-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~- - o U =~~~~~~~ _ *_ r , m2 r ''-V t ^ - _~~~~~~~~~~~* Us ? fD~~~~/ -'~~~~~ k I ':1. r/S k H, .,^IW 1 ~~~~~~~~~~~~ 3~ Imm w N: | v _ _ _ _ _ _ _ _ _ _ _ _ _ _ _._ _ _ _ _ _ _ _ _ _ _ _ _ _,'_ _ _-_ _ _ _ _ _ _ _ _ _ _................................................................................................. ,] Sustainable Settlement in the Brazilian Amazon Anna Luiza Oz6rio de Almeida* and Joao S. Campari** Oxford University Press OXFORD NEWV YORK TORONTO DELHI BOMBAY CALCUTTA MADR_AS KACHI KUALA LUMPUR SINGAPORE HONG KONG TOKYO NAIROBI DAR ES SALAAAI CAPE TOWN MELBOURNE AUCKLAND and associated companies in BERLIN IBADAN © 1995 The International Bank for Reconstruction and Development / THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. Published by Oxford University Press, Inc. 200 Madison Avenue, New York, N.Y. 10016 Oxford is a registered trademark of Oxford Universitv Press. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or othervise, without the prior permission of Oxford University Press. Manufactured in the United States of America First printing December 1995 The late Anna Luiza Oz6rio de Almeida was a staff member in the World Bank's Education and Social Policy Department at the time of the writing of this volume. Her co-author, Joao Campari, is a Ph.D. candidate in the Economics Department of the University of Texas at Austin. The findings, interpretations, and conclusions expressed in this study are entirely those of the authors and should not be attributed in anv manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. Lilnbaiy of Congress Cataloging-in-Publication Data Almeida, Anna Luiza Oz6rio de. Sustainable settlement in the Brazilian Amazon / Alna Luiza Oz6rio de Almeida and Joao S. Campari. p. cm. Includes bibliographical references. ISBN 0-19-521104-9 1. Land settlement-Amazon River Region. 2. Land settlement- Brazil. 3. Sustainable agriculture-Amazon River Region. 4. Sustainable agriculture-Brazil. 5. Deforestation-Amazon River Region. 6. Deforestation-Brazil. I. Campari,Joao S. II. International Bank for Reconstruction and Development. III. Title. HD499.A44A457 1995 333.3'181 1-dc2O 95-10600 CIP Contents Acknowledgments iv Foreword vi 1. Introduction and Principal Findings I 2. Population Shifts in the Amazon during the 1980s 11 3. Economic Shifts in the Amazon during the 1980s 21 4. Land Markets and Sustainable Frontier Farming 37 5. Productive and Speculative Deforestation by Frontier Small Farmers 63 6. Policy Implications: Institutional Improvements for Sustainable Settlement in the Amazon 75 Appendixes: A. Descriptive Tables 87 B. Empirical Procedures 133 C. A Model of Amazonian Deforestation 139 D. Econometric Results 151 Bibliography 169 Index 185 iii Acknowledgments 'T'his work began as a report for the position of Visiting Research Fellow in the Latin America and the Caribbean Environment Division of the World Bank from April to August 1992. Its objective is to discuss policy implica- tions for the World Bank derived from empirical analysis of a panel of small farmers surveyed in 1981 and 1991 in directed colonization projects in the Brazilian Amazon. Nancy Birdsall extended the original inVitation for this research to be done at the World Bank. Dennis Mahar and Oey Astra Meesook provided encouragement and institutional support for the project. Other members of the Bank staff who were especially generous with their time and criticism were Sergio Margulis, Carlos Prinio Braga, and Robert Schneider. Hans Binswanger made many helpful comments, as did Michele de Nevers, John Dixon, Philip Hlazelton, and William Partridge. Members of the Agricul- ture and Natural Resources, Agricultural Policies Division also gave helpful suggestions, especially Flarold Alderman, Jock Anderson, Ant6nio Salazar Brandao, and Gershon Feder. Daniel Gross, Robert Kaplan,John Redwood, Alfredo Sfeir-Younis, and Shekhar Shah also helped us deal with different aspects of this research. Meetings at the World Bank of the Brazil Agricultural and Resource Policy workshop, led by Malcolm Bale and Kreszentia Duer, provided stimulating discussions on many of the topics addressed in this text. In addition, many presentations were made outside the World Bank at the invitation of Michael Conroy and Dan Slesnick (Department of Eco- nomics, the University of Texas at Austin); Betsy Kuznesof (University of Kansas); Richard Reed (Department of Anthropology, Trinity University, San Antonio, Texas); Ruibens Ricuipero (Brazilian Embassy); Gert Rosenthal (Economic Commission for Latin America and the Caribbean); Margaret Sarles (U.S. Foreign Service Institute); and Charles Wood (Population Center and Department of Sociology, the University of Texas at Austin), all of whom provided stimulating discussions and highlighted important aspects of this research. i.: .A1ckI7o-7edirnents v Informal discussions with David Feeny (Centre for Health Economics and Policv Analvsis at McMaster Universitv); Richard Graham (the IJniver- sitv of Texas at Austin); Donald Ilarris (Stanford LUniversitv); Juan C(arlos Lerda (Economic Commission for Latin America and the Caribbean); I lenrique Monteiro de Barros (Organization of American States); Fredricka Santos andJohn "'ilson (AVorld Bank); C. Peter Timmer (Harvard Institute for International Development); Steve Vosti (International Food Policy Research Institute); and Peterj. Wilcoxen (Department of Economics, the University of Texas at Austin) helped sharpen different parts of the argu- ment. Students in the course "The Economics of the Amazon" at the University of Texas at Austin provided manv insightful criticisms and suggestions. 'WVe are also grateful to the reviewers for their insightful comments. Several research assistants prepared the core factual basis for this research. At the WVorld Bank, Pamela Stedman researched and organized an extensive annotated bibliographv, while Alex Panagides did the painstaking work of splicing together the 1981 and 1991 samples, figuring out and running countless tests and estimations, and putting together the statistical appen- dices to this report. At the Institute of Applied Economic Research (IPEA) of the Brazilian Planning Ministrv. Adriana Alves, Gustavo Gontijo. Manuel Augusto Magina. Paulo Arthur Moneto, Paulo Sergio Monteiro, Carlos Ozorio de Almeida. Flivio Paim Freaza, Maria da Piedade Morais, and Luciano Sobrinho Porto, plus a large team of field researchers, organized the survey research and questionnaires and performed the basic program- ming on the data base for 1991. Angela Moulin Penalva Santos, although conducting research of her own, took over the management of data retrieval for this project at IPEA. She also graciously provided special tabulations on frontier merchants for this work. Sergei Soares provided valuable unpub- lished information on fiscal revenues and transfers from federal to state and municipal governments in Brazil during the 1980s. Carlos Eduardo F. Young made detailed comments and suggestions on an earlier version, incorporated into this one. Roberto das Chagas Campos, Diva Rodrigues de Mattos, and Carlaile Pina Meireles at IPEA aided in typing, logistics, and communications. At the World Bank, Maria Abundo, Laura Alvarez, Chris- tos Georgiou, an(i Bobak Rezaian walked us safelv through the perilous jungle of office technology. We are also grateful to the small farmers who endured our relentless questioning, as well as to all the kind people who helped us in the Amazon itself. WVe alone are responsible for anv errors and omissions in the text and would welcome readers' comments and criticisms. Foreword This book is about small farmers in