University of Southern California Dissertations and Theses

Insurance mechanisms, forest clearance, and the effect of government policies in rural economies

(USC Thesis Other)

Insurance mechanisms, forest clearance, and the effect of government policies in rural economies

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INSURANCE MECHANISMS, FOREST CLEARANCE, AND THE EFFECT OF GOVERNMENT POLICIES IN RURAL ECONOMIES Copyright 2002 by Elbert Lance Howe A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements of the Degree DOCTOR OF PHILOSOPHY (ECONOMICS) December 2002 Elbert Lance Howe Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3093772 INFORMATION TO USERS The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. ® UMI UMI Microform 3093772 Copyright 2003 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 90089-1695 This dissertation, written by under the direction o f h / 3 dissertation committee, and approved by all its members, has been presented to and accepted by the Director of Graduate and Professional Programs, in partial fulfillment o f the requirements fo r the degree o f DOCTOR OF PHILOSOPHY Director Date 4 ,Lf ) ° ° Dissertation Committee Chair Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DEDICATION Whatever Good and Useful Ideas Flow Out of this Research, I dedicate to the Tsimane People. May Their Gracious Support of Years of Academic Research One Day Yield Abundant Gains in Quality of Life. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. iii ACKNOWLEDGEMENTS I thank my committee members, Professors Nugent, Pendleton, Day, Perrigne, and Jeong, for their valuable input throughout the research process. In particular, I thank Prof. Jeong for his valuable comments and suggested revisions for Chapter 1. I also thank fellow graduate students who participated in the development workshops for their helpful comments, and particularly, I thank Sripad Motiram and Mehdi Farsi. I also thank Ricardo Godoy, who provided the survey data which forms the basis for the empirical work in Chapters 1 and 2. The model presented in chapter 3 was inspired by Professor Samar Datta and I thank him along with Professors Nugent and Milindo Chakrabarti for allowing me to participate in the discussions that have led to its current form. I thank the USC Economics Department for funding throughout my USC graduate studies, the support staff, Young Miller, Sheila Williams, and Marie Reyes, the College of Letters, Arts, and Sciences for the Summer Dissertation Fellowship, and John and Alice Tyler for the Tyler Environmental Fellowship. I also thank my mentors at Central Michigan University and Liberty University, Michael Shields and Maurice Zaffke, respectively, for their significant input into my academic and personal development. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. iv I thank my wife Carina for her love and patience throughout this focused research process and my parents, Lynn and Peggy Howe, for their unwavering support and prayer through many years of education. I am especially grateful to Professor Linwood Pendleton. He worked with me as a new Ph.D. student, and guided me through the process of turning a research paper into polished, professional work. Chapter 2 is a modified version of the material that appears in Land Economics. The co-author listed in this publication directed and supervised research which forms the basis for Chapter 2. I extend my deep thanks to my chairperson, Dr. Nugent, from whom I have learned so much and to whom I will always be grateful. His breadth of knowledge and critical insight proved invaluable in forming the research questions, undertaking the investigation, and analyzing the results presented in this dissertation. Never reluctant to give of his time, nor heavy handed in his demands, he motivated through his own tireless work habits and optimistic and amiable nature. He is the graduate student’s ideal mentor. Finally, my deepest thanks I give to God, and thank Him for providing this wonderful opportunity and for sustaining me through to the end of this chapter in life. May I be a good steward of what He has blessed me with in the years to come. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V TABLE OF CONTENTS Dedication ii Acknowledgements iii List of Tables vii List of Figures viii Abstract ix Introduction 1 Chapter 1. Risk Coping Among Indigenous Smallholders in Sub-Tropical Bolivia 7 1.1 Introduction 7 1.1.1 Consumption Smoothing 7 1.1.2 Risk Coping Mechanisms 11 1.2 Data 18 1.3 Savings 27 1.4 The Labor Market 35 1.4.1 Agricultural Production in a Subsistence Economy 35 1.4.1.1 Household Constraints in the Dry and Rainy Season 36 1.4.1.2 Utility Maximization in the Dry and Rainy Season 38 1.4.2 Estimating Labor Market Participation 41 1.5 Informal Credit and Remittances 45 1.6 Discussion and Conclusions 48 Chapter 2. Market Integration, Development, and Smallholder Forest Clearance 51 2.1 Introduction 51 2.2 Background 53 2.3 Literature Review 56 2.4 The Model 63 2.5 Comparative Statics 71 2.5.1 Forest Clearance and Market Integration 75 2.5.2 Forest Clearance and Impatience 78 2.5.3 Forest Clearance, Technology, and Agricultural Productivity 79 2.6 Data and Analysis 81 2.7 Policy Implications and Discussion 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vi Chapter 3. A Co-Management Model of Timber Resources 89 3.1 Introduction 89 3.2 The Model 91 3.2.1 Background to the Model 91 3.2.2 The Three Sectors 97 3.2.2.1 The Forest Community 99 3.2.2.2 The Residual Sector 102 3.2.2.3 The Forest Department 102 3.2.2.4 Government Grants 104 3.2.2.5 Market Clearing Equations 105 3.2.3 Constrained Optimum by Sector 106 3.2.3.1 The Forest Community 106 3.2.3.2 The Residual Sector 108 3.2.3.3 The Forest Department 109 3.3 Simulations and Discussion 112 3.4 Conclusion 121 References 124 Appendices Appendix 1. Co-Management Simulations: Functional Forms 132 A. 1.1 Forest Community 132 A. 1.2 Forest Department 132 A. 1.3 Residual Sector 133 Appendix 2. Co-Management Simulations: GAUSS Code 134 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vii LIST OF TABLES Table 1.1 Summary Statistics: General 21 Table 1.2 Summary Statistics: Savings 24 Table 1.3 Summary Statistics: Labor Market Participation, Credit, and Remittances 25 Table 1.4 Summary Statistics: Health Shocks 26 Table 1.5 Generalized Least Squares Savings Estimates - Entire Sample 29 Table 1.6 Generalized Least Squares Savings Estimates - phwrs 30 Table 1.7 Generalized Least Squares Savings Estimates - dhwrs 32 Table 1.8 Labor Market Participation, 1995 (censored normal regressions with robust standard errors) 44 Table 1.9 Informal Loans and Remittances 48 Table 2.1 Overview of Selected Theoretical Models of Smallholder Forest Clearance 58 Table 2.2 Relationship Between Total Forest Clearance and Key Model Variables in Selected Theoretical Models of Smallholder Deforestation 75 Table 2.3 Regression Variables 82 Table 2.4 Summary Statistics 83 Table 2.5 Estimation Results 85 Table 3.1 Simulation Results 113 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. viii LIST OF FIGURES Figure 1.1 Map of Survey Area 20 Figure 2.1 Choosing the Optimal Level of Forest Clearance 74 Figure 2.2 Choosing the Optimal Level of Forest Clearance Effort 73 Figure 3.1 A Circular Flow Diagram 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ix ABSTRACT This dissertation treats three critical linkages, namely, those between (a) the objectives of national development and rural development, (b) the rural economy and rural households, and (c) rural households and their natural resource base. Both theoretical and empirical methods are applied in exploring these linkages in three interrelated stages. The first two sections focus on the micro-foundations of the rural household and the third extends these observations through a general equilibrium model which explores co-management of village resources. Taken together, the three approaches are intended to provide the reader unique insight into specific challenges to rural economic development, the existing research surrounding such issues, and a set of novel methodologies and results specific to each issue. Chapter one begins the analysis by exploring the type of mechanisms employed by rural households to cope with idiosyncratic shocks. Building on the consumption smoothing and risk coping literature, both analytical and empirical methods are used to explore the household’s use of savings, the labor market, and informal credit and remittances to cope with idiosyncratic health shocks. The second model and chapter, also employing analytical and empirical models, examines household forest clearance and its linkages to market integration. The theoretical model develops the traditional agricultural household model in a unique two season framework in which the household treats deforestation labor as an investment in future labor productivity. Both analytical and empirical results indicate that market integration may have a number of differing effects on forest clearance in the long Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. run. Also, patient households are found to invest more in forest clearance. The third and final stage in the analysis uses a general equilibrium framework to explore the co-management of environmental resources. Resource use under exclusive de jure state control (which is modeled as de facto open access) is contrasted with co management schemes under varying degrees of enforcement. Results support the notion that there are large differences between regimes with respect to biomass production and the negative externality effect associated with forest degradation. Furthermore, simulation results highlight the importance of a local communities’ dependence on the resource base in ascertaining benefits to the user group under different regimes. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 Introduction Promoting economic development among the world’s rural poor remains a policy focus for many countries and organizations. Both in terms of meeting basic human needs and exercising global environmental stewardship, the rural sector of developing countries remains a top candidate for innovative and successful economic development. Recent World Bank data indicate that of the 1.2 billion poor consuming less than a dollar-a-day, 75% live and work in rural areas (IFAD, 2001). In Bolivia, the country particular to the data used in this dissertation, the rural poverty headcount index is 81.7% (with a rural to urban poverty ratio of 2.42) and rural destitution rates are estimated to be 59.1% (IFAD, 2001; MECOVA, 2001).1 At the same time, the rural population’s close link to the environmental resource base is well documented; that is, households in the rural sector are largely biomass dependent (DasGupta, 1993). The current rural population comprises roughly 53% of the world’s population, and of the rural economically active population, 51% are employed in agriculture (FAOSTAT, 2002). In Bolivia, smallholder farmers comprise roughly 80% of the rural economically active population (ECLAC, 2000). Besides explicit agricultural production, rural households also rely heavily on the natural resource base for meeting subsistence needs. 1 While income measures o f poverty do not account for subsistence gathering and other non-monetary assets, a consistent account emerges when using alternate measures o f well being. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 Because of this tight linkage to local environmental resources, a strong correlation between rural poverty and environmental degradation is often observed. Poorer households rely more heavily on local resources, oftentimes leading to environmental degradation, which, in turn, results in greater levels of poverty and human suffering. Indeed, rural smallholders are often considered to account for some of the greatest levels of deforestation in the developing world.2 This combination of persistent rural poverty and environmental degradation, has presented a formidable challenge to development economists and policymakers alike. Nevertheless, as will be discussed in more detail throughout this dissertation, an exciting stream of new research is emerging which may inform national, regional, and local policies of rural economic development to the end that rural poverty and environmental degradation are greatly reduced. This dissertation, therefore, treats three linkages critical to understanding the relation between rural poverty and environmental degradation, namely, those between (a) the objectives of national development and rural development, (b) the rural economy and rural households, and (c) rural households and their natural resource base. Both theoretical and empirical methods are applied in exploring these linkages in three interrelated stages. The first two sections focus on the micro foundations of the rural household and the third extends these observations through a regional /macroeconomic model which explores co-management of village 2 According to Angelson (1995), estimates o f the share o f deforestation from shifting cultivators range from 45% to 60%. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. resources. Taken together, the three approaches are intended to provide the reader unique insight into specific challenges to comprehensive rural economic development, the existing research surrounding such issues, and a set of novel methodologies and results specific to each issue. Chapter one begins the analysis by exploring the type of mechanisms employed by rural households to cope with idiosyncratic shocks. Building on the consumption smoothing and risk coping literature, both analytical and empirical methods are used to explore the household’s use of savings, the labor market, and informal credit and remittances to cope with idiosyncratic health shocks. The results of Chapter 1 are consistent with existing literature, but are also unique. Similar to existing literature from very different contexts, this research also find that rural Bolivian household’s rely on savings, informal credit, and the labor market to cope with shocks. At the same time, the sum of the literature points to the great differences in operative risk coping mechanisms across regions. Uniquely, this research finds that even among a tight knit, ethnically homogenous group, households rely on very different risk coping mechanisms. Households closest to the market appear to rely primarily on the drawing down of animal assets and durable goods to cope with shocks, a mechanism most consistent with the permanent income model. In contrast, households in more remote regions appear to rely on informal credit and remittances and temporary labor market participation, evidence consistent with both risk sharing and permanent income models. We contend that these Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. findings highlight the importance of the particular institutional context in framing effective rural development policy. The second model and section examine household forest clearance using both theoretical and empirical models. While this model of forest clearance builds on existing literature and therefore in many ways is consistent with existing models, thereby affording meaningful comparisons, it is also unique in certain ways. A two- period, two-season model of the agricultural household is constructed in such a way that households decide whether or not to clear forest, and if so, to choose between primary and secondary forest on the basis o f (among other things) the off-farm wage rate, transport costs and technological factors. Similar to some studies, deforestation is modeled as an investment strategy; but unlike such studies, it is not represented as a strategy for establishing title to the land. The approach is also distinct in that the results of the analytic model are tested empirically. Both analytical and empirical results indicate that the integration of the Tsimane into the market economy may have a number of differing effects on forest clearance in the long run. If such integration into the market results in increased levels of education and increased off-farm wages, rates of forest clearance should decline. However, if greater integration leads to increases in the relative price of agriculture, or to improvements in agricultural or forest clearance technology, rates of forest clearance are likely to increase. Greater levels of impatience among households are also found to be associated with relatively lower levels of clearance in both the analytical and empirical results. More impatient households prefer to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 obtain the more immediate benefits of wage payments relative to the uncertain future return on agricultural investment and forest clearance. The third and final stage in the analysis uses a general equilibrium framework to explore the co-management of environmental resources. Resource use under exclusive de jure state control (which is modeled as de facto open access) is contrasted with a co-management regime (between a forest department and a forest community). Moreover, the model is simulated and results are discussed in light of the analytical findings of the model. The straightforward comparison between the benchmark case and co management with high enforcement supports the hypothesis that co-management produces greater levels of the positive externality (in this case forest biomass) and less of the negative externality. This follows from the fact that under co management, the Forest Community allocates less time to biomass gathering and more time to protective labor. Consequently, the urban or “Residual Sector” always prefers co-management to pure state control (or de facto open access) and, given co management, the “Forest Department” prefers co-management regimes with greater levels of enforcement. In contrast, for low levels of valuation of the forest good, the “Forest Community” prefers co-management and the accompanying wages for protective labor, but for communities that place a higher value on the forest, the “open access” regime is favored for obvious reasons. As co-management reaches a relatively lower level of enforcement, however, the utility of communities that highly value the forest is greater than or equal to the utility under the open access regime. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 In conclusion, the three parts of this analysis fit together and serve to explore the link of the rural household to their environment. It is hoped that the ideas and results expressed in this dissertation uniquely contribute to the field of economic development, in addition to stimulating further dialogue with respect to comprehensive policies of economic development that enhance both rural quality of life and environmental stewardship. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 Chapter 1 Risk Coping Among Indigenous Smallholders in sub-Tropical Bolivia 1.1 Introduction The ability of poor, rural households to maintain relatively constant consumption is a remarkable feature of the village economy. In spite of incomplete credit and insurance markets, and a limited array of coping and management mechanisms, several empirical studies have found that a majority of village households, while not completely insulated from idiosyncratic and aggregate shocks, maintain remarkably smooth consumption (Townsend, 1994; Ravallion and Chaudhuri, 1997; Morduch, 1990; Grimard, 1997; Deaton, 1992). Extending this work, other novel research is beginning to explore the specific exante and expost coping mechanisms in light of economic theory. 1.1.1 Consumption Smoothing The literature that deals specifically with household savings and consumption typically adopts a model consistent with either the permanent income hypothesis or the Arrow Debreu approach. While each model is unique, short-run predictions are similar. Both models predict that individual household consumption is independent of most idiosyncratic shocks over the short-run. In the former model, household consumption follows a lifetime or permanent income process, and complete credit and factor markets are generally assumed. This Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. allows the household to use credit or asset sales to smooth consumption over time. In a modified version of the model, households save out of unexpected transitory income for future income shocks rather than out of expected permanent income (Rosenzweig, 2001; Deaton and Paxson, 2000; Deaton, 1992; Paxson, 1992). Households in this framework are relatively immune to common village level shocks that do not affect the household’s expectation of permanent income; only negative idiosyncratic shocks strong enough to affect permanent income will lower individual consumption (Rosenzweig, 2001). That is, household consumption is a function of the individual income processes, not aggregate village income (Ligon, 1998). In contrast, risk is pooled across village members in Arrow Debreu type models, and households are primarily insured against idiosyncratic risk. Assuming perfect markets, households in the village are tightly linked through a full set of complete contingent markets (reciprocal credit and gift exchange are examples of informal mechanisms) to reduce the effects of consumption shocks. If full insurance indeed exists, individual income follows aggregate village consumption, and, consequently, individual consumption is independent of idiosyncratic shocks (Diamond, 1967; Wilson, 1968; Townsend, 1994).1 Diamond’s (1967) result, f/;(c,(0))h,(0) = XiP(8) => U',(c,(Q)) = X, can be rewritten as (Bardhan and Udry, 1999), U ' . j c X Q ) ) X, Ufa®)) Xj 1 Other important assumptions behind this result include village households sharing the same utility function, the same subjective probability o f the shock, and the same prices (Diamond, 1967). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 for j households; the marginal utility of household i relative to household j is equal to the weight of household i in the “Pareto program” relative to household j. That is, the ratio of marginal utilities should remain at a constant rate or consumption should co-move across village households. This result indicates that household consumption follows aggregate village income, consequently, the model predicts that household consumption is constant even when the household encounters • . . . 9 idiosyncratic shocks large enough to affect lifetime income. In contrast to the permanent income model, however, negative aggregate shocks large enough to affect average village income reduce individual consumption. A modified version of the full-insurance model replaces the perfect information assumption with imperfect information. The second best but non-Pareto efficient allocations predicted by this “private information” model appear to be more consistent with the empirical findings of strong but imperfect consumption smoothing in village economies (Ligon, 1998; Townsend, 1995b). Evidently, regardless the model selection, empirical evidence of the independence of individual consumption and idiosyncratic shocks (or individual •5 income) is consistent with both models in the short-run. Under the permanent income model, so long as the idiosyncratic shock does not affect expectations of permanent income, present consumption will not change. Similarly, present 2 In the Arrow Debreu model each household receives a set portion o f average village consumption based on the households weight in the Pareto program. Alderman and Paxson (1994) highlight this point by noting that in the pure version o f this model, households with individual incomes falling to zero continue to receive a constant portion o f average village consumption. 3 This point is a basic implication o f the model, and is articulated in Bardhan and Udry (1999, p. 104) and Alderman and Paxson (1994). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 consumption is shielded from idiosyncratic shocks through full insurance in the Arrow Debreu framework. Consequently, empirical evidence of the independence of individual consumption and individual income is consistent with both models. In the long-run, however, the empirical predictions of the pure versions of the models diverge (Ligon, 1998). The permanent income model predicts that expected changes in marginal utility of incomes are equal across village households each period (Ligon, 1998). In reality, however, realized incomes diverge from expected incomes increasing income inequality between households. The resulting income inequality is the source of the resulting consumption inequality over time (Deaton and Paxson, 1994; Deaton and Paxson, 2000; Ligon, 1998). The full insurance model, in contrast, predicts that actual marginal rates of substitution equal across time for all households in the village. Ligon (1998) rigorously explores the appropriateness of models of permanent income, full insurance, and imperfect information in explaining consumption smoothing within the village India context. Using the Indian ICRISAT data (which includes data on household consumption) and a technique to estimate the coefficient of relative risk aversion, he finds that in two of three villages used in the Townsend (1994) study, the best fit is provided by the private information model (the full insurance model with incomplete information), where expected marginal rates of substitution vary across households. The permanent income model is found to be the best fit for the third village. Villages appear to employ a combination of strategies consistent with both the permanent income and full insurance models. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 Correctly identifying the relevant risk coping strategies for a particular village or region is particularly important for policymakers. Effective policies aimed at reducing risk may take very different forms in communities with strong informal risk sharing arrangements compared to those where households are relatively independent and rely on asset or labor markets to smooth consumption. In either case, identifying the operative coping mechanisms allow policymakers to build on the existing institutional structure. With adequate consumption data, the process can be clarified using a test similar to Ligon (1998); few data sets, however, offer such detailed consumption data. In such cases, exploring the operative risk coping mechanisms may prove a useful alternative. 1.1.2 Risk Coping Mechanisms An evolving focus of the literature has been on identifying the sources and extent of consumption smoothing. And, as mentioned, ascertaining the coping mechanisms in a particular institutional setting may provide important clues as to the relevance of one model over another. Expost risk coping mechanisms most consistent with implications of the permanent income model include sale of assets, credit, and temporary changes in labor market participation. As mentioned, in the permanent income model household consumption is a function of the individual income processes and asset and credit markets are used to save and borrow in light of permanent income. In villages with relatively well-developed factor and credit markets one would expect to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 find households relying heavily on credit and sales of assets to cope expost with shocks. This seems more likely in villages that are relatively integrated to the market or within a reasonable distance from a market center. Consumption of animal and grain stocks can also account for asset draw-down, and do not require fully developed factor markets. Evidence supporting a strong version of this hypothesis includes changes in a household’s stocks large enough to completely cover the transitory idiosyncratic shock. Empirical evidence on asset draw-down is mixed. Rosenzweig and Wolpin (1993), using Indian ICRISAT data, find that asset sales are sensitive to negative income shocks. In contrast, in a West African context Fafchamps, Udry, and Czukas (1999) find that livestock sales are insensitive even to extreme idiosyncratic and aggregate shocks among farmers in Burkina Faso. Udry (1995) finds that rural Northern Nigerian households do not draw down livestock (which are subject to diminishing returns) but rather reduce grain stocks in response to production shocks. Nevertheless, the drawing down of grain stocks is not sufficient to cover the loss in income due to the shock. Goldstein (1999) finds evidence for risk sharing within groups among rural households in Southern Ghana. Fafchamps and Lund (1999) find that rural Filipino households do not use crop or livestock sales to cope with aggregate or idiosyncratic shocks but rely on mechanisms more consistent with risk sharing. In terms of temporary labor market participation, existing empirical studies have studied the link between household labor market participation and shocks. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 Using Indian ICRISAT data, Kochar (1999) finds that labor force participation increases for males in village India following negative idiosyncratic production shocks. Fafchamps and Lund (1999) fail to find evidence of labor market coping mechanisms among households in rural Philippines. Rose (2001), using a broader Indian data set (2115 hh) and a measure of aggregate shocks (relative to Kochar (1999)), finds that the likelihood of household participation in the labor market is higher with a negative shock, and separately, that households facing a riskier distribution of incomes due to weather shocks are more likely to participate in the labor market. Reciprocal credit and gift giving spread risk and serve as evidence more consistent with the full insurance model. Villages without formal insurance markets but possessing a tight network of information serve as a fertile environment in which risk sharing arrangements can arise. At the same time, however, risk must be primarily idiosyncratic implying that the village size must be great enough to support uncorrelated risk (Das Gupta, 1993). A wide variety of such mechanisms can emerge depending on the institutional context; examples range from reciprocal credit and gift exchange to households providing labor and productive assets to members of their risk sharing network. Those studies using hard-to-find contingent credit data, have found contingent credit to be an important mechanism for risk sharing in certain contexts. Udry (1994), using data from Northern Nigeria, finds that state contingent loan transactions allow households to cope with idiosyncratic risk (however, he rejects Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 Pareto efficient risk pooling). Fafchamps and Lund (1999) also find evidence for risk sharing within small groups of family and friends in rural Philippines. Negative idiosyncratic shocks are found to increase gifts and informal loans to the household, while shocks to individuals within the household’s risk sharing group reduces gifts and loans; a result consistent with the notion that individual consumption is a function of aggregate consumption. Kurosaki and Fafchamps (1999) find evidence for risk sharing in the Pakistan Punjab but complete insurance is rejected. Using data from Cote d’Ivoire, Grimard (1997) finds evidence for partial risk sharing by ethnic group. And using data from Southern Ghana, Goldstein (1999) finds that men and women utilize different risk sharing groups; the evidence indicates consumption is generally independent of income shocks. Finally, because a fully Pareto efficient allocation of risk is not achieved, household’s make use of risk management or risk spreading techniques at the household level to protect income. These exante mechanisms include such measures as crop diversification, permanent labor market participation, and production of crafts for the market.4 Furthermore, risk management options may actually be preferred over risk-sharing schemes if the efficiency loss associated with diversification is less than the expected loss due to moral hazard associated with the risk-sharing scheme (DasGupta, 1993). That is, risk management schemes favor subsistence crops over cash crops and labor diversification over pure household farm 4 This strategy, “income smoothing,” is nicely described in Morduch (1995), Bardhan and Udry (1999, ch.6), and DasGupta (1993, ch.8). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 labor, and, consequently, result in lower expected profits for the household but also a lower variance in income (Bardhan and Udry, 1999; DasGupta, 1993). The few studies that have explicitly explored risk management strategies in light of risk pooling have found that risk does indeed affect crop production choice. For instance, Morduch (1990) finds that households facing borrowing constraints are less likely to adopt more risky (but more profitable) agricultural techniques in order to smooth consumption. Rosenzweig and Bingswanger (1993) find that a lack of expost consumption smoothing instruments (such as credit) lead to a portfolio selection reflecting risk aversion. Using the same Bolivian data set employed in this paper, Godoy, Jacobson, and Wilkie (1998) find that relatively remote Tsimane households are more likely to cut old-growth forest following a death in the household or a crop loss. Finally, Kurosaki and Fafchamps (1999) conclude that, although significant risk sharing occurs within groups, the production choices of Pakistani Punjabi farmers are influenced by price risk. To the extent that the efficiency loss identified in these studies is less than the risk of default in reciprocal credit or gift giving schemes, households choose exante protection of risk (DasGupta, 1993). As indicated, the evidence of risk coping and management schemes is supportive of partial insurance and limited savings of transitory income in different settings. That is, household risk coping mechanisms appear to be highly context dependent. Because villages differ in terms of the types of productive activities, land rights, culture, information, trust and a vast array of other factors, the diversity of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 risk coping and risk sharing mechanisms naturally follows. Villages within the same region or ethnic group may rely on an entirely different portfolio of risk coping mechanisms based on the degree of market integration, the existence of formal village institutions and other village characteristics (Townsend, 1995b). As mentioned, some African households may never draw down productive assets to cope with even the most dramatic of shocks (Fafchamps, Udry, and Czukas, 1998) whereas rural Indian households may frequently trade bullocks to smooth consumption (Rosenzweig and Wolpin, 1993). Such distinctions may be unimportant in terms of identifying the specific coping mechanism used in a particular region (especially in the short-run), however, they may be very important in terms of policy-making prescriptions that are relevant to other contexts. Policies that inadvertently substitute for successful mechanisms can reduce net social welfare (Morduch, 1995); whereas, policies explicitly crafted to replace inefficient mechanisms and to complement the existing institutional context can strengthen successful mechanisms. Also, by recognizing how particular mechanisms are used in a diversity of contexts, policymakers can shape more appropriate policies which allow the rural poor to more effectively protect themselves from adverse shocks expost and to engage in more profitable production activities exante. As noted, the existing literature has provided rich insights into unique, context dependent, household risk coping mechanisms yet there remains, nevertheless, a need to bring fresh empirical results to bear on existing theory. As Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 Alderman and Paxson (1994) note in their survey article, “there is simply not enough literature from enough countries to draw general conclusions about the scope and importance of household risk sharing” (p. 71). A primary objective of this research, therefore, is to contribute to existing understanding by highlighting the bundle of coping mechanisms used in a unique context. In particular, this paper explores household reliance on the labor market, assets, and informal credit and remittances following idiosyncratic health shocks among the Tsimane Indians, an indigenous people group of central Bolivia, in light of the permanent income and full insurance models. Results indicate that, even among a tightly linked homogenous group, coping mechanisms vary greatly across regions, especially when factor, credit, and insurance markets are underdeveloped. Households in villages closest to trade routes and the market city appear to rely relatively more on the drawing down of assets, a result most consistent with the permanent income model. In contrast, more remote villages, but close to timbering operations, appear to rely relatively more on the labor market to manage risk, and, to some degree, on informal loans and remittances; evidence consistent with both the permanent income model and risk sharing models. Section 2 introduces the data, and sections 3 through 5 explore the use of assets, the labor market, and informal credit, respectively, among Tsimane households to cope with risk. The paper concludes in section 6. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 1.2. Data Data used for the empirical analysis is based on a household survey of the Tsimane Indians of Bolivia. Numbering around 5,000 individuals, the Tsimane live in a vast expanse of sub-tropical evergreen forests from the lowland eastern slopes of the Bolivian Andes to the center of the department of the Beni (northwest of San Borja). The Tsimane’s traditional homeland is the Chimanes Forest (1.2 million hectares), and within the Chimanes Forest, the majority of Tsimane villages are located in “Chimane Territory” (392,000 hectares), official Tsimane homeland legally established in 1990 (Lehm and Kudrenecky, 1995; Chicchon, 1995).5 Remaining Tsimane share “Multi-ethnic Indigenous Territory” (355,000 hectares within the Chimanes Forest) with the Mojeno, Yuracare, and Movima peoples.6 A small number of Tsimane (approximately 800) live in the Beni International Biosphere Reserve (135,000 hectares), the northeastern most region of Tsimane ethnic territory (Chicchon, 1995; Piland, 1991; Lehm and Kudrencky, 1995). Conducted from July to August 1996, the survey covers some 209 households in 19 villages and includes specific village and household level data. As indicated in Figure 1.1, the survey was conducted in villages of varying degrees of proximity to rivers, roads, other villages, and the largest market town of San Borja (pop. 13,000). Eight surveyed villages are south of San Borja in Chimane Territory. 5 The territory is recognized as common property and is “inalienable, indivisible, and cannot be lent or mortgaged” (Lehm and Kudrenecky, 1995, p. 80). 6 The Multi-ethnic Indigenous territory is part o f a “Forest Concession” area (579,000 hectares) made by the Bolivian Government with timber companies. Commercial forestry has been allowed on 422,000 hectares o f Chimanes Forest by non-indigenous timber companies since 1986 and will be banned in 2010 (Lehm and Kudrenecky, 1995). Indigenous groups are allowed to develop a commercial forestry industry. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 Seven villages are further south and east of San Borja, are close to or within Chimane Territory, and are adjacent to the area of Timber Concessions. Three villages are within Multi-ethnic Indigenous Territory and closest to the area of n Timber Concessions and the remaining settlement is on the outskirts of San Borja. Villages 15 and 2, indicated in Figure 1, are, respectively, 12.8 and 80 km from San Borja. To more clearly capture differences in risk coping based on location and levels of market integration, villages are sorted into three separate groups. Villages in group 1 are closer to the market of San Borja, or to the main commercial highway (which, nevertheless, is still dirt) from La Paz, relative to villages in groups 2 and 3.8 Typically, group 1 villages also lie along the Maniqui River, the largest river in the area, which provides an important trade outlet especially during the rainy season. Villages in group 2 are close to an area of commercial forestry, a long distance from San Borja, and are not close to a major river. Villages in group 3 are the most remote of all villages but are also close to timbering operations, and are situated close to large rivers. Household characteristics used in the empirical analysis are summarized in Table 1.1. Standard measures include average education of the household adult respondents, the number of years living in the village, the number of household 7 Based on the data, it appears that this is a settlement (Horeb) used primarily by the Tsimane Council (the official tribal representatives) and by Tsimane temporarily visiting San Borja (for medical help, etc.). Because this settlement is not a traditional Tsimane village it is excluded from the regression results. 8 Villages in group 1 include village 1, 15, 17, 14, 12, 11,9, 10, and 21; villages in group 2 include village 16, 19, and 13; villages in group 3 include village 5, 1 8 ,4 ,2 0 , 3, 6, and 2. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 Figure 1.1 Map of Survey Area J C U T I 0 I M Q rV ^ t) ■ 4- I , I, I . - ■ \ , . \ \ > Credit: Core Software Technology; ImageNet Service; www.imagenet.com members, and their average age. Significant differences can be observed between groups especially with regard to education. The data also provides a unique measure of time preference based on a data collection method common in social anthropology. Survey participants were given the choice of having one candy at the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. beginning of the survey or two pieces at the end of the survey. “Patient” individuals opted to wait until the end of the survey (patience = 0). Individuals with “average Table 1.1 Summary Statistics: General____________________________________ G roup 1 (100 hh; 8 vill.) G roup 2 (59 hh; 3 vill.) G roup 3 (41hh; 7 vill.) M ean . . . . . M ean . . . . . M ean . . . . . / j \ M m M ax / . j j \ M in M ax . ^ , , . M m M ax _____________ (std.dev)__________________ (std.dev)_________________ (std.dev)_______________ aveeduc horeb avgage num tot fem pop num kids yrsinvil depth 1 depth2 holdings dim pat dpat vpop 1.804 0 10 1.345 0 A 0.181 (1.973) (1.266) 4 (0.463) 0 .8 8 8 0 15 0.659 0 8.25 0.137 (2.061) (1.511) (0.501) 32.380 16 74.5 28.959 15 53.5 35.081 (12.192) (9.104) (11.227) 5.480 1 16 5.186 13 6.780 (2.699) 1 (2.232) Z (3.461) 1 .2 1 0 0 o 1.068 1 9 1.512 (0.880) O (0.254) 1 z (0.978) 3.110 0 1 1 3.000 0 1 1 3.585 (2.305) 1 1 (2.093) 1 1 (2.775) 24.718 0 78 14.526 0.5 43 10.461 (18.204) (11.073) (14.543) 0.090 0 1 0.034 0 1 0.024 (0.288) 1 (0.183) 1 (0.156) 0.810 0 1 0.831 0 1 0.902 (0.394) 1 (0.378) 1 (0.300) 226.550 3 4000 141.271 800 72.444 (493.415) (155.673) Z (50.245) 0 .2 1 0 0 1 0.136 0 1 0.098 (0.409) 1 (0.345) 1 (0.300) 0.850 0 | 0.915 0 1 0.780 (0.359) 1 (0.281) I (0.419) 198.400 30 300 277.034 50 350 88.390 (107.850) (88.551) (67.182) 0 0 20 2 0 0 2 2.25 65 15 6 1 1 0.08 60 0 1 0 3 0 0 25 200 1 1 200 impatience,” initially opting for candy immediately, agreed to wait until the end of the interview when offered more candy (patience = 1). “Impatient” individuals always opted for candy delivered immediately (patience = 2). Dummies were used to identify very patient and very impatient households (dpat and dimpat) and are equal to 1 if 50% or more of adult respondents are patient (impatient) and 0 9 Sugar is known to be a valuable commodity to the Tsimane (most adults lack front teeth because o f a love for sugar cane that begins at an early age) and in the majority o f villages, processed sugar is a rare commodity (this is especially true in relatively remote villages). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 otherwise. “Holdings” is the total amount of fallowed secondary forest controlled by the household. Depth 1 is a dummy that is equal to one if one of the parents of the household heads live in the same house, and depth2 is a dummy indicating if grandchildren live with the household heads. Dummies are also used to indicate households in the top and bottom 10% of group wealth for each year. Village leaders were asked price levels for eight “durable” commodities and five animal assets for the current year (1996) and the prevailing prices in 1994 and 1995.1 0 As prices were not available in every village for each year, an average price level is constructed for each year based on the average group price. It is assumed that when a price is not available in the village trade can easily occur with neighboring villages. Households can also trade directly with river traders or in the market center of San Borja. To capture the effect of missing village prices, however, a variable is created to indicate the percentage of complete prices in the village for a particular year (the number of prices available in the village out of the total number of commodities). Using price information, the values of total assets in 1994, 1995, and 1996 are calculated and presented in Table 1.2. Values of asset stocks are highest among households in group 1. However, savings as a percentage of total wealth is highest among group 2 (35% in 1995 and 29% in 1996) and lowest among group 3 households (roughly 20% in both years). The value of animal stocks is relatively lower than holdings of durable goods (the difference between total wealth and animal wealth). Wealth is lowest for groups 2 1 0 “Durable” commodities include canoes, pistols, shotguns, short-wave radios, mosquito nets, machetes, bows, and drift nets. Animal assets include cattle, ducks, hens, pigs, and dogs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 and 3, the most remote of all groups. As indicated by the values for andm and dgdm villages in group 1 have the most complete price information (markets) for durable type goods, group 2 the most information for animal assets, and prices for most goods are missing in group 3 villages. Information on labor market participation and informal credit and remittances is presented in Table 1.3. As indicated, labor market participation among survey respondents is widespread. Roughly 63% of all Tsimane households reported labor income in 1995. O f those who participated in the labor market, 71% worked outside of the village and 49% worked inside the village (19% worked both inside and outside of the village). Employment in the timber industry accounted for nearly 65% of outside the village labor earnings, fewer than 25% of those working outside of the village worked principally for commercial ranchers or farmers. Between regions labor market participation appears to be highest in group 2 and lowest in group 3; the average wage is highest in group 2 (situated in closest proximity to the timbering operations). Households receiving loans constitute a minority of Tsimane households. Only 23% of all households received loans or remittances during 1995 (7.7% of households received remittances). The majority of loans were informal, as indicated by the fact that of the 36 households who received loans, 86% received loans from relatives. Only 7% of households in region 3 received loans or remittances while nearly 36% of households in group 2 received loans. Average loan values are highest for group 1. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 Table 1.2. Summary Statistics: Savings G roup 1 (100 hh; 8 vill.) M in M ax Mean (std.dev) G roup 2 (59 hh; 3 vill.) M ean (std.dev) M in Max G roup 3 (41 hh; 7 vill.) M ean (std.dev) M in Max with 1642.063 0 6763 1045.237 24.5 5188 706.887 29 2526 Tf as (1389.733) (1002.748) (639.956) O s w * anwlth 513.159 (783.175) 0 4940 281.305 (415.326) 0 1750 212.133 (377.775) 0 1767 with 2203.183 (1796.597) 91 9174 1645.258 (1309.505) 72.4 6553 899.672 (895.980) 57.1 5205 anwlth 741.088 (922.657) 0 4644 425.085 (473.487) 0 2550 277.920 (576.960) 0 3724 svwl 561.121 (1232.662) -4377 5779 600.021 (904.737) -1023 4660 192.785 (660.310) -1251 3412 1995 svanwl 227.929 (827.670) -2390 3150 143.780 (504.558) -1300 1975 65.786 (533.631) -1767 2572 svdgwl 333.191 (631.002) -1987 2690 456.241 (690.084) -870 2685 126.998 (349.188) -635.1 1134 andm 0.642 (0.267) 0 .2 1 0.976 (0.065) 0 .8 1 0.380 (0.147) 0 0 .6 dgdm 0.741 (0.238) 0.375 1 0.614 (0.294) 0.375 1 0.396 (0.128) 0.125 0.625 with anwlth svwl 'O o \ c\ 3033.737 (2625.477) 1138.685 (1901.388) 830.554 (1838.210) 397.597 (1363.065) , , 432.957 svdgw | ( ] 0 2 2 926) 0.798 (0.162) , , 0.828 dg dm (Q.1 9 4 ) andm 1 1 0 12575 0 9440 -2812 7945 -2270 6745 -2782 3587 0.4 1 0.375 1 2319.735 (1471.914) 481.763 (700.633) 674.477 (1324.911) 56.678 (728.518) 617.799 (1125.587) 0.976 (0.065) 0.614 (0.294) 144 7531 1127.301 (772.946) 115 2554 0 3180 179.088 (197.127) 5 907 -4498 3190 227.630 (746.268) -2936 1844 -2055 3030 -98.832 (561.473) -3325 742 -3493 3135 326.461 (426.015) -436.0 1337 0 .8 1 0.449 (0.125) 0 .2 0 .6 0.375 1 0.515 (0.226) 0.25 0.75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 Table 1.3. Summary Statistics: Labor Market Participation, Credit, and Remittances G roup 1 (100 hh; 8 vill.) G roup 2 (59 hh; 3 vill.) G roup 3 (41hh; 7 vill.) M ean (std.dev) Min M ax M ean (std.dev) Min Max Mean (std.dev) Min Max dslabor 0.580 0 1 0.729 0 1 0.610 0 1 (0.496) 1 (0.448) I (0.494) 1 outw rkd 23.745 (54.549) 0 298 41.000 (79.394) 0 360 37.622 (49.255) 0 168 iwrkd 27.185 (75.535) 0 360 15.941 (40.440) 0 2 0 0 2.927 (18.741) 0 1 2 0 clear 15.777 (15.790) 0 1 1 0 14.110 (9.677) 0 60 6.585 (5.920) 0 30 new outw g 22.392 (14.754) 14.4 75.7 48.424 (27.504) 12.5 70.9 22.345 (8.625) 7.1 36.1 new inw g 19.924 (4.232) 10.4 25.7 21.433 (10.341) 8.3 37.0 16.638 (0.030) 16.6 16.7 drem itcr 0 .2 1 0 0 1 0.356 0 1 0.073 0 1 (0.409) (0.483) 1 (0.264) 1 rloans 95.210 (570.901) 0 4802 34.331 (106.195) 0 600 1 .2 2 0 (5.566) 0 30 nrloans 7.350 (53.675) 0 500 0.847 (6.509) 0 50 0 .0 0 0 (0 .0 0 0 ) 0 0 caploans 5.150 (50.007) 0 500 25.610 (101.910) 0 505.5 0.024 (0.156) 0 1 Finally, information on shock terms is presented in Table 1.4. The shock terms used in the savings regressions (phwrs and dhwrs) are based on a qualitative question. Survey respondents were asked if their health in 96 was good, ok, poor and a separate question asked if health this year relative to 1995 was better, worse or the same. A variable (phwrs) was created to indicate the percentage of household adults that had worse health in 1996 and in 1995. A separate dummy was also created and set equal to 1 for phwrs > 0. Dummies in 1996 and 1995 equal to one if 50% or more of household respondents had good health (dhagd) or bad health (dhabd) were also generated. The number of days sick during the forest clearance season (schaco) in 1994 was used as the shock term for labor market participation. And, the shock term used for the informal credit and remittances received in 1995 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 was the number of days the household was sick during the forest clearance season in 1995. As is evident in Table 1.4, households in group 3 appeared to have the worse health in terms of the qualitative question (dhabd) and the shock terms (phwrs, dhwrs) but on average missed fewer days of forest clearance due to sickness. A majority of respondents from groups 1 and 2 considered themselves to be in generally good health. Table 1.4. Summary Statistics: Health Shocks G roup 1 (100 hh; 8 vill.) M ean . . . , , . ^ . M in Max (std.dev) G roup 2 (59 hh; 3 vill.) M ean , . , , . Min Max (std.dev) G roup 3 (41 hh; 7 vill.) M ean . . . . . . . M in Max (std.dev) 1994 14.390 schaco (2 7 8 9 3 ) 0 180 14.831 (23.341) 0 150 9.293 (17.572) 0 90 phw rs 1 32.917 0 1 0 0 14.407 0 1 0 0 37.398 0 1 0 0 (35.541) (27.018) (33.702) dhwrs 0.540 0 0.271 0 0.683 0 (0.501) 1 (0.448) 1 (0.471) 1 < r > 15.845 150 13.966 1 0 .2 2 0 C\ O S schaco (25.003) 0 (20.470) 0 1 2 0 (17.186) 0 90 dhagd 0.740 0 0.814 0 0.561 0 (0.441) 1 (0.393) 1 (0.502) 1 dhabd 0.270 0 I 0.305 0 I 0.659 0 | (0.446) (0.464) (0.480) phwrs 1 13.250 0 1 0 0 16.384 0 1 0 0 40.854 0 1 0 0 (26.060) (30.401) (37.119) dhwrs 0.240 0 1 0.254 0 1 0.610 0 1 so (0.429) 1 (0.439) 1 (0.494) 1 os O S schaco - - - - - - - - - dhagd 0.720 0 0.763 0 0.463 0 (0.451) 1 (0.429) 1 (0.505) 1 dhabd 0.300 0 1 0.288 0 1 0.585 0 1 (0.461) (0.457) (0.499) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 1.3 Savings As discussed, the drawing-down of assets is consistent with a permanent income type approach to risk coping. Savings at time t can be defined as the difference in assets from the previous period, sn = Ajt - Al(l 1 } . Consistent with an interpretation of the permanent income model, savings is also equal to transitory income, sjt = Y ‘ t , where Yjt = Y‘‘ + Y] . Households experiencing negative shocks draw-down assets in order to supplement current consumption. A standard econometric specification has been to assume that permanent income is approximated by a vector of household characteristics, and that the idiosyncratic shock is a proxy for transitory income. A time dummy is used to account for village level shocks. The equation can be written as, sjt = a 0 + a 1 r|„ + a2H j + a 3 d, +s„ (1.1) where r|„ is the idiosyncratic shock, HI is a vector of time invariant household characteristics, and df is a time dummy. If households use changes in assets to cope with negative idiosyncratic shocks ai should be negative and equal to the amount of the shock, whereas the proxy for permanent income, c x , 2, should not influence savings. To identify savings and fixed effects at the group level a slightly modified version of equation (1.1) is used in the estimation. In particular, the group dummies are interacted with the shock variable and with household level characteristics. The modified version of (1.1) can be written as, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 s„ = a 0 + a,G / + a2 r\H + a 3 r| itGj + a 4H j +asH jGj + a 6 t/, + a1dlGj + zjl (1.2) where GJ , j = 1,2,3 is the group dummy. The results of equation (1.1) presented in Table 1.5 are based on feasible generalized least squares estimates. The first column represents overall savings and the second and third represents savings in terms of durable goods and animal assets, respectively.1 1 Across all Tsimane, the null hypothesis that savings is not affected by the health shock can be rejected with nearly 99% degree of confidence. A one percentage increase in the number of adult household respondents receiving a health shock reduces savings by 4.68 Bolivianos; if the shock is specified as the dummy in Table 1.5, when one or more adult respondents experience worse health, households draw-down savings by roughly 277 Bolivianos, or roughly 12.6% of average household wealth (2,192 B). Using group interactions, this hypothesis is tested on the group level. Feasible generalized least squares estimates of equation (1.2) are given in Table 1.6. As expected, only in villages along the Maniqui river and closest to the town of San Borja (group 1) is the null hypothesis rejected, and the effect is much stronger relative to the entire group case (a draw-down of assets of 6.52 Bolivianos per one- percent increase in worse health or 311 Bolivianos (14.1% of wealth) with the discrete measure). At the same time, the coefficients constructed to capture the degree of market completeness in the village (andm and dgdm) are not significant. 1 1 While one measure is redundant, both are provided for convenience. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 Table 1.5 Generalized Least Squares Savings Estimates - Entire Sample*_________ Entire Sam ple (p h w rsl) Entire Sam ple (dhwrs) svwl svdgwl svanwl svwl svdgwl svanwl shock -4.682** -2.244* -2.438* -276.655** -152.040* -124.616 -(2.43) -d -7 9 ) -(1.76) -(2.09) -(1.77) -(1.31) tim edm y 145.049 133.825* 11.224 136.017 126.945 9.072 (1.19) (1.70) (0.13) ( 1. 1 1 ) (1.60) (0 . 1 0 ) andm l -0.578 -0.257 -0.321 -0.552 -0.311 -0.241 -(0 .2 2 ) -(0.15) -(0.17) -(0 .2 1 ) -(0.18) -(0.13) dgdm l 2.472 2.723* -0.251 2 .6 8 6 2.807* -0 .1 2 1 (0.99) (1.67) -(0.14) (1.07) (1.73) -(0.07) dhabd 260.385 145.673 114.712 280.448* 159.714 120.734 (1.71) (1.47) (1.05) (1.82) (1.60) (1.08) dhagd 207.719 218.596** -10.877 218.005 226.929** -8.924 (1.34) (2.17) -(0 . 1 0 ) (1.40) (2.24) -(0.08) holdings -0.186 -0.245 0.059 -0.189 -0.245 0.056 -(0.45) -(0.92) (0 .2 0 ) -(0.46) -(0.92) (0.19) holding 2 0 .0 0 0 0 .0 0 0 0 .0 0 0 0 .0 0 0 0 .0 0 0 0 .0 0 0 (0.45) (0.96) -(0.24) (0.39) (0.92) -(0.28) aveeduc 265.403*** 97.476* 167.927*** 264.298*** 97.008* 167.289*** (3.26) (1.84) (2 .8 6 ) (3.24) (1.83) (2,84) aveeduc2 -33.935*** -16.649** -17.286* -33.847*** -16.594** -17.253* -(2.64) -(1.99) -( 1 .8 6 ) -(2.63) -(1.99) - ( 1.8 6 ) yrsinvil 10.597*** 3.555 7.042** 10.920*** 3.766 7.155** (2.52) (1.30) (2.32) (2.58) (1.37) (2.35) depth 2 -426.632*** -217.615** -209.017* -424.686*** -218.783** -205.904* -(2.52) -(1.98) -(1.71) -(2.50) -(1.98) - ( 1.6 8 ) pdtopyrg 1622.000*** 548.224*** 1073.776*** 1634.047*** 557.554*** 1076.493*** (8.09) (4.21) (7.43) (8 .1 0 ) (4.26) (7.41) pdlow yrg -865.972*** -656.337*** -209.635 -883.840*** -665.029*** -218.812 -(4.42) -(5.15) -(1.48) -(4.50) -(5.22) -(1.55) dim pat 129.215 -23.576 152.791 115.161 -30.128 145.288 (0.75) -(0 .2 1 ) (1.23) (0.67) -(0.27) (1.17) dpat -323.930* -164.658 -159.272 -336.567* -174.516 -162.051 -(1.73) -(1.35) -(1.18) -(1.78) -(1.43) -(1.19) _cons 285.990 141.279 144.711 277.653 154.116 123.536 (0.76) (0.58) (0.53) (0.73) (0.63) (0.45) 1 1 -3235 -3070 -3110 -3236 -3071 -3111 wald C hi2 128.9*** 69.3*** 917 * * * 126.8*** 5 9 2 *** 90.0*** * 191hh, 2 years (1995, 1996) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.6 Generalized Least Squares Savings Estimates - phwrs* G roup 1 (100 hh; 8 vill.) svwl svdgw l svanwl G roup 2 (59 hh; 3 vill.) svwl svdgw l svanwl G roup 3 (41 hh; 7 vill.) svwl svdgw l svanwl g 2 1 g l 1 gl 1882.524 1999.881 -117.357 -1882.524 -1999.881 117.357 461.004 966.932 -505.927 (0.87) (1.43) -(0.08) -(0.87) -(1.43) (0.08) (0.36) (1.18) -(0.56) g3 I g3 | g2 -461.004 -966.932 505.927 -2343.528 -2966.813** 623.284 2343.528 2966.813** -623.284 -(0.36) -(1.18) (0.56) -(0.98) -(1.93) (0.37) (0.98) (1.93) -(0.37) phw rs 1 -6.518** -3.380** -3.138* -0.711 0.032 -0.742 -2.802 -0.252 -2.550 -(2.39) -(1.93) -(1.63) -(0.17) (0 .0 1 ) -(0.25) -(0.71) -(0 . 1 0 ) -(0.92) tim edm y 274.096 137.534 136.563 78.136 173.224 -95.088 -47.923 104.085 -152.008 (1.52) (1.19) (1.08) (0.38) (1.30) -(0.65) -(0.15) (0.52) -(0.70) andm l -3.567 -5.496 1.929 -18.418 -23.418* 5.001 -1.271 1.318 -2.589 -(0.61) -(1.46) (0.47) -(0.94) -(1.87) (0.36) -(0.09) (0.15) -(0.26) dgdm l 0.530 3.536 -3.007 5.402 7 736*** -2.335 1.576 4.278 -2.703 (0.09) (0.95) -(0.73) (1.32) (2.95) -(0.81) (0.18) (0.77) -(0.44) dhabd 254.243 86.166 168.077 230.308 86.182 144.126 102.399 103.660 -1.261 ( 1 .2 0 ) (0.63) ( 1. 1 2 ) (0.73) (0.43) (0.65) (0.28) (0.44) -(0 .0 1 ) dhagd 289.860 145.588 144.273 -110.474 21.555 -132.029 76.426 150.768 -74.342 (1.35) (1.06) (0.96) -(0.32) (0 . 1 0 ) -(0.54) (0 .2 1 ) (0.64) -(0.29) holdings -0.557 -0.386 -0.171 -1.372 -1.090 -0.282 0.123 2 .2 0 0 -2.077 -(1.17) -(1.27) -(0.51) -(0.51) -(0.63) -(0.15) (0 .0 1 ) (0.34) -(0.29) holding2 0 .0 0 0 0 .0 0 0 0 .0 0 0 0.003 0 .0 0 2 0 .0 0 0 0 .0 0 0 -0.009 0.009 (1.13) (1.40) (0.34) (0.74) (0.99) (0.15) (0 .0 0 ) -(0.28) (0.25) aveeduc 225.667** 133.282* 92.385 161.457 72.874 88.583 537.398 334.367 203.032 (2.05) (1.89) (1.19) (0.55) (0.39) (0.43) (0.44) (0.43) (0.24) aveeduc2 -30.391** -19.969** -10.422 -13.704 0.757 -14.461 -303.700 -214.075 -89.624 -(1.98) -(2.03) -(0.97) -(0.17) (0 .0 1 ) -(0.25) -(0.43) -(0.48) -(0.18) yrsinvil 12.551** 5.180 7.372* 3.816 8.005 -4.189 0 .0 2 2 -1.082 1.104 (2.28) (1.47) (1.90) (0.36) (1.18) -(0.56) (0 .0 0 ) -(0.16) (0.15) * Group interactions terms for each variable are not included in table, however group dummies are (g 1 | g2 | g3) — 51 estimated coefficients, 191 hh, 2 years (1995, 1996). O J o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.6 (cont.) Generalized Least Squares Savings Estimates - phwrs* G roup 1 (100 hh; 8 vill.) G roup 2 (59 hh; 3 vill.) G roup 3 (41hh; 7 vill.) svwl svdgwl svanwl svwl svdgwl svanwl svwl svdgw l svanwl depth 2 -533.965** -259.11* -274.858* -343.526 -109.592 -233.934 116.717 211.625 -94.908 -(2.33) -(1.76) -(1.70) -(1.09) -(0.55) -(1.06) (0.19) (0.53) -(0 .2 2 ) pdtopyrg 2296.704*** 326.504* 1970.199 1748.602*** 1080.290*** 668.312*** 520.832 397.226 123.606 (7.60) (1.69) (9.27) (4.61) (4.44) (2.50) (1.06) (1.27) (0.36) pdlow yrg -889.104*** -641.05* -248.055 -1420.88*** - 1129.11*** -291.771 -183.549 -180.969 -2.580 -(3.21) -(3.61) -(1.27) -(3.84) -(4.77) - ( 1. 1 2 ) -(0.40) -(0.62) -(0 .0 1 ) dim pat 9.199 14.316 -5.117 77.384 -27.357 104.740 61.347 5.057 56.289 (0.04) (0 . 1 0 ) -(0.03) (0 .2 1 ) -(0 . 1 2 ) (0.41) (0 . 1 2 ) (0 .0 2 ) (0.15) dpat -333.079 -29.636 -303.442* -391.221 -415.387 24.166 -86.776 -51.790 -34.986 -(1.29) -(0.18) -(1.67) -(0.91) -(1.51) (0.08) -(0.19) -(0.18) -(0 .1 1 ) _cons 579.143 362.985 216.158 2461.667 2362.866* 98.801 118.139 -603.947 722.086 (1.06) (1.04) (0.56) (1.17) (1.75) (0.07) (0 .1 0 ) -(0.81) (0 .8 8 ) 1 1 -3221 3051 -3087.6 -3222 -3052 -3089 -3220 -3051 -3088 w ald C hi2 167.1*** 116.6*** 151.9*** 167.1*** 116.6*** 151.9*** 167.1*** 116.6*** 151.9*** * Group interactions terms for each variable are not included in table, however group dummies are (g l | g2 | g3) - 51 estimated coefficients, 191 hh, 2 years (1995, 1996). Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.7 Generalized Least Squares Savings Estimates - dhwrs*___________________________________________ G roup I (100 hh; 8 vill.) G roup 2 (59 hh; 3 vill.) G roup 3 (41hh; 7 vill.) svwl svdgw l svanwl svwl svdgwl svanwl svwl svdgwl svanwl g 2 I g l 1 g l 1953.45 2052.21 -98.77 -1953.5 -2052.2 98.77 440.14 982.97 -542.82 (0.89) (1.47) -(0.06) -(0.89) -(1.47) (0.06) (0.34) (1.18) -(0.59) g3 1 g3 | g2 -440.14 -982.97 542.82 -2393.6 -3035.2 641.59 2393.59 3035.18** -641.59 -(0.34) -(1.18) (0.59) -(0.99) -(1.96) (0.38) (0.99) (1.96) -(0.38) dhwrs -311.05* -168.42 -142.62 -174.92 -154.55 -20.37 - 1 1 1 .2 0 32.81 -144.00 -(1.67) -(1.42) -(1.09) -(0.65) -(0.90) -(0 . 1 1 ) -(0.37) (0.17) -(0 .6 8 ) tim edm y 310.70* 154.32 156.38 75.21 172.84 -97.63 -61.68 108.01 -169.69 (1.71) (1.33) (1.23) (0.36) (1.30) -(0.67) -(0 .2 0 ) (0.54) -(0.77) andm l -4.35 -5.89 1.53 -17.86 -22.82 4.96 -1.13 1.39 -2.52 -(0.74) -(1.57) (0.37) -(0.91) -(1.82) (0.36) -(0.08) (0.15) -(0.25) dgdm l 1.32 3.93 -2.61 5.46 7.77 -2.31 1.72 4.38 -2 .6 6 (0.23) (1.05) -(0.64) (1.33) (2.96) -(0.80) (0 .2 0 ) (0.79) -(0.43) dhabd 262.85 92.62 170.23 241.76 100.98 140.78 120.97 100.14 20.83 ( 1.2 1 ) (0.67) ( 1. 1 1 ) (0.76) (0.50) (0.63) (0.33) (0.43) (0.08) dhagd 260.10 132.44 127.66 -119.54 6.99 -126.53 79.90 149.57 -69.67 ( 1.2 0 ) (0.96) (0.84) -(0.35) (0.03) -(0.52) ( 0 .2 2 ) (0.64) -(0.27) holdings -0.59 -0.40 -0.19 -1.60 -1.32 -0.28 0.30 2.32 -2.03 -(1.23) -G .31) -(0.56) -(0.59) -(0.76) -(0.15) (0.03) (0.36) -(0.28) holding 2 0 .0 0 0 .0 0 0 .0 0 0 .0 0 0 .0 0 0 .0 0 0 .0 0 -0 .0 1 0 .0 1 ( 1. 1 0 ) (1.37) (0.32) (0.84) (1.16) (0.14) -(0.04) -(0.31) (0.23) aveeduc 220.76** 130.58* 90.18 154.34 62.30 92.04 523.83 356.56 167.27 (2 .0 0 ) (1.85) (1.16) (0.53) (0.33) (0.45) (0.43) (0.46) (0.19) * Group interactions terms for each variable are not included in table - 51 estimated coefficients, 191 hh, 2 years (1995, 1996) - dhwrs, discrete measure o f health shock. to Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.7 (cont.) Generalized Least Squares Savings Estimates - dhwrs* G roup 1 (100 hh; 8 vill.) G roup 2 (59 hh; 3 vill.) G roup 3 (41hh; 7 vill.) svwl svdgwl svanwl svwl svdgw l svanwl svwl svdgw l svanwl aveeduc2 -29.96** -19.72** -10.24 -10.41 5.42 -15.83 -303.65 -227.68 -75.97 -(1.95) -(2 .0 0 ) -(0.95) -(0.13) (0 . 1 0 ) -(0.27) -(0.43) -(0.50) -(0.15) yrsinvil 1 2 9 3 ** 5.39 7.54** 3.97 8.09 -4.11 -0.18 -1.43 1.25 (2.33) (1.52) (1.93) (0.38) (1.19) -(0.55) -(0 .0 2 ) -(0 .2 1 ) (0.16) depth 2 -522.63** -253.81* -268.82* -365.20 -138.47 -226.73 105.40 205.47 -100.07 -(2.27) -(1.73) -( 1 .6 6 ) -(1.17) -(0.69) -(1.03) (0.17) (0.51) -(0.23) pdtopyrg 2295.75*** 326.94* 1968.80*** 1797.72 1135.79 661.93 539.72 396.87 142.85 (7.56) (1.69) (9.22) (4.67) (4.61) (2.44) ( 1 . 1 0 ) (1.26) (0.41) pdlow yrg -907.38*** -650.6*** -256.80 -1432.2 -1139.4 -292.81 -184.84 -184.44 -0.40 -(3.26) -(3.66) -(1.31) -(3.85) -(4.80) - ( 1 .1 2 ) -(0.40) -(0.63) (0 .0 0 ) dim pat -20.24 -0.96 -19.28 107.34 1 .8 8 105.46 16.80 9.00 7.80 -(0.09) -(0 .0 1 ) -(0 . 1 2 ) (0.29) (0 .0 1 ) (0.41) (0.03) (0.03) (0 .0 2 ) dpat -349.27 -40.29 -308.99* -406.92 -434.92 27.99 -79.82 -48.49 -31.33 -(1.33) -(0.24) -(1.67) -(0.95) -(1.58) (0.09) -(0.18) -(0.17) -(0 . 1 0 ) _cons 527.07 341.45 185.62 2480.52 2393.66 8 6 .8 6 86.93 -641.52 728.45 (0.95) (0.97) (0.48) (1.17) (1.77) (0.06) (0.07) -(0.85) (0 .8 8 ) 1 1 wald C hi2 -3223 -3052 -3088 -3223 -3052 -3088 -3223 -3052 -3088 163.04*** 115.48*** 149.25*** 163.04*** 115.48*** 149.25*** 163.04*** 115.48*** 149.25*** * Group interactions terms for each variable are not included in table - 51 estimated coefficients, 191hh, 2 years (1995, 1996) - dhwrs, discrete measure o f health shock. L O O J 34 Relative to groups 2 and 3, savings is not significantly higher in group 1, as indicated by the interactions. Columns 2 and 3 of Tables 1.6 and 1.7 provide results for savings in terms of durable goods and animal assets. In Table 1.6, savings falls by roughly equivalent amounts for animal assets and durable goods. When one or more adult respondent has worse health, as indicated in Table 1.7, durable goods are drawn down 152 Bolivianos or roughly 10% of the total value durable goods holdings, while animal savings are unaffected. For group 1, both types of savings are drawn down by roughly equivalent amounts. The fact that the drawing-down both productive assets and assets subject to diminishing returns is occurring among households in a subsistence economy further highlights the significance of this finding. Durable goods described in the survey are used primarily as inputs to subsistence production. That is, canoes and drift nets are used in fishing, guns, bows and arrows in hunting, and machetes in brush clearance. In addition, animal assets, while not used as inputs in the agricultural production process, are subject to diminishing returns. The larger the stock of animal assets the greater must be the labor allocated for supervision, food provision, grazing land, etc. (Udry, 1995). As one would expect households to deplete productive assets and assets subject to diminishing returns as a last resort, the empirical estimates indicate households in group 1 use dramatic means to cope with the negative shock to health. While it is not possible to directly test if such savings is used to smooth the decline in household consumption associated with the health shock, these results Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 represent a relatively strong case for the importance of savings in smoothing consumption among group 1 households. Households receiving shocks draw down both productive assets and assets subject to diminishing returns. While the measure of savings used here does not include all household level assets it is fairly comprehensive for this subsistence economy, and, to the extent that there are other assets used to cope with shocks (such as grain) it is unlikely that savings would respond in an opposite direction of the shocks (Udry, 1995). 1.4 The Labor Market Labor market participation is another mechanism Tsimane households may use to cope with idiosyncratic shocks. For households experiencing a shortfall in yields due to a health shock in the preparation season, the labor market provides an alternative to agricultural production. In effect, a household uses the labor market to manage risk or to deal expost with a shortfall in income. This possibility is formalized by a simple agricultural household model and then empirically tested. The model is similar to the general household model outlined in Chapter 2, in addition to following the dynamic framework of Saha (1994). 1.4.1 Agricultural production in a subsistence economy As has been discussed, rural households must make decisions in an environment of extreme uncertainty. Input and savings decisions in one season affect production and consumption possibilities in the next (Saha, 1994). That is, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 household’s make choices in both the dry and rainy season that determine the overall agricultural output of the household. The investment of time devoted to agriculture in the dry season (the planting season), time allocated to agriculture in the rainy season (the harvest season), in addition to output and health shocks determine agricultural output in this simple subsistence economy. Health shocks in the clearance season reduce the amount of possible agricultural production in the harvest season, as do expectations and realizations of productivity shocks in the harvest season. 1.4.1.1 Household constraints in the dry and rainy season Time spent by the household in agricultural production during the dry season is modeled as an investment in future agricultural returns. In a manner consistent with the data used in the empirical portion of this paper, households are assumed to invest in agriculture through time spent clearing old growth forest ( T ( n " ) and secondary growth forest (7 )/') in the dry season. Time is also spent working in the labor market ( T l t)) and enjoying leisure (lD ) during the dry season. The household’s time constraints for work and total time can thus be characterized by, (1.3) T - T w +1 1D 1D T ll) (1.4) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. where a > 1 indicates old growth forest is more difficult to clear than secondary season. The household earns income in the dry season from wage labor, the sale of market goods (which must be replenished in the rainy season), and from sale of stored grain (decided in the harvest season), where is the price of the agricultural good in the dry season, S * ., SM are savings of grain and the market good (the price of the market good is normalized to one), wD is the wage rate, and Cf} , are the consumption of agricultural and market goods, respectively. In the rainy season, time spent in agricultural production in the rainy season is simply a function of the agricultural production times a constant (c) times the realization of a random shock 8 * (where s* < 1), The uncertainty associated with investment of clearance labor is resolved as output shocks are realized. Again the household can spend its’ time in the rainy season working or enjoying leisure, where the amount of time working is split between agriculture and wage labor growth, r|* < 1 is a health shock realized in the dry season, and Id is leisure in the dry (1.5) TR A= cQ Y = cQ (yT Z °,T f/)e ( 1.6) (1.7) ( 1.8) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 Income earned from agricultural production and wage labor is equal to consumption, savings in the rainy season, in addition to replenishment of the asset stocks depleted in the dry season, Pl? (cQ V ) + wl(( T ^ = P,?(Cf< + SG) + C% + S l (1.9) where the amount grain storage ( SG) is now a choice variable. 1.4.1.2 Utility maximization in the dry and rainy season In the dry season, household utility is a function of current consumption and leisure in addition to the discounted expected value of consumption and leisure in the next season, U{Cf}, CG ,lD) + 8E(U(CZ, C " , /„)) (1.10) Combining the constraints defined above, the Lagrangian for the household’s maximization problem can be written as, r r < '() r p C x L = U(Cfy, C " , Td - (— f - + + T{;)) + 5 E(U(C/f, C " , TR - TR " )) V [ T ) - X D( C ^ - S M + PG (CG- S f o - w J , 0) ( 1.11) - 5 E [lR(C% +Sm -P r (g (yTfi°,T% x)& -CR - S G) - w X ) ] As modeled by Saha (1994), because of uncertainty about agricultural output in the next season, production is modeled as a function of labor input in the dry season in addition to the random shock (which is not realized until the rainy season), Pr Q(Y V)" ■ > Tj,' )e ■ 1° contrast, the health shock is realized in dry season, y > 0 is a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 parameter indicating old growth clearance is more productive for agriculture. First order conditions with respect to household choice variables include: — j-.U, - dC i a D " : U, - L P n = 0 (1.12) 0-13) - ^ ■ . l D-5E{X„] = 0 (1.14) M U . u (Za } + g E [K P ‘i] = 0 (1.15) dTD ° r| 0 : uh (4 ) + 6 c Q £[X„P;E -] = 0 (1.16) dT^ '' r\ +XDwn =0 (1.17) dT 'D D D in addition to the multipliers. Substituting to solve for marginal products of forest clearance in the dry season yield: Qfo = . Y L - i d-18) rj cy E[Pr e ] Qrr.