) have a formal mo nitoring structure in place. Many claim that they informally monitor their VSS ar ea because their farmland is adj acent to it, or the only access is through their farmland or village. The Visakhapatnam Sample Table 4-5 summarizes the data for the dem ographic and economic variables from the Visakhapatnam Division sample (20 VSS total). The total number of households for each VSS village in the Visakhapatnam samp le ranges from a low of 12 in Ch. Konda Veedhi to a high of 1,362 in Darlapudi. In general, the sample can be characterized as bei ng comprised mainly of

PAGE 101

101 very-small to small VSS villages: 75% of the sample (15 VSS) is comprised of fewer than 80 households. Thus, from the sta ndpoint of VSS size the Visakhap atnam and Adilabad samples are similar. Table 4-5. Selected demographic and econom ic variables of the Visakhapatnam sample. ID VSS NAME N CST EDU ELC DGI MU YFP YAL YOS 201 Appampalem 133 OC 1.530 10 3.8 5,115 0 2,561 13,564 202 Kunnempudi 33 ST 1.096 5 9.1 5,884 10,510 720 11,409 203 Poolabanda 42 ST 1.556 5 4.8 4,391 4,381 888 6,954 204 Kothavalasa 75 ST 1.941 7 6.7 6,021 4,102 3,863 9,644 205 Yerrabandalu 84 OC 1.647 10 4.8 6,523 0 5,567 10,141 206 K. Boddapadu 57 ST 2.464 0 3.5 7,117 3,969 3,265 4,440 207 Ch. Konda Veedhi 12 ST 2.053 1 8.3 4,834 9,821 501 3,672 208 Kandulapalem 70 ST 1.884 3 2.9 5,765 3,117 5,771 5,482 209 Vemagiri 344 BC 3.997 10 1.5 7,466 0 4,999 19,849 210 Somannapalem 117 BC 4.670 10 4.3 10,983 0 4,110 21,332 211 Chinauppalam 71 OC 3.206 5 5.6 7,004 0 3,588 12,129 212 Darlapudi 1,362 OC 3.913 5 6.6 10,161 1,699 1,906 26,874 213 Kondiba 70 ST 2.249 10 4.3 4,527 2,195 1,095 10,564 214 Medaparthi 54 ST 2.811 10 5.6 4,563 4,848 1,210 255 215 Punyagiri 60 ST 4.046 9 3.3 8,425 6,766 184 9,944 216 Latchannadorapalem 46 ST 2.442 10 6.5 5,491 4,385 2,310 12,073 217 Dellipadu 17 ST 0.321 3 11.8 3,474 3,191 970 2,385 218 Urumula 42 ST 1.477 0 4.8 3,968 6,818 1,133 3,077 219 Thodubanda 28 ST 0.985 0 3.6 4,173 1,692 95 1,703 220 Kandulaguddi 17 ST 2.326 3 5.9 3,689 1,468 763 3,024 Mean 137 3.36 5.8 5.4 8,390 1,869 2,621 19,690 Standard Deviation 297 0.34 3.9 2.4 938 366 544 3828 Coeff. of Variation 2.17 0.10 0.67 0.44 0.11 0.20 0.21 0.19 The Visakhapatnam sample is predominantly tribal: nearly 75% (14 VSS) of the VSS villages listed in Table 4-5 are primarily compri sed of Scheduled Tribes households. Of these, 10 VSS villages consist entirely of Scheduled Tr ibes households and the other four remaining villages range from 83.3% to 95% ST-designate d households. Additional caste groups that form a majority in a given VSS include Other Castes (four VSS) and Backward Castes (two VSS). Three of these VSS are completely ho mogeneous (Appampalem, Yerrabandalu, and

PAGE 102

102 Somannapalem), while the other three VSS have varying degrees of caste mixture. These six VSS tend to correlate with a larger village size. EDU values (average years of education acr oss households) are rather low across the Visakhapatnam sample, although the weighted mean (3 .36) is greater than the weighted mean of the Adilabad sample (2.67). Four of the five hi ghest EDU values correlate with VSS villages classified as either Other Castes or Backward Castes. In addition, the four lowest EDU values tend to be among the smallest VSS (in terms of si ze), while the five highest EDU values contain three of the four largest VSS villages (Darlapudi, Vemagiri, and Somannapalem). In general, the average educational level of the VSS Chairman is not very high, although ELC is greater than or equal to 10 years for seven of the VSS. Of the remainder, more than half (seven VSS) have Chairman that completed only th ree years of school or less. MU, the average measure of the value of i ndividual consumption, ranges from Rs. 3,474 (Dellipadu) to Rs. 10,983 (Somannapalem); or roughly $76 to $239, based on the September 19, 2006 exchange rate. The weighted mean is Rs. 8, 390, with a relatively low cluster-adjusted standard deviation (938) and co efficient of variation (11%). The weighted mean of MU is approximately 1,200 Rupees less than in the Adilabad sample. The weighted mean of income from other sources, YOS, (Rs. 19,690) is similar to the weighted mean YOS of the Adilabad sample (R s. 20,848). In contrast, the average mean of income from both forest products and agricultural inco me (YFP and YAL, respectively) in the Visakhapatnam sample differ substantially from the Adilabad sample. For example, Table 4-5 illustrates that the weighted mean YFP equals Rs. 1,869, which is more than six times as much as mean YFP for Adilabad. Clearly, dependence of the VSS villages on forests resources is greater in Visakhapatnam. Like the Adilabad sample though, YFP is associated with the tribal VSS: five

PAGE 103

103 of the six non-tribal VSS report zero income from forest products Interestingly, the weighted mean of YAL (Rs. 2,621) is three times lower than in Adilabad. More than a third of the sample (seven VSS) has a weighted mean of YAL that is less than 1,000 Rupees, while the maximum YAL is only Rs. 5.771 (Kandulapalem). Table 4-6 presents data on the bio-physical and institutional variables obtained from the Visakhapatnam sample. The length of time the village has been under JFM/CFM (LT) ranges from two to 11 years for the Visakhapatnam sample and averages 7.2 years. The size of the VSS forest area (FA) ranges from 41 to 363 hectares. Only seven VSS are greater than 200 hectares. The mean and median (159 ha and 131.5 ha, respec tively) for FA indicate average VSS size in the Visakhapatnam sample is about half that seen in Adilabad (275 ha and 267 ha, respectively). Table 4-6. Selected bio-physic al and institutional variables for the Visakhapatnam sample. ID VSS NAME LT FA DTF PFC PQF PMU EP PG FP 201 Appampalem 7 126 43 5.4 2 2 0.889 1 0 202 Kunnempudi 7 150 71 6.9 3 4 0.789 2 0 203 Poolabanda 9 225 51 7.1 3 3 0.765 2 1 204 Kothavalasa 9 100 106 5.1 1 4 1.000 1 0 205 Yerrabandalu 7 115 59 4.6 1 1 1.000 2 0 206 K. Boddapadu 11 75 81 9.0 1 4 0.889 2 1 207 Ch. Konda Veedhi 9 250 13 3.0 3 4 1.000 2 0 208 Kandulapalem 5 50 43 6.1 2 4 0.950 3 0 209 Vemagiri 6 60 26 0.3 2 1 0.667 1 0 210 Somannapalem 9 139 28 2.0 2 1 1.000 2 1 211 Chinauppalam 7 41 18 3.2 4 0 0.533 1 0 212 Darlapudi 2 70 29 3.9 2 1 0.400 2 1 213 Kondiba 7 124 27 8.3 3 4 0.688 2 0 214 Medaparthi 7 313 35 4.6 1 1 1.000 2 0 215 Punyagiri 9 200 21 3.7 2 4 0.944 2 1 216 Latchannadorapalem 9 137 52 3.8 2 4 0.944 2 0 217 Dellipadu 5 363 18 11.0 2 4 0.824 3 0 218 Urumula 9 352 14 7.6 2 4 0.882 2 1 219 Thodubanda 5 200 20 6.1 3 4 0.750 2 1 220 Kandulaguddi 5 92 19 9.4 3 1 0.875 2 1 Mean 7.2 159 39 4.0 0.839 0.40 Standard Deviation 2.1 98 5.5 0.6 0.17 0.50 Coeff. of Variation 0.29 0.62 0.14 0.14 0.20 1.26 Mode 2 4 1.000 2 Median 2.0 4.0 0.886 2.0

PAGE 104

104 DTF, the distance to the VSS forest area, ranges from 13 to 106 minutes, with a weighted mean of 39 minutes (which is n early identical to the Adilabad sa mple). PFC (a proxy for forest dependency) indicates that the consumption of forest products from 0.3% to 11.0% of total consumption. The weighted mean is 4.0%, which is twice that of the Adilabad sample. The tiny VSS village of Dellipadu, which only has 17 house holds, has the maximum PFC value; this VSS also has the lowest MU value (m ean per-capita consumption). Intere stingly, the same scenario is observed in the Adilabad sample. The mode and median of the measure of prio r forest quality (PQF) both equal (i.e., slightly degraded). The mode is representative of nine VSS villages, while another seven VSS rated PQF as either or (i.e., degraded or ve ry degraded). A majority of the villages of the Visakhapatnam sample (11 VSS) indicated that NT FP collection was the main use of their forest prior to JFM/CFM; six VSS indicated fuelw ood collection was the previous main use. EP is the equity in participation index, wh ich averaged 0.839 but ranges from 0.4 to 1.0 in the sample. Seven VSS have an EP value below 0.8. The magnitude of the mean is similar, though less, than the mean EP for Adilabad (0. 867). PG proxies the perception of government; almost the entire sample (18 VSS) indicated that the relationship with the APFD field officer was either good or very good. The other two VSS villages, Kandulapal em and Dellipadu, both indicated a fair relationship with their APFD officer. FP is a dummy indicating a formal patrol of the VSS area. Similar to the Adilabad data, on ly eight VSS villages (40% of the sample) have a formal monitoring structure in place. The Chittoor West Sample The Chittoor West sample is comprised mainly of small and medium sized VSS villages: 75% of the sample (15 VSS) consists of VSS villages with 70 or more households. The sample

PAGE 105

105 ranges from a low of 26 in Kondakindaindlu to a high of 224 in Muddanapalli. The coefficient of variation (0.57) indicates a much tighter distribution of N for this sample than for either the Visakhapatnam and Adilabad samples (CV = 2.17 and 1.20, respectively). Table 4-7. Selected demographic and economic variables for the Chittoor West sample. ID VSS NAME N CST EDU ELC DGI MU YFP YAL YOS 301 Krishnapuram 200 SC 4.165 10 4.0 7,853 664 21,103 13,614 302 Nadimuru 99 OC 3.093 12 2.0 4,435 0 10,954 11,710 303 Jowkupalli 58 OC 3.249 5 5.2 5,973 0 10,203 17,060 304 Kudavagadda 157 SC/BC 3. 616 12 4.5 4,141 346 11,132 10,580 305 Muddanapalli 224 BC 3.052 5 4.5 4,223 79 11,433 14,138 306 Urinayanipalle 84 BC 1.961 5 4.8 4,423 330 575 12,373 307 Diguvathanda 70 ST 2.510 5 5.7 7,919 0 4,482 7,021 308 Kondakindaindlu 26 ST 1.830 0 7.7 5,674 278 8,170 9,268 309 Mudiveducross 87 SC 4.175 7 2.3 6,428 0 4,362 13,331 310 Papepalli 190 BC 3.163 0 9.5 6,319 0 1,978 13,258 311 Rajugaripalli 85 OC 2. 998 7 4.7 5,978 0 7,724 8,458 312 Siddhartha Colony 48 ST 3.046 0 8.3 5,389 15,408 1,994 8,396 313 Atukurallapalle 70 OC 3.672 5 1.4 8,800 0 15,691 21,633 314 Chappidipalle 137 SC 3.046 7 2.9 5,016 140 5,646 11,985 315 Motlapalle 38 SC 3.107 10 5.3 5,417 0 10,159 11,154 317 Pengaragunta 190 BC 3.206 4 2.1 6,417 240 5,459 17,817 318 Adivilopalle 70 ST 2.746 5 0.0 7,252 395 10,056 9,511 319 Gowthumakulapalle 105 OC 3.591 5 6.7 6,225 50 10,037 9,654 320 Singiriguntapalle 103 BC 2.643 10 2.9 5,446 2,606 15,671 12,188 321 Irlapalli (81) 42 ST 2. 466 10 7.1 3,962 5,936 1,555 8,597 Mean 104 3.20 6.2 4.6 5,708 764 9,044 12,714 Standard Deviation 59 0.13 3.7 2.5 324 373 1487 745 Coeff. of Variation 0.57 0.04 0.59 0.54 0.06 0.49 0.16 0.06 Unlike the other two samples, Table 4-7 shows th at the Chittoor West sample can only be characterized as a poly-caste sample. There is an almost equal division of the CST variable: 25% (five VSS) of the VSS listed are majority Sche duled Tribes (ST) households, 25% are majority Backward Castes (BC) households, and 25% ar e majority Other Castes (OC) households. Another four VSS are majority Scheduled Cast es (SC) households. The remaining VSS is a bimodal BC/SC village. Three of the four smallest communities in the sample are Scheduled Tribes-designated VSS villages (Kondakindaindlu, Irlapalli ( 81), and Siddhartha Colony).

PAGE 106

106 EDU values, measuring the average level of education by household, are again rather low across the VSS; the weighted mean (3.20) falls between the Visakhapatnam mean and (3.36) and the Adilabad mean (2.67). The distribution (CV = 0. 04) is also somewhat tighter than the other two samples (CV = 0.10 and 0.11, respectively). In general, the average educational level of the VSS Chairman (6.2 years) is similar to the Vi sakhapatnam sample (5.8), with six of the VSS having an ELC value greater than or equal to 10 years. Of the remainder, more than 70% (10 VSS) have a VSS Chairman that completed five years of school or more. MU, which measures the average level of cons umption at the individua l level, ranges from a low of Rs. 3,962 Rupees in Irla palli (81) to a high of Rs. 8, 800 in Atukurallapalle. The dollar equivalent for the range is US$86 to US$191 ba sed on the 19 September 2006 exchange rate. The weighted mean is Rs. 5,708, which is 32% and 41% less than the Visakhapatnam and Adilabad samples, respectively. MU also exhibits a low cluster-adjusted st andard deviation (324) as indicated by the coeffici ent of variation (6%). The weighted mean of average forest income (Y FP) is Rs. 764. This is more than twice as large as the weighted mean for YFP in the Adilab ad sample, but less than half of the weighted mean for YFP in Visakhapatnam. YFP is concentr ated in three VSS villa ges (Siddhartha Colony, Irlapalli (81), and Singiriguntapalle ). These three VSS contain 41 of the 58 households (out of 396 total households in the Chittoor West sample) that report non-zero YFP. Moreover, of the 16 VSS villages that record weighted mean YFP values less than Rs. 500, half of these (eight VSS) derive no income at all from forest products. In contrast, Adilabad has only seven VSS with mean YFP less than Rs. 500, with only three havi ng zero YFP. In Visakhapatnam, there are five VSS villages with zero mean YFP but the rest all have a mean for YFP greater than Rs. 1,400.

PAGE 107

107 Table 4-7 shows that the weight ed mean of agricultural income (YAL) is Rs. 9,044, which is similar to the weighted mean YAL of the Ad ilabad sample (Rs. 8,290). Both are substantially greater than the Visakhapatnam weighted mean YAL (Rs. 2,621). Only one VSS village has a weighted mean for YAL that is less than Rs. 1,500 (Urinayanipalle); fully half the sample (10 VSS) records a weighted mean for YAL greater than Rs. 10, 000. In contrast to both the Visakhapatnam and Adilabad samples, the weighted mean of other income (YOS) in the Chittoor West sample is substantially less (approxima tely 35% to 40%), alt hough the distribution is tighter. The weighted mean of YOS equals Rs. 12, 714, with a cluster-adjusted standard deviation of 745 (CV = 6%). Thus, the minimum YOS (Rs. 7,021) for Chittoor West is higher than the minimum for the other samples. In Table 4-8, the length of time under VSS ma nagement averages 7.4 years, with a range of 3 to 12 years. The size of the VSS forest area (FA) ranges from 150 ha to 521 ha. There are no small VSS areas in the Chittoor West sample : the minimum of 150 ha (Jowkupalli) exceeds the median FA for Visakhapatnam by almost 20 ha. The mean and median (312 and 300, respectively) indicate that the average VSS forest size in the Chittoor West sample is roughly twice that of the Visakhapatnam sample (159 an d 131.5, respectively). FA size in the Chittoor West sample is also greater than the Adilabad sample, although the difference in mean and median is less than 12%. DTF (average distance from household to fo rest) ranges from 18 to 56 minutes, with a weighted mean of 37 minutes (which is slightly less than the other two samples). The forest dependency proxy, PFC, indicates that the consump tion of forest products as a percentage of total household consumption ranges from 1.4% to 9.6%. The weighted mean is 4.3%, which is twice that of the Adilabad sample and slightly higher than the Visakhapatnam sample. Irlapalli

PAGE 108

108 (81) is the VSS with the maximum PFC value; this VSS also has the lowest MU value (mean per-capita household consumption). Interestingly, the same was observed in both of the other samples. The mode and median of PQF both equal , in dicating that eight villages rated the forest quality before VSS protection as degraded. A nother four VSS rated PQF as (i.e., very degraded). The data for the PMU variable indicate that a large majority of the Chittoor West sample (14 VSS) report that fuelwood collection was the main use of their forest area prior to JFM/CFM, while NTFP collection was listed by five VSS as the previous main use. Table 4-8. Selected bio-physic al and institutional variables for the Chittoor West sample. ID VSS NAME LT FA DTF PFC PQF PMU EP PG FP 301 Krishnapuram 8 345 34 3.1 3 4 1.000 2 1 302 Nadimuru 8 200 29 5.2 3 3 0.857 1 0 303 Jowkupalli 7 150 48 5.9 4 4 0.900 3 0 304 Kudavagadda 7 200 31 6.4 2 1 0.917 2 0 305 Muddanapalli 10 500 28 4.6 2 1 0.846 2 1 306 Urinayanipalle 3 200 41 6.2 2 4 0.824 4 0 307 Diguvathanda 5 200 23 2.8 3 1 0.750 2 0 308 Kondakindaindlu 7 402 18 7.8 3 1 0.706 2 0 309 Mudiveducross 7 265 24 1.4 4 1 1.000 2 1 310 Papepalli 3 199 54 3.9 2 1 0.800 2 0 311 Rajugaripalli 8 457 52 2.5 4 1 0.867 2 0 312 Siddhartha Colony 7 287 56 3.9 3 4 0.938 2 1 313 Atukurallapalle 7 390 32 2.3 3 1 0.889 2 0 314 Chappidipalle 7 300 37 3.1 3 1 0.571 2 0 315 Motlapalle 9 300 32 2.7 2 1 0.412 2 0 317 Pengaragunta 9 521 46 2.8 3 1 0.667 3 1 318 Adivilopalle 9 350 41 3.8 4 1 0.667 3 0 319 Gowthumakulapalle 3 200 42 3.4 2 1 0.571 3 0 320 Singiriguntapalle 7 380 35 8.7 2 1 0.444 1 1 321 Irlapalli (81) 12 390 41 9.6 2 4 1.000 1 0 Mean 7.4 312 37 4.3 0.781 0.30 Standard Deviation 2.1 111 2.1 0.4 0.18 0.47 Coeff. of Variation 0.28 0.35 0.06 0.09 0.23 1.57 Mode 3 1 1.000 Median 3.0 1.0 0.835 The range of the equity in participation (EP) proxy is 0.412 to 1.0, with a mean of 0.781. EP values less than or equal to 0.75 are recorded for eight VSS villages. The magnitude of the

PAGE 109

109 mean is less than the mean EP for both Adilaba d (0.867) and Visakhapatnam (0.839). Table 4-8 indicates that 75% (15 VSS) report PG values of either or , indicating that the relationship of the EC with the APFD officer was either very good or good. Four VSS record a PG value of (i.e., fair relationship). Th e only VSS to report a bad relationship is Urinayanipalleduring our interview they accused th e Forest Beat Officer of serious bribery and corruption charges. The FP dummy indicates that a formal patrol or a watchman is employed in only six VSS (30%) of the sample from Chittoor West. This percentage is lower than in both the Adilabad (44%) and Visakhapatnam (40%) samples. Summary Direct examination of the economic and di stributional impacts of JFM/CFM over time would ideally be made using a panel of data. As th is was not possible, analytical focus is placed upon factors influencing the relative success of JFM/CFM that can be captured with crosssectional data. Models representi ng different dimensions of soci al welfare are analyzed using economic indicators as dependent variables. The overarching purpose of such regressions is to evaluate selected explanatory variables for their c ontribution, or lack thereof, to the ability of JFM/CFM to reduce poverty and in equality. Highlighting key relationships that drive or hinder this ability is important for providing empirical evidence to researchers seeking to refine the facilitating institutions of JFM/CFM, or similar forest management programs. Results derived from the regression models specified in the first section of this chapter are presented and discussed in the following chapter.

PAGE 110

110 CHAPTER 5 RESULTS AND DISCUSSION Introduction This chapter presents, discusses, and summarizes the estimation results of the study. Table 5-1 contains the mean, standard deviation, maxi mum, and minimum for all of the variables used in the regressions. The total number of observati ons for the data set is 58 VSS villages. The logged average consumption model (ln ) model and the poverty-gap ratio (PGR) model both provide good results. For the ln regression, the joint test for all parameter coefficients equal to zero is F(19, 38) = 10.83, with a p-value13 of 0.0000. For the PGR regression, the LR Chi2(19) = 110.99 with a p-value of 0.0000. These resu lts indicate that th e null hypothesis (H0: all coefficients = 0) for each regression is reject ed. In contrast to these two models, the null hypothesis that all coefficients e qual zero cannot be rejected for the consumption inequality (ln ) regression: F(18, 39) = 1.11 (p = 0.382); indicating the overall results of this regression are very poor. The estimation results for ea ch variable of the three regressions are displayed below in Table 5-2; the statistically significant coefficien ts (and their p-values) are highlighted in bold type for p 0.050. The regression results will only be discussed br iefly for the inequality equation, while the results for the consumption and poverty regression s will be examined in further detail in the sections that follow; included are brief interpreta tions of the results and supporting analyses (e.g., correlations and joint significance tests) for selected variables. After a discussion of the results of the economic models, the results of an auxiliar y model are presented which explain changes in forest quality under JFM/CFM in Andhra Prades h. Unlike the previous models, the unit of 13 The probability value (or p-value) is defined by Gujarati to be the lowest significance level at which a null hypothesis can be rejected (1 995, p. 132). It can also be thought of as the exact probability of committing a Type-I error (i.e., the probability of rejecting a true hypothesis).

PAGE 111

111 analysis is the household and the de pendent variable proxies for the effects of forest restoration. The chapter concludes with a detailed discussion of the length of time (LT) variable and its impact on average consumption and forest quality. Table 5-1. Basic statistics for VSS-le vel regression analyses (58 observations). Variable Variable Description Mean Std. Dev. Maximum Minimum Dependent: Mean per-capita household consumption 6,850 2,236 12,232 3,474 ln Natural log of 8.7804 0.3242 9.4118 8.1531 Gini coefficient 0.2476 0.0553 0.3777 0.1097 ln Natural log of -1.4224 0.2389 -0.9737 -2.2100 PGR Poverty-gap ratio 0.0421 0.0551 0.2648 0 Explanatory Demographic: N Number of households 137 210 1,362 12 ST Scheduled Tribes village (dummy) 0.483 0.504 1 0 EDU Mean HH education level 2.474 1.018 4.670 0.321 NGO NGO assistance for forestry (dummy) 0.724 0.451 1 0 DGI DWCRA group index 0.050 0.024 0.118 0 Explanatory Economic: YFP Income from forest products 1,930 3,033 15,408 0 YAL Income from agriculture and livestock 5,970 5,215 23,669 95 YOS Income from other sources 12,073 6,442 31,539 255 IHW Investment per HH for VSS works 6,118 6,735 27,388 411 TRN Number of training events 5.05 2.15 12 1 Explanatory Bio-physical: LT Length of time under JFM or CFM 7.50 1.99 12 2 FA VSS forest area size 248 132 621 40 RRE VSS forest area per member 3.44 3.22 15.25 0.15 PFC Percent forest product consumption 4.59 2.55 11 0.3 DTW Depth to groundwater 188 157 500 0 Explanatory Institutional: VBA VSS boundary awareness ratio 0.232 0.140 0.579 0 GAI General awareness of institutions 2.303 0.692 4.5 0.700 CCA Collective-choice arrangements proxy 0.653 0.144 1 0.273 FP Formal patrol dummy 0.379 0.489 1 0 GS Graduated sanctions dummy 0.621 0.489 1 0

PAGE 112

112 Table 5-2. Estimated regression coe fficients and associated p-values. ln model ln model PGR model Variable Coefficent P-value Coe fficent P-value Coefficent P-value N -0.000267 0.218 0.000077 0.803 --ST 0.197503 0.018 0.133558 0.271 -0.053690 0.001 EDU 0.101853 0.019 -0.054342 0.378 -0.002280 0.750 NGO 0.101585 0.188 ---0.045400 0.012 DGI 1.01758 0.410 0.782015 0.646 -0.504483 0.052 YFP -0.000004 0.708 -0.000024 0.161 0.000001 0.456 YAL 0.000007 0.297 -0.000001 0.952 0.000001 0.592 YOS 0.000015 0.021 0.000009 0.307 -0.000004 0.000 IHW -0.000009 0.026 -0.000008 0.166 0.000002 0.022 TRN 0.010360 0.398 -0.039771 0.036 --LT -0.204649 0.016 0.097490 0.418 0.012810 0.259 LT2 0.013876 0.010 -0.005351 0.478 -0.001013 0.175 FA -0.0001683 0.504 -0.000013 0.972 --RRE ----0.005740 0.000 PFC -0.050075 0.001 -0.015716 0.435 0.012500 0.000 DTW 0.000174 0.481 -0.000030 0.934 -0.000097 0.026 VBA -----0.056620 0.152 GAI 0.049750 0.179 ----CCA -0.471046 0.020 0.011689 0.968 0.135157 0.000 FP --0.121990 0.117 -0.010741 0.239 GS -----0.012444 0.198 ADIL 0.431535 0.000 -0.257758 0.042 -0.130066 0.000 CHIT -0.146724 0.177 -0.104503 0.513 -0.034326 0.031 _CONS 9.18943 0.000 -1.47981 0.009 0.044168 0.443 Economic Models Average Consumption Model Including a constant, twenty explanatory va riables are regressed on the natural log of which measures mean per-capita household consump tion and is hereafter re ferred to as average consumption (recall that average consumption is used in this study to measure the economic

PAGE 113

113 component of social welfare, per Equation 4-4) The regression was tested for omitted variable bias using the Ramsey RES ET test; the null hypothesis (H0: no omitted variables) could not be statistically rejected (p = 0.668) As the presence of heteroskedas ticity is possible, most likely due to the income variables, th e Breusch-Pagan/Cook-Weisberg test was applied. However, the null hypothesis of constant variance could not be statistically rejected (p = 0.657) either. The adjusted R-squared value of 0.766 indicates that the model is successful in explaining a large portion of the variation observ ed in the logged values of Ten of the twenty variables analyzed were statistically significant at p-values 0.050. These variables are discussed individually in the paragraphs that follow. The Scheduled Tribes dummy (ST) is statisti cally significant (p = 0.018) with a positive sign on the coefficient. This indicates that ST-des ignated VSS villages are associated with higher values of mean per-capita household consumption, which supports the a priori expectation that ST-designated VSS are likely to have benef ited economically from other tribal welfare programs. The coefficient on EDU (educationa l index) is statistically sign ificant (p = 0.019), with the anticipated positive sign. Thus, higher mean edu cation levels are correlated with economically higher levels of average consumption. This is cons istent with the expectation that an increase in human capital through education will translat e into economic benefits to a community. As expected, income from other sources (YOS) is statistically signifi cant (p = 0.021) with a positive sign on the coefficient. Thus, an increase in YOS will result in higher levels of average consumption for the VSS in the sample. Although YOS includes income from household enterprises and remittances, this variable is mainly comprised of employment income derived outside the household. Therefore, this estima tion result indicates that local employment

PAGE 114

114 opportunities are an economically significant component of the we lfare and development of the VSS sampled. Of the three income variables, only YOS is statistically significant. However, a positive Pearson correlation coefficient ( r = 0.279) between YOS and YAL (income from agriculture) indicates a complementary relati onship between these two variables. Indeed, employment income for many households is deri ved from providing agricu ltural labor to other farms; YAL is jointly significant with YOS: F(2, 38) = 3.34 (p = 0.046).14 The IHW variable proxies for ex ternal investment in VSS villages in the form of average wages paid (per household) for VSS works projects. This variable is statis tically significant (p = 0.026) in the regression; however, the negative sign contradicts the a priori expectation that external investments will increase average consump tion. This result may signify that poorer VSS villages have been targeted to receive greater amounts of assistance money. (The same argument can also be applied to Scheduled Tribes villages ). The rationale for this might be that other economic opportunities are relatively limited. Th is interpretation is supported by negative correlations of IHW with both YAL ( r = -0.247) and YOS ( r = -0.229), and the joint statistical significance (p = 0.023) of these three variables. The length of time (LT) variable and its squared term (LT2) are significant with p-values of 0.016 and 0.010, respectively. These variables are jointly significant: F(2, 38) = 4.06, with a pvalue of 0.025. LT has a negative coefficient, wh ile LT2 has a positive coefficient. The observed signs of the coefficients were opposite of what was expected. Instead, the estimation results indicate a convex (i.e., U-shaped) relations hip between and the length of time under the JFM/CFM program; average consumption initially d eclines before increasing as the cross-section 14 The degrees of freedom available precl uded the testing of interaction terms. Joint significance tests using linear restrictions on the models were used to determine whether or not selected groups of variables had collective explanatory power.

PAGE 115

115 of the VSS sample increases in age. This relations hip is explored in further detail at the end of this chapter. The variable PFC is negative and highly statistically significant (p = 0.001), meaning that as forest consumption (as a percentage of tota l consumption) decreases, average consumption also increases. This is expected because PFC is used as a proxy for forest dependency. As the collection of forest products (particularly NTFPs) has a high opportunity cost in terms of time, greater dependency on forest resources likely signifies a lack of other remunerative opportunities. Empirical data presented by Reddy and Chakravarty (1999) show that forestry income is important for poor households, especially those that own less land. However, Heltberg (2001) used a measure of forest dependency as a dependent variable in a study of forest conservation and management, instead of as an explanatory variable. As this raises the possibility of PFC being an e ndogenous variable, a Wu test wa s performed. The endogeneity of PFC was statistically rejected.15 The collective-choice institutional proxy (CCA) is statistically significant (p = 0.020) and has a negative sign, which is opposite of the a priori expectation that VSS villages that actually place an emphasis on forest-dependent households would have a positive influence on average consumption. However, the negative sign of this variable might reflect that the CFM institution being proxied has more relevance in the poorer VSS villages for which CFM is designed and oriented towards. The location dummy ADIL is highly statistically significant (p = 0.000) with a positive coefficient, while CHIT (p = 0.177) is statisti cally insignificant. This means that the Chittoor 15 A Wu test was performed by regressing PFC on some of the exogenous variables of the equation and some other available variables. The predicted values of PFC were then placed in the original equa tion. The coefficient of PFChat was statistically insignificant (p = 0.828), indicating there is no correlation of PFC with the error term.

PAGE 116

116 sample cannot be statistically distinguished from the Visakhapatnam sample in terms of average consumption, while the Adilabad sample is statis tically distinct. Specifically, the VSS villages of the Adilabad sample are economically better off (as measured by average consumption levels) than those of the Visakhapatnam and Chittoor samp les: for example, the data show that 60% (24 of 40) of the VSS villages of the combined Visakhapatnam and Chittoor samples have average consumption levels that are less th an those in the Adilabad sample (ln = 8.705). This regional effect could be one reason why the coefficient of the ST variable is positive and significant. In terms of average consumption, the nine ST-des ignated VSS of the Adilabad sample rank above all but four or the nineteen ST -designated VSS of the other two regional samples. This seems to exert a positive influence on the direction of the relationship, even though the correlation coefficient between ST and ln is negative ( r = -0.294). Several variables that displa y no statistical significance are somewhat surprising. The size of the VSS village (N) was predicted to have a positive relationship with average consumption. In general, the economic activity of a given villa ge should be more robust the greater number of households. Income from forest pr oducts (YFP) was expected to ha ve a negative association with average consumption, because poorer VSS villages are more likely to be dependent on forestry related income than other VSS villages. Su ch an association was observed by Reddy and Chakravarty (1999) in their study of poverty and inequality in U ttar Pradesh, India. Except for the VSS villages of the Visakhapatnam sample, however, income from fo rest products was found to play a relatively small economic role in most of the VSS sampled; perhaps the statistical insignificance of YFP should have b een expected for this reason. It is interesting to note that YAL (inc ome from agriculture) shows no statistical significance in this regression because agriculture generally plays an important role in the local

PAGE 117

117 economies of rural Andhra Pradesh. A possible explanation for this result is that home consumption of agricultural products is predominant in many households (i.e., many households are subsistence-oriented). As such, many of the households that were sampled only produced small surpluses (if any) available to be ma rketed. Thus, many agricu ltural households often showed little or no income. This effect may be enough to effectively blunt any statistical relationship between YAL and average consumption. Consumption Inequality Model The inequality model regresses nineteen expl anatory variables on th e natural log of the Gini coefficient ( ), which is an economic measure of ine quality (in this study, consumption is the basis for measuring the Gini coefficient). Tests for omitted variable bias and heteroskedasticity were statistically rejected; ho wever, the results displayed in Table 5-2 show that this model is almost completely devoi d of any explanatory power. The statistical insignificance of most of the variables is unlikely due to multicollinearity, as this property was only detected for the LT and LT2 variables, whic h (as expected). The only variables found to be statistically significant, in addition to the c onstant term, are TRN (training events) and ADIL (Adilabad dummy). Although the over all F-test rejects the model, these two variables are briefly discussed. TRN (p = 0.036) is significant and negative, indicating lower values of inequality are associated with VSS villages that reported rela tively more training for their VSS members. This may be indicative of CFM havi ng a beneficial effect on VSS equity, but the statistical insignificance of LT and LT2 cast some doubt on su ch an interpretation. Furthermore, joint significance tests for TRN and LT (p = 0.101) an d for TRN, LT, and LT2 (p = 0.199) do not indicate any collectiv e explanatory power for these groupings. The negative sign on ADIL (p =

PAGE 118

118 0.042) means that the VSS village s of the Adilabad sample have less inequality than the VSS sampled in either Visakhapatnam or Chittoor. Du e to the non-significance of the CHIT variable, the latter two samples cannot be distinguished statistically in te rms of inequality. The results of this model are surprising becau se almost every explanatory variable was statistically insignificant. That some variable s would display no impact on inequality was not unexpected due to the complex na ture and interaction of distributional issues; however, it was expected that at least some of the demographic variables and perhaps one of the income variable would register an appreciable impact on inequali ty. In particular, the variables ST, EDU, and YOS were predicted to have a significan t negative relationship with inequality. Poverty-gap Ratio Model The third regression model includes twenty e xplanatory variables regressed on PGR, the poverty-gap ratio. As shown in Equation 48a, PGR is based on per-capita household consumption, and essentially measures the degr ee of poverty for those VSS villages having at least one individual below the poverty line. Thus, the measured poverty-gap ratio will be larger for greater levels of poverty and zero for thos e VSS villages that had no sampled households below the poverty line. Because nearly a thir d (18 of 58 observations) of the VSS villages sampled have no households below the poverty line, their PGR values are equal to zero.16 This means that these 18 observations are censored at zero, which necessita tes the use of Tobit estimation to overcome the bias that censoring cr eates. The estimation results are shown in Table 5-2; ten of the independent variables ha ve statistically significant coefficients The Scheduled Tribes dummy (ST) is statisti cally significant (p = 0. 001) with a negative sign on the coefficient. This means that ST-desig nated VSS are more likely to be less poor than 16 The observed maximum PGR value in the sample is 0.2648 (Table 5-1). The theoretically potential maximum value of PGR is 1, which is essentially equivalent to all individuals in a village having zero consumption.

PAGE 119

119 other VSS with households below the poverty line. As with the ST variable in the average consumption equation, this result is likely due to the presence of other tribal welfare programs. For example, several remote tribal communities that were surveyed in the Visakhapatnam district were being served by an European Union-supporte d project implemented by the Indian branch of CARE, which is an international non -governmental organization (NGO). The NGO variable is also statistically signifi cant (p = 0.012) and negative. This sign was anticipated and suggests that th e involvement of NGOs that provi de forestry assistance to VSS communities contributes to lower levels of poverty. This is somewhat difficult to explain given the non-significance of some of the other forestry -related variables (e.g., LT, YFP, etc.), but may instead reflect an influence that is indirect and not readily observed. YOS is highly statistically significant (p = 0.000) with a negative coefficient. This sign was anticipated, and means that income derive d from sources (mainly employment) other than forestry or agriculture are more important am ong the less poor VSS that have households below the poverty line. The external funding proxy (IHW) is positive a nd statistically significant (p = 0.022). The sign on IHW does not agree with the a priori expectation that external money for VSS works has a positive effect on poverty allevi ation. As with the average c onsumption model, however, this result may simply mean that poorer VSS village s have received greater amounts of assistance money (via wages for VSS works) per household. The coefficient on RRE (relative resource endowment) is positive and statistically significant (p = 0.000), which is contrary to a priori expectations. Thus, hi gher levels of poverty are associated with higher values of per-member VSS forest area. This suggests that larger VSS forest areas are being allotted to the poorer VSS villages, per th e development objectives of the

PAGE 120

120 CFM program. Support for this proposition is fo und by observing the correlation coefficients between average consumption ( ) and the VSS forest area si ze (FA). Although the correlation for the pooled sample ( r = -0.137) is weak, both of the co rrelations of FA with the Adilabad sample ( r = -0.429) and the Visakhapatnam sample ( r = -0.493) are much stronger. Percent forest consumption (PFC) is highl y significant (p = 0.000) and positive, as predicted. This signifies the poverty level of a given VSS village is higher the greater is the forest dependency. In conjunction with the strong statistical significance of PFC in the average consumption equation, this result reinforces the contention that fore st dependency is a good proxy for socio-economic status. DTW (depth to water) is significant (p = 0.026) and negative, which is also contrary to the predicted result. This result sugge sts that higher depths to groundw ater are associated with less poverty among the VSS having at least some house holds below the poverty line. This could be the case if, for a given VSS village, irrigation from deep bore-wells is only possible for households that are economically better off. In general, VSS villages with high DTW values are likely to be relatively more reliant on other so urces of income than income from agriculture; recall that YOS is also statistically significant in this regression, also. The CCA (collective-choice proxy) variable is statis tically significant (p = 0.000) and has a positive sign, which is opposite of the a priori expectation that VSS villages placing an emphasis on forest-dependent households would have a bene ficial influence on pove rty. Like CCA in the average consumption equation, the wrong sign of this variable may only mean that the proxied institution is associated with the poorer VSS v illages for which CFM is designed and oriented towards.

PAGE 121

121 Both location dummies in the PGR regressi on model are statistically significant. The negative signs on ADIL (p = 0.000) and CHIT (p = 0.031) signify that the poorest VSS in the Adilabad and Chittoor samples are notably less poor than the poorest VSS villages of the Visakhapatnam sample (e.g., these four VSS villages are located in the mountainous Araku valley area of northern Visakhapatn am district, near the border w ith the state of Orissa). Forest Quality Change Model With the exception of the inequality equati on, the above models provide a reasonable amount of statistically significant variables that explain the varia tion in observed measures of average consumption and the poverty-gap ratio. Most of these variables ar e of the demographic, economic, or bio-physical type. Only one is an institutional variable (CCA). Despite the statistical significance of CCA (with an unexpect ed coefficient sign), the other institutional variables included in the three regressions had no explanatory power. As these variables either directly describe, or act as a proxy for, various institutional fact ors directly related to CFM in Andhra Pradesh, one may conclude that they exert no influence on average consumption, inequality, or poverty in the sample. It is important to note, however, that the three economic indicators serving as dependent variables are not direct outcome variables in terms of measuring the success of the JFM/CFM program. Such a variable would need to quantify forest productivity change s, or perhaps income derived from forests over time. Therefore, it seems entirely plausible that the institutional variables would be statistically insignificant in the models that explain economic outcomes.17 17 This may also be why some of the statistically significant program-related explanatory variables (e.g., IHW, CCA, RRE) show coefficient sign s that contradict the a priori predictions. The average consumption and PGR regressions may be capturing relationships with these variables that are descriptive in nature, as opposed to relationships that might otherwise be hypothesized (and inferred) as causalespecially if the data were time-series or panel data, instead of cross-section data. For exam ple, it was suggested earlier that poorer VSS villages may have received more external monetary assistance (IHW) in an effort to compensate their relatively lower economic status. This is

PAGE 122

122 Thus, an alternative model with a different depe ndent variable is requi red in order to better analyze the suite of institutional variables listed in Table 5-1. The household survey provides data on forest qu ality that can be used as an outcome, or performance, variable for the JFM/CFM program. Forest quality change (FQC) is likely to be more directly influenced by the program institutions described in Chapter 2 than were the previous dependent variables that measured economic outcomes. Thus, a fourth model was analyzed using many of the same explanatory variables; this equation is based on householdlevel data from which a measure of forest quali ty change under JFM/CFM can be calculated. The dependent variable, FQC, is derived fr om two subjective questions in the household questionnaire that separately i nquired about: (1) the level of fore st quality prior to JFM or CFM in their village, and (2) the current level of forest quality in the forest ar ea being protected by the VSS. Respondents were presente d with responses representing the subjective quality rating of the VSS forest area, which were coded as follows : 1 = good, 2 = slightly degraded, 3 = degraded, and 4 = very degraded. As the same scale was us ed for both the prior and current forest quality levels, an FQC value was obtained for each hous ehold by subtracting the current quality rating from the prior quality rating. The highest possible level of positive forest quality change (i.e., FQC = 3) is recorded if the respondent indicat ed a very degraded pr ior quality and a good current level of forest quality. The potential range for FQC is -3 to 3, although the survey data actually yield a six-category scale that ranges from -2 to 3. An ordered probit model was used for this auxi liary regression because the discrete values of the dependent variable render ordinary least sq uares as a poor choice to model the subjective forest quality change. The ordinal nature of th e dependent variable requires a model structure perhaps why the relationship between IHW and average consumption is negative, when economic theory would suggest a positive impact.