the Brazilian Amazon and how to raise their incomes while reducing resource degradation. The starting point of the book is the observation that whereas those farmers who deforested the Amazon during the 1970s and 1980s had migrated there from outside the basin, most of the small farmer deforesters of today have apparently come from within the region. Hence the greatest threat from small farmers to the Amazon forest is coming from intraregional migration. The central paradox is this: unsuccessful farmers tend to be unstable, selling out and moving on to new frontiers where they will deforest again. Parallel to this, successful farmers are led to increase deforestation in the place where they remain.The issue is thus no longer how to prevent migration to the Amazon from the rest of the country but how to ensure that farmers already in the Amazon stay where thev have already deforested, thus reducing migration and deforestation farther inland. The main argument in the book is that intra- regional migration can be stemmed only by establishing sustainable farming in already deforested areas. Such sustainabilitv would not likely attract interregional migrants because of irreversible changes that occurred during the 1980s in Brazil's population and economy. A deliberate policy of Amazonian settlement was initiated in Brazil during the earlv 1970s in response to massive dislocations of poor farmers because of drought in the northeast and expulsion of tenants, squatters, and small landowners in the south. Over the next two decades, vast areas were deforested. In the 1990s a large part of the original settlers of the 1970s have left their plots, which are now increasingly owned by the local urban middle class. Some of those who left profited from the sale of their lands, but many were soon as poor as when they started. Thus, although much of Amazon deforestation has been done by small farmers, they have not necessarily kept the land they cleared. Small farmers are encroaching upon Indian and forest reserves set up by federal and local governments-some with support from the World Bank, other international organizations, and nongovernmental organizations- and have even reached the political boundaries of Brazil, spilling over into Forevrord vii neighboring countries. One approach for those concerned with environ- mental conservation would be to support obstructions to the migratory flow. Such conservationists may oppose infrastructure that is necessarv for sus- tainable farming systems, such as roads and energy, social overhead invest- ments, and anv form of directed settlement. Thev mav also encourage the establishment of forest reserves and increasingly protect them from en- croachment. This approach has limitations, however. It does not address what is causing small farmers to leave old frontiers, and it bars them from the new ones they desperately need. Increasing the sustainability of settlement in old Amazonian frontiers requires learning from the errors of the past, which led to unsustainable settlements, and avoiding those of the present, which oppose any new settlement. Based on original field panel data spanning two decades and representing the largest and most comprehensive data set ever produced on the economic variables that influence small farmer deforestation in the Amazon, this book makes a major contribution to the debate by outlining a policy prescription to increase the sustainability of small farming in the already settled parts of the Amazon. This approach amounts to a frontier poverty reduction program that uses economic policies to raise agricultural incomes and reduce environmental aggression. It requires environmental agencies to share in the design and enforcement of economic policy instru- ments and economic agencies to support environmental objectives. None of this can be achieved without the cooperation, and willingness to change, of all involved agencies, local and central, national and international, and governmental and nongovernmental, including the World Bank. Given the acceleration of intra-Amazonian migrations, institutional change is increas- ingly urgent. This book was the last major piece of work that Anna Luiza Oz6rio de Almeida completed before she died in September 1994. As important as this book is, I want its readers to know how much of herself Anna Luiza gave to this book and how much it meant to her when she heard in June 1994 that it had been accepted for publication. I still have the electronic mail that she sent me from the hospital on August 9, 1994, saying how happy she was that the final manuscript had been approved and asking me to write the foreword for the book. Anna Luiza was a very special person, and we were fortunate to have known her and to have worked with her. And she also collaborated on this book with someone special-Joao Campari-who has had to carry the final burden of seeing the manuscript to press. Oey Astra Meesook lManager, Poverty Analysis and Social Assistance Poveiy and Social Policy Department The World Bank July 1995 Acronyms and Abbreviations bdias-frias poor urban dwellers who earn their living working in the fields betifeitorias physical improvements made on farms cabloclo type of indigenous technology CELADE Centro Latinoamericano de Demografia CPD Centro de Processamento de Dados DPE Diretoria de Pesquisas (de IBGE) FAO Food and Agricultural Organisation of the UN FPM Fundo de Participac,o dos Municipios GEF Global Environment Facilitv GTA Gr(upo de Trabalho da Amazonia IBAMA Instituto Brasileiro do Meio Ambiente e Recursos Naturais Renoviveis IBGE Fundacao Instituto Brasileiro de Geografia e Estatistica imediatis7no high rates of time preference of small farmers INCRA Instituto Nacional de Colonizaq5o e Reforma Agraria INPA Instituto Nacional de Pesquisas Amaz6nicas IPEA Instituto de Pesquisa Econ6mica Aplicada MARA Ministerio da Agricultura e Reforma Agraria MIRAD Ministerio da Reforma Agraria e Desenvolvimento miniflindiarios small landholders NGOs nongovernmental organizations VXiii Acr-on-y7ms anyd Abbr7eviations ix PNUD/BRA Programna das Na,6es Unidas para o Desenvolvimento/ Brasil Polonoroeste Programa de Desenvolvimento Integrado do Noroeste do Brasil (integrated development program for northwest region of Brazil) PROBOR Programa de Incentivo a Produ,co de Borracha SAFEAI Sistema de Acompanhamento das Finan,as dos Estados e Municfpios SIVAM Servico de Informac,o e Vigilincia da Amaz6nia ULFRJ/USP Universidade Federal do Rio de Janeiro/Universidade de Sio Paulo UNDP United Nations Development Programme CHAPTER 1 Introduction and Principal Findings The current migration of small farmers from old to new frontiers within the Amazon poses an important threat to the forest. In the long run, such intraregional migration-which is increasing and spilling out over Brazil's borders-will eventually put pressure on forest reserves now being estab- lished in the Amazon by local and federal governmental agencies. Some of these reserves are being organized with substantial international resources and World Bank support. In order to protect forest reserves from future encroachment, agriculturally sustainable frontiers must be promoted in conjunction with direct forest conservation projects. Speculative demand for land by a growing, prosperous Amazonian middle class is one of the important forces impelling growing numbers of small farmers to sell their plots and move on to settle temporarily and deforest new frontiers. In response to the national economic crisis during the 1980s, an emergent, local, urban middle class increasingly began buying frontier land-originally deforested by the pioneers of the 1 970s-for the purpose of holding it as a store