= . (1-19) r| cE[Pr e ] Recalling that agricultural production is assumed to follow diminishing returns to scale, equation (1.18) indicates that for an increase in the difficulty of clearing old growth forest ( a ), a decrease in the productivity of old growth forest (y ), an increase in the wage rate in the current season ( wD), or a decrease in expected future Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. price of agriculture E[P‘‘] , labor allocated to old-growth forest clearance in the dry season (the current period) decreases. More importantly, with respect to the increase in the expected productivity shock (a lower s ) decreases time allocated to forest clearance in the dry season, and in turn, time allocated to agriculture in the rainy season. Similar to the dry season, household utility in the rainy season is a function of current consumption and leisure and the discounted expected value of future consumption and leisure, First order conditions with respect to household choice variables in the rainy season include: empirical estimation, an increase in the realized health shock (a lowerr)*), or an U(CR,C R ,lR) + 8E(U(Cf^Cy ,lD)) ( 1.20) The Lagrangian for the household maximization problem can be written as, L = U(CR , C“ , Tr - 7 f ) + 5 E(U(Cf}, C " , lD )) - kR (C : +S*M+ P: (Sa + C a r - cQ'e ‘) - wltT;t ) - 5 E[Xn(C % -S i +PA(CA- S G) - w X ) ] ( 1.21) ( 1.22) (1.23) dL (1.24) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 (1.25) In contrast to the dry season, first order conditions for the rainy season yield Leisure time, and correspondingly, time spent working in the rainy (harvest) season, shock. 1.4.2 Estimating labor market participation As is evident from equations (1.18) and (1.19), expectations of the productivity shock in the dry season or the realization of the health shock lower time allocated to clearance, T// s , in the dry season. This implies that time allocated to agricultural during the rainy season will be relatively lower, while time allocated to the labor market should be higher. Kochar (1999) uses a similar idea when she uses the type of crop planted by farm households (in the dry season in this context) as an estimate of the expectation of a harvests, and then tests if household labor market participation is sensitive to the idiosyncratic shock (derived based on the expected The approach used here is consistent with Kochar (1999), however, the focus is on determining the effect of the health shock in the dry season on labor market 1 2 See Kochar (1999) and Rose (2001) for a thorough treatment o f the effect o f idiosyncratic shocks on labor market participation. (1.26) are not affected by the realized productivity shock or by the expectation of the health 1 2 harvest). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 participation in the rainy season.1 3 Days of labor market participation (outside the village) is regressed on explanatory variables including group dummies, the health shock from the previous period, a vector of household demographic variables, clearance in the current season, the wage rate, and interactions with the group dummies. A censored tobit model is used to estimate the following equation, T‘* = a 0 +a,G / + a 2 r|,_u + a 3 r|/G/ + a 4Hi +asHiGJ +a(T . +a1Tit G/ + a 8 w( +a9wiGj +e, T.<-=T'* if 7)7 * > 0 T'' = 0 otherwise (1.27) where T‘* is the unobserved days of wage labor, T*' is the observed days of wage labor, T( : is days spent clearing forest, Hi are the household specific variables, andr) is the idiosyncratic shock. Because panel data is not available for labor market participation, and observations are limited, groups 2 and 3 are merged, consequently j = 1,2. Results of equation (1.27) are presented in Table 1.8. As indicated, across all villages the health shock in the previous year has no effect on labor market participation. However, for groups 2 and 3 the null hypothesis that savings is not affected by the health shock in the previous season can be rejected. The results indicate that labor supply is sensitive to the shock for households in groups 2 and 3; for each 10 days sick during the previous forest-clearing season, households work approximately 6 more days in the labor market. This result also holds when using a ' ’ Data constraints did not allow for an estimate o f the agricultural shock. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 discrete variable for labor market participation and the logit model, households in groups 2 and 3 receiving the shock in the previous season are more likely to increase labor force participation in the next season. The group shock interaction terms in Table 8 (gl | g2 under the shock term) also indicate that households in group 1 experiencing a shock work significantly fewer days in the labor market relative to households in groups 2 and 3 experiencing the shock. Overall group dummies (gl | g2 in row 1) indicate households in group 1 work fewer days relative to households in groups 2 and 3. As expected, clear (the number of days spent clearing forest during the current year) is negative and significant and the interaction term (not listed) indicates that households in group 1 clear significantly more forest than households in group 2 (subsistence agricultural production is more dominant among households in group 1). Education variables (aveduc and horeb) are highly significant and positive, as is ptimber, the percentage of days worked in the timber industry out of total labor market participation, and the level of household wealth. Curiously, the wage rate is negative and significant for groups 2 and 3. Quite possibly, the sign of the wage rate is consistent with a strong income effect in a subsistence economy where leisure is valued very highly. A higher male population is positive and significant for group 1 households. The finding that relatively remote Tsimane rely more heavily on the labor market following periods of adverse shocks may also partly explain their lack of reliance on savings instruments. As discussed in Kochar (1999), if households have Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. access to flexible labor markets it may be that accumulation of savings assets is not necessary to protect against negative shocks. While summary statistics indicate that households in group 3 save more as a percentage of wealth, households in groups 2 and 3 do not rely on savings to cope with shocks. Table 1.8 Labor Market Participation, 1995 (censored normal regressions with robust standard errors)*_______________________________ Com bined Croup 1 G roups 2 & 3 g2 | g l 159.617** -159.617** (2.041) -(2.041) schaco_l 0.263 -0.192 0.610* ( 1. 1 1 1 ) -(0.671) (1.649) _g 2 |_ g 1 .802** -.802** --- (1.70) -(1.70) clear -1.193 -0.235 - 2 3 9 9 *** -(1.535) -(0.404) -(2.556) ptim ber 0.014*** 0 .1 1 1 *** 0 .0 1 1 *** (3.490) (4.078) (5.836) with 0 .0 0 0 -0 .0 0 1 0.014** -(0.007) -(0.166) (1.993) m alepop 8.564 51.667** -17.144 (0.767) (1.934) -(1.399) -1.265 -2.106** -0.142 avgage -(1.393) -(2 .0 1 0 ) -(0.154) aveeduc 3.000 3.620 11.476 (0.448) (0.595) (1.044) horeb 9.698 -5.776 41.488*** (1.030) -(0.926) (5.942) dpat 15.540 80.301*** -55.145** (0.632) (2.301) -(2 . 1 0 1 ) dim pat -2.832 -8.737 -30.054 -(0.114) -(0.408) -(1.059) holdings 0 .0 0 0 0.007 -0 .0 0 1 -(0.015) (0.829) -(0.023) 0.179 -0 .2 2 1 -0.761** new outw g (0.484) -(0.545) -(1.930) cons -5.304 -80.862 78.755* -(0.117) -(1.304) (1.656) II -583.0 -549.4 -549.4 wald C hi2 29.31 *** 206.1*** 206.1*** However, their reliance on the labor market (if indeed temporary) is consistent with the permanent income model. Combined with the evidence from the next section on * 100 left censored observations (at zero), 88 uncensored observations Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 informal credit and remittances it appears that savings is not as important to households in groups 2 and 3, not only because of poor asset markets, but, because of stronger labor markets and a heavier reliance on risk sharing mechanisms. 1.5 Informal Credit and Remittances The use of reciprocal credit and gift giving is a type of risk sharing consistent with the full insurance model discussed earlier. The Tsimane survey allows one to test for a version of this hypothesis by using data on loans or remittances from family or friends in addition to a shock variable (the number of days sick during the present (1995) clearing season). While information on reciprocal credit is not available, the results indicate that households in groups 2 and 3 tend to rely more heavily on risk sharing via informal loans and remittances to cope with current idiosyncratic shocks relative to the households situated closer to the market in group 1. Direct extensions of Diamond (1967) and Wilson (1968) show that if full insurance exists, individual income follows aggregate village consumption, and, consequently, individual consumption is independent of idiosyncratic shocks. Using a model for contingent commodity markets, Diamond (1967) shows that for individuals facing the same prices and subjective probabilities consumption variation depends only on risk aversion, U](c,(0 ))*,<?) = = K 1 4 (1-28) 1 4 Where hj is subjective probability and p(0 ) is the price if stateG occurs (equation (33) in Diamond (1967)). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 That is, for households with the same utility functions, subjective probabilities, and prices the weighted marginal utilities are equal across all households in the village (Townsend, 1994; Ligon, 1998; Bardhan and Udry, 1999). Individual consumption is a function of the households weight in the Pareto program ( A ,.) and moves with aggregate village consumption over time. Consequently, the consumption smoothing hypothesis can be tested by regressing individual consumption on individual income and aggregate village consumption (Townsend, 1994). Assuming that utility functions of village households are equal and follow a constant absolute risk aversion form, Diamond’s (1967) result can be written as for k individuals and i households where A , represents the households weight in the Pareto program (Bardhan and Udry, 1999; Fafchamps and Lund, 1999; Townsend, 1994). Consumption of the household is a function of aggregate village consumption plus the time invariant fraction of the households weight in the Pareto program. If the households budget constraint is given by income and informal loans and remittances, cjt = yjt + btl + rn , a simple estimation equation can be used to test if loans and remittances are enough to cover the income shortfall associated with the sickness shock. As in equation (1.2), income can be divided into a transitory component and a permanent component, where the transitory component is estimated Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 by the health shock and the permanent component along with the welfare weights by a vector of household characteristics (Fafchamps and Lund, 1999). Given the available survey data the equation to be estimated can be specified as, bt +rj =a0 +atGj + a 2 r|/ +<x3 r|/G. + a iHi + a sHiGj +e( (1.30) Both censored normal regressions and logit estimates (where bj + r, is changed to a discrete variable that takes the value o f one for households receiving remittances or loans from family or friends) are presented in Table 9. Regrettably, the data does not allow one to test over time. Flowever, the results can signal a rough proxy as to the degree to which risk is shared at one point in time. As in the case of the labor market participation estimations, groups 2 and 3 are combined and compared to group 1. If efficient risk sharing exists, coefficients of the shock variables should be significant and of the same order of magnitude as income shock. As indicated in Table 1.9, group 2 and 3 households experiencing greater levels of sickness during the clearance season are more likely to receive loans and remittances. At the same time, households in groups 2 and 3 receive significantly fewer loans relative to group 1. However, the hypothesis that the probability of receiving an informal loan or remittance is unaffected by the health shock cannot be rejected for group 1 households or across all Tsimane households. As expected, over all households, households in villages with a greater population are more likely to receive a loan, as are households with higher average education, and those with a higher percentage of “patient” adults. In groups 2 and 3 households with a higher Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 average age are less likely to receive loans. While these results for informal credit and remittances are not as robust relative to savings and the labor market (adding wealth overpowers any effect of sickness), they provide limited evidence that households in more remote Tsimane villages rely on risk sharing mechanisms to cope with negative shocks whereas households in closer proximity to the market do not. Table 1.9 Informal Loans and Remittances, 1995* Censored Normal Regressions (with robust standard errors) C om bined G roup 1 G roups 2 1 Logit Regressions C om bined G roup 1 G roups 2 & 3 g 2 1 gl -3192.203*** 3192.203*** -(2.33) (2.33) cr»nor»A 7.229 5.925 10.501* 0.009 0 .0 0 0 0.026* stndco (1.51) (0.75) (1.83) (1.15) ( 0 .0 2 ) (1.84) _g 2 ] _ g l 4.576 -4.576 0.026 -0.026 (0.51) -(0.51) --- (1.40) -(1.40) avgage 3.723 20.245 -24.100* -0 .0 0 2 0.030 -0.052* (0.41) (1.45) -(1.71) -(0.13) (1.39) -(1.63) -19.550 -3.421 -56.241 -0.018 0.048 -0.126 num tot -(0.49) -(0.06) -(1.08) -(0.27) (0.52) -(1.13) 209.920* 253.859* 64.218 0.198* 0.229* 0.129 aveeduc (1.71) ( 1.6 6 ) (0.55) (1.81) ( 1.6 6 ) (0.51) 334.979 -265.209 5075.918*** 1.096* 0 .0 0 2 4.245*** dpat (0.89) -(0.45) (4.57) (1.67) (0 .0 0 ) (2.49) pdlow yrg -249.729 63.658 -639.248 -0.409 0.622 -1.641 -(0.67) (0.14) - ( 1.0 2 ) -(0.64) (0.76) -(1.41) vpop 1.487 1 .0 1 0 1.426 0.003* 0.003 0 .0 0 2 (1.41) (0.56) (1.08) (1.78) ( 0 .8 8 ) (0.79) cons -1947.342*** -2129.997** -5322.201*** -3.089*** -3.654*** -3.654*** -(3.14) -(2.37) -(3.89) -(3.08) -(2.94) -(2.94) 1 1 -430.5 -425.8 -425.8 -96.99 -90.52 -90.52 wald C hi2 12.77* 4 2 3 * * * 42 3*** 16.17** 2 9 ] * * * 2 9 i *** pseudoR 2 . . . . — . . . . .0769 .1385 .1385 1.6 Discussion and Conclusion A principle finding, most consistent with the permanent income model, is that some Tsimane households rely on changes in assets to cope with health shocks. * For censored normal regressions: 149 censored, 45 uncensored observations. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 Similar to the findings of Udry (1995) and Rosenzweig and Wolpin (1993), changes in assets among Tsimane households are found to be sensitive to idiosyncratic shocks. Across all villages, the null hypothesis that savings is unaffected by the negative health shock is rejected. In contrast to the African context, however, Tsimane households are found to draw down animal assets in addition to productive durable assets. Furthermore, when analyzing savings by group, this result is only robust among group 1 households who possess the most complete asset markets, and, relative to groups 2 and 3, face equal opportunities for informal credit, and possibly labor. Savings is unaffected by the shock in groups 2 and 3. A second result, similar to the findings of Rose (2001) and Kochar (1999) using Indian data, is that idiosyncratic shocks appear to affect labor market supply among some Tsimane households. Households from groups 2 and 3 that experience health shocks during the forest clearance season are more likely to participate in the wage market in the next season. Temporary labor market participation on the one hand serves as a mechanism for the household to cope with the shock, a result consistent with the permanent income hypothesis. At the same time, the labor market can also serve as a mechanism used by the household to manage risk. For group 1 households, and across all Tsimane households, labor supply is unaffected by the shock. Finally, the study finds results consistent with the risk sharing type models in the most remote of Tsimane villages. That is, remote Tsimane households rely on informal loans and remittances to cope with current health shocks, whereas villages Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 closer to San Borja and the major trade river do not. Mean loan values are less for groups 2 and 3 relative to group 1, but households in groups 2 and 3 are more likely to receive such loans following a season of sickness whereas group 1 households are not. Moreover, while the results in this section are not as highly robust, they quite possibly provide good evidence that more remote Tsimane, with limited access to formal goods markets, rely more heavily on other Tsimane to meet shortfalls in income due to sickness. In conclusion, using data from subsistence level villages in semi-tropical Bolivian forests, this study finds similar results as empirical studies from rural India and Africa. Indigenous households experiencing health shocks draw-down savings, increase labor market participation, and receive greater informal loans and remittances when experiencing idiosyncratic shocks. In addition, the study also finds that health shocks result in the use of different risk coping mechanisms among a highly homogenous ethnic group. More remote Tsimane tend to rely most heavily on the labor market and informal credit, whereas Tsimane located closer to the Maniqui River and the market center of San Borja, depend primarily on the drawing down of animal assets and productive durable goods to cope with negative health shocks. These results suggest that policies directed toward reducing the inefficiencies that result from risk among the rural poor should be carefully crafted in light of the local institutional context. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 Chapter 2 Market Integration, Development, and Smallholder Forest Clearance 2.1 Introduction In the rural areas of developing countries households are particularly dependent on the environmental resource base for consumption and production needs. Worldwide, the rural population comprises roughly 53% of total population and 51% of the economically active population is employed in agriculture (FAOSTAT, 2002) In Bolivia, the country particular to the data used in this study, roughly 80% of the economically active population in rural areas are smallholder farmers (ECLAC, 2000). At the same time, smallholders and the rural indigenous poor are among the poorest of the poor. Recent World Bank data indicates that of the 1.2 billion poor consuming less than a dollar-a-day, 75% live and work in rural areas (IFAD, 2001). In Bolivia, the rural poverty headcount index is 81.7% (with a rural to urban poverty ratio of 2.42) and rural destitution rates are estimated to be 59.1% (IFAD, 2001; MECOVA, 2001). Taken together, poverty and biomass dependence are highly correlated with the high rates of forest clearance frequently observed among smallholder farmers.1 This relation has become increasingly apparent to policymakers who are consequently shaping rural development policies to satisfy the dual objectives of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 alleviating poverty and reducing environmental degradation. As such, this chapter is intended to not only contribute to the literature on smallholder deforestation, but also to explore the linkages between economic development and the rural smallholder’s use of forest resources. In this paper, both analytical and empirical methods are used to explore the effect of market integration, time preference, technology, and forest productivity on the use of forest resources at the household level. In terms of market integration, the effect of market integration on the opportunity costs of forest clearance (in terms of the effect on prices of agriculture and the wage rate) is particularly explored. Changes in the discount rate accompanying market integration are analyzed in light of a household’s willingness to invest in future clearance and technology, and agricultural productivity (along with forest productivity) are both highlighted in the model and explored empirically. Throughout, the attempt is made to make the model and assumptions particularly relevant to Latin American, rural, ethnic indigenous smallholders, where destitution and biomass dependence is especially great. The model is a partial equilibrium model; it is not a comprehensive model of forest clearance. Because the focus is on policy parameters, no attempt is made to explicitly model every factor in the household production function. Certain simplifying assumptions are also made in order to make the comparative statics more clear. Specifically, it is assumed that a) the smallholder is risk-neutral, b) the 1 According to Angelson (1995), estimates o f the share o f deforestation from shifting cultivators range from 45% to 60%. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 smallholder is rational and has full information about prices and wages in the present and following year, and c) there exists some contribution to household welfare from non-timber forest products collected from standing forest; these non-timber forest products act as complements to subsistence agriculture. The result is a highly flexible model that can be adapted and expanded easily to explore smallholder forest clearance in a variety of settings. In Section 2 the context of development in Latin America is discussed. Section 3 surveys the literature on smallholder forest clearance and, in light of this literature, a two-period, two-season agricultural household model is developed in section 4. Hypothesis generated from the model with respect to market integration, patience, technology, and forest productivity in section 5 are empirically tested in section 6. Results and conclusions are presented in section 7. 2.2 Background Latin America and the Caribbean alone are home to approximately 27.5% of the world's forests (FAO, 1999). Given the relative size of Latin American forests, forest clearance levels have been among the highest in the developing world. The change in forest area from 1980 to 1995 in Latin America was -9.7% (compared to an average -9.1% for the developing world during the same period) (FAO, 1999). Forest clearance in tropical South America accounts for the majority of forest clearance in the Latin American region. Regional forest clearance from 1990 to 1995 was estimated to be 5.3 million hectares per year, of this amount, the loss of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 tropical forests in South America averaged nearly 4.7 million hectares per year (FAO, 1997); smallholders account for the majority of this forest clearance. Traditional and ethnic indigenous smallholder populations still comprise a significant portion of the population in developing countries and especially Latin America. In 1988 smallholder farmers (households with up to three hectares of crop land) accounted for 52% of the rural population in 114 least developed countries and 38% in Latin America and the Caribbean (IFAD, 1992). The ethnic indigenous population also makes up a non trivial constituent of the rural population in least developed countries (7.3%) and account for over 27% in Latin America and the Caribbean (IFAD, 1992, p. 49-50). In Bolivia, 95% of the rural population is ethnic indigenous (IFAD, 1992), and as already mentioned, over 90% of the economically active population in rural Bolivia are smallholders. Official Bolivian statistics estimate indigent poverty among rural indigenous households to be 62.7% and overall poverty to be 85.1% (MECOVA, 2001). Traditional approaches to development, popular from the 1940’s through the 70’s, emphasized using the rural sector as a means of expanding the modern sector. For example, in the influential Lewis (1954) model, excess labor from the traditional sector supplies country capitalistic "islands" (islands within the modern sector that are not fully developed.) Development diffuses from these centers. Following the debt crisis experienced by many Latin American countries in the early 1980's, macroeconomic stabilization and short-run policy objectives prevailed as the dominant focus of many governments throughout the decade (de Janvry et. al., Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 1993). More recently, with the growth of the new institutional economics, approaches emphasizing tailoring development strategies to particular institutional contexts have grown in influence. Recognizing the significant barriers that exist between the rural and urban sectors, government policies have focused on improving these linkages, and consequently, have implemented policies to reduce transaction costs between rural and urban sectors. While it is arguable that a majority of such policies have been crafted in light of the local institutional context, they have, nevertheless, reduced transactions costs thereby improving market integration. A subsequent evolution of markets in rural areas in combination with rapid technology growth has meant that even the most traditional of societies are interacting to some extent with market economies. In remote Amazon villages where tuberculosis, malaria, and dysentery are pandemic, village leaders use GPS technology for staking-out tribal boundaries. Households may also have limited, but increasing, access to modern pesticides and herbicides. Such technological expansion brings with it not only direct effects but gradually changes incentive structures in the village thereby affecting how land, labor and cash are used (Godoy et. al., 1996). In the paper that follows, the ways in which market integration influences the economic decisions made by households in transitional economies are explored. Unlike other more “descriptive models” of smallholder forest clearance, we develop a simple agricultural household model in which we explicitly consider the balance between cash crops and subsistence agriculture, the links between standing forest Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 and the provision of subsistence goods, the tradeoffs between forest clearance labor and off-site wage, and the role of time preference in the smallholder’s allocation of time across farming, wage labor, and forest clearance. The fact that many rural farmers have the option of converting both old growth and secondary growth forests, is recognized and also included in the model. The model also accounts for differences in the productivity (and costs of clearance) of old growth and secondary growth forests for both non-timber products and for agricultural land that influences the rates at which each type of forest should be cleared for agriculture. The model also highlights the roles that technological change, especially improvements in the farmer’s ability to clear forest, influences agricultural and forest clearance decisions at the household level. Finally, insights are provided into the way that shrinking transaction costs, a hallmark of market integration, influence forest clearance. 2.3 Literature Review Consistent with Kaimowitz and Angelsen (1998), a partial sample of the recent literature with respect to models of smallholder deforestation is summarized in Table 1. Typically, these models adopt either the subsistence or the open economy approach to explore smallholder forest clearance. Under a standard subsistence approach, the household’s objective is to satisfy a minimum nutritional requirement with the least amount of labor; rates of clearance are fueled mainly by population 2 Kaimowitz and Angelsen (1998) give a thorough summary o f analytic household models o f deforestation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 growth and declining agricultural productivity (Angelsen, 1995; Angelsen, 1999; Dvorak, 1992; Stryker, 1976). In contrast, the open economy approach highlights the role of prices, technology, and accessibility to the market on rates of deforestation (Angelsen, 1995; Angelsen, 1999; Bluffstone, 1995; DeShazo and DeShazo, 1995). The majority of these approaches recognize forest clearance as a means of increasing arable land and securing property rights on primarily open access land. Similar to Angelsen (1999), we model deforestation as an investment. However, in contrast to much of the existing literature summarized in Tables 1 and 2, we do not view household clearance as strictly a title establishment strategy. Rather, we model deforestation as a short-term investment in agriculture. We assume relatively secure short-term tenure3 where time is the household’s only clearance constraint and the decision to clear is based on the expected profitability of agriculture in the next period. This approach seems consistent with the empirical context. Tsimane villages are located in abundant and officially recognized Tsimane forests or in the Beni Biosphere Reserve. Encroachment onto Tsimane land by abutters has been shown to be minimal (Godoy et. al. 1998). 3 Tenure in our empirical case is secure only as long as the plot is cultivated - usually two years. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 2.1 Overview of Selected Theoretical Models of Smallholder Forest Clearance4 Category Optimization Problem Contributions of Model Labor Market Property Rights Regime C urrent M odel tw o-period open econom y m odel hh m ax utility s.t. tim e and incom e constraints tw o-season; accounts for distance, tim e preference, gathering, prim ary vs secondaiy clearance costs perfect open access — clearance secures short-term tenure Angelsen (99) M4 dynam ic open econom y m odel hh m ax agricultural profit accounts for tenure security, tim e preference, and distance perfect open access - clearance secures long-term rights O 4 ) Southgate (90) static open econom y m odel hh max agricultural profit explores conservation labor perfect open access - clearance secures long-term rights 2 © S ^ Larson (91) static open econom y m odel hh m ax agricultural profit expands on Southgate (90) perfect open access - clearance secures long-term rights C 4 > o U cS * Mendelsohn (94) M l static open econom y m odel hh m ax profit accounts for distance na frontier open access - clearance /defense expenditures secure S s long-term rights s - s Angelsen (95) M2 static open econom y m odel hh m ax agricultural profit accounts for distance perfect open access — clearance secures long-term rights Stryker (76) M2 static open econom y m odel hh m ax utility s.t. tim e and incom e constraints m odels travel tim e betw een village and fields na open access DeShazo & DeShazo (95) static open econom y m odel hh m ax agricultural profit distinguishes betw een land types and crop types na established property rights or clearance secures rights 4 See Kaimowitz and Angelsen (1998) for comprehensive overview o f the tropical deforestation literature and Angelsen (1999, 190) for a similar table comparing the four modeling approaches used in the agricultural household literature. C O O O Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 2.1 (continued) Overview o f Selected Theoretical Models of Smallholder Forest Clearance Category Optimization Problem Contributions o f Model Labor Market Property Rights Regime Vi -= D D E u 1 £> Angelsen (99) M2 static open econom y model hh m ax utility s.t. budget and tim e constraints production and cons, decisions not separable because o f im perfect labor mkt. constrained com m unal - hh owns constant fraction o f tribal land s * rvi Q. O Angelsen (99) M3 static open econom y m odel hh m ax agricultural profit accounts for distance perfect com m unal - hh owns constant fraction o f O. tribal land o | o Mendelsohn (94) M2 dynam ic open econom y m odel hh m ax present value o f sustainable forestry and sustainable and unsustainable activities; na clearance guarantees only short-term ag. agriculture pr(losing land title) > 0 rights V 0 ) ■ o o % Dvorak (92) static subsistence m odel hh m inim izes labor given subsistence requirem ent output fn(fallow length, cropping, and w eeding); prim ary vs. secondary clearance perfect shifting cultivation C J u e a > V I Angelsen (99) M l static subsistence model hh m inim izes labor given subsistence requirem ent standard na com m unal - hh owns constant fraction o f tribal land •a 3 Angelsen (95) Ml* subsistence m odel descriptive m odel standard na na Vi Stryker (76) M l static subsistence model hh m inim izes labor given subsistence requirem ent travel tim e per farm er constrains production na open access sO 60 Our methodology differs from much of the literature in that we adopt a two- season, two-period framework (DeShazo and DeShazo, 1995, also use a two period model) that combines elements of both the open economy and subsistence models. We also include changing transport costs and our model allows the household to choose between clearance of primary and secondary forest. The two-season approach captures the seasonal nature of the smallholder’s choice problem: the decision to farm (e.g. preparing the soil or land) precedes sowing and harvesting by one season (Saha, 1994).5 Forgone wages are the opportunity cost of clearing forest to increase future agricultural production in the model. The impact of transport costs on wages, agricultural prices, and the prices of consumption goods also factor into the household’s time allocation decisions. Households are also given a choice between clearing old and secondary growth forest; plots derived from secondary growth are relatively less productive than those derived from primary growth forest, but require less time to clear. Assuming a short-term investment horizon, even though land is relatively secure, appears to suit the cultural context. The majority of agricultural production is presently derived through shifting cultivation; the Tsimane still view agriculture as inferior to hunting and gathering (Godoy et al. (1996) citing Riester (1976) and Piland (1991)). Plots are typically used only one to two seasons before being returned to fallow (Godoy et al. 1996). Once abandoned to fallow, land essentially 5 See Saha (1994) for a two-season agricultural household modeling approach that also incorporates price and yield uncertainty. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 becomes an open-access resource once again.6 The two period model we use to highlight the smallholder’s decision at the margin is consistent with the existing literature. We could develop this same model as an n-period model and use Bellman’s equation to model the marginal decision; the results would be the same. The extension of the model to n periods does not change the smallholder’s forest clearance behavior at the margin (i.e. the maximum principle is met each year). Therefore we keep the model simple for the sake of clarity. We also attempt to keep the model in the simplest terms possible in order to focus on specific policy parameters. We concentrate on the most basic applications of prices, smallholder labor, output, and agricultural production. We also include parameters that allow for changes in deforestation capabilities, changes in agricultural production, changes in non-timber forest yield, prices, and even impatience. O f course, all of these factors themselves may be influenced by other factors. For instance, agricultural productivity in any period may depend upon the application of fertilizers or the type of soil. Rather than clutter the model with every factor relevant to the farmer’s decision, we limit ourselves to primary factors and leave it to the reader to explore how other factors influence our analysis. In fact, the simplicity of the model makes such investigations straightforward and represents an important strength of our approach. We remind the reader that the model we provide is not a comprehensive model of total forest clearance. Instead, we present a model 6 The household does maintain use rights, however, over perennial fruit trees planted during their tenure. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 that allows us to look at the way in which specific policy parameters influence forest clearance decisions. Our study also differs from much of the existing literature on smallholder clearance in that we utilize an empirical model to test the predictions of the theoretical model. In a series of articles, Godoy et al. (1996 and 1998) explore linkages between development and forest clearance among South American indigenous peoples. In their first paper, “The effects of economic development on neotropical forest clearance: household and village evidence from Amerindians in Bolivia” (Godoy et a l, 1996) the authors test several hypothesis of smallholder clearance among the Tsimane, Mojeno, and Yuracare People of Bolivia. In this paper they include some of the basic assumptions behind the subsistence and open economy approach. Godoy et al. find that there exists a negative relation between wage income and clearance and a positive relation between farm income, technology, and clearance. In a second article, Godoy, Jacobson, and Wilkie (1998) more explicitly examine the link between forest clearance and development by exploring the role of time preferences in a descriptive empirical model. They find that more patient individuals (those with a lower discount rate) engage in more old-growth cutting while less patient individuals work relatively more outside of the village and cut n more new growth forest. n The method used to measure impatience will be discussed in the empirical section o f this paper. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 We offer a micro-economic foundation to explain many of the findings of Godoy et al. (1998). Like Godoy et al., we develop an empirical model, but we focus specifically on important explanatory variables from our theoretical model. We also include additional variables used by Godoy et al. and others (as summarized by Kaimowitz and Angelsen (1998)) in the empirical model for ease of comparison and to test for the relevance of the subsistence hypothesis. 2.4 The Model To begin, household utility (or welfare) is assumed to be a simple linear function of the present amount of the market good consumed in the initial period, X°m , and the discounted future value of consumption of the market good consumed in the next year, X)n , u = U{Xl) + (±-)U(X'j (2.1) o More patient households place a relatively greater weight on the future and thus more patient households have a relatively lower discount rate, 5 . The household maximizes utility subject to a production function, a time constraint, and a budget constraint. Agricultural production at time t, Q’ A, is a function of the time the household devotes to agriculture at time t , TA, / = 0,1 and the total area of land devoted to agriculture, A . & = ( V , A ) (2.2) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 New agricultural land can be derived from old growth forest and secondary growth forest. (We denote the absolute level of each type of forest in period 0 as F0 and Fs respectively.) We can distinguish the contribution to agriculture of land derived from old growth forest ( Ac)) and secondary or new growth ( As ) and rewrite (2.2) for period 0 and 1, respectively, as, Q0 =Q(T°a ,A o,As) (2.3) Q'=Q(T x a,A 0 + A I A s + 4 ) (2-4) where A x a and Aa represent the initial stock of arable land and A() and As represent the addition to arable land through forest clearance. We assume that the agricultural productivity of land derived from old growth forest (<r) differs from that of secondary growth (p ) , where cr,p> 0 and in practice < r > p . We represent the , ^ j • • dQ ,dQ . , dQ rdQx represent the ditterences in productivity a s = cr(— ) a n d = p (— ). dAa dA dAs dA Agricultural land can be augmented in the second period by forest clearance (of possibly open access lands) in the initial period. Production is assumed to be a mix of cash crops and staples and excess agricultural goods can be sold at market to acquire market goods. To specify the household's time constraint, assume that the household uses its time in the initial period for agriculture, forest clearance or for working outside of the village for a wage rate (TA, 7) , and Tw, respectively.) The household also chooses how to allocate its time differently in the rainy and dry seasons in each Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 period (rainy seasons precede dry seasons in each period.) In the rainy season (represented by subscript R) the household chooses between time spent working outside of the village (TwR) and time spent on smallholder agriculture (TAR). In the dry season (represented by subscript D), the household allocates time between working for an outside wage rate (Tw l)) and time devoted to the clearance of primary (TrOD) and secondary (TFSD ) forest. For simplicity, it is assumed that gathering from the forest is done by other members of the household.8 Therefore, the household’s time constraint for the initial period in the rainy season is, T° + T° < T° (2 51 A R w R R and in the dry season T° +T° + T° < T ° (2 6) 1 P O D + 1 P S D ^ 1 w D - 1 D ■ The household's choices are the same in the rainy season of the second period. However, since the model ends after two periods, households will not choose to clear forest in the second period since there is no gain to forest clearance. The time constraints for the second period are therefore, <2.7) T:.„<Tl. (2.8) Total household income includes income earned from crop sales (see below) and from “outside the village” employment. Agricultural production sells at the market This simplification does not change the marginal decision o f the household in the model. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 for a price of PA in periods / =0,1. Similarly, the market good can be purchased in the market for a price of P 'm in periods t = 0,1. The household, however, does not face the market prices defined above because of the transactions costs that exist between the modern and the rural sectors. In the setting of a remote rural economy it is especially important to consider transactions costs.9 We define agricultural prices in the village as (PA - cA ) and consumption commodity prices as (P * m - cm ), where agricultural and commodity-specific transaction costs are captured by cA and cm. Also, we assume that the price received for outside the village labor, w' m, is influenced by transactions costs. The take home wage received for labor is defined as {PA - cA ) since often individuals must travel long distances to work outside of the village. As mentioned earlier, transactions costs may include transport costs and are thus functions of both the distance to the market and conditions of the roads or paths. Households receive the prices defined above for agricultural production beyond subsistence and for wage labor. All income earned is spent in the same period for market goods, X'm . There is no savings or storage in the model across periods. Excess agricultural production is defined as the amount of production beyond subsistence agricultural production, Q' - Qs , where subsistence agriculture, Qs , is one means of satisfying a minimum nutritional level. 9 See Omamo (1998a; 1998b) and Oczkowski and Philp (1994). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Subsistence consumption, N , is the households minimum nutritional level and is met by gathering from the forest, N ' FR, and by subsistence agricultural p ro d u ctio n ^ in the rainy seasons of both periods. Products gathered from the forest (herein referred to simply as gathering) substitute for subsistence agriculture at a fixed rate a , thus subsistence agriculture can be written as, If no gathering occurs agricultural production must at least be equal to N in order to support a basic nutritional level. If the quantity of products gathered from the forest is an increasing function of the amount of primary and secondary forests available, gathering can be represented as, in the final period, where F < > and Fs are initial levels of old growth and secondary clearance, and 0 is a constant that indicates that old growth forest and secondary growth forests differ in productivity for subsistence gathering (in practice 6 > 1 .) Reducing the size of the forest reduces the subsistence value of gathering. Qn = N - a N 'h . lr (2.9) N R R = dFo + F s , (2.10) in the initial period and (2 .11) growth forest areas respectively, F( \ and Fs ! indicate the loss of forest due to forest Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 Equating spending in each period to earned income and substituting (2.10) and (2.11) into equation (2.9), the household’s budget constraint for periods 0 and 1 can be rewritten as, ( p : + c ° m )x:=(p:-co A mT:R , - (2-12) (K +c'm )X'm = ( P '- c\)[Q(T\r , A0 + A [ 0 , As +A I ) - Q n(F o - F ( \,F s - F ■ ■ ) ] + W m-c l)(T iD+rw R ) In both periods, expenditures are equal to earned income (i.e. there is no savings in the model). In our model, forest clearance essentially is an investment in future agricultural output. The cost of the investment is foregone wage earnings, but the payoff is increased agricultural production. If we were to model the farmer’s problem for more than two periods, we would need to include costs of protecting this investment (e.g. markers, fences, vigilance). Furthermore, we take only the opportunity cost of the farmer’s time as the cost of investment. If forest clearance required material investment, we would need to incorporate these costs. Similarly, if the farmer were to borrow to pay for this investment, we also would include the cost of interest. The traditional Tsimane tend to clear forest areas for short-term agricultural production (e.g. one or two years.) More modern Tsimane, however, may use cleared areas for longer-term agricultural production. The changes in arable land and consequently the relative size of the forest in the second period are both functions of time spent clearing forest in the initial period. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 Specifically, the amounts of old growth and secondary growth forest cleared are defined as linear functions of the time allocated to forest clearance, (2.i4) where y/ and y are exogenous parameters that describe the efficiency of forest clearance and y/ < y since clearance of primary forest requires more time to clear than secondary growth forest. The change in arable land is equal to the area of forest cleared, y,=Fo=vK,„ (2-10 4 = 0 1 = ) < • „ (2.17) Since forest clearance only occurs in the initial period, equations (2.14) - (2.17) indicate that the household's time allocation to forest clearance in the initial period determines the amount of new arable land available in the second period. Given the above, the budget equations can be rewritten as, x °.= 7 ^ 4 ( [ e ( 0 , . 4 , . 4 s ) - ( ^ - “ ( « o + o » ] ( P n, + C m ) ( 0 _ 0 ^ ^m ) fT’ Q I T1 ® \ / nO . 0 \ V w i) i wR) ( P n, + C „ , ) (2.18) K = T T k u [Q(rA it ’ 4; + < p T :o »As +rn sD ) ( ^ 1 + 4 ) C N -a {G (F 0 -(pT°O D ) + F s - f O ) ] (2.19) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 Assuming utility is a linear function10 that follows U=X°„+jK (2-20) where X°m is defined by (2.18) and (2.19), the household chooses TL » tIr> Tfod»Ti'si)■ Tw d■ TA R > ’Th>t0 maximize utility subject to time constraints (2.5), (2.6), (2.7), and (2.8). The Lagrangian can be formulated as, i = (221) - K (r i« + - r «) - 4 , (r,'„ - ) where (2.18) and (2.19) substitute for X m . Taking the first order conditions with respect to the choice variables, yields the following: CL CQ _ K = 0 < « K + < d T M dT° w R in m > n t P°+c° 1 m m dL . 1 P \-c 'A{ dT° U 1 R O D ' t K + c l , dL 1 P 1 - c 1 . 1 r A A r dT° ui F S D ' 8 P' +c' m m dL 0 0 dA dA jpU po 0 v t - 7 9 '« ,r (2.23) (2.24) (2.25) (2.26) 10 We assume linearity in utility for the sake o f simplicity. In doing so, our model becomes similar to a constrained profit maximization model (constrained by time and subsistence) except we account for changes in the real buying power o f income and the effects o f discounting. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 dL . 1 P \ - c \ . dQ JT'ar S K + c, dT, (2.27) A lt dL 1 w > - c„ 8P»+ c'm - T = 0 (2.28) < iL 1 wl - cl dTlD S P l+ c \ -A ' = 0 , (2.29) dL d X , ■ T + T 1 * 1 A lt ^ 1 w it -7)1=0 _dL_ i _ i • 1 w D 1 D U (2.30) (2.31) 2.5 Comparative Statics At the optimal level of forest clearance, if the smallholder maximizes utility, marginal benefits of adding arable land will equal to the marginal costs. This general result can be seen more clearly when substituting (2.27) into (2.24) and (2.26), old growth forest clearance: A p ; -4 )[< jp A - a m = (w °- c l ) (2.32) 8 dA Pm+cm secondary growth forest clearance: V ; -c'A) [ p r ( % - a r ] = ( w : , (2.33) 8 dA Pm+cm The left hand side of equation (2.32) represents the discounted return to an additional unit of land; in this case the village price for the cash crop times additional Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 agricultural output less the reduction in subsistence production from lost forest. The right hand side is the marginal cost of forest clearance in the initial period in terms of the foregone wage, weighted by the relative future price of consumer goods. We assume that the marginal returns to agriculture are a decreasing function dQ_ > Q d^Q dA ’ dA2 of land, that is, — ~ > 0,— y f < 0 . Since the rate at which forest clearance labor produces arable land is linear, we can write > 0, y < 0. Additional time dTp dTF allocated to forest clearance increases agricultural production at a decreasing rate. Substituting in A = F and simplifying equations (2.32) and (2.33) yields the contribution to agricultural output of the marginal unit of arable land and of time spent clearing forest, = — [ S (w" ; ~ c; } + C 'l) + ad(p] = — Q 0 (2.34) dF0 ocpV { P \- c [ A){ P :+ c l) oq> = — [ S ~ c^ + ay] = — ns (2.35) dFs p y l { P \- c \) { P : + c D p y s where Q 0 = [S ^ ~ ^ + + a0(p\ , Q s = [S (Wm~ ° ^ + ° ^ +ay] and ( p i - c i A) ( p : + c i) s ( p ' - c ’) ( ^ : + c:) = (2.36) /,()l) dQ = Q s , . (2.37) dT/w Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure One demonstrates that old growth forest will be converted to agriculture as represents the equilibrium amount of old growth forest a utility maximizing household would choose to clear. Also following from above, Figure Two shows The equilibrium amount of time spent clearing primary growth forest is given by T^)D in Figure Two. We examine the way in which policy and economic variables influence forest clearance by examining the way in which — Q a , G s ,, Q.(), Q s change with ay/ p y changes in these variables. If any of these quantities fall, forest clearance increases. In Figure Two, such a change is illustrated by a move from QtoQ' and a subsequent change in forest clearance effort from T* to T* . While we focus only on a subset of policy variables, the approach we use illustrates the theoretically hypothesized influence on deforestation rates of a variety of technological and economic factors. A comparative statics approach is common in the theoretical literature concerning smallholder deforestation. Table 2 places the comparative static predictions of our model in the context of other models of smallholder forest clearance. Generally, our predictions are in keeping with other open-economy models of smallholder decision making. Our model differs from other open economy models in that we can easily long a s - ^ - > — Q,, and secondary when— > — — n s , . In Figure One, F * }x = A*} that labor is invested in forest clearance as long a s ------- dTpoo > Q (; and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. generate comparative statics for changes in variables across two periods, especially changes in prices, market integration (through transaction costs), and technology. Figure 2.1 Choosing the Optimal Level of Forest Clearance dA a y / a : 1 A*' A = F Figure 2.2 Choosing the Optimal Level of Forest Clearance Effort dQ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 Table 2.2 Relationship Between Total Forest Clearance and Key Model Variables in Selected Theoretical Models of Smallholder Deforestation wage price ag. tech patience distance subsistence risk current m odel - + + + _\\ + o r - _ I 2 ■ « ■ 2 E A ngelsen (99) M4 - + + + 1 3 N a 14 ® 0 Southgate (90) - + (+) na na N a na « 0 Larson (91) na + o r - + o r - na na N a na s “■ M endelsohn (94) + N a 2 » C 0* 0 0 M l na + na na W « 5 s A ngelsen (95) M2 - + + na 16 N a H 17 O -2- Stryker (76) M2 na + + o r - na c + na D eShazo & (+ or - ) N a DeShazo (95) + na ( - ) na L . A ngelsen (99) + o r - na na f + na >7* J S M2 Open conom lels (ot Angelsen (99) M3 - + na na f N a na M endelsohn (94) + N a S M2 na + na na Dvorak (92) na na - na na + na (J e & ( / > ” 0 A ngelsen (99) M l na - - na na + na 3 Angelsen (95) M l 1 8 na - - na na N a 4- 2.5.1 Forest Clearance and Market Integration An important result of market integration is that transaction costs, especially those associated with distance to and from the market, tend to decline. All things See Angelsen (1999, 201) for a similar table that compares the four modeling approaches used in the agricultural household literature. 1 1 change in distance from village to market center 1 2 risk as reflected in the discount rate 1 3 distance from village center to field 1 4 probability o f losing the land /eviction 1 5 distance from village to market center 1 6 transport costs 1 7 expected future land scarcity (when clearance gives land rights) 1 8 descriptive model Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 being equal, declining transaction costs increase the prices earned for cash crops and reduce the prices paid for market goods. If we think of market integration in terms of transaction costs, the model shows clearly how market integration influences the decision to clear forest or to participate in agricultural production. In particular, when the first order conditions are satisfied, transaction costs affect equilibrium P ] + c' levels of forest clearance in two separate ways. First note that — | — 'f-represents the Pm f n (undiscounted) relative future price of consumption goods at the village level compared to current prices. Declining transactions costs, associated with increased market integration (e.g. decreasing distances to markets or decreasing transport P' +c' costs), cause the ratio — ^— f- to decline (since c 'm falls) thus increasing the Pm +Cm (expected) future buying power of farmers and thus the real value of income derived from future agricultural production. Similarly, note that the ratio of wages to 0 0 w — c agricultural prices,— ^ f-, also declines with increased integration P a ~ c a (since c\ declines), causing the amount of time allocated to forest clearance to rise. Wages foregone during time spent clearing forest in the initial period can be offset 0 0 w —c by higher prices for cash crops in the following period. In a sense, — f- P A ~ C A represents the relative opportunity cost of foregone wages. Individuals will substitute away from wage labor and toward forest clearance labor in the initial Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. period when the relative value of future agricultural production increases. Summarizing these two effects we see that the inverse of their product, 77 19 V + c ' Y w ° - c oVH in 1 ^ m i M W , P° +c° V r m ^ P' - c ' V 1 A A J = 7 , is a good proxy for market integration, where 7 represents the rate of market integration (increasing rates of market integration result in increasing 7 . Rewriting (2.34) through (2.37) in terms of the rate of market integration illustrates this more clearly, ^ - = — [ - + a0y/] = — Q () (2.38) dF0 cry/ rj cry/ dQ 1 8 , 1 ^ - T7 = — [ - + <*/] =— (2.39) dFs p y rj p y dQ = [ - + ad(p\ = Q() (2.40) dTF (m 7 d Q =[- + ar] = ns. (2.41) dTlw 7 Taking the derivative of O. with respect to the rate of market integration we find = < 0 . From Figures One and Two it can be seen that declining Q results in drj rj increased levels of both forest clearance and time spent clearing the forest. Thus, 19 Note that a change in the degree o f market integration in the second period does not affect wages in the first period. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 increased rates of market integration unambiguously increase rates of forest clearance. 2.5.2 Forest Clearance and Impatience The model shows that in our two period world, forest clearance is an investment in future agricultural production and thus consumption. Not surprisingly, the value of this future pay-off will vary across smallholders depending on how much they discount the value of future consumption. The literature on discounting and time preference for consumption is immense and will not be summarized here. Nevertheless, we posit that an important component of an individual’s discount rate, or their time preference for consumption, depends critically on their general level of patience/impatience (see Godoy et al., 1998). The comparative statics of the model show quite clearly that increased levels of impatience diminish the value of future consumption and thus decrease the benefit of forest clearance. Specifically, we see that - — > 0. Once again, from Figures One and Two we see that increased dd rj levels of impatience, increasing 8 , cause the farmer to stop clearing the forest when marginal products of arable land and time are higher than levels chosen by more patient farmers. The result is that impatient farmers clear less land for agriculture, patient farmers clear more. The model also predicts that there is a synergistic effect between the rate of market integration and time preference. The influence of patience on forest Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 clearance is exacerbated by higher rates of market integration. Taking the derivative of Q with respect to impatience and the rate of market integration gives, d(.— ) , — d d l = - ± < 0 . drj rj 2.5.3 Forest Clearance, Technology, and Agricultural Productivity The efficiency with which farmers convert forest to farmland and the productivity of that land in agriculture also influence the optimal rate of forest clearance for the smallholder. Comparative statics can be derived to examine explicitly the way in which tree cutting efficiency on old growth and secondary forests, if/ and y respectively, influences forest clearance. Similarly, we can derive comparative statics that relate changes in forest clearance to different levels of agricultural productivity, subsistence production, and the relative differences in these levels between old growth and secondary forest. While we leave the simple calculus to the reader, we note several interesting outcomes here. Not surprisingly, anything that increases the efficiency with which forest is converted to farmland increases forest clearance. Recall that our production function for forest clearance models the relationship between effort and forest clearance as a dF 1 dF l simple linear relationship F() = < p T °m , F' = yT%R , where = < p, and = y . l'( ) h Changes in the marginal productivity of forest clearance effort can come about Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 through the introduction of chainsaws, logging roads, and tractors. Furthermore, the marginal productivity of effort also can vary across farmers and may depend on relative experience and knowledge of forested areas. Anything that increases the productivity of forest clearance effort will influence forest clearance in two ways. Most obviously, anything that makes it easier to clear the forest increases forest 2 d dQ 2 clearance (— — — = ^ ----- < 0.) Comparative static results for forest clearance dFady/ dy/ effort, however, show that the optimal level of forest clearance effort actually declines, since farmers can clear more forest in less time (— — ----- = ^ 0 > 0 .) dTF O D dy/ dy/ Briefly, it is also interesting to note that a more productive old growth forest, in terms of satisfying the subsistence requirement, lowers forest clearance, - a y / . An increase \n6 increases the opportunity cost of removing “biomass dO productive forest” for agricultural land. Given our approach, higher agricultural productivity unambiguously increases forest clearance. In the model, agricultural productivity is characterized as the output per unit of arable l a n d ( ^ y ^ ) . O f course the marginal output of land itself is a function of many other factors (e.g. irrigation, technology, inputs, experience); we could write this = fn(irrigation, technology, inputs, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 experience). Higher agricultural productivity simply implies that is higher for a given amount of land. From Figure 1, we see that if is everywhere higher for any given quantity of land cultivated, the smallholder will choose to spend more time clearing forest to add to his land. Simply speaking, the payoff per unit area for forest clearance is higher for higher levels of productivity. (A similar comparative static for agricultural productivity measured as a function of labor effort, , also could be developed along the same lines with the same result.) This finding does not depend on relative improvements in agricultural productivity between periods. Because forest clearance comes at the expense of off-season activities, increased intensification of agriculture only affects the return to forest clearance, but not the costs of clearance. A variety of factors could improve agricultural productivity including irrigation, fertilizer, pesticides, biological pest control, or even agricultural extension. Contrary to popular belief, efficient farmers would be even more likely than inefficient farmers to clear the forest. 2.6 Data and Analysis The data described more thoroughly in Chapter 1 is used to empirically test the relationships explored above. The household survey was conducted in 1996 among Tsimane villages of differing degrees of proximity to the closest market town Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82 of San Borja, Bolivia. In all, 209 households in 18 villages are included in the data set. O f particular interest to this research are variables used in the model section above. Direct measures that relate to the amount of old growth and secondary growth forest clearance are provided and related to the model in Table 2.3 below. Table 2.3 Regression Variables V ariable Relation to M odel Description clearfo95 F Total forest cut by the household in 1995 (in tareas, where 1 hectare = 10 tareas) oldgrw95 Fo Total old grow th forest cut by the household in 1995 (in tareas) secgrw95 Fs Total secondary growth forest (fallow ) cut by the household in 1995 (in tareas) dpat (dim pat) 5 Patience (im patience) o f adult respondents (dum m y variable) prcorn95 P ° (corn) Village price o f corn in 1995 (Bolivianos) prrice95 P ° (rice) Village price o f rice in 1995 (Bolivianos) banratio A T / / 0 0 / PA~ CA Village price o f bananas in 1996 over price in 1995 walkSB / A A c (San Borja) H ours walking from the village to the closest m arket center o f San Borja (1996) walkrd c (road) Hours w alking from the village to the nearest road (1996) hunt possibly N p R N um ber o f days o f successful hunting in the past w eek tim es the num ber o f household participants (1996) hhsize influences N Size o f the household in 1996 avgeduc influences w age (w) A verage num ber o f years o f education o f household heads Summary statistics of relevant variables are given in Table 4, where data are based on the responses of heads of households. As described in more detail in chapter 1, the survey villages are located in varying degrees of proximity to the market. Table Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83 2.4 gives distances from each survey village to the nearest market center of San Borja range from less than a one-hour walk to nearly 84 hours (in the dry season.) To estimate impatience, Godoy et. al. (1998) used a method common to social psychologists; survey participants were given the choice of having one candy at the beginning of the administration of the survey or two pieces at the end of the survey. Individuals were classified as patient if they were willing to wait until the end of the survey for candy, individuals with moderate impatience were those that required a greater incentive to wait (e.g. more candy), while individuals who declined to wait were labeled most impatient. Dummies were used to identify very Table 2.4 Summary Statistics V ariable Obs M ean Std. Dev. Min M ax clearfo95 208 13.116 12.955 0 1 1 0 oldgrw95 208 6.640 7.843 0 80 secgrw95 208 6.476 8.281 0 60 dpat 208 0.861 0.347 0 1 dim pat 208 0.183 0.387 0 . 1 prcorn95 208 9.278 3.442 3 15 prrice95 208 9.553 2.977 4 20 ban ratio 189 1.558 0.378 1 2.5 prsoap 173 1.165 0.909 .5 5 walkSB 208 12.103 13.208 .6 84 walkrd 200 2.604 5.906 0 20 hunt 207 1.300 2.333 0 19 hhsize 208 5.625 2.763 1 16 avgeduc 207 1.366 1.832 0 9 patient and very impatient households (dpat and dimpat) and are equal to 1 if 50% or more of adult respondents are patient (impatient) and 0 otherwise. Prices for corn and rice are given as general indicators of the price of agricultural commodities in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 each village. Measures of hunting and household size are included to reflect subsistence factors. Hunting may also indicate familiarity with the forest. Level of education is included as a factor that could potentially have a positive influence on wage. Several variables important to the model were not available explicitly. For instance, the stocks of forest or agricultural land prior to the period of deforestation are not known. Distances to market are known, but changes in these distances for the periods relative to distances before or after the period for which there exists deforestation data are not. These empirical results are presented as a preliminary test of the model. The estimated model is not intended to be a comprehensive model of deforestation. Many variables that would have been useful in the analysis (e.g. fertilizers, pesticides, and credit) were available only for a small subsample of the households sampled. A tobit model, with robust variance estimates, is used for the regressions since forest clearance is censored at 0. Table 5 presents the results of regressing forest clearance in 1995 (total, old growth, and secondary growth) on the explanatory variables summarized in Tables 3 and 4. The results of the model estimation are robust, most coefficients are of the expected sign and many are statistically different from zero (using Huber-White sandwiched estimates of variance.) The estimated coefficients of the old growth forest clearance model are significant more often than the model for secondary growth forest clearance. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 The model predicts that the forest clearance decision should be sensitive to expected rates of market integration via the impact of integration on agricultural and commodity prices. From the start, understand that the survey does not provide straightforward data on market integration nor data on a farmer’s expectations of crop prices. Nevertheless, actual changes in village prices from 1995 to 1996 for Table 2.5 Estimation Results____________________________________ Total Prim ary Forest Secondary Growth ___________________C learance C learance Clearance dpat prcorn95 prrice95 banratio prsoap walksb walkrd hunt hhsize avgeduc constant log likelihood w a ld chi2 left censored uncensored “significant at the 1% level bsignificant at the 5% level Significant at the 10% level drobust standard errors in parenthesis for two important cash crops (bananas and yucca) can be ascertained from the data and therefore be used as a proxy for market integration. Changes in farmgate prices 5 .14b 2.38 3.48b (2.12) (2.01) (1.55) -0.04 -0.57b 0.28 (0.24) (0.26) (0.18) 1.94“ 2.11- 0.90° (0.66) (0.70) (0.490 7.22 9.46b -1.28 (5.33) (4.39) (3.69) -0.77 0.84 -1.50° (0.76) (0.71) (0.89) 0.13“ 0.15- 0.06 (0.04) (0.05) (0.05) -0.09 -0.28° 0.17 (0.16) (0.15) (0.11) 0.89c 0.17 0.86b (0.54) (0.29) (0.38) -0.14 0.4 l c -0.53b (0.27) (0.26) (0.25) -2.21- -2.35- -0.90b (0.69) (0.85) (0.46) -16.13 -26.39b -2.03 (12.94) (11.61) (9.45) -660.53 -487.79 -550.31 36. W ' 23.81“ 31.98“ 7 51 25 166 122 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 (in a way that is insensitive to the units used), are estimated by using a ratio for the price of bananas (P 1996/P 1995). Our reasoning is this: 1) the model assumes that the rate of market integration influences transport costs and thus village-level prices; therefore, whatever the reason for the change in actual agricultural prices at the village, if the model is correct there should exist a relationship between price ratios and deforestation, and 2 ) changes in agricultural prices at the village can be caused by changes in price (via changes in supply, demand or pure inflation) at the market (San Borja) and changes in transaction costs (we assume little or no market for these goods within the village); if this is the case then changes in San Borja would affect all villages equally and differences in price ratios between villages would result only from changes in transaction costs. The estimated models indicate that increasing rates of market integration (the ratio of future and present banana prices) lead to a large and significant increase in forest clearance of old growth areas. The impact of price changes on secondary growth is insignificantly different from zero. Distance to the road has a negative impact on the forest clearance of old growth areas (and an insignificant, but positive impact on cutting in secondary growth areas.) The empirical estimations also show that forest clearance increases with distance from the closest major town of San Borja. One explanation consistent with these findings is that the availability of old growth forest is itself a function of the distance from major developed areas like San Borja. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 Finally, the estimations confirm an important result of the model: patient farmers clear more forest than impatient farmers. While the coefficients for impatience are remarkably similar for forest clearance in both old growth and secondary growth areas, the coefficient on impatience is significant only for secondary growth. As found in other studies, education has a negative impact on forest clearance (perhaps through its influence on wages). The level o f subsistence hunting, possibly an indicator of familiarity with the forest, is positively correlated with forest clearance. 2.7 Policy Implications and Discussion At face value, implications of this research indicate that, ceteris peribus, higher rates of market integration increase forest clearance. To the extent that market integration increases the returns to forest clearance through higher farm-gate prices (due to a decrease in transportation costs) forest clearance is predicted to rise in the analytical model, a result supported in part by the empirical results. Similarly, the model also predicts that improvements in agricultural productivity also increase returns to agriculture and consequently to forest clearance. Finally, both the theoretical and empirical models indicate that deforestation rates are sensitive to smallholder impatience and thus discount rates; lower levels of impatience or personal discount rates lead to increased forest clearance. At the same time, both the analytical and empirical models indicate that market integration may have a dampening effect on forest clearance. To the extent Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 that market integration increases wages, the opportunity cost of agriculture and forest clearance rises, lowering clearance. The empirical estimates also indicate a strong negative relation between education and forest clearance. When market integration improves education, rates of forest clearance are likely to fall (this effect may come through the wage.) Finally, the analytical model also indicates that a more productive old growth forest (in terms of satisfying the subsistence requirement) reduces rates of forest clearance as households forgo agriculture in favor of traditional hunting and gathering. As discussed, extreme poverty and high rates of forest clearance are pandemic among rural smallholders in much of the developing world. Consequently, balanced policies which address both poverty and the environment are increasingly sought after by policymakers. The results of this research indicate that the effect on the environmental resource base of a policy encouraging market integration is largely dependent on prevailing incentives. To this extent, policymakers would be wise to consider the particular institutional context of a given state or region. Flouseholds in regions with a relatively strong labor market may face different incentives relative to more remote households. At the same time, market integration policies that also seek to enhance education, improve the productivity or returns to forest biomass, and strengthen the labor market will have a tendency to counterbalance the increase in forest clearance by increasing the opportunity cost of forest clearance. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89 Chapter 3 A Co-Management Model of Timber Resources 3.1 Introduction The preceding chapters have used empirical and analytic methods to explore different dimensions of the rural household using a modified agricultural household model. In this chapter, a general equilibrium model is used to further investigate the linkages of the rural household to their environment. Broadly, this chapter will explore the co-management of resources between local user groups and an outside authority (state governments, private corporations, etc.). Co-management, as such, is a relatively recent approach to rural resource management, anticipated by some to be a mechanism capable of promoting both quality of life enhancements for the rural poor and sound environmental stewardship. Under co-management regimes, communities receive greater autonomy in managing local natural resources under varying degrees of supervision. An appealing aspect of the approach is that the strengths inherent in the property regimes of private ownership, direct state control, and communal property are combined to creatively manage the environmental resource base (Baland and Platteau, 1996). Current programs range from large game wildlife management in Africa (Bulte and Horan, 2002), to Japanese fisheries (Baland and Platteau, 1996; Kenneth, 1989), to forests in Mexico (Klooster, 2000; Munoz-Pina, de Janvry, and Sadoulet, 2002), India (Richards, 2000; Kumar, 2002), and Nepal (Agrawal and Ostrom, 2001; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 Edmonds, 2002), to management of all village resources in Burkina Faso (Baland and Platteau, 1996). The U.S. Federal Government has also experimented with versions of co-management among Arctic Alaskan communities with respect to select marine mammals and large game. A good deal of descriptive literature has been written on the viability of such programs, and particularly on the Joint Forest Management program instituted by India. As will be discussed in more detail, research generally finds that local or joint management of forest resources leads to larger efficiency gains relative to pure State management in certain contexts. At the same time, however, many studies have less promising news regarding the effect of such schemes on rural poverty, where it is argued benefits often accrue to a local elite (Kumar, 2002; Klooster, 2000). Nevertheless, co-management appears to be a viable mechanism through which scarce resources may be more effectively managed, and, with careful forethought, the rural poor empowered. The purpose of this chapter, then, is to contribute to the growing body of literature exploring co-management of local resources through the development of an analytical model. In particular, a general equilibrium model is used to investigate the linkages between the local Forest Community, the Forest Department, and a Residual Sector under a co-management arrangement. We focus on the general flows of resources between these sectors under the benchmark case of exclusive State management versus a co-management approach with both a direct payments and in- kind incentives, wherein the Forest Community receives both wage payments and a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 share of forest biomass. The model is then simulated under various assumptions of income transfers and preferences, and while this approach does not explore group heterogeneity and its influence on the distribution of benefits and the associated optimal regime, it does capture the effect of different incentives and the degree of local dependence on the resource base. The model therefore allows the utility of the Forest Community, the Residual Sector, and the Forest Department to be compared under different degrees of “dependence” and a variety of co-management arrangements. After explaining the model’s background, and its implications, section three develops the framework used in the simulations and presents the results. The chapter concludes by discussing the implications of the model and simulations and by comparing them with the existing literature. 3.2 The Model 3.2.1 Background to the Model In 1990 the Indian Central Government mandated that State governments formally adopt Joint Forest Management (JFM) as the primary mechanism through which the State managed and owned forest resources. The policy was reportedly motivated by a desire to both reduce environmental degradation (which, according to Kumar (2002), the Central Government attributed largely to local communities using Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92 the forests as de facto open access property) and to reduce rural poverty. States, however, were left with a great deal of flexibility with respect to the particular approach they would adopt. Incentives strategies offered by various State Forest Departments to local village forest communities range from wage payments for protective labor services, to in-kind and revenue shares of non-timber forest products that are collected, to revenue shares of timber sales, to combinations of each, in the 26 of 28 States that have formally adopted JFM (Kumar, 2002). While there is less direct government involvement, a similar form of co management was recently adopted in Nepal. Due to increasing rates of forest clearance and growing environmental degradation, the Nepalese government began a process of transferring ownership and control of all forests to local communities or “Forest User Groups.” The central government provides the user groups with both the framework and resources necessary to reduce resource extraction (Edmonds, 2002). In contrast to the Indian case, however, user-groups in Nepal receive a greater share of the return from successful management in that land is held as common property by the village. Limited empirical evidence appears to support the hypothesis that forest resources are managed more efficiently under Joint Forest Management schemes relative to pure central management. In an excellent empirical study of such programs, Edmonds (2002) tests the robustness of relatively lower mean levels of resource extraction in Nepalese forests managed by “Forest User Groups” relative to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 areas managed purely by the central government. Using several different estimation techniques, he finds that the difference is indeed robust, supporting the view that Nepalese forest communities are able to effectively manage resources more efficiently relative to communities where resources are managed solely by the central government. Consistent with Edmond’s (2002) findings, Kumar (2002) finds similar evidence in India. While he argues that the distribution of benefits from Joint Forest Management among the Forest Community are highly unequal (a rural elite capturing most of the economic benefits) and that much of the gain in lower resource extraction comes at the expense of the most poor, he reports that extraction of timber and non-timber resources are lower under JFM relative to non-JFM communities. Successful examples of co-management have also been identified in Mexico. As a result of land reform that followed the 1910 peasant-led revolution in Mexico, roughly 80% of Mexico’s forests are currently held as de jure common property (Klooster, 1998). However, until legislative changes in the 1980’s, local communities did not have the autonomy to collectively manage timber resources but were required to contract with state approved commercial logging companies. The amendment allowed communities to form cooperatives to harvest and manage logging operations under specified criteria, a context akin to the Nepalese case given that communities both own and manage resources with considerable State oversight. As a result of the legislative changes, several successful examples of community co management have emerged in Mexico since the 1980’s. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94 In outlining the institutional differences between successful and relatively unsuccessful community programs in Mexico, Klooster (1998) finds that in seven of the eight cases, community managers have been successful in increasing forest area and, in contrast to the Indian case, maintained generally equal distribution of benefits to community members. “Successful” communities in the Klooster (2000) studies were indigenous, ethnically homogenous communities. Consistent with these findings, Kumar (2002), Kant (2000), and Kant and Berry (1998) argue that group heterogeneity may result in a less efficient outcome both in terms of natural resources and income distribution. That is, shared institutions at the community level reduce the degree of moral hazard and adverse selection therein serving as an important element in the stability of the co management scheme. Homogenous groups are more likely to share common goals and values with respect to subsistence harvest amounts, enforcement mechanisms, and the distribution of benefits. To the extent that roughly 26% of the Amazonian forests are recognized (either officially or as historic territory) as Indigenous Territory (Richards, 1997), the co-management framework may be prove to be particularly appealing to South American indigenous peoples.1 A second element, also identified in the literature as important to the success of a co-management regime, is the degree of dependence by the user group on the resource base in question (Kant, 2000; Kant and Berry, 1998). Groups highly dependent on non-timber forest products, for example, are likely to have strong 1 See Rangan and Lane (2001), Becker and Leon (1998), and Chase Smith (2000) for examples. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 incentives to cooperate with the State or some other entity in managing the forest to achieve an “optimal” harvest level, one that meets subsistence needs. Consistent with these aspects of a successful regime, the incentive mechanisms selected by the State can be a final element important to the success of co-management programs. That is, given that a particular forest area is held by the State, the central government must decide on the degree of new local ownership or management, in addition to the community compensation for cooperation and enforcement of a particular plan. In the case of Nepal, select communities obtained ownership to village land and the government heavily subsidized enforcement costs. While communities in village India do not, as a rule, receive common property rights over local forests, State governments have flexibility in developing unique incentive schemes which promote the realization of the Forest agreement between the State and the local community. Such arrangements in India include a share of profits from harvested forest, direct wage payments for enforcement effort, and /or a share of forest biomass. As argued by Richards (2000) and modeled by Kant and Berry (2001, 1998), and Kant (2000) optimal resource allocation strategies may differ significantly on a continuum from pure private ownership, to State control, to open access regimes. That is, even within a state or province community incentives may vary with respect to the type of land tenure, institutions, and goals for a particular village or region. In the language of Kant and Berry (1998), for instance, a user group in region A may be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 quite heterogeneous but very reliant on the natural resource base, whereas the neighboring user group in region B may be ethnically homogenous but not dependent on the resource base. The incentives and type of management arrangement, therefore, selected by the government will be critical in determining the outcome of a particular policy in these different regions. As will be discussed in further detail, it has been argued that group homogeneity and dependence on the resource base correspond toward more community management approaches whereas heterogeneity and independence correspond with more private property arrangements (Kant and Berry, 1998). The research in this paper is therefore intended to extend the inquiry into the important linkages between a local community, the central government, and a residual or urban sector. In particular, a general equilibrium framework is used to capture the major features of a prototypical co-management program. The model highlights the interactions between a Forest Community, a Forest Department, and an urban or Residual Sector. As such, it traces the effects of price and incentive changes brought about by changes in exogenous parameters on resource management in the Forest Community. With respect to incentive schemes, the model provides a degree of flexibility. We assume that the Forest Department may use both direct wage payments and a “transfer” 2 of biomass production to the Forest Community as a payment for participating in the co-management scheme. The 2 Under joint management, the Forest Community may remove a certain percentage o f biomass production without penalty - forest products removed beyond the set amount are subject to fines. Both the fine and the percentage o f removal allowed are flexible. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 model also captures in a general way the Forest Community’s dependence or valuation of the forest. 3.2.2 The Three Sectors Figure 3.1 illustrates the linkages between the Forest Department, the Forest Community, and the Residual Sector. As detailed in the analytical model that follows, the figure presents the basic flows of goods, services, incomes, and expenditures in the model. Briefly, the Forest Community provides protective labor, , to the Forest Department, and labor for production of the market good, N s r , to the Residual Sector in exchange for payments a and w , respectively. Labor income, in addition to government transfers, Gc, flow out of the Forest Community back to the Residual Sector in the form of payments for good X , PxX c , and to the Forest Department in terms of fees for excess gathering, PfF(N df). Income flows into (out of) the Forest Department (Residual Sector) through sales (purchases) of timber, PfFr. The central government also provides transfers to the Residual Sector and Forest Community through taxes on Residual Sector profits and Forest Department timber revenues. Gathering by the Forest Community, N df, or sales by the Forest Department reduce the size of the forest and so impose a general negative externality on each sector. Biomass production by the Forest Department increases the size /quality of the forest and is therefore a positive externality for each sector. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3.1 A Circular F lo w Diagram 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 3.2.2.1 The Forest Community Utility in the Forest Community is a function of goods collected from the forest, Fdf (which may be thought of as either fuel-wood, non-timber forest products, (NTFP’s), or timber), government grants, Gc (which are often in-kind and may be used for infrastructure improvements),3 consumption of the market good produced in the Residual Sector, X c, and a “negative externality” effect that captures forest degradation. The externality, FI;, is a function of Fdf and Fr , where Fr is the amount of the forest good consumed by the Residual Sector, Fe = Fh (Fdf > F r ) > Fm > 0 ’ F,n 1 > 0 > FU 2 > FE 22 > 0 • The Forest Community’s utility can therefore be written as, U, = UAF„,G',X,)-F,:(F„r F,) (3.1) where X c and Fdf are choice variables. Fdf is simply a function of time allocated to removal of timber resources by the Forest Community (which may be legal or illegal), N df, F„=F„(NJr),F0>O,F^<O (3.2) Consumption spending on market goods, PxX c, is constrained by wage earnings from the Residual Sector, wNs r , plus income earned from the Forest 3 Apart from government in-kind transfers to the Forest Community, some Indian State governments mandate that a certain percentage o f profits earned by Village Forest Committees be allocated to community development projects (Kumar, 2002). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 ps Department, a (N t + — L -), (which can be thought of as labor payments for protective c services and payments for timber removal). The Forest Community is also forced to pay for removal beyond that stipulated by the government agreement, FA (which can be thought of as a percentage of the biomass produced), where total payments are equal to, ijPf (Fdf (Ndf) - FA) , where rjPf (Fl)l< (N/If) - FA) > 0. The budget constraint is therefore, PxX c+j1 Pf (FD I,,(Ndf) - F A) = w N ;+ a(N d dl+ ^-), w here/' =1 (3.3) ■ ' c The rj term in equation (3.3) is a proxy for the effectiveness of enforcement of a particular policy. When a policy is perfectly enforced, 7 = 1, and the Forest Community pays for all harvesting beyond an agreed upon amount. When 7 = 0, the Forest Community faces no penalty for removal of timber. Similarly, a lower (higher) value of FA is associated with a higher (lower) penalty for forest clearance. A clear advantage of the co-management framework discussed in the literature is that enforcement mechanisms under JFM may be stronger because of shared local institutions (which are positively correlated with group homogeneity). Under co-management arrangements, the government may grant local communities complete autonomy in choosing the means of achieving the government objectives (and, as in the Nepalese case, the government often provides significant funding for enforcement of “Forest User Group” objectives) in addition to providing incentives Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 in terms of harvest shares and sometimes wages for protective labor. These incentive effects are partly captured by setting 7 7 = 1. Without enforcement, 7 7 = 0, and the Forest Community is not constrained in terms of the amount of extraction. The case of no enforcement is akin to Mexico, India, and Nepal before co management. As discussed, in Mexico, while land was technically de jure common property, the combination of a lack of enforcement and an absence of local autonomy over village resources led to high levels of abatement. In Nepal, ownership of forests was transferred to the government in 1957, however, enforcement mechanisms were not able to curb forest degradation and, in fact, deforestation and degradation probably increased under government control.4 It is assumed, then, that under direct government control 7 7 = 0; that is, de jure State control and de facto open access.5 Labor time supplied by the Forest Community to the Residual Sector and Forest Department is constrained by total time. Time is allocated to JFM . . . . F supervisory and harvest activities (AC.,— ) , Residual Sector labor (AC), and in c collection of fuel-wood or non-timber forest products (Ndf) , Nm r = K , + ^ + K + N il (3.4) 4 Edmonds (2002) citing Pal it (1996). 5 A final argument for assuming a higher level o f enforcement under more local management is simply that collective action becomes much more feasible as group size decreases - see Mancur Olson’s forward in Baland and Platteau (1996). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 3.2.2.2 The Residual Sector Households in the Residual Sector seek to maximize utility, which is a function of the consumption of the forest good purchased from the Forest Department, F? , consumption of the market good, X r , the government grant, Gr , and the “negative externality,” which is associated with forest degradation. Utility in the Residual Sector can therefore be written as, where choice variables are Fd and X r ; and, Ur is increasing in all arguments. Spending on the consumption good in the Residual Sector, PxX r , and the forest good, PfF d , is constrained by after-tax profits earned in the production of X, (PxX - wNd r )(1 - 1 ) . The budget constraint for the Residual Sector is therefore, 3.2.2.3 The Forest Department The final stakeholder is the Forest Department. Under the benchmark case of exclusive Forest Department control it is assumed that the objective is simply to maximize profits. The Forest Department’s costs of protective labor and timber (3.5) Pf F d +PxX r ={PxX - w N d . \ \ - t ) (3.6) where the production of X is simply a function of labor demand, X = X ( N d),X ' <0,X" < 0 (3.7) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 F s removal services, a (N t + - 1 - ) , used to produce biomass must be met by income c derived from a government grant, Gf , plus timber sales,PfF* (of which the government keeps a share, ( j ) , to distribute as grants), less any fixed costs, H. The majority of timber sales are domestic, Frd, but some timber is exported, F x. While exports earn a price Pfx it is assumed that the residual (Pfx - Pf )F* is kept by the central government and distributed as a government grant. Forest department profit is therefore given by, Gf + (1 - < /> )P fF; - a (N d. + Z - ) - H (3.8) c where H are fixed costs of production. In contrast, under JFM it is assumed that the Forest Department chooses the optimal amount of forest good to supply and biomass to produce to maximize its utility. In particular, the Forest Department’s utility is considered to be increasing in forest biomass, revenue from the sales of timber, and the government grant. Furthermore, as with the Residual Sector and the Forest Community, the negative externality now influences the Forest Department’s optimal choice of F* and N d f . Production of forest biomass is a simple function of “protective labor” provided by the Forest Community, N d dj, FB=FR{Nd dj), where, Fb >0,F'b <0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 and the negative externality is increasing in Fdf and F* The Forest Department therefore seeks to maximize, (3.9) where N d d J and F* are choice variables. The Forest Department’s costs of protective labor and timber removal services, are identical to the direct control case, however, the Forest Department also receives revenue from removal of timber beyond the agreed upon amount. The corresponding budget constraint is therefore, 3.2.2.4 Government Grants The grant from the Government received by each stakeholder is a fraction of the tax receipts from the Residual Sector, (PxX - wNr )(1 - I) , plus a fraction (j) of (Pfx - Pf )Fr x. The total tax revenue, therefore, collected by the Government is G, + (1 - 4,)P,f; + Pf K F „ - F i,) = a(N]l +P-) + H (3.10) timber sales.6 As discussed, the Government also receives the revenue from exports, (1 - t){PxX - wNr) + (PJ X - Pf )F; + (1 - < fi)P f Fr (3.11) 6 Government grants, while endogenous to the model, are exogenous to each sector. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 The Forest Community receives a share f t , the Residual Sector y , and the Forest Department receives the residual, (1 - ( 3 - y ) . 3.2.2.5 Market Clearins Equations For simplicity, it is assumed that markets clear for labor, N d r =N*r (3.12) good X , X{N r) = X c+ X r (3.13) and for the forest good (sold by the Forest Department) F; = F? (3.14) The total amount of forest good is equal to a fixed stock plus additions due to protective labor services less sales of timber and removal by the Forest Community, FTor=F + FH{N4j) - F r - F df (3.15) ^ F - F r0T=Fr+ F(lf- F B{Ndj) where F > Fmr . The time constraint in equilibrium is given by Nr< n = N * + N Jf + K + — 13.16) c Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 3.2.3 Constrained Optimum by Sector 3.2.3.1 The Forest Community The Forest Community’s utility maximization problem can be written as, max„ Uc(Fdf(Ndf),Gc,FR( N J - X R) - F E(Fdf(Ndf),F;) (3.17) /v < / / r »'vdi Cl/’",. ,M( /j s.t. K , + — + K + N , , - N ror=0 (3.18) c F + F„{Nl)-Fr-F„(.Nt ) - F m = 0 (3.19) wN: + a(N-j + 5 - ) + np, (F„ (N4 ) - Fa) - F„(N'r) + X , = 0 (3.20) Combining (3.18) and (3.20) and substituting into (3.19) for A (); allows for the following Lagrangian function, L = Uc(F4f(N< (f),Gc,FR( N J - X R) - F ^ i N ^ F ; ) (3.21) -A, [F„ (N*j ) + F - F r - F N(Ndf) - Fror ] where AT* = --------- ^----------------------------------------------------------------------------------- d j (l + «) Solving this system for the choice variables Ndf and N s r yields Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 (1 +a)Ul f K( K ,).™ = ------------- 7 ----------------------------- ; - (3.22) F „ ( K ) ’ F , w ( N d, Y lU ' F *)l F I N " W W f - — * — 1 blANr< and rewriting in terms of Ndf gives F„ (K„ ).»,< = ------------- r C3.23) (if - if + ul[tjp. - — - f „ ( n ; ) ' f b ( n - , ) ' With specific functional forms, (3.22) and (3.23) are used to solve for optimal values of Njj , N df, N s r , and X c under joint forest management. If, as previously discussed, it is assumed that r/ = 0 under direct government control, the marginal products of protective labor and forest clearance in the benchmark case are, respectively, F»(K,),x = ------------------ (1 + ° > ,;( L -------- 773---- (3-24) F„(K) FD I,(Kr ) 77----- ^ , , . , - 7 777^ (3'25) 1 _ , l + w - F R(Nr) _ Uc(l + a) c /; K ( K ) K ( K ) Comparisons between the direct control outcome (DC) and Joint Forest Management case (JFM) are largely dependent on functional forms and specific coefficient values selected. Assuming that there exists increasing returns to protective labor, Fb'(Nj.) > 0,Fh"(Njj) > 0, and decreasing returns to forest clearance, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 FDF(Ndf) > 0,FD / ( N df) < 0, (as assumed in the simulations), in order for the Forest Community to allocate more time to protective labor and less time to clearance it should be the case that FB(Ndj)JF M >FB(NdJ)nc and FD F(Ndf),F M > FDF(Ndf)lx.. The simulation results also provide further insight into equilibrium values of Nd j and Ndf by treating rj as a continuous variable between zero and one, comparing direct control with co-management under varying degrees of enforcement. The first order conditions also illustrate that utility for the Forest Community is dependent on the functional form and the specific coefficient values placed on parameters in the utility function. Although utility is increasing in the amount of the forest good, FB, and decreasing in the amount of the externality, Fi;(Fdf (Ndf -),F*), it is also increasing in Fd f . 3.2.3.2 The Residual Sector The objective of the Residual Sector is to max (3.26) s.t. A ,. + P,F;j - (1 - t)[FK ( < ) - wN1 ' ] = 0 (3.27) X , + X c- F R(Nd r ) = 0 (3.28) (3.29) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 Using (3.28) to substitute for X r and combining (3.27) and (3.29) to substitute for results in the following unconstrained maximization problem: maX< Uc(F?\Gc,FR{ N i ) - X c) - F ,(F # , 0 (3.30) where, c(Xe-tFR{Nd r ) - { \ - t ) w N t - N dj- N df- N : + Ntot) F?' = ■ Pfc + 1 Optimizing (3.30) with respect to the choice variable AUyields, (3.31) tc(F?-U'r) + Ui(Pf c + i) Labor demand in the Residual Sector is unaffected by 77and the Residual Sector’s decision calculus for the optimal amount of labor is the same under JFM or direct control. 3.2.3.3 The Forest Department The final stakeholder is the Forest Department. Under the benchmark case without JFM we assume that the Forest Department is simply interested in choosing F* and N d d j to maximize profits, max; . v (Gf + (1 - (/> )P f F* +?1 Pf (Fdf - FA)- a (N ^ + - ^ ) - / / ) (3.32) s.t. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110 N J j, + — + N- +N* - N tot=° < 3-33) c F + FB( N ^ ) - F ; - F df- F roT=0 (3.34) Combining constraints (3.33) and (3.34), and writing in terms of F/results in the following unconstrained maximization problem m ax,;,(G/ + ( l - « / > , ^ : " + ^ , ( F „ - F A) - a ( N ^ + F - ) - H ) (3.35) where, c ( - N i - N , „ - N r + Nm r- F ^ + F ,,- F + Fmr) F' = _ 0 ^ ^ ' Taking first order conditions with respect to the choice variable N d d j results in, , a - Pf (1 - (b ) F M , ) n c = (3.36) P f Q - t ) - - c In contrast, under JFM, we assume that the Forest Department derives utility from biomass production and disutility from the externality. The Forest Departments objective then is to m a (3.37) s.t. K +F + Nr +Ndr- N m,.=0 (3.38) G, + (1 - 4,)P,f; + nP,(F4 - F A)-a(N^J + F ) . H = 0 (3.39) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I l l F + FH { N ^ - F; - F df- F rOT= 0 (3.40) Combining constraints (3.38), (3.39) and (3.40) and substituting for F)'yields the following unconstrained maximization problem: As evident in equation (3.36), demand for protective labor in the Forest Department under the profit maximization case is simply a function of the prices and the shares of revenues collected by the government. In contrast under JFM, demand for protective labor is a function of prices, share parameters, and preferences. In addition, note that the share parameter does not affect the Forest Department’s decision with respect to hiring protective labor. Similar to the Forest Community, direct comparisons between equations (3.36) and (3.42) are largely contingent on parameter values and functional forms selected. Again, assuming increasing returns to labor in biomass production, in order for biomass production to be higher under co-management we should observe (3.41) where c ( - N j ( I - a ) - N df - N r+ Nmr - G, - r,Pf ( f y - FA) + H - FB( N ^ + F , — F + Fmr) (1 + c (l- < f> )P f - a - c ) Solving for the remaining choice variable, N d dj, yields: (1 -a )(U 2 f Pf ( l - t ) - F Z ) (3.42) d j > . i m Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 3.3 Simulations and Discussion The results of simulating the analytical model are described in this section. A CES utility function is assumed for each sector and production functions for goods X , Fdf, and biomass are assumed to follow a standard Cobb-Douglas functional form. For production o f X , marginal product is assumed to be decreasing in labor provided to the Residual Sector, F(N r) > 0, F(N r) < 0, and F(N df)' > 0, F(Ndf) < 0 in forest clearance labor; whereas, biomass is an increasing function of protective labor, F(Nd /.) > 0, F{Ndj) > 0. FI: is a simple linear function. The specific functional forms and initial values are given in Appendix 1 and the GAUSS program used in the simulations is given in Appendix 2. To compare the two cases, competitive equilibrium prices and quantities are generated under the benchmark case of exclusive state management (where rj = 0) and the co-management regime (where 7 7 = 1). Associated policy outcomes are compared for changes in Forest Community preferences (/?', /?( 2, (3) in Appendix 1) and the elasticity of substitution, 6 . Finally, policy outcomes are compared within JFM under various levels of enforcement (0 < 7 7 < 1) and for changes in FA, the share of forest good the Forest Community may harvest without a penalty (as a percentage of biomass). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.1 Simulation Results ¥ Market Good C om m ercial Tim ber Protective t- • ^ 1* 7 • i_i U tility » » « .* . T , A E nvironm ental V ariables Production Sales Labor M arket fil Uc Ur Uf N r x Pf F ; a N d j Fdf FB(N dj) F™ F*s 0.400 175.9573 185.9192 — 0.999997 12.26762 94.40923 1 .6 9.642232 0.999961 15.20903 100.2962 9.938462 102.5264 102.3486 0.450 216.6494 187.2268 - 0.999998 12.39402 95.00216 1 .6 9.740253 0.999961 15.22504 100.1977 9.937903 102.4901 102.2659 0.500 333.129 313.2894 - 0.999999 21.53099 158.3999 1.620592 23.84846 0.999961 19.74325 89.45395 13.3024 100.5141 93.11936 0.550 369.1987 316.8928 - 1 .0 0 0 0 0 1 21.85858 160.1129 1.620592 24.17744 0.999961 19.81715 89.15821 13.33565 100.4253 92.87292 0.600 407.3027 316.9177 - 1.000002 21.86114 160.1243 1.620592 24.17935 0.999961 19.81741 89.15624 13.33559 100.4245 92.87128 0.650 450.4358 316.9276 - 1.000003 21.86216 160.1288 1.620592 24.18011 0.999961 19.81751 89.15545 13.33556 100.4242 92.87063 0.700 499.4278 316.9344 - 1.000004 21.86286 160.1319 1.620592 24.18063 0.999961 19.81759 89.15491 13.33554 100.424 92.87018 0.750 555.2608 316.9358 - 1.000006 21.86301 160.1326 1.620592 24.18074 0.999961 19.8176 89.1548 13.33554 100.4239 92.87009 0.800 619.1185 316.9385 -- 1.000007 21.86329 160.1338 1.620592 24.18095 0.999961 19.81763 89.15458 13.33553 100.4239 92.86991 O © c 0.400 423.6955 976.3434 480.3218 0.999992 74.08253 481.6641 1.509419 72.02311 2.213592 25.89734 45.91349 17.9366 84.61663 68.12566 0.450 455.9511 976.3662 480.3201 0.999992 74.09192 481.719 1.509419 72.02422 2.213592 25.89381 45.9108 17.93502 84.61399 68.12355 0.500 459.0385 976.3697 480.3199 0.999992 74.09337 481.7275 1.509419 72.02439 2.213592 25.89327 45.91039 17.93478 84.61359 68.12322 0.550 462.1533 976.3702 480.3198 0.999992 74.09359 481.7288 1.509419 72.02442 2.213592 25.89319 45.91032 17.93474 84.61352 68.12317 0.600 465.2947 976.3703 480.3198 0.999992 74.09362 481.729 1.509419 72.02442 2.213592 25.89317 45.91031 17.93473 84.61351 68.12316 0.650 468.4629 976.3703 480.3198 0.999992 74.09363 481.729 1.509419 72.02442 2.213592 25.89317 45.91031 17.93473 84.61351 68.12316 0.700 471.6582 976.3703 480.3198 0.999992 74.09363 481.729 1.509419 72.02442 2.213592 25.89317 45.91031 17.93473 84.61351 68.12316 0.750 474.8807 976.3703 480.3198 0.999992 74.09363 481.729 1.509419 72.02442 2.213592 25.89317 45.91031 17.93473 84.61351 68.12316 0.800 478.1308 976.3703 480.3198 0.999992 74.09363 481.729 1.509419 72.02442 2.213592 25.89317 45.91031 17.93473 84.61351 68.12316 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.1 (cont.) Simulation Results Utility M arket Good Production Com m ercial Tim ber Sales “Protective” L abor M arket E nvironm ental V ariables uc Ur Uf W K P f K a F* FB(Ndj) p F D 1 E p R S 1 E 0.400 423.3504 946.9969 467.9694 0.999992 71.72093 467.8228 1.514695 69.78544 2.259194 25.64987 47.96213 17.74757 85.37022 69.26409 - * ■ > 0s- o 0.450 457.4622 947.017 467.9681 0.999992 71.7295 467.8731 1.514695 69.78647 2.259194 25.64666 47.95968 17.74616 85.36784 69.26212 c o £ 1 ) 0.500 462.2664 947.02 467.9679 0.999992 71.73078 467.8805 1.514695 69.78663 2.259194 25.64618 47.95932 17.74595 85.36748 69.26183 © C D 1 0.550 467.1348 947.0204 467.9679 0.999992 71.73097 467.8817 1.514695 69.78665 2.259194 25.64611 47.95927 17.74592 85.36743 69.26178 s c e -J 0.600 472.0676 947.0205 467.9679 0.999992 71.73099 467.8818 1.514695 69.78665 2.259194 25.6461 47.95926 17.74591 85.36742 69.26178 O' O s 0.650 477.0658 947.0205 467.9679 0.999992 71.731 467.8818 1.514695 69.78665 2.259194 25.6461 47.95926 17.74591 85.36742 69.26177 s * © 1 ! 