PAGE 123

123 that allows the data to be analyzed by utiliz ing the cumulative normal function to represent the probabilities that each discrete category (e.g., forest quality change) is observed. There are a total of 815 usable observations (only VSS-member households were asked questions regarding CFM), with all 58 VSS villages represented as primary sampling units from which the households were randomly selected. The standard errors of the parameter coefficients were corrected for the effects of clustering at the village-level; this cluster correction simultaneously corrects for any heteroskedasticity that may be present as well (StataCorp, 2005). Thus, the design degrees of freedom for the model is 57, and the effective degrees of freedom for the regression equation is 39 (58 clusters less 19 explanatory variables) instead of 796. The joint test for all coefficients equaling zero is F(19, 39) = 13.12, which has a p-value of 0.000. This indicates that the regression has statistical vali dity, despite the fact that only 8 of the 19 explanatory variables (42%) are statistically significant at p 0.050. The model only predicted slightly more than half of the values correctly. This can be observed in Table 5-3, which shows the joint frequency of the actual FQC data versus the predicted values. The 409 correct predictions are hi ghlighted in bold type. Table 5-3. Actual versus predicted Forest Quality Change (FQC) values. Predicted FQC values Actual FQC -1 0 1 2 3 Total 3 0 0 16 8 1 25 2 0 3 155 28 2 188 1 0 39 338 40 0 417 0 1 40 87 2 0 130 -1 2 35 16 0 0 53 -2 0 0 2 0 0 2 Total 3 117 614 78 3 815 Comparison of the row and column totals indi cate that the predicted values are more concentrated in the FQC = 1 category than ar e the actual values, and less so in the other

PAGE 124

124 categories. In terms of the prediction bias, a ne t of 66 observations moved from higher categories (i.e., FQC = 2 or 3) into Predicted FQC = 1 th an from the lower categories. The percentage of correctly predicted FQC = 1 values is relativel y high (81%) while the correctly predicted FQC = 0 values is relatively low (31% ), although the total predicted (117) is similar to the number of actual FQC = 0 values (130). There are eight (42%) statistica lly significant variables in this regression, including the length of time variable and its square (i.e., LT a nd LT2). The other variables are relative resource endowment (RRE), prior-use value (PUV), microplan reflects forest users (MRFU) dummy, access regulations clearly unders tood (ARCU) dummy, graduated sanctions (GS) dummy, and the Adilabad (ADIL) dummy. Unlike the previous regressions, the estimated parameter coefficients of the ordered probit model cannot dire ctly be used to evaluate the impact of the explanatory variables on the dependent variable.18 Instead, one has to calculate the marginal effects of each independent variable on the proba bility of being selected into each of the response categories (e.g., FQC = 2). Table 5-4 pres ents the marginal effects and p-values for the statistically significant variables listed above (e xcept LT and LT2, which are discussed later), and for five of the six FQC response categories. RRE is negative and statistically significant (p 0.025) for two of the positive FQC categories. This indicates that an increase in the relative size of the VSS forest area (measured in ha per member) will decrease the probability that a VSS village will be sel ected into the FQC = 1 category or be selected into the FQC = 2 categor y (i.e., a one to two category impact to forest quality under JFM/CFM). RRE is also pos itive and statistically significant (p 0.012) in both of the non-positive FQC categories displayed in Table 5-4, which means that an increase in the size 18 For this reason, the parameter coefficients are not presen ted as before. The more meaningful results to present are the marginal effects; these ar e calculated and presented for the statistically significant variables of the regression.

PAGE 125

125 of the VSS forest area per member will increase the probability that a VSS village will be selected into either the FQC = 0 or FQC = -1 cat egory (i.e., forest quality did not change or got worse under JFM/CFM). Taken together, the margin al effects coefficients for RRE support the inference that smaller-sized VSS forest areas per member are more likely to have positive changes in forest quality. This interpretation is plausible considering th e monitoring required for successful management, as well as the restora tive works projects undertaken. Smaller resource endowments allow for relatively more inte nsive monitoring and physical improvements. Table 5-4. Estimated marginal effects and p-values on categor ical probabilities. Variable Prob. [FQC = 3] Prob. [FQC = 2] Prob. [FQC = 1] Prob. [FQC = 0] Prob. [FQC = -1] RRE -0.000879 p = 0.081 -0.012120 p = 0.008 -0.004428 p = 0.025 0.011269 p = 0.011 0.006133 p = 0.012 PUV 0.000002 p = 0.034 0.000034 p = 0.009 0.000012 p = 0.050 -0.000031 p = 0.018 -0.000017 p = 0.009 MRFU 0.009628 p = 0.067 0.137789 p = 0.002 0.078048 p = 0.073 -0.135794 p = 0.010 -0.089142 p = 0.035 ARCU 0.005342 p = 0.058 0.078045 p = 0.002 0.038670 p = 0.007 -0.076127 p = 0.002 -0.045707 p = 0.008 GS -0.006964 p = 0.116 -0.086305 p = 0.004 -0.022402 p = 0.012 0.076009 p = 0.003 0.039506 p = 0.005 ADIL -0.023280 p = 0.048 -0.246004 p = 0.000 -0.085030 p = 0.003 0.214743 p = 0.000 0.138633 p = 0.000 Note: The Probability of FQC = -2 category is not shown b ecause all of the marginal effects were statistically insignificant (p 0.349). PUV is the prior-use value (ex-ante JFM/CF M) of the forest ar ea currently under VSS protection, as reported by each VSS-member household interviewed in the survey. It is an estimate of the total value of pr oducts collected, consumed, and/or sold (by the household) from the VSS forest area in the year prior to its protection under JFM or CFM. This variable controls for the baseline quality and productivity le vel of the forests area being protected and

PAGE 126

126 rehabilitated. PUV is positive a nd statistically significant (p 0.050) for the three positive FQC categories, significant and negative (p 0.018) for the two negative FQC categories. As one would expect, this indicates that the more productive the forest area was prior to VSS protection, the greater is the probability of a positive forest quality change. MRFU is the household-level equivalent of th e CCA variable that was analyzed in the VSS-level regressions. MRFU is a dummy that e quals one if respondents indicate that forestdependent households are being served by the VSS micro-plan. Previously it was shown that CCA has explanatory power in both the averag e consumption and poverty-gap ratio models, although the signs of the coeffici ents were different than pred icted. In the FQC model, the marginal effects coefficient MRFU is positive and statistically significant (p 0.002) only for the FQC = 2 category. The positive marginal eff ects for the other positive categories (i.e., FQC = 3 and FQC = 1) are si gnificant only at p 0.073. Both of the negative FQC categories have marginal effects coefficients that are negative and significant at p 0.035. Thus, there is some evidence to suggest that VSS with management pl ans oriented towards households that are more forest-dependent are more likely to have positive changes in forest quality. The ARCU variable is also a dummy. ARCU is equal to one if the access regulations regarding entry to the VSS ar ea are clear and easy to unders tand, as indicated by household respondents, and zero otherwise. Thus, this is a proxy for one as pect of the clearly defined boundaries design principle. The marginal eff ects coefficients for ARCU are positive and statistically significant (p 0.007) for the FQC =2 and FQC = 1 categories, and negative and statistically significant (p 0.008) for the negative FQC categor ies. This signifies that access awareness in a VSS village increases the probab ility of a positive change in forest quality from JFM/CFM.

PAGE 127

127 The graduated sanctions (GS) variable is negative and statistically significant (p 0.012) for the FQC = 2 and FQC = 1 categories, and positive and significant (p 0.005) for the negative categories in Table 5-4. In theory, the presence of graduated sanctions ought to have a positive impact on the success of JFM/CFM. In practice, however, the graduated sanctions mechanism is largely unobserved by most VSS membersas there ha ve been very few (if any) rules violations by community members in most of the VSS that were surveyed. Repeat offenders seem to be rare, as well. Thus, it is difficult to know exactly what GS represents at the field level; it could perhaps serve as a proxy for the re lative level of complexity or s ophistication of the local VSS institutions. The regional dummy ADIL is negati ve and statistically significant (p 0.048) for all three of the positive FQC categories, and positive and significant (p 0.000) for both of the negative FQC categories shown. This indicates that VSS v illages of the Adilabad sample are more likely to be selected into the negative FQC categor ies. The CHIT dummy was not statistically significant; thus, the Visakhapatnam and Chittoor sa mples are statistically indistinguishable from each other with respect to change in forest quality. Discussion of the Effect of Time under JFM/CFM Aside from the institutional variables, which ar e discussed in further detail in the next chapter (which examines CFM a dherence to the Ostrom design pr inciples), the most important variable analyzed by this study is LT. This variab le represents the length of time that a given VSS has officially practiced JFM or CFM (as registered with the APFD). Paired with its square term (LT2), this variable appears in all f our of the regressions presented aboveand is statistically significant in two of the four. In the average consumption (ln ) equation, the statistically significant signs on the coefficients for LT () and LT2 (+) signify a convex (or U-

PAGE 128

128 shaped) relationship with average consumption; su ch a relationship is indicative of an initial decline in the economic welfare of VSS age-coho rts, followed by an eventual increase as the length of time under JFM/CFM increases. The suggestion, however, that this quadra tic representation of LTand by inference JFM/CFMhas a definitive economic influence on VSS villages in the sample is, by itself, rather weak. This is because the inferen ce of any causal relationship between average consumption and LT must be made with caution, fo r at least two reasons. First, regression results alone give no assurance that a particular rela tionship is causal. Thus, there must be some underlying basis to assert that a statistical association implie s a causal relationship. Second, unlike time-series or panel data, the data at hand are not tracking VSS villages dynamically through time. If they were, there would be a soli d basis on which to infer causality. Given these concerns, the statistical link identifying a consum ptionduration (i.e., ln LT) relationship is either valid or an artifact of the data in which th e association of the variables is more descriptive in its essence. In either case, additional ev idence is required to es tablish a more robust interpretation of the relationshi p between LT and either average consumption or forest quality change. Each will be explained further in turn. Average Consumption Model The imposition of JFM/CFM institutions theore tically places severe restrictions on access and/or use of the VSS forest area that has been designated for restoration. If forest resource benefits are still able to be derived from th e VSS forest area at the inception of the JFM/CFM program in a given village (i.e., LT = 0), then it is plausible that the flow of such benefits will be reduced in the near-term (e.g., LT = 2 to 5) as forest resources are conserved through access restrictions, etc. Once forest productivity has incr eased to a level that permits some sustainable

PAGE 129

129 extraction (e.g., LT > 7), economic benefits ought to increase as forest products are once again collected. Establishment of additional VSS institutions might also directly impact economic activity in the VSS villages if restrictions on gr azing are imposed (which is almost always the case), and/or if forest lands en croached under shifting cultivation are re-established as forest under the protection of the VSS. Both of these examples are important locally. The introduction to this discussion raised th e question of whether or not the inferred economic impact of JFM/CFM (through LT and LT2 ) on the households of the VSS villages sampled was a valid interpretation. By itself, the statistical significance of these two variables in the average consumption model merely provides evidence for such a link, while the signs indicate the general shape of the relationship (e.g., convex). In order to observe the true nature of the link between LT and mean per-capita household consumption, Figure 5-1 plots the predicted values of as a function of LT and while holdi ng the other variables constant. 0 5,000 10,000 15,000 20,000 0246810121416 LT (years) Pred. Mu Figure 5-1. Actual (dots) and predicted (line) in Rupees, versus LT. This graph confirms the convex relationship implied by the coefficient signs, but also indicates an overall negative impact of JFM/CFM on the mean per-capita household

PAGE 130

130 consumption value of the VSS villages in the sample. The initial steep decline in observed for LT = 0 to 4 suggests that the program has an ex tremely adverse impact on local households in the early years as predicted drops nearly 50% by year 4. The marginal impact of the program is zero at 7.35 years and, although there are positiv e marginal impacts beginning thereafter, the values of predicted do not surpass the initial value (at LT = 0) until after year 14. Thus, there seems to be a large opportunity cost of participating in this program.19 Although these results are compelling, there ar e a few caveats to note. First, many of the VSS communities have yet to harvest any timber, while others indicate d they do not wish to harvest any. Depending on the species grown, timb er harvesting would likely occur after ten or more years of forest protecti on (or plantation management), which would help to offset the negative impacts of the early years. Second, any positive externalities associated with the program, such as increased groundw ater recharge and soil cons ervation, are unlikely to be captured in this study since the duration under JFM/CFM is too short in the VSS villages sampled. Therefore, it may be that predicted at higher values of LT are biased downward. Another important factor that potentially aff ects the results can be tested. The consumption of forest products is generally a small fraction of the value of household consumption, usually less than 5%. As is ultimately based on consumpti on values as reported by individual households, the impact of JFM/CFM (through LT ) on mean per-capita household consumption might be obscured if a significant number of households do not consume many forest products. Therefore, it is necessary to examine a subset of the data that only includes those VSS villages that are most dependent on forest products. This was accomplished by running a regression on a 19 Note, however, that the predicted values of outside the range of the actual values are likely to have a high level of uncertainty. Moreover, a large majority of the samp le data (86%) lies between LT = 5 and LT = 9, which will diminish the confidence in the predicted values of for both the low and high values of LT depicted in Figure 5-1.

PAGE 131

131 restricted set of variables for a subset of VSS villages based upon PFC (percent forest consumption) values greater than 4.5 (which is slightly below the full-sample mean of 4.59). LT was statistically significant and positive with a p-value of 0.010 in this regression. The implicit regression equation is: ln = f (N, ST, EDU, Y, IHW, LT, FA, PFC, CCA, ADIL), where Y is an aggregation of the three income va riables (YFP, YAL, YOS). The adjusted-R2 is 0.711 for this specification. In Figure 5-2, the predicted values of from this subset regression are plotted against LT. 0 2,500 5,000 7,500 10,000 024681012 LT (years) Pred. Mu Figure 5-2. Actual (dots) and predicted (line) in Rupees, versus LT: using a subset of VSS where PFC > 4.5 (n = 26). The linear prediction in Figure 5-2 controls for the other variables, and a positive relationship between and the JFM/CFM program can be in ferred. Note that this subset comprised the most forest-dependent VSS villages, and also tends to represent the poorest VSS villages. In addition, the subset sample is comp rised of 18 VSS (69%) that are identified as STdesignated VSS villages. Because a key objective of this program is to increase the opportunities and economic well being of the rural poor, the in stitutional structure of CFM in Andhra Pradesh

PAGE 132

132 has explicitly focused attention on the incorpora tion of Scheduled Tribes. Thus, the results from this subset provide evidence that the most ec onomically disadvantaged VSS are benefiting from JFM/CFM. In contrast to the convex pattern observed in Figure 5-1, the positive relationship between and the JFM/CFM program observed in Figure 52 is more akin to a direct relationship between FQC and LT (discussed in the next secti on). As such, improved forest quality ought to affect local economic well being in the following manner: as degraded forests with little productive capacity are put under protection through JFM/CFM arrang ements, restoration proceeds to the point where the forest resources once again become productive enough to have discernable beneficial effects on th e level of household consumption. Nevertheless, this analysis still calls for some caution. The implementation of JFM/CFM often results in the collection of forest products being shifted away from newly protected areas to nearby forest areas that are unprotected. Some ev idence of this was encountered in the household survey, as well as in field observations made by the principal investigator. The result is that while the output of forest products from th e VSS forest area is diminished, the observed consumption of forest products may not change much. (The same would be true of income derived from forest products.) This may preven t an accurate assessment of the overall average consumptionLT relationship if significant amounts of forest products are in fact derived from non-protected areas. Forest Quality Change Model How might JFM/CFM be positively linked to improved economic well being in the VSS villages sampled in this study? The simple answ er, of course, is through improvements in forest quality and productivity that the CFM program is designed to achieve. Consider Table 5-5,

PAGE 133

133 which presents the probability of selection for each FQC category and the compound marginal effects of LT and LT2 on the observed probabilities displayed. Table 5-5. Probabilities of sel ection and marginal effects of LT. FQC Probability of Selection Compound Marginal Effects Standard Error p-value 3 0.0057 0.57339 0.257 0.029 2 0.1518 1.81799 0.564 0.002 1 0.5849 -0.79101 0.409 0.058 0 0.2008 -1.57306 0.626 0.015 -1 0.0567 2.02957 0.580 0.001 -2 0.0001 -2.05689 0.604 0.001 Total 1.0 0.0 --The desire to link the institutional variables more directly to an outcome variable prompted the inclusion of the FQC regression in the study but it also provides an opportunity to assess forest quality as a function of the JFM/CFM prog ram. As Table 5-5 illustrates, there is a high probability (74%) that a positive category of FQ C will be selected. Moreover, LT and LT2 are also key variables analyzed by the forest quality change regression, and the calculation of their compound marginal effects offers some evidence to support that the leng th of time of JFM/CFM has a positive impact on the change in forest qual ity in VSS forest areas. However, the marginal effects of LT on the probability of FQC = 1 are stat istically insignificant, despite the fact that this category (i.e., the one indicating a one-category im provement in forest qu ality) has the highest number of responses and the highe st probability of selection. Conve rsely, the marginal effects of LT for both of the adjacent categories (i.e., FQC = 2 and FQC = 0) are stat istically significant at p 0.015. Together, these three FQC response categor ies (i.e., FQC = 2, FQC = 1, and FQC = 0) contain 90% of the total FQC responses (cf. Table 5-3).

PAGE 134

134 Figure 5-3 displays predicted probability as a function of LT for these three FQC categories. This graph illustrates why the margin al effects of LT on the probability of FQC = 1 are statistically insignificant. Following an initia l increase in probability in years 0 to 3, the probability declines slightly, from the maximum ( 36%) at year 4, for several years before rising slightly in the final years of the time period depicted. From the maximum (year 4) to the secondary rise (years 10 to 12), the probability remains in a narro w 5% band; this means that LT has little overall impact on the probability of a one-category impr ovement in forest quality. The reason for this is unclear, although it may be due to the large number of the FQC values that are predicted as despite the actual distribution of FQC values (cf. Ta ble 5-3). Another likely contributing factor is that n early three-quarters of the actual FQC values (586 of 815 total observations) correspond to LT values of 7 to 9 years. 0.00 0.15 0.30 0.45 0.60 0.75 024681012 LT (years) Predicted Probability FQC=2 FQC=1 FQC=0 Figure 5-3. Predicted probabilities of FQC values as a function of LT. In contrast, both of the predicted probabiliti es for FQC = 2 and FQC = 0 display a salient relationship with LT. The predicted probability of FQC = 2 increases steadily with LT and reaches a maximum (49%) in year 9 before d eclining somewhat thereafter. The opposite is

PAGE 135

135 observed for the predicted probability of FQC = 0. The initial probability of zero forest change being reported is well over 60%; this value fall s steadily until year 9 wh en it reaches a minimum of 14%, and begins to increase beyon d this point. Note that each of the curves flatten and change direction in the same time period (years 8 to 9), and the point at which th e derivative equals zero is the same (i.e., year 9). This might indicate th at the concentration of the data around this time period is affecting all of the proba bilities, and not just the predicte d probability of FQC = 1. Nevertheless, there is a robust and unequivoc al relationship between LT and the predicted probability of FQC = 2 and the predicted probability of FQC = 0 that lends some credence to the hypothesis that positive forest quali ty change results from the adoption of JFM/CFM. The former relationship (i.e., for predicted probability of FQC = 2 vs. LT) is suggestive of positive forest quality changes initially occurring in response to VSS protecti on, but then the change in quality subsequently declines somewhat for the VSS vill ages that have practi ced JFM/CFM for a longer length of time. This is mirrored by the latter relationship (i.e., predicted probability of FQC = 0 vs. LT): as time progresses it is more likely that forest quality will change, presumably for the better as the frequency data and the probabilities in Table 5-4 indicate. As mentioned previously, the data being modeled do not directly represent a dynamic association through time as time-series or panel data would; however, the relationship between LT and the predicted probabilities of selection (into the positiv e FQC categories) can be thought of as approximating a direct temporal association because forest growth (essentially being proxied by FQC) is itself an increasing function of time.20 20 At least for the time periods considered in the sample (max. LT = 12 years). Generally tree growth is represented as a cubic function of time: initial growth (from planting) is slow but followed by rapid growth during maturation, with the eventual decline in the rate of growth as a plateau is reached. The pe riod of slow early growth may not be as prominent if tree growth is regenerated from coppicing, where stem growth proceeds directly from cut stumps. This is largely the case in th e VSS forest areas surveyed.

PAGE 136

136 Why does the predicted probability of FQC = 2 appear to decline somewhat for higher values of LT (years 10 to 12)? It may be that forest quality change reaches a plateau, with resource use once again increasing; or there may be some downward bias of FQC values in the upper ranges of LT. Recall that the FQC variable is an ordinal and relative measure of forest quality change based on the subj ective perceptions of respondents to the HH questionnaire, with memory recall of the forest quality prior to pr otection a critical component. As such, one can intuitively understand that FQC valuesas as sessed by individuals of a specific VSSwill increase with LT (for VSS age-cohorts) in the early to middle years of the VSS age scale, as CFM institutions and VSS works are initiated to protect and restore the VSS forest area. In addition, early in the program initial small impr ovements will likely be perceived as a great success. Respondents are more likely to have good recall in the short run, when they are not far removed in time from when the VSS forest ar ea was unprotected and degraded. The recall period for respondents is much longer at the highest valu es of LT than it is for those members of VSS villages that have recently adopted CFM. This may cause an upward bias in what the perceived forest quality prior to VSS prot ection was, especially if forest regeneration has been successful. For older VSS, the initial rapid grow th of some species (e.g., eucalyptus21) will have likely slowed somewhat; and if the forest has alr eady grown in in terms of spatial and canopy density, then respondents may unconsciously sk ew the ex-ante JFM/CFM forest quality perception towards forest quality values more indicative of the cumulative change that they currently observe. 21 The principal investigator observed cloned varieties of eucalyptus in the field that ranged anywhere from 12 to 20 feet in height; these trees were 2 years old or less.

PAGE 137

137 What drives the FQCLT relationship? The change in forest quality through time is derived from the participation of VSS members in the protection of the VSS forest area, and in the forest restoration works unde rtaken by the VSS. Although the participation proxy (PRT) is statistically insignificant in the FQC model, a Wald Test indicates that the joint significance of PRT with LT is robust: F(2, 56) = 13.04 (p = 0.000). The joint significance of PRT with both LT and LT2 is also strong: F(3, 55) = 10.91 (p = 0.000). Likewise, the monitoring proxy (FP) is statistically insignificant in the FQC model, but a Wald Test indicates that the joint significance of FP with LT is robust: F(2, 56) = 14.06 (p = 0.000). The same is also true of FP with both LT and LT2 as well: F(3, 55) = 10.11 (p = 0.000). Despite the statistical insignificance of FP in the FQC equation, there are two other institutional variables that are statistically si gnificant with positive coefficient signs that are indicative of an effective impact on forest rehabil itation. ARCU represents one aspect of the first design principle (clearly defi ned boundaries), which are fundame ntally characteristic of successful CPR management schemes (Ostrom, 1990). MRFU proxies another important design principle: collective-choice arrangements (i .e., effective representation and input of stakeholders). Together, the join t statistical significance of these two institutional variables with LT and LT2 is strong: F(4, 54) = 9.02 (p = 0.000). This offers some additional support for the causal relationship between LT and FQC. Summary In general, the analyses presented above o ffer evidence that JFM/CFM has had a positive impact on the forest resource of the VSS sampled, in terms of improving the quality of the VSS forest areas that have been under local prot ection and management. The range of time under management for the VSS sampled is 2 to 12 years. In terms of economic impacts, VSS protection has an adverse impact on the average consumption of the VSS villages sampled, at least for the

PAGE 138

138 full sample of 58 VSS that were pooled for anal ysis. Specifically, it app ears that the opportunity cost of the program is high as it relates to bot h the short run economic impact on consumption, and the long run impact on the su stainability of the CFM program in Andhra Pradesh in terms of the participation and involvement of the stakeholders. However, the analysis of a subset of the da ta that represents more forest-dependent households provides evidence to the contrary by showing a positive relationship between the length of time of VSS protecti on (i.e., the LT variable) and aver age consumption. As the main target populations of the JFM/CFM program are the poorer, more forest-dependent VSS villages and households (especially Scheduled Tribes), this result implies that the program is at least having some success in benefiting the people for whom it was designed. Caution is advocated when extrapolating the results, however, due to the limited number of observations in this subset. Further investigation is require d to confirm these results and test for additional hypotheses. Finally, there also is evidence that some of th e CFM institutional variables analyzed in the FQC regression have a positive impact on rehabilitati on of VSS forest lands protected by the VSS sampled. A more detailed investigation of the institutional variables is addressed in the next chapter, with additional data collected in the survey augmenting the regr ession results described in the previous sections.

PAGE 139

139 CHAPTER 6 ANALYSIS OF CFM INSTITUTIONS Introduction The long-term efficacy of CFM as a resource management paradigm is dependent upon several factors. Such factors include the active interest and continued engagement of the VSS, their adherence to Ostroms critical design principles, outside assistance from NGOs, and the effective facilitation of the program by the A ndhra Pradesh Forest Department (APFD). The degree to which these factors are realized wi ll largely determine the spatial and temporal sustainability of the program, and the effective r ealization of the program goals: improved forest cover and productivity, improve d livelihoods and equity, and reduced rural poverty. As a unified program for forest restoration and community development, JFM/CFM is still in a nascent stage of developmen t: in Andhra Pradesh (and most of India) the program is still only in its first generation. This is exemplified by th e fact that official legislation has evolved, in the first decade of JFM in Andhra Pradesh, to the extent where more direct control has been given to local stakeholders under the new moniker CFM. In addition, the first rush of demand for the implementation and registration of VSS has b een impressive, but it seems likely that the growth rate of VSS formation has peaked and annual additions will be relatively modest. The challenge moving into the future is to retain the VSS that are currently practicing CFM by making sure that their efforts are successful. Th is is because the long-term viability of the CFM program is dependent upon the ability of the VS S to sustain their efforts in the absence of external financial support. For example, Khare et al. (2000) state that s upport for JFM in West Bengal waned following the completion of the World Bank project that was providing support. One of the salient elements of this resear ch is to address the degree to which the institutions of CFM (having succeeded JFM) fo llow the design principles that Ostrom (1990)

PAGE 140

140 devised in her seminal work. The reason for this is that these principles were specifically distilled from case studies of different CPR management schemes that have endured through timein some cases for hundreds of years. Thus, they re present successful institutions common to a diverse array of resource situ ations and are meant to serve as guiding principles in the development of new CPR manageme nt programs. By incorporating the design principles into the institutional set of CFM, the long-term success of this endeavorwhile not assuredis at least more likely. On the other hand, non-adherence to these design principles will likely limit the efficacy of any CPR management scheme (not just JFM/CFM) because of their demonstrated theoretical and empirical strength (Ostrom, 1990). Some of the ke y features of JFM as originally devised, however, violate these design principles. For in stance, self-initiated institutions for the sustainable management of a CPR are the ideal circumstance for a group of local stakeholders who actually own the resource. This is not the case in India, which is an important distinction that may also limit the efficacy of CFM in the future. Unlike the self -initiated ideal, the Government of India (GOI) is responsible for introducing an overarching institutional set that defines the forest resource mana gement paradigm while retaining ownership of locally managed forest lands. As such, the long-term security of rights may perhaps be considered questionable and this may eventually impact participation ra tes, whatever amount of trust there is in the government, and ultimately the viability of collective action through the CFM program. In addition, although the original legislation for JFM spelled out the conditional benefits that VSS are entitled to, it did no t guarantee rights to the forest pr oduce or the forest land. While subsequent revisions by the Gove rnment of Andhra Pradesh (GOAP) seem to strengthen the rights of the VSS to access forest produce (and expa nd the range of forest products), it is still

PAGE 141

141 clear that such rights are condi tional on the discharge of the VSS duties and responsibilities. This is one of the main limitations of this paradigm, which is underscored by the fact that forestland tenure remains with the government. Under the proper legal and institutional e nvironment, community-level organization directed at sustainable resour ce management can be realized. Ostrom (1990) defined eight critical institutional elements that are require d (at a minimum) for transforming an open access resource into a CPR. As CFM is undertaken to ch eck deforestation and rehabilitate forest lands through a similar process, it is necessary to revis it each element and evaluate the institutions of CFM against them using the empirical results from the regression analysis in Chapter 5 and the data described in Chapter 4. DP 1: Clearly Defined Boundaries According to Ostrom (1990), the CPR being mana ged must be clearly defined in terms of the spatial boundaries. In addition, the definition of rights of access and us e by a specific set of stakeholders must be clearly de fined. Together these components reduce uncertainties regarding who has legitimate access to the CPR, and clearly establishes the exclusion of non-stakeholders from accessing the defined area. Although these instit utions would ideally be self-initiated by the local stakeholders, the APFD and GOAP are responsible for proscribing the overarching institutional set that includes these components. For example, the GOAP orders on CFM define the conditions under which a VSS is formed and, to some extent, membership eligibility (e.g., all Scheduled Tribes households ar e automatically granted member ship in the VSS). To fully analyze this design principle, the spatial bounda ry awareness, awareness of access regulations, and the general awareness of institutions are examined in turn.

PAGE 142

142 Spatial Boundary Awareness In actual practice, clearly defined spatial bounda ries are not a simple issue for JFM/CFM. As forest areas allotted for protection are ultimately public land, th e APFD has the greatest amount of knowledge concerning the spatial boundaries of the VSS forest area in terms of descriptions of the north, east, west, and south (NEWS) boundaries They have descriptions of these boundaries in both documents and maps that define the location of boundaries in reference to physical landmarks and administrative territory In some cases, a physical characteristic makes one or more of the boundaries obvious (e.g., the cr est of a ridge or a paved road). However, a typical boundary might be listed as Reserve Fo rest Compartment 502, which may be important for the APFD but likely has little meaning to a VSS member. Thus, the main problem with respect to this as pect of DP 1 is that the VSS members need to be well informed of the bounda ries of their VSS forest area because they are the primary stakeholders responsible for resource ma nagement. Although the NEWS boundaries are normally identified in the VSS micro-plan, most VSS members are unlik ely to have read it (notwithstanding the fact that ma ny rural people are illiterate). Sometimes a map is provided to the VSS, either in the micro-plan or displayed on a wall in the village. In practice, however, the VSS members often seem to have little or no knowledge of some or all of the boundaries, especially the administrativetype boundaries that do not co rrespond to easily recognizable physical landmarks. Data from the household questionnaire (A ppendix B) support the anecdotal evidence described above: respondents were asked, Do you know the exact boundarie s of the VSS forest area? (Question 15, Forest Resources Module, Pa rt A). This question was scored yes if the respondent was judged credible in the identification of two or more of the VSS boundaries, and

PAGE 143

143 no otherwise.22 Despite a high level of uncertainty invol ved with the reporting of the data, only 24% (202 of 839 respondents) of VSS-member hous eholds were determined to have boundary awareness (BA). The BA variable was statistically insignifican t in the FQC regression, which attempts to explain variation in the change in forest quality under VSS. The variable VBA ranks the relative boundary awareness of the VSS villages. It is calculated, for a given VSS, as the percent of VSS households that indica ted a general level of boundary awareness in the household-level ques tionnaire. Table 6.1 shows the frequency distributions of this variable. Table 6-1. VSS boundary awareness. VBA Frequency Relative Frequency Cumulative Frequency 0 to 0.10 11 18.97 18.97 0.101 to 0.20 16 27.59 46.55 0.201 to 0.30 15 25.86 72.41 0.301 to 0.40 7 12.07 84.48 0.401 to 0.50 8 13.79 98.28 0.501 to 0.60 1 1.72 100.0 Total 58 100.0 -Most of the VSS villages (42 of 58) had a VBA va lue of less than or equal to 0.30 (i.e., 30%); only one VSS village, K. Boddapadu (# 206), had a majority of households that were deemed boundary aware (VBA = 0.579). This is perhaps due to the fact that a reservoir borders their VSS forest area, and a boat is required to acce ss their site. Overall, VBA was not found to explain poverty, as it was statisti cally insignificant in the poverty-gap ratio (PGR) regression. 22 Note, however, that the response to this question is po tentially biased towards the no answer for two reasons. First, the presence of the administrativ e-type boundaries made it impossible to evaluate the respondents knowledge of all NEWS boundaries. Second, the principal investig ator and the interview teams often lacked NEWS data themselves; therefore the interviewer had to subjectively de termine if the respondent knew at least some the VSS forest area boundaries.

PAGE 144

144 Awareness of Access Regulations The rights of access and use by a specific set of stakeholders are meant to define the group of individuals that are active as forest protectors or resource users. Definition and knowledge of these rights, and who they pertain to, strengthen the control and monitoring of the CPR. This is one institution with which the VSS has some degree of latitude to establish. While all of the VSS that were sampled have banned the felling of tree s in their VSS forest areas (and most have bans against the grazing of animals), other types of acc ess restrictions or prohibitions vary according to the local institutions devised. For example, a frequent rule observed is a prohibition against carrying cutting tools in the forest area; two VSS have banned smoking in the forest, etc. This aspect of DP 1 is perhaps the most important based on th e relative length of discussion by Ostrom (1990) on this topic. Access restrictions th at reserve benefits for members only are an important institutional incentive that helps to strengthen participation by explicitly discouraging the free-riding of non-members. To understand this aspect of the first design principle, Question 25 in Part A of the Fore st Resources Module inquired about this access awareness by asking: Regulations concerning entr y to [the] VSS forest area, are they clear and easy to understand? (HH questionnaire, Append ix B) An overwhelming majority (82%) answered affirmatively, suggesting that at least one of the first steps toward effective collective action is being followed. This question also form s the basis for the dummy variable ARCU that was analyzed in the forest quality change (FQC) re gression. The marginal effects of this variable are positive and statistically significant (p 0.007) for the FQC = 2 or FQC =1 categories. This indicates that access awareness in a VSS village increases the probability of positive forest quality changes resulting from protection and restoration.

PAGE 145

145 As with boundary awareness, observations in th e field must be considered when evaluating this data. Three factors must be addressed: Sc heduled Tribes (ST) membership, access of nonmembers, and paid wages for VSS works. Access and use rights for a specific set of stakeholders are not as critical in smaller-si zed VSS villages where every household in the village belongs to the VSS. This situation is most likely to occur in ST-designated VSS, not only because the small VSS villages tend to be tribal villages, but because CFM legislation automatically grants VSS membership to ST households. Thus, all ST house holds have permission to access the forest and use forest products that the VSS allows collectio n of, and everyone is likel y to know this fact. A breakdown of the ARCU variable by cas te is presented in Table 6-2. Table 6-2. Access awareness of VSS-member households by caste. Caste Designation ARCU Frequency Relative Frequency ST Households No 93 18.83 Yes 401 81.17 Total 494 100.0 SC Households No 17 19.54 Yes 70 80.46 Total 87 100.0 BC Households No 28 19.18 Yes 118 80.82 Total 146 100.0 OC Households No 11 9.48 Yes 105 90.52 Total 116 100.0 The relative frequency of access awareness by the Scheduled Tribes (ST), Scheduled Castes (SC), and Backward Castes (BC) households (80% to 81%) are almost identical to those

PAGE 146

146 observed in the pooled sample (82% ). The only notable difference is with the Other Castes (OC) households, which indicate a higher awareness of the access restrictions ( 91%). Thus, the ARCU variable is not unduly influenced by the au tomatic VSS membership of ST households. The rights of access and use by a specific set of stakeholders are more critical in larger VSS villages that have a fair amount of non-VSS me mbers. It was observed that, while all of the VSS villages in the survey exclude village outsi ders from accessing their VSS forest area, most seem to allow non-members to access their protected forest areawhich may undermine the commitment of the participating VSS members and erode their future efforts toward CFM. Only one VSS sampled, Kancherabai (# 120), explicitly reported the prohibition of non-members in the VSS forest area. Moreover, their rules stated that only VSS members were allowed to collect deadwood for fuelwood, and/or coppiced wood fo r construction. In terms of this design principle, Kancherabai VSS seems to be a pproaching the idealindeed, the principal investigator observed that this VSS had good leadership, was well-endowed with forest resources (especially bamboo), and functioned very well in terms of forest protection, management, and remunerative production (e.g., selling bamboo). Undoubtedly there are social reasons for allowing access to VSS non-members of the village, but in practice this likely weakens th e VSS in the long-run for the reasons presented above. Such adverse impacts might not be ma nifested immediately, however, because the benefits and costs of protection are not always apparent, especially in the early years of JFM/CFM programs. The benefits of forest prot ection and regeneration will often take several years to be realized, while the labor costs are bein g subsidized in the initial stages of the APCFM project.

PAGE 147

147 Since the wages paid for VSS works are derived from external funding23 from the government and/or the World Bank, there is a clear disconnect be tween the cost-benefit incentive structure of the institutions and the rights of access of individuals. It is possible that problems will arise when the external funding for these wo rks is no longer available. In most of the CPR management systems documented by Ostrom (199 0), individuals or households contribute unpaid labor to maintain the CPReither through works and/or monitoring, which entitles them to the benefits of the resource. For the VSS of Andhra Pradesh, future maintenance works will probably involve mostly the coppicing and pruning of trees and the maintenance of fire-breaks. Although such work is not as physically grueling as the soil-moisture conservation (SMC) works (e.g., digging of contour trenches), it will most li kely have to come from the unpaid labor of VSS members, which will reinforce the need to clearly define the rights of acce ss to the forest and its produce. This is especially so since most VSS currently have provisions that allow non-members to be employed for their works projects, a lthough many VSS do give first preference to members. General Awareness of Institutions A variable that examines the first DP in terms of the general awareness of VSS institutions and parameters is labeled HGAI. This variable is an additive index having a theoretical scale of 0 to 7, and is used as a proxy for access awareness in order to augment the previous variables that proxy for DP 1. HGAI measures the level of aw areness that VSS-member households have vis-vis two institutions and two quantitative paramete rs of the VSS. The evaluation criteria include: the type of process used to select the VSS Management Committee (ballot or consensus), the 23 The impending termination of th e funds for these works has necess itated an emphasis on the economic sustainability of the VSS through livelihood augmentation and development of alternate livelihood opportunities. This topic will be discussed in more detail in the next chapter.

PAGE 148

148 length of the Management Committees term (in y ears), the size of the allotted VSS forest area (in hectares), and the year of VSS establishm ent. Household responses are compared against objective data for these four criteria, and scored according to the accuracy of the responses. The distribution of HGAI scores for each VSSmember household is shown in Table 6-3. In general, the respondents have a rather low leve l of general awareness of the VSS, in terms of the criteria evaluated. More than half scored a va lue of 2 or less, meaning that these households have little or no knowledge of the VSS beyond how the VSS Management Committee is elected (because 92% of respondents knew the MC election process). Less than 3% of respondents scored a high level of knowledge about their VSS (GAI 5); of these, approximately 65% are current or former members of the Management Committee. Table 6-3. HGAI for V SS-member households. HGAI Scale Frequency Relative Frequency Cumulative Frequency 0 29 3.24 3.24 1 357 39.84 43.08 2 100 11.16 54.24 3 220 24.55 78.79 4 80 8.93 87.72 5 85 9.49 97.21 6 10 1.12 98.33 7 15 1.67 100.0 Total 853 100.0 -The HGAI index was included in the FQC regr ession, and the parameter coefficient was found to be statistically insignificant. Mean HGAI values for each VSS village were also calculated in order to create the variable GAI. This variable was included in the average consumption model but was statistically insignificant as well.

PAGE 149

149 DP 2: Congruence between Rules and Local Conditions This design principle recognizes the importan ce that local institutions must accommodate the specific characteristics of a given CPR, which include (but are not limited to) the heterogeneity of resource endowments and soci al circumstances. Ostom stresses that the implementation of well-tailored appropriati on and provision rules will help enable CPR management systems to flourish and endure (199 0). Even amongst separate user-groups of the same geographic region, variabil ity must be accounted for by adap ting the local institutions as necessary. This DP is difficult to observe without an extended amount of time to study each VSS; as such, isolating the factors that reflect this DP is more conduciv e to a case study format. Because of the limited time spent in each VSS, only anecdotal evidence is presented. In general, and to the extent that this DP could be assessed in a short interview, it appears that the VSS sampled conform fairly well to this DP. Explicit exampl es are very limited, but a few are notable. For example, K. Boddapadu VSS (# 204) has proh ibited grazing in part because a reservoir borders their VSS forest area. Other VSS v illages such as Kunnempudi VSS (#202) have restricted or prohibited the tr aditional practic e of shifting cultivation known locally as podu. As mentioned previously, two VSS have explicit bans against smoki ng in the forest. In addition, while not pertaining directly to a rule or regulation, Puttiguda VSS (# 119) maintains a tree inventory that includes 760 gum-pr oducing trees in their VSS forest area. This reflects a sitespecific organizational response pertaining to the management of an important local NTFP resource. DP 3: Collective-choice Arrangements Functionally effective institutions must be re sponsive to the individuals they serve by giving a voice to all stakeholde rs, and by giving them the opport unity to participate in the

PAGE 150

150 modification of the institutions. Much like th e first design principle (i.e., clearly defined boundaries), collective-c hoice arrangements are fundamenta lly important for the long-term viability and success of CPR management scheme s. This is because the success of collective action is based on the active and effective partic ipation of the stakeholde rs. Their interests and views must not only be heard, but actively incorporated into th e process in order to ensure institutional flexibility, and (ideally) promote equitable re presentation and the equitable distribution of benefits deri ved from collective action. For the VSS of Andhra Pradesh, this design principle is particularly re levant given that local stakeholders can set some of the locallevel CFM institutions (e.g., rules like access restrictions). It not only allows the VSS some flex ibility to tailor their loca l rules to their CPR (as much as possible, at least theo retically), but also provides th e opportunity for all members to participate in the direction of resource manageme nt. The latter of these is extremely important and was examined by focusing on the extent to which the VSS Management Committee adequately represents the members of the VSS General Body, if the VSS micro-plan addresses the interests of the poorest households, and if participation in the VSS is equitable. The respondent in each VSS-member house hold was asked: Do you think [the] VSS Managing Committee considers views, needs of ma jority of its VSS members? (Question 8, Forest Resources Module, Part A). The res ponse was strong: 88.9% (746 of 839 households) answered yes to the query. Th is is a positive finding, of cour se, because it indicates that the group of stakeholders entrusted w ith the day-to-day management responsibilities of the VSS act in accordance with the spirit and ideals of CFM. These data also offer support that the theoretical design principle is being adhered to, as well.