of value. The potential gains from speculation in the frontier land markets came to compete, therefore, with the land's agricul- tural potential, and few small farmers withstood the pressure to sell. Those who did not were mainly the outstandingly productive, whose profits were plowed back into expanding agriculture, and the outstandingly indebted, whose repayment obligations locked them into their shrinking farms. Small farmers in the Amazon have been responding to economic condi- tions that reward speculation and encourage deforestation. Without new policies designed to change these conditions, therefore, the 1990s will probably wvitness continued deforestation, fueled by growing intraregional expulsions and migrations. The basic policy prescription that emerges from this work is that Amazonian deforestation can be reduced by inverting current economic rewards for speculation and intraregional migration so that farmers are discouraged from selling their cleared land and moving on to clear forest farther inland and from deforesting more extensively their 1 2 Suestainable Settlement in the Brazilian Amnzon current plots. In this book, a prescription that promotes agriculturallv sustainable frontiers is combined with the judicious use of pricing and fiscal policies in order to elevate agricultural productivitv, punish speculation in land transactions, and penalize deforestation directiv in the Brazilian Ama- zon. Ideallv, such policies must be created and enforced at the local level and be flexible enough to deal with a variety of Amazonian circumstances. (The cost of decentralization would be efficiencv variations across loca- tions.) The policies should also generate sufficient local revenues to pav for their enforcement, as well as for the expenses of direct conservation inea- sures. such as demarcation and protection of native Indian, extractive, biological, and other reserves. New local taxes on capital gains, agricultural income (from farms above a certain size), fallow land, and stumpage could be established toward this end. TIhe federal government could serve in a coordinating capacity, and the World Bank, other international organiza- tions, and nongovernmental organizations (NGOs) could provide technical and financial assistance. Based on original recent field data, this work outlines a policv prescription to increase the sustainability of small farming in the alreadv settled parts of the Amazon. This approach amounts to a frontier poverty reduction pro- gram that uses economic policies to raise agricultural incomes and reduce environmental aggression. It requires environmental agencies to share in the design and enforcement of economic policy instruments and economic agencies to support environmental objectives. None of this can be achieved without the cooperation, and willingness to change, of all involved agencies, local and central, national and international, and governmental and non- governmental, including the WVorld Bank. Given the acceleration of the intra-Amazonian migrations, institutional change is increasingly urgent. Success in this area will require that local and federal governments, as well as international organizations, work with local people in order to develop institutions that will support the innovative use of policy to achieve envi- ronmental ends. It is of equal importance that good farming settlements not attract a new wave of interregional migrants in a perverse demonstration effect. This does not appear likely. The economic and demographic ad'just- ments that originally led to declining interregional migration in the 1980s have probably become irreversible, indicating that eventual recovery from Brazil's current economic crisis will probablv not trigger anothei Amazon- bound migration comparable to that of the 1970s. This work also discusses two other important issues: the relationship between Amazonian deforestation and general macroeconomic conditions in Brazil, and the relationship between returns to frontier farming and the cost to the global economy of reducing Amazonian deforestation. Although the original intention of this work was not to address either question, empirical findings bear upon these issues in important ways. Introduc-tion and Principal Findin7gs 3 Macroeconomic instability appears to fuel deforestation in two ways. On the one hand, it drives up land prices and/or changes the demand for land from that of a productive factor into that of a speculative asset. This shift leads to the buying and hoarding of land by the nonfarming middle class, and the small farmers are pushed farther into the forest. On the other hand, low wages and the low probability of finding a job in a depressed urban economv reduce the opportunity cost of frontier farming, encouraging settlers to continue farming and deforesting the frontier. On both counts, therefore, economic recovery may tend to reduce deforestation in the Amazon. The cost to the global economv of reducing deforestation may turn out to be much higher than land prices alone would suggest. Farmers manage a portfolio of agricultural, nonagricultural, and speculative incomes, not all of which get capitalized into frontier land prices. Since in many frontier locations agricultural income is a small and declining part oftotal household income, the income forgone when land is sold is greater than the net present value of the income derived from agricultural production alone. Addition- ally, frontier farmers' transactions in varied markets sustain a large and growing nonfarming economy, inside and outside the Amazon. This con- tingent would also forgo income if Amazonian land were to be kept out of agricultural production. So the cost to the global economy of containing Amazonian deforestation is underestimated if the full range of forgone incomes is not taken into account. These conclusions are based upon an analysis of twenty years of Ama- zonian settlement in Brazil, using secondary data, institutional interviews, surveys of the literature, and a large panel survey of small farmers (500), merchants (100), and institutions (80) in representative settlement projects in the Brazilian Amazon. Farmers, merchants, and institutions were inter- Viewed in 1981 and 1991 in public projects in the state of Para and private projects in the state ofMato Grosso. This sample captured much of the large differences across settlement locations and highlighted some broad under- lying processes. Differences appeared in natural resources; in political, legal, and institutional frameworks; and in settler origin characteristics, including economic and cultural ties. Background of Amazonian Settlement Since colonial times, Brazil has dealt with the social problem of landlessness by extending the agricultural frontier westvard. Successive federal and state-level land distribution agencies, such as the National Institute for Colonization and Agrarian Reform (INCRA) and the former Ministry of Agriculture and Agrarian Reform (MIRAD), have traditionally been respon- sible for bringing together agrarian reform and colonization. Rather than 4 Sutahinable Settlement in the Brazilia72 Amazon tackle the political cost of taking from wealthy landowners and giving to the landless poor, land distribution in Brazil has mostlv taken from nature and given to the rich. Directed colonization projects are the exception to this rule. For the last few years, since the rapid rise'and fall of MIRAD in the mid-1980s, the issue of government-directed, small farmer settlements has lain relatively dormant in Brazil in general and in the Amazon in particular. Experience gained in the 1970s and 1980s, whether with colonization projects in the frontier or agrarian reform projects (assentamentos de refo'ma agrdria) in established areas, was disappointing. Many directed settlements, though made in the name of distribution, quickly regressed into consoli- dated and concentrated land property structures whose social benefit was questioned by leading domestic and foreign policv communities. In addi- tion, massive deforestation and inappropriate tropical farming technologies that caused soil degradation generated conservationists' outcry against the environmental impact of colonization projects. Small farmer settlement, however, is likely to become a governmental priority again soon under the pressure that arises from rural violence and because of a growing local demand for stabilization. Since the end of the military regime in 1985, the rural workers' unions, once severely repressed, have become larger, more widespread, and better organized. Landowners' associations have become more influential as well. The low priority in the allocation of government resources to land distribution policies in recent vears has escalated the conflict between these two groups. An increasing incidence of violence has followed not only in the Amazon but also through- out Brazil, as landless farmers press for, and landowners resist, any form of land redistribution. Since the promulgation of the 1988 constitution, fiscal reform and in- creasing decentralization of the government have given more power to local Amazonian communities that benefit by settlement schemes. Demand for colonization is a growing political issue in local and regional elections. Small farm surveys in the Amazon and in the rest of Brazil are showing that the distributive effects of directed colonization projects are greater and more long-lasting than was once believed. The errors of the past have been learned, and top-down, unassisted, federal colonization projects that spread unsustainable farming over land areas too large to manage are recognized as being undesirable; the movement is toward local-level programs aimed at relieving poverty and social conflict. Principal Findings The aim of this project was to attain long-term and generalizable policy implications from an original longitudinal data set, while taking into account the variability of the Amazonian experience. Chapters 2 and 3 provide background on the evolution of population and economic conditions in Introditction and Principal Finzdinzgs 5 Brazil during the 1980s. Chapters 4 and 5 analyze a large panel survey conducted in 1981 and 1991 of small farmers in representative locations of the Amazon frontier. The originality of the work as a whole lies in the unique primarv information it brings to light on the evolution of frontier farming in the Brazilian Amazon during the 1980s. Chapter 2 uses Brazilian demographic and economic census to trace the path of migrations to and within the Amazon during the 1970s and 1980s. Small farmers who deforested the Amazon during the 1970s and 1980s had mostly migrated there from outside the basin. Most small farmer deforesters of the 1990s have apparently come from within the region, however. The greatest threat from small farmers to the Amazon forest, therefore, now seems to come from intraregional migrations. The issue is no longer how to prevent deforesting farmers from migrating to the Amazon from the rest of the country, but how to ensure that farmers already in the Amazon stay where they have already deforested, thus reducing migration (and defores- tation) farther inland. Intraregional migration can be stemmed only by establishing sustainable farming in already deforested areas. Chapter 2 argues that the Amazon is not likelv to suffer overwhelming population influx from outside the region. Fertilitv rates for Brazil are declining, overall and regional population growth rates are falling, a general process of urbanization is under way, and experience from the past two decades has taught migrants that settling the Amazon is arduous. On the other hand, rural exodus is accelerating within the Amazon itself. Chapter 3 uses fiscal and other sources, plus direct field research among Amazonian merchants in 1991 to analyze the growth of the public and private sectors in the Amazon during the 1980s. The chapter shows that economic conditions shifted dramatically during the 1980s, inside and outside the Amazon, changing the course of intraregional migrations. As the na- tional economic conditions weakened and the federal government weak- ened, local economies grew and provincial governments became stronger. The transition to democracy and fiscal decentralization brought far more revenues into the Amazon via local governments than were lost with the demise of federal fiscal incentives, credit incentives, and other programs. Meanwhile, primary activities in the frontiers of the 1970s shifted from agroextractive to a form of urban ranching in the 1980s. By the 1990s, a large part of the original settlers of the 1970s had left their plots, which are now increasingly owned by the local urban middle class. Some of those who left profited from the sale of their lands, but many were soon as poor as they started. Thus, though much of Amazon deforestation has been done by small farmers, they have not necessarily kept the land they cleared. The new urban middle class, linked by trade to the industrial Southeast, started to purchase local land as a store of value mainly in response to the national economic crisis. These urbanites constitute a local elite that may offer resistance to the policy changes this work recommends. 6 Sustainmale Settlem7ent in the Brazilian Amazon The chapter concludes that local governments, recentlv empowered politically and financially, must begin to take responsibilitv for devising local economic policy frameworks and incentives that would check the flow of small farmers to new frontiers in the Amazon. Increasing the sustainability of settlement in old Amazonian frontiers requires learning from the errors of the past, wvhich led to unsustainable settlements, and avoiding those of the present, which oppose anv new settlement. Over the 1980s, as the Brazilian general economic crisis deep- ened, employment opportunities and real wages decreased, so settlers covered their opportunity costs of labor and capital. However, as macroeconomic instability drove up land prices, thev did not cover the opportunity cost of land. Under such conditions, it became more worthwhile to become an itinerant farmer, selling in old frontiers and buying in new ones, than to remain in one single plot. To reduce deforestation, then, the frontier economv must be changed so as to reduce its embedded rewards to small farmer itinerancy. Chapter 4, based on a field survey of Amazonian small farmers in 1981 and 1991, indicates that small farmer outmigration can be reduced bv reversing local economic costs and benefits of speculation and itinerancv. Itinerant agriculture has not proven necessarily harmful to small farmers themselves. Although many had insufficient agricultural income to resist selling land at prices inflated by local land speculation, they were still better off than if they had pursued alternatives, given the low wages in the rest of the economy. This indicates that stabilization and generallv rising incomes in the rest of the economy would, most likely, reduce deforestation because the opportunity cost of frontier farming would increase. In other words, there would be less incentive both for outsiders to migrate to the frontier and for insiders to continue farming there. The small farmers surveved made far more income than if thev had been employed in Brazil's swollen labor market and accumulated net worth much faster than if they had put their savings into available financial markets. Given the economic conditions and policies existing at the time, it was worthwhile for them to farm successive frontier plots, reap capital gains from selling each one, and move on. The distributive benefit of settling small farmers in the Amazon was thus sub- stantial, but it was not accompanied by a stabilization of the population in the original deforested plots. The only small farmers who did not move were those with high agricul- tural productivity. These, however, were also the greatest deforesters. Thus. good farming does keep small farmers from deforesting new frontiers, but only at the cost of more extensively deforesting old ones. Local, enforceable policies are needed, therefore, that not only promote good farming but also reduce deforestation in older Amazonian frontiers. Chapter 5 uses the survey described in chapter 4 to analyze economic motives for deforestation. The evidence indicates that farmers whose main Introduction and Principal Findings 7 source of income is agriculture curb their deforestation when land prices rise and incomes fall. Speculators, on the contrary, deforest more, not less, under the same circumstances. Over the 1980s, as land prices rose in old frontiers, speculative motives surpassed agricultural motives for deforesta- tion. A form of agricultural involution took place, increasing unproductive land holding and the proportion of fallow to cropped area and decreasing the proportion of agricultural earnings in total family incomes. 1 Amazonian settlers became less agriculturally active and less sensitive to economic policies aimed at stimulating agriculture. The chapter argues that deforestation since arrival (that is, the accumu- lated area cleared of forest by settlers from the moment they arrive on the plot) is heavily influenced by the migrants' origin characteristics, especiallv the extent of prior itinerancy. In the current year, however, the extent to which farmers deforest is influenced more bv local characteristics, especiallv access to credit. Policies that improve the agricultural performance of small farmers could thus prepare the ground for economic policies aimed at curbing deforestation in old frontiers. Chapter 6 draws policy implications from the findings of chapters 2 to 5. The chapter proposes that innovative, local-level economic policies in old Amazon frontiers can contain intraregional migrations to new frontiers. These include (1) promoting productive agriculture by appropriate zoning, extension, marketing, and credit; (2) taxing agricultural incomes, so as to penalize the increasing tendency of successful farmers to deforest; (3) punishing speculation by taxing capital gains based on land transactions; and (4) penalizing deforestation directly by taxing stumpage, levying fines, and so on. Such policies would indirectly provide the economic conditions necessary for sustainable frontier settlement, correcting the shortcomings of directed colonization projects attempted during the 1970s and held over during the 1980s. The successful implementation of these policies would demand much institutional learning: environmental authorities must under- stand and support new economic instruments; economic authorities must do the same for new enlvironmental objectives; local governments must take on new executive responsibilities; federal governments must assume new coordinating roles; international organizations, such as the W'Vorld Bank, must contribute more broadly to the building of institutions; and settlement agencies must learn from the mistakes of the past. Dimensions and Limitations of the Study The objective of this book is to analyze the extent to which economic policy, in particular that of the World Bank, can be used to reduce the share of Amazon deforestation caused by local small farmers. It draws on twventy years of regional observation and experience and makes use of analytical tools from diverse fields of economics-macro- and microeconoinics, 8 Sustainabl,e Settlemen7t in the Brazilian7 Amazon econometrics, resource and environmental economics, institutional eco- nomics, public finance, economic demography, and others-to offer a policy prescription for reducing Amazon deforestation. The empirical section of the book (chapters 4 and 5) assess how well, or how badly, small farmers are covering their opportunity costs of frontier farming. These opportunity costs are rather abstract. Overall, they represent alternative rates of remuneration of factors of production in the economy as a whole. Concrete alternatives, such as gold prospecting, drigs, extrac- tivism, and so on, are not dealt with explicitly. Some terms are used interchangeablv in the text, glossing over important differences, such as "frontier farmers," "settlers," "colonists," and "pioneers." The terms "Ama- zon," "Amazonia," and "the North region" are also substituted for one another, as are "family farming," "poor farming," and "small farming." The most difficult notion to pin down was that of "sustainability," a term both overused and misused in the development literature. Instead of at- tempting to redefine the notion, the conceptual framework used in this book allows for a method that identifies movements toward and awav from sustainability; that is, sustainability is said to increase as the ratio of private benefits to environmental costs rises and to decrease as the ratio declines. As used here, the term has a time dimension, as do policies aimed at changing farmer behavior to increase sustainability. I he originality of this work lies in its empirical observation of indicators of changes in sustainability. Empirical findings of the work bear upon some broader policv issues, such as how Amazonian deforestation relates to general macroeconomic condi- tions and the global economy. The evidence suggests the following: * Recovery from economic depression in Brazil mav be conducive to decreasing deforestation in the Amazon because (a) stabilization would reduce speculative demand for land and compress capital gains to small farmers in selling out and moving inland, and (b) if recoverv creates jobs and raises wages, it would increase the opportunity cost of frontier farming, thus reducing the pressure of small farmers on the forest. The relationship between the environment and macroeconomic conditions is thus a complex one, subject to specific local conditions. * The cost to the world economy of buying Amazonian land to prevent deforestation may be considerably greater than expected because (a) in selling land, frontier farmers forgo not only agricultural income, but also a whole portfolio of agricultural, nonagricultural, and specu- lative incomes and (b) farmers' transactions generate nonagricultural incomes inside and outside the Amazon that would also be forgone. The relationship between the environment and the global economy is thus also a complex one, subject to national and local conditions. There are several things that this study does not attempt to do. It does not analyze data on Amazonian occupation as a whole, or on the general bitroduction and Principal Findings 9 process of Amazon deforestation, beyond that caused by small farmers. It does not discuss some of the broader consequences of Amazonian defores- tation, such as its global environmental impact or the political and social dimensions of Amazonian deforestation. (However, extensive reference to the literature is made.) Although the text does not attempt to offer an in-depth analysis of human capital, the data set used does contain informa- tion that pertains here. Oz6rio de Almeida (1992) offers more extensive research in this area. Likewise, health conditions, which are not explicitly examined in the text but are known to correlate to economic performance, are taken into account in the book's conclusionis. Finallv, it was not possi- ble to undertake an empirical qualitative analysis of the soils of sampled locations. There is no evidence, however, that the colonization projects sampled have endogenized soil quality. The sample, therefore, is not biased in regard to soil type. To cover all of the above was beyond the scope of our study. This book raises issues heretofore unexplored in the literature and is largely based on observations made in the Amazon in the 1970s and 1980s. The new information and novel interpretations it offers are based on an analysis of the largest and most complete data set ever produced on the economic variables that influence small farmers in the Amazon. Such observation of the economic behavior of a panel of small farmers over a span of twenty years, from the major thrust toward Amazonian colonization in the early 1970s until the chaotic aftermath of the early 1990s, is unprece- dented (see appendix B for a description of procedures). The main lesson learned is that changing Aimazonian sustainability is an intra-Amnazonian matter; interregional migrations no longer pose the threat they once did. If frontier farmers are not settled where they are now, they will eventually move to areas currently being placed under reserve. Conser- vation of these areas, therefore, will not be possible unless intraregional migrations are stopped. To increase sustainability, the Amazonian region must be viewed as whole over the long term. The compartmentalization of academic knowledge, executive agencies, and financing institutions makes it verv difficult to deal with the problem of Amazonian deforestation in its full complexity. Targeted lending operations and grants tend to focus on specific localities and to neglect overall processes that in the long run are necessary if individual projects are to succeed. This work addresses several issues that are generally overlooked. These are: * the current intraregional nature of migrationis to new frontiers, * the importance of good farming in old frontiers for containing defor- estation in new frontiers, • the need for innovative economic-agronomic-environmenital-social policies and institutional change at all levels, including at the NWVorld Bank, to ensure that Amazonian settlement becomes sustainable, 10 Sufstain1able Settlement in the Br-aziliani Amazon * the role of local governments, organizations, and communities in enforcing such policies. By revealing the economic mechanisms at work in deforesting the A_ma- zon today, we hope to contribute to the design of appropriate policies for use by the World Bank, other international organizations, and national and regional governments. Summoning the political will to implement such policies will depend on the recognition that they are necessary. This recognition is all that we are trving to accomplish. If we can do this, then further analysis, debate, negotiation, and experience should build the politi- cal will necessary to achieve effective implementation. Note 1. The term "agricultural involution" is used throughout this book to indicate a decline in the qualitv of farming bv frontier settlers. WVhen farming covers its opportunity costs, settlers tend toward "evolution," meaning they improve as farmers and are likely to remain on the land. W-Vhen the opportunitv costs of farming rise, settlers tend toward "involution"; they stop being good farmers and mav eventually sell out and leave the land. It is important to note that we are not using "agricultural involution" as Clifford Geertz used it in his b)ook by that title. That author used the term to refer to the development of labor-intensive rice cultivation in Asia. He applied that concept with precision to the phenomenon to which he was referring. It is important to note that in this book we are making use of the same term with equal precision to refer to a different phenomiienon. Another observation worth noting is that agricultural involution does not neces- sarily mean abandonment. Abandonment of the plot does not occur immediately when opportunity costs turn against frontier farming. A form of low-level equilib- rium farming begins, which, ifcompared to that which prevailed prev.ious to farmers' arrival, characterizes "involution." A frontier is a dyniamic place, and comparisons in time must be made. Agricultural involution compares negativelv to the farming that was there before. Abandonment mav or may not occur in consequenice. CHAPTER 2 Population Shifts in the Amazon during tie 1980s It is widely believed that the pressure of a growing population in the Amazon is causing the destruction of the rain forest. Conservation would thus require reducing the size of such population or, at least, not attracting new entrants. This is not realistic, however. The Amazon population now numbers more than 19 million people, about one-third ofwhom live in rural areas and show no intention of leaving the region. Forced relocation in such massive numbers would be impossible since neither resources nor the political will to do so exist. Anv realistic Amazon conservation scheme, therefore, must deal with the fact that for the foreseeable future, there will be approximately six million people farming the Amazon.' The issue, then, is how to do so sustainably. The purpose of this chapter is to show that Amazon deforestation is largely responding to intraregional population shifts and to local-level dynamics, as opposed to countryvide pressures. Fertility rates for Brazil are declining, overall and regional rates of population growth are falling, and a general process of urbanization is under way. As a result, in the near future, Amazonian settlement is not expected to experience significant population pressure from the outside. Rural exodus, however, is accelerating within the Amazon, and small farming is becoming less sustainable. Both these factors have spurred intra-Amazonian migrations and deforestation farther inland to the very borders of Brazil, where farmers are adding to the flow of prospectors and other migrants now spilling over into neighboring coun- tries. Thus, reduction of population pressure on the forest requires policies that curb intraregional rather than interregional migrations. Fertility Decline High fertility is often believed to contribute to pressure on fragile environ- ments by creating a large stock of potential migrants. Rising fertility rates would suggest, other factors holding constant, a possible increase in the 11 Table 2.1 Population, Percentages, and Average Rates of Growth: Brazil and Regions (1960-91) Average annual geometric Population Perrcentage over country's total grozth rate Growth 1960- 1970- 1980- Area considered 1960 1970 1980 1991 1960-91 1960 1970 1980 1991 70 80 91 Frontier North 2,561,782 3,603,860 6,619,317 10,146,218 7,584,436 3.65 3.87 5.57 6.94 3.47 5.02 3.96 Center-West 2,963,715 5,099,787 6,805,746 9,419,896 6,456,168 4.23 5.47 5.72 6.45 5.58 3.99 3.00 Total 5,525,497 8,703,647 13,425,063 19,566,114 14,040,617 7.88 9.34 11.29 13.39 4.65 4.43 3.48 Other regions Northeast 22,181,880 28,111,927 34,812,356 42,387,328 20,205,448 31.65 30.17 29.25 29.00 2.40 2.16 1.81 Southeast 30,630,728 39,853,498 51,734,125 62,121,357 31,490,629 43.70 42.78 43.47 42.50 2.67 2.64 1.68 South 11,753,075 16,496,493 19,031,162 22,079,703 10,326,628 16.77 17.71 15.99 15.11 3.45 1.44 1.36 Total 64,565,683 84,461,918 105,577,643 126,588,388 62,022,705 92.12 90.66 88.71 86.61 2.72 2.26 1.66 Brazil 70,091,180 93,165,565 119,002,706 146,154,502 76,063,322 100.0 100.0 100.0 100.0 2.89 2.48 1.89 Source: IBGE, Censos Demograficos, various years. Popul1ation Shifts in the Amazon durinxg the I 980s 13 propensity to migrate, whereas declining fertility rates would suggest a decrease. These hypothesized relationships have not held in Brazil during the past thirty years, however. The weighted averages of total fertilitv rates for Brazil were 5.6 for 1965, 3.2 for 1990, and the projected rate for 2000 is 2.4. The presumed vear of achieving a reproduction rate of I is 2005. These figures indicate that the overall fertility rate has been falling and will probably continue to decline in the 1990s to less than half of what it was in the 1960s.2 Despite important regional differentials (as described by Martine 1992b, page 17), this decline has been significant in all regions and social classes.) The drop in fertility rates, however, has not resulted in a reduction in the rates of migration and urban growth. On the contrarv. during the 1960s and 1970s throughout much of the countrv, internal migration has remained high despite severe declines in fertility rates. Internal migrations, in general, and to the Amazon in particular, seem to have been spurred more bv economic and social factors than bv population increase.4 The Populous Frontier From 1960 to 1991 the growth of total population in the old cerrado frontier5 and in the new Amazon frontier was more than 14 million people (see table 2.1). During the 1980-91 period, however, rates of population growth signifi- cantly decreased in all regions relative to the previous decade. This phe- nomenon can be largely attributed to the general decline in fertilitv rates. Table 2.1 shows that, although the population growth rates for the North and the Center-West were relatively high in the 1980-91 period, these rates experienced the greatest declines relative to the 1970s. The cen-ado frontier of the Center-West had the highest rates of population growth during its hevday in the 1960s, as did the Amazonian frontier during the 1970s. Since then, both regions have experienced far lower population growth rates. These trends suggest that, until the earlv 1980s, migrations from old to new frontiers were interregional, from the cer-ado and from other regions to the Amazon.6 Nowadays, migrations from old to new frontiers seem to have become an intraregional, Amazonian phenomenon. The Urban Frontier Ever since the 1960s, the urban population in the frontier has grown at a much higher rate than the rural population (see Martine 1992b for detailed data). The frontier, therefore, is no exception to the process of urbanization that has been occurring throughout Brazil since the 1960s.7 What is unique about Amazonian urbanization, however, is that it occurs despite a low 14 Snsaianable Settlement in the Brazilian Amnazon population density of 0.23 inhabitants per square kilometer in the region as a whole.' For this reason, Amazonian urbanization mav be considered precocious9 (see table 2.2 and Oz6rio de Almeida 1991b, p. 622). During the 1980s, frontier urbanization gained momentum. Shrinking opportunities in frontier agriculture stimulated the massive migration of would-be settlers to the region's cities and gave rise to an urban network more geographicallv balanced overall.'0 Urbanization, therefore, is an overriding demographic tendencv nation- wide, against which frontier expansion is but a weak countertrend Increas- ingly, potential colonists choose to earn a living in urban settings, rather than venture into the forest. Nowadays, the significance of the shifting frontier tends to come less from its demographic, agricultural, and social magnitudes than from its environmental consequences. Shifting Agricultural Frontier The Brazilian agricultural frontier has historicallv been short-lived. (Oz6rio de Almeida 1992 and Sawver 1990 expand on this point.) The main sending areas, once frontier areas themselves, soon began expelling emigrants at rates greater than those at which they were taking in immigrants. During the 1970s, the total growth of the rural population in the frontier was 1,281,575 as opposed to 1,325,223 in the previous decade (see table 2.2).'' During the 1960s and 1970s, the geographical spread of occupied frontier areas was considerable. As the frontier shifted, it left behind low-densitv population pockets where small farming actiVities had once been carried out. Small landowners sold their plots, either moVing farther inland to reestablish their family farming or going to nearbv towns, seeking nonfarm employ- ment. Small tenants and squatters were evicted, and commercial agriculture took over, often converting farmland into pasture. 12 Table 2.3 shows broad trends during successive decades. During the 1960-70 period, the Southeast experienced the greatest rural exodus in Brazil, the rural population declining by 1.224.574 persons. During the 1970-80 period, rural exodus in the Southeast increased but was topped by the exodus in the South, vhere the rural population declined by 2,023,200 Table 2.2 Variation in Rural Population in Frontier Areas, 1960-80 F7 ontier area 1960-70 1 970-80 Amazon 547.745 1,047.912 Ce17ado 77,7.478 233.668 Total 1,325.223 1,281.565 Source: IBGE, Cen.os Dem_ograficos 1980, 19.(0. 1960, and Oz6rio de .Almeida 1902. Popllation Shifts in the Amazon duning the 1980s- 1 I Table 2.3 Variation in Rural Population by Region, 1960-80 Region 1960-70 1970-80 North 383,076 924,532 Northeast 1,945,981 957,853 Center-West 720,432 -178,430 Southeast -1,224,574 -1.963,936 South 1,826,351 -2,023,200 All Brazil 3,651,266 2,283,181 Source: IBGE, Censos Demograficos 1980, 1970 1960, and Oz6rio de Almeida 1992. persons. This massive rural exodus in the South-Southeast during the 1960s and 1970s was caused bv a number of factors, including the penetration of soybean and sugarcane plantations, changes in land laws and rural labor legislation, the effect of business cycle variations on real estate appreciation, and fiscal and credit policies favoring large farmers. 13 The tables and map 2.1 show that during the 1970s the Center-West became an old frontier, expelling more farmers from the countryside than it absorbed, especially in the states of Mato Grosso do Sul and Goias.14 Map 2.1 is based on the agricultural census and refers to units of agricultural production. Given that bean cultivation is strongly associated with small farmers, a reduction in the area harvesting this crop, as indicated bv the dark area in the map, implies a decline of small farming. This is what occurred in the cer7ado frontier during the 1970s. Unfortunately, the lack of 1991 census information on rural population makes comparison impossible for the 1980s (see Oz6rio de Almeida 1992, chapter 2, for further data). Hollow Frontier This section attempts to trace the movements of small farmers in the Amazon during the 1980s by looking at similarities between their movement and the area harvested with beans during the 1970s. Map 2.2 suggests rural population trends for the 1980s. According to this map, the size of the intra-Amazonian rural exodus assumed impressive proportions during the 1980s.15 The centripetal force of this exodus created a demographic hollow at the core of the frontier as it pushed migrants awav from the Trans-Amazonian highway, which had drawn them in one decade before. Migrants concentrated in northern Mato Grosso and along national boundaries, often going beyond them into eight neighboring countries. 16 Elsewhere in the countrv, as noted by Martine (1992a and b), there was an inverse tendencv to this intraregional post- frontier phenomenon: rural areas increased their absorption of small farm- 1 6 Snstainable Settlement in the Briazilia2 A4mazon Map 2.1 IBRD 27149 ;° (VENEZUELA )s o 3 ( : C OOmBIA r 0 A H ATLANTIC OCEAN k ~~~~ PIAW( ~~~~DO NX"T PERU~~~~~~~~~~~~~~~~ P E MAO SAO C PAAHOA Hil O L' t ~/PRAN S t _ ? MINAS 20' 1_ SANTO i tUt ' J : <' Cs PsRA AUAY' R,,t DE t < AONFIRtO PACIFIC CHiLE/ OCEAN* I RiO GftANVE A ATLANTIC OCEAN j ARGENTINA D UEUGUAi,, BRAZI L HARVESTED AREAS-BEANS-i 970-75 PERCENT VARIATION - 0 200 1000 KOC OMEVR S i 500 __ _ _ __ _ _ _ 100-500 0 ~~~~~~200 400 600 MILES : )I 0-500 . :50-99 TNonoiscbs 0-49 J n,in-o,n- on, ,or <0 on Rh,,i rap do noP 00ly-' 1hpor of NON-PROJECT AREAS rI. Worl8d onh Groop. * NATIONAL CAPITAL uT t,,. o STATE BOUNDARIES or coz ---INTERNATIONAL BOUNDARIES oondor,s NOVEMBER1995 Note: The (lark areas in the northwestern part of the Brazilian Amazon indicate a long-term decline in extractive activity, mainly rubber tapping, and urbanization in the region. This is a pre-frontier phenomenon; it is very different from the post-frontier phenomenon in which farmer migration rapidly emptied the frontiers. Source: Oz6rio de Almeida (1992). Popuilation Shifts in the Amazon during the 1980s 17 Map 2.2 IBRD 27148 - l )--- - x T r .m--- - C COLOMBIA V A Z "Eir AM/JA ATLANTIC OCEAN 1'~~~~' , ' \ ARAUAY, / ~, t~,8;RPODCPANIt R tA - OCFAN: . t~~~~~~~~~~~~~~~~--rRAD PIAUL~~~ P* R AAHIA MINAS ; { / R1O GRANDEi' AfL ~~~~ANTCOCA 2c,~~~~ AA PAU J { OSJ w ~~ & s .f ..'.t.. ve. "~~~~ ~TLANTIC OCEAN A R G E N/RiO GIANDA BRAZI L HARVESTED AREAS - BEANS, 1980-85 PERCENT VARIATION: >500 o soo Io0 00 ILEERS [Z 1 )0-500 0 200 000 000 MIlES 50-99 The. bordores,. chor.,' _ c0 on th-0 1k,,p do wto -.ply, us the port of [II] NON-PROJECT AREAS The World bont Group. GoY; do,daton ou 'he legoI * NATIONAL CAPITAL 0trrtuj 0 onr terrirory STATE BOUNDARIES or 0ycepfdorrnenttot -. -- INTERNATIONAL BOUNDARIES bhondornes NOVEMBER 1995 Source: Oz6rio de Almeida (1992). IS Snifstaina'le Settement in the Brazilian7 Amazon ers. Map 2.2 shows that, during the early 1980s, even the ce7rado frontier of the Center-West reabsorbed the bean-producing (small) farmers who had left this area during the 1970s. In the absence of a 1991 agricultural census, why small farming has been making a comeback is not yet understood. The issue is an important one, however, not only in its own right, but also for interpreting the intra-Amazonian rural exodus of the 1980s. This exodus probably has intra-Amazonian causes, including the chronic itinerancy of frontier farmers and the unsustainability of frontier settlement. Although it is difficult to change such a long-lived. century-old phenome- non, the urgency of doing so is increasing, as the ecological and geographical limits of the frontier are being exhausted.'7 Sustainable settlement of small farmers in the Amazon is now imperative, not only to preserve the unexploi- ted portion of the forest, but also to maintain good international relations with Brazil's neighbors. Summary and Policy Implications Based on the most recent data set available, the present chapter showed that (1) national and regional fertility rates are declining; (2) overall and regional rates of population growth are falling; (3) a process of precocious urbaniza- tion of the frontier has been under way for a long time and may be accelerating; (4) old frontiers in the ce-rado and in the eastern and southern parts of the country have begun to reabsorb small farmers; and (5) the shifting Amazonian agricultural frontier is adding to presstires from pros- pectors and other migrant groups and emptying out vast areas that once attracted settlements. Given the large population in the Amazon and the rapidity with which this frontier population shifts, there is an urgent need for policies that could contribute to increasing the sustainabilitv of Ama- zonian settlement. Notes 1. Several recent authors (Partridge and Schumann 1989) assert that peasant farmers are not the main agents of forest clearance. Both Denevan (1978, page 67) and Foweraker (1981, page 208), on the other hand, argue that peasants do clear most of the original forest, which then passes on to cattle ranchers. In fact, small farmers occupy a smaller area than anv other deforester in the Amazon. In 1980, for instance, total cropped area covered only I percent of total Amazon area, as compared to 4 percent covered by ranching (Oz6rio de Almeida 1992). However, in spite of this apparent insignificance, small farmers are the most mobile of deforesters; so over their lifetimes they clear an area far larger than what they currently occupy. A simple numerical example illustrates this point. According to U'orld Resources (1992, table 19.1), during the 1981-85 period, the yearly rate of deforestation in the Amazon was approximately 0.5 percent of total area in the basin. Dividing total Popuilation Shifts in the Amazon dri4ng the 1980s 19 cropped area (I percent) by total yearly deforested area (0.5 percent) implies that crops remain, on average, only two years on each deforested plot. Many farmers, even small farmers, use their lands for considerably longer periods of time, as will be seen in chapter 4. So the underutilization of, and turnover on, deforested plots in the Amazon is very high indeed. In this sense, the most effective wav to reduce small farmer deforestation would be to increase the duration of farming on each deforested plot. This is the thesis of this book and, in particular, the focus of chapter 4. 2. Quoting from a 1992 internal World Bank report: "Until the mid-1970s. Brazil had a pro-natalist policy. After that, its laissez-faire stance opened the way to considerable activity by NGOs .. which were active in public education to popularize and legitimize family planning.... The contraceptive pill was mainly provided by private pharmacies whose prices were very low.... The social security health system paid for Caesarean deliveries, which provided an opportunity for tubal ligation as a side arrangement, for an additional payment." From the mid-1970s on, therefore. concerted efforts were made to reduce the fertilitv rate in Brazil. 3. For a discussion of urban growth in Brazil, see Martine (1992h). IBGE has data concerning regional decline in fertility rates. 4. In support of this position, see Martine (1992b, especially page 9) and Sawyer (1990, page 14). Contrary to this position, see Thiele (1990, introduction). 5. The cerrado region is basically a savanna ecosvstem that covers most of the Center-West. The Amazon frontier consists mostly of tropical rainforest soils. 6. See Oz6rio de Almeida (1992, chapters 2 and 12) for a discussion of the literanire and evidence on interregional migration and the advancing agricultural frontier. For a discussion of how migration during the 1970s affected deforestation- especially that caused bv small farmers-see Browder (1988). 7. A thorough historical review of the urbanization process in Brazil can he found in Martine and Camargo (1983) and Martine (1992a and b). These sources analyze the impact of urban groxwth at the regional level, focusing particularly on frontier areas. 8. See table 2.1 and Oz6rio de Almeida (1991b, page 622). 9. In general, rapid urbanization occurs in the face of high population density. By precocious, we mean the process of urbanization that occurs despite low regional population densitv. 10. For a discussion of the impact of urban growth on local populations, refer to Moran (1983) and Penalva Santos (1993). 11. Unfortunately, because data on rural population are still not available from the population census of 1991, the 1980s are not included in table 2.2 This section does refer, however, to the 1985 agricultural census. 12. For a microeconomic model that discusses the conversion of land, given tenure insecuritv