0.700 482.1303 947.0205 467.9679 0.999992 71.731 467.8818 1.514695 69.78665 2.259194 25.6461 47.95926 17.74591 85.36742 69.26177 c uj 0.750 487.2621 947.0205 467.9679 0.999992 71.731 467.8818 1.514695 69.78665 2.259194 25.6461 47.95926 17.74591 85.36742 69.26177 0.800 492.4625 947.0205 467.9679 0.999992 71.731 467.8818 1.514695 69.78665 2.259194 25.6461 47.95926 17.74591 85.36742 69.26177 0.400 420.8134 892.4889 444.7875 0.999992 67.33141 441.9732 1.524309 65.64891 2.356063 25.17947 51.74133 17.39024 86.7584 71.3806 6s- o 0.450 458.1314 892.5082 444.7866 0.999992 67.34019 442.0251 1.524309 65.65 2.356063 25.17619 51.73884 17.38884 86.75599 71.3785 s 4 > £ I I 0.500 466.068 892.5109 444.7865 0.999992 67.3414 442.0322 1.524309 65.65015 2.356063 25.17574 51.7385 17.38864 86.75566 71.37821 © O X ) I 0.550 474.1781 892.5113 444.7865 0.999992 67.34156 442.0332 1.524309 65.65017 2.356063 25.17568 51.73845 17.38862 86.75561 71.37817 c 6S —I 0.600 482.4655 892.5113 444.7865 0.999992 67.34159 442.0333 1.524309 65.65017 2.356063 25.17567 51.73844 17.38861 86.75561 71.37817 00 O N 0.650 490.9348 892.5113 444.7865 0.999992 67.34159 442.0333 1.524309 65.65017 2.356063 25.17567 51.73844 17.38861 86.75561 71.37816 c * © I I 0.700 499.5907 892.5113 444.7865 0.999992 67.34159 442.0333 1.524309 65.65017 2.356063 25.17567 51.73844 17.38861 86.75561 71.37816 c C d 0.750 508.4382 892.5113 444.7865 0.999992 67.34159 442.0333 1.524309 65.65017 2.356063 25.17567 51.73844 17.38861 86.75561 71.37816 0.800 517.4825 892.5113 444.7865 0.999992 67.34159 442.0333 1.524309 65.65017 2.356063 25.17567 51.73844 17.38861 86.75561 71.37816 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3.1 (cont.) Simulation Results T T ,..., Market Good Commercial Timber “Protective ^ A . , 7 . , , Utility j ^ i t l ™ i x Environmental Variables J Production Sales Labor Market Pi Uc Ur Uf Nr X Pf F; a Nd } Fdf FB{Ndj) Ffd F c r s 0.400 422.4644 921.5201 457.17 0.999992 69.66934 455.7614 1.519204 67.84886 2.28126 25.43178 49.7328 17.58165 86.02105 70.25336 0.450 458.1143 921.5391 457.1691 0.999992 69.67778 455.8111 1.519204 67.84992 2.28126 25.42862 49.73039 17.58031 86.01875 70.25137 0.500 464.3928 921.5411 457.1693 0.999992 69.67895 455.8179 1.519204 67.85011 2.28126 25.42818 49.73006 17.58017 86.01847 70.25106 0.550 470.7801 921.5396 457.17 0.999992 69.679 455.8177 1.519204 67.85023 2.28126 25.42816 49.73005 17.58027 86.01856 70.25096 0.600 477.278 921.5391 457.1702 0.999992 69.67889 455.8175 1.519204 67.85023 2.28126 25.4282 49.73008 17.58031 86.0186 70.25097 0.650 483.8883 921.5391 457.1702 0.999992 69.67886 455.8174 1.519204 67.85023 2.28126 25.42821 49.73009 17.58031 86.01861 70.25098 0.700 490.6134 921.5391 457.1702 0.999992 69.67886 455.8174 1.519204 67.85022 2.28126 25.42822 49.73009 17.58031 86.01861 70.25098 0.750 497.4557 921.5391 457.1702 0.999992 69.67886 455.8174 1.519204 67.85022 2.28126 25.42821 49.73009 17.58031 86.01861 70.25098 0.800 504.4177 921.5391 457.1702 0.999992 69.67886 455.8174 1.519204 67.85022 2.28126 25.42821 49.73009 17.58031 86.01861 70.25098 116 Values of coefficients under the changes described above are listed in Table 3.1. The straightforward comparison between the benchmark case and co-management with perfect enforcement appear to be consistent with the analytical results previously described. Under co-management, the Forest Community allocates less time to biomass gathering and more time to protective labor (and consequently there is relatively greater biomass production). The negative externality effect is also lower for all three sectors under co-management. Sales of harvested timber are relatively higher under co-management, but the combination of sales and illegal gathering as a percentage of biomass production is less under co-management. When analyzing co-management under strong and weak enforcement, we find that the results of the simulations in Table 3.1 are consistent with the assumptions made in section 3.2. Namely, while the equilibrium level of protective (forest clearance) labor is higher (lower) under co-management than under the pure state control regime, the levels decrease (increase) as enforcement under co management declines. Similarly, as the share of biomass removal stipulated under the JFM agreement increases for the Forest Community, levels of protective labor (forest clearance labor) decrease (increase). Because protective labor decreases with a reduction in enforcement levels, biomass production also falls, as indicated in Table 3.1; likewise, with an increase in time allocated to gathering , the negative externality, FI: (Fd[ (Ncl/), /(;'), increases. Notice, however, that in all co- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 management scenarios both biomass production and the negative externality are improvements relative to the benchmark case. Holding all other variables constant and increasing the Forest Community’s dependence on the resource base, or the weight of Fdf (f l in Appendix 1) in the Forest Community’s utility function, strongly affects the Forest Community’s utility under different regimes, and will be discussed in more detail in light of the existing literature. In brief, when valuations of Fdf (and consequently dependency) are low, the Forest Community’s utility gain from the additional “protective labor” income and the positive externality effect under co-management outweigh any utility loss from reductions in time spent gathering under the benchmark case. That is, Forest Community utility is higher under JFM for low levels of valuation, p \ . However, as the weight of Fdf rises and the Forest Community is increasingly characterized as highly dependent on the resource base, Forest Community utility is higher under the benchmark case (de facto open access wherein timber removal is penalty free) in contrast to JFM (where the Forest Community is penalized for any excess removal of timber). Surprisingly, however, increasing the weight of Fdf in the Forest Community’s utility function has relatively little effect on the Forest Community’s allocation of time spent gathering within a particular regime. If anything, it appears that the Forest Community allocates slightly less labor to biomass gathering, N df, F which is accounted for by the slight increase in N s r and — . This is due to an c Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 increase in the opportunity cost of Ndf which comes through an infinitesimal and unobserved increase in the respective wage rates, w and a , which increase because • 7 • of an increase in demand for labor in the respective sectors. The fact that there is a sharp decline in Ndf under co-management relative to direct control is not surprising given that enforcement mechanisms are relatively more successful under co management. Note, however, that as enforcement under co-management falls, NJf increases because the penalties for removal also decrease. Finally, based on the utility functions and parameter values chosen, the Residual Sector unambiguously prefers JFM to the benchmark case, and the Forest Department prefers co-management with relatively higher levels of enforcement. This result is straightforward as the positive externality associated with biomass increases under JFM and increases with higher levels of enforcement while the negative externality associated with timber sales falls. These results are also consistent with existing analytic models to allow for meaningful comparison. In a study using a dynamic model to compare JFM with direct state control over time and under different enforcement levels and contract lengths, Ligon and Narain (1999) find a result alluded to above. Namely, that the Forest Community does not unambiguously prefer either JFM or exclusive State control. In their dynamic framework, when JFM contracts are long, or villagers are sufficiently patient, and the probability of government harvests is high, JFM is 7 Note that the production o f both X and Fr increase with increases in /? '. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 preferred to State control. In contrast, local communities prefer State control when contracts are short and the probability of government timber harvest is low. Our analysis finds that State control is preferred to co-management by the Forest Community when the enforcement of the JFM contract is perfect - that is, penalties associated with excess removal under JFM do not exist under direct State control - and the Forest Community highly values the forest good. Relative to the approach of Ligon and Narain (1999) this result is somewhat stronger in that we do not assume that the Forest Community shares in the profits of harvested timber; rather, these profits are exogenous to each sector and return indirectly to the Forest Community as a grant. In a related study, using a dynamic model of forest change, Kant (2000) explores the nature of an optimal resource regime over time. He argues that development and changing community valuations of the forest bring concomitant changes in the resource regime. An increase in heterogeneity of the Forest User Group and /or a decrease dependency on the forest for a staple goods results in a movement toward a private resource regime, whereas an increase in dependence or homogeneity results in an movement toward a community management regime. In a similar paper Kant and Berry (1998) argue that the open access (private property) regime is “optimal” in the extreme case of homogenous (heterogeneous) users and high (low) dependence on the natural resource base. Less than complete homogeneity of the user group or dependence on the resource base likewise yields an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 alternative “optimal” regime. Alternative arrangements described lie on a continuum between direct state control and open access and include community management, joint management between the State and the local community, between the State and private companies, and between the private companies and local communities. The simulation results are comparable with the Kant and Berry (1998) and Kant (2000) analysis above with respect to “dependence” on the resource and the corresponding optimal regime. In our framework, a low community valuation of the forest good (FJf) corresponds to a low dependence on the natural resource base, and the benchmark case (under which there exists zero enforcement) corresponds to de facto open access property. We find that under the benchmark case, as Forest Community dependence on the forest rises, so does Forest Community utility. Indeed, under very high levels of dependence, de jure state control (de facto open access) is preferred to the co-management arrangement. Similarly, note that as enforcement decreases under the JFM arrangement, at the lowest level of dependence (/?j = .4) Forest Community utility decreases; in contrast, under relatively high valuations of the forest good (a high value of f}\), Forest Community utility increases as enforcement of the JFM arrangement decreases. The Forest Community harvests greater levels of Fdf in a co-management regime with relatively lower enforcement levels. It is very likely that for communities that place a high value on the forest good, under sufficiently low levels of enforcement, utility under co-management will be greater than utility under direct control. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 3.4 Conclusion The model and analysis present evidence that, for a given set of utility functions and parameter values, there exist conditions under which a representative Forest Community prefers co-management to exclusive state control. The simulation results illustrate that when the Forest Community places very high values on biomass collection, the appeal of Joint Forest Management is relatively lower than the de facto open access regime. Flowever, keeping in mind the limited range of enforcement parameters tested, if the results of simulations follow the same pattern, the Forest Community that highly values the forest would prefer co-management to direct state control under co-management regimes with relatively low levels of enforcement. Communities with a low valuation of the forest good always prefer co management. Finally, the Residual Sector unambiguously prefers Joint Forest Management to the benchmark case, and similar to the Forest Department, prefers higher levels of enforcement. With respect to the production of biomass, greater levels occur under JFM and increase as enforcement increases, whereas protective labor is relatively insensitive to changes in valuation ofF^ . Likewise, equilibrium values of Fdf are higher under exclusive state control relative to co-management, increasing with relatively lower levels of enforcement, and are essentially constant with respect to changes in preferences for Fdf. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 These results suggest, that under certain conditions, a unanimous coalition could emerge in support of co-management; that is, JFM may be welfare enhancing for both the Residual Sector and the Forest Community. This appears to be particularly true in a community where dependence on the resource base is minimal. In a Forest Community where subsistence gathering is highly valued, these results suggest that other incentives must be in place in order for the Community to support an initial move from de jure State control and de facto open access to a co management regime (in certain contexts, possibly guarantees of unlimited harvests for certain staple crops or game). The loss of subsistence harvest (due to an enforced limit on the amount of extraction) outweighs any increase in income from the protective labor wages. In contrast, given the conditions of the model, the Residual Sector always stands to benefit from JFM - a result due primarily to the reduction in the negative externality and the increased production of biomass. Finally, the Forest Department will always support a JFM policy with higher levels of enforcement and extraction penalties relative to a co-management policy with low levels of enforcement and extraction penalties. In conclusion, the model presented in this chapter is intended to provide a general framework for understanding co-management. Broadly, the model indicates that co-management can lead to higher levels of biomass production and reductions in forest degradation under certain conditions. Furthermore, results support the notion that the utility of user groups with high levels of dependence on the forest Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. increase as penalties for subsistence gathering decline, consequently such users prefer state control, that is the defacto open access regime. For sufficiently low levels of enforcement under co-management, however, such user groups may prefer co-management. Users with a low valuation of the forest good always prefer co management to direct state control as do those in the Residual Sector. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 References Alderman, H. and C. H. Paxson, 1994. Do the Poor Insure? A Synthesis on the Literature on Risk and Consumption in Developing Countries. In: Atkinson, A.B. (Ed.), Economics in a Changing World, Vol. 110. St. Martins Press: New York, 48-78. Angelsen, A., 1995. Shifting Cultivation and ‘Deforestation’: a Study from Indonesia. World Development, 23 (3), 1713-1729. Angelsen, A., 1999. Agricultural Expansion and Deforestation: Modelling the Impact of Population, Market Forces and Property Rights. Journal of Development Economics, 58 (1), 185-218. Argrawal, A. and E. Ostrom, 2001. Collective Action, Property Rights, and Decentralization in Resource Use in India and Nepal. Politics and Society, 29 (4), 485-514. Arnold, J.E.M., 1998. Managing Forests as Common Property. Food and Agricultural Organization Forestry Paper, 136, United Nations, Rome. Baland, J.M. and J.P. Platteau, 1996. Halting Degradation of Natural Resources. Is There a Role for Rural Communities? Oxford University Press, New York. Bardhan, P. and C. Udry, 1999. Development Microeconomics. Oxford University Press, Oxford. Becker, C.D. and L. Rosario, 1998. Indigenous Forest Management in the Bolivian Amazon: Lessons from the Yuracare People. International Association for the Study of Common Property, http://www.indiana.edu/~iascp/Final/becker.pdf, downloaded (05/27/2002). Bluffstone, R.A., 1995. The Effect of Labor Market Performance on Deforestation in Developing Countries Under Open Access: an Example from Rural Nepal. Journal of Environmental Economics and Management, 29 (1), 42-63. Bulte, E. and R.D. Horan, 2002. Habitat Conservation, Wildlife Extraction, and Agricultural Expansion. Tilburg University Department of Economics Working Paper, Tilburg University, Netherlands. Chase Smith, R., 2000. Community-Based Resource Control and Management in Amazonia: a Research Initiative to Identify Conditioning Factors for Positive Outcomes. Presented at the 8’th IASCP Conference, June 2002, pp 1-31. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 Chicchon, A., 1995. Faunal Resource Use by the Chimane of Eastern Bolivia: Policy Notes on a Biosphere Reserve, in Sponsel, L.E. (Ed.), Indigenous Peoples and the Future of Amazonia: an Ecological Anthropology of an Endangered World. University of Arizona Press, Tucson, AZ, 225-243. Chomitz, K.M. and C. Griffiths, 1997. An Economic Analysis of Woodfuel Management in the Sahel: the Case of Chad. World Bank Environmental Impact Assessment Document, 1-26. Dasgupta, P., 1993. An Inquiry into Well-Being and Destitution. Oxford University Press, Oxford. Dasgupta, P., 1995. Population, Poverty and the Local Environment. Scientific American, 272 (2), 40-46. Deaton, A. and C. Paxson, 1994. Intertemporal Choice and Inequality. Journal of Political Economy 102 (3), 437-467. Deaton, A. and C. Paxson, 2000. Growth and Saving Among Individuals and Households. The Review of Economics and Statistics 82 (2), 212-225. Deaton, A., 1992a. Saving and Income Smoothing in the Cote d’Ivoire. Journal of African Economies 1 (1), 1-24. Deaton, A., 1992b. Understanding Consumption. Oxford University Press, Oxford. DeShazo, R.P. and J.R. DeShazo., 1995. An Economic Model of Smallholder Deforestation: a Consideration of the Shadow Value of Land on the Frontier. In Sandbukt, O. (Ed.) Management of Tropical Forests: Towards an Integrated Perspective. Center for Development and the Environment, University of Oslo, Oslo. Dvorak, K.A., 1992. Resource Management by West African Farmers and the Economics of Shifting Cultivation. American Journal of Agricultural Economics, 74, 809-815. Edmonds, E., 2002. “Government-Initiated Community Resource Management and Local Resource Extraction from Nepal’s Forests.” Journal of Development Economics, 68 (1), 89-115. Eghenter, C., 2000. What is Tana Ulen Good For? Considerations on Indigenous Forest Management, Conservation, and Research in the Interior of Indonesian Borneo. Human Ecology, 28 (3), 331-357. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 Eswaran, M. and A. Kotwal, 1989. Credit as Insurance in Agrarian Economies. Journal of Development Economics, 31 (1), 37-53. Fafchamps, M. and S. Lund, 1999. Risk-Sharing Networks in Rural Philippines. Mansfield College, Oxford University and Department of Economics, Stanford University Working Paper. Fafchamps, M., 1999. Rural Poverty, Risk, and Development. Food and Agricultural Organization Economic and Social Development Paper Number 144, 1-137. Fafchamps, M., C. Udry and K. Czukas, 1998. Drought and Saving in West Africa: Are Livestock a Buffer Stock? Journal of Development Economics, 55 (2), 273- 305. FAO (United Nations Food and Agricultural Organization), 1997. State of the World's Forests. Food and Agricultural Organization, Rome. FAO (United Nations Food and Agricultural Organization), 1999. State of the World's Forests. Food and Agricultural Organization, Rome. FAOSTAT (United Nations Food and Agricultural Organization Online Statistical Database), 2002. http://apps.fao.org/default.htm. Accessed 07-15-2002. Godoy, R. and M. Jacobson, 1999. Covariates of Private Time Preference: a Pilot Study among the Tsimane’ Indians of the Bolivian Rain Forest. Evolution and Human Behavior, 20 (4), 249-256. Godoy, R., J.R. Franks, D. Wilkie, M. Alvarado, G. Gray-Molina, R. Roca, J. Escobar, and M. Cardenas, 1996. The Effects of Economic Development on Neotropical Deforestation: Household and Village Evidence from Amerindians in Bolivia. Harvard Institute for International Development, Development Discussion Paper No. 540. Godoy, R., M. Jacobson, D. Wilkie, 1998b. Strategies of Rain-Forest Dwellers Against Misfortune: the Tsimane Indians of Bolivia. Ethnology, 37 (1), 55-69. Godoy, R., M. Jacobson, J. De Castro, Vianca Aliaga, J. Romero, and A. Davis, 1998a. The Role of Tenure Security and Private Time Preferences in Neotropical Deforestation. Land Economics, 74 (2), 162-170. Goldstein, M., 1999. Chop Time, No Friends: Intrahousehold and Individual Insurance Mechanisms in Southern Ghana. Department of Agriculture and Applied Economics, University of California, Berkeley, Working Paper. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 Grabowski, R. and M. Shields, 1996. Development Economics. Blackwell Publishers, Cambridge, MA. Grimard, F., 1997. Flousehold Consumption Smoothing Through Ethnic Ties: Evidence from Cote d’Ivoire. Journal of Development Economics, 53 (2), 391 - 422. Hill, I. and D. Shields. 1998. “Incentives for Joint Forest Management in India: Analytical Methods and Case Studies.” World Bank Technical Paper no. 394. Washington DC: World Bank. Holden, S.T., J.E. Taylor, and S. Hampton, 1998. Structural Adjustment and Market Imperfections: a Stylized Village Economy-Wide Model with Non-Separable Farm Households. Environment and Development Economics, 4, 69-87. IFAD (International Fund for Agricultural Development), 1992. The State of World Rural Poverty: an Inquiry into its Causes and Consequences. New York University Press, New York. Kaimowitz, D. and A. Angelsen, 1998. Economic Models of Tropical Deforestation - a Review. Center for International Forestry Research, Bogor, Indonesia. Kant, S. and R.A. Berry, 1998. Community Management: an Optimal Resource Regime for Tropical Forests. Department o f Economics, University of Toronto, Working Paper UT-ECIPA-BERRY-98-01. Kant, S. and R.A. Berry, 2001. A Theoretical Model of Optimal Forest Resource Regimes in Developing Economies. Journal of Institutional and Theoretical Economics, 157 (2), 331-355. Kant, S. 2000. A Dynamic Approach to Forest Regimes in Developing Economies. Ecological Economics, 32, 287-300. Klooster, D., 1998. Institutional Choice, or a Process of Struggle? A Case Study of Forest Co-Management in Mexico. Presented at the 7’th IASCP Conference June, 1998, http://www.indiana.edu/~iascp/Drafts/klooster.pdf, (downloaded 05/27/02), 1-28. Klooster, D., 2000. Institutional Choice, Community, and Struggle: A Case Study of Forest Co-Management in Mexico. World Development, (28) 1, 1-20. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 Kochar, A., 1999. Smoothing Consumption by Smoothing Income: Hours-of-Work Responses to Idiosyncratic Agricultural Shocks in Rural India. The Review of Economics and Statistics, 81 (1), 50-61. Kumar, S., 2002. Does ‘Participation’ in Common Pool Resource Management Help the Poor? A Social Cost-Benefit Analysis of Joint Forest Management in Jharkhand, India. World Development, 30(5), 763-782. Kurosaki, T. and M. Fafchamps, 2002. Insurance Market Efficiency and Crop Choices in Pakistan. Journal of Development Economics 67 (2), 419-453. Larson, B.A., 1991. The Causes of Land Degradation Along ‘Spontaneously’ Expanding Agricultural Frontiers in the Third World: Comment. Land Economics, 67 (2), 260-266. Lehm, Z. and J. Kudrenecky, 1995. Reflections on a Proposal for Forest Management and Harvest in the Multi-Ethnic Indigenous Territory of Beni, Bolivia. In: Madison, WI: IIES, Land Tenure Centure (Eds.) Case Studies of the Community-Based Forestry Enterprises in the Americas. University of Wisconsin, Madison, 71-97. Ligon, E., 1998. Risk Sharing and Information in Village Economies . Review of Economic Studies, 65, 847-864. Ligon, E. and U. Narain, 1999. Government Management of Village Commons: Comparing Two Forest Policies. Journal of Environmental Economics and Management, 37, 272-289. Mendelsohn, R., 1994. Property Rights and Tropical Deforestation. Oxford Economic Papers, 46 (supplement Oct.), 750-756. Morduch, J., 1990. Risk, Production and Saving: Theory and Evidence from Indian Households. Department of Economics, Harvard University, mimeo. Morduch, J., 1991. Consumption Smoothing Across Space: Tests for Village Level Responses to Risk. Department of Economics, Harvard University, mimeo. Morduch, J., 1994. Poverty and Vulnerability. American Economic Review, 84 (2), 221-225. Morduch, J., 1995. Income Smoothing and Consumption Smoothing. The Journal of Economic Perspectives, 9 (3), 103-114. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 Munoz-Pina, C., A. de Janvry, and E. Sadoulet, 2002. Recrafting Rights Over Common Property Resources in Mexico: Divide, Incorporate, and Equalize. Department of Economics, University of California, Berkeley Working Paper, http://are.berkeley.edu/~sadoulet/papers/Carlos.pdf, downloaded (06/06/2002), 1- 33. Oczkowski, E. and N.E. Philp, 1994. Household Expenditure Patterns and Access to Consumer Goods in a Transitional Economy. Journal of Economic Development, 19 (1), 165-183. Omamo, S.W. 1998a. Farm-to-Market Transaction Costs and Specialisation in Small-Scale Agriculture: Explorations with a Non-Separable Household Model. The Journal of Development Studies, 35 (2), 152-163. Omamo, S.W., 1998b. Transport Costs and Smallholder Cropping Choices: an Application to Siaya District, Kenya. American Journal of Agricultural Economics, 80 (2), 116-123. Palit, S., 1996. Comparative Analysis of Policy and Institutional Dimensions of Community Forestry in India and Nepal. ICIMOD MNR Series No. 96/4. ICIMOD, Kathmandu. Paxson, C., 1992. Using Weather Variability to Estimate the Response of Savings to Transitory Income in Thailand. American Economic Review, 82 (1), 15-33. Piland, R.A., 1991. Traditional Chimane Agriculture: a Starting Point for Agricultural Development. Latin Americanist, 26 (2), 5-9. Piland, R.A., 1991. Traditional Chimane Agriculture and its Relation to Soils of the Beni Biosphere Reserve, Boliva. M.A. thesis, University of Florida, Gainesville. Ravallion, M. and S. Chaudhuri, 1997. Risk and Insurance in Village India: Comment. Econometrica, 65 (1), 171-184. Reardon, T. and S.A. Vosti., 1995. Links Between Rural Poverty and the Environment in Developing Countries: Asset Categories and Investment Poverty. World Development, 23 (9), 1495-1506. Reister, J., 1976. En busca de la loma santa. Los Amigos del Libro, La Paz, Bolivia. Richards, M., 1996. Stabilising the Amazon Frontier: Technology, Institutions and Policies. Natural Resource Perspecitives, 10, 1-7. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 Richards, M., 1997. Tragedy of the Commons for Community-Based Forest Management in Latin America? Natural Resource Perspectives, 22, 1-9. Richards, M., 1999. Can Tropical Forestry be Made Profitable by ‘Internalising the Externalities’? Natural Resource Perspectives, 46, 1-6. Richards, M., 2000. Can Sustainable Tropical Forestry be Made Profitable? The Potential Limitations of Innovative Incentive Mechanisms. World Development, 28 (6), 1001-1016. Rose, E., 2001. Ex Ante and Ex Post Labor Supply Response to Risk in a Low- Income Area. Journal of Development Economics, 64 (2), 371-388. Rosenzweig, M. and H. Binswanger, 1993. Wealth, Weather Risk and the Composition and Profitability of Agricultural Investments. The Economic Journal, 103, 56-78. Rosenzweig, M. and K. Wolpin, 1993. Credit Market Constraints, Consumption Smoothing, and the Accumulation of Durable Production Assets in Low-Income Countries: Investments in Bullocks in India. Journal of Political Economy, 101 (2), 223-244. Rosenzweig, M., 2001. Savings Behavior in Low-Income Countries. Oxford Review of Economic Policy, 17 (1), 40-54. Ruddle, K., 1989. Solving the Common-Property Dillemma: Village Fisheries Rights in Japanese Coastal Waters. In: F. Berkes (Ed.), Common Property Resources: Ecology and Community-Based Sustainable Development. Belhaven Press, London, 168-184. Saha, A., 1994. A Two-Season Agricultural Household Model of Output and Price Uncertainty. Journal of Development Economics, 45 (2), 245-269. Sehkar, N.U., 2000. Decentralized Natural Resource Management: From State to Co- Management in India. Journal of Environmental Planning and Management, 3(1), 123-138. Shiferaw, B. and S.T. Holden, 2000. Policy Instruments for Sustainable Land Management: the Case of Highland Smallholders in Ethiopia. Agricultural Economics, 22, 217-232. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 Southgate, D., 1990. The Causes of Land Degradation Along ‘Spontaneously’ Expanding Agricultural Frontiers in the Third World. Land Economics, 66 (1), 93-101. Stryker, D.J., 1976. Population Density, Agricultural Technique, and Land Utilization in a Village Economy. American Economic Review, 66 (3), 347-358. Sundar, N., 2000. Unpacking the ‘Joint’ in Joint Forest Management. Development and Change, 31, 255-279. Townsend, R.M., 1994. Risk and Insurance in Village India. Econometrica, 62 (3), 539-591. Townsend, R.M., 1995a. Consumption Insurance: An Evaluation of Risk-Bearing Systems in Low-Income Countries. Journal of Economic Perspectives, 9 (3), 83- 102. Townsend, R.M., 1995b. Financial Systems in Northern Thai Villages. Quarterly Journal of Economics, 110 (4), 1011-1046. Udry, C., 1994. Risk and Insurance in a Rural Credit Market: an Empirical Investigation in Northern Nigeria. Review of Economic Studies, 61 (3), 495-526. Udry, C., 1995. Risk and saving in Northern Nigeria. American Economic Review, 85 (5), 1287-1300. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 Appendix 1. Co-Management Simulations: Functional Forms A.1.1 Forest Community Functional Form < 7 c- \ < J C ~\ cr(. — 1 < J (. Uc = {Ac + £ W C - X n V 7 ) ^ ' -BcH a ^ dfN ^ + a } F j ' } Start Values: A =io Pdf = 0.8; = 0.9 if}], A ,P]) = (.4, .3, .3); (.45, .25, .3); (.5, .225.275); (.55, .2, .25); (.6,. 175.225); (.65,. 15, .2); (.7,. 125,. 175); (.75,. 1,. 15); (.8, .075,. 125) cr, = 0.8 B = .2\a\ = .6;a l - A \ e f =1.5 A.1.2 Forest Department Functional Form (Utility Maximization Case) <7,-1 o y -l (J j — \ Gf Uf = {Af *(Pf ((1_ share)*Pf F) + (32 f ( P ^ N ^ Y ^ 1 -Bf * (a] f /3dfNdfe ,i r +a] fFeilYf } Start Values: A f = 10 P4 = 0 .5 ;^ =1.1 ( f i , /}},#) = (. 1,.6,.3) a { = 0.8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 c = 1.2 share =. 1 Bf = .2;al f =.6; a j =A;ef =1.5 Functional Form (Profit Maximization Case): U j = {(1 - share) * Pf Fr * + Gf - a N d d ] - A.1.3 Residual Sector Functional Form: Gr - \ < J r - I <7,.-1 < 7 ,. Ur = (4 *(/W ^ +P2 r(PxN ^ ~ X L ) ~ + Start Values: 4=10 cr. = 0.8 Br = .2;a] f =.6 ;< 2 2 =.4;ef =1.5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 Appendix 2. Co-Management Simulations: GAUSS Code /* BEGINNING */ new; /* CONSTRAINED OPTIMIZATION MODULE uses sequential quadratic programming parameters are updated in a series of iterations beginning with starting value provided NON LINEAR DIFFERENTIAL EQUATION SOLVER */ library co, simplex, nlsys, pgraph; #include co.ext; i/include nlsys.ext; #include simplex.ext; coset; nlset; lpset; output file = jfm2results.out reset; outwidth 256; els; output file = jfmchanges.out reset; outwidth 256; els; /^UPDATED VALUES FOR THE SECOND AND THIRD DO LOOPS */ kkm = 3; kk = 2; count 1=0; /* 2 Sector model - Forest sector maximizes production of biomass */ /* START VALUES SOME WILL BE UPDATED________________________________________________________ */ /* df: de facto activities - gathering (labor yields payment in kind to forest community) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 dj: de jure protective services (supplied by forest community demanded by forest dept, as an input to production of F - in-kind payment) r: residual sector activities - production of good X (labor provided by forest community demanded by residual sector as an input to production of X - cash payment of W) */ /* PRICES (RELATIVE TO price of good X */ alpha = 1; /* 1 (1.8 with utility max) alpha is the wage paid to the forest community for labor services */ pricedj = 1; /* 4.3820662 CONSTANT wage paid to forest workers - 40 in WB*/ w = 1; /* .6 1.5 0.60357145 0.42152523 0.095280567 2 UPDATED WAGE RATE - ? in WB */ wage = 1; Pf = 1.6; /* 1.6 UPDATED PRICE OF THE FOREST GOOD - */ PriceF =1.6; Px = 1.7; /* 1.5 price of export - when p f < px exports are some fraction of Frs-Frd */ rundmy = 1; MA11 = 0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0 ~0~0~0~0~0~0~0~0~0~0~0~0~0; TA11 = o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o ~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0; /* in choosing remember that N = Ndjd + N df + Nrs and that Ndjd = (Gf-H)/alpha (by definition) and that F = Fbar + Frs + Fdf where Fdf = betadf*NdfA elasdf */ Xr = 80; /* 18.9 1.6 20 200 UPDATED USED IN THE FOREST COMMUNITY FIRST RUN */ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 Ntot = 250; /* 200 TOTAL LABOR SUPPLY - WHERE Ntot = Nrs + N df + Ndjs */ Ndjeq = 20; /* 42 intial value for equilibrium laborers */ Nreq = 24; /* 24 ilibrium labor supply used in calculating G - updated until equilibrium labor supply reached then kept constant */ Fbar = 1100; /* 20 RUN THIS WITH A POSITIVE VALUE OF FBAR - 10 CONSTANT ADDED TO MODEL WHERE Ftot = Fbar + Frs + Fdf*/ Ftot = 1000; /* 100 USED IN THE RESIDUAL SECTOR AND FOREST SECTOR FIRST RUN - updated */ Feq =10; /* 8 52 Feq is used as Nreq is used */ Frs = 10; Frx = 0; /* difference between Frs and Frd when price is less than one */ /* GOVERNMENT GRANTS */ betag = .25; /* .2 .25 FRACTION TAX REVENUE (GRANT) GOING TO THE FOREST DWELLERS */ gammag = .35; /* .5 .35 FRACTION TAX REVENUE (GRANT) GOING TO THE RESIDUAL SECTOR */ G f = 7; /* 7 .5 UPDATED implies that Ndjd = 104.16 for H = 76 */ Gc = 3; /* 3 1 */ G r= 15; /* 15 6 */ /* TAXES AND PENALTIES */ /* TAX ON RESIDUAL SECTOR */ t = . 1; /* TAX RATE ON RESIDUAL SECTOR PROFIT */ /* FOREST COMMUNITY PENALTY FOR STEALING */ Fa = 1; /* 1 4.1350204 initial value of Fa - updated after to be equal to 10% of biomass production*/ perbio = .048; /* .9 .048 percentage of biomass produced - Fdf cannot exceed this without penalty */ zf = 1; /* 1 = fc pays for excess gathering; 0 = fc does not pay for excess */ /* PRODUCTION PARAMETERS - EXOGENOUS TO SYSTEM */ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 /* FOREST COMMUNITY GATHERING /STEALING */ betadf = .8; /* changed from .8 .6 .3 .9 2 (.6 works for profit max case - .3 for utility max case)increase THE COEFFICIENT FOR Fdf = betadf*NdfA elasdf */ elasdf = .9; /* changed from .9 1.1 .8 THE EXPONENT ON THE COBB DOUGLAS PRODUCTION FUNCTION FOR Fdf*/ /* PRODUCTION OF GOOD X */ betax =10; /* 10 3 THE COEFFICIENT FOR Xtot = betax*NrdA elasx = betax*Nrs */ elasx = .9; /* .9 THE EXPONENT ON THE COBB DOUGLAS PRODUCTION FUNCTION FOR Xtot */ /* PRODUCTION OF BIOMASS */ betadj = .5; /* changed from .5 (.3 will work for profit max case .5 for utility max case) */ elasdj = 1.1; /* changed from 1.1 .8 keep this as an increasing function 1.8 this was 1.4 .8 elasticity of production of biomass */ /* NEGATIVE EXTERNALITY EFFECT */ /* MULTIPLIERS FOR CES PRODUCTION FUNCTION (FOR COBB DOUGLAS USED .5, -.1, -.1) */ alphafl = .6; /* .6 .7, 1.7 N df Multiplier - */ alphaf2 = .4; /* .4 .3, -.4 Frs or Frd Multiplier*/ expf= 1.5; /* 1.5 1.9 should be > 1 exponent for externality growth */ Be = .2; /* .2 .03 */ Br = .2; /* .11 .03 */ B f = .2; /* .2 .03 */ /* CES UTILITY FUNCTION PARAMETERS */ /* RESIDUAL SECTOR */ Ar = 10; /* 2.5 constant multiple of utility function */ alpharl = .4 alphar2 = .5 alphar3 = .1 /* .8 .6 .5 VALUE OF THE FOREST GOOD Frd */ /* .3 .4 VALUE OF THE MARKET GOOD Xr */ /* VALUE OF THE GOVERNMENT GRANT Gr */ sigmar = .8; /* 1.5 4/5 8/9 .8 RESIDUAL SECTOR ELASTICITY OF SUBSTITUTION BETWEEN GOODS */ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 rhor = ((sigmar - 1) / sigmar); /* FOREST COMMUNITY */ A c= 10; /*. 10 2.5 1.2 CES Utility Function Multiplier */ alphacl = .4; alphac2 = .3; alphac3 = .3; /* .4 .5 BENEFIT OF Fdf*/ /* .3 .25 BENEFIT OF Gc */ /* .3 .25 BENEFIT OF Xc */ sigmac = .8; /* .8 ELASTICITY OF SUBSTITUTION BETWEEN GOODS */ rhoc = ((sigmac - 1) / sigmac); /* FOREST DEPARTMENT UTILITY FUNCTION */ A f = 10; /* 10 */ betaf 1 = . 1; betaf2 = .6; betaO = .3; /* .01 .2 .1 .05 MULTIPLIER FOR Frs */ /* .98 .6 .6 .01 .0001 MULTIPLIER FOR Ndjd */ /* .01 .2 .3 .49 MULTIPLIER FOR THE GRANT */ sigmaf = .8; /* .8 .6 4/5 .75 .8 .65 ELASTICITY OF SUBSTITUTION BETWEEN GOODS */ rhof = ((sigmaf - 1) / sigmaf); /* FOREST DEPARTMENT OTHER PARAMETERS */ share = .1; /*. 