PAGE 151

151 In addition, each VSS-member household was asked the following question: In practice, [the] VSS micro-plan reflects interests of whic h section of society? (Question 9, Forest Resources Module, Part A). Res pondents were given four possibl e answers to choose from; both the response choices and the fre quency distribution of responses are shown below in Table 6-4. More than two-thirds of the respondents believe that forest dependent households are being served by the VSS micro-plan, a lthough nearly 20% of the responde nts believe that local elites are favored. The overall responses to this questio n and the previous question seem to support the presence of collective-choice arrangements, which means that the VSS Management Committees are responsive to the stakeholders and the micro-pl an reflects the interests of forest dependent households. Table 6-4. Response frequency indicating that the VSS Micro-pl an reflects the interests of: Code Response Categories Frequency Relative Frequency Cumulative Frequency 1 Forest Dependent Households 582 68.23 68.23 2 Other Households 63 7.39 75.62 3 Local Elites 167 19.58 95.19 4 Forest Department 41 4.81 100.0 Total 853 100.0 -In order to represent collective-choice ar rangements in the econometric equations, the frequency data of Table 6-4 was used to develo p a dummy variable. The variable MRFU = 1 if the VSS micro-plan represents the interests of forest-dependent households, and MRFU = 0 otherwise. Analyzed in the FQC equation, the marginal effects of MR FU are statistically significant (p = 0.002) with a positive sign for the FQC = 2 category. This signifies that the VSS that are oriented towards forest-dependent house holds have a higher probability of positive forest quality change. This is an important finding, as it offers some statistical evidence that the third

PAGE 152

152 design principle is effectively represented in the CFM institutions of the VSS sampled by this study. The variable CCA proxies for collective-choice arrangements in all three of the VSS-level equations; it is a ratio measuring the percent of households in a gi ven VSS that indicate the VSS micro-plan reflects the interests of forest depe ndent households. CCA was statistically significant in two of the three regressions that attempted to capture variation in economic indicators of social welfare (i.e., average consumption and th e poverty gap), but the coefficient signs were opposite of the a priori expectation in both cases. The interp retation of such results seems to indicate that VSS villages that place an emphasis on forest-dependent households have a deleterious influence on local ec onomic well-being and poverty, but that could be an incorrect conclusion. For example, the causality implied could be a result of the fact that CCA is measuring the degree to which the VSS micro-plan is oriented towards the poorest communities, which are typically the most fo rest dependent. In a ddition, both the average consumption and poverty-gap variables are not outcome variable s in terms of directly measuring JFM/CFM success. Conflicting interests are always present in a co mmunity and, as such, the potential exists for rural elites (or other interest groups) to co -opt development schemes such as JFM/CFM for their own benefit. To further examine whether DP 3 is well represented by the VSS of the sample, data was also collected as to the equita ble representation of stak eholders in terms of electing the VSS Managing Committee. The re spondent in each VSS-member household was asked: In general, what do you think about the way [the] VSS Management Committee elections are held? (Question 13, Forest Resources Module, Part A). In Table 6-5, the five response categories provided to respondents are shown, along with the frequency distribution of

PAGE 153

153 responses. The household data suggest that consci entious representation in the community is the norm: 82% of VSS-member households report ed that elections for the VSS Managing Committee are free and fair. Table 6-5. Response frequency regarding elections for the VSS Managing Committee. Code Response Categories Frequency Relative Frequency Cumulative Frequency 1 Free and Fair 698 81.92 81.92 2 Dominated by Local Elites 86 10.09 92.02 3 Dominated by Interest Group 48 5.63 97.65 4 Other (specify) 0 0 97.65 99 Dont Know 20 2.35 100.0 Total 852 100.0 -DP 4: Monitoring Stakeholder compliance with th e appropriation rules is necessary to achieve effective and sustainable CPR management, and this can only be guaranteed through effective monitoring. Monitoring not only ensures stakeholder compliance, but also helps to en force the exclusion of non-stakeholders. Ostrom (1990) discusses how the CPR management systems profiled in her book are not dependent upon external authorities for the monitori ng and enforcement of the CPR institutions. Monitors are accountable to the stakeholders, and are fr equently the resource appropriators themselves. This design principl e is a vital component of the CPR management system, and this study tried to determine if regu lar monitoring is undertaken by the VSS. Note that the CFM institutions as legislated by th e state do not mandate formal patrols by VSS members; these institutions transfer the monito ring responsibilities to the VSS and, as such, only implicitly refer to monitoring of the forest for access violations.

PAGE 154

154 The FP variable is a dummy variable that identi fies the presence of a formal patrol and/or a watchman, and was included in two of the VSS-level regressions (the consumption inequality and poverty-gap ratio models). It was also included in the FQC re gression, by assuming the same value for all households of a given VSS. FP was not statistically signif icant in any of the regressions, which implies that this variab le has no explanatory power on consumption inequality, poverty, or a ch ange in forest quality. The data for this variable indi cate that only 38% of the samp le (22 VSS) organized formal patrols or employed a watchman. Five of these VSS had at least one watchman, and some VSS employed a watchman in addition to patrols by members. The rest engaged in informal monitoring by either casually watching the VSS forest area (often it is near their agricultural land), or by incidentally monitoring while they are in the forest for other reasons. Some VSS reported that they monitor informally because ther e is only one (or a few) access points into their VSS forest area and these can be readily observe d. While these efforts represent a low-cost method of monitoring, they may not be effective. Of all the information collected, the data on FP is perhaps the most surprising. It cannot be overemphasized how important monitoring is to effective CPR management. A formal patrol and/or a watchman should be a mainstay in e ach VSS village according to the fourth design principle, regardless of the proximity to agricu ltural lands. Most VSS fore st areas are too big for that approach, and effective deterrence means incr easing the odds of detecti ng violations. In fact, the PI uncovered at least two seri ous violations on impromptu fiel d visits to VSS forest areas: timber theft from the Puttiguda VSS forest, and illeg al goat grazing in the Rampur VSS forest. In both cases, the offenders were outsiders from other communities. In addition, the VSS in both cases reported that they engage in monitoring activities. Puttiguda VSS (# 119) has organized

PAGE 155

155 formal patrols that are comprised of VSS member s split into four groups of five members each. Data from the VSS interview shows that Ra mpur VSS (# 106) reported having a watchman, although the PI did not observe a watchman dur ing the impromptu tour of their VSS area. In general, approximately 50% of the VSS reported some type of violation had been committed, usually the smuggling of timber or d eadwood. Most offenders appear to be people from outside the community; in other words, th e general compliance rate within the population of the VSS village seems to be very high. This anecdotal evidence is supported by the fact that 85.5% of VSS-member households thought that th e majority of VSS members follow the rules and regulations. DP 5: Graduated Sanctions The fifth design principle has important imp lications for the sustainability of a CPR management regime, because circumstances will ev entually arise where stakeholders themselves are tempted to cheat. The consequences of rules violations should be designed to act as effective deterrents without being so draconian as to encourage non-compliance. For example, the first rule violation could result in a modest penalty to remind the offender of the importance that compliance serves to the greater public good. Re peat offenders, however, must face increasingly severe penalties to discourage wanton fr ee-riding according to Ostrom (1990). The CFM institutions as legislated by the state do not explicitly call for gr aduated sanctions but, similar to monitoring (DP 4), this design principle can perhaps be observed as implicit within the institutional framework. The presence of graduated sanctions in a given VSS was measured by the GS dummy variable. Of the VSS sampled, 62% (38 VSS) re ported having this penalty structure. When analyzed in both the poverty-gap ratio (PGR) and forest quality change (FQC) models, this variable was statistically significant in only the FQC regression. However, the marginal effects

PAGE 156

156 of this variable are negative (p 0.012) for two of the positive forest quality change categories (FQC = 2 and FQC = 1). This contradicts the a priori expectation that graduated sanction should have a beneficial influence on forest quality. As stated above, compliance within the VSS seems to be very high with most violations appeari ng to occur from offenders outside the community. This means that there have been few opportunities for the graduated sanctions mechanism to be observed by most members of the VSS that were surveyed and is, therefore, difficult to know what kind of influence this design principle has on the stakeholders themselves. In practical and statistical terms, this would mean that no causative effect is expected. It is possible that a purely descriptive association is being mistaken for a casual relationship. If so, it is not readily apparent sin ce there is no obvious desc riptive reason why this variable should be associated negatively with a change in forest quality Given the likelihood that the presence of graduated sanctions is not wi dely known by the members of the VSS, it seems plausible that GS might be more representative of some other property or pr ocess that is related. In such a case GS would be indirectly acting as a proxy for something else. The application of the graduated sanctions mechanism in many of the VSS sampled is also disconnected from theory somewhat. Roughly half of the VSS repor ting graduated sanctions indicated that, for second offences, offenders would be handed over to the APFD instead of facing locally-based sanctions. DP 6: Conflict-resolution Mechanisms According to Ostrom (1990), conflict resolution mechanisms that are characterized by low transactions costs are a necessary aspect of an effective CPR management scheme. To ensure both the fairness and the continuity of the institutions, each stakeholder must have recourse to a forum established for the purpose of dispute resolu tion. As any set of rules and regulations is

PAGE 157

157 subject to different interpretation by different individuals, and because all organizations are bound to have some free-riders, it is important th at a forum exist for the dispensation of justice and punishment to those accused of non-compliance. As with monitoring and graduated sanctions, an external source of governan ce is not required for this function. Again, the fact that most of the rule violations were not from the VSS villages sampled means that this design principle was difficult to assess. As compliance with the VSS institutions appears to be rather high am ongst community members, there is ostensibly little need for a dispute resolution forum. Indeed, preliminary fi eld interviews with VSS Management Committee and General Body members indicated that such matte rs would generally be referred to the Forest Department. Because of the socio-cultural cohesi veness of rural India, especially in the Scheduled Tribes communities, the absence of a VSS-organized disput e resolution forum may not be a critical loss in terms of VSS e ffectiveness in many of the smaller VSS. DP 7: Minimal Recognition of Rights to Organize Effective CPR management regimes need an over-arching governmental policy framework that facilitates, or even encourages, community -level institutions capab le of natural resource management. As discussed previously, the legislation enabling CF M provides the broad framework for local forest protection institut ions. The most basic requirement is for the government to recognize the rights of local peopl e to devise their own institutions regarding CPRs; to this narrow extent, CFM in Andhra Pr adesh adheres to this design principle. However, the distinction between this applie d case and the theoretical case described by Ostrom is that CFM goes beyond enabling: it act ually defines many of the key institutions. Moreover, a huge distinction between theory and reality is that land te nure remains with the governmentalthough this does not necessarily render CFM ineffectual. Nevertheless, the spirit of this design principle as applied to CFM is in the recognition of VSS autonomy to manage their

PAGE 158

158 allotted forest area. Thus, VSS autonomy was us ed as a proxy for this design principle. In addition, the general tenor of the relationship be tween the VSS and the APFD officer with whom they normally interact was also used to gauge the overall influence of the government in this study. The VSS questionnaire asked the VSS Chai rman or the Vice-Chairman the following question: Does the Forest Department give the Executive Committee24 enough autonomy to manage the VSS forest area in a way that best serves the community? Ninety-three percent answered affirmatively. A similar question wa s included in the HH que stionnaire; 765 of 853 VSS-member households (89.7%) indicated th at the VSS receives enough autonomy from the APFD. The variable PG (perception of the government) was used in the FQC regression to control for this design principle. This variable se rves to proxy for the in terference by external government agencies (i.e., the APFD). Thus, PG measures government interaction with the VSS by qualitatively rating the relationship that th e VSS Management Committee has with the APFD field officer with whom they work most closely. PG is an ordinal subjective ranking made by the VSS Chairman or the Vice-Chairman when asked to describe this relationship. The rating scale is presented in Table 6-6, along with the distribution of the responses. An overwhelming majority of VSS (84%) indica te that they have a good or very good relationship with their APFD field officer. The re gression results were statistically insignificant, however, indicating that PG has no explanatory power on forest quality change. Nevertheless, the data describing the perception of autonomy suggest s that the spirit of th is design principle is 24 The VSS Management Committee is often referred to as the VSS Executive Committee, which was the term used in the questionnaires.

PAGE 159

159 adhered to rather well, even though it does not di rectly affect the measures of economic welfare and forest quality change as defined in this study. Table 6-6. Relationship between the VSS Mana gement Committee and their APFD field officer. Scale Response Categories Frequency Relative Frequency Cumulative Frequency 1 Very Good (Officer really supports VSS efforts) 14 24.14 24.14 2 Good (Officer is helpful) 35 60.34 84.48 3 Fair (or indifferent) 6 10.34 94.83 4 Bad (There are problems with this officer) 3 5.17 100.0 Total 58 100.0 -Summary The design principles (DPs) for successful common-pool resource management regimes (Ostrom, 1990) offer a benchmark against which the CFM institutions of Andhra Pradesh can be examined. Quantitative data were used to assess the following five design principles: DP 1: Clearly defined boundaries. DP 3: Collective-choice arrangements. DP 4: Monitoring. DP 5: Graduated sanctions. DP 7: Minimal rights to organize. Overall, this set of design principles is represented throughout the sample of VSS villages surveyed as a whole, but the presence of each individual design principle varies among the VSS sample. Moreover, the degree to which a given CFM institution (as implemented at the community-level) adheres to its counterpart design principle is also variable, and in some cases clearly deficient (e.g., monitoring). The implications of the specific findings will be discussed in the following chapter, and recommendations will be offered in terms of how to address these deficiencies.

PAGE 160

160 CHAPTER 7 CONCLUSIONS, POLICY IMPLICATIONS, AND FUTURE WORK Overview This last chapter contains four sections. The first section presents conclusions drawn from the analytical results of the regression anal ysis, as well as the results derived from the institutional analysis discussed in the previous chapter. In particular, the impact of the JFM/CFM program is assessed using the leng th of time (LT) variable and the degree to which the Ostrom design principles are incorporat ed into the institutions of the CFM program being implemented in Andhra Pradesh. The second section discusses the broader policy implications of this study, especially as it relates to the ab ility of this program to alleviat e poverty and remain sustainable in the long run in Andhra Pradesh and elsewhere. Some recommendations are also provided that are designed to strengthen the program and its instit utions. The third section describes some of the limitations of this study. The fourth and final se ction discusses some planned future work, and other areas of focus that are wort h investigating further. Summarized Conclusions Regression Analysis The first of the models to examine any impr ovements in the economic indicators of social welfare due to JFM/CFM attempted to explain vari ations in mean per-capita consumption values (average consumption) at the VSS-level. Estim ation results indicate that this regression performed fairly well in terms of explanatory po wer. Especially notewor thy are the statistical significance of LT () and LT2 (+) a nd the direction of their signs: the coefficients signs signify a U-shaped predicted relationship with averag e consumption. The graph of this predicted relationship shows that the ove rall impact of JFM/CFM (as pr oxied by the duration under this relatively new forest management regime) on av erage consumption is adverse because of the

PAGE 161

161 steep decline in predicted av erage consumption in the initi al years following adoption of JFM/CFM. Only after 14 years does the predicted level of average consumption increase to the original value at year zero. This suggests th at VSS villages face a sign ificant opportunity cost when implementing the program, both in terms of the short run economic impact and the long run sustainability of stakeholder participat ion and involvement in VSS collective action. This average consumption analysis was conduc ted on the full sample of 58 VSS villages, which include VSS villages that have a relatively low dependency on the forest as measured by the percentage of total consumption that fore st products comprises. Positive socio-economic impacts of JFM/CFM should, however, be derived from improvements in forest productivity and because average consumption served as the depende nt variable, the relationship described in the previous paragraph is likely to be absent in many of the low-dependency VSS villages. Therefore, an additional regression was performe d on a subset of the sample data and these results indicated a positive predicted relationshi p between average consumption and the length of time under JFM/CFM. This suggests that JFM/ CFM is having a benefi cial socio-economic impact on the VSS villages of the sample subset that is comprised mainly of poorer households, many of which are Schedule Tribes. These are the principal stakeholde rs of CFM in Andhra Pradesh, as identified by the legislative orders and targeted by the implementing agencies. As such, this evidence implies the program is successful in at least this dimension of its objectives. The second model to examine any improvement s in the economic indicators of social welfare due to JFM/CFM attempte d to explain variations in c onsumption inequality using the Gini coefficient. This model performed poorly such that no results are summarized here. The third model to examine any improvements in the economic indicators of social welfare due to JFM/CFM attempted to explain variations in poverty as measured by the poverty-gap ratio.

PAGE 162

162 Estimation of this model revealed that poverty le vels are distinctly le ss in the Adilabad and Chittoor samples than they are for the Visakha patnam sample. The results also suggest, in particular, that NGOs providing fore stry-related assistance contribute to lower levels of poverty. However, the results also show there is no rela tionship between the poverty-gap ration and the length of time under JFM/CFM. The regression to analyze forest quality chan ges (FQC) as a function of the length of time that a given village has participated in the JF M/CFM program (i.e., the LT values) found that the compound marginal effects were statistically si gnificant for a two category improvement (i.e., FQC = 2). This result suggests that a relations hip between the JFM/CFM program and improved forest quality can be asserted. To summarize, if the length of time sin ce JFM/CFM was (1) positively related to the average level of consumption and (2) positivel y related to a change in forest quality, the interpretation is intuitive; fo rest quality improvements increase forest productivity through time, which eventually leads to increased benef its to households (as measured by average consumption). This was the finding for households dependent on forest products for livelihoods or subsistence needs in this study. As CFM institutions should have a direct effect on the physical resource, seven institutional variables were analyzed in the FQC regression. These variables represented five of Ostroms design principles. Three variables repres ented the first design pr inciple; one of which was statistically significant and positive (ARCU ). This indicates that awareness of access restrictions (ARCU) had a positive impact on fore st quality change. In addition, the statistical significance of MRFU indicated that the orientation of the local forest management plan towards forest-dependent households also had a positive impact on forest quality change. Thus, there is

PAGE 163

163 evidence that some of the CFM institutional variab les analyzed have a beneficial impact on the regeneration of forest lands protected under CFM in Andhra Pradesh. Institutional Analysis One of the main objectives of this study wa s to examine how well the CFM institutions conform to the design principles delineated by Ostrom (1990), both as they are defined in the legal orders that enable the program and in the actual field implementatio n by local stakeholders. All of the design principles for successful common property resour ce (CPR) management regimes are presentmore or lessin the CFM in stitutions enacted and being promoted by the Government of Andhra Pradesh. This is notable because they are meant to function together to reinforce the participation, commitment, and mo nitoring of stakeholders, without which success cannot be sustainable in the long run. However, most of the de sign principles are either only partially or implicitly represented and this ne eds to be addressed in future policy revisions. Moreover, two of the Ostrom design principles are not directly applicable to the CFM institutions examined by this study and can only be evaluated in terms of the degree to which they are approximated by analogous CFM institutions or participant behavior and perceptions. These two design principles (minimal recogni tion of rights to organize and collectivechoice arrangements) are important because they represent where the JFM paradigm deviates from the CPR management theory as expounded by Ostrom, due to the intrinsic governmental oversight of the program. First, unlike the case studies from which Ostroms design principles are synthesized, the vast major ity of forest protection taking place under CFM in Andhra Pradesh (and indeed JFM throughout India) is not self-originating becaus e the basic institutions enabling the program are derived from legislative orders. Thus, the extent of government involvement is far greater than Ostrom (1990) suggests it s hould be, which means that the seventh design

PAGE 164

164 principle (minimal recognition of rights to organi ze) is effectively moot within the institutional set of JFM/CFM. For this reason it was difficu lt to assess in its original context. Nevertheless, it is necessary to account for the s pirit of this design pr inciple and, as such, some proxy indicators were used to approximate it in terms of measuring VSS autonomy and government interference. An overwhelming percen tage of respondents re ported that the APFD (i.e., the government) allowed the VSS sufficient autonomy to manage their VSS forest area. In addition, a large majority of the VSS sampled i ndicated that the relatio nship with their local Forest Officer was either good or very good; th is measure intended to capture the degree of inference. Thus, it appears that the spirit of the se venth design principle is adhered to as much as can be expected given the circumstances that prev ent its full realization as Ostrom had intended. Secondly, collective-choice arrangements (the third design principle) are ostensibly guaranteed in the national forest policy a nd the state legislation defining CFM in Andhra Pradesh. This is because the CFM program pref erentially targets the most forest-dependent people, as seen by the automatic membership th at tribal households ar e granted in the VSS General Body and the VSS Management Committ ee. However, despite this, and although the local communities have some latitude to devise their own VSS-specific rules and regulations, they have no direct input into the creation and modification of the basic CFM institutions. This is, of course, a violation of collective-choice arrangements th at is a consequence of the government origins of the program. Note the interrelation between these two desi gn principles as a result of their weak representation (or re lative absence). Because the basic inst itutions that define JFM are exogenous to the local communities for which the program is designed, individual stakeholders do not have the opportunity to directly devise and revise the key institu tions that define the program and

PAGE 165

165 govern forest resource management. This is a st ructural design weakness that may eventually affect the functioning of the other design principl es and, thus, may have a long-term impact on the success of the program. For exam ple, the ability to revise the institutions to adapt to changing local conditions may be hampered due to the violation of collective-choice arrangements mentioned above. Notwithstanding this problem, though, it appear s that the program is responsive to the needs of those for which it is designed; im plying that collective-choice arrangements are conformed to well, if not totally in theory then at least in practice. For example, a majority of VSS households indicated that forest-dependent h ouseholds are indeed being represented in the local management plan. In addition, the el ection of the Management Committee by the VSS general body was deemed to be free and fair by a wide margin, and the Management Committee was reported to be accurately representi ng the rest of the VSS members well. The fact that many of the design principles are pr esent, to a greater or lesser degree, in the institutions of CFM is encouraging. This signifies, theoretically, that the foundation for successful long-term forest management is in pl ace. But clearly weaknesses remain at the statelevel that affect the implementation and effectiv eness at the local level. To its credit, the Government of Andhra Pradesh has been responsi ve to institutional ch ange by modifying their original JFM institutions to better facilitate th eir functioning, and to im prove the incentives for participation by local stakeholders. Despite th e relatively new CFM institutions in Andhra Pradesh that are more devolved, however, this updated program will never qualify as a fully selforiginated system of CPR manage ment for which Ostroms design pr inciples apply as an ideal. But the design principles that ar e directly applicable to JFM/CFM (e.g., monitoring) must be readdressed in order to strengthen them.

PAGE 166

166 The question at present is how effective ar e the relevant design principle-associated institutions being implemented in the field (i.e., at the local level) This was examined in detail in the previous chapter and the answer is similar to the evaluation at the state-level: overall, the entire set of the five design principles that we re targeted for analysis are represented throughout the sample of VSS surveyed as a whole, but no t in each individual VSS. Furthermore, the degree of adherence to each design principle as implemen ted as a CFM institution was variable, and was found to be deficient entirely lacking in some cases. Again, however, most of the basic foundations of successful CPR management (as per Ostrom) are there, albeit in varying degrees, both in the legislative orders and implem ented at the field level. The key is to begin making progress towards the full realization of the design principles, as represented by the specific institutions of CF Mto the extent possibl e given the governmental origin of the program. This is a difficult task be cause the standardization of certain institutions, in order for them to be more congruent with the design principles, must also preserve the ability of individual VSS villages to adapt the local-level institutions to their own local circumstances. Indeed, field observations underscore the need fo r flexibility and creativity of each individual VSS. The sheer variability in resource endowme ntswhether in terms of human capital, natural capital, physical capital, and/or fi nancial capitalrequire that c ookie-cutter approaches to rural development be eschewed. Thus, the CFM program w ill be constrained in its ability to evolve if it cannot reconcile the need for flexibility at the local level with uniformity at the state level, which is inherently needed for the wider implementation of the program. It is interesting to observe, then, that the fundamental nature of the JFM paradigm is somewhat paradoxical: a decentralized system of forest management is designed to empower loca l people, but is guided by mostly centralized institutions devised and promoted by the government.

PAGE 167

167 Policy Implications Recommendations from the Economic Analysis The main conclusions from the economic analys is are that the JFM/CFM program seems to have had a beneficial influence on the most forest -dependent VSS villages, but it also appears to have had a significant opportunity cost for thos e VSS villages that are better off economically. As the alleviation of poverty is a key objective of the program, it is encouraging to find some evidence that the poorer forest-dependent VSS ar e being reached. However, more needs to be done if the individuals in such villages are to be lifted out of poverty. Alternative and/or improved livelihood opportunities ar e needed to increase the low incomes and consumption of the households in the poorer VSS villages. This is a rural development issue as much as it is a forestry issue. In terms of the CFM program overall, the hi gh opportunity cost of the program also indicates the need for development or impr ovement of alternate or existing livelihood opportunities. Without a concerted fo cus on this topic, the long run sustainability of the program, in terms of stakeholder participation and commit ment, will be in jeopardy. Currently, the main pecuniary incentives for many VSS villages are derived from an external source (e.g., World Banks funds distributed by the APFD). In fact, Reddy et al. (2004) state that the major benefit from participatory forestry in Andhra Pradesh appears to be from externally supported wage employment, which is not sustaina ble in the long run. As such, pa rticipation is likely to wane when external funding is removed (Kumar, 2002), and Khare et al. (2000) point out that this situation was indeed observed in West Benga l when a World Bank supported project ended. To underscore the emphasis on livelihoods, a conference was held in March 2005 at the APFDs demonstration center in Ja nnaram (Adilabd district), where representatives of the World Bank addressed field officers of th e APFD. They advised these fore st officers that funding from

PAGE 168

168 the World Bank would only last a few more years and they should relay this information to the VSS members. The development and enhancem ent of livelihood opportun ities was a central theme of this conference; many di fferent ideas, techniques, and oppor tunities were discussed. It was clear that promotion of this agenda wa s important to the APFD and the World Bank, because they wanted the local stakeholders to sustain their forest management efforts in the absence of external funds. The only way to ensure this outcome is by increasing the incomes of stakeholders through the development of their skills and opportunities. Despite the World Bank presentation and th eir concern for livelihoods, the subsequent field visits during data collection for this study found that most VSS memb ers are unaware of the impending termination of funds. Worse still, ma ny VSS villages are currently without viable alternatives (i.e., in te rms of livelihood opportuniti es and/or marketable timber) to the external money that has so far supported th e forest rehabilitation efforts. It is obvious that a crucial deficiency on the part of the VSS members is an inability to plan more than one or two years in the future. Thus, it is extremely important that the APFD and various NGOs effectively assist in the planning of management schemes that: Enhance existing livelihood opportun ities through value-addition. Develop new livelihood opportunities through value-addition and marketing linkages. Procure and distribute physic al capital that allows value-addition in the above. Assist VSS villages and VSS members with the marketing and distribution of output. Devise management plans that stagger timb er production to ensure a steady source of income through time. A good example is the valued use of bamboo, which is being promoted by the APFD. This forest product is a great source of income for rural people in th e different areas where it grows well. Some value-added items produced using ba mboo include: baskets, floor and wall mats, sericulture mats for cultivating silkworms, and sm all slivers used for making incense sticks. The enhancement of bamboo production is not only ti ed to better propagation and silvicultural

PAGE 169

169 techniques, but also requires th e allocation of physical capital in the form of a small machine used to cut bamboo in different sized strips de pending on the ultimate use. By western standards these machines are relatively cheap and save a lo t of time in the manufacturing of the different end products. Training on these machines and for learning how to produce different bambooderived products is an important focus th at needs to be continually emphasized. As mentioned previously, the site-specific nature of re source endowments precludes a cookie-cutter approach to both forestry mana gement and related livelihood options. Other options for VSS villages in Andhr a Pradesh include the production of lac, which is a natural resin used as a dye and as the base component of shellac. It is de rived from the secretions of the tiny insect, Laccifer lacca as they colonize host trees and de posit their resinous pigment (a deep scarlet color). A local NGO employee has traine d various rural people in how to produce lac by preparing certain tress such as Moduga ( Butea frondosa ) in order to better cultivate the insect. Moduga is locally abundant and produces othe r important NTFPs (e.g., leaves, flowers and seeds) besides lac, but the development of this particular livelihood opti on could benefit certain VSS villages that lack other options (due to its relatively high remunerative potential). Two other livelihood options being promoted by the APFD are important because they have a wider range of availability in terms of the raw material (i.e., tree species) and the lower cost of the physical capital. The first is ba sed on a common activity in India that involves stitching leaves together by hand to make small di sposable plates for eati ng (both at home and in local food stalls or restaurants). Severa l tree species commonly found throughout Andhra Pradesh, such as Moduga and Adda (Bauhinia vahlii ), are used for these l eaf plates and offer an alternative source of income for VSS village s with few other opportunities. Another small machine promoted by the APFD can produce these pl ates by using pressure and heat to fuse the

PAGE 170

170 leaves into the shape of a plat e. Like the bamboo-cutting machin e, this is another example of how a small machine can increase the pr oduction efficiency of VSS members. Another livelihood option being pr omoted by the APFD is a tamarind-cake machine. It is common for rural Indians to collect tamarind fru its for cooking purposes and to sell in local markets. Use of a very simple mechanical press can quickly convert a large bulky bag of tamarind fruits into small bricks (i.e., cakes) that are easily transportable. When selling these tamarind cakes at market, the local people can ob tain up to twice the price per kilogram of tamarind than for the bulk product. Thus, this liv elihood option is important especially since the cost of the tamarind-cake machine (approximately US$300) is much less than the other machines mentioned. These options are strategies th at are known and available, but what is lacking is sufficient funding at the field level to provide each VSS v illage with the opportunity to obtain at least one of the above machines or skills (e.g., lac production). Also lacki ng are personnel that can assess the site-specific needs and opportunities faced by each VSS village, which is necessary to match these communities with the prope r capital (be it human capital, phys ical capital, etc.) that they require for economic improvement. While it is not feasible to simply give a machine to every village, those that can afford one can be put on an extended, interest-f ree repayment plan to cover the basic cost of the mach ine. Those VSS villages that ar e especially poor ought to have only a token repayment schedule. Recommendations for Training and Extension Although the APFD is aware of th e necessity of promoting some of these activities, their efforts need to be redoubled. In addition, such efforts need to be placed in a more integrated framework of development that involves effectiv e partnerships with other government agencies and local NGOs. Obviously, training and extens ion are at the root of any efforts towards

PAGE 171

171 developing skills and livelihood options. In gene ral, the APFD does a good job of promoting CFM and developing the knowledge and skills of VSS members through tr aining programs and local demonstration centers. The APFD is to be commended because, although CFM is forestrybased, it is actually more of a rural development scheme. This means that there are limits to what they are able to accomplish because the implemen tation of such a program is obviously not the specialty of the APFD. The formal training provided by the APFD is conducted at the Andhra Pradesh Forest Academy near Hyderabad. Courses taught ar e grouped into five subject areas: Forest Management, Financial Management, Busine ss Management, Social Development and Information Technology. This academy instructs a large number of people (a few thousand) throughout a given year, includi ng VSS Chairpersons and other VSS members, Forest Section Officer trainees, members of other governme nt agencies, members of various NGOs, and representatives from other state forest depa rtments and forestry institutes in India. Training of new Forest Section Officers at the academy needs to be geared more towards CFM and the field techniques that support it. Give n all the attention paid to CFM by the APFD (especially in the upper levels of management), it appeared as if the pers onnel in training were not given enough instruction in this area. This fact was acknowle dged by the then-Director of the Academy who indicated he was pushing for such changes in the curriculum to address this situation. As with any large organization, there are spa tial heterogeneities in terms of personnel and their commitment to CFM and the local stakeholde rs. The challenge is to make sure that the commitment and adaptation (from traditional fore st management to CFM) required are actually being registered throughout th e lower levels of the APFDb oth administratively, with

PAGE 172

172 committed Divisional Forest Officers, and at the fi eld level. Field officers working directly with the VSS must not only be committed to CFM, but they must have proper training to be effective facilitators and extension worker s. Most of the APFD personnel appear to doing a good job, but resources at the field-level seem to be stre tched thin and more training is necessary. A particular aspect of concern in this ar ea is the long-term commitment to the forest department of new recruits. Unlike the United Stat es or other countries, forestry programs are generally limited at the university level and so personal preference s for a career in forestry are unlikely to be an especially significant influence on new APFD recruits. For example, during this study, many of the officers undergoing training at the Academy were also taking government placement exams in order to obtain a different (non-forestry) job within the government. Either there is little inherent desire for a career work ing in forestry, or the remuneration in the forest department is lacking in relation to other governmental jobs. Whiche ver is the case, this is likely to affect the necessary commitment to the idea ls of the CFM program in terms of its rural development focus. Thus, the APFD may want to consider a modified se lection process for new employees; perhaps by developing a special internsh ip or recruitment program with the forestry department at the college in Nizam abad or the Indian Institute of Forest Management in Bhopal, for example. In addition, an internship or recr uitment program should also be set up with rural development related programs at other unive rsities in Andhra Prad esh or other states. The training of VSS Chairper sons and other VSS members at the Andhra Pradesh Forest Academy appears to be very good. Not only are important management skills taught to these individuals, but they also have an opportunity to interact with a diverse array of people from all parts of the state. This networking aspect is perhaps underappreciated by the APFD, although they do promote study tours for VSS members to visit other VSS villages in the state and

PAGE 173

173 occasionally other parts of India (e.g., Maharashtra, West Bengal). This is a critical element of training and extension because it can help VSS Chairpersons to fo ment ideas and strategies based on the successes of other VSS, especially if they have similar site-specific resource endowments and/or other bio-physical or soci al characteristics. Thus, more study tours should be promoted given available funding; but more importantly, additional networking strategies need to be developed to provide links for V SS villages to alternative live lihood opportunities that other VSS villages and/or NGOs know about. This job could be accomplished by several small extension teams of two or three people that travel around and coordinate the linkage of opportunities and recipients, and assess the various financial, human, and physical capital needs to support the alternative livelihood options. The need for such teams was prev iously outlined above. Another key element of the APFDs knowle dge-dissemination strategy is the regional demonstration center just outside the town of Ja nnaram, in Adilabad district. This center has a well developed nursery for raising different tree species (including cloned varieties of eucalyptus), a small demonstrati on garden of locally important medicinal plants, and a small demonstration plot filled with examples of the various technical works (e.g., soil-moisture trenches) that VSS members undertake in their fo rest areas. Most importantly, however, this demonstration center hosts differe nt training events and fora fo r the dissemination of skills and knowledge to VSS members, especially in the ar ea of livelihood opportun ities. A high priority ought to be placed on funding and developing more of these demonstration centers; at the very least there should be a similar type of demonstrat ion forum in every forest department circle. Recommendations from the Institutional Analysis Theory, empirical evidence, and field observations were all used to identify institutional and structural strengths and weaknesses of th e JFM/CFM program in order to increase the potential of this paradigm as a development t ool that can alleviate poverty and inequity. In

PAGE 174

174 particular, the institutional an alysis was undertaken specifica lly to highlight certain key institutions that needed attention in terms of being augmented where deficient or incorporated where absent. The following paragra phs offer some recommendations. A large majority of VSS members in the sa mple indicated that the access regulations concerning their protected forest area are clear and understandable. Of concern, however, is that non-members of the VSS are allowed access and use in the protected forest area of many VSS villages. While wider social considerations like ly inform this practi ce, in the long run the violation of the basic economic incentive struct ure for participation can potentially undermine the future commitment of VSS members if free-ri ding becomes widespread. This issue must be addressed by each VSS where it is common practi ce to have loose access requirements. At the very least non-members of the VSS should be charged a user fee to offset the cost of their forest use. Such a fee would distinguish them from VSS members, who would then know that others were not free-riding on their effort s. Only four VSS sampled indica ted that any type of user fee was collected; however, it appear s as if these fees apply to both members and non-members alikewhich is acceptable, theo retically, as long as the f ee structure is two-tiered. Awareness of the physical boundaries of the protected forest area is low, as is the general awareness of local institutions and parameters. The overall awareness on the part of VSS members (particularly boundary awareness) mu st be increased through improvements in the dissemination of information in VSS meetings. A dditional maps that are better suited to local knowledge, and which display boundaries that are more understandable to local stakeholders, would be helpful as well. This will help create a greater sense of resource ownership and perhaps lessen conflicts with other VSS that are related to boundaries. Saxena (199 7) recounts instances of inter-village disputes resulting from confusion over boundaries between communities.

PAGE 175

175 The state-level institutions grant VSS member s the power to apprehe nd violators, so the theoretical element of monitoring is present. As local protection of the forest is the cornerstone of CFM, it cannot be said that monitoring is bein g complied with fully: less than half of the VSS sampled had either formal patrols or a watchman. Vigilant monitoring is especially important since most rule violations seem to have occurr ed from people outside of the village. Long-term sustainability of VSS-member participation in fo rest protection requires effective monitoring so that members know their efforts are not in vain. The choices involve contr acting a third party to discharge the monitoring, which requires all hou seholds to share the costs; or to have the villagers themselves perform the duties, which is the usual practice because most households prefer to provide the labor themselves. A community self-patrol has the added benefit that it likely decreases inequality because the richer ho useholds may opt to hire the poor to fill their shift. At the very least, VSS Management Comm ittees need to be encouraged to set up daily patrols to ensure that there is a deterring presence in the forest everyday. An active monitoring program can reinforce the sense of community re sponsibility for forest management, as well as help instill this value into the next generation of VSS members. Although the government orders on CFM allow fi nes to be collected for minor offences, the particular legislative clause is wordy and seems unlikely to be followed in practice. For example, many VSS have set their own fine struct ure. In any event, the state-level institutions cannot truly be considered gra duated sanctions. In contrast, many of the VSS villages sampled report that they have a two-tiered structure of punishment; usually for the second offense, however, the violator is hande d over to the APFD. This can perhaps be considered quasigraduated sanctions, and ought to be the minimum level of acceptable sanc tions as encoded in the legislative orders. Although the current need for gr aduated sanctions appears to be relatively low,

PAGE 176

176 it may be of importance in the near future when external funding for CFM is removed. As such, this issue needs to be addressed in the form of legislative orders that mandate the graduated sanctions structure but leave the particular details to each individual VSS. Conflict-resolution mechanisms were not able to be formally analyzed as part of this study, but it is important to briefly address them because a low-cost forum for resolving conflicts seems to be necessary for CPR institutions to flourish, especially over long periods of time (Ostrom, 1990). Most VSS villages in the sample reported that disputes are adjudicated by the APFD; indeed, the government orders on CFM provide fo r a higher level (i.e., Forest Division-level) coordination committee to resolve inter VSS c onflicts and conflicts between the VSS and non VSS conflicts [sic] (GOAP, 2004). Nevertheless, the establishment of a local forum to deal with disputes internal to a VSS ought to be encouraged so that su ch matters can be dealt with promptly and transparently, as conflicts naturally arise when a community enforces the exclusion of access. This is especially so given th e recommendation above that access and use requirements need to be restri cted to VSS members only. In addition, an effective local mechanism for dispute resolution reinforces the ot her institutions that collectively work together to ensure effective commitment and monitoring. This is because complex institutions are never unambiguous in their interpretation and/ or implementation (Ostrom, 1990). A forum for conflict resolution would also a dd some gravitas to the overall institutional structure of the VSS and, as such, could help act as a deterrent to local elites who may be tempted to high-jack the VSS for their own intere sts. Finally, a forum of this type is probably more important for larger VSS villages that are li kely to have more internal disputes among VSS members, and that are also likely to ha ve a higher percentage of non-members.

PAGE 177

177 Caveats of the Study Survey The most limiting factor in terms of the su rvey was that preliminary fieldwork was insufficient prior to the initiation of this phase of the study. Budget consid erations had precluded this activity; however, given a similar opportunity in the future, all efforts would be made to ensure that some preliminary field visit could be arranged. Certainly the survey instruments (in particular the VSS questionnaire) would have be en strengthened by adva nce work, and important contacts could also have been established a priori As mentioned in Chapter 3, the field-testi ng of the questionnaires and training of the interviewers were cut short due to time consider ations. The incomplete fi eld-testing and training were limiting factors in terms of data accuracy for certain questions. For example, the household questionnaire is a good instrument in terms of th e information it is designed to collect but the wording, response codes, and/or recall periods of some questions are sub-optimal and needed to be refined. In particular, the module for data on the collection, consumpti on, and sales of timber and NTFPs required modifications to improve how these data could be reported and recorded more effectively. Additional testing would have uncovered these and other problematic questions, and been beneficial in how best to revise them and be tter deal with contingencies as they arose. Additional time could perhaps have allowed for more in-depth visits and interviews. This would have been informative in terms of collecting more deta iled information on the forest resources available in each VSS village, the VSS works conducted (and when), and the associated livelihood options a nd potential opportunities availabl e. Specifically, these data improvements would have allowed for a better pr ofile of each VSS village vis--vis the number and types of forest resources available, and the number and types of forest products consumed or

PAGE 178

178 sold at the household level. As it stands, exis ting household data on fuelwood collection could be hampered by measurement errors due to a lack of uniform and precise measures used in the instrument (i.e., how much does a bundle or bunch weigh and cost in Adilabad versus Visakapatnam?). Data The lack of baseline economic and forestry-rel ated data for the VSS participating in the JFM/CFM program precluded this study from tr acking direct tempor al changes within individually specified VSS villages. In this study a cross section of VSS villages was chosen in which the relative differences within this sample were used to evaluate the economic impacts of the JFM/CFM program. This is obviously a second-best method for conducting economic analyses having a temporal element, but at least offers one way to investigate this forest management paradigm in the absence of an appropriate comparative baseline. VSS-level data were also problematic in term s of availability from the APFD. The field officers of the APFD routinely collect data in order to monitor the compliance and progress of the participating VSS. Examples of data collected include the finances of each VSS (e.g., expenditures for labor employed for VSS works), si ze of treatment areas, ot her measures of VSS works (e.g., length of contour trenched dug), gene ral measures of inputs and outputs of forest management, and demographic information such as literacy rates and the number of VSS members. Data compilation and management by the APFD is generally poor. Crucial measures used in this study that were obtained from the A PFD (i.e., size of the VSS area and year of VSS establishment) are basic measures found in the micro-plan of each VSS; these data were not always readily accessible from electronic database s at the division levelthey should be easily available there and at all higher levels. Important input and output data regarding VSS finances

PAGE 179

179 or VSS works could not be obtained from the APFD due to the lack of cogent and timely compilation. Such data had to be estimated from information collected by the survey questionnaires, which subsequently impacted its accuracy and reliability. The general unavailability of data that ostens ibly exists limited some of the m odels in terms of incorporating certain factors as explanatory variables. The compilation, management, and availability of data are critical as pects that the APFD cannot overlook and must upgrade statewide at the division level. This means more and better equipment where needed but, more importantly, guidelines and training ar e primarily what is required. Note that the computer services and data management section of the Visakhapatnam Circle office did an excellent j ob providing data for this study, a nd could be used as an example of how to compile and manage data and inform ation correctly. The computer personnel in the Chittoor West and Jannaram division offices we re very helpful also, but their resources (especially office space) seemed to be limited and that appeared to hamper their efficiency. Future Work Many opportunities for future work abound regarding the topic of CFM in Andhra Pradesh, as well as for the broader subject of JF M in India. Additional wo rk can be undertaken to examine different subsets of the average consum ption model, especially in the context of interaction terms and slope differences among the three regional samples. The forest quality change model can be reformulated in two diffe rent ways by rescaling the dependent variable, FQC. The first way would be to simply collapse the ordinal dependent va riable into a simple binary variable without any qualitative s cale, where 1 = positive change and 0 = no change/negative change. An alternative model would be to examine forest quality change separately as a positive change (with FQC range of 0 to 3) and a negative change (with FQC measured as 0 or 1, where 1 = negative change).

PAGE 180

180 A number denoting the specific VSS village, the individual house hold number, and the actual person who conducted the interview uniquely identifies each observation in the household dataset. During the data compilation phase, the quality assurance and c ontrol checks uncovered two interviewers with sub-par performance. As such, it would be informative to explore the potential effects that intervie wer bias may have on the estima tion results. Either the proposed analysis of household participation in JFM/CFM or the modified FQC equation described above could incorporate such a tangential component, as both of th ese models are based on householdlevel data. Because the consumption inequality regres sion essentially had no explanatory power, it might be informative to remodel this equation using a binary dependent variab le to see if that has any effect on the statistical significance of the explanatory variables. For example, the twenty VSS villages with the lowest Gini coefficients could be assigned a value of zero (i.e., no inequality) while the twenty VSS vi llages with the highest Gini va lues would be given a value of one (i.e., inequality). Alternative choices for expl anatory variables could al so be sought as well. The mixed results of the poverty-gap regres sion suggest an opportunity for additional work. Deaton (1997) discusses some limitations of using Tobit models for the regression analysis of data pertaining to devel opment research, and offers median regression as one alternative. Using the median of the PGR for the regressand, instead of the mean value, would effectively blunt the effect of the 18 observations in the sa mple that are censored at zero (because they have no households below the poverty line). This may perhaps improve the estimation results as compared with the Tobit model. An alternative a pproach would be to use a selection model if a suitable specification of explanatory variables co uld be found for the selection equation, because the selection and outcome equations must di ffer by at least one independent variable.

PAGE 181

181 Another focus could be to use the existing da ta to examine the determinants of household participation in JFM/CFM as members of a VSS. A selection model framew ork could be used to: (1) isolate the decision to participate as a binary choice dependent variable (i.e., the selection equation), and (2) analyze the level of participation in the VSS (i.e., the outcome equation) as a continuous dependent variable measured by the per-capita number of days of labor that a household was employed for VSS works projects in the previous year. The preceding paragraphs detail additional work that the data collected in this study can support. However, there are other important areas of investigation beyond the immediate focus of this study and for which additional data woul d be required. For example, the ideal manner in which to model the economic impact of JFM/CFM would be to undertake a longitudinal analysis based on panel data. Thus, a future study could re visit the VSS villages (if not households) that were sampled by this study and collect new data that would then be inco rporated into a panel design with the data collected previously. Another approach would be a cohort analysis where additional survey data on selected VSS villages are periodically coll ected and analyzed. As the data of this study is a cross-section of cohorts, as defined by the establishment year of the VSS (i.e., LT), additional data would simply expand the number of VSS villages of each age cohort and expand the total number of cohorts as new villages adopt the program each year. The addition of randomly selected nonVSS villages as a control group would perhaps pr ovide a better evaluation strategy, especially for a single cross-section. This w ould allow the two groups to be test ed for a treatment effect that the adoption of the JFM/CFM program would represent. Additional focus should also be made towa rds better measurement and analysis of the institutional variables that reflect the Ostrom design principles, especially for those that are

PAGE 182

182 proxy measures. In general, a larger study with more observations is needed to better investigate the institutional aspect of JFMin India as a whole, and not just Andhra Pradesh. A particularly worthy area of focus would be a comprehensive analysis of the different JFM institutions across the states of India, both theore tically and empirically, in terms of the design principles of Ostrom as well as the response of each st ate with regard to revisions a nd evolution of the program. Part of such a study would also include a deeper examination of the paradoxical institutional dichotomy of JFM, as was described previously which is that local participatory forest management is enabled and dire cted by state-level institutions.

PAGE 183

183 APPENDIX A THE VSS QUESTIONNAIRE UNIVERSITY OF FLORIDA COMMUNITY FOREST MANAG EMENT RESEARCH STUDY ANDHRA PRADESH, INDIA SPRING 2005 VSS-Level Questionnaire DISTRICT ID VSS ID VSS VILLAGE NAME INTERVIEW DATE: START TIME: END TIME: NAME OF EC CHAIRMAN: CHAIRMANS AGE: CHAIRMANS LEVEL OF EDUCATION: NAME OF EC VICE-CHAIRMAN: Part A: Historical Background Questions 1) Thinking back to before the CFM program be gan here, how would you rate the quality of the forest area at that time? 2) Regarding the forestland now protected by the V SS, what were the main uses of this area by members of the community before CFM bega n in this village? (Circle the most important.) 3) Did this village have any prior experience with community management of any forest areas? (e.g., sacred forest, temple lands, irrigation networks, etc.) YES NO If yes, specify: GOOD (manchiga)......................1 SL. DEGRADED (palsaga undedi)........2 DEGRADED (podaluga undedi)...........3 V ERY DEGRADED (banjaruga undedi).....4

PAGE 184

184 Part B: Essential Questi ons for Econometric Analyses 1) Are there access restrictions to the VSS forest area? YES NO If no, go to 4. 2) What are the restrictions imposed? 3) How are these restrictions implemented? (e.g., by activity, household, user fee, etc.) 4) How does the community monitor the VSS forest area? 5) What sanctions does a violator of the rules encounter? GRADUATED SANCTIONS? 1st Violation: ____________________________________________ Check if YES: _______________________________________________ 2nd Violation: ___________________________________________________ ______________________________________________________________________ 6) Have there been any violations either within the VSS or from non-members and/or other communities? YES NO If yes, prompt for details.

PAGE 185

185 7) Have there been any other type s of conflicts or disputes with non-VSS members and/or other communities? YES NO If yes, specify: 8) Does any NGO work specifically with th e VSS on forest related issues? YES NO 9) How many NGOs in total work with this community? NUMBER (e.g., with respect to heal th care, education, etc.) 10) What is the total number of training events (e.g., Dullapalli, Vallur) and/or field visits (e.g., West Bengal, other VSS areas) that members of the Executive Committee have participated in, since the inception of the VSS? 11) Did these experiences help improve the VSS? YES NO 12) VSS Work Project Details : Please describe the types of works taken up by the VSS and the expenditures incurred. 2004-2005 2003-2004 2002-2003

PAGE 186

186 13) How many man-days of employ ment did the VSS provide in the past 3 years for all works projects? 2002 2003 VSS works: ________ persons ________ days 2003 2004 VSS works: ________ persons ________ days 2004 2005 VSS works: ________ persons ________ days 14) What are the criteria for sele cting workers for VSS projects? 15) What is the daily wage rate offered to workers employed in VSS project work? RUPEES 16) How many DWCRA groups are in this VSS village? NUMBER 17) Has this village received assistance in the pa st from the ITDA? If yes, was it helpful? YES NO 18) Has this village received assistance in the past from any other Govt agency? YES NO

PAGE 187

187 Part C: Participation/Equity Questions 1) Is the selection of the VSS Executive Committ ee made by consensus, or is an election held? 2) What is the length of term of the Executive Committee? YEARS 3) Describe the relationship between the Ex ecutive Committee and the FD officer with whom you normally interact? 4) Does the FD give the Executive Committee enough autonomy to manage the VSS forest area in a way that best serves the community? YES NO 5) VSS Future Livelihood Plans/Investments : Please describe how the VSS plans to sustain itself in the future, gi ven the decline and elimination of external funding in a few years. What investments are being made and/or requested? CONSENSUS...................0 ELECTION....................1 V ERY GOOD (OFFICER REALLY SUPPORTS VSS EFFORTS)......................1 GOOD (OFFICER IS HELPFUL)............................................2 FAIR (OR INDIFFERENT)................................................3 BAD (THERE ARE PROBLEMS WITH THIS OFFICER)...........................4

PAGE 188

188 6) Does the VSS micro-plan adopted by the V SS contain any items that the FD insisted upon, even though the community does not want or need? (e.g., particul ar tree species or NTFP, etc.) YES NO If yes, specify item(s): 7) Does this village have any Social Forestry? If so, what types of tree species are grown. 8) Is any of the VSS area under plantation? If so, please specify the tree species. 9) What is the frequency of meetings for th e Executive Committee, and for the General Body? 10) Typically, what is the attendance for a meeting of the VSS General Body? 11) What is the current balance of the VSS Account, approximately? RUPEES 12) How would you describe the overall particip ation of VSS members in VSS meetings, projects, and works? V ERY INVOLVED......1 INVOLVED...........2 SOMEWHAT INVOLVED..3 N OT INVOLVED.......4 DONT KNOW........99

PAGE 189

189 Part D: Final Questions 1) Have there been any problems employing people for the VSS project works? YES NO If yes, specify problem(s): 2) Have there been any conflicts within the VSS? YES NO If yes, specify: 3) Have there been any encroachment problems from other communities or other VSS ? If so, how have they been solved? YES NO 4) What is the extent of the encroachment area? ACRES 5) What are the main crops grown here and the main market(s) for them? 6) Approximately how many functioning bore-wells ar e there in this village? NUMBER 7) What is the average depth of water in these bore-wells? FEET 8) Do people from this village migrate to other places for work? YES NO If yes, specify:

PAGE 190

APPENDIX B THE HOUSEHOLD QUESTIONNAIRE

PAGE 191

191 UNIVERSITY OF FLORIDA COMMUNITY FOREST MANAGE MENT RESEARCH STUDY RURAL HOUSEHOLD SU RVEY, SPRING 2005 ANDHRA PRADESH, INDIA Principal Investigator: Frederick Rossi Food and Resource Economics Department University of Florida Research Associate: R. Vishnu Vardhan Reddy M.Sc. (Agricultural Economics) With assistance from: Andhra Pradesh Forest Department

PAGE 192

192UNIVERSITY OF FLORIDA ANDHRA PRADESH COMMUN ITY FOREST MANAGEMENT RESEARCH STUDY Rural Household Survey: Informed Consent Protocol (Verbal) I (Name of Investigator) am working for Frederick Rossi, Research Scholar of University of Florida. Dr. Janaki Ram Reddy Associ ate Professor at School of Forest Resource and Conservation, University Florida is the guide for Frederick Rossi. Fred Rossi is doing research on particip ation of rural communities in Community Forest Management and benefits accruing to rural families in Andhra Pradesh. The data that will be collected from yo u will be available only to Fred Rossi. There are no risks / benefits for participating in this research. You will not receive any compensation / payment f or participating in this research. The questions will be in detail on your knowledge of CFM, on projects, your pa rticipation, and your family's economic activitie s. For example: In last 2 weeks, is your family purchased Red gram? What is the price paid? How much quantity you have purchased? Other questions may be like this one: Have you voted in Vss Management Committee elections? How many Women members are there in Vss Management committee? If you don't feel like answering any of the questions you can do so. It may take approximately one hour to answer the questions If you want the interview to be scheduled at other time I will do so. I will be giving you a copy having summary of the details given to you now. If you need any details regarding this research, yo ur rights, if you have questions or any other doubts. You can call the phone number given in the copy We will answer you. If you dont like to sign the copy, you can give oral consent. May I start the interview? DISTRICT ID VILLAGE ID INTERVIEWER HOUSEHOLD ID DATE: START TIME: END TIME: RESPONDENT NAME: CASTE:

PAGE 193

193HOUESEHOLD ROSTER 1. Before beginning this survey, I request you to enumerate your family details. LIST ALL HOUSEHOLD MEMBERS BEFORE GOING TO QUESTIONS 4 9. 3. Their relationship with the Head of the Family? 4. Age of the member? 5. What is the educational qualification of the member? (Highest educational qualification). 6. Past year, how many months did the member live here? 8. To know the member's contribution to the family, give two important occupational details of the member. NAME OF HOUSEHOLD MEMBER 2. Sex (gender) of the member? CODE CODE MONTHS 7. BASED ON THE CRITERIA IN Q. #6, IS THIS PERSON A MEMBER OF THE HOUSEHOLD? #1 #2 IF 10,SPECIFY 9. Who among the family, two members belong to the VSS? 1 2 3 4 5 6 7 8 9 10 11 INDICATE WITH A CHECK MARK, THEN GO TO NEXT PAGE I F < 3 Y EARS: () RELATIONSHIP CODES: HEAD.....................1 SPOUSE OF HEAD...........2 SON/DAUGHTER.............3 SON/DAUGHTER-IN-LAW......4 GRANDCHILD...............5 FATHER/MOTHER............6 SISTER/BROTHER...........7 NIECE/NEPHEW.............8 FATHER/MOTHER-IN-LAW.....9 BROTHER/SISTER-IN-LAW...10 SERVANT/EMPLOYEE........11 OTHER...................12 OWN FARM PRODUCTION (AGRI)..........1 HOUSEHOLD ENTERPRISE.......2 LONG-TERM AGRI. EMPLOYEE.............3 SALARIED EMPLOYEE..........4 CASUAL LABOR...............5 COLLECTION/FORAGING........6 HOUSEHOLD CHORES...........7 ATTENDS SCHOOL.............8 DOES NOT WORK..............9 OTHER ....................10 MALE..1 FEMALE. ......2

PAGE 194

194 HOUSEHOLD SURVEY: CONSUMPTION MODULE PART A: DAILY EXPENSES 1. Following questions will be on purc hases made for your family. Purchaser can be any member of the family. In the past 7 days did any member in the house spend money to purchase any of the items given below? 2. How much did the member spend on ()? 3. In the past 7 days, How many times did household members eat out (Tiffin or snacks / lunch or dinner)? 4. In the past 7 days, what is the amount spent on eating out? ITEM YES RUPEES ITEM NUMBER RUPEES 1 Chutta (local kind of cigar), Beedi, Cigarettes. Tiffin/Coffee and Tea 2 Gutka, Khaini, Panmasala, Tambaku (tobacco products). Madhyana Bhojana (lunch) 3 Tamalapakalu (betel leaves). Rathri Bhojana (dinner) 4 Newspapers or Magazines (e.g., Swathi) Thinu Bandaralu (food items) 5 Lottery Tickets Mandu (e.g., Kallu, beer, wine, whiskey) 6 Prayana Kharchulu (e.g., bus, autorickshaw) 7 Devudiki Kanukalu (donation to god) 8 Toddy/Kallu IF YES, PUT A CHECK MARK IN THE YES BOX FOR EACH ITEM, THEN ASK Q. 2 FOR THOSE ITEMS.

PAGE 195

195Following questions will be on purchases made for your family. Purcha ser can be any member of the family. PART B: FOOD AND FUEL PURCHASE FREQUENCY HOME PRODUCTION 1. In the last 12 months, give details of purchases/ usage of food items by your family. Exclude the items that are purchased for business purpose. 2. How often do your family members purchase (...)? 3. How much did they purchase? 4. Total amount spent on the purchase? 5. In the last 12 months, how much quantity of your own production of (...) did you consume? 6. What is the value of own production of (...) you have consumed? ITEM YES CODE FREQUENCY QUANTITY RUPEES QUANTITY UNIT RUPEES Biyyam (Rice) PDS 1 Biyyam (Rice) Market 2 Jonnalu (Sorghum) 3 Mokkajonnalu (Maize) 4 Wheat 5 Kandi (Red Gram i.e., Red Lentils) 6 Sanaga (Bengal Gram) 7 Minumulu (Black Gram) 8 Pesalu (Green Gram) 9 Pillipesearea 10 Verusanaga (Groundnut) 11 Other Grains 12 13 14 15 UNIT CODES: KILO.....1 LITER....2 DOZEN....3 PACKET...4 BUNDLE...5 GRAM.....6 BOTTLES..7 BUNCH....8 OTHER....9 WEEKLY.....1 FORTNIGHTLY....2 MONTHLY....3 QUARTERLY..4 BIANNUALLY...5 YEARLY.....6 IF YES, PUT A CHECK MARK IN THE YES BOX BELOW FOR EACH FOOD ITEM, THEN ASK Q. 2 6 FOR THOSE ITEMS.

PAGE 196

196PART B: FOOD AND FUEL PURCHASE FREQUENCY HOME PRODUCTION 1. In the last 12 months, give details of purchases/ usage of food items by your family. Exclude the items that are purchased for business purpose. 2. How often do your family members purchase (...)? 3. How much did they purchase? 4. Total amount spent on the purchase? 5. In the last 12 months, how much quantity of your own production of (...) did you consume? 6. What is the value of own production of (...) you have consumed? ITEM YES CODE FREQUENCY QUANTITY RUPEES QUANTITY UNIT RUPEES Palm Oil 16 Other Cooking Oils 17 Palu (Milk) 18 Curd 19 Chakkera/Panchadara (Sugar) 20 Eggs 21 Mutton 22 Chicken 23 Chepalu (Fish) 24 Vankaya (Brinjal i.e., Eggplant) 25 Tomato 26 Ullipaya (Onion) 27 Karivepaku (Curry Leaves) 28 Chinthapandu (Tamarind) 29 Benda (Lady Fingers i.e., Okra) 30 UNIT CODES: KILO.....1 LITER....2 DOZEN....3 PACKET...4 BUNDLE...5 GRAM.....6 BOTTLES..7 BUNCH....8 OTHER....9 WEEKLY.....1 FORTNIGHTLY....2 MONTHLY....3 QUARTERLY..4 BIANNUALLY...5 YEARLY.....6 IF YES, PUT A CHECK MARK IN THE YES BOX BELOW FOR EACH FOOD ITEM, THEN ASK Q. 2 6 FOR THOSE ITEMS.

PAGE 197

197PART B: FOOD AND FUEL PURCHASE FREQUENCY HOME PRODUCTION 1. In the last 12 months, give details of purchases/ usage of food items by your family. Exclude the items that are purchased for business purpose. 2. How often do your family members purchase (...)? 3. How much did they purchase? 4. Total amount spent on the purchase? 5. In the last 12 months, how much quantity of your own production of (...) did you consume? 6. What is the value of own production of (...) you have consumed? ITEM YES CODE FREQUENCY QUANTITY RUPEES QUANTITY UNIT RUPEES Pachimirapa (Green Chillies) 31 Ithara Kuragayalu (other vegetables) 32 (Specify) 33 34 35 Aratipandu (Banana) 36 Jamakaya (Guava) 37 Seethaphal (Custard Apple) 38 Mamidi (Mango) 39 Ithara Pallu (other fruits) 40 (Specify) 41 42 43 Jeelakarra/Avalu (Mustard) 44 Gasagasalu (Poppy Seeds) 45 UNIT CODES: KILO.....1 LITER....2 DOZEN....3 PACKET...4 BUNDLE...5 GRAM.....6 BOTTLES..7 BUNCH....8 OTHER....9 WEEKLY.....1 FORTNIGHTLY....2 MONTHLY....3 QUARTERLY..4 BIANNUALLY...5 YEARLY.....6 IF YES, PUT A CHECK MARK IN THE YES BOX BELOW FOR EACH FOOD ITEM, THEN ASK Q. 2 6 FOR THOSE ITEMS.

PAGE 198

198PART B: FOOD AND FUEL PURCHASE FREQUENCY HOME PRODUCTION 1. In the last 12 months, give details of purchases/ usage of food items by your family. Exclude the items that are purchased for business purpose. 2. How often do your family members purchase (...)? 3. How much did they purchase? 4. Total amount spent on the purchase? 5. In the last 12 months, how much quantity of your own production of (...) did you consume? 6. What is the value of own production of (...) you have consumed? ITEM YES CODE FREQUENCY QUANTITY RUPEES QUANTITY UNIT RUPEES Allam/Vellulli (Ginger/Garlic) 46 Endumirapa (Dried Red Chillies) 47 Tea 48 Coffee 49 Bottled drinks (Pepsi, etc.) 50 Beer 51 Ithara Madyapanalu (other Alcohol) 52 Snacks (Samosa/Biscuits/etc) 53 Pillalaki Palapodi (Baby Foods) 54 Canned Foods 55 Kerosene 56 Cooking Gas 57 58 59 60 UNIT CODES: KILO.....1 LITER....2 DOZEN....3 PACKET...4 BUNDLE...5 GRAM.....6 BOTTLES..7 BUNCH....8 OTHER....9 WEEKLY.....1 FORTNIGHTLY....2 MONTHLY....3 QUARTERLY..4 BIANNUALLY...5 YEARLY.....6 IF YES, PUT A CHECK MARK IN THE YES BOX BELOW FOR EACH FOOD ITEM, THEN ASK Q. 2 6 FOR THOSE ITEMS.

PAGE 199

199PART C: NON-FOOD GOODS PURCHASES LAST 30 DAYS LAST 12 MONTHS 1. Following questions will be on purchases made for your family. Purchaser can be any member of the family. Exclude the items that are purchased for business purpose. In the last 12 months, have you purchas ed or amount spent on or received as gift any of the following (. Item.) 3. How much did you spend on (...)? 4. In the last 12 months, How much did your family spend on (...)? ID ITEM YES 2. Did any member of the family purchase (...) In the last 30 days? RUPEES RUPEES 1 Ladies clothing + tailoring charges 2 Gents clothing + tailoring charges 3 Childrens clothing + tailoring charges 4 Aadavalla Cheppullaki (Ladies footwear) 5 Magavalla Cheppullaki (Mens footwear) 6 Pillala Cheppullaki (childrens footwear) 7 Personal Services (e.g., hai r-cuts, shaving, etc.) 8 Sabbulu (soap), Shampoolaki (shampoo), Toothpaste 9 Make-up, Other Products for Women 10 Kitchen Utensils, Dishes Pots/Pans, etc. 11 Small Kitchen Appliances/Other Small Electrical Items 12 Electrical Supplies (light bulbs, batteries, etc.) 13 Household Linen (towels, bed sheets & others) 14 Repair and Maintenance of Appliances 15 Repairs, Maintenance, and Improvements to House 16 Vehicle Repairs and Maintenance, Parts and Licenses YES NO..2 4 IF YES, PUT A CHECK MARK IN THE YES BOX FOR EACH ITEM, THEN ASK Q. 2 4.

PAGE 200

200PART C: NON-FOOD GOODS PURCHASES LAST 30 DAYS LAST 12 MONTHS 1. In the last 12 months, have you purchased or amount spent on or received as gift any of the following (. Item.) 3. How much did you spend on (...)? 4. In the last 12 months, How much did your family spend on (...)? ID ITEM YES 2. Did any member of the family purchased (...) In the last 30 days? RUPEES RUPEES 17 Pustakala Kharchulu (expenditure on books) 18 Post-officeki/Cell phone/Telep hone Kharchulu (expenses) 19 Cinema Kharchulu (expenses) 20 Toys, Sports Equipment, Small Games 21 Travel Expenses (marriag e, pilgrimage, jatara) 22 Ceremonies (child birth, marriages) 23 Dowry 24 Doctor, Hospital, Traditional Medicines 25 Insurance (auto, property, LIC) 26 Taxes (housing, property and others) 27 Legal Services, Registrations, etc. 28 Contributions (temples/m osques/churches/charities) 29 Funeral Expenses 30 Savings Account Deposits 31 Gambling Losses 32 YES NO..2 4 IF YES, PUT A CHECK MARK IN THE YES BOX FOR EACH ITEM, THEN ASK Q. 2 4.

PAGE 201

201CONSUMPTION MODULE PART D: DURABLE GOODS 1. Do you own any of the following articles / Goods? 2. How many (...) do you own? 3. How many years back did you purchase the (...)? 4. Have you purchased the article brand new or used one? 5. How much did you spend to purchase the article? If you have received it as gift what is the value? ITEM YES NUMBER YEARS NEW USED RUPEES 1 Stove 2 Fridge (refrigerator) 3 Grinder 4 Sewing machine 5 Fan 6 T.V. 7 VCD 8 Tape Recorder/Radio 9 CD Player 10 Camera 11 Video Camera 12 Cell Phone 13 Cycle 14 Motorcycle/Scooter 15 Car 16 Lorry (truck) 17 Computer 18 Washing Machine IF YES, PUT A CHECK MARK IN THE YES BOX FOR EACH ITEM. THEN ASK Q. 2 4.

PAGE 202

202FOREST RESOURCES MODULE PART A: CFM AWARENESS AND PARTICIPATION IS THE RESPONDENT ONE OF THE HOUSEHOLDS VSS MEMBERS? YES.1 NO..2 IF YES, GO TO QUESTION 1. IF THE ANSWER TO THE ABOVE QUESTION IS NO, ASK TO SPEAK TO A VSS MEMBER. RESCHEDULE FOR LATER IF NECESSARY TO INTERVIEW A VSS MEMBER. This module consists of questions to know about your family's involvement and opi nion with regard to Community Forest Managemen t (CFM) of your village. First few questions are to know about your family's usage of forest (currently the fore st area under VSS) prior to the introduction of CFM. CODE 1 What is the quality of forest prior to the introduction of CFM. Following few questions are to know your opinion about family's participation in CFM and VSS micro-plan. YES1 NO.2 CODE 2 Prior to the introduction of CFM did your family use forest produce, currently they are under VSS protection. 5 Is election to VSS Management Committee by agreement / consensus or is it election by secret ballot method? YES.1 NO..2 6 In the last VSS Management Committee elections, did both members from your family cast their votes? YES.1 NO..2 3 ASK THEM TO IDENTIFY WHICH USE WAS MOST IMPORTANT: 7 Do you or any one of your family is a member of the VSS Managing Committee? RUPEES YES.1 NO..2 4 Prior to the introduction of CFM, what is the average value of the forest produce used in a year for different needs of your family. 8 Do you think VSS Managing Committee considers views, needs of majority of its VSS members? CODE 9 In practice, VSS Micro-plan reflects inte rests of which section of the society? GOOD (manchiga)......................1 SL. DEGRADED (palsaga undedi)........2 DEGRADED (podaluga undedi)...........3 V ERY DEGRADED (banjaruga undedi).....4 IF YES, SPECIFY USES; THEN 3 IF NO, ASK WHY NOT; THEN 5 CONSENSUS (ekagrivamuga)...0 7 E LECTION (ennikalu)........1 FOREST DEPENDENT HHs (adavi pi adarapadina).....1 OTHER HOUSEHOLDS (ethara kutambalu).............2 LOCAL ELITES (grama peddalu)....................3 FOREST DEPARTMENT (atavi shaka).................4

PAGE 203

203 CODE 10 How do you describe VSS member's participation in CFM micro-plan preparation? Now I would ask few questions on VSS, current usage of VSS forest area. MINUTES 14 How much time it takes to walk to forest area managed by CFM from your home? YES.1 NO..2 11 In your opinion, what are the reasons for non-participation of VSS members in micro-plan preparation? 15 Do you know exact boundaries of VSS forest area? CODE CODE 16 What is the improvement in quality of VSS forest area since formation of CFM? YES.1 NO..2 YES.1 NO..2 12 Do you think Forest Department is giving sufficient autonomy to VSS, to manage forest area and to provide best services to community? 17 Do You pay any membership fee to VSS? RUPEES 13 In general, what do you think about the way VSS Management Committee elections are held? 18 How much do you pay for a year? CODE 19 For which important use do you pay this fee (say grazing, and so on). IF 3 OR 4, 12 V ERY INVOLVED (andaru sabyulu palgontara)..........1 INVOLVED (ekkuva sabyulu palgontara)...............2 SOMEWHAT INVOLVED (thakkuva sabyulu palgontara)....3 N OT INVOLVED (sabyulu palgonara)...................4 DONT KNOW (thelavadu)...........................99 THE PROCESS WAS DOMINATED BY LOCAL ELITES............1 THE PROCESS WAS DOMINATED BY INTEREST GROUP...........2 THE PROCESS WAS DOMINATED BY THE FOREST DEPT..........3 THE MEMBERS WERE INDIFFERENT (sabyulu pattichukoru)...4 OTHER (Specify).......................................5 DONT KNOW...........................................99 FREE and FAIR (vattidi lekunda swatantramuga)...1 DOMINATED BY LOCAL ELITES.......................2 DOMINATED BY INTEREST GROUP (VARGAMU)...........3 OTHER (Specify).................................4 DONT KNOW.....................................99 I F NO: 19 GOOD.(manchiga)...............1 SL. DEGRADED.(palsaga)........2 DEGRADED.(podaluga)...........3 V ERY DEGRADED.(banjaruga).....4 PROBE RESPONDENT ABOUT THE B OUNDARIES OF THE VSS FOREST: NEWS

PAGE 204

204 RUPEES YES.1 NO..2 20 Currently what is the approximate balance amount in the VSS account? 25 Regulations concerning entry into VSS forest area, are they clear and easy to understand? HECTARES YES.1 NO..2 21 What is the extent of area in hectares under CFM managed by VSS? 26 Do you think majority of VSS members follow rules and regulations? YEARS CODE 22 What is the tenure of VSS Management Committee? 27 According to you, what is the main reason for majority of people to follow rules and regulations? YEAR 23 In which year VSS was established? Before going to next section, I will ask few questions on community participation in CFM. 28 In your opinion, what is the main reason for people not following rules and regulations? CODE RUPEES 24 Approximately, what is the level of par ticipation of your family in VSS (like participation in General Body meeti ng, VSS works, Projects and so on)? 29 If all the members of VSS contribute some amount for the development of community, what amount is your family ready to contribute for the same? I F NO, 28 THEY RESPECT THE LAW.................1 THEY RESPECT THE FOREST AND NATURE...2 THE THREAT OF SOCIAL DISAPPROVAL.....3 THE THREAT OF THE MONETARY FINES.....4 THEY KNOW THERE ARE BENEFITS FOR COMMUNITY AND INDIVIDUAL HH......5 OTHERREASON(SPECIFY)6 V ERY ACTIVE (CHURUKUGA ANNI SABHALU/PANULU PALGONTAMU)..1 SOMEWHAT ACTIVE (THARACHUGA PALGONTAMU).................2 OCCAISIONALLY ACTIVE (APPUDAPPUDU PALGONTAMU) ..........3 N OT ACTIVE AT ALL (PALGONAMU)...........................4

PAGE 205

205FOREST RESOURCES MODULE PART B: INCOME FROM FOREST PRODUCTS 1. In the last 12 months, did any one of your family members collected any of the following (NTFP Item) from the nearby forests? 2. In one-year period, for how many months your family members collect (.)? 3. Generally in a month how much quantity of (.) do your family collect? 4. How much quantity of (.) collected in a month do you sell? 5. Last time when you sold () what price did your family received? 6. Who purchases (.) from your family? 7. From which region / forest area do you collect ()? ID NTFP NAME YES MONTHS QUANTITY UNIT CODE QUANTITY UNIT CODE RUPEES UNIT CODE CODE CODE 1 Tapsi Gum 2 Kondagogu Gum 3 Chirumanu Gum 4 Gumpena Gum 5 Honey 6 Karakkaya 7 Shikakai 8 Kunkudukaylu 9 Kanuge Ginjalu 10 Mahua (Ippa) Ginjalu 11 Mahua (Ippa) Puvvu 12 Nallajeedi 13 Sarapappu 14 Vepa Ginjalu 15 Jeddipikkalu GOVT (GCC)....1 A CONTRACTOR...2 M KT VENDOR.....3 LOCAL PEOPLE...4 OTHER(specify..5 In this section, I will be asking about your family's collection of Forest Produce. LIST EACH ITEM, THEN ASK Q. 2 7. V SS FOREST.....1 OTHER FOREST...2 OWN LAND.......3 OTHER(specify).4 UNIT CODES: KILO.....1 LITER....2 DOZEN....3 CARTS....4 BUNDLE...5 GRAMS....6 POLES....7 BUNCH....8 PACKS....9 OTHER...10

PAGE 206

2061. In the last 12 months, did any one of your family members collected any of the following (NTFP Item) from the nearby forests? 2. In one-year period, for how many months your family members collect (.)? 3. Generally in a month how much quantity of (.) do your family collect? 4. How much quantity of (.) collected in a month do you sell? 5. Last time when you sold () what price did your family received? 6. Who purchases (.) from your family? 7. From which region / forest area do you collect ()? ID NTFP NAME YES MONTHS QUANTITY UNIT CODE QUANTITY UNIT CODE RUPEES UNIT CODE CODE CODE 16 Thuniki (Beedi Leaf) 17 Adda Akulu 18 Isthara Aaku (Mothuka) 19 Rayla 20 Lakka 21 Cheepurlu 22 Eetha Chapalu 23 Usiri 24 Seethaphalalu 25 Chinthapandu 26 Maredu Gaddalu 27 Nelavemu 28 Sathavari Gaddalu 29 Ithara NTFPs 30 31 32 33 GOVT (GCC)....1 A CONTRACTOR...2 M KT VENDOR.....3 LOCAL PEOPLE...4 OTHER(specify..5 V SS FOREST.....1 OTHER FOREST...2 OWN LAND.......3 OTHER(specify).4 UNIT CODES: KILO.....1 LITER....2 DOZEN....3 CARTS....4 BUNDLE...5 GRAMS....6 POLES....7 BUNCH....8 PACKS....9 OTHER...10

PAGE 207

2071. In the last 12 months, did any one of your family members collected any of the following (NTFP Item) from the nearby forests? 2. In one-year period, for how many months your family members collect (.)? 3. Generally in a month how much quantity of (.) do your family collect? 4. How much quantity of (.) collected in a month do you sell? 5. Last time when you sold () what price did your family received? 6. Who purchases (.) from your family? 7. From which region / forest area do you collect ()? ID ITEM NAME YES MONTHS QUANTITY UNIT CODE QUANTITY UNIT CODE RUPEES UNIT CODE CODE CODE 34 Small timber and poles 35 Fuelwood 36 Ithara wood products 37 38 39 40 41 42 Wild Game 43 44 45 46 47 48 49 Grasses/Fodder 50 Do you graze animals? GOVT (GCC)....1 A CONTRACTOR...2 M KT VENDOR.....3 LOCAL PEOPLE...4 OTHER(specify..5 V SS FOREST.....1 OTHER FOREST...2 OWN LAND.......3 OTHER(specify).4 UNIT CODES: KILO.....1 LITER....2 DOZEN....3 CARTS....4 BUNDLE...5 GRAMS....6 POLES....7 BUNCH....8 PACKS....9 OTHER...10

PAGE 208

208 AGRICULTURE MODULE PART A: LANDHOLDINGS 1. Currently, do any one in your family own agriculture land, or forestland, or crops, or wells, lakes and so on? 2. How many acres of land do your family own? 3. Do your family has any land without proper legal documents? 4. What is the extent of such land? 5. What is the extent of leased land? 6. What is the extent of leased out land? 7. What is the extent of irrigated land? 8. Previous year, what is the extent of cultivated land? 9. What is the market value of land owned by your family? ACRES YES.1 NO.2 ACRES ACRES ACRES ACRES ACRES RUPEES AGRICULTURE MODULE PART B: FARM CAPITAL INVENTORY 1. Do your family own (.)? 2. How many do your family own (.)? 3. Do your family have any (.) joint ownership? 4. How many do your family have (.) joint ownership? 5. What is your family's share in joint ownership of (.)? 6. If you sell it (.) now, what price you will receive? 7. Previous year, has your family derived any income by renting it (.)? 8. What is the income? CODE TYPE OF FARM EQUIPMENT YES NUMBER YES.1 NO..2 () NUMBER PERCENT RUPEES YES.1 () NO..2 RUPEES 1 Tractor (include implements) 2 Animal pulled plow (Nagali) 3 Pump-set/Submersible 4 Sprinkler/Drip irrigation 5 Thresher/Harvester 6 Rice/Flour Mill 7 Chopper (to process feed) 8 Power sprayer 9 Hand sprayer 10 Ox cart YES..1 NO...2 IF YES, PUT A CHECK MARK IN THE YES BOX FOR EACH ITEM, THEN ASK Q. 2 4. I F NO, GO TO PART B BELOW

PAGE 209

209AGRICULTURE MODULE PART C: DISPOSITION OF CROPS 1. What are the crops cultivated previous year? 2. What is the yield of (.) in the last 12 months? 3. What quantity of () produced in last 12 months, you have sold? 4. What is the price received for the () sold by you? ID CROP NAME YES QUANTITY UNIT CODE QUANTITY UNIT CODE RUPEES UNIT CODE 1 Vari (paddy) 2 Jonnalu (sorghum) 3 Mokkajonnalu (maize) 4 Kandi (redgram) 5 Chanaga (Bengal gram) 6 Pesalu (Green gram) 7 Palli (Verushanaga) (groundnut) 8 Minumulu (Blackgram) 9 Sunflower 10 Sugarcane (Cheraku) 11 Cotton (Patti) 12 Ithara Pantalu (other crops, specify) 13 14 15 16 17 18 19 20 21 PROMPT RESPONDENT FOR RE C ALL AND/OR CROPS NOT LISTED IF 0, NEXT CROP UNIT CODES: KILO.....1 LITER....2 DOZEN....3 CARTS....4 BUNDLE...5 GRAMS....6 POLES....7 BUNCH....8 PACKS....9 QUINTAL.10 BALES...11 OTHER...12

PAGE 210

210 AGRICULTURE MODULE PARTS D and E: INPUT PURCHASES AND LIVESTOCK 1. Did your family purchased (input) in the last 12 months? 2. In last 12 months, how much did you purchase ()? ( what is the total cost) 1. Any member of your family owns ()? 2. How many does your family own ()? 3. If you sell this one () what price does it fetch you? 4. Are you deriving any income from this () (like milk, eggs and so on). 5. What is the income derived from () in the last two weeks? ID INPUT DESCRIPTION YES RUPEES ID ANIMAL YES NUMBER RUPEES YES.1 ( 5) NO..2 RUPEES 1 Fertilizer #1 Urea 1 Gedelu (Buffaloes) 2 Fertilizer #2 D.A.P. 2 Aavulu (cows) 3 Fertilizer #3 M.O.P. 3 Yeddulu (oxen) 4 Fertilizer #4 S.S.P. 4 Pandulu (Pigs) 5 Fertilizer #5 20:20:0:15 5 Gorrelu (Sheep) 6 Manure FYM 6 Mekalu (Goat) 7 Pesticide #1 Monocrotophos 7 Kollu (Chicken) 8 Pesticide #2 Quinolphos 8 Bathulu (Ducks) 9 Pesticide #3 Chloropyriphos Ithara Jantuvulu (other animals) 10 Pesticide #4 Endosulfon 11 Herbicide #1 Butachlor 12 Herbicide #2 2,4-D 13 Fungicide #1 Bavistin 14 Fungicide #2 Thiram/Captan 15 Bio-fertilizers Rhizobium/PSB UNIT CODES: KILOGRAM.........1 LITER............2 CARTLOAD.........3 OTHER (Specify)..4 IF YES, MARK ALL INPUT PURCHASES BEFORE ASKING QUESTION 2. LIST ALL ANIMALS BEFORE ASKING Q. 2 5 FOR EACH ANIMAL

PAGE 211

211 EMPLOYMENT MODULE [REFER TO HOUSEHOLD ROSTER] In this module, few questions to the family head on the various works and the family members engaged in it. If any member were employed outside, I would like to talk to the member now. If it is not possible for the member to talk to me I would like to talk to any other member of the family who can provide details of employment of such member. WRITE THE HOUSEHOLD MEMBER ID CODE IN THE FIRST COLUMN, IF THE PERSONS OCCUPATION CODE (FROM Q.10 OF THE HOUSEHOLD ROSTER) EQU ALS 3, 4, 5, OR 6. THEN WRITE THE PERSONS JOB DESCRIPTION NEXT TO THEIR ID CODE. IF A PROXY RESPONDENT IS ANSWERING THIS SECTION, INDICATE THEIR PERSONAL ID CODE HERE: 1. IF THE PERSON HAS MORE THAN ONE TYPE OF JOB, INDICATE THIS BY MARKING THEIR ID CODE IN COLUMN 1 FOR EACH JOB DESCRIPTION LISTED. THEN PROCEED TO ASK QUESTIONS 3 9 FOR EACH JOB LISTED. 2. IS THIS PERSON ANSWERING FOR HIMSELF OR HERSELF? 3. Last year for how many months you have worked in this job? 4. What is the salary per month you received for this job? 5. How many days in a month do you work? 6. Normally what is your earning per day? 7. Last year have you received any payment not in cash form 8. What is the value of payment received in noncash mode? 9. IF THE RESPONDENT RECEIVES NO TYPE OF PAYMENT FOR WORK, THEN ASK THIS QUESTION : What is the reason for not receiving any kind of payment for your work? ID CODE JOB DESCRIPTION YES..1 NO...2 MONTHS RUPEES DAYS PER MONTH RUPEES PER DAY YES..1 NO...2 RUPEES CODE (Specify if CODE = 5) NON-PAYMENT CODES: APPRENTICESHIP O R UNPAID TRAINEESHIP....1 LABOR EXCHANGE...........2 PAYING OFF A DEBT........3 WORKING FOR A RELATIVE...4 O THER (S p ecif y )..........5 IF NO: 9 IF NO MONEY WAGE IS COLLECTED: 7

PAGE 212

212HOUSEHOLD ENTERPRISES MODULE PART A: GENERAL INFORMATION In last one year any one in your family ow ned a shop or ran/ managed business enterprise? Previous year any one in your family op erated business enterprise (such as sculpting, tailoring, mechanic) not related to Manufacturing, service sector, or agriculture IF THE ANSWER TO EITHER QUESTION ABOVE IS YES, PROCEED TO QUESTION 1 BELOW. IF BOTH ANSWERS ARE NO, INFORM THE RESPONDENT THAT YOU WILL NOW PROCEED TO THE FINAL SECTION. PROCEED TO THE REMITTANCES MODUL E. 1. I will be asking some important questions related to the business operated by your family. If it is convenient, I would like to talk to the person who is familiar with day to day operations of the enterprises. Please tell what kind of enterprise is run by the family. 2. Are you the sole owner of this enterprise? 3. What is your family's share in this enterprise? 4. Who manages this enterprise? 5. IS THE RESPONDENT THE MANAGER? 6. How many days in month do you work to manage this enterprise? 7. In normal month, how many days in a month does your enterprise work? 8. In normal month how many members of your family are employed in the enterprise? 9. In a normal month, how many persons are employed who are not members of your family? 10. Since how many years this enterprise is in operation? 11. Is this enterprise started with loan component? ENT. CODE ENTERPRISE DESCRIPTION YES..1 NO...2 PERCENT NAME YES..1 NO...2 DAYS DAYS NUMBER NUMBER YEARS YES..1 NO...2 1 2 3 4 5 6 YES.1 NO..2 YES.1 NO..2 IF NO: 7 IF YES: 4 IF THE MANAGER BELONGS TO ANOTHER HOUSEHOLD: WRITE 7 BRIEFLY INDICATE THE TYPE OF ENETRPRISE OWNED BY THE HOUSEHOLD (E.G., METALWORKING)

PAGE 213

213 HOUSEHOLD ENTERPRISES MODULE PART B: REVENUES The following questions are related to income from your occupat ional business enterprise. Please furnish gross income (i.e., in clusive of product price or service cost received in cash and or in kind form). Do not exclude expenses on operating business, or expenses for family. 1. In a year, how many months do your business has higher sales? 2. What is the monthly income in months of higher sales? 3. In a year, how many months do your business has normal sales? 4. What is the monthly income in months of normal sales? 5. In a year, how many months do your business has no sales? 6. Previous year, in business have you received any payments in kind or in the form of service? 7. In months with normal sales, what is the value of payments received not in cash (i.e., in kind form or service form)? 8. Whether your family consumed products or services of your enterprise in previous year? 9. In a month with normal sales, what is the monthly consumption of your family of product / service of this enterprise? ENT. CODE NUMBER OF MONTHS RUPEES NUMBER OF MONTHS RUPEES NUMBER OF MONTHS YES..1 NO...2 RUPEES YES..1 NO...2 RUPEES 1 2 3 4 5 6 IF NO: 8

PAGE 214

214 HOUSEHOLD ENTERPRISES MODULE PARTS C and D: INPUT COSTS AND BUSINESS ASSETS PART C: INPUT COSTS In this section, questions will be on investments made on enterprise. PART D: BUSINESS ASSETS Lastly, few questions in brief to know details of assets used for operating household enterprise. 1. Last year, how much you have invested on all inputs (include all such as labor costs, raw materials, transport, electricity, water, fuel, rent, tax, registration fee, insurance, loans, operation costs and so on). 2. In a month with normal sales what is the total expenses on all inputs? 1. Have you acquired any of following assets for household enterprise. Please answer yes / no. Please take care not to include assets already mentioned in consumption or agriculture module. NOTE: PROMPT THE RESPONDENT TO MAKE SURE THAT NO ITEMS ARE DOUBLECOUNTED IF YOU SUSPECT THIS PROBLEM. FOR EXAMPLE, A TRACTOR ALREADY LISTED AS AN AGRICULTURAL ASSET SHOULD NOT BE INCLUDED HERE. THE SAME APPLIES TO A BICYCLE LISTED IN PART E OF THE CONSUMPTION MODULE. 2. What is the market value of this asset? 4. Have you acquired any of the following assets in the past oneyear? 5. What is the total amount spent to purchase Household enterprise assets? 6. Have you sold any of these assets in the past oneyear? 7. What is the amount received for selling enterprise asset? ENT. CODE RUPEES RUPEES ASSET YES RUPEES 3. Is your family sole owner of this asset or having joint ownership with other family? YES.1 NO..2 RUPEES YES.1 NO..2 RUPEES 1 Building/Shop 2 Small vehicles (bicycle, cart, etc.) 3 Large vehicles (truck/car/boat/etc.) 4 Equipment and machinery 5 Tools 6 Other (Specify) SOLE OWNER..1 SHARED ITEM...2

PAGE 215

215 REMITTANCES MODULE PART A: EXPENDITURES ON INTER-HOUSEHOLD TRANSFERS 1. In last 12 months, any one in your family has given some amount to other member of the family. For ex: to relatives living at other places, or for children's education or friends. PART B: INCOME FROM INTER-HOUSEHOLD TRANSFERS 1. Last year, give names of persons who have helped your family? 2. In the last 12 months, give name and details of person whom your family members helped. 3. What is the relationship of this person with head of the family? 4. In the last 12 months, what is the amount given to this person by family members? 2. Last year, give names of persons who have helped your family? 3. What is the relationship of this person with head of the family? 4. Last year, what is the amount received by your family from this person? ID NAME RELATIONSHIP RUPEES ID NAME RELATIONSHIP RUPEES 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 YES NO..2 IF NO ( PART B) YES NO..2 IF NO: END THE INTERVIEW BY THANKING THEM (ENTHUSIASTICALLY) FOR THEIR TIME AND CO-OPERATION.

PAGE 216

216Informed Consent Protocol Title: Economic Analysis of Joint Forest Management (JFM) in In dia: Impacts on Poverty and Inequality in Andhra Pradesh. Please read this consent document carefully befo re you decide to participate in this study. Purpose of the research study: This research will study the participatio n of rural communities in Joint Forest Management (JFM), and how JFM provides benefits to households within villages of Andhra Pradesh. What you will be asked to do in this study: I would like to ask you some questions regarding your household welfare and livelihood, and your participation in the forest management program in your community. You do not have to answer any question if you do not want to. If you would p refer, I can schedule the interview for another time. Time required: Approximately one hour. Risks and Benefits: There are no direct risks or benefits to you for participating in this study. Compensation: There is no compensation for participating in this study. Confidentiality: Your identity will be kept confidential to the extent provided by law in any report produced. Only the principal investigator and interviewer will know your identity. Voluntary participation: Your participation in this study is completely. There is no penalty for not participating. Right to withdraw from the study: You have the right to withdraw from the study at any time without consequence. Whom to contact if you have questions about the study: Whom to contact about your rights as a participant in this study: Supervisor: Principal Investigator: UFIRB Office Dr. Janaki Alavalapati, Associate Professor Frederick Rossi, PhD student Po Box 112250 School of Forest Resource & Conservation Food and Resources Economics Department University of Florida P.O. Box 110410 PO Box 110240 Gainesville, FL USA 32611-2250 Gainesville, FL USA 32611-0410 Gainesville FL USA 32611-0240 Phone: 352-392-0433 Ph: 352-846-0899; Fax: 352-392-1707 Ph: 352-392-1826 ext. 428 e-mail: janaki@ufl.edu e-mail: frossi@ufl.edu Agreement: I have read the procedure described above. I voluntarily agree to participate in th e procedure and I have received a copy of t his description. Participant signature: ____________________________________________ Date: Interviewer signature: ____________________________________________ Date: Principal Investigator signature: ____________________________________ Date: A pproved B y University of Florida Institutional Review Board 02 Protocol # 2004-U-890 For Use Through 11/17/2005

PAGE 217

217 LIST OF REFERENCES Agrawal, A. Common Property Institutions and Sustainable Governance of Resources. World Development 29(2001):1649-1672. Allanson, P., and L. Hubbard. On the Compar ative Evaluation of Agricultural Income Distributions in the European Union. Dund ee Discussion Papers in Economics: Working Paper No. 93. Dept. of Economic Studies, Un iversity of Dundee. Dundee, Scotland, 1998. Asia Forest Network. Community-Based Natural Regeneration Workshop Proceedings pp. 1-19. Regional Field Workshop, Andhra Pradesh, India, 11-15 March 2002. Andhra Pradesh Forest Department. Current Status of Implementation of JFM in Andhra Pradesh. Andhra Pradesh Forest Depart ment website. Hyderabad, India, 2006. Accessed: September 2006. http://www.ap.nic.in/apforest/JFM%20CFM/CURRENTSTATUS.htm Andhra Pradesh Forest Department. Andhra Pradesh Community Forest Management Project: Project Implementation Plan, Vol. 1. Government of Andhra Pradesh. Hyderabad, India, n.d. Obtained from Andhra Pradesh Forest Department, January 2005. Available at: http://forest.ap.nic.in/JF M%20CFM/CFM/PIP/01_PIP.htm Balaji, S. Forest Policy in India In Retrospe ct and Prospect. Paper presented at the IUFRO Science/Policy Interface Task Force regiona l meeting. Chennai, India, 2002. Chagari, S. Information Capability Building: Role of Information Literacy Programmes A Study. World Library and Information Congr ess: 71th IFLA General Conference and Council. Oslo, Norway, August 14-18, 2005. Accessed: June 2006. http://www.ifla.org/IV/ifla 71/papers/043e-Chagari.pdf Bray, D., L. Merino-Prez, P. Negreros-Castillo G. Segura-Warnholtz, J. Torres-Rojo, and H. Vester. Mexico's Community-Managed Fore sts as a Global Model for Sustainable Landscapes. Conservation Biology 17(2003): 672-677. Bray, D., C. Antinori, and J. Torres-Rojo. The Mexican Model of Community Forest Management: The Role of Agrarian Policy, Forest Policy and Entrepreneurial Organization. Forest Policy and Economics 8(2006):470-484. Deaton, A. Understanding Consumption Clarendon Lectures in Economics. Oxford University Press. Oxford, England, 1992. Deaton, A. The Analysis of Household Surveys: A Microeconometric Approach to Development Policy The World Bank, Johns Hopkins University Press. Baltimore, MD, 1997. Deaton, A. and M. Grosh. Consumption in Designing Household Survey Questionnaires for Developing Countries: Lessons from 15 Year s of the Living Standards Measurement

PAGE 218

218 Study pp. 91-133. Edited by M. Grosh and P. Glewwe. The World Bank, Washington DC, 2000. Deaton, A. and S. Zaidi. Guidelines for C onstructing Consumption A ggregates for Welfare Analysis. Living Standards Measurement Study, Working Paper No. 135, World Bank, Washington DC, 1999. Fowler, F. Survey Research Methods (3rd edition). Applied Social Research Methods Series. Sage Publishing Company. Thousand Oaks, CA, 2002. Government of Andhra Pradesh. Andhra Pradesh Community Forest Management Project: Executive Summary, Social and Environmental Assessment Government of Andhra Pradesh. Hyderabad, India, 2002. Obtained from Andhra Pradesh Forest Department, January 2005. Available at: http://forest.ap.nic.in/JFM%20CFM/CFM /PIP/02_SEA/01_Executive%20Summary.htm Government of Andhra Pradesh. Comprehensive Orders on Community Forest Management: G.O.Ms. No. 13. Government of Andhra Pradesh. H yderabad, India, 2004. Obtained from Andhra Pradesh Forest Departme nt, January 2005. Available at: http://forest.ap.nic.i n/JFM%20CFM/GO13-2004.htm Grosh, M. and P. Gle wwe. Introduction in Designing Household Survey Questionnaires for Developing Countries: Lessons from 15 Year s of the Living Standards Measurement Study pp. 5-19. Edited by M. Grosh and P. Glewwe. The World Bank, Washington DC, 2000. Grosh, M., P. Glewwe, and J. Muoz. Des igning Modules and Assembling Them into Questionnaires in Designing Household Survey Questionnaires for Developing Countries: Lessons from 15 Years of the Living Standards Measurement Study pp. 43-74. Edited by M. Grosh and P. Glewwe. The World Bank, Washington DC, 2000. Grosh, M. and J. Muoz. MetadataInformati on about Each Interview and Questionnaire in Designing Household Survey Questionnaires for Developing Countries: Lessons from 15 Years of the Living Standards Measurement Study pp. 77-90. Edited by M. Grosh and P. Glewwe. The World Bank, Washington DC, 2000. Gujarati, D. Basic Econometrics (3rd edition). McGraw-Hill, Inc. New York, NY, 1995. Hardin, G. The Tragedy of the Commons. Science 162(1968):1234-8. Heltberg, R. Determinants and Impact of Local Institutions for Common Property Management. Environment and Development Economics 6(2001):183-208. Hildyard, N., P. Hedge, P. Wolvekamp, and S. Reddy. Same platform, different train: the politics of participation. Unasylva 194(1998). Accessed: September 2004. http://www.fao.org/docrep/w8827E/w8827e06.htm

PAGE 219

219 Hill, I. and D. Shields. Incentives for Joint Fo rest Management in Indi a: Analytical Methods and Case Studies. World Bank Techni cal Paper No. 394, Washington DC, 1998. Jayashanker. K. Telangana: Education for th e Doubting Toms. from Telangana RashtramOka Demand, 2nd Printing: March 13, 2004. Accessed: June 2006. http://www.telangana.org/TelanganaFAQ.asp Joshi, A.L. Community Forestry in Nepal ( 1978 to 2010): Forest Policy. Government of Nepal, Ministry of Forests and Soil Cons ervation. Kathmandu, Nepal, 1997. Accessed: June 2006. http://www.mtnforum.org/oldocs/164.doc Khare, A., M. Sarin, N.C. Saxena, S. Palit, S. Bathla, F. Vania and M. Satyanarayana. Joint Forest Management: Policy, Practice and Prospects Policy that Works for Forests and People Series No. 3. World Wide Fund for Nature, New Delhi and International Institute for Environment and Development, London, 2000. Kumar, S. Does Participati on in Common Pool Resource Management Help the Poor? A Social Cost-Benefit Analysis of Joint Fo rest Management in Jharkhand, India. World Development 30(2002):763-782. Mitra, K. Dynamics of Forest Product Consumption: The Role of Economic Development and Forest Policies in West Bengal, India. Ph.D. dissertation, University of Florida. Gainesville, FL, 1995. Misra, D. and S. Kant. Production Analysis of Collaborative Forest Management Using an Example of Joint Forest Manage ment from Gujarat, India. Forest Policy and Economics 6(2004):301-320. Ministry of Environment and Forests. State of the Environment Report: India 1999 Government of India, Ministry of Environment and Fore sts. New Delhi, India, 1999. Accessed: June 2006. http://www.envfor.nic.in/soer/1999/soer.html Ministry of Environment and Forests. Proceedings: National Consultative Workshop on Joint Forest Management (JFM) pp. 1-42. July 14-15, 2005. Edited by J. Kishwan, S. Datta, R. Pai, and S. Bose, India Habita t Centre. New Delhi, India, 2005. Ministry of Environment and Forests. Report of the National Forest Commission 2006 Government of India, Ministry of Envir onment and Forests. New Delhi, India, 2006. Accessed: June 2006. http://www.envfor.nic.i n/divisions/nfr.html Odedokun, M., and J. Round. Determinants of Income Inequality and its Effects on Economic Growth: Evidence form African Countries . Discussion Paper No. 2001/103. World Institute for Development Economics Res earch. United Nations University, 2001.

PAGE 220

220 Odera, J. Lessons Learnt on Community Forest Management in Africa. Report prepared for the project: Lessons Learnt on Sustainable Fore st Management in Africa, 2004. Accessed: November 2006. http://www.afornet.org/images/pdfs/ Community%20forest%20management.pdf North, D. Institutions, Institutional Change and Economic Performance Cambridge University Press. Cambridge, England, 1991. Ostrom, E. Governing the Commons: The Evolution of Institutions for Collective Action Cambridge University Press. Cambridge, England, 1990. Rangachari, C. and S. Mukherji. Old Roots, New Shoots: A Study of Joint Forest Management in Andhra Pradesh. Winrock International/Ford Founda tion. New Delhi, India, 2000. Reddy, V., M. Shekar, P. Rao, and S. Gillesp ie. Nutrition in India. UN ACC/SCN Country Case Study, Prepared for the XV Internat ional Congress of Nutrition, Adelaide, Australia, 26 September 1 October, 1993. December 1992. Accessed: June 2006. http://www.unsystem.org/scn/archives/india/ch16.htm Reddy, S.R.C. and S.P. Chakravarty. Fores t Dependence and Income Distribution in a Subsistence Economy: Evidence from India. World Development 27(1999):1141-1149. Reddy, V., M. G. Reddy, V. Saravanan, M. Ba ndi, and O. Springate-Baginski. Livelihoods Impact of PFM in AP. Centre for Economic and Social Studies (CESS) Working Paper. Hyderabad, India, 2004. Saxena, N. The Saga of Participatory Forest Management in India CIFOR Special Publication, Center for International Forestry Research. Jakarta, Indonesia, 1997. Schmink, M. Communities, Forests, Markets, and Conservation in Working Forests in the Neotropics pp. 119-129. Edited by D. Zarin, J. Alavalapati, F. Putz, and M. Schmink. Columbia University Press. New York, NY, 2004. StataCorp. Stata Statistical Software: Release 9 StataCorp LP. College Station, TX, 2005. Tata Consultancy Services. Spirit of AP we bpage, AP Online Website, 2002. Accessed: June 2006. http://www.aponline.gov.in/quick%20 links/apfactfile/a pfactmain.html Tewari, D. The Effectiveness of State Fore st Development Corporat ions in India: An Institutional Analysis. Forest Policy and Economics 8(2006):279-300. United Nations. World Population Prospect s: The 2004 Revision and World Urbanization Prospects: The 2005 Revision. Urbanization Webpage for 2005. Population Division of the Department of Economic and Social Affair s of the United Nations Secretariat, 2005. Accessed: June 2006. http://esa.un.org/unup

PAGE 221

221 Vanhoudt, P. An Assessment of the Macroeconomi c Determinants of Inequality. Stockholm School of Economics: Working Paper Seri es in Economics and Finance, No. 271. Stockholm, Sweden. October, 1998. Accessed: September 2004. http://swopec.hhs.se/ha stef/papers/hastef0271.pdf World Bank. India: Alleviating Poverty Through Forest Development Evaluation Country Case Study Series. World Bank, Washington DC, 2000.

PAGE 222

222 BIOGRAPHICAL SKETCH A native of Michigan, Frederick Rossi grew up w ith a keen interest in maps and the natural world. Educated in geological sciences at Mi chigan State University, he soon moved to the western United States in search of mountains. After working for the man for several years, he enrolled in the graduate program at Colorado St ate University in order to pursue interests in economics and foreign countries. After obtaining a masters degree in agricultural and resource economics, he moved to Florida to continue the study of both development economics and natural resource economics. Despite the limited snowboarding opportunities in Florida, benefits of the move to this state have included home ownership, finding a wife, and a passage to India.


Permanent Link: http://ufdc.ufl.edu/UFE0017574/00001

Material Information

Title: Socio-economic impacts of community forest management in rural India
Physical Description: Mixed Material
Creator: Rossi, Frederick John ( Author, Primary )
Publisher: University of Florida
Place of Publication: Gainesville, Fla
Publication Date: 2007
Copyright Date: 2007

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0017574:00001

Permanent Link: http://ufdc.ufl.edu/UFE0017574/00001

Material Information

Title: Socio-economic impacts of community forest management in rural India
Physical Description: Mixed Material
Creator: Rossi, Frederick John ( Author, Primary )
Publisher: University of Florida
Place of Publication: Gainesville, Fla
Publication Date: 2007
Copyright Date: 2007

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0017574:00001


This item has the following downloads:


Full Text

SOCIO-ECONOMIC IMPACTS OF COMMUNITY FOREST MANAGEMENT IN RURAL
INDIA





















By

FREDERICK JOHN ROSSI


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

UNIVERSITY OF FLORIDA

2007

















Copyright 2007

by

Frederick John Rossi

















I dedicate this to my parents, Joseph and Irene Rossi, for the unconditional love and enduring
support they have always given me. I also wish to dedicate this work to my uncle and godfather,
David Rossi. His words once challenged me to action, which ultimately lead to this dissertation.
He will be missed always.





ACKNOWLEDGMENTS

I give thanks to God for granting me the ability, the opportunity, and the means for

pursuing and completing this degree, and for safely conducting the fieldwork in India. I also

thank my wonderful wife, Rosa Cossio, for her patience, support, and encouragement when I

needed it most. Her presence in my life has given me great joy, and has kept my spirit aloft even

during times of mental and physical stress.

I thank my parents for always being there to help me improve myself; without their support

during the years my education would not have been possible. More importantly, however, I

thank them for the positive influence that they have had on my life and on the lives of my

brothers and sister. I also thank my siblings for their encouragement and for the examples that all

three of them have set for me in their own different ways.

I would also like to take this opportunity to thank the members of my committee: Ridwan

Ali, Robert Emerson, and Stephen Perz. Their guidance, valuable suggestions, and general

support were crucial to the improvement and completion of this study. I also wholeheartedly

thank the co-chairs of my committee: Janaki Alavalapati and Sherry Larkin. Janaki provided me

with much knowledge about India and its people and culture, as well as important contacts in

Andhra Pradesh. More importantly, he was always there with both guidance and enthusiasm for

this work.

A special note of appreciation is reserved for Sherry. She has been tremendous in her

support and her willingness to accommodate my all of my questions, concerns, and doubts. I

acknowledge her significant contribution to the final form of this dissertation; her abilities to edit

and re-structure my work will never be under-appreciated, and this dissertation is a testament to

that fact. In addition, her good nature and optimism always buoyed my spirits whenever I felt as





if I hit a dead-end. I am truly grateful for having her support and guidance as both my advisor

and friend.

A special word of thanks is reserved for Vishnu Reddy for all of the effort and dedication

he put into the fieldwork. I also thank Vishnu and his family for their friendship, generosity, and

hospitality; words cannot begin to convey how much I appreciated and enjoyed being a guest in

their home: I thank them from the bottom of my heart. Namaste.

I would like to thank the many other people in India that I befriended and were important

to the furtherance of this research in many different ways. In particular, I want to thank all of my

interviewers; especially Ashok, Kiran, Suresh, and Sugunakar: this work would not have been

possible without their dedication and effort. For their hospitality, conversation, and willingness

to help me conduct this research, I also want to thank members of the Andhra Pradesh Forest

Department; especially Ramesh Kalaghatgi, P. Malikarrjuna Rao, P.V. Padmanabham,

Siddhanand Kukrety, M. Waheed, Vinod Kumar, A.S. Rao, Raj Rao, and Prabakar. I would also

like to thank Gopinath Reddy at the Center for Economic and Social Studies (CESS) in

Hyderabad for both his insights and valuable assistance in helping me to hire my staff.

Finally, I would like to thank Doris Capistrano and the Center for International Forestry

Research (CIFOR). Their financial contribution helped to support the fieldwork that is the basis

of this work; for this I am deeply appreciative of their generosity.





TABLE OF CONTENTS

page

ACKNOW LEDGM ENTS ........................................ ............................................................4

LIST OF TABLES.........................................................................................................................9

LIST OF FIGURES .................................................................. ...........................................10

ABSTRACT............................................................................................................ ...............................11

CHAPTER

1 INTRODUCTION: PARTICIPATORY FOREST MANAGEMENT.................................. 13

Challenges of Resource M anagem ent .......................................................... .................. 13
The Role of Institutions ........................................................................................................ 16
A Case Study in India................................................ ........................................................20
Forest M anagem ent .................................................................................................... 20
Research Objectives .................................................................. .............................23
M methods .............................................................................................. .......................... 24
Sum m ary and Preview ..........................................................................................................26

2 BACKGROUND INFORMATION ON THE STUDY AREA ............................................. 28

Overview .................................................................. ..................................... ........................28
History of Participatory Forest M anagem ent in India............................................................ 29
Pre-Colonial Era .............................................................................................. ........... 29
Social Forestry Program s .......................................................................................... 31
The Arabari Experim ent ................................................................................................ 33
Developm ent of Joint Forest M anagem ent (JFM )..................................................34
Advent of Com m unity Forest M anagem ent (CFM )......................................................38
Com prison to Participatory Forestry in Other Nations.........................................................40
N ep al....................... ...... ................................................................ ............ .............. .. 4 0
China and Other Asian Nations....................................................................................40
Africa...........................................................................................................................41
M exico ............................................... ......................................................... ................42
Description of Andhra Pradesh...............................................................................................44
Location................................................... ............ ..........................................................44
Languages and Castes..........................................................................................46
Geographic/Political Regions.........................................................................................47
The Forest Departm ent..................................................................................................50
Sum m ary......................................................................................................................................52

3 QUESTIONNAIRE DEVELOPMENT AND SURVEY ADMINISTRATION .................55

Introduction......................................... .. .....................................................................55





The V SS Questionnaire ..........................................................................................................55
The H household (HH) Questionnaire................................................................................... 56
The LSM S M odules .......................................................................................................56
Design and Structural Guidelines............. ................................. .........................61
Form atting Guidelines ............................................................................................... 63
Survey Structure and Im plem entation................................................................................. 64
HH Questionnaire Testing and Revision......................................................................... 64
Sam ple Selection ........................................................................................................... 65
Interviewer Training and Implementation of the HH Survey .........................................68
Im plem entation of the V SS Survey...............................................................................69
Data Com pilation.......................................................................... .................................73

4 M ODELS, HYPOTHESES, AN D D ATA ......................................................................76

Overview ............................. ............................... ....................... .......................................76
Em pirical M odels................................... ..................... ................... ...........................77
Social W welfare Function..................................................................................................79
Average Consum ption M odel.......................... ...................... .......................................80
Consum ption Inequality M odel..................................................................................... 81
Consumption-based Poverty Model ....................................................................................83
Hypotheses for the Explanatory Variables ........................................ ..................85
Dem graphic Variables ...................................................................................................85
Economy ic Variables........................................................................... ........................87
Bio-physical Variables .................................................................... ...................... 88
Institutional Variables ....................................................................................................90
Description of the Data................................................................. .......................................92
The Adilabad Sam ple ................................................................................................94
The Visakhapatnam Sam ple .......................................................................................... 100
The Chittoor W est Sam ple .............................................. ......... ...............................104
Sum m ary ............................................................................................................................... 109

5 RESU LTS AN D DISCU SSION ...........................................................................................110

Introduction............................................................................................................................110
Economic Models ........................................................................................................................ 112
Average Consum ption M odel......................................................................................112
Consum ption Inequality M odel................................................... .............................117
Poverty-gap Ratio M odel ........................... .............................................................. 18
Forest Quality Change M odel........................................................................................121
Discussion of the Effect of Tim e under JFM /CFM .............................................................127
Average Consum ption M odel......................................................................................128
Forest Quality Change M odel ..................................................... ............................132
Sum m ary ...................................................... .................. ................................... ..... ...... 137

6 AN ALY SIS OF CFM IN STITU TION S.................................................. .........................139

Introduction............................................. ......... ...............................................................139





DP 1: Clearly Defined Boundaries .....................................................................................141
Spatial Boundary Awareness................................................. ....................................142
Awareness of Access Regulations..............................................................................144
General Awareness of Institutions................................................................................147
DP 2: Congruence between Rules and Local Conditions............... .........................149
DP 3: Collective-choice Arrangements .............................................................................149
DP 4: M onitoring....................................... ...................................................................... 153
DP 5: Graduated Sanctions ....................................................................... .....................155
DP 6: Conflict-resolution M echanisms ............................................................................156
DP 7: M inimal Recognition of Rights to Organize ...........................................................157
Summary............................................................. .........................................................................159

7 CONCLUSIONS, POLICY IMPLICATIONS, AND FUTURE WORK ............................160

Overview............................................................................................................................................. 160
Summ arized Conclusions ..................................................................................................160
Regression Analysis ....................................................................................................160
Institutional Analysis...................................................................................................163
Policy Implications .......................................................................................................... ........ 167
Recommendations from the Economic Analysis ............................. .........................167
Recommendations for Training and Extension............................................................170
Recommendations from the Institutional Analysis ....................................................... 173
Caveats of the Study ............................................. ............. ................................................177
Survey................................................................. ....................................... ................177
D ata ........................................................................................................ ........................ 17 8
Future W ork ...................................... ....................................................................... .............179

APPENDIX

A THE VSS QUESTIONNAIRE.........................................................................................183

B THE HOUSEHOLD QUESTIONNAIRE................................... .................................. 190

LIST OF REFERENCES ........................................................................................................... 217

BIOGRAPHICAL SKETCH .............................. ................ ............................................. 222





LIST OF TABLES


Table page

3-1 Composition of the Household Questionnaire................................................................. 57

4-1 Explanatory variables utilized in the VSS-level regressions.............................................77

4-2 Mean and standard deviation for VSS-level data: comparison by the region sampled.....93

4-3 Selected demographic and economic variables for the Adilabad sample........................95

4-4 Selected bio-physical and institutional variables for the Adilabad sample .....................98

4-5 Selected demographic and economic variables of the Visakhapatnam sample.............101

4-6 Selected bio-physical and institutional variables for the Visakhapatnam sample .........103

4-7 Selected demographic and economic variables for the Chittoor West sample..............105

4-8 Selected bio-physical and institutional variables for the Chittoor West sample. ..........108

5-1 Basic statistics for VSS-level regression analyses (58 observations).............................111

5-2 Estimated regression coefficients and associated p-values...........................................112

5-3 Actual versus predicted Forest Quality Change (FQC) values.......................................123

5-4 Estimated marginal effects and p-values on categorical probabilities...........................125

5-5 Probabilities of selection and marginal effects of LT....................................................133

6-1 V SS boundary aw areness..........................................................................................143

6-2 Access awareness of VSS-member households by caste.................................................145

6-3 HGAI for VSS-member households. .......................................................................148

6-4 Response frequency indicating that the VSS Micro-plan reflects the interests of:..........151

6-5 Response frequency regarding elections for the VSS Managing Committee................153

6-6 Relationship between the VSS Management Committee and their APFD field officer..159




LIST OF FIGURES


Figure page

2-1 Map of India with Andhra Pradesh (shaded). ................................................................45

2-2 D district m ap of Andhra Pradesh.........................................................................................48

5-1 Actual gt (dots) and predicted t (line) in Rupees, versus LT...........................................129

5-2 Actual pt (dots) and predicted [t (line) in Rupees, versus LT: using a subset of VSS
w here PFC > 4.5 (n = 26). ..........................................................................................131

5-3 Predicted probabilities of FQC values as a function of LT. ..........................................134





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

SOCIO-ECONOMIC IMPACTS OF COMMUNITY FOREST MANAGEMENT IN RURAL
INDIA


By

Frederick John Rossi

May 2007

Chair: Sherry L. Larkin
Cochair: Janaki Alavalapati
Major Department: Food and Resource Economics Department

This work provides an economic and institutional analysis of Joint Forest Management

(JFM) as applied in Andhra Pradesh, India. JFM is a natural resource management paradigm that

actively involves local stakeholders in the protection and management of local forest resources.

In essence, JFM is an institutional framework in which state forest departments work in

partnership with local communities to restore degraded forests to a productive and sustainable

capacity. Concomitant goals of this program are to increase the incomes of rural participants, and

provide an equitable distribution of program benefits throughout local communities. In 1992,

Andhra Pradesh enacted legislative orders adopting JFM; these orders were subsequently revised

in 2002 to authorize Community Forest Management (CFM), a more devolved version that

further involves local communities in the decision-making process of forest management.

Primary data collected from three regions of Andhra Pradesh are used to analyze

community-level impacts that this forest management program has on indicators of economic

well-being, inequality, and poverty. Results suggest there is an appreciable opportunity cost of

adopting the program when all 58 of the sampled villages are analyzed. When restricting the

analytical focus to more forest-dependent villages, results indicate the JFM/CFM program has a




beneficial impact on economic well-being. This implies the program is successful in achieving

one of its objectives: increasing economic benefits for the forest-dependent poor.

The CFM institutions of Andhra Pradesh are evaluated using design principles that Ostrom

identified as key institutions common to successful, long enduring common-pool resource

management regimes. Taken as a group, all of the design principles are found to exist in the state

legislative orders and at the community-level (to varying degrees). However, some of the CFM

institutions defined by the state only reflect these design principles in general terms, while the

institutions implemented by communities are often incomplete or otherwise deficient.

Policy implications and recommendations drawn from this study are directly applicable to

the CFM program of Andhra Pradesh. This study also provides useful information to those

interested in the performance of CFM in Andhra Pradesh, or others interested in the general

application and performance of JFM in India.





CHAPTER 1
INTRODUCTION: PARTICIPATORY FOREST MANAGEMENT

Challenges of Resource Management

Most natural resource systems are considered to be common-pool resources (CPRs). A

CPR is typically defined by some type of natural resource stock (e.g., a forest or a fishery) from

which individual units (e.g., trees or fish) are extracted. The resource units are rival goods, which

means that use is mutually exclusive (i.e., use by one individual precludes use by another). This

theoretically allows the possibility of excluding specific individuals or groups from benefiting

from the CPR. The ownership of a given natural resource is often held by the government,

although resource users themselves sometimes hold collective ownership. CPRs that lack an

owner or clearly defined ownership (e.g., international fisheries) are often characterized as being

an open access resource, which essentially means there are no regulations governing access,

extraction, etc.

From the use of both renewable resource and game theoretic models of harvesting,

economists are aware of the dynamics and incentives that drive people to over-exploit natural

resources. These include, for example, the lack of regulation, a lack of enforcement, and zero

user-costs borne by each individual. When a government cannot effectively exclude local people

from accessing a natural resource, and if there are no communal rights and responsibilities

governing use, a CPR effectively becomes an open access resource. The fundamental problem of

open access resources will then follow: the benefits gained from resource extraction are realized

individually, while the costs of extraction (in terms of decreased reserves, or environmental

degradation) are shared by all who use the resource. As there is no guarantee that reserves will

remain in the future, or that productivity will be sustained, extraction will proceed as fast as

possible. Therefore, there is no incentive for any individual to practice conservation, because the





resource has no imputed value. The irony is that there is a collective social value to conservation:

if the community of resource users can effectively organize and cooperate, they can establish

stakeholder-devised institutions for the productive management of the shared resource.

The phrase 'tragedy of the commons' is often used to refer to open access over-

exploitation resulting from the (perverse) incentives faced by each individual, which themselves

are a function of misguided or non-existent institutions governing the resource (Hardin, 1968;

Ostrom, 1990). In India, for example, government ownership of forest land, coupled with the

historical policy of commercial timber extraction, allowed little consideration for the needs of

forest-dependent people (Khare et al., 2000). As the government could not effectively exclude

rural inhabitants from accessing public forestland (and with no communal rights and

responsibilities for them to adhere to), the forests became a defacto open access resource. The

resulting local pressures that a community exerts on a natural resource, such as a forest, are

compounded over time by high population growth rates. In India, the impact of deforestation and

forest degradation caused by local inhabitants is significant (Khare et al., 2000).

India is not the only nation facing threats to forest sustainability and the related poverty of

forest-dependent people. Around the world, many developing countries are involved in forest

conservation efforts due to concerns for both the long-term sustainability of forest resources and

the wider socio-economic benefits that healthy forest provide to local communities. Beginning in

the 1970s, the international development community has been aware of the important

interdependence between forests and rural peoples (Odera, 2004). As traditional forest

management policies became increasingly powerless to stem deforestation, different agencies

and donors began trying more "holistic" approaches to forest management that included

collaboration with local communities (Odera, 2004). As such, community-based forest





management eventually emerged as a paradigm for forest conservation and rural development in

many developing nations; although what actually constitutes "community forest management" is

in fact quite fluid and varies widely. Schmink (2004) stresses that the diversity in ecology,

resource endowments, socio-cultural systems, and political and economic histories of each

community (let alone nation) require approaches that are adapted to particular situations.

Interest in forest conservation in particular is now widespread around the world among

both developed and developing countries. This is largely due to visually perceptible

environmental degradation caused by deforestation, such as clear-cuts and their after effects

(e.g., soil erosion, etc.). In addition, the controversial issue of anthropogenic global warming has

recently sparked a renewed interest in forest conservation, including research addressing carbon-

sequestration programs. Concurrent with the interest in forest conservation is the attention that

the development community is giving to decentralized and participatory processes that embrace

local human resources. Empowering local people to be at the forefront of their own development,

in the context of forest conservation and management, has become a paradigm for sustainable

socio-economic and environmental development (Schmink, 2004).

Forest degradation has also been a continuing problem for India, adversely impacting both

the environmental and economic stability of rural communities that depend upon forests (World

Bank, 2000; Balaji, 2002; Tewari, 2006). India is a country of approximately 1.1 billion people

(2005 estimate) that is facing the challenge of forest resource sustainability for approximately

200 to 275 million people that depend wholly, or in part, on income derived from forests

(Saxena, 1997; Khare et al., 2000). The future need for timber, fodder, non-timber forest

products (NTFPs), and other environmental amenities and services will require a resource

management paradigm that ensures productive and sustainable forests. Thus, the challenge of





reversing forest degradation, improving forest resource productivity and management, and

improving rural communities is a daunting task, given that India has 678,333 square kilometers

(km) of forest land, of which an estimated 42% is classified as "open forest", meaning that it is

degraded (Ministry of Environment and Forests [MoEF], 2006).

The Role of Institutions

Institutions are the formal and informal rules and regulations that serve as humanly devised

constraints. They shape human interaction and help structure the incentives of human exchange,

thereby decreasing uncertainty in transactions (North, 1991); and their underlying social

organizations are becoming increasingly relevant for the attainment of socio-economic

development and natural resource sustainability goals. This is particularly so in rural areas where

a lack of cooperation, cohesiveness, and coordination frequently hampers economic progress.

For this reason, strong local organizations are often touted as a key determinant of successful

rural economic development. They are especially important in mitigating the deleterious effects

caused by the over-exploitation of CPRs such as public forests. Institutions often originate

spontaneously in response to the particular needs of a group of individuals. Healthy institutions

(e.g., those that are functioning and appropriate) also grow and evolve to fit the needs and

demands of the society they serve (North, 1991).

Although population pressure has induced the unsustainable use of key natural resources

such as forests in countries such as Nepal and India (Joshi, 1997; MoEF, 1999), the threats to

natural resource sustainability are not limited to population growth alone. The absence of

effective governmental and local institutions that manage natural resource use is just as

detrimental to environmental sustainability, if not more so, than population pressures. Indeed, a

viable and appropriate institutional framework that is responsible for natural resource





stewardship is a necessary condition for the achievement of socio-economic development and

natural resource sustainability goals.

Hill and Shields (1998) and Rangachari and Mukherji (2000) provide empirical evidence

for the reversal of severe land degradation and deforestation following the adoption of Joint

Forest Management (JFM) in different states of India. Given the proper legal and institutional

environment, communities have an incentive to organize and manage their resource base in a

sustainable manner. The work of Elinor Ostrom has been instrumental in explaining how the

strength and resiliency of community-level resource management organizations is derived from

institutions that have evolved over time. In her book Governing the Commons (1990) she uses

several case studies from around the world to analyze successful institutional arrangements that

govern a diverse array of CPRs, including forests. Some of these case studies include CPR

management regimes that have persisted for several hundred years.

Delineated from the case studies that Ostrom (1990) presents are a particular set of seven

institutions referred to as "design principles" that she suggests all successful, long-enduring CPR

management regimes are likely to exhibit.1 Most of the design principles are self-reinforcing,

and thus function as an integrated system for equitable governance of access to, and extraction

from, CPRs. Presented below is a brief description of each design principle (DP) and how it

provides an incentive to collective action, or otherwise affects the successful functioning of CPR

management.

Clearly defined boundaries (DP 1) are basic to the problem and consist of two components:

the spatial delineation of a given resource as common property, and the definition of rights of

1 An eighth design principle (nested enterprises) is auxiliary and only relevant for larger and more complex CPR
management systems that have an ordered hierarchy (Ostrom, 1990). For example, irrigation associations that must
temporally co-ordinate allocation of water amongst several groups of users, and at several spatial scales in a region,
often require the other seven design principles to be organized in multiple layers of nested enterprises. This design
principle has no applicability to the scope of the present study and will not be discussed further.





access and use to a specific set of stakeholders. When taken together, these components lead to a

marked reduction in uncertainty regarding who has legitimate access to the CPR, and clearly

establishes the exclusion of non-stakeholders from access. These boundaries can be thought of as

necessary conditions (though not sufficient) for conversion of an open-access resource to a CPR

through the imposition of resource management institutions.

Congruence between appropriation and provision rules and local conditions is the second

design principle (DP 2). The heterogeneity of resource endowments, even amongst separate user-

groups (i.e., villages or communities) of the same geographic region emphasizes the importance

of designing institutions that reflect the specific characteristics of a given CPR. For example, no

two physical environments are exactly the same, even for the same type of CPR such as forests.

Therefore, human interactions with CPRs will display considerable spatial and temporal

variability, and will require institutions tailored to the specific parameters and nuances of a

particular CPR.

Effective and equitable governing structures will give a voice to all stakeholders and

requires that most members are given the opportunity to participate in the modification of the

institutions. These collective-choice arrangements (DP 3) ensure that the institutions are

responsive to the very individuals that they serve. In addition, this responsiveness also helps to

ensure that DP 2 (congruence) is functioning well: by being responsive to its members, the

institutions are able to respond and adapt to changing conditions. This is because the users of the

CPR are in the best position to assess the local environment (political and physical) and any

changes therein.

Monitoring of the CPR (DP 4) is necessary to keep all stakeholders in compliance with the

appropriation rules, and to enforce the exclusion of non-stakeholders. This design principle





addresses the fundamental problem of open access by eliminating the free-rider and forcing each

user to pay the cost of their resource usage. Monitors are accountable to the stakeholders, and are

frequently the resource appropriators themselves. Effective CPR institutions do not rely on

external enforcement. The community itself is responsible for the compliance of each individual

through both formal and informal (i.e., cultural) institutions.

Circumstances will arise where temptations to cheat lead to infractions; thus, effective

compliance is a function of graduated sanctions (DP 5). There is a certain amount of fairness that

is implied with graduated sanctions. They are important so as not to discourage future

compliance when minor infractions occur, or when first-time offenders are discovered cheating.

However, sanctions should be heavy for repeat offenders in order to ensure increased future

compliance by those who follow the rules.

Mechanisms for conflict resolution (DP 6) that are characterized by low transaction costs

are required. To ensure both the fairness and the continuity of the institutions, each stakeholder

must have recourse to a forum established for the purpose of dispute resolution. As any set of

rules and regulations is subject to different interpretation by different individuals, and as the

design principles are attempting to manage an open access resource, a forum is necessary for the

dispensation of justice or punishment to those accused of non-compliance. Again, an external

source of governance is not required for this function.

Minimal recognition of rights to organize (DP 7) is the seventh and last design principle

discussed in this study. This design principle relates directly to an external presence. It refers to

the need for an over-arching governmental policy framework that facilitates, or even encourages,

community-level institutions that are capable of natural resource management. The most basic





requirement is for the government to recognize the rights of local people to devise their own

institutions regarding CPRs.

These design principles provide a context within which participatory forestry programs,

including those in India, can be evaluated. Although the particular aspects of JFM institutions in

India do not allow perfect adherence to all of the design principles advanced by Ostrom, these

principles nevertheless provide a framework for later discussion in this study.

A Case Study in India

Forest Management

The response to the problem of widespread forest degradation in India has been the

development and implementation of Joint Forest Management (JFM). The Government of India

introduced legislation in 1988 that was specifically designed to promote forest regeneration

while addressing the needs of forest-dependent people living in rural areas. This change in forest

policy subsequently led to the adoption of JFM in the individual states of India.

JFM attempts to provide for the sustainability of public forests by incorporating local

stakeholders into the planning, management, and protection of public forests. Economic

incentives provide a foundation for community involvement, and help to improve the socio-

economic outlook of the participants. The cornerstone of the JFM paradigm is the participatory

involvement of village stakeholders working (in collaboration with the state forest department) to

reforest degraded local public forest lands. This was not always the case: forest departments

historically followed a reactive approach to forest conservation and management that often

alienated local communities. In contrast, JFM is a much more proactive approach.

In order to participate in the program, a community must establish a forest protection

committee and register this organization with the forest department. Membership (by

households) in this body is voluntary. Once approved and registered with the forest department,





the forest protection committee is given a tract of degraded forestland to rehabilitate, manage,

and protect. Silvicultural techniques and other works (e.g., soil and water conservation) are

employed to promote the natural regeneration of the area protected by the forest protection

committee. Although ownership of the land is retained by the government, economic benefits

conferred to the forest protection committee act as incentives for sustainable management and

continued participation. As such, the forest protection committee is granted usufruct rights for

most of the NTFPs that the designated forest area yields. Moreover, a critical aspect of JFM is

their right to a portion of timber revenues (from future plantation harvesting), which serves as

one of the main incentives for community participation in the program.

The basic objectives of JFM are to increase the livelihoods of local forest-dependent

inhabitants by restoring degraded public forest lands, and to equitably distribute the related

benefits throughout local communities. Since implementation of this forest management

paradigm began in the early 1990s, adoption of JFM by local communities has been widespread.

Recent data indicate more than 99,000 registered JFM committees are managing an estimated

214,300 square km of forest land (32% of total forest land) in all of the 28 states that comprise

India (MoEF, 2006). The JFM program has changed throughout its tenure (MoEF, 2005),

however, and continues to evolve over time.

On balance, the basic theory behind JFM is sound: addressing the restoration and

sustainability of forests through incentive-driven community participation. The result is a move

away from open access over-exploitation and toward collective conservation efforts that will

realize future dividends to the community. Some empirical evidence exists to support this

outcome, both in terms of forest resource restoration and income generation. For example, Khare

et al. (2000) states that JFM is successful when measured in terms of the spatial spread of the





program, the number of hectares of forest managed under the program, and the regeneration of

degraded forests that has occurred. Other researchers have documented increased village welfare

derived from timber revenues, NTFPs, and environmental services (e.g., increased groundwater

recharge) following implementation of JFM programs (Hill and Shields, 1998; Rangachari and

Mukherji, 2000; Asia Forest Network [AFN], 2002). Rangachari and Mukherji recount clear, if

preliminary, local success following the initial implementation of JFM in the state of Andhra

Pradesh. Although they caution against premature claims of achieving sustainability after only a

few years of JFM, their discussion specifically references improvements in forest protection and

regeneration, NTFP yields, water table levels, agricultural production, and forest employment. In

this lies the strength of JFM: demonstrating that community-level participatory involvement in

resource management has the potential to reverse environmental degradation, even when the

institutional arrangements are sub-optimal.

Nevertheless, some of the same research raises legitimate concerns of the distributional

aspects of JFM by discussing how the most disadvantaged people appear to be those most

adversely affected by restricted access to forests (Hill and Shields, 1998; Khare et al., 2000).

Reddy et al. (2004) discuss how the livelihood impact of participatory forest (i.e., JFM

programs) "falls short of expectations" in selected villages of Andhra Pradesh that were studied.

Hildyard et al. (1998) raise concerns that "...the rhetoric of participation is not matched by

realities on the ground." They are highly critical of projects that are participatory in name only.

Overall it appears that the broad success of JFM has been mixed. Considering the multiple

objectives that JFM is intended to address, it is perhaps not surprising that success has been

variable; especially if management of the resource takes defacto priority over socio-economic

concerns, as some authors have suggested (Saxena, 1997; Rangachari and Mukherji, 2000).





Despite these shortcomings, JFM is the preeminent forest management paradigm in India,

and one of many participatory forest programs currently practiced around the world. As the

generic label "community forest management" has become ubiquitous in describing most (if not

all) of these programs, it is increasingly apparent that there is no universal specification of a

"community forest management" model (Odera, 2004). This is largely because the programs and

institutions being implemented in different countries around the world reflect the specific

situations unique to a given country or region. This is as it should be, though; as each country (or

greater/lesser region) will be defined by its own natural resource endowments (e.g., geology,

climate, etc.), cultural practices, and socio-cultural and political history (Schmink, 2004). Hill

and Shields (1998) provide an example from their case study of two villages, one each in two

different states of India. They found that, in general, site-specific factors (e.g., tree species,

locally important NTFPs, etc.) helped to determine the differential economic success they

observed between the two cases.

Research Objectives

Even though its popular appeal is greater in some states than others, and may vary locally,

there is little doubt that the JFM paradigm has spread throughout India in impressive fashion.

This is advantageous for research, as continued efforts to study JFM in different locations can

help to shed light on how spatial heterogeneities (whether environmental, institutional, and/or

socio-cultural, etc.) affect the socio-economic objectives of the JFM program. Given that the

main overarching goals of JFM are to restore degraded forest lands and to increase and equitably

share benefits derived from forests, it is important that contemporary research analyze the

impacts of JFM.

Although much has been written about community forestry in general, and JFM in

particular, relatively few studies (e.g., Hill and Shields, 1998; Reddy et al., 2004) investigate the





economic efficiency and equity outcomes of JFM in India. It is generally believed that JFM has

been successful in rehabilitating degraded forest lands (Khare et al., 2000; Rangachari and

Mukherji, 2000); yet many questions still remain: Does JFM have an identifiable economic

impact on the villages participating in the program in Andhra Pradesh? Is there any impact on

inequality or poverty that can be measured? Are the design principles of Ostrom reflected in the

institutions that currently define Community Forest Management (CFM)2 in Andhra Pradesh;

and, if so, are they being implemented and adhered to in the local communities?

In order to answer these questions, this dissertation has four specific research objectives:

1. To calculate estimates of several economic indicators in JFM/CFM villages of Andhra
Pradesh; including the mean per-capita household consumption (g), the Gini coefficient (y),
and the poverty-gap ratio (PGR);

2. To analyze the influence of demographic, economic, bio-physical, and institutional factors on
the economic indicators in Objective 1;

3. To analyze the institutions of Community Forest Management (CFM) as implemented in
Andhra Pradesh; and

4. To identify the structural strengths and weaknesses of the Community Forest Management
(CFM) institutions based on the findings of Objectives 2 and 3.

Methods

Two separate questionnaires were utilized in the field survey of this research study. A

village-level questionnaire was designed mainly to collect information about the key institutions

of each JFM/CFM village selected for the survey, such as the way in which they monitor their

protected forest area, for example. Other questions inquire about the basic functioning of their



2 Note that in the state of Andhra Pradesh, JFM was modified in 2002 to further devolve the program to allow local
stakeholders more control of forest management. As such, the JFM program there is currently referred to as
Community Forest Management (CFM). The compound acronym JFM/CFM is used in the remainder of the text to
denote when the continuity of the program is being referenced (e.g., for communities that began JFM in the early
1990s, and continue their activities under CFM in the present). Use of the acronym JFM will henceforth denote the
program as applied throughout India, while use of the CFM acronym will generally represent the current set of
institutions that define participatory forestry in Andhra Pradesh.





forest protection committee, the historical quality and uses of their protected forest area, and

about the forest improvement works they have undertaken.

A household-level questionnaire was developed specifically to collect socio-economic

information from the households that were sampled in each of the villages surveyed. It is based

upon the modular format of the Living Standards Measurement Study (LSMS) questionnaire,

which was initially designed and tested by the World Bank in the 1980s (Grosh and Glewwe,

2000). LSMS surveys have been administered around the world; and, in fact, the development of

the module structure was mindful of the standardization of questionnaire formatting and the

flexibility needed to accommodate surveys with different research goals. LSMS surveys are

composed of numerous independent modules, each one corresponding to particular topics

covered in surveys administered in developing countries that seek to gather information related

to various development issues. One of the central objectives of LSMS surveys is to measure

household consumption in order to document living standards and poverty. Deaton and Grosh

(2000) explain that consumption directly generates a state of well-being (i.e., an actual

condition), whereas income and wealth connote a power dimension (or potential). Thus, the most

important concern of LSMS in estimating consumption is for measuring the distribution of living

standards, including poverty (mainly) and inequality as well.

The compilation and analysis of the socio-economic data allows for the empirical

examination of the interrelated relationship among CFM, its institutions, and the economic

welfare and equity of the local communities that participate in the program. Dependent variables

for regression analysis are economic indicators that represent components of social welfare and

the level of poverty in the communities surveyed; they are analyzed because the JFM/CFM

program is largely concerned with poverty alleviation and improving the livelihoods of rural,





forest dependent communities. In this study, social welfare is represented as W = tp (1 y),

where t is mean per-capita household consumption (for a given community), and y is inequality

(as measured by the Gini coefficient). Because both tp and y are two separate components of

social welfare, they are analyzed as distinct dependent variables in different regression equations.

A third equation utilizes the poverty-gap ratio (PGR) as the dependent variable. PGR is a

measure that quantifies the incidence of poverty in a given community; it is based on an

objective consumption-based poverty line.

Demographic, socio-economic, bio-physical, and institutional variables represent the four

types of explanatory variables drawn from a survey of individual households and a survey of the

forest protection committee leaders. Examples of each variable type include: number of

households (demographic); income from forest products (socio-economic); size of the protected

forest area (bio-physical); and the presence or absence of a formal patrol (institutional). In

particular, the JFM/CFM program is represented by a variable that controls for the length of time

that a community has participated in the program; as such, any impact on the dependent variables

can be discerned.

Summary and Preview

Chapter 2 begins with the history of forest management in India, including a discussion of

the revolutionary experiment that JFM was later modeled on. The next section describes in detail

the institutional framework of JFM (and later CFM) as it has been implemented in Andhra

Pradesh. Then, a general overview of participatory forestry in other countries is offered to

provide a basic comparison with JFM in India. The final section of Chapter 2 provides an

introduction to the southern Indian state of Andhra Pradesh. A general description of the

linguistic diversity, the social caste composition, and geographic regions of this state is presented





in order to supply the reader with some basic knowledge of the study location, as well as

illustrate some of the challenges posed by the scope of this research study. Included is a brief

overview of the role of the Andhra Pradesh Forest Department (APFD) in administering the

CFM program throughout the state.

The third chapter describes both the development and implementation of the community-

level and the household-level questionnaires, and the overall structure and administration of the

survey. The first section describes the questionnaire used to interview the leaders) of the forest

protection committees in each village of the sample. The second section presents the composition

and layout of the household-level questionnaire, discussing the design, structure, and formatting

in detail. The final section of this chapter examines some important aspects of how the survey

was implemented, including the spatial structure of the survey and field methods, etc.

Chapter 4 begins by presenting details of the specification of the empirical models. The

second section discusses the hypotheses on the independent variables, while the final section

presents and describes the data collected by the field survey. Chapter 5 presents the regression

results and discusses the general implications. Chapter 6 returns to the design principles outlined

by Ostrom, discussing them in the context of CFM in Andhra Pradesh and the empirical data that

were collected and analyzed.

The final chapter summarizes the study results and offers overall conclusions and

implications of the research. Policy recommendations and opportunities for future work in this

area are also provided.





CHAPTER 2
BACKGROUND INFORMATION ON THE STUDY AREA

Overview

This chapter is divided into three main sections. The first section describes the historical

management of natural resources and institutions in India, beginning with the pre-colonial era.

Then, the immediate predecessor of JFM in India is discussed: the so-called "social forestry"

programs of the mid 1970s to the late 1980s. Due in part to its lack of long-term success, social

forestry in India was succeeded by JFM around 1990, although the origin of JFM can be traced

back to a pilot program undertaken in the state of West Bengal in 1972 (which is also discussed).

Thus, the initial foray into JFM actually predates social forestry in time, but eventually the

success of this pilot program in regenerating degraded forest lands became widely publicized

(Mitra, 1995). This led to JFM ultimately supplanting social forestry as it began being replicated,

first in West Bengal, and later throughout the rest of India. This first section ends with a detailed

description of the development of JFM in Andhra Pradesh, and its later transformation into the

CFM program.

The second section of this chapter presents an overview of community-based forest

management in other areas of the world. Nepal, China, and a few other countries in Asia are

discussed; this is followed by a general description of participatory forestry in Africa.

Community forest management in Mexico is also presented because, although there are some

similarities to JFM in India, there are also some important institutional and production

differences.

The third section provides an overview of Andhra Pradesh, which includes a general

description of the state's linguistic diversity, social caste composition, and its three geo-political





regions. The role of the forest department in administering JFM/CFM throughout Andhra

Pradesh is discussed as well. The final section summarizes the discussion of this chapter.

History of Participatory Forest Management in India

Pre-Colonial Era

The historical antecedents of JFM lie in the distant past of India itself. Two examples

illustrate that community-level control of natural resources is not a foreign concept to the people

of India. Indeed, it becomes clear that institutional structures can provide the incentives, or the

disincentives, which govern the behavior of people.

Rangachari and Mukherji (2000) state that, like many other indigenous peoples around the

world, the tribal peoples of India practiced shifting cultivation (also known as swidden or "slash

and bum" agriculture). Forest land is typically cleared, burned, and planted with crops until soil

productivity declines. The process then begins again on a new parcel, or an area left fallow is

reutilized. Agricultural output is relatively low compared to high-input, intensive farming

techniques used today. However, swidden cultivation requires no external inputs (other than

labor) and, for that reason, is sustainable in the long-run (given relatively low population

densities and an abundance of land). Besides raising crops under this system, the forest is utilized

for other subsistence activities like hunting and gathering. Forests are an integral part of the

cultural and economic life of such a community, and the institutions that govern access and use

to the resource will reflect the values of the people who depend on it for survival. As land is held

in common (i.e., there are no private property rights), resource sustainability and community

welfare become the implicit objectives of resource management.

The southern kingdoms of India offer historical evidence of agricultural surplus and wealth

in the post-Sangam era, according to Rangachari and Mukherji (2000), who discuss how village-

level autonomy and community-level social organization was the norm for medieval southern





India. The institutional structure of this era (A.D. 300 to colonization) emphasized the de-

centralized administration of planning, investment, and management of agriculture and natural

resources. As an example of an appropriate and functioning institution, Rangachari and Mukherji

(2000) discuss the structure of institutions that governed local water resources. Apart from tax

collection, the King would preclude himself from any local administration, but would provide

such public goods as large-scale irrigation infrastructure. The local organization and control of

smaller-scale irrigation infrastructure was facilitated by the appropriate set of incentives and

enforcement that bind together the stakeholders of common property resources.

The traditional institutions of forest management became incompatible with the centralized

and bureaucratic governance structure of the British following the advent of their political and

economic hegemony. The imposition of western ideas (such as the nationalization of forests)

naturally supported and furthered their control, enabling them to consolidate their power and

increase their wealth. For instance, the British promoted the distribution of private property

rights for continuously cultivated land, and actively encouraged the clearing of forest land for

agricultural production in order to generate farm incomes and subsequent tax revenues (Khare et

al., 2000).

Another consequence of the changing institutional structure was the adoption of the

prevailing forest management paradigm of the time, which viewed timber primarily as a factor of

production. Before the introduction of steel and plastic, wood was utilized in the production of a

wide variety of consumption goods. In particular, aside from use as a fuel source, large quantities

of timber were used to produce railway ties, ships, and houses. Following Independence in 1947,

the government of India essentially carried over this "industrial forestry" paradigm in the interest

of its own national development objectives. They maintained exclusive control over the





production and management of forests, with concessions to industry providing a steady flow of

income to the states.

Social Forestry Programs

After India gained independence, the forest policies of the government continued both the

land tenure regime and the management practices of the British. In particular, the colonial-era

paradigm of 'scientific' forest management proceeded well into the 1970's; there was little

regard for the natural diversity of tree species as forests were converted to monoculture

plantations (Khare et al., 2000). The needs of local people continued to be subordinate to

national economic imperatives. However, burgeoning pressure on the state forests from rural

people led to a policy change designed to stem forest degradation and ensure future productivity.

Although official forest policy rhetoric had always paid only token respect to forest-dependent

people, the 1976 National Commission on Agriculture (NCA) explicitly stated that management

of government forests was to further the production of timber for industrial purposes (Khare et

al., 2000). But in order to do so productively, the government realized the necessity of relieving

the local pressures that were increasingly being placed on the public forest lands. What became

known as "social forestry" was the resulting vehicle designed to accomplish this objective.

The idea behind social forestry was that the rural demand for small timber, fuelwood, and

other forest resources would be met through production on village communal lands, private

farms, and degraded or unproductive government land. Government subsidies and technical

assistance were provided as incentives to mobilize the local population to adopt this paradigm.

More importantly (from the government's perspective), the industrial demand for timber would

be met by plantation forestry on government forests, which would now be free from rural

predation. Thus, the continuation of a national forest policy that served industry still left the

welfare of rural inhabitants as a secondary consideration. In fact, Khare et al. (2000) state that





"the major benefits [of social forestry] had been to the central and state governments in the form

of cheap raw material and unbridled access to forests. Both types of benefits were at the expense

of forest-dependent people" (p. 59).

Despite this reality, social forestry was important because it foreshadowed the concept of

JFM by allowing tree cultivation on degraded government land, with at least some management

responsibilities in the hands of local people. For example, the "community plantations"

component of social forestry established government-sponsored plantations on community

and/or government lands. Communal grazing areas, degraded government forests, and roadsides

are examples of lands that were planted to meet the small timber and fuelwood needs of local

inhabitants.

This program could not be sustained beyond the first harvest, however; due in part to a

lack of institutions defining both the rights to the trees and the distribution of benefits (Khare et

al., 2000). In addition, other factors that contributed to the demise of the program were that the

forest departments exhibited a poor understanding a priori of the actual perceptions and practices

of local people vis-a-vis trees, timber, and related forest issues. For example, the assumption that

the community plantations would be utilized for fuelwood was proven false when it became clear

that local authorities viewed these plantations as a source of communal income.

"Farm forestry" was the name applied to the other component of social forestry. The

defining feature of farm forestry was tree production on private lands. Farmers were given free

or subsidized seedlings, and no restrictions were placed on what to do with the output (e.g., use

as fuelwood or timber). Initially this program was very successful, thriving beyond the original

expectations of planners, who had anticipated social forestry as a scheme for locals to produce

their own fuelwood (mainly) and small timber for their own use. In reality, however, participants





were cultivating commercially valuable species in response to the economic incentives provided

by the program. As they did not regard the production of trees for fuel as a valuable use of the

resource, they instead viewed trees as an investment. Thus, farmers preferred to focus on the

production of poles, small timber, and/or pulpwood for commercial gain.

The subsequent initial success of farm forestry was so great that the government was

unable to assist with the marketing of surplus timber. Thus, timber prices fell drastically in many

states, which negatively impacted further program participation. Pulpwood subsidies to paper

mills, and legal restrictions on the sale and transport of timber, also contributed to the low prices

that eventually signaled the end of farm forestry in the late 1980s. Although "farm forestry

neither significantly met local needs nor improved private wastelands", an important realization

was that farmers responding to the right incentives and prices could meet most of the raw

material needs of industries based on wood (Khare et al., 2000, p. 59).

The Arabari Experiment

The origin of JFM can be traced to what Saxena (1997) refers to as the Arabari experiment.

In 1972, an enlightened Divisional Forest Officer in the state of West Bengal took control of

1,272 hectares of deforested land. This area had once been very productive as a good source of

fuelwood, food, and livestock fodder. By 1972, however, it no longer produced commercial

timber, and the soil erosion associated with the loss of tree cover was adversely affecting the

local agriculture. In order to ensure successful forest regeneration, this officer realized the need

for an integrated development plan to eliminate the cycle of degradation. Because the collection

and selling of fuelwood in local markets would hinder any afforestation efforts, the officer

banned such activity. Income lost from this source was replaced by the employment of local

people for the forest restoration work.





The majority (55%) of this land was managed for the re-growth of coppiced sal (Shorea

robusta), an important tree species that provides NTFPs in addition to timber. Most of the

remaining land was cultivated as plantation of acacia (Acacia spp.), eucalyptus (Eucalyptus

spp.), cashew nut, etc. In addition to the wages received for their labor, the participants were

promised 25% of the final timber harvest (subject to the success of the project). Moreover, to

ensure that the needs of the participants were met during the maturation of the forest, rice and

fuelwood were grown on the land. Rice was subsequently sold at cost, while fuelwood was sold

for a token price. Rotational grazing was also allowed.

By creating incentives to regenerate, manage, and protect this plot of land, the Arabari

experiment placed local people at the center of efforts to rehabilitate a portion of the state's

forest. The program incentives provided for their intermediate needs through wage income and

low cost subsistence goods, while the share of timber revenues gave them a stake in the long-

term health and productivity of the forest. Saxena states that "this arrangement made it clear to

the people concerned that they had a right to enjoy the enhanced benefits of forests, but this right

was accompanied by their duty to nurture and protect the forests" (1997, p. 99). In 1987, 618

families were awarded a share of the timber revenues following the harvest of 97 hectares of sal

and eucalyptus. In addition, each family received wages paid for harvesting the timber. The

success of this model is evident in that over 2,000 villages had adopted the program by 1993, and

that approximately 60% of the forest area in the region is managed under this model (Saxena

1997).

Development of Joint Forest Management (JFM)

Recognizing the value of grassroots participation to successful afforestation projects, the

Government of India (GOI) enacted the 1988 National Forest Policy that subsequently enabled

the adoption of JFM programs throughout India (World Bank, 2000). This policy change shifted





the use of public forests away from commercial exploitation and towards the subsistence needs

of forest-dependent people (Saxena, 1997; Khare et al., 2000). This was a radical departure from

past forest policies and reflected a new understanding that the "tragedy of the commons" is not

necessarily an inevitable outcome of resource use by locals, but rather a function of institutions

which are either non-existent or ineffective. This new understanding came in part from

demonstrated successes such as the Arabari Experiment.

Policy makers were also influenced by the changing political winds. Although

"environmental activism is not a new phenomenon in India, but is rooted in the past," the plight

of forest dwellers was not promoted by "intellectuals and activists" until the early 1970s (Saxena

1997, pp. 40-41). Widespread attention to the dissent and unrest of forest communities began

with the non-violent Chipko movement in northern India in 1973, and continues to the present-

day with the armed rebellion by so-called "Naxalite" groups in Andhra Pradesh and other states.

Although the 1988 National Forest Policy reoriented forest policy to the needs of the local

stakeholders, the June 1, 1990 Circular actually specified the basic rights that people have in

relation to forests under their protection (Khare et al., 2000). It explicitly enabled the state forest

departments to engage locals in the management of forests. Thus, this resolution directly

facilitated the implementation of JFM without ever mentioning the words "joint forest

management". Clearly JFM is the reference point, as the text mentions how "Village Forest

Protection" in West Bengal receives a 25% share of timber revenues. In addition, the text states

that similar institutions may be adopted by the other states (Government of India [GOI], 1988),

and also encourages the forest departments to work with non-governmental organizations

(NGOs) as intermediaries between the government and the local communities (Saxena, 1997).





With this legal order, state governments now had the authority to draft their own legislation that

specified the particular institutions that would define JFM in their respective states.

JFM in Andhra Pradesh was initiated under G.O. MS. No. 218 in August 1992. With

reference to the 1990 Circular of the GOI, this order commands the Andhra Pradesh Forest

Department (APFD) to introduce and implement the JFM concept to all districts of the state. It

provides some general instructions for implementation, most notably calling for frequent reviews

so that analysis can lead to beneficial amendments. The order also directs the local village

community to be organized into a Vana Samrakshana Samithi (Sanskrit for "forest protection

committee") if the community collectively desires to participate in JFM. VSS is the acronym for

Vana Samrakshana Samithi, and it is commonly used in Andhra Pradesh to denote the forest

protection committee general body. The Annexure to G.O. MS. 218 contains the specific

institutions of JFM that spell out the composition, functions, responsibilities, and rights of the

VSS and its managing committee; these details are summarized next.

Given a quorum of 50% of village households, a Forest Officer will explain the concept of

JFM to the assembled community. A VSS will be formed if sufficient interest exists, and every

household in the village has the opportunity to join. Any two members from a given household

are allowed to join the VSS; however, one must be a woman. Each member of the VSS general

body will, individually and collectively, protect the forest area against grazing, fire, and theft of

forest products. In addition, members will assist the forest department in implementing a jointly-

developed forest management plan (known as the "micro-plan"). The VSS general body will

meet every six months to review the plan.

Every VSS shall have a Management Committee (MC) that is charged with the

responsibility of carrying out the approved JFM micro-plan. The MC will convene monthly, and





the term of service is one year. It will be composed of six to ten members elected from the

general body of the VSS, the president of the local government council, and two members of the

forest department (who do not have voting privileges).

The micro-plan is to be developed by the APFD in consultation with the MC, and will

apply to a specifically designated tract of degraded public forest land selected by the MC and the

APFD. It will focus on supporting the demand for traditional forest products local to the area,

and include measures designed to aid the regeneration of the forest-soil and water conservation

measures, in particular, are to be an integral component. Planting of low-valued fruit trees such

as tamarind is allowed, but horticultural species like mango and guava are not permitted. The

conservation and development works of the micro-plan are to be coordinated by the MC, as are

the paid and un-paid labor inputs. First preference for paid labor is to go to VSS members. The

micro-plan will be in effect for 10 years and is subject to revision by the forest department.

Given adherence to its duties and responsibilities, the VSS is granted usufruct rights to the

forest. The VSS, acting through the MC, is responsible for the equitable distribution of the

usufruct benefits entitled to VSS members. Discretion is given to withhold or lessen the share of

benefits according to the contributions, or lack thereof, of individual members. Each household

is considered as one member for the dispensation of the usufruct benefits. Rights are divided into

two classes. Non-reserved rights are granted to leaf and grass fodder, thatching and other grasses,

thorny fencing material, and deadwood. Reserved rights apply to certain NTFPs under contract

to a parastatal, and to timber and poles. After three years, access for timber and poles are

afforded to the community subject to the JFM micro-plan: harvest is shared between the VSS

and the APFD, with each receiving 50%. Usufruct rights shall only apply to VSS members, and





any disputes are to be adjudicated by the MC. The Conservator of Forests (a high-ranking APFD

official) has the authority to relax these rules and regulations.

The first revision of the initial legal orders came in December 1996 with G.O. MS. No.

173. As well as changing some elements of the original order, it also provides more detail to the

composition, functions, responsibilities, and rights of the VSS and its management committee.

Important changes included:

* Ensuring participation of all tribal households and households from the lowest castes by
making their membership automatic. Such households are also allotted a certain percentage
of membership in the MC.

* Increasing the number of elected MC members to 10 15 (of which 30% are to be
women), and increasing the term of the MC to two years.

* Allowing the VSS to apprehend offenders and turn them in to the concerned authorities.

* Devolving more power to the VSS to prepare the micro-plan, especially with the goal of
including the input of women and more disadvantaged groups.

* Allowing the VSS to select the species for plantations, and relaxing other restrictions on
what type of trees can be planted.

* Specifying that all labor contributions be paid.

* Relaxing the restriction on most NTFPs, and specifying that 50% of net income from
collection of Beedi leaf (for locally produced cigarettes) be paid to VSS members.

* Increasing the share of timber and bamboo harvest received by the VSS to 100%.

Advent of Community Forest Management (CFM)

In February 2002, the legal orders for JFM were re-written to place greater emphasis on

community participation and autonomy. As such, JFM was superseded by Community Forest

Management (CFM) under G.O. MS. No. 13. According to the APFD, CFM "...aims at

decentralizing the entire process of planning and implementation with APFD and Government of

Andhra Pradesh (GOAP) acting more as facilitators and providers of technical and infrastructure

support" to local stakeholders (APFD, undated, p. 4). This is in contrast to JFM, where the





VSS/govemment relationship is characterized as being more like a partnership. The main

changes included revising the membership and functions of the MC, and creating other support

councils for the VSS. Some minor revisions to NTFP, timber, and bamboo rights were made

also. The important changes were:

* The VSS can now collect fines (less than 100 Rupees) for minor forest offences.

* VSS are entitled to all intermediate yields obtained from silvicultural operations.

* The MC is now comprised of 15 elected members from the VSS, and at least eight must be
women. The MC tenure is increased to three years.

* The MC will elect a Vice-Chairperson. Either the Chairperson or Vice-Chairperson must
be a woman. The Vice-Chairperson is to work closely and assist the Chairperson with the
dispensation of his/her duties.

* The Chairperson will maintain VSS account books, micro-plans, minutes books, etc.

* The MC will account for and manage the VSS funds and other resources.

* There are two accounts. The 'Government Account' contains funds received from the
government, and is jointly operated by the Chairperson, Vice-Chairperson, and APFD
representative. The 'VSS Account' contains internally generated funds, and those derived
from non-governmental sources. It is operated jointly by the Chairperson and Vice-
Chairperson.

* APFD staff, NGO representatives, etc. are no longer part of the MC, but will form an
Advisory Council to review micro-plans, and advise the VSS on strategies and resources.

* Other councils will be created to review the implementation of CFM and provide direction
to the APFD regarding the rural development. Representatives from the APFD, other
government agencies, NGOs, and selected VSS will be included. These councils will be
created at the district, forest division, and state levels.

G.O. MS. No. 13 represents the current set of institutions authorizing and defining CFM in

Andhra Pradesh. This document is a significant improvement over the original order authorizing

JFM in 1992, both in terms of the actual institutions and the clarity of presentation. Although the

general program parameters are specified by the government, as opposed to being organic

institutions self-derived by locals, the clear intention of CFM is to recreate a similar institutional





environment. The continued involvement of the government has helped the program evolve

towards a framework within which local stakeholders can fine-tune certain parameters to fit their

specific situation. Thus, from initiation in 1992 to the advent of CFM in 2002, the various

updates have ensured and/or strengthened, in principle, the institutions of JFM/CFM, which

characterizes participatory forest management in Andhra Pradesh today.

Comparison to Participatory Forestry in Other Nations

Nepal

The causes of deforestation in Nepal, and subsequent adoption of community forestry, are

similar to the experiences of its neighbor India. Population pressures had already begun to

impact forest resources in Nepal by 1957, at which time the nationalization of the country's

forests took place-ultimately leading to a degraded (de facto) open access resource (Joshi,

1997). The legal orders for participatory forest management in Nepal date to 1978. According to

Joshi, however, orientation of the policy at this time was more towards management by local

leaders or local political units (i.e., panchayats) than collective management. Later policy

revisions ensured that local stakeholders were empowered to manage their forest resources

through the formation of Forest User Groups (Joshi, 1997). These legally recognized groups are

given rights to timber and NTFPs from the forest area under their management, and there is also

an emphasis placed on the participation of women. Unlike JFM in many states of India, however,

even well-established forests in Nepal are eligible for the program in addition to degraded lands.

China and Other Asian Nations

The 1981 Forest Policy enacted by China began to loosen the government monopoly over

forest resources. Essentially, policy changes promoted afforestation by leasing forestland to

individual households, and forest farmers were given greater flexibility and control of their

management (MoEF, 2006). In general, the model became less strictly collective and more a





combination of collective and individual management. In 2002, new laws were passed to

strengthen the security of collective forests, which account for approximately 60% of Chinese

forests (MoEF, 2006). In addition, forestry income is shared with the farmer based on labor input

and other factors.

Other Asian nations are also promoting various degrees of community forestry. In Bhutan,

small groups of at least five people can obtain use rights to forestland if they agree to regenerate

it according to a management plan. Changes to forest policy in Myanmar have enabled

reforestation cooperatives at the village level, and community forest management. The program

allows all benefits derived from the community-managed lands to remain with the stakeholders.

Vietnam is decentralizing the management rights and responsibilities to provincial and local

authorities (MoEF, 2006).

Africa

Odera states that community forest management in sub-Saharan Africa dates from the late

1980s and early 1990s, but that these early efforts were generally focused on "...a narrow band

of linkages between people and trees" (2004, p. 16). Such linkages included exchanging forest

access for labor, and joint management schemes that borrowed principles from wildlife

management services. In the intervening years, community forest management has continued to

develop and evolve as the "open-ended" definition of this paradigm has allowed different

countries to devise and/or adapt institutions that are specific to their own experiences and

circumstances. However, Odera also notes that different management regimes have emerged that

vary from full-fledged participatory management to only token representation by local people.

Nevertheless, experience has shown that community forest management has been successful in

reducing deforestation and improving forest cover and benefits-but only when people have

been empowered with responsibility and have been given secure tenure rights (Odera, 2004).





Most sub-Saharan countries have taken steps toward implementing community forest

management, although most of these programs are clearly in the developing stages with many

less than five years old. Like JFM in India, the general rule seems to be that the local

stakeholders must register with a government agency. This allows them official recognition and

formal use and management rights that, combined with a management plan, usually cover a

period of five to fifteen years (Odera, 2004). Unlike India, however, some countries have granted

permanent land rights or ownership titles to local communities; while others have granted

permanent titles to forest resources. Management plans are often required for such tenure as well.

However, most benefit flows in Africa have been confined to NTFPs because restrictions on tree

felling have been imposed in order to rehabilitate the forest, even when rights have otherwise

been transferred to the local people.

There appears to be wide variation in terms of the types of forest to which community

forest management in Africa is applied. For example, only unclassified forests in Cameroon are

eligible (under a 10-year agreement), with a maximum size of 5,000 hectares. Other countries

allow community forest management in reserved forest areas, even those that are classified as

"high-priority" in terms of conservation (Odera, 2004).

Despite the advent of community forest management in Africa, there are significant

problems. Support for community forest management in many countries is lacking due to

internal problems in national forest departments. A lack of funding and training are also key

constraints. Thus, the role of non-governmental organizations (NGOs) is critical in helping to

serve these needs, and to better implement this management paradigm in general.

Mexico

Agrarian land reforms in Mexico throughout the twentieth century created substantial

amounts of common property in the rural sector, much of it forested (Bray et al., 2006). Despite





legal rights granting control over common property to local communities, the government

maintained significant control over forest resources (and also retained ownership of the land until

1992). From 1940 to 1970, government control was exerted in the form of ineffective bans on

logging and the granting of concessions to harvest timber from community lands (Bray et al.,

2003). During this period, modest attempts to involve locals arose through government efforts

that trained local communities to manage community forest enterprises (CFEs). However, these

actions were undertaken mainly in support of industries dependent upon timber and do not

constitute a serious attempt to empower locals stakeholders by involving them in the direction,

planning, and management of forests for their own benefit.

In the 1970s, grassroots activism and legal reform stimulated interest in true participatory

forest management at the community level, and led to the development of new CFEs. Subsequent

legal reforms lent additional support by devolving more power to locals, and by funding various

projects that directly focused on community forest management and development. Thus, in many

respects the precursor conditions leading to community forestry in Mexico were similar to those

in India.

The particular manifestation of community forestry in Mexico is quite different from that

of JFM, however. The main difference is that many of the CFEs in Mexico are engaged in

commercial production of timber and timber products. Bray et al. (2006) cite an unpublished

study that reports over 2,400 communities in Mexico were engaged in commercial logging in

2002. Furthermore, in a previous paper Bray et al. (2003) state:

What is unique about the Mexican case is the large number of communities that are
managing common-property forests for the commercial production of timber, as well as
finished timber products in some cases, in industrial processes that are thought to be
beyond the reach of most poor, rural communities.





Moreover, Bray et al. (2003) discuss how communities are managing their forests for multiple

benefits and with due consideration for the increasingly robust environmental laws. They also

say that CFEs often voluntarily harvest timber at rates less than their management plans allow.

Why such a focus on timber production? Mexican forestry policies, rural activism, and the

traditional system of rural organization have combined in such a way that allows some CFEs to

respond competitively to market forces (Bray et al., 2003). In addition, a history of past

experience with commercial logging (i.e., concessions), and a legacy of resource use, may have

left a residual attitudinal imprint in rural areas. The size of communal forest areas in Mexico is

also relatively large (at least by Indian standards), averaging 3,074 hectares, according to

calculations from the unpublished data cited in Bray et al. (2006). Therefore, large tracts can

better accommodate competitive commercial logging without being destructive of the

environment, because a proportionately larger volume can be harvested.

Description of Andhra Pradesh

Location

Andhra Pradesh is situated along the Bay of Bengal in the southeastern part of the Indian

subcontinent (Figure 2-1). The fifth largest state in terms of area (276,754 sq. km), Andhra

Pradesh comprises 8.4% of the total territory of India; and, at 972 km, has the longest coastline

of any Indian state (Tata, 2002). The 2001 population of Andhra Pradesh is 75.7 million, of

which nearly 73% (55 million) are rural inhabitants (Tata, 2002). With such a large rural

population, agriculture is a key sector. Not only is Andhra Pradesh a net producer of cereals, but

it also leads all Indian states in the production of rice (Reddy et al., 1992). Cash crops important

to Andhra Pradesh include sugarcane, tobacco, cotton, and groundnuts (peanuts). Commercial

enterprise and industrial production can also be found in the two largest cities: Hyderabad and

Visakhapatnam.



































The capll city, Hydrabad, is loated in thenohw estp aofe htte adisa pr ng
metropols with a 2005popula of61 milon(UN 2006) Hydeabad rivals Bagalears a
Iladag cete of ifoarson tLenology m Inda, ad is s- d m film product only to
'Bollywod" (n Bombay) Hydeabad s also a majorcesntrfpharacauticals and bi-
tehnology The moal cty ofVl paa (A- populaal of 1 5 millt ) ha, ma
ecoaomIe actl tes as steel produdtm peh-ia soa ad shipping (Chagan 2 W, N206)





1 r





-


Flgu 2-1 Ma





Languages and Castes

The principal language of Andhra Pradesh is Telugu, which is a Dravidian language that is

very similar to the languages of the other states of southern India (e.g., Tamil). Urdu is widely

spoken in Hyderabad, owing to the large Muslim population (nearly 50%) in this city. English is

spoken by many as a second language, usually by people who have been educated in "English

medium" schools where it is the focal language. In the rural countryside, however, English is

likely to be known by very few individuals.

The social division of people into castes is a well known aspect of the Hindu religion.

There are literally hundreds of castes, each defined by specific names and often indicative, at

least historically, of an occupation. The GOI has delineated four main categories from the many

castes and tribes: Scheduled Tribes (ST), Scheduled Castes (SC), Backward Castes (BC), and

Other Castes (OC). This classification is for legal and bureaucratic purposes, as affirmative

action-type programs in India are based on this division. These categories are commonly referred

to by their initials.

Scheduled Tribes include the various tribal populations in India that have their own

languages, which are essentially verbal only. They have their own cultural customs as well,

although many elements of Hinduism have been absorbed. In addition to their own language,

most tribal people in Andhra Pradesh speak Telugu. Near the border of an adjacent state, they

may speak the official languages of that state instead of Telugu. Scheduled Tribes are generally

poorly educated and engage in subsistence farming and/or casual labor activities. Most of the

more remote villages in the mountainous regions of Andhra Pradesh are Scheduled Tribes

villages, and their dependence upon NTFPs is usually higher than other social groups.

Scheduled Castes is the term given to the castes that were formerly identified as

"Untouchables" by the rest of Indian society. Scheduled Castes typically occupy the most menial





jobs (which are often stigmatized), they usually own no land, and are generally poorly educated

(as compared to the higher castes). In order to improve the long-term welfare of the Scheduled

Tribes and Scheduled Castes, the Indian government has written into the constitution an

"affirmative action"-type program that guarantees access to civil service jobs for members of

these two social designations. Reddy et al. (1992) estimate that Scheduled Castes and Scheduled

Tribes account for 5.9% and 14.9%, respectively, of the population of Andhra Pradesh. Thus,

nearly 80% of the population of Andhra Pradesh is comprised of Backward Castes and Other

Castes.

Backward Castes include the grouping of castes that socially lie between the lower level of

Indian society (i.e., Scheduled Tribes and Scheduled Castes) and the upper castes. Backward

Castes are largely analogous to "middle class" society in the United States; a typical occupation

might be a merchant or a military officer.

Other Castes comprise the top level of Indian society. They are generally better off,

economically and educationally, than the Backward Castes, and certainly more so than the

Scheduled Tribes and the Scheduled Castes.

Geographic/Political Regions

Figure 2-2 is a map of Andhra Pradesh that is divided into 23 districts. The basic socio-

political unit of a given Indian state is the district, which is roughly analogous to counties in the

United States. More broadly, Andhra Pradesh is comprised of three different socio-political

regions known as Telangana, Rayalaseema, and Coastal Andhra. Andhra Pradesh was formed as

a state in 1956, along the linguistic bounds of Telugu, from these three regions.3



3 As such, it is quite possible that one day Andhra Pradesh will split into three separate states: Jayashanker (2004)
discusses how the recent demands for a separate state of Telangana predate the formation of Andhra Pradesh by a
decade.



























Most of Andra Pradesh ren nds e of the southwestern United States in terms of
physical features and climate, although the coasntl oneis much more humid and lush in terms of
Iropical vegetation. In general, Andhra Pradesh an be characterized as follows forested
mountains in the north, nrth-cnt, and northeast areas, dry upland plateau with smaller
mountain ranges in tme central-we and nsoumt and humid plains along the coast

-ae ano ~ann ona








-5 -




Figure 2-2. District map of Andhra Pradesh
The physical-climatic description above broaly conforms to the Telangana, Rayalaseema,
and Coastal Andhra regions Telangana is compsed of ne dishics mn the north-northwest parts
of the state (the triangular shaped area from Mahbubnagar-Khammam-Adilabad in Figure 2-2).





Of these, the four northernmost districts (Khammam to Adilabad) contain forested mountain

slopes where teak (Tectona grandis) is the naturally predominating tree species, as well as the

main tree species managed by most VSS villages in the region. Bamboo grows well locally and

is also cultivated in such areas for its multiple uses. The other districts of Telangana are

somewhat drier and steppe-like, having more exposed bedrock and less forest cover (as opposed

to the more mountainous and forested terrain in the north). Overall, Telangana receives

approximately 890 mm (35 in.) of rain annually, with most of this occurring in the summer

monsoon of June to September (Reddy et al., 1992).

The four southern in-land districts of the state (Kumool to Chittoor) comprise the

Rayalaseema region, and are similar to the southern districts of Telangana in that rock outcrops

are ubiquitous. Much of this region lies on a rocky plateau interspersed with higher mountainous

areas and valleys; the climate on the plateau is somewhat more agreeable than surrounding

lowlands due to the elevation. Nevertheless, Rayalaseema receives only 670 mm (26 in.) of

precipitation per year and is prone to drought conditions, like much of Telangana (Reddy et al.,

1992). The density of forest cover on the hillsides is generally quite low (extremely low where

degradation is advanced), and these slopes mainly contain low quality tree species (which are

often called "scrub" forest). Native tree species of great value, such as sandalwood (Santalum

album) and Red Sanders (Pterocarpus santalinus), are largely gone except for individual villages

that are attempting to cultivate Red Sanders. Although non-native to India, eucalyptus is widely

cultivated there as part of the CFM program, largely because it is a fast growing species used for

pulp, and because cows do not like to graze on it.

Coastal Andhra is comprised of the districts adjacent to the Bay of Bengal. The low

mountain ranges in the western part of the southern districts (Nellore to Guntur), and the high





mountains in the northwestern parts of East Godavari, Visakhapatnam, Vizianagaram, and

Srikakulam districts form the physical boundary to the west. In between these mountains and the

ocean are humid, tropical plains (with some steep peaks locally) that receive ample moisture to

ensure a lush and productive agriculture. Coastal Andhra receives approximately 1,000 mm (39

in.) of rain annually (Reddy et al., 1992), and this region is generally regarded as being more

prosperous than Telangana and Rayalaseema. Much timber is grown commercially here owing to

the beneficial growing conditions. Species include eucalyptus and casuarina (Casuarina spp.),

and sal in the district of Srikakulam. Fruit plantations are ubiquitous as well, and bananas,

coconuts, cashews, papaya, etc. are all cultivated there.

The Forest Department

As the agency commissioned with facilitating the VSS in their implementation of CFM,

the APFD is the most important entity other than the local stakeholders themselves. As such, it is

important to summarize their main role under CFM. The job of the field level staff is to work

closely with the VSS, providing technical guidance when appropriate. For example, the forest

department is tasked with preparing estimates for VSS project works until the VSS has the

capacity to do so. However, the "VSS which are able to take up this responsibility are

encouraged to do so at the earliest" (Government of Andhra Pradesh [GOAP], 2002). APFD field

officers also ensure that VSS are in compliance with the CFM rules and regulations, and that

they conform to all other pertinent GOI and GOAP laws.

In addition, the forest department is responsible for providing training for forest

management and planning, including specialized technical training for aspects such as the

grafting of high-yielding clonal varieties, raising and management of tree nurseries, etc. This is

an important component of the program because extension work is crucial for creating a viable

program that can endure in the long-run. The transfer of skills and knowledge, especially in





support of the development and enhancement of livelihood opportunities, is necessary for

improving the incomes of the program participants. This will help to sustain participation in the

CFM program when the external funding ends.

Concurrent with the transition to CFM is the development and implementation of the

Andhra Pradesh Community Forest Management project (APCFM), supported by major funding

provided by the World Bank. This project leaves no doubt that the focal objectives of CFM are to

alleviate rural poverty through the improvement of livelihoods, by improving forest management

and productivity. The APCFM Project Implementation Plan also acknowledges the need for

developing alternate livelihoods, and the need for providing coordinated technical and financial

inputs to the VSS (APFD, undated). Therefore it is important to note that this project is

essentially a rural development scheme, set in the context of participatory forest management.

In many respects the APFD is put in an unenviable position by being the lead agency

implementing the APCFM project. Outside of their tradition role, they are now charged with

running a rural development program-something that is not the forte of this organization. It is

worth noting that the APFD is aware of this fact, at least rhetorically, and is making efforts to

adapt:

One of the prerequisites for successful CFM is attitudinal change in forest department from
one of 'command and control' to that of 'recognizing communities as equal partners'.
...With the introduction of CFM they will be required to don the role of 'Facilitators and
Extension workers'. Foresters will be mostly performing regulatory role and they will be
facilitating community participation and providing technical and infrastructure support.
This also warrants greater shift in mindset, which can be ensured only through massive
training. (APFD, undated, p. 9)

These observations indicate that the APFD management structure is committed to CFM,

and takes its implementation and facilitation responsibilities seriously. Indeed, others agree that

the APFD has made substantial progress in adapting from the traditional 'command and control'





mode of operation, to the role of 'Facilitators and Extension workers' that participatory forestry

requires (AFN, 2002).

Summary

The legacy of the institutional changes ushered in by the British is still being felt today. A

near complete erosion of the traditional institutions of forest management, coupled with dramatic

increases in population in the twentieth century, has resulted in the deforestation and degradation

that has left millions in poverty. The poor and other marginalized people of rural communities

dependent upon forests for their livelihoods have been the principal losers from government

policies that traditionally favored timber production (Hill and Shields, 1998; Hildyard, et al.,

1998; Khare et al., 2000). Often these people have no other choice but to illegally harvest

firewood for both income generation and domestic consumption (Khare et al., 2000).

Since implementation of JFM in India officially began in the early 1990s, progress has

been evident in terms of the adoption of the program by local communities. Citing data from

MoEF and others, Khare et al. (2000) report that by the end of the decade there were nearly

35,000 registered forest protection committees in 16 states of India; the area under JFM was

estimated to be over 7 million hectares, and perhaps as high as 9 million hectares. Some of the

forest protection committees recorded are not due strictly to JFM, however; as many of these

committees were established, in some states, under other participatory management regimes

(e.g., van panchayats in Uttar Pradesh) or were self-organized as in Bihar and Orissa.

Recent data shows that the JFM program is still expanding, as over 99,000 registered JFM

committees now manage an estimated 21.4 million hectares in all 28 states (MoEF, 2006).

Several states in particular have shown growth of JFM in terms of numbers of forest protection

committees and total area under joint management, including Andhra Pradesh, Madhya Pradesh,

West Bengal, and Rajasthan. Andhra Pradesh not only possesses one of the largest JFM





programs in India, but the APFD has made "significant progress" in adapting to the changing

priorities in forestry that JFM represents (AFN, 2002, p. 1). Recent figures indicate that there are

8,343 VSS in Andhra Pradesh; the coverage area is about 2.3 million hectares (APFD, 2006)4.

Some of the initial JFM institutions implemented in Andhra Pradesh are robust. One strong

point is that participation by communities is clearly voluntary. It is inclusive in that everyone in a

participating community has an opportunity to join the forest protection committee. Another

strength is that the benefits are clearly spelled out. In addition, several of Ostrom's design

principles are represented to varying degrees: clearly defined boundaries [mainly in terms of who

has resource access] (DP 1); congruence between rules and local conditions (DP 2); collective-

choice arrangements (DP 3); monitoring (DP 4); and dispute resolution (DP 6).

A main weakness of the program, however, is that the JFM institutions are exogenous at

the village level as they are originated with government legislative orders. Although the state

governments provide for minimal rights to organize (DP 7), their influence on the whole process

is much greater than Ostrom's DP has it in mind. This will be discussed in more detail later.

There are several other important weaknesses, as well. For example, women are not accorded a

significant role in VSS management; they are only mentioned in reference to household

membership in the VSS general body. The government role has not devolved enough power to

the local stakeholders: sufficient VSS autonomy is lacking, as evidenced by the fact that the

APFD is represented on the MC. In addition, there are too many restrictions on the usufruct

rights given the VSS as incentives for program participation.

That changes were made to the JFM program in Andhra Pradesh is important not only for

the improvement of the institutions themselves, but because they demonstrate the responsive


4 These data were obtained from the APFD website on Sept. 7, 2006.





involvement of the government in terms of its willingness to make amendments. As a result,

equity concerns were promoted by attempting to better facilitate the participation of the

disadvantaged (e.g., tribals and other poor) and women. More control over enforcement and

planning was given to the VSS, and improvements to the economic incentives were made. For

example, the share of timber revenues was increased to 100%. In addition, further revisions now

allow the VSS to receive a percentage of fees collected (by the authorities) from smugglers

apprehended by the VSS. The share of fees (or the forest produce in question) to be paid to the

VSS was initially set at 25%; it was later increased to 50%. With the evolution of the policy,

graduated sanctions (DP 5) are seen to be represented implicitly.

In conclusion, India is not alone in promoting community-based forest management

programs in order to redress forest degradation and ensure future sustainability of forests. Many

countries around the world, both developed and developing, are involved in various forms of

participatory forest conservation efforts. Although the definition of "community forest

management" can vary widely depending on the location, the international development

community has looked to community-based forest management as a paradigm for forest

conservation and rural development in many developing nations. Thus, it is important that

different approaches under the umbrella of "community forest management" reflect the

particular context within which they operate by considering the local ecology, resource

endowments, socio-cultural systems, and political and economic history of an area (Schmink,

2004). As such, a brief examination of some of the different models of participatory forestry

being implemented in different countries is useful for comparison with JFM.





CHAPTER 3
QUESTIONNAIRE DEVELOPMENT AND SURVEY ADMINISTRATION

Introduction

A significant element of this research study was the collection of primary data for

empirical analyses. Given the lack of comprehensive data on the economic aspects of JFM/CFM

in Andhra Pradesh, it was necessary to design questionnaires and administer a survey of VSS

villages and households (HH) within them in rural areas. This chapter begins with brief

description of the VSS-level questionnaire, and is followed by a detailed description of how the

HH-level questionnaire was developed, structured, and formatted. The following section of this

chapter examines some important aspects of the implementation of the survey, including the

spatial structure of the survey and field methods. Finally, the chapter concludes with a brief

description of how the data were compiled and reviewed for accuracy and completeness.

The VSS Questionnaire

Analysis of VSS institutions is one of the main points of this study. Therefore,

development of this questionnaire was initially derived from the design principles identified by

Ostrom (1990) as defining the institutional foundations of stable, well-functioning CPR

management regimes. The comparative discussion of CPR management institutions, as presented

by Agrawal (2001), was also helpful. Moreover, the APCFM Project Implementation Plan

(APFD, undated) was important to the actual understanding of VSS functioning in Andhra

Pradesh (i.e., in applied sense), and guided the overall construction of the questionnaire. Finally,

field interviews with VSS Management Committee members in villages of Medak district in late

February 2005 were invaluable in ensuring that the questionnaire was grounded in reality.

The VSS questionnaire was designed to elicit information about the institutions and

functions of each VSS selected for the survey. For example, key institutional questions inquired





about monitoring of their VSS area and about sanctions imposed for rules violations, while an

important question regarding the functioning of the VSS included queries about the forest

improvement works undertaken. Other questions asked about the historical quality and uses of

their VSS forest area, number and types of APFD training, any assistance from NGOs, etc. The

VSS questionnaire is 7 pages in length and is written entirely in English (with a few Anglicized

Telugu words); it is reprinted in Appendix A.

The Household (HH) Questionnaire

The LSMS Modules

The LSMS modular format was used as the foundation for constructing the HH

questionnaire. Based upon actual experiences with the LSMS, Grosh and Glewwe (2000) and

other researchers provide detailed advice on how to design multi-topic household surveys for

developing country research; included are discussions by module (i.e., information category),

examples, and even Microsoft Excel templates on an accompanying compact disc. Several

independent modules comprise an LSMS survey, with each module corresponding to particular

topics of interest that a given research study is either directly or indirectly addressing. For

example, household consumption is one of their main modules; others include income,

employment, education, health, etc.

The HH questionnaire contains eight modules and is 27 pages in length (24 pages of

questions). Table 3-1 presents the contents of this questionnaire with a brief description of each

section. The ordering of the modules reflects not only a logical sequence (e.g., household roster

near the beginning), but also the relative importance of each module in the study. For example,

the two most crucial modules of the survey, in terms of the research objectives, are the

Consumption and Forest Resources modules. A copy of the HH questionnaire used in this

research study is presented in Appendix B.





Table 3-1. Composition of the Household Questionnaire.

Module Item/Module I

Cover Page

1 Metadata Location identifier

2 Household Roster Key demographic

3 Consumption Detailed collection

4 Forest Resources Qualitative / quant

5 Agriculture Farm production, e

6 Employment Employment type,

7 HH Enterprise Enterprise manage

8 Remittance Remittances to and

Informed Consent (English) Signature page

Informed Consent (Telugu) Telugu copy given


descriptionn


s, date; informed consent

information

Sof expenditure data

itative forest-related data

expenditure & livestock data

time employed, income

ment, income, & cost data

I from the household



to respondent

Total


Metadata (Module 1) are the key identifying information about how the survey was

conducted. This module contains some crucial pieces of information, such as the unique

household identification number, the VSS identification number, and the identification code of

the interviewer. Certain qualitative aspects of the interview that are recorded include the name of

respondent, the date of the interview, and the starting and ending time of the interview.

According to Grosh and Mufioz (2000), there are three main reasons why the collection of

metadata is important. First, for substantive analysis: metadata is often crucial for certain

purposes, such as the calculation of sample weights. Second, for survey management: metadata

helps to assess the time needed to complete implementation, to anticipate replacement

households, etc. Third, methodological research can be assisted with use of metadata, and this

module can even incorporate research questions to help improve future surveys.


Pages


1

1

8

6

3

1

3

1

1

1

27





The Household Roster (Module 2) records the name and sex of each member of the

household, as well as the relationship with the head of the household (e.g., wife, daughter,

grandson, etc.). Key demographic data collected here for each individual include their age,

education, and time spent living in the household. The latter information is important to

determine if the person is counted as an official member of the household; the LSMS format

guidelines requires nine months per year of residence to officially be considered a household

member. Other associated details that are important are the VSS membership status of each

individual, and their occupation or main daily activity (e.g., school, chores, etc.).

The Consumption Module (Module 3) is designed around items familiar to respondents:

flows of goods and cash flows (e.g., taxes or remittances). Deaton and Zaidi (1999) define

household consumption as all reported expenditures on individual goods and services, and all

non-market consumption (e.g., own production and/or in-kind transfers). Consumption must be

comprehensive for accuracy in measuring welfare; thus, reliance on one or a few items as a

proxy is invalid.

The level of disaggregation of the consumption list will vary based on the requirements of

a particular study. For instance, the comprehensive Indian National Sample Survey (NSS) has

used long lists with good results (i.e., little respondent fatigue, good accuracy). On the other

hand, LSMS survey lists are typically much shorter: examples include a Pakistani survey with 33

food items and 20 non-food items; and a Vietnamese survey with 45 food and 46 non-food items.

The trade-offs in time efficiency versus accuracy are implied and will be site-specific; research

from LSMS surveys have not settled the debate as to whether shorter lists provide sufficient

accuracy, although some surveys have indicated this is so. The LSMS draft module, which is

typical of past LSMS surveys, included 70 to 100 total items. This study incorporated





approximately 50 food items and 35 non-food items, which falls within the recommended

guidelines (Deaton and Grosh, 2000).

Recall periods are one of the most difficult, yet important, design issues of the

consumption module. The objective is to obtain a reasonably accurate estimate of the rate of

household total consumption expenditure over the previous year. The proto-typical LSMS format

asks about consumption expenditures for the past year, in addition to the recall period chosen. In

general, however, there are different recall periods for different items. High-frequency non-food

items such as tobacco, newspapers, etc. use recall periods of 1 to 2 weeks. Low-frequency items

may have recall periods of 1, 3, 6, or 12 months; or different recall periods for individual items

(e.g., soap in a month, vacations in a year). Deaton and Grosh (2000, pp. 112-13) discuss this in

greater detail.

Basic options for a LSMS survey are single-visit or multiple-visit; and the choice will help

determine the design of the recall period. Typical LSMS surveys are large undertakings that

incorporate multiple visits; thus, the standard LSMS format is to use two recall periods-the

time since the last visit (typically two weeks), and the "usual month" period. According to

Deaton and Grosh, "...including multiple visits is probably not the highest priority for improving

the typical LSMS survey." The following guidelines are given: if one is to compare results with

other surveys, then use the previously established recall period; if not, use the general LSMS

design format with modifications if necessary. If using a single visit, the "within the past two

weeks" recall period can be substituted for "since last-visit" recall period of multiple-visit

surveys. In general, a single-visit strategy is acceptable because consumption is smoothed

throughout the year.





For logistical reasons, each household sampled in this study was only visited once.

Therefore, the food purchases component utilized a recall format that allowed the respondent to

self-identify the purchase frequency of each item, prior to the interviewer inquiring about the

purchase quantity and units, and the total purchase price.

Imputing the values of non-market transactions is difficult in countries that are not highly

monetized (Deaton and Grosh, pp. 116-17). For LSMS surveys, food is the most important

imputed item in household budgets, and generally comes from home production or sometimes as

gifts. The recall period can be a year or a "usual month". The main problem is with valuation

because respondents are asked to hypothetically assign a value to items that are often rarely

purchased or traded. The recommendation is to collect quantities of such goods, ask respondents

about prices, and cross-check with other data. For the this survey, the HH questionnaire asked

the respondent to recall if there was any home production during the previous year for the food

goods listed. For positive responses, they were asked the total quantity and the units of measure;

they were also asked to estimate the total value of this home production.

The Forest Resources Module (Module 4) consists of two parts. Part A contains 29

questions, loosely divided into four parts, which focus mainly on CFM awareness and

participation. Because specific information related to VSS details were sought, Part A was only

administered to VSS-member households; and only if the survey respondents) was an actual

member of the VSS. The number of questions were about evenly split between questions

requiring yes/no responses, Likert scale or other categorical responses, and magnitude responses

(e.g., how many hectares of VSS forest area are there?). Part B quantifies the sales value and

home consumption value of NTFPs, fuelwood, and small timber; the basic format is styled on the

part of the LSMS agriculture module that quantifies crop production and sales. In addition to





fuelwood and small timber, Part B lists approximately 30 NTFP items that might be collected by

respondent households, and (if any are sold) inquires about where and to whom these products

were sold.

Farm size and income derived from agriculture are the two main pieces of information that

the Agriculture Module (Module 5) collects. The Employment Module (Module 6) tabulates

income from work performed outside the household. The Household Enterprise Module (Module

7) is included mainly to gather information on household income derived from home production

activities. The Remittance Module (Module 8) concludes the HH questionnaire by inquiring

about monetary flows into and out of the household that involve cash exchanges with household

members, usually in the form of gifts to or from relatives (although pension income is

occasionally listed here also).

Design and Structural Guidelines

The ordering of the modules within the questionnaire will depend upon the size and

implementation of the survey. Respondents themselves may indicate the best ordering during

field testing, especially if return visits are scheduled for later in the day or week. However, all of

the interviews are completed in a single trip to a primary sample unit (i.e., VSS villages) in

smaller, scaled-down LSMS surveys such as this study. In general though, different factors such

as best recall period, the natural or logical location (e.g., end of survey for sensitive information),

etc. will help determine where certain data are to be collected in the interview. The essential

point is to ensure that data important to the study is collected in at least one of the modules.

All surveys begin, however, with the metadata and household roster modules. Following

these modules, the primary respondents are usually determined; the other modules are collected

as applicable. Education, housing, and migration are good topics to open with, once the metadata





and household roster are obtained. Employment and other sensitive topics (savings, credit,

transfers) should be covered at the end of the interview (Grosh et al., 2000).

The issue of survey length is extremely important: a general goal is to keep actual survey

interviews for any particular respondent to no more than an hour per day. This can vary given the

tolerance of the people in a given country, and may depend on local conditions. LSMS

experience shows that tolerance for a long interview is less in urban versus rural areas, lower for

wealthy versus poor households and lower in wealthier countries. Grosh et al. (2000) offer

guidelines that can be used to pare down a survey to the specific needs of a particular research

study. For example, many LSMS prototype modules come in both a long and short version, and

the particular needs and resources of a given research study will determine which version is

preferred. Although choosing the shortest version of modules will allow the analysis of more

objectives, depth should not be sacrificed for those objectives deemed most important to the

particular study. For instance, the Consumption Module was the most extensive module in terms

of length (in time required to complete) and depth (number of items/questions)-reflecting its

central importance to this study. In addition, the Agriculture Module was based on the "short"

version of the LSMS agriculture module, and it was modified even further to simplify and

shorten it.

Draft modules for field testing are recommended to help determine and judge the trade-offs

being made with respect to data collection. Their acceptability will either be confirmed, or will

suggest that corrections are necessary, or that a suitable alternative must be found. Draft modules

also allow for the review of the coding and nomenclature to ensure consistency throughout the

questionnaire, especially with regard to similar questions (Grosh et al., 2000). A draft

questionnaire is also important to recognize gaps and overlaps between modules, and testing





allows them to be modified or corrected as necessary. A draft questionnaire also helps to ensure

that at least one of the modules is collecting data important to the study.

Formatting Guidelines

Below are some important issues pertaining to the formatting of selected questionnaire

components as related by Grosh et al. (2000) per the LSMS format:

* Pre-coding and code boxes should be used extensively to increase efficiency and reduce
data entry error.

* Response codes should be located next to the question. A code key can be placed
somewhere on the page.

* Response codes corresponding to answers must be clear: simple, mutually exclusive, and
exhaustive. They should be designed so they are not likely to provoke the same response.

* Lists of items (e.g., food goods in the consumption module) that respondents are asked
questions about help the efficiency/accuracy of the survey. By first enumerating all items
purchased before collecting details on each item, the temptation for the respondent to not
list something is avoided when they realize there are several questions about each item.

* Use uppercase letters for instruction to the interviewers, while lowercase letters are for the
actual questions asked of the respondents (Fowler, 2002).

* The questionnaire should be designed so that interviewers always ask verbatim questions
to ensure uniformity among interviewers and between respondents.

* Two or three simple questions should be asked instead of one long, complicated question.
Qualifiers are important (e.g., "What was the main reason...") to help obtain mutually
exclusive answers where more than one answer could apply. When appropriate, the
following convention is also used: other (specify).

* Probe questions are common in consumption, agricultural, and similar modules that
attempt to get at "how much" of something. Interviewers need to know what to probe for,
and how to do it. This technique should reduce the number of "I don't know" responses,
for which the "DK" abbreviation should be the proper response code.

* Code tables for different units of quantity allow the respondent to choose the unit for
which they are most comfortable. This will also tend to reflect the unit in which the
action/item discussed occurred-which may differ from household to household. Code
tables are of key importance for use in "quantities produced" questions.





Survey Structure and Implementation

HH Questionnaire Testing and Revision

Grosh et al. (2000) emphasize the importance of field testing the survey instrument and

list, by component, some of the key considerations that testing ought to cover. For the overall

questionnaire, field tests should ensure that all the necessary information is being collected, and

that there is internal consistency to the instrument without any needless double-counting. Testing

will also reveal if individual modules collect the intended information, cover all major activities,

and avoid any redundancy. For individual questions, field tests will help to know if the wording

is clear, if the coding works well or not, and if there are any ambiguous responses due to either.

In addition, Fowler (2002) states that questionnaire testing also helps to determine how long it

takes to complete a survey instrument.

For the present study, the HH questionnaire was tested in the Medak district of Telangana

region in late February 2005. The main finding of this test was that the questionnaire was too

long, and redactions were necessary because respondents were showing signs of interview

fatigue. Changes included shortening and simplifying the Consumption Module. For example,

the extensive list of food items in Part B was reduced by eliminating less common items. In

conjunction with this, more use was made of blank spaces for "other" items not specified.

Further reductions were made by consolidating separate items into like categories in Part C

(Non-Food Goods).

A more significant change was to revise Part B to allow the respondent to self-identify the

purchase frequency of each item. The initial design of the Consumption Module relied on the

LSMS frequency format: a 7-day recall period for daily expenditures (e.g., tobacco); both a 2-

week and a typical-month recall period for food goods; and both a 30-day and a 1-year, recall

period for non-food goods. Changing this avoided the time consuming (and potentially error-





prone) process of inquiring about purchases in the previous two weeks-followed by asking

about the purchases in a typical month. For example, the question became: "How often does your

family purchase [...]?" with given responses pre-coded (e.g., weekly = 1, fortnightly = 2, etc.).

Changing to this configuration for food items (Part B) also allowed greater flexibility for the

respondent, by allowing them to better answer the query in a format closer to how they mentally

relate their actual usage of a given item.

Further technical efficiency gained could only have come from extensive field trials that

were not logistically possible at the time. Therefore, a second round of formal field testing was

foregone, due to time constraints. This mainly had to do with the fact that the survey needed to

be completed by June, when the monsoon season begins. Moreover, interviewers had yet to be

hired and 1,200 questionnaires had to be printed. In addition, as several different individuals in

India contributed to the development of the household questionnaire, it was decided to proceed

with the instrument as revised.

Sample Selection

This research study was designed to obtain a sample of VSS villages from each of the three

regions of Andhra Pradesh. Thus, the field survey is based on the random selection of 20 VSS

villages from a district chosen to represent each one of these regions. Selection of villages was

made using the random number generator in Microsoft Excel. The total number of 60 VSS was

arrived at in order to ensure sufficient degrees of freedom when conducting the subsequent

econometric analyses. For the household survey, each of the 60 VSS villages had 20 households

randomly selected for interviews. Thus, the total possible number of observations is 1,186

households (which is less than 1,200 because three VSS villages had less than 20 households).

Random selection of households was made using either Microsoft Excel or a random number

sheet in the field.





The APFD apportions the state into administrative units (or so-called "Circles");

occasionally these circles conform to the districts observed in Figure 2-2, but in general they are

aggregated somewhat into larger units. Each circle is comprised of smaller divisions: the

Visakhapatnam Circle contains five divisions, for example. As such, the following APFD forest

divisions were selected for sampling from within each of three circles: 1) Nirmal and Jannaram

Divisions (Adilabad Circle); 2) Visakhapatnam Division (Visakhapatnam Circle); and 3)

Chittoor West Division (Anantapur Circle). Random probability sampling was employed to

select 20 VSS villages within each circle. The two divisions in Adilabad had their sample split

based upon the relative number of VSS villages present in each division. Thus, 14 VSS were

sampled in the Nirmal Division and 6 VSS were sampled in the Jannaram Division. Within each

sampled VSS village, 20 households were also selected using random probability sampling

techniques in order to ensure statistical validity when drawing inferences from the results.

It is important to point out, however, that the selection of the study areas (circles) within

each region (i.e., Telangana, Rayalaseema, and Coastal Andhra) was not random. The Adilabad

Circle was ostensibly chosen by the APFD on the basis of security concerns, and the Nirmal and

Jannaram divisions of this circle were selected to represent two different forest regimes within

the Telangana region. The forested area of the Nirmal Division is dominated by teak, while the

Jannaram Division has more bamboo present for economic and consumptive activities. The

Chittoor West Division of Anantapur Circle was also chosen by the APFD for inclusion in the

survey-as was the Visakhapatnam Circle, although the principal investigator (PI) had originally

selected this area to represent the Coastal Andhra region. (The representative division of this

circle that was actually sampled [i.e., the Visakhapatnam Division] was randomly selected,

however.) The implication of the selection process is that each regional part of the overall survey





is essentially representative only of that specific location within Andhra Pradesh, and

aggregation of all three parts reduces the validity of inferring results to the state as a whole.

Nevertheless, each village within a given region was still selected at random, and the evaluation

of the specific hypotheses of the study takes precedence over the ability to apply the results on a

wider scale.

For comparison, the Centre for Economic and Social Studies (CESS) in Hyderabad

conducted a survey of Andhra Pradesh in 2004 to investigate the impact of JFM/CFM on

livelihoods (Gopinath Reddy, personal communication). Much like the present research, the

objectives of the CESS study included examination of the institutional and economic dynamics

of CFM at the micro level (Reddy et al., 2004). Although much smaller, the CESS survey is also

structurally very similar to the present study-selection of one district from each of the three

regions of Andhra Pradesh (Adilabad, Visakhapatnam, Kadapa) with six villages from each

district (3 VSS villages and 3 non-VSS villages). At the final sampling stage, 25 households

were interviewed in the VSS villages (225 total), while 15 households were interviewed in the

non-VSS villages (135 total).

Reddy and Chakravarty (1999) used a much smaller survey to investigate forest

dependence and income distribution in villages of northern India. Their survey was based upon

four "development blocks" (out of 15 total blocks in a single district) selected because they were

contiguous to forest areas. Twelve villages were randomly selected from the four blocks,

followed by individual households serving as the final sampling unit. Households were selected

by simple random sampling with replacement, and the total number of usable household

questionnaires equaled 233.





Interviewer Training and Implementation of the HH Survey

During the first week of March 2005, four enumerators were hired in Hyderabad to

conduct the household interviews. Staff members of CESS were instrumental in identifying

qualified interviewers. Three men that were hired each held masters degrees and had extensive

experience with conducting rural household interviews, for both CESS and other rural welfare

agencies. In particular, each had done similar work in the district of Adilabad, where the first leg

of the survey was to commence. Although the fourth man had limited survey experience, he had

served as an agricultural extension agent in his home district of Adilabad.

Training took place in the town of Nirmal, where the purpose of the survey and the format

of the questionnaire were explained in more detail. In order to become familiar with the

instrument and procedures, mock interviews were conducted where each member interviewed

another-taking turns as both interviewer and respondent. This follows one recommendation of

Fowler (2002), who also emphasizes that interviewer training should cover from two to five

days. Following this practice session, Vishnu Reddy (the research associate of the PI) had an

extensive discussion, in Telugu, with the interview team about their questions, problems, and

other concerns. The PI was also there to answer questions and to comment about procedures.

Subsequently the next phase of training consisted of actual data collection in the field-an extra

village had previously been selected for this purpose, so each person interviewed a total of 20

households. In addition, the first two VSS villages surveyed are dropped from the analyses and

effectively become additional training villages. Therefore, the total number of observations for

analysis is 58 VSS villages.

The second round of sampling took place in the Visakhapatnam Division of the

Visakhapatnam Circle in early April 2005. It was determined that two teams were required to

reduce the amount of time needed to complete the research effort. Thus, prior to the beginning of





the second round of sampling, an additional four enumerators were hired. These men were also

college graduates and most possessed previous survey experience as well. The pre-field training

followed the procedures discussed above, with the original sample team assisting in the mock

interviews and post-exercise discussion. For this leg of the survey, however, the field training of

the new enumerators was increased and consisted of household interviews in two villages (for a

total of 40 households per interviewer). When the actual sampling commenced, each new

interviewer was paired with a member of the original sampling team.

The overall survey proper officially began on March 13, 2005; data collection was

completed in the Adilabad Circle on April, 1 2005. Following the hiring and training of

additional personnel, fieldwork resumed on April 10, 2005; data collection was completed in the

Visakhapatnam Circle on April 19, 2005. Fieldwork for the final round of sampling commenced

on May 4, 2005, in the Chittoor West Division of the Anantapur Circle. Like the previous round,

the household surveys were again conducted by two interview teams. However, one team was

now comprised of three enumerators, due to an unrelated injury sustained by one of the

interviewers during the hiatus between the Visakhapatnam portion of the survey and the

resumption of fieldwork in the Chittoor District. Data collection fieldwork for the study was

completed on May 16, 2005.

Implementation of the VSS Survey

Prior to visiting villages selected for sampling, attempts were made to obtain secondary

information from local APFD offices, local Velugu (a statewide rural poverty reduction project)

offices or, on occasion, local magistrates. APFD division and field offices were visited to obtain

VSS micro-plans for relevant information (e.g., VSS size or date of VSS inception), to observe

topographic maps, and/or to gain useful anecdotal information. Local Velugu offices were often

able to provide access to a list of households, and/or hand-drawn "social maps" of dwelling





units. This information was usually photocopied, or occasionally borrowed. If unable to do either

of these, then this information was used to randomly select households at that moment in the

Velugu office by recording the names of the household heads into a laptop or a notebook, for use

in implementing the random selection techniques previously mentioned.

Obtaining secondary information from these sources (especially from Velugu) was often a

time consuming and frustrating job, but proved extremely beneficial when successful. For

instance, the household lists offered an objective estimate of the number of households for a

given village, and meant that the random selection of households could be made prior to the field

visit. This was the main benefit, as it allowed the interview team to avoid this process in the

field, saving time and minimizing potential errors: they simply had to contact the pre-selected

households.

Once the secondary information was in hand, the PI and his research associate would visit

each village in order to conduct the VSS interview. The first task was to locate either the VSS

Chairman and/or Vice-Chairman for introductions and to give an explanation of the research

study. For cultural and practical reasons the PI and his research associate usually interviewed

whichever person was male. If the VSS Chairman was a woman, she may not feel comfortable

being interviewed: in such cases the interview would be conducted with the Vice-Chairman, with

or without the presence of the Chairman.5 In one village, the husband of the VSS Chairman was

the former Chairman himself. Thus, this man was interviewed as he was obviously the de facto

VSS Chairman of his village. Often during the interviews, a few other members of the





5 When the women did participate in the interview, whether they were the Chairman or the Vice-Chairman, they
almost always provided little input. Often it seemed that the women who were VSS Chairman or Vice-Chairman
were mere figure-heads serving only to comply with the CFM institutions.





community (whether or not they were VSS Management Committee members) would usually be

present and offer input to the discussion.

The interview was quite informal in terms of adhering to a rigid survey format of

verbatim questions. The respondents) would be engaged in a quasi-structured conversation (in

Telugu) based on the VSS questionnaire, which allowed for a certain amount of flexibility in

terms of question order and overall flow of the interview.6 Following the interview, the

researchers would ask to take a walk through the VSS forest area (schedule permitting). Often

the conversations would continue during the inspection of the protected VSS forest areas and/or

plantation areas, as the villagers were usually enthusiastic to show off their forest area and the

works they completed. Of the 62 VSS villages surveyed (including training villages), the

researchers were able to inspect at least two-thirds of the VSS forests and/or plantations.

The visits to each village were intended to be unannounced in order to obtain an

independent and unbiased assessment of each VSS-the goal being to conduct the interviews

without any positive or negative bias that could potentially result from the outside influence of

the APFD. Occasionally, field officers of the APFD would already be present upon arrival in a

village; this was because they knew of the research effort when VSS micro-plans had been

borrowed. In such cases, the PI would ask the forest officers to show him the VSS forest area in

order to draw them away from the on-going interview: for this reason the PI was not present for

approximately five or six of the interviews proper. In addition, for visits to the VSS villages of

the sample, the PI declined transportation provided by the APFD so as to similarly avoid any

direct or official association with the government. This is an important consideration for



6 Although the PI could not understand the words, he could follow along with the general mood and tenor of the
interview. Upon request, important passages were translated and the PI would pose additional questions to expand
on the explanations.





obtaining unbiased interviews in the rural villages of Andhra Pradesh, according to CESS

Research Fellow Gopinath Reddy (personal communication). Therefore, the PI generally rented

a vehicle or an auto-rickshaw, rode the bus if the location was close enough.

Before leaving a village, the PI (through his research associate) would inform the people

interviewed that a sample team would be coming within a few days to conduct the household

interviews. The research associate would ask one of the villagers to assist the sample team in

locating the households that were selected for interviews. Sometimes a sample team would

accompany the PI to a village (usually in remote areas, or at the beginning or end of a sample

leg) and administer the HH questionnaire while the VSS interview was being conducted; but

most of the time the sample teams) worked independently of the PI and his research associate.

For this reason, the PI made every attempt to secure a list of households in advance, but in the

event that they were unobtainable it was necessary to do the random selection of households in

the field. The desire of the PI to avoid this situation by using pre-selecting households was based

on two observable facts: 1) many villages lack an orderly layout, and 2) it is often difficult to

distinguish distinct dwelling units amongst a collection of structures. Thus, the likelihood of

deviating from a completely random sample increases in such a scenario, especially the larger

the village and the less orderly its lay-out.

In the first few days of the survey, the PI was afforded the opportunity to demonstrate to

the interview team how to randomly select households without a list. For example, in the VSS

village of Rampur (Adilabad), the 340 households were divided by 20 (i.e., the intended sample)

to calculate a selection interval of 17. For each interviewer, the PI selected a number ranging

from 1 to 17 that was taken from the random number sheet. The village was divided into four

sections and each enumerator was to interview every seventeenth household, beginning with





whichever household corresponded to the random number that was selected for them. This

procedure was subsequently followed (to the best of the PI's knowledge) by the sample team

leader when the PI was not present for the situations where households were randomly selected

in the field.

Each household interview generally took between 45 and 60 minutes to complete,

depending upon the size of the household, their level of consumption, and sources of income.

Each team would usually complete one village per day (i.e., five households per interviewer),

although at times it was feasible to finish two villages if they were located near each other. Non-

cooperation (respondent refusal) was not a problem whatsoever, although occasionally a

household would be unavailable (nobody home), said to have migrated, or non-existent for some

reason. Such households were initially replaced by interviewing the next household, but later

alternate random selections were made for this scenario.

Data Compilation

Following completion of the interviews, a template file was created using Microsoft Excel

that essentially mirrored the HH questionnaire. This allowed data from each household to be

easily transcribed into this format for later compilation into master files for each village, and

eventually to an overall aggregate file. Each household file contains internal calculations that

compute, for example, total household consumption expenditures and total household income. In

addition, there is summary page where important data for the household are compiled into a

single row of cells, in order that this line of data can easily be copied and transferred to a master

file for the corresponding village.

For the job of transcribing the data, the PI retained two of the interviewers that were

available and also hired a woman specifically to do this work. These three people did the

majority of the transcription, although others (including the PI) occasionally assisted also.





Initially, the PI was able to supervise the transcriptions for quality assurance (QA) by instructing

the data entry personnel on how to handle specific ambiguities in questionnaire responses and/or

information as recorded in the field. By checking completed files to identify and correct entry

errors, instructive feedback to the transcription crew was also provided. After the PI departed

from India, the research associate continued the supervision activities. Prior to sending the

completed files to the PI, the research associate conducted initial quality control (QC) checks of

completed household files (approximately 5 per village).

The PI completed the final QC check for all household data files. When assembling the

village master files, it was necessary to open each file to copy and transfer the summary line of

data mentioned above. Thus, this opportunity was used to compare the HH questionnaire with its

associated data file to ensure correct transcription and to rectify errors, which was a very time

consuming process. However, it was necessary to perform this function because of persisting

ambiguities with certain sections of the HH questionnaire with which the transcription team had

problems. For example, NTFP consumption in Part B of the Forest Resources Module is prone to

errors, in terms of the frequency and amounts collected, that need to be corrected. The confusion

stems from the rigidity of the format: respondents were asked about their consumption on a

monthly basis only. As most items collected from the forest are done so seasonally, or on an

otherwise irregular basis, the self-identification of frequency (adopted for the Consumption

Module following pre-testing) should have been the standard here also. Thus, it is often difficult

to discern the total yearly consumption amounts from the recorded answers; interviewer

heterogeneity could also compound the problem.

Other QC activities were also conducted by the PI. As they were basically transferring data

verbatim, the transcription team was necessarily given little responsibility to interpret the data.





Thus, QC consisted of checking for the correct placement of information (e.g., income from a

home enterprise in the HH Enterprise module instead of the Employment Module), and

monitoring the internal consistency of calculations. The correction of all types of errors was

made as objectively as possible in an effort to "clean" the data for subsequent analyses.





CHAPTER 4
MODELS, HYPOTHESES, AND DATA

Overview

The analytical focus of this study is concerned with factors influencing the economic

impacts of participatory forest management in Andhra Pradesh, India (i.e., JFM/CFM). This

study will capture such influences to the extent that the three measures representing different

aspects of social welfare vary among the cross-section of VSS villages that comprise the sample.

The three indicators include the mean per-capita household consumption value of a given village,

the inequality in consumption among households in the village, and the level of poverty within

the village. The first two are used to define social welfare within an economic context, while the

third can be considered an alternative measure of social welfare since it focuses on the portion of

a village (i.e., individuals in households) below a specified poverty level. Evaluating whether,

and to what extent, various explanatory variables affect each individual economic indicator will

help to empirically assess if there has been any economic impact of JFM/CFM.

To accomplish this task, each of the economic indicators will be used as dependent

variables in separate models that will largely rely on a common set of explanatory variables. The

explanatory variables used in the econometric models are summarized in Table 4-1. The

variables are categorized into the following four groups: Demographic, Economic, Bio-physical,

and Institutional. There are five to six variables in each group and 21 in total. The following

section begins with a brief discussion of the selection of the basic welfare measure and its

application. Social welfare is defined in an economic context and then each model is described in

more detail, including the implicit assumptions behind each of the indicators chosen to represent

these aspects of social welfare. Next, a section discussing the hypotheses of each variable shown

in Table 4-1 is presented. Lastly, a detailed description of the data concludes this chapter.





Table 4-1. Explanatory variables utilized in the VSS-level regressions.
Variable Variable Description Unit Type


Demographic:
N Number of households (HH) in a VSS village
ST VSS predominately comprised of Scheduled Tribes (1 if yes)
EDU Mean HH Education level
NGO Number of NGOs working in the community
DGI Index: number of DWCRA Groups divided by N


Economic:
YFP
YAL
YOS
IHW
TRN


Mean HH Income from Forest Products
Mean HH Income from Agriculture and Livestock
Mean HH Income, Other Sources (employment, enterprises, etc.)
External Investment per HH (avg. wages for VSS works/HH)
Number of Training events/field trips by VSS MC members


Bio-physical:
LT Length of Time under JFM/CFM
LT2 Squared value of LT
FA VSS Forest Area size in hectares (ha)
RRE Relative Resource Endowment (VSS forest area per VSS member)
PFC Percent Forest Product Consumption
DTW Depth to groundwater (proxy for spatial environ, heterogeneities)
Institutional:
VBA Percent of VSS HHs with general Boundary Awareness
GAI Mean HH General Awareness of Institutions
CCA Mean HH proxy for Collective-Choice Arrangements
FP Presence of Formal Patrol of VSS forest area (1 if yes)
GS Graduated Sanctions for rules violations (1 if yes)


Number
0, 1
Years
Number
Percent


Rupees
Rupees
Rupees
Rupees
Events


Years
Years
Hectares
ha/mem.
Percent
Feet


Percent
0to7
Percent
0, 1
0, 1


Integer
Dummy
Continuous
Integer
Ratio


Continuous
Continuous
Continuous
Continuous
Number


Integer
Integer
Integer
Continuous
Ratio
Continuous


Ratio
Continuous
Ratio
Dummy
Dummy


Empirical Models

The empirical modeling of this study necessarily begins with the construction of the three

economic indicator variables. First, a suitable measure to represent the living standards of

individuals or households must be selected. The standard choices to measure economic well-

being are either income or consumption. Deaton (1992, 1997) and Deaton and Grosh (2000)

explain that while income is a superior welfare proxy for households in developed countries,





consumption serves as a better proxy for social welfare in less developed countries. Income data

are generally inferior to consumption data for the measurement of living standards of rural

people: data on consumption over a period shorter than a year gives a more accurate estimate

than income. This is because consumption is affected less by seasonality than is income, which

can be highly seasonal (especially in agriculture). Moreover, it is more cost-effective to gather

consumption data and it is more reliable, as people are more likely to inaccurately report income

and assets (for instance to lower tax liabilities). The Indian National Sample Survey (NSS) also

focuses on consumption rather than income. For these reasons, consumption is used to measure

social welfare in this study.

Household welfare is representative of utility as attained by the consumption expenditures

of the household budget constraint. Therefore, the main purpose of the HH questionnaire

described in Chapter 3 was to collect household-level consumption data so that the economic

measures of social welfare could be constructed for each VSS village. Deaton (1997) emphasizes

using individuals as the basis for a welfare measure because "it is hard to think of households as

repositories for well-being" (p. 150). For example, he discusses how to transform household

consumption data into individual welfare measures, including the complexities and practical

constraints involved in assigning different consumption values to different members of the same

household. In the end, however, Deaton (1997) recommends a simpler method which he deems

to be the best practice: assigning the per-capita household consumption value to each individual

in a given household. This is the procedure followed in this study, where the total consumption

value of the household (tcv)7 is divided by the total number of household members (hhs) to

obtain x, the per-capita household consumption value.


7 The household total consumption value equals all consumption expenditures, plus the imputed consumption of
food items produced or collected, and the own consumption value of goods produced by a household enterprise.





Social Welfare Function

As explained by Deaton (1997), a measure of social welfare (W) transforms individual or

household consumption data of a population into a single summary value that is useful for policy

analysis. The general form of W that Deaton (1997) presents is decomposable into two parts

(Equation 4-1), so that the social welfare of a given population is represented by the average

level of consumption (gp) and its distribution (1):

W = 4( 1 -I ) (4-1)

Equation 4-1 implies that if each household in the population has the mean level of welfare

(i.e., the case of perfect equality), then I = 0 and social welfare itself is equivalent to p. 8 This

expresses a societal preference that more equal distributions of social welfare, for a given level

of p, are superior to less equal distributions. Thus, any deviation from a totally equal distribution

of welfare will necessarily result in social welfare (W) being less than the mean value (p). In

other words, gt is the highest level of welfare that is attainable, ceteris paribus. This identifies gt

as a purely economic indicator of social welfare that remains uncorrected for distributional

inequality.

An important aspect of Equation 4-1 is the explicit illustration that inequality is not

synonymous with social welfare. Indeed, it is possible for W to increase while the inequality

measure is also increasing, but only if average consumption (p) increases enough to offset the

decrease caused by I (i.e., the rich gain more than the poor, although everyone gained). Such a

situation is still a Pareto improvement from the initial scenario. Nevertheless, given that social

welfare consists of these two elements, it is necessary to decompose W into two separate



8 Note that because perfect equality does not exist, empirically I will always be non-zero so that W = jt is merely a
theoretical point of abstraction.





variables in order to analyze them individually. Thus, the present study utilizes each component

of W as a dependent variable for regression analysis, where gL is measured as the mean per-capita

household consumption value for a given VSS village; and I is measured by the Gini coefficient

(y), which is typically used to represent inequality.

Average Consumption Model

The first measure, which is the focus of this subsection, is p. It is calculated for each VSS

village as the mean per-capita value of household consumption. As data on consumption were

collected at the household level, the per-capita household consumption value x was first applied

to each individual in a household following Deaton (1997) as discussed previously. In order to

derive gL as based on individual consumption, Equation 4-2 below was used:

J
(x, -hhs,)
t =G (j = 1,...,J) (4-2)
(hhs, )


where xi denotes that the per-capita household consumption value x is applied to an individual i;

hhs is the household size (i.e., number of household members) of the /h household sampled in

the VSS village; and J is the total number of households sampled in the VSS village. By using

random sampling techniques to select the households in each village, Equation 4-2 is used to

calculate an estimate of g for each of the 58 villages surveyed. The model used to estimate (t is

specified as:

= e(+E) (4-3)

and, by taking the natural log of ,. Equation 4-3 is transformed into a semi-log model:

In X = X'/ + e (4-4)





that can be estimated using the ordinary least squares regression procedure. Equation 4-4 is

comprised of the following components: X is a matrix of independent variables, P is a vector of

coefficients associated with X, and e is the disturbance term. The matrix X is composed of the

following demographic, economic, bio-physical, and institutional variables listed in Table 4-1.

Demographic: N, ST, EDU, NGO, DGI

Economic: YFP, YAL, YOS, IHW, TRN

Bio-physical: LT, LT2, FA, PFC, DTW

Institutional: GAI, CCA

Two location dummies (ADIL and CHIT) are also included in the matrix X in order to control for

the different spatial regions of Andhra Pradesh from which the data were collected.

Consumption Inequality Model

As described in the previous section, an important aspect of Equation 4-1 is that inequality

I is a separate component of W, and is therefore not valid as an independent measure of social

welfare. As a dependent variable, I evaluates the degree to which welfare is distributed in an

inequitable manner throughout a given community. As mentioned, the typical measure used to

represent inequality is the Gini coefficient (y). This is a ratio measure most commonly associated

with the Lorenz curve, which is a graphical depiction of percentage social welfare distribution

(e.g., by income or consumption) in terms of population quintiles. Following Deaton (1997), the

equation below is used to calculate an estimate of the Gini coefficient for each VSS village:


(= [(M+1, 2 / (P, x, ) (4-5)
M- 1 M(M- 1) ,

where M is the total number of individuals i that belong to households sampled within a VSS

village, and p is the relative rank of each individual, starting with p = 1 for the richest person





sampled, and ending with the poorest person (p = M) sampled.9 As defined previously for

Equation 4-2, xi is the per-capita household consumption value (which is identical for all

members of the same household) while p is the mean of all xi (i.e., mean per-capita household

consumption value). The expression pix1 is summed over all persons sampled in a given VSS

village. The use of random sampling techniques to select the households in each village allowed

the calculation of an estimate of y for each of the 58 villages surveyed.

According to Vanhoudt (1998), a measure of distributional inequality can be

econometrically modeled as a function of non-conventional factors (e.g., institutional

parameters) in addition to typical neo-classical factors such as labor and both physical and

human capital. Odedokun and Round (2001) discuss how recent studies of inequality have

investigated "a broad range of factors" affecting inequality, specifically mentioning institutional

factors. Regressing y on a collection of explanatory variables is undertaken to evaluate the

contribution of these variables to inequality; the model used to estimate inequality is specified as:

y = e(Z'+) (4-6)

which is transformed into a semi-log model specification:

In y = Z'a + u (4-7)

where Z is a matrix of independent variables, a is the vector of coefficients associated with Z,

and u is the disturbance term. The matrix Z is composed of the following demographic,

economic, bio-physical, and institutional variables listed in Table 4-1; Z also contains the two

location dummies introduced earlier:



9 For instance, if the household with the highest per-capita consumption value x has four members, then those four
individuals would have p values of 1, 2, 3, and 4. The Gini coefficient is an example of where it is necessary to use
individual consumption values, following the procedure that Deaton (1997) recommends (as just described), in order
to calculate a socio-economic indicator that is representative of inequality in a given community.





Demographic: N, ST, EDU, DGI

Economic: YFP, YAL, YOS, IHW, TRN

Bio-physical: LT, LT2, FA, PFC, DTW

Institutional: CCA, FP

Location dummy: ADIL, CHIT

Consumption-based Poverty Model

As mentioned previously, measures of poverty can be thought of as a special class of social

welfare measures that specifically address the proportion of the population below some given

poverty line. To the extent that there is a significant difference between poverty and social

welfare in the villages sampled (i.e., poverty is not a uniform condition), then it would be useful

to estimate a model that tries to explain any observed variations in poverty across villages.

Deaton (1997) discusses the derivation and characteristics of several different poverty

measures. One of these measures is the headcount ratio, which is simply the number of people

below a predetermined poverty line (e.g., a threshold level of consumption). The headcount ratio

is a poor social welfare metric, however, because it has the disadvantage of violating the so-

called "principle of transfers" (i.e., a transfer from a poor person to one who is less poor could

conceivably lift the latter above the poverty line). The poverty-gap ratio (PGR) rectifies this

deficiency by calculating the difference in an individual's welfare (which can be based upon

either a measure of consumption or income) with the given poverty line, and normalizing to this

line. Summing this over all individuals below the poverty line, and dividing by the total

population, results in a ratio that (when multiplied by -1) makes a suitable measure of social

welfare. The PGR is a poverty measure that is commonly used by development researchers (e.g.,

Reddy and Chakravarty, 1999) and is the measure that will be used in this study. Following

Deaton (1997), the poverty-gap ratio is estimated for each VSS village by calculating:






PGR = -l 1-L for all xi < z; or (4-8a)


PGR = 0 if all xi 2 z (4-8b)

where z is the predetermined poverty threshold. It should be noted that considerable debate

surrounds the construction and use of poverty thresholds and researchers generally rely on

existing estimates (Deaton, 1997); this study utilizes a consumption-based poverty line already

established for rural Andhra Pradesh. Ceteris paribus, the PGR will increase the greater the

difference between xi and z, or if more individuals fall below z. Thus, as an economic measure of

welfare for the population below the poverty line, higher PGR values indicate a greater level of

inequality in the distribution of consumption within a VSS village.

Allanson and Hubbard (1998) discuss how to empirically estimate the income-gap ratio

from a random sample of different income classes, and clearly this technique also applies to

consumption data. Additionally, they relate the income-gap ratio to second-degree stochastic

dominance, a concept whereby integration of cumulative distribution functions allows for the

ranking of welfare distributions. Moreover, second-degree stochastic dominance is equivalent to

generalized Lorenz dominance (Deaton, 1997), which itself permits the ranking of standard

Lorenz curves by scaling them up using the mean of the distribution. Thus, randomly sampled

household consumption data can be aggregated into a PGR value (estimated for each VSS

village), which serves as the dependent variable in the equation below:

PGR = Y' + v (4-9)

where Y is a matrix of independent variables, 6 is the vector of coefficients associated with Y,

and v is the disturbance term. The matrix Y is composed of the following variables:

Demographic: ST, EDU, NGO, DGI





Economic: YFP, YAL, YOS, IHW

Bio-physical: LT, LT2, RRE, PFC, DTW

Institutional: PBA, CCA, FP, GS

Location dummy: ADIL, CHIT

Hypotheses for the Explanatory Variables

Demographic Variables

For the empirical analysis, N is the total number of households belonging to the VSS

village regardless of whether the household is a VSS member or a non-member. This variable is

expected to be positive with respect to p, meaning that economic well-being ought to be greater

the larger the VSS. As such, an inverse relationship with PGR is expected, as small communities

are more likely to be poor.10 This is because small communities were generally observed to be

more isolated and remote, and with employment opportunities that were generally more limited

as compared to larger villages. The relationship with y is also anticipated to be positive, which

would reflect greater inequality due to a larger, and presumably more socially diverse,

community.

ST is a binary dummy variable equal to 1 if a given VSS village consists entirely or

predominately (as based on the sample mode) of households of the Scheduled Tribes caste-

designation. As one of the main stakeholder groups identified in the JFM legislative orders, it is

important to control for the Scheduled Tribes VSS villages in the sample. A negative relationship

with gt may be predicted because Scheduled Tribes villages are often more remote than other

villages and thus less-integrated with the larger local economy. (A positive relationship with

PGR might be predicted for the same reason). The relationship between ST and y ought to be

10 For variables that are included in both the In ai equation and the PGR equations, the relationships with the
dependent variables will usually be opposite of each other as the prediction for N illustrates.





negative, as Scheduled Tribes communities are usually very homogeneous both ethnically and

economically. However, as Scheduled Tribes have generally benefited from other tribal

development projects sponsored by the government or external donors, it is possible that the

opposite relationships (i.e., ST having a positive affect on t and a negative effect on PGR and y)

will be observed (Janaki Alavalapati, personal communication).

EDU represents the educational level of a given VSS. It is computed by calculating the

weighted mean" household education level for each VSS village; where the education level of a

household is the sum of school years completed, divided by the total number of household

members greater than three years of age. A positive relationship with g is anticipated, and an

inverse relationship with PGR is expected. The relationship with y is likely to be positive,

meaning that more education is associated with greater inequality.

NGO is a binary dummy that equals 1 if a non-governmental organization has provided

forestry-related assistance to a given VSS. Misra and Kant (2004) point out that the particular

focus of a given NGO must be taken into consideration; for example, the way in which an NGO

affects a given village will depend upon whether their orientation is more towards conservation

than economic development. The NGO variable as defined is, thus, more conservation oriented;

therefore, one might expect a negative relationship with gt according to Misra and Kant (2004).

However, it is possible that NGO could have a positive relationship on gt through improvements

in human and social capital. For these reasons, a negative relationship with PGR is posited.

The recent CFM legislation specifically intends to empower women within the political

economy of the VSS. In addition, the World Bank (2002) places much emphasis upon the critical



11 VSS-level data derived from the household interviews is weighted to take into account the differential probability
of selection of households for different VSS villages (i.e., large versus small villages in terms of N).





importance that women's representation has in the VSS. The extent to which women participate

in (and influence) the collective action decision-making is extremely important because women

likely know more about forest produce than men in many communities. The DWCRA group

index variable (DGI) is the number of women's self-help groups (DWCRA) divided by N, the

total number of households in the VSS. DGI represents the degree of self-reliance and self-

organization of women, and proxies for the general empowerment of women in a given VSS.

Empirical evidence from Misra and Kant (2004) indicate that there is a positive relationship

between women's participation and the economic output of JFM. Thus, the present study

anticipates a positive relationship between DGI and p, meaning a greater amount of self-

organized women have a beneficial influence on average consumption (i.e., mean per-capita

household consumption). In addition, a larger DGI is likely to be inversely related to poverty (as

measured by the PGR variable). A negative relationship with y is anticipated, suggesting that a

larger DGI results in lower levels of consumption inequality.

Economic Variables

Annual income is measured with three variables in this study. YFP is the weighted mean

household income derived from all forest products, YAL is the weighted mean household

income from agriculture and livestock, and YOS is the weighted mean household income derived

from all other sources. YOS is mainly comprised of income from employment but it also

includes (to a much lesser degree) income from household enterprises and remittances.

Considering these measures independently differentiates low-income households from those that

are relatively more affluent, and also categorizes income by its source; this is important because

evidence suggests that the poorest households in rural communities are the most reliant upon

forests for their livelihoods, and for supplementing their household consumption (Kumar, 2002).





As such, YFP is anticipated to be inversely related to p, while YAL and YOS are predicted to be

positively related to p.

Financial support for the APCFM project is derived from the World Bank, and funds are

distributed to the VSS by the APFD. IHW is an important variable because it proxies outside

investment to the community in the form of the household average of total payments to

individuals participating in VSS works projects, such as silvicultural treatments. IHW is

calculated as the man-days of labor employed for VSS works in the three previous years as

estimated by the VSS Chairman, divided by N (total number of households). This value is then

multiplied by the wage rate paid by the VSS to the laborers. IHW is expected to have a positive

effect on pt; a negative effect on both y and PGR is anticipated because the VSS works are more

likely conducted by the poorer households.

TRN is a count of the number of training events and field trips that VSS members (mainly

Management Committee members) have participated in through the auspices of the APFD. It is a

proxy for the amount of human capital investments made to a given VSS. This measure was also

estimated by the VSS Chairman. Anticipated relationships for TRN are a positive sign for the

average consumption and poverty-gap equations, and a negative sign for the consumption

inequality equation.

Bio-physical Variables

The length of time (LT) that a community has participated in JFM as a registered VSS is

extremely important because it is the key variable in terms of evaluating the success, or lack

thereof, of the economic impact of JFM. This is because the NTFP benefits of JFM are expected

to increase over time from an initial state of low returns characteristic of the degraded nature of

the landscape being protected; high values of LT theoretically imply income is being derived





from timber harvests. Therefore, in relative terms among the cross-section of VSS sampled, a

positive relationship between LT and pt (and/or a negative relationship between LT and PGR)

would likely indicate that JFM has had a beneficial economic impact on the communities

engaged in this program. Misra and Kant (2004) found empirical evidence to support such a

relationship. However, a negative sign on LT (vis-A-vis pt) could be indicative of weakened

social cohesion through time. Kumar (2002) relates that some foresters fear that interest in JFM

may wane when outside support inevitably declines, affecting project sustainability-though this

is unlikely to be important here, as funding under the APCFM project was still active at the time

of the survey. Community malaise may have more to do with local factors, poor or inappropriate

management plans, and/or poor engagement by the certain field officers of the forest department.

A negative sign on LT for y would support an interpretation that JFM has a beneficial impact on

distributional equity, again relative to the cross-section of VSS villages sampled.

FA is the size of VSS forest area under CFM protection. Forests are potentially an

important form of natural capital in rural areas, and Misra and Kant (2004) found that total forest

area has a direct and positive effect on the economic output of JFM. As such, FA is expected to

have a positive statistical influence on the economic measure of welfare in this study (i.e.,

average consumption, g), and the anticipated relationship between FA and inequality is negative.

RRE represents the relative forest resource endowment managed under CFM, for the VSS

sampled in the survey. RRE is calculated as the VSS forest area (in hectares) divided by the

estimated total number of VSS members for the corresponding VSS village. This variable can be

thought of as a population-normalized measure of the relative forest asset base of each of the

villages selected for the survey. This variable is utilized in the poverty-gap equation because

poorer VSS villages are more likely to have a greater dependence on their forest resources. Thus,





a negative relationship with PGR is expected, as larger proportionate areas under CFM

protection should mean that more forest resources are available to help mitigate poverty in these

communities. Damodaran and Engel (2003) state that "a minimum per capital land allotment of 1

to 2 hectares per VSS member can be considered a must for the success of JFM" (p. 30).

PFC is a proxy of the forest dependency of a community, as based on consumption at the

household level. It is calculated as the consumption of forest products as a percent of total

household consumption and is averaged (and weighted) over the households sampled in a given

VSS village. Reddy and Chakravarty (1999) provide data indicating forest dependency is greater

among poorer households. Thus, the following relationships are predicted for PFC: a negative

relationship in the average consumption and consumption inequality models, and a positive

relationship in the PGR model.

It is important to control for spatial differences in environmental and/or physical qualities

and processes that may impact the components of social welfare that communities possess. One

variable is included to represent a control of this type. DTW is the average depth to groundwater

(in feet) for a given VSS, as estimated by the VSS Chairman or the Vice-Chairman. As

agricultural production is a key component of local economies in rural India, this is an important

variable. Depths of several hundred feet are common. Thus, smaller DTW should equate with

higher gL and lower PGR. A positive relationship with y is likely as greater depths would tend to

preclude the poor from accessing the resource.

Institutional Variables

The institutional variables are based on the Design Principles established by Ostrom

(1990) as described in Chapter 1. These variables represent a subset of the unconventional

factors that may influence the welfare measures; Misra and Kant (2004) describe the use and





importance of such "non-neoclassical" factors, in addition to conventional economic factors, in

equations defining a joint production model of JFM.

VBA is a variable that ranks the relative "boundary awareness" of the VSS villages. It is

based on data gathered in the HH questionnaire; specifically, it is the percent of VSS households

(within a given VSS) that were recorded as being aware of the spatial boundaries of their VSS

forest area. This variable is used only in the poverty-gap equation, because the poorest

households tend to utilize forests more than households that are economically better-off. A

negative relationship is therefore anticipated, indicating that VSS villages with higher levels of

boundary awareness are likely to have a lower PGR value because the households of these VSS

theoretically have better control over their CPR.

The variable GAI measures the average general awareness (that households have) of four

quantifiable VSS institutions and parameters; two each, respectively. It is based on an additive

index (with a scale of 0 to 7) derived from scoring HH questionnaire data against objective data

that describe the VSS (e.g., hectares of forest managed). GAI measures the average index score

for each VSS village and, thus, has the same scale (i.e., 0 to 7). It is only included in the average

consumption equation because of the possibility that it may have a statistical influence on pt. If

so, a positive relationship is anticipated as higher levels of GAI are likely due to greater

engagement of VSS-member households in CFM activities. As such, GAI would act as a proxy

for level of participation, and theoretically would exert a positive influence on the average level

of consumption in a village.

CCA is a proxy for collective-choice arrangements and is measured in terms of whether or

not the micro-plan for each VSS is oriented towards the needs of forest-dependent households.

This variable is also based on data collected in the HH questionnaire. CCA is a ratio that





measures the percent of households indicating that the VSS micro-plan for their VSS village

reflects the interests of forest dependent households. Because conflicting interests are always

present in a community, and the APFD has a large hand in developing the VSS micro-plan, this

document may potentially favor rural elites, other interest groups, or perhaps even the APFD,

instead of the poorer households that are the focus of CFM. To the extent that higher values of

CCA signify greater adherence to the theoretical principle of responsive and representative

institutions, a positive relationship with tp is expected; conversely, a negative relationship is

anticipated with respect to both y and PGR.

Monitoring of the resource is one of the key design principles identified by Ostrom (1990).

The FP variable indicates whether or not the VSS has a formal patrol to monitor the VSS forest

area. This variable is defined as a 0/1 dummy, where FP takes the value of 1 if the VSS has a

regularly scheduled patrol, or employs a watchman to monitor the VSS forest area, and 0 if

neither. Negative relationships for FP are expected with both y and PGR.

GS represents graduated sanctions applied to locals who break the rules and regulations

established by the VSS. This is captured as a dummy; GS equals 1 if the VSS has a two-tiered

structure for punishing offenders, and 0 if not. The efficacy of GS as an explanatory variable is

likely dependent upon the presence of graduated sanctions being widely known among members

of the VSS. Although this is unlikely to be the case, the poorest households should be affected

the most; thus a negative relationship with PGR is expected if GS is statistically significant.

Description of the Data

Table 4-2 displays the mean and standard deviation of the variables that will be used in the

empirical analysis of the three economic measures examined in this study. These descriptive

statistics are presented for each of the three regional samples collected and are intended to





Table 4-2. Mean and standard deviation for VSS-level data: comparison by the region sampled.
Adilabad (n = 18) Visakhapatnam (n = 20) Chittoor (n = 20)


Variable Mean


Std. Dev.


Mean


Std. Dev.


Maximum Minimum


Dependent:

t1 9,604 422 8,390 938 5,708 324
In I 9.032 0.036 8.864 0.111 8.538 0.051

Y 0.2396 0.0137 0.2704 0.0411 0.2320 0.0601
Iny -1.457 0.059 -1.319 0.152 -1.495 0.273
PGR 0.0048 0.0023 0.0826 0.0702 0.0353 0.0319

Explanatory Demographic:
N 174 206 137 297 104 59
ST* 9 -- 14 -- 5 --
EDU 2.67 0.30 3.36 0.34 3.20 0.13
NGOt 0.333 0.485 1.0 0.0 0.800 0.410
DGI 5.1 2.5 5.4 2.4 4.6 2.5

Explanatory Economic:
YFP 300 119 1,869 366 764 373
YAL 8,290 2,750 2,621 544 9,044 1,487
YOS 20,848 2,568 19,690 3,828 12,714 745
IHW 8,662 7,618 6,024 6,397 3,922 5,657
TRN 4.9 1.7 4.5 1.9 5.8 2.6

Explanatory Bio-physical:
LT 8.0 1.7 7.2 2.1 7.4 2.1
FA 275 138 159 98 312 111
RRE 3.9 3.0 3.2 4.3 3.3 2.1
PFC 1.9 0.3 4.0 0.6 4.3 0.4
DTW 120 97 79 55 359 121

Explanatory Institutional:
VBA 0.176 0.125 0.279 0.169 0.234 0.105
GAI 2.12 0.72 2.53 0.83 2.24 0.45
CCA 0.663 0.131 0.658 0.140 0.640 0.164
FPt 0.444 0.511 0.400 0.501 0.300 0.470
GSt 0.722 0.461 0.700 0.470 0.450 0.510
* The value listed for the ST variable is the mode.
SDenotes a 0/1 dummy variable.





provide the reader with a broad comparison of the regions, which justify the inclusion of the

regional dummy variables. Weighted means and cluster-adjusted standard deviations are

calculated for those variables for which these measures are directly derived from the household-

level data (i.e., R, In g, EDU, YFP, YAL, YOS, PFC).12 In addition, the mode of the ST variable

is shown in place of a mean.

The remainder of this chapter presents and describes the data of some of these variables in

further detail. An additional six variables of interest to the synoptic description of the data are

also presented. These data are presented by region and variable type such that each of the three

regions of Andhra Pradesh is represented by two tables, with each region discussed individually.

The first table contains demographic and economic variables, while the second table contains

bio-physical and institutional variables.

The Adilabad Sample

Table 4-3 summarizes the data for the demographic and economic variables collected from

18 VSS villages sampled in the Nirmal and Jannaram Divisions of the Adilabad Circle. The

variables in this table are briefly described as follows. ID is the unique field identification

number given to each VSS village sampled. N is the total number of households in each VSS

village sampled. CST is the predominant caste group of a VSS village, as determined by the

mode of the households sampled. EDU is the average household education level, which is

measured in years and can range from 0 (no education) to 20 (a Ph.D.). ELC is the educational

level of the VSS Chairman. DGI is the average number of women's self-help groups across

households. MU is the weighted mean per-capita household consumption value of the VSS



12 For the region-level descriptive statistics in Table 4-2 that are derived from household data, it is necessary to
correct the standard deviations in order to account for the effects that the clustered sampling of households has on
the variance of data collected in this manner.





village. YFP is the weighted mean household income derived from forest products. YAL is the

weighted mean household income derived from agriculture and livestock. YOS is the weighted


mean household income derived from other sources (e.g., employment, household enterprises).


Table 4-3. Selected demographic and economic variables for the Adilabad sample.
ID VSS NAME N CST EDU ELC DGI MU YFP YAL YOS
101 Dounelli Thanda 77 ST 1.34 12 2.6 6,435 731 4,327 15,405
103 Chincholi (B) 641 BC 3.93 10 4.7 9,133 0 3,368 31,539
104 Dongir Gama 80 BC 2.36 10 6.3 9,708 0 9,101 13,616
105 Kalva 643 BC 2.56 4 2.3 8,786 5 5,840 17,719
106 Rampur 340 BC 3.15 12 6.8 12,232 0 5,532 27,165
107 Arepalli 168 ST/BC 2.09 0 6.5 10,365 885 6,164 17,585
108 Mamada Thanda 34 ST 1.22 10 8.8 7,776 1,270 16,461 16,250
109 Rachchakota 80 ST 1.62 0 1.3 9,034 1,674 6,229 11,571
110 Ankena 80 ST 2.07 15 7.5 11,579 1,508 11,149 12,196
111 Badhankurthi 475 BC 2.19 10 3.4 10,506 100 23,669 20,326
113 Dildarnagar 150 SC 2.65 12 5.3 8,495 247 4,274 12,264
114 Kothagudem 78 BC 1.94 5 1.3 10,326 690 3,231 23,988
115 Danthanpalle East 80 BC/OC 3.48 12 7.5 10,800 325 11,753 10,706
116 Janguguda 41 ST 0.91 10 7.3 6,271 991 6,987 4,108
117 Gandi Gopalpur East 34 ST 0.63 5 2.9 6,843 1,397 798 10,602
118 Islampur (K) 22 ST 0.65 0 9.1 6,033 4,317 5,403 9,122
119 Puttiguda (Kotha) 35 ST 1.49 10 5.7 7,583 573 2,735 5,540
120 Kancherabai 72 ST 1.27 7 2.8 8,523 1,799 5,354 10,271
Mean 174 2.67 8.0 5.1 9,604 300 8,290 20,848
Standard Deviation 206 0.30 4.6 2.5 422 119 2,750 2,568
Coeff. of Variation 1.18 0.11 0.58 0.50 0.04 0.40 0.33 0.12


The total number of households for each VSS village in the Adilabad sample (N) ranges

from a low of 22 in Islampur (K) to a high of 643 in Kalva. In general, the sample can be


characterized as being comprised mainly of very-small to small VSS villages: 75% of the sample


(12 VSS) is comprised of fewer than 80 households. The other six VSS, however, all have more


than 150 households.


The Adilabad sample can generally be described as being tribal as well. Fully half (nine


VSS) of the VSS villages listed in Table 4-3 are comprised of a majority of Scheduled Tribes

(ST) households. If a given VSS is listed as majority Scheduled Tribes, it will usually be the case





that 100% of the households are of the Scheduled Tribes designation. An exception is Arepalli

VSS, where the number of Scheduled Tribes and Backward Castes households is equal. The

other caste group that appears with some frequency (six VSS) as a majority is the Backward

Castes; it also shares the majority in two other VSS. The majority-Backward Castes VSS villages

appear to be associated with larger villages (i.e., greater values of N).

The educational variable EDU is a measure of the average number of school years

completed in each household that is then averaged over all of the households sampled in a given

VSS village. While the potential maximum for this variable is 20 years, the observed maximum

in this sample was less than 4 years. The lower education levels appear to be correlated with

Scheduled Tribes. In addition, the four lowest EDU values are among the five smallest VSS

villages in terms of N, while the five highest EDU values are in the three largest VSS (i.e.,

Kalva, Chincholi (B), and Rampur). In general, the educational level of the VSS Chairman is

high, with ELC equal to or greater than 10 years for eleven of the VSS. Three VSS have a

Chairman with no education.

MU (average per-capita consumption) ranges from a low of 6,033 Rupees (Rs.) in

Islampur (K), to a high of Rs. 12,232 in Rampur. This range is equivalent to roughly US$131 to

US$266 based on the September 19, 2006 exchange rate. The weighted mean is Rs. 9,604, with a

very low cluster-adjusted standard deviation (422) and coefficient of variation (4%). This means

that the variation of the value of per-capita consumption among households in the Adilabad

sample is low. The six lowest values of MU are all in VSS villages with a majority of Scheduled

Tribes.

The income variables are derived by averaging at the household level, which is why they

appear relatively large compared to the mean levels of consumption at the individual level (i.e.,





MU). As mentioned previously, income data in developing countries are subject to inaccuracy

due to various factors. The use of YFP, YAL, and YOS (measures of income from forest

products, agriculture, and other sources, respectively) must, therefore, be cautioned. The

weighted mean YFP is greater than Rs. 500 in 11 of the 18 VSS and reached a maximum of Rs.

4,317 in Islampur (K). Only three VSS report zero income from forest products. Income from

forest products (YFP) is associated with VSS villages that are generally both tribal and remote

(similar to the VSS villages with the lowest values of MU); and those VSS villages with little or

no YFP appear to be larger in size (i.e., larger N). The exception is Dongir Gama, a small VSS

village with commercially worthless trees. They were given poor quality land in the 1980s when

the government relocated their village due to the construction of a dam.

Comparison of the weighted income means also illustrates that agricultural income (YAL)

and income from other sources (YOS), which is mainly derived from employment, are the main

economic drivers of the VSS in the Adilabad sample. This actually applies to rural Andhra

Pradesh as a whole. Mean YAL equals Rs. 8,290; only one VSS village averaged less than Rs.

1,000 from agriculture and livestock (Gandi Gopalpur). YOS appears to be very important in

Adilabad: the mean is Rs. 20,848 and only three VSS have mean YOS values that are less than

Rs. 10,000.

Table 4-4 summarizes data on the bio-physical and institutional variables obtained from

the Adilabad sample. LT is the length of time that the VSS has been practicing JFM/CFM. LT

was calculated by subtracting the year of VSS establishment from the year 2005; as such, the

maximum possible age is 12 years, as field implementation of JFM in Andhra Pradesh began in

1993. FA is the total forest area (in hectares) that the VSS manages. Data on both LT and FA

were obtained from the VSS micro-plans. DTF is the walking distance (in minutes) to the VSS





forest area, averaged over the households sampled (within a given VSS). PQF and PMU are the


quality and main use, respectively, of the forest area before protection was established under the


VSS. The data for these two variables were derived from subjective ratings made by the VSS


Chairman or Vice-Chairman on Likerd scales. PFC is the percentage of household consumption


that is derived from forest products, averaged over the households sampled. EP is an index that


measures the equity in participation of the VSS; it is the percentage of VSS member households


that believe the process of selecting the VSS Management Committee was free and fair, as


opposed to being dominated by local elites or other groups. PG measures the perception of the


government, as based on the relationship between the VSS Management Committee and the


APFD, according to a Likerd scale measure made by the VSS Chairman or Vice-Chairman. The


FP variable indicates whether or not the VSS has a formal patrol to monitor the VSS forest area.


Table 4-4. Selected bio-physical and institutional variables for the Adilabad sample.
ID VSS NAME LT FA DTF PFC PQF PMU EP PG FP
101 Dounelli Thanda 6 275 40 4.4 3 4 0.875 2 1
103 Chincholi (B) 5 180 40 0.6 2 3 1.000 2 0
104 Dongir Gama 7 40 33 1.4 4 0 0.583 2 0
105 Kalva 11 621 33 2.0 3 2 0.778 1 0
106 Rampur 9 164 36 1.2 3 2 0.813 2 1
107 Arepalli 9 280 30 1.8 2 4 1.000 2 1
108 Mamada Thanda 8 343 30 3.9 2 4 0.882 1 0
109 Rachchakota 7 227 54 2.7 3 4 0.941 1 0
110 Ankena 6 150 23 2.0 1 4 1.000 1 0
111 Badhankurthi 9 154 60 1.2 3 1 0.750 1 0
113 Dildamagar 9 385 47 3.1 3 4 1.000 1 1
114 Kothagudem 7 150 20 2.9 3 3 0.625 2 0
115 Danthanpalle East 8 294 71 4.1 2 4 0.714 2 0
116 Janguguda 7 385 54 6.5 3 3 0.944 2 1
117 Gandi Gopalpur East 7 294 42 6.7 1 4 0.900 2 0
118 Islampur (K) 8 257 52 10.1 1 4 1.000 4 1
119 Puttiguda (Kotha) 9 380 39 4.5 3 4 0.900 4 1
120 Kancherabai 12 500 53 5.7 3 4 0.895 1 1
Mean 8.0 275 40 1.9 0.867 0.44
Standard Deviation 1.7 138 3.1 0.3 0.13 0.51
Coeff. of Variation 0.22 0.50 0.08 0.17 0.15 1.15
Mode 3 4 1.000 2
Median 3.0 4.0 0.898 2.0





As shown in Table 4-4, LT (length of time under JFM/CFM) ranges from 5 to 12 years for

the Adilabad sample; the mean is 8 years. The size of the VSS forest area (FA) ranges from 40 to

621 hectares, with a mean of 275 hectares. The minimum (40 hectares) is unusually low for the

Adilabad sample, as the next largest FA value is nearly four times as large. In addition, three-

quarters of the sample (12 VSS) is greater than 200 hectares.

The distance to the VSS forest area (DTF) ranges from 20 to 71 minutes, with a weighted

mean of 40 minutes. PFC is a proxy of the forest dependency of a community, based on

consumption at the household level. PFC indicates that the consumption of forest products as a

percentage of total household consumption ranges from 0.6% to 10.1%; the weighted mean is

1.9%. Comparison with the CST variable in Table 4-3 indicates that each of the six highest

values of PFC correspond to VSS that are wholly or predominately composed of Scheduled

Tribes households.

PQF represents a subjective quality rating of the forest area prior to protection by the VSS,

where 1 = good, 2 = slightly degraded, 3 = degraded, and 4 = very degraded. The mode and

median both equal "3" (degraded). This result is not surprising since the legal basis for JFM rests

upon rehabilitation of degraded land. However, some VSS clearly were given charge of non-

degraded areas in order to avoid potential future degradation, and to support the development of

communities such as Islampur (K), for example.

PMU identifies the main uses of the forest area prior to protection by the VSS, where 0 =

no previous use, 1 = fuelwood, 2 = construction material, 3 = agricultural implements, and 4 =

NTFPs. The majority of the Adilabad sample (11 of 18 total VSS) indicated that NTFP collection

was the main use of their forest prior to JFM/CFM.





EP is an index created to proxy for the equity in participation within a given VSS. The

potential range is 0 (no equity) to 1 (perfectly equitable participation), but the range for EP in the

sample is 0.583 to 1.000; only five VSS villages have EP values below 0.800. The magnitude of

the mean (0.867) reflects a high level of equity with regard to the internal political and

institutional process.

The perception of government (PG) was measured with an ordinal subjective ranking

where 1 = very good, 2 = good, 3 = fair, and 4 = bad. The vast majority of the sample (16 VSS)

indicated that the relationship with the APFD officer was either "good" or "very good". The

other two VSS, Islampur (K) and Puttiguda (Kotha), both indicated a "bad" relationship. The

VSS Chairman of Puttiguda described in detail the corruption of the local APFD Beat Officer:

this person took bribes from two other villages to allow them to cut wood in the VSS area. To

underscore this point, the VSS Chairman and his associate also intercepted some timber thieves

and confiscated their axe during the inspection of the VSS forest area by the principal

investigator.

FP is a 0/1 dummy variable that takes the value of 1 if the VSS either undertakes a formal

patrol of their VSS area, or employs a watchman to monitor it. Surprisingly, only eight VSS

villages (44% of the sample) have a formal monitoring structure in place. Many claim that they

informally monitor their VSS area because their farmland is adjacent to it, or the only access is

through their farmland or village.

The Visakhapatnam Sample

Table 4-5 summarizes the data for the demographic and economic variables from the

Visakhapatnam Division sample (20 VSS total). The total number of households for each VSS

village in the Visakhapatnam sample ranges from a low of 12 in Ch. Konda Veedhi to a high of

1,362 in Darlapudi. In general, the sample can be characterized as being comprised mainly of