1 .855 .2 .05 1 = all; share of revenues going to the federal government */ H = 1; /* 1 5 76 EXTERNAL COSTS TO FOREST SECTOR*/ frcut =1.2; /* 1.2 = c in equations = 1/frcut = cutting cost */ expfr = 1; /* 1 DON'T USE THIS - KEEP AS 1 costs of forest removal are increasing */ fdxmult = 100; /* 100 value to multiply by when Frx >0 */ /* START VALUES ALL */ /* RESIDUAL SECTOR */ /* Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 rsstart = {21,86,97,.7,-.4}; /* { Frd, Nrd, Xr, rsmultl, rsmult2 } */ rrsstart= {21,86,97,.7,-.4}; */ rsstartl = {10,24,80,.7,-.4}; rsstart = {10,24,80,.7,-.4}; rrsstart= {10,24,80,.7,-.4}; /* FOREST COMMUNITY */ /* fcstart = {86,100,60,90,.04,-.2,-.02}; /* { Nrs, Xc, Ndf, Ndjs, fcmultl, fcmult2, fcmult3 } */ rfcstart = {86,100,60,90,.04,-.2,-.02}; */ case */ fcstartl = {54,80,47,139,.04,-.2,-.02,1}; /* worked with profit max fcstart = {54,80,47,139,.04,-.2,-.02,1}; rfcstart = {54,80,47,139,.04,-.2,-.02,1}; fcstartl = {57,70,14,140,.04,-.2,-.02,1}; fcstart = {57,70,14,140,.04,-.2,-.02,1}; rfcstart = {57,70,14,140,.04,-.2,-.02,1}; */ /* FOREST DEPARTMENT */ fdstartl = {169,150,0,-.8}; fdstart = {169,150,0,-.8}; /* UTILITY MAX CASE */ rfdstart = {169,150,0,-.8}; /* { Frs, Ndjd, fdmultl, fdmult2, fdmult3 } */ fdstartl = {90,20,-.5,-.8}; /* PROFIT MAX CASE */ fdstart = {90,20,-.5,-.8}; /* PROFIT MAX CASE */ rfdstart = {90,20,-.5,-.8}; /* { Frs, Ndjd, fdmultl, fdmult2 } */ fdstartl = {169,150,-.5,-.8}; /* PROFIT MAX CASE */ fdstart = {169,150,-.5,-.8}; /* PROFIT MAX CASE */ rfdstart = {169,150,-.5,-.8}; /* { Frs, Ndjd, fdmultl, fdmult2 } */ */ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 /* Failed Iterations */ fcfail = 0; fdfail = 0; rsfail = 0; /* Variables for to Control Sector Shift */ m = 1.0e+15; n = 1.0e+15; p = -l; o = -1; mmm = 1.0e+15; /* = = = #2 D 0 LOOP - THIS IS USED TO ADJUST PARAMETER VALUES = = = = = = = = = * / do while kkm < mmm; mm = 1.0e+15; = _ #3 D 0 L 0 0 p ===== */ k = 1; do while kk < mm; m = 1.0e+15; n = 1.0e+15; p = -i; o = -1; /* ===== #4 D 0 LOOP ===== */ do while k < m; /* RESET FTOT AND FBAR TO INITIAL VALUES */ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 Fbar = 1100; /* 20 RUN THIS WITH A POSITIVE VALUE OF FBA R- 10 CONSTANT ADDED TO MODEL WHERE Ftot = Fbar + Frs + Fdf */ Ftot = 1000; /* 100 USED IN THE RESIDUAL SECTOR AND FOREST SECTOR FIRST RUN - updated */ output file = jfm2results.out off; print; print t» print; print "YEAR" k; print; print "BEGINNING OF RUN " k; print; print; print "CONSTANT PARAMETERS"; print; print "Fbar =" Fbar; print "Ftot =" Ftot; print "Ntot =" Ntot; print; print "PARAMETER VALUES ADJUSTED TO EQUATED SUPPLY AND DEMAND OF Nr AND Nd"; print; print "w =" w; print "Pf =" Pf; print "alpha =" alpha; print; print "VARIABLES BELOW WILL TAKE THE OPTIMAL VALUES CALIBRATED IN PREVIOUS YEAR"; print; print "Nreq =" Nreq; print "Feq =" Feq; print "G f=" Gf; print "Gc =" Go; print "Gr - ' Gr; print; /* output file = jfm2results.out off; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 */ /* FOREST COMMUNITY CHOICE PROBLEM choose Ndjs and Nrs to max Uc = Uc(Fdf, Fdf + Frd, Gc, Xc) s.t. Xc = (w*Nrs) + alpha*Ndjd Xtot = Xr + Xc; Fdf = betad f* Nd fA e 1 as fdf Fdjd = demand determined in the forest sector; Nrs = Ntot - Ndjd - Ndf F = F(Ndjd, Fdf, Frs) F = Fbar + Fdjd + Fdf */ /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++ FOREST COMMUNITY USING NON-LINEAR SOLVER Nrs = x[l]; Xc = x[2]; N d f= x[3]; Ndjs = x[4]; Fdf = betadf* N dfA e 1 asd f; fcmultl = x[5]; fcmult2 = x[6]; fcmult3 = x[7]; fcmult4 = x[8]; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 */ proc fc(x); local fcl, fc2, fc3, fc4, fc5, fc6, fc7, fc8; /* dL/dNrs */ fcl = x[5]*(elasx*betax*NreqA (elasx-l)) - x[6] + x[7]*w; /* dL/dXc */ fc2 = (Ac*(l/rhoc)* (alphacl *(betadf|!x[3]A elasdf)A rhoc + alphac2*GcA rhoc + alphac3*x[2]A rhoc)A ((l-rhoc)/rhoc)) * (alphac3*rhoc*x[2]A (rhoc-l)) - x[5] - x[7]; /* dL/dNdf */ fc3 = (Ac*(l/rhoc)* (alphacl*(betadfK x[3]A elasdf)A rhoc + alphac2*GcA rhoc + alphac3*x[2]A rhoc)A ((l-rhoc)/rhoc)) * (alphacl *rhoc*(betadPx[3]A elasdf)A (rhoc-l)) * (betadPelasdfH x[3]A (elasdf-l)) - ( Bc*expf!(alphafl*(betadf!x[3]A elasdf) + alphaf2*Feq)A (expf-l)) * ( alphafl *elasdf*(betadPx[3]A (elasdf-l ) ) ) -x[6] - x[7]*(Pf*zf|!(betadf! ! elasdPx[3]A (elasdf-l))) - x [8] * (-betadf* elasdf*x[3 ]A (elasdf-1)); /* dL/dNdjs */ fc4 = -x[6] + (x[7]*alpha) - x[8]*(betadj*elasdj*x[4]A (elasdj-l)); /* dL/dfcmultl */ fc5 = Xr + x[2] - (betax*x[l]A elasx); /* dL/dfcmult2 */ fc6 = (Feq/frcut) + x[4] + x[3] + x[l] - N to t; /* dL/dfcumlt3 */ fc7 = x[2] + (zfH P f,c ((betadf|!x[3]A elasdf)-Fa)) - w*x[l] - alpha*(x[4] + ((F eq A expfr)/frcut)); /* dL/dfcmult4 */ fc8 = Fbar - Ftot + (betadj*x[4]A elasdj) - (betadf|:x[3]A elasdf) - Feq ; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 retp(fc 11 fc21 fc31 fc41 fc51 fc61 fc71 fc 8); endp; /* output file = jfm2results.out on; */ if k <=2; xO = fcstart; _nltypx = fcstart; else; xO = rfcstart; _nltypx = rfcstart; endif; altnam = { Nrs, Xc, Ndf, Ndjs, fcmultl, fcmult2, fcmult3, fcmult4 }; title = "Forest Community: Non-linear equation solver"; _nlmaxit = 100; output = 1; _nlalgr = 2; _nlchpf = 1; /* /* STARTING JACOBIAN */ REDO - THIS IS ONLY FOR CONSTRAINT let feel [1,1] = -elasx*betax*x[l]A (elasx-l); let fce2[l,l] = -zf*Pf*elasdfK betadf|:x[3]A (elasdf-l); let fcjl [1,4] = -1 feel 0 0; let fcj2[l,4] = 0-1 -1 -1; let fcj3[l,4] = -1 w alpha fce2; fcJac = fcjl | fcj2 | fcj3; _nlstjc = fcJac; /* output file = jfm2results.out on; */ print fcJac; output file = jfm2results.out off; nlmtol =10; */ {x,fcV,fcJ,tcode}=nlprt(nlsys(&fc,xO)); Nrs = x[l]; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 Xc = x[2]; N df = x[3]; Fdf = betadf*NdfA elasdf; Ndjs = x[4]; fcmultl = x[5]; fcmult2 = x[6]; fcmult3 = x[7]; fcmult4 = x[8]; if tcode == 1; fcret = 0; rfcstart = Nrs|Xc|Ndf]Ndj s| fcmult 11 fcmult21 fcmult31 fcmult4; else; fcret = tcode; fcfail = fcfail + 1; /* output file = jfm2results.out on; */' print; print "****** fc failed on iteration" k; print "****** rc" tcode; print; output file = jfm2results.out off; endif; print; print "+++++++++++++++ FOREST COMMUNITY USING NON-LINEAR SOLVER ++++++++++++++++++++++ "; print; print "Nrs" Nrs; print "Xc" Xc; print "N df’ Ndf; print "Ndjs" Ndjs; print "fcmultl" fcmultl; print "fcmult2" fcmult2; print "fcmult3" fcmult3; fctempu = Ac*( alphacl *(betadfH NdfA elasdf)A rhoc + alphac2*GcA rhoc + alphac3*XcA rhoc )A (l/rhoc); fctempe = Bc*(alphafl*Fdf + alphaf2*Feq)A (expf); fcbiom = (betadj*NdjsA elasdj); print; print "forest community utility" fctempu; print "biomass production" fcbiom; print "externality" fctempe; print; output file = jfm2results.out off; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 FOREST DEPARTMENT UTILITY MAX CASE Frs = y [1]; Ndjd =y[2]; fdmultl =y[3]; fdmult2 = y[4]; fdmult3 = y [5]; = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = */ ifpx>pf; FrxDiff = ((l-share)*(Px-Pf)*Frx); else; FrxDiff = 0; endif; proc fd(y); local fdl, fd2, fd3, fd4; /* dL/dNdjd */ fdl = A P(l/rhof)*( betafl*(betadj*y[2]A elasdj)A rhof + betaf2*((l- share)*PPy[l])A rhof + betaf3*GfA rhof )A (( 1 -rhof)/rhof) *rhoPbetafl*( (betadj*y[2]A elasdj)A (rhof-l) )* (elasdj*betadj*y[2]A (elasdj- 1)) -y[3]*alpha - y [4]*(betadj*elasdj* y [2] A (elasdj -1)); /* dL/dFrs */ fd2 = A P(l/rhof)*( betafl* (betadj *y [2] A elasdj)A rhof + betaf2*((l- share)*PPy[l ])A rhof + betaf3*GfA rhof )A ((l-rhof)/rhof) * rhoP betaf2 * ((1 -share) * P P y [ 1 ]) A (rhof-1) *((l-share)*Pf) - ((BPexpP(alphafl*Fdf + alphaf2*y[l])A (expf-l))*alphaf2) - (y[3]*(-(l-share)*Pf + (alpha/frcut))) -y[4]; /* dL/dfdmultl */ fd3 = - Gf - ((l-share)*PPy[l]) - FrxDiff - (zP P P (F d f - Fa)) + H + alpha*( y[2]+(y[l]/frcut)); /* dL/dfdmult2 */ fd4 = (betadj*y[2]A elasdj) - Fdf - y[l] + Fbar - Ftot; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 /* /* dL/dfdmult3 */ fd5 = (y[l]/frcut) + y [2] + Ndf + Nreq - Ntot; /* OLD CASE WITHOUT LABOR CONSTRAINT */ /* dL/dNdjd */ fdl = A P (1 /rhof)*( betafl *(betadj *y[2]A elasdj)A rhof + betaf2*((l-share)*PPy[l])A rhof + beta£3*GfA rhof )A ((l-rhof)/rhof) *rhoPbetafl*( (betadj*y[2]A elasdj)A (rhof-l) )* (elasdj*betadj*y[2]A (elasdj- 1))+ y[3]*alpha - y[4]*(betadj*elasdj*y[2]A (elasdj-l)); /* dL/dFrs */ fd2 = A P(l/rhof)*( betafl *(betadj*y[2]A elasdj)A rhof + betal2 * ((1 - share) * P P y [ 1 ]) A rhof + betaO*GfA rhof )A ((l-rhof)/rhof) *rhoPbetaf2*((l-share)*PPy[l])A (rhof-l) *((l-share)*Pf) ((BPexpP(alphafl*Fdf + alphaf2*y[l])A (expf-l))*alphaf2) - (y[3]*((l- share)*Pf - (alpha/frcut))) - y[4]; /* dL/dfdmultl */ fd3 = Gf + ((l-share)*PPy[l]) + (zP P P (F d f - Fa)) - H - alpha*( y[2]+(y[l]/frcut)); /* dL/dfdmult2 */ fd4 = (betadj *y [2]A elasdj) - Fdf - y [1 ] + Fbar - Ftot; */ retp (fd 11fd 2 1 fd31 ; endp; altnam = { Frs, Ndjd, fdmultl, fdmult2 }; title = "Forest Department: Utility Maximization Case"; _nlmaxit = 500; output = 1; nlalgr = 2; _nlchpf= 1; if k <=2; yO = fdstart; nltypx = fdstart; else; yO = rfdstart; _nltypx = rfdstart; endif; {y,fdV,fdJ,tcode}=nlprt(nlsys(&fd,yO)); if tcode == 1; fdret = 0; else; fdret = tcode; endif; if tcode /= 1; fdfail = fdfail + 1; output file = jfm2results.out on; print; print "****** fy failed on iteration" k; print "****** fd" tcode; print; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 output file = jfm2results.out off; endif; Frs = y [1 ]; Ndjd = y [2]; fdmultl =y[3]; fdmult2 — y[4]; if tcode == 1; fdret = 0; rfdstart = Frs|Ndjd|fdmultl|fdmult2; else; fdret = tcode; fdfail = fdfail + 1; output file = jfm2results.out on; print; print "****** fy failed on iteration" k; print "****** fd" tcode; print; output file = jfm2results.out off; endif; /* output file = jfm2results.out on; */ fdprofit = G f + (l-share)*Pf*Frs + FrxDiff + zP P P (F d f - Fa) - alpha*( Ndjd + ((FrsA expfr)/frcut)) - H; fdtempu = A P ( betafl * (betadj *Ndj dA elasdj )A rhof + betaf2 * ((1 - share)*PPFrs)A rhof + betaG*GfA rhof )A (l/rhof); fdtempe = B P(alphafl*Fdf + alphaf2*Frs)A (expf); fdbiom = (betadj *NdjdA elasdj); print; print "fd utility" fdtempu; print "biomass production" fdbiom; print "externality" fdtempe; print; output file = jfm2results.out off; /* /*+++++++++++++++++++++++++++++++++++++++++++++++++ FOREST DEPT. PROFIT MAX CASE ++++++++++++++++++++++++++++++++++++++++++++++++ */ /* output file = jfm2results.out on; */ Ndjd = (alpha / ( elasdj * betadj *(( 1 - share) *Pf - (alpha/frcut)) ) ) A ( 1 /(elasdj -1)); Frs = (betadj *NdjdA elasdj) - Fdf + Fbar - Ftot; if Frs < 0; Frs = 0; endif; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 fdret = 0; print; p rin t"+++++++++++++++++++++ FOREST DEPT PROFIT MAX CASE + +++ + + + ++ + + + ++ + + ++ "; print; print "Ndjd" Ndjd; print "Frs" Frs; fdret = 0; fdtempu = G f + (1 - share)*PfH Frs + zf*Pf*(Fdf - Fa) - alpha*( Ndjd + ((FrsA expfr)/frcut)) - H; fdbiom = (betadj *NdjdA elasdj); fdtempe = Bf*(alphafl*Fdf + alphaf2*Frs)A (expf) ; print; print "fd profits" fdtempu; print "biomass production" fdbiom; print "externality" fdtempe; print; output file = jfm2results.out off; */ /* /*+++++++++++++++++++++++++++++++++++++++++++++++++ FOREST DEPT. PROFIT MAX CASE NLSYS Frs = y [1 ]; Ndjd = y [2]; Nmult = y [3]; Fmult = y[4]; ++++++++++++++++++++++++++++++++++++++++++++++++ * / proc fd(y); local fdl, fd2, fd3, fd4; /* dL/dFrs */ fdl = (1-share) *Pf - (alpha/frcut) - y[3]*(l/frcut) + y[4]; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 /* dL/dNdjd */ fd2 = -alpha - y[3] - y[4]*( elasdj *betadj*y[2]A (elasdj-l)); /* dL/dfdmultl */ fd3 = (y[l]/frcut) + y [2] + N df + Nreq - Ntot; /* dL/dfdmult2 */ fd4 = (betadj *y[2]A elasdj) - Fdf - y[l] + Fbar - Ftot; /* FOC'S EXCLUDING THE LABOR MARKET CONSTRAINT AND INCLUDING THE MULTIPLIER fdl = (l-share)*Pf - (alpha/frcut) + y [3]; fd2 = - alpha - y [3 ]* (elasdj* betadj *y [2] A (elasdj-1)); fd3 = (betadj *y[2]A elasdj) - Fdf - y[l] + Fbar - Ftot; FOC’S EXLUDING THE LABOR MARKET CONSTRAINT AND COMBINING THE MULTIPLIER fdl = - alpha - ((alpha/frcut) - (1-share)*Pf)*( elasdj*betadj*y[2]A (elasdj-l) ); fd2 = (betadj*y[2]A elasdj) - Fdf - y[l] + Fbar - Ftot; fdl = -alpha + ( (l-share)*Pf - (alpha/frcut) )*( elasdj*betadj*y[2]A (elasdj-l)); fd2 = (betadj *y [2]A elasdj) - Fdf - y [1 ] + Fbar - Ftot; FOC'S INCLUDING THE LABOR MARKET CONSTRAINT fdl = (l-share)*Pf - (alpha/frcut) - y[3]*(l/frcut) + y[4]; fd2 = -alpha - y [3] - y[41*( elasdj*betadj*y[2]A (elasdj-l)); fd3 = (y[l]/frcut) + y[2] + N df + Nrs - Ntot; fd4 = (betadj *y[2]A elasdj) - Fdf - y[l] + Fbar - Ftot; */ retp (fdl|fd2|fd3|fd4); endp; /* output file = jfm2results.out on; */ if k <=2; yO = fdstart; nltypx = fdstart; else; yO = rfdstart; nltypx = rfdstart; endif; altnam = { Frs, Ndjd, fdmultl, fdmult2 }; title = "Forest Department: Profit Maximization Case”; nlmaxit = 5000; output = 1; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. nlalgr = 2; nlchpf = 1; /* /* STARTING JACOBIAN */ let fd je l[l,l] = -1/share; let fdje2[l,l] = -elasdj*betadj*y[2]A (elasdj-l); let fdj 1 [1,2] = fdjel -1; . let fdj2[l,2] = 1 fdje2; fdJac = fdj 1 | fdj2; nlstjc = fdJac; output file = jfm2results.out on; print fdJac; output file = jfm2results.out off; nlmtol = 100; */ {y ,fdV,j dF, tcode} =niprt(nlsys(&fd,yO)); Ndjd = y [2]; if y[l] < 0; Frs = 0; else; Frs = y[l]; endif; fdmultl = y[3]; fdmult2 = y[4]; Frstest = (betadj *NdjdA elasdj) - Fdf + Fbar - Ftot; /* fdmultl = ((alpha/frcut) - (l-share)*Pf) ; */ print; p rin t" ++++++++++++++++++ FOREST DEPT. PROFIT MAX CASE print; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 print "Ndjd" Ndjd; print "Frs" Frs; print "Frstest" Frstest; print; print "fd resource multiplier" fdmultl; if tcode == 1; fdret = 0; rfdstart = Frs|Ndjd|fdmultl|fdmult2; else; fdret = tcode; fdfail = fdfail + 1; output file = jfm2results.out on; print; print "****** fd fajie(j on iteration" k; print "****** fd” tcode; print; output file = jfm2results.out off; endif; if px>pf; FrxDiff = ((l-share)*(Px-Pf)*Frx); else; FrxDiff = 0; endif; fdprofit = G f + (l-share)*Pf*Frs + FrxDiff + zP PP ^Fdf - Fa) - alpha*( Ndjd + ((FrsA expfr)/frcut)) - H; fdtempu = 0; fdbiom = (betadj *NdjdA elasdj); fdtempe = Bf*(alphafl*Fdf + alpha!2 * Fr s) A (expf); print; print "fd profits" fdprofit; print "biomass production" fdbiom; print "externality" fdtempe; print; output file = jfm2results.out off; */ /* ========================= RESIDUAL SECTOR PROBLEM: max Ur(Frd, Xr, Gr, Fdf+Fr) s.t. Pf*Fr + Xr = (Xtot-w*Nrd)( 1 -t) Xtot = betax * NrdA elasx profit = (Xtot-w*Nrd)(l-t) z[l] = Frd; z[2] = Nrd; z[3] = Xr; z[4] = rsmultl; z[5] = rsmult2; z[6] = rsmult3; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 proc rs(z); local rsl, rs2, rs3, rs4, rs5; /* dL/dFrd */ rsl = (l/rhor)*Ar*( (alpharl*z[l]A rhor + alphar2*z[3]A rhor + alphar3*GrA rhor )A ((l-rhor)/rhor)) * (rhor * alphar 1 * z [ 1 ] A (rhor-1)) - (Br * expP (alphaf 1 * Fdf + alphaf2*z[l])A (expf-l))*alphaf2 - z[4]*Pf; /* dL/dNrd */ rs2 = - z[4]*( -(1-t) * ( (betax*elasx*z[2]A (elasx-l)) - w ) ) + z[5]*( betax*elasx*z[2]A (elasx-l)); /* dL/dXr */ rs3 = (l/rhor)*Ar*( (alpharl *z[l]A rhor + alphar2*z[3]A rhor + alphar3*GrA rhor )A ((l-rhor)/rhor)) *( alphar2*rhor*z[3]A (rhor-l) ) -z[4] -z[5]; /* dL/drsmultl */ rs4 = z[3] + (Pf*z[l]) - (l-t)*( (betax*z[2]A elasx) - w *z[2]) ; /* dL/drsmult2 */ rs5 = z[3] + Xc - (betax*z[2]A elasx); /* OLD CASE THAT INCLUDES THE LABOR MARKET CONSTRAINT /* dL/dFrd */ rsl = (l/rhor)*Ar*( (alpharl*z[l]A rhor + alphar2*z[3]A rhor + alphar3*GrA rhor )A ((l-rhor)/rhor)) *(rhor*alpharl*z[l]A (rhor-l)) - (Br*expf*(alphafl*Fdf + alphaf2*z[l])A (expf-l))*alphaf2 - z[4]*Pf - z[6]*(l/frcut); /* dL/dNrd */ rs2 = - z[4]*( -(1-t) * ( (betax*elasx*z[2]A (elasx-l)) - w ) ) - z[5]*( -betax*elasx*z[2]A (elasx-l)) - z[6]; /* dL/dXr */ rs3 = (l/rhor)*Ar*( (alpharl*z[l]A rhor + alphar2*z[3]A rhor + alphar3*GrA rhor )A ((l-rhor)/rhor)) *( alphar2*rhor*z[3]A (rhor-l) ) -z[4] z[5]; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154 /* dL/drsmultl */ rs4 = z[3] + (Pf*z[l]) - (l-t)*( (betax*z[2]A elasx) - w *z[2]) ; /* dL/drsmult2 */ rs5 = z[3] + Xc - (betax*z[2]A elasx); /* dL/drsmult3 */ rs6 = (z[l]/frcut) + Ndjeq + N df + z[2] - N to t; END OF OLD CASE */ retp(rs 11 rs21 rs31 rs41 rs5); endp; /* output file = jfm2results.out on; */ if k <=2; zO = rsstart; nltypx = rsstart; else; zO = rrsstart; nltypx = rrsstart; endif; altnam = { Frd, Nrd, Xr, rsmultl, rsmult2 }; title = "Residual Sector: Non-linear equation solver"; nlmaxit = 500; output = 1; nlalgr = 2; nlchpf= 1; /* /* STARTING JACOBIAN */ let rsjel 1 [1,1] = elasx*betax*z[2]A (elasx-l)-t*elasx*betax*z[2]A (elasx-l); let rsj e 12 [ 1,1 ] = w-t*w; let rsj e 1 [ 1,1 ] = rsjel l-rsjel2; let rsje2[l,l] = elasx*betax*z[2]A (elasx-l); let rsj 1 [1,3] ~ -1 -Pf rsjel; let rsj2[l,3] = -10 rsje2; rsJac = rsjl | rsj2; _nlstjc = rsJac; output file = jfm2results.out on; print "RESIDUAL SECTOR JACOBIAN"; PRINT; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 155 print rsJac; PRINT; output file = jfm2results.out off; _nlm tol=100; */ {z,rs V ,r s J, tcode} =nlprt(nl sy s(&rs,zO)); Frd = z[l]; Nrd = z[2]; Xr = z[3]; rsmultl = z[4]; rsmult2 = z[5]; if tcode == 1; rsret = 0; rrsstart = Frd|Nrd|Xr|rsmultl|rsmult2; else; rsret = tcode; rsfail = rsfail + 1; output file = jfm2results.out on; print; print "****** rs failed on iteration" k; print "****** rc" tcode; print; output file = jfm2results.out off; endif; print; p rint"+++++++++++++++++++++++++ RESIDUAL SECTOR UTILITY MAX CASE +++++++++++++++++++++++++"; print; print "Frd" Frd; print "Nrd" Nrd; print "Xr" Xr; print "rsmultl" rsmultl; print "rsmult2" rsmult2; rstempu = Ar*(alpharl*FrdA rhor + alphar2*XrA rhor + alphar3*GrA rhor )A (1/rhor); rsbiom = (betadj *NdjeqA elasdj); rstempe = Bf*(alphafl*Fdf + alphaf2*z[l])A (expf) ? rsprofit = (l-t)*( (betax*NrdA elasx) - w *N rd) ; print; print "rs utility" rstempu; print "biomass production" rsbiom; print "externality" rstempe; print; output file = jfm2results.out off; /* THIS IS THE BIOMASS, EXTERNALITY AND FOREST GOOD PRODUCED IN THE PERIOD FTOT WILL BE RESET BEFORE NEXT RUN Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 156 */ Fadd = (betadj *NdjeqA elasdj); Fa = perbio*Fadd; if Fa>Fdf; Fa = Fdf; else; Fa = perbio*Fadd; endif; FaRev = Pf*zf*(Fdf-Fa); Fext = (Bf*(alphafl*Fdf + alphaf2 * Feq) A (expf)); Ftot = Fbar + Fadd - Fdf - Feq; if Nrd > Nrs; Xtot = (betax*NrsA elasx); else; Xtot = (betax*NrdA elasx); endif; LabAll = "Period kkm" $ - "Period kk" $~ "Period k" $~ "countl" $~ "Rundmy" $~ "alphacl" $~ "Nrs" $~ "Nrd" $~ "w" $~ "Frs" $~ "Frd" $~ "Frx" $~ "Pf' $~ "Ndjs" $~ "Ndjd" $~ "alpha" $~ "Gc" $~ "Gr" $~ "G f’ $~ "Xc" $~ "Xr" $~ "Xtot" $~ "Ndf' $~ "Fdf' $ - "Ftot" $~ "FD Profits" $~ "FD Utility" $~ "FD Biomass" $~ "FD Ext." $~ "FD Fdf_Rev" $~ "Penalty" $~ "zf' $~ "FC Biomass" $~ "FC Utility" $~ "FC Ext." $~ "RS Profit" $~ "RS Utility" $~ "RS Biomass" $~ "RS Ext." $~ "fcfail" $~ "fcret" $~ "fdfail" $~ "fdret" S~ "rsfail" $~ "rsret"; MA11 = MA111 kkm ~kk ~k -countl -rundmy - alphacl -N rs -N rd ~w -F rs -F rd -Frx -P f -N djs -N djd -alpha -G c -G r -G f -X c -X r -X tot -N d f -F d f -Ftot -fdprofit -fdtempu -fdbiom -fdtempe -Farev -perbio - z f -fcbiom -fctempu -fctempe -rsprofit -rstempu -rsbiom -rstempe -fcfail -fcret -fdfail -fdret -rsfail -rsret; output file = jfm2results.out reset; outwidth 256; els; output file = jfm2results.out on; print; print LabAll; print Mall; print; output file = jfm2results.out off; /*FINISH OFF LOOPS */ /* PUT OUTPUT FILES BACK HERE WHEN THINGS RUN */ diffnrl = Nrd - Nrs; diffnr2 = Nrs - Nrd; difffrl = (Frd+Frx) - Frs; difffr2 = Frs - (Frd+Frx); Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 157 diffndl = Ndjd - Ndjs; diffnd2 = Ndjs - Ndjd; if kk <=5; diffeq=.5; elseif kk>5; diffeq=. 3; endif; if kk>25; diffeq=.l; endif; if Px>Pf; FrxRev = (Px-Pf)* share* Frx; else; FrxRev=0; endif; /* #5 DO LOOP ======== (w) Nrd and Nrs ===== */ do while k < n; rundmy = 1; if Nrd > Nrs; /* if kkm == 4; diffeq = .5; endif; */ if diffnrl >= diffeq; if diffnrl >= 1; wage = w + ((Nrd-Nrs)/1.0e+6); else; wage = w + ((Nrd-Nrs)/1.0e+8); endif; /* output file = jfm2results.out on; */ print "inside do Nrd > Nrs " k; print "old value of wage " w; output file = jfm2results.out off; w = wage; Xtot = (betax*NrsA elasx); Nreq = Nrs; if Frs > (Frd+Frx); Feq = Frd+Frx; else; Feq = Frs; endif; if Ndjs > Ndjd; Ndjeq = Ndjd; else; Ndjeq = Ndjs; endif; G f = (l-betag-gammag)*(share*(PfH Feq) + t*(Xtot - (w*Nreq)) + FrxRev); Gr = gammag*(share*(Pf!Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Gc = betag*(share*(Pf*Feq)+t*(Xtot - (w*Nreq)) + FrxRev); /* output file = jfm2results.out on; */ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 158 print "Nrd =" Nrd; print "Nrs =" Nrs; print "difference =" diffnrl; print "updated wage (used next round)" w; print "value of n" n; print; output file = jfm2results.out off; else; n = -l; p = 1.0e+15; endif; elseifNrd <N rs; /* if kkm == 4; diffeq = .5; elseif kkm — 5; diffeq = .5; elseif kkm == 6; diffeq = .5; elseif kkm — 7; diffeq = .5; elseif kkm == 8; diffeq = .5; elseif kkm == 9; diffeq = .5; elseif kkm ==10; diffeq = .5; elseif kkm ==11; diffeq = .5; endif; */ if diffnr2 >= diffeq; if diffnr2 >= 1; wage = w - ((Nrs - Nrd)/1.0e+6); else; wage = w - ((Nrs - Nrd)/1.0e+8); endif; /* output file = jfm2results.out on; */ print "inside do Nrd < Nrs" k ; print "old value of wage " w; output file = jfm2results.out off; w = wage; Xtot = (betax*NrdA elasx); Nreq = Nrd; if Frs > (Frd+Frx); Feq = Frd+Frx; else; Feq = Frs; endif; if Ndjs > Ndjd; Ndjeq = Ndjd; else; Ndjeq = Ndjs; endif; G f = (1-betag- gammag)* (share* (Pf*Feq)+t* (Xtot - (w*Nreq)) + FrxRev); Gr = gammag*(share*(Pf*Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 159 Gc = betag*(share*(Pf|:Feq)+t*(Xtot - (w*Nreq)) + FrxRev); /* output file = jfm2results.out on; */ print "Nrd =" Nrd; print "Nrs =" Nrs; print "difference =" diffnr2; print "updated wage (used next round)" w; print "value of n" n; print; output file = jfm2results.out off; else; n = -1; p = 1.0e+15; endif; else; wage = w; Xtot = (betax*NrdA elasx); Nreq = Nrs; if Frs > (Frd+Frx); Feq = Frd+Frx; else; Feq — Frs; endif; if Ndjs > Ndjd; Ndjeq = Ndjd; else; Ndjeq = Ndjs; endif; G f = (l-betag-gammag)*(share*(Pfi < Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Gr = gammag*(share*(Pf*Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Gc = betag*(share*(PfK Feq)+t*(Xtot - (w*Nreq)) + FrxRev); n = -1; p = 1.0e+15; /* output file = jfm2results.out on; */ print "FINISHED inside do Nrd = Nrs" k ; print "value of n (= -1 when this finishes)" n; print "Nrd =" Nrd; print "Nrs =" Nrs; print "wage " w; print; output file = jfm2results.out off; endif; /* output file =jfm2results.out on; */ print; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 160 print "inside wage determination" k ; print "value of n" n; print "Nrd =" Nrd; print "Nrs =" Nrs; print "wage " w; print; output file = jfm2results.out off; break; endo; /* #5 DO LOOP — THIS IS FOR THE WAGE, Nr DO STATEMENTS */ do while k < p; /* #6 DO LOOP ======== (Pf) Frd and Frs ===== */ rundmy = 2; if Frs > (Frd + Frx); if difffr2 >= diffeq; if difffr2 >= 1; PriceF = Pf - ((Frs - (Frd+Frx))/(1.0e+3)); else; PriceF = Pf - ((Frs - (Frd+Frx))/(1.0e+3)); endif; /* output file = jfm2results.out on; */ print "inside do Frs > Frd " k; print "old value of P f " Pf; output file = jfm2results.out off; Pf = PriceF; if Pf < Px; Frx = fdxmult*(Px-Pf); else; Frx = 0; endif; Feq = (Frd + Frx); endif; if Ndjs > Ndjd; Ndjeq = Ndjd; else; Ndjeq = Ndjs; endif; if Nrs < Nrd; Nreq = Nrs; elseif Nrs > Nrd; Nreq = Nrd; Xtot = (betax*NreqA elasx); G f = (l-betag-gammag)*(share*(Pf|!Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Gr = gammag*(share*(PfH Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. FrxRev); else; 161 Gc = betag*(share*(Pf*Feq)+t*(Xtot - (w*Nreq)) + output file = jfm2results.out on; print "Frd =" Frd; print "Frs =" Frs; print "Frx =" Frx; print "difference =" difffr2; print "new price of F (used next round)" Pf; print "value of p" p; print; output file = jfm2results.out off; p = -l; o = 1.0e+15; endif; elseif Frs < (Frd + Frx); if difffrl >= diffeq; if difffrl >= 1; PriceF = Pf + (((Frd+Frx) - Frs)/(1.0e+3)); else; PriceF = Pf + (((Frd+Frx) - Frs)/(1.0e+3)); endif; /* output file = jfm2results.out on; */ print "inside do Frs < Frd " k; print "old value of P f " Pf; output file = jfm2results.out off; Pf = Pricef; if Pf < Px; Frx = fdxmult*(Px-Pf); else; Frx = 0; endif; Feq = Frs; if Ndjs > Ndjd; Ndjeq = Ndjd; else; Ndjeq = Ndjs; endif; if Nrs < Nrd; Nreq = Nrs; elseif Nrs > Nrd; Nreq = Nrd; endif; Xtot = (betax*NreqA elasx); G f = (l-betag-gammag)*(share*(PPFeq)+fK (Xtot - (w*Nreq)) + FrxRev); Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 162 Gr = gam m ag*(share*(Pf!Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Gc = betag*(share*(PfTeq)+t*(Xtot - (w*Nreq)) + FrxRev); /* output file = jfm2results.out on; */ print "Frd =" Frd; print "Frs =" Frs; print "difference =" difffrl; print "new price of F (used next round)" Pf; print "value of p" p; print; output file = jfm2results.out off; else; p = -l; o = 1.0e+15; endif; else; if Ndjs > Ndjd; Ndjeq = Ndjd; else; Ndjeq = Ndjs; endif; if Nrs < Nrd; Nreq = Nrs; elseif Nrs > Nrd; Nreq = Nrd; endif; Xtot = (betax*NreqA elasx); G f = (l-betag-gammag)*(share*(Pf|!Feq)+f|!(Xtot - (w*Nreq)) + FrxRev); Gr = gammag*(share*(Pfi:Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Gc = betag*(share*(PP!Feq)+t*(Xtot - (w*Nreq)) + FrxRev); P = -i; o = 1.0e+15; /* output file = jfm2results.out on; */ print "FINISHED inside do Frs = Frd " k; print "Frd =" Frd; print "Frs =" Frs; print "price of F" Pf; print; "value of p" p; print; output file = jfm2results.out off; endif; /* output file = jfm2results.out on; */ print; print "inside price of F determination" k ; print "value of p (= -1 when finished)" p; print; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 163 output file = jfm2results.out off; break; /* this will break to the end of the do statement */ endo; /* #6 DO LOOP = = THIS IS FOR THE Price of Forest Good, Fr DO STATEMENTS */ do while k < o; /* #7 DO LOOP ======== (ALPHA) Ndjd and Ndjs ===== * rundmy = 3; if Ndjd > Ndjs; Ndjeq = Ndjs; if diffndl >= diffeq; if diffndl > = .5 ;/* .5 */ pricedj = alpha + ((Ndjd-Ndjs)/(1.0e+3)); else; pricedj = alpha + ((Ndjd-Ndjs)/(1.0e+4)); endif; /* output file = jfm2results.out on; */ print "inside do Ndjd > Ndjs " k; print "old value of alpha " alpha; output file = jfm2results.out off; alpha = pricedj; if Frs > (Frd+Frx); Feq = Frd+Frx; else; Feq = Frs; endif; if Nrs < Nrd; Nreq = Nrs; elseif Nrs > Nrd; Nreq = Nrd; endif; (w*Nreq)) + FrxRev); FrxRev); FrxRev); Xtot = (betax*NreqA elasx); G f = (l-betag-gammag)*(share*(PPFeq)+t*(Xtot - Gr = gammag*(share*(Pf!Feq)+t*(Xtot - (w*Nreq)) + Gc = betag*(share*(Pf*Feq)+t*(Xtot - (w*Nreq)) + /* output file = jfm2results.out on; */ print "Ndjd =" Ndjd; print "Ndjs =" Ndjs; print "difference =" diffndl; print "new price of dj (used next round)" alpha; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 164 print "value of o" o; print; output file = jfm2results.out off; else; o = -l; m = 0; endif; elseifNdjd <Ndjs; Ndjeq = Ndjd; if diffnd2 >= diffeq; if diffnd2 >= .5; /* .5 */ pricedj = alpha - ((Ndjs - Ndjd)/(1.0e+3)); else; pricedj = alpha - ((Ndjs - Ndjd)/(1.0e+4)); endif; /* output file = jfm2results.out on; */ print "inside do Ndjd < Ndjs " k; print "old value of alpha " alpha; output file = jfm2results.out off; alpha = pricedj; if Frs > (Frd+Frx); Feq = Frd+Frx; else; Feq = Frs; endif; if Nrs < Nrd; Nreq = Nrs; elseif Nrs > Nrd; Nreq = Nrd; endif; Xtot = (betax*NreqA elasx); G f = (1-betag- gammag)* (share* (Pf*Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Gr = gammag*(share*(Pf*Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Gc = bctag*(share*(Pf*Feq)+t*(Xtot - (w*Nreq)) + FrxRev); /* output file = jfm2results.out on; */ print "Ndjd =" Ndjd; print "Ndjs =" Ndjs; print "difference =" diffnd2; print "new price of dj (used next round)" alpha; print "value of o" o; print; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 165 output file = jfm2results.out off; o = -1; m = 0; endif; Ndjeq = Ndjs; if Frs > (Frd+Frx); Feq = Frd+Frx; else; Feq = Frs; endif; if Nrs < Nrd; Nreq = Nrs; elseif Nrs > Nrd; Nreq = Nrd; endif; Xtot = (betax*NreqA elasx); G f = (l-betag-gammag)*(share*(Pf|!Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Gr = gammag*(share*(Pf|:Feq)+t*(Xtot - (w*Nreq)) + FrxRev); Gc = betag*(share*(Pf*Feq)+t*(Xtot - (w*Nreq)) + FrxRev); o = -1; m = 0; /* output file = jfm2results.out on; */ print "inside do Ndjd = Ndjs " k; print "value of o" o; print "Ndjd =" Ndjd; print "Ndjs =" Ndjs; print "price of dj - alpha" alpha; print; output file = jfm2results.out off; endif; /* output file = jfm2results.out on; */ print; print "inside alpha determination" k ; print "value of o" o; print; output file = jfm2results.out off; break; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 166 endo; /* #7 DO LOOP ==== THIS IS FOR THE ALPHA, Ndj DO STATEMENTS */ /* output file = jfm2results.out on; */ p rin t" ; print; prpricesl = "Prices" $~ "alpha" $~ "W" $~ " P f ; prprices2 = 0 ~ alpha ~ w ~ P f ; print prpricesl; print prprices2; print; prresidl = "Residual Sector" $~ "" $~ "Nrd" $~ "Frd"; prresid2 = 0 ~ 0 ~ Nrd ~ Frd ; print prresidl; print prresid2; print; prforl = "Forest Dept." $~ "Ndjd" $~ " " $~ "Frs"; prfor2 = 0 ~ Ndjd ~ 0 ~ Frs ; print prforl; print prfor2; print; prdwelll = "Forest Dwellers" $~ "Ndjs" $~ "Nrs" $~ "Fdf' $~ "Ndf' ; prdwell2 = 0 ~ Ndjs ~ Nrs ~ Fdf ~ N d f; print prdwelll; print prdwell2; print; print; prgovl = "GovernmentGrant" $~ "Gc" $~ "Gr" $~ "Gf'; prgov2 = 0 ~ Gc ~ Gr ~ Gf; print prgovl; print prgov2; print; prxtotl = "BP & Ext" $~ "FD Biomass" $~ "FD Ext." $~ "FC Biomass" $~ "FC Ext." $~ "RS Biomass" $~ "RS Ext."; prxtot2 = 0 ~ fdbiom ~ fdtempe ~ fcbiom ~ fctempe ~ rsbiom ~ rstempe; print prxtotl; print prxtot2; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 167 print; prxtotl = "Total Market Good" $~ "Xtot" $~ "Xc" $~ "Xr"; prxtot2 = 0 ~ Xtot ~ Xc ~ Xr; print prxtotl; print prxtot2; print; print; p rin t" ====================================== print; /* PerK = k; LabFr = "Period k" $~ "Frs" $~ "Frd" $~ "P f; MFr = MFr | k~Frs~Frd~Pf; LabNdj = "Period k" $~ "Ndjs" $~ "Ndjd" $~ "alpha"; MNdj = MNdj | k~Ndjs~Ndjd~alpha; LabG = "Period k" $~ "Gc" $~ "Gr" $~ " G f ; MG = MG | k~Gc~Gr~Gf; LabX = "Period k" $~ "Xc" $~ "Xr" $~ "Xtot"; MX = MX | k~Xc~Xr~Xtot; LabMisc = "Period k" $~ "N df $~ "F df; MMisc = MMisc | k~Ndl~Fdf; */ /* if k > 5; print "graph Pf and Frs"; plot(MFrs, MPf); print; print; endif; */ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 168 print; prftotl = "Total Forest Good" $~ "Fadd" $~ "Fext" $~ "Ftot" "Fdf’ $~ "Frs" $~ "Frd"; prftot2 = 0 - Fadd ~ Fext ~ Ftot ~ Fdf ~ Frs ~ Frd; print prftotl; print prftot2; print; k; k = k + 1; /* if k == 25; end; endif; */ endo; /* 1 .Oe-01 #4 DO LOOP */ if diffnrl >= .2; mm = 1.0e+15; elseif diffnr2 >= .2; mm = 1.0e+15; elseif difffrl >= .2; mm = 1.0e+15; elseif difffr2 >= .2; mm = 1.0e+15; elseif diffndl >= .2; mm = 1.0e+15; elseif diffnd2 >= .2; mm = 1.0e+15; else; mm = 0; endif; /* output file = jfm2results.out on; */ print; print "value of mm is =" mm; print "value of kk is =" kk; print; output file = jfm2results.out off; kk; kk = kk + 1; endo; /* #3 do loop */ LabTAll = "Period k" $~ "perbio" $~ "zf’ $ - "countl" $~ "Alphacl" $~ "sigmac" $~ "alphafl" $~ "sigmaf' $~ "Be" $~ "Br" $~ "Bf' $~ "Rundmy" $~ "Nrs" $~ "Nrd" $~ "w" $~ "Frs" $~ "Frd" "Frx" $~ " P f $~ "Ndjs" $~ "Ndjd" "alpha" $~ "Gc" $~ "Gr" $~ "G f $~ "Xc" $~ "Xr" $~ "Xtot" $~ "Ndf' $~ "Fdf' $~ "Ftot" $~ "Fbar" $~ "FD Profits" $~ "FD Utility" $~ "FD Biomass" $~ "FD Ext." $~ "FD Fdf Rev" $~ "Penalty" $~ "zf' $~ "FC Biomass" $~ "FC Utility" $~ "FC Ext." $~ "RS Profit" $~ "RS Utility" $~ "RS Biomass" $~ "RS Ext." $~ "fcfail" $~ "fcret" $~ "fdfail" $~ "fdret" $~ "rsfail" $~ "rsret"; TA11 = TA111 k -perbio - z f -countl -alphacl -sigmac -alphafl -sigm af -B e -B r -B f -rundmy -N rs -N rd ~w Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 169 -F rs -F rd -Frx - P f -N djs -N djd -alpha -G c -G r -G f -X c -X r -X tot -N d f -F d f -Ftot -Fbar -fdprofit -fdtempu -fdbiom -fdtempe -Farev -perbio - z f -fcbiom -fctempu -fctempe -rsprofit -rstempu -rsbiom -rstempe -fcfail -fcret -fdfail -fdret -rsfail -rsret; /* put tall at beginning */ /* if alphacl == .4; sigmac=.8; alphacl = .5; alphac2=.2; alphac3=.3; mmm = 1.0e+15; fcstartl = {54,80,47,39,.04,-.2,-.02,1}; /* worked with profit max case */ fcstart - {54,80,47,39,.04,-.2,-.02,1}; rfcstart = {54,80,47,39,.04,-.2,-.02,l}; */ if alphacl == .4; alphacl = .45; alphac2=.25; alphac3=.3; mmm = 1.0e+15; output file = jfmchanges.out on; outwidth 256; print LabTall; print Tall; output file = jfmchanges.out off; output file = jfm2results.out reset; outwidth 256; els; MA11 = 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~0 ~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0 ; elseif alphacl == .45; alphacl = .5; alphac2=.225; alphac3=.275; mmm = 1.0e+15; output file = jfmchanges.out on; outwidth 256; print LabTall; print Tall; output file = jfmchanges.out off; output file = jfm2results.out reset; outwidth 256; els; MA11 = o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o ~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0 ; /* NORMAL CASE elseif alphacl == .5; alphacl = .55; alphac2=.2; alphac3=.25; mmm = 1.0e+15; output file = jfmchanges.out on; outwidth 256; print LabTall; print Tall; output file = jfmchanges.out off; output file = jfm2results.out reset; outwidth 256; els; MA11 = 0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0 ~ 0~ 0~ 0~ 0- 0~ 0~ 0~ 0~ 0~ 0- 0~ 0~ 0 ; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 170 elseif alphacl == .55; alphacl = .6; alphac2=.175; alphac3=.225; mmm = 1.0e+15; output file = jfmchanges.out on; outwidth 256; print LabTall; print Tall; output file = jfmchanges.out off; output file = jfm2results.out reset; outwidth 256; els; MA11 = 0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0 ~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0 ; */ /* NEW ADD ON */ elseif alphacl == .5; alphacl = .6; alphac2=.175; alphac3=.225; mmm = 1.0e+15; output file = jfmchanges.out on; outwidth 256; print LabTall; print Tall; output file = jfmchanges.out off; output file = jfm2results.out reset; outwidth 256; els; MA11 = 0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0 ~0~0~0~0~0~0~0~0~0~0~0~0~0; elseif alphacl == .6; alphacl = .65; alphac2=.15; alphac3=.2; mmm = 1.0e+15; output file = jfmchanges.out on; outwidth 256; print LabTall; print Tall; output file = jfmchanges.out off; output file = jfm2results.out reset; outwidth 256; els; MAU = o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o ~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0 ; elseif alphacl == .65; alphacl = .7; alphac2=.125; alphac3=.175; mmm = 1.0e+15; output file = jfmchanges.out on; outwidth 256; print LabTall; print Tall; output file = jfmchanges.out off; output file = jfm2results.out reset; outwidth 256; els; MA11 = 0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0~0 ~0~0~0~0~0~0~0~0~0~0~0~0~0; elseif alphacl == .7; alphacl = .75; alphac2=.l; alphac3=.15; mmm = 1.0e+15; output file = jfmchanges.out on; outwidth 256; print LabTall; print Tall; output file = jfmchanges.out off; output file = jfm2results.out reset; outwidth 256; els; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. elseif alphacl == .75; alphacl = .8; alphac2=.075; alphac3=.125; mmm = l.Oe+15; output file = jfmchanges.out on; outwidth 256; print LabTall; print Tall; output file = jfmchanges.out off; output file = jfm2results.out reset; outwidth 256; els; MA11 = o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o~o ~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0~ 0 ; elseif alphacl == .8; alphacl = .8; alphac2=.075; alphac3=.125; mmm = 0; endif; kkm; kkm = kkm + 1; endo; /* #2 do loop === associated with kkm */ output file = jfmchanges.out on; outwidth 256; print LabTall; print Tall; output file = jfmchanges.out off; print; print; print "end of program"; print; print; print; output file = jfm2results.out on; print; print M print; print "LAST YEAR IS" k; print; print; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 172 print "Forest Community Failures:" fcfail; print "Forest Department Failures:" fdfail; print "Residual Sector Failures;" rsfail; print; /* output file = jfmchanges.out on; outwidth 256; els; print LabTall; print Tall; output file = jfmchanges.out off; */ /* alphacl = .4; alphac2=.3; alphac3=.3; alpha=l; Pf=1.5; w =l; Nreq=Nreql; Ndjeq=Ndjeql; Xr=Xrl; Feq=Feql; Frs=Frsl; Frx=Frxl; Gf=Gfl; Gc=Gcl; G r=G rl; fdstart^fdstartl; rsstart=rsstartl; fcstart=fcstartl; */ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Health interventions, health status, and labor market outcomes in Indonesia: Three essays

PDF

How financial statements in Mexican firms reflect changes in the financial/economic environment from 1978 to 1996

PDF

Essays on regulation of public utilities and the provision of public goods

PDF

Buscando la prosperidad: Migration as long -term investment in El Salvador

PDF

Asymmetry and risk aversion in first -price sealed bid auctions: Identification, estimation, and applications

PDF

Three essays on the determinants of firm structure and contractual choice

PDF

Essays on auctions

PDF

Individual heterogeneity and program evaluation

PDF

Production Functions In Korean Manufacturing: An Analysis Of Trade Policyand Its Effect On Technological Change

PDF

Essays on consumption behavior, economic growth and public policy

PDF

Dual labor markets, public debt management, and exchange rate movements

PDF

Essays on exchange rates, prices, and corporate behavior during the Asian currency crisis

PDF

Income inequality and economic growth: A theoretical and empirical analysis

PDF

Intellectual property rights, quality control and incentives

PDF

Endogenous entry equilibrium in auctions and markets with standards

PDF

An enquiry into the effectiveness of export assistance organizations serving southern California SMEs

PDF

Dependence on foreign labor, quality of education and unemployment in the GCC countries: In search of solutions

PDF

Essays on agrarian institutions

PDF

Complex economic growth

PDF

Business group affiliation in India

Asset Metadata

Creator Howe, Elbert Lance (author)

Core Title Insurance mechanisms, forest clearance, and the effect of government policies in rural economies

School Graduate School

Degree Doctor of Philosophy

Degree Program Economics

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

Tag Economics, Agricultural,Economics, General,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Nugent, Jeffrey B. (committee chair), Day, Richard H. (committee member), Jeong, Dr. (committee member), Pendleton, Dr. (committee member), Perrigne, Isabelle (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-252113

Unique identifier UC11339414

Identifier 3093772.pdf (filename),usctheses-c16-252113 (legacy record id)

Legacy Identifier 3093772.pdf

Dmrecord 252113

Document Type Dissertation

Rights Howe, Elbert Lance

Type texts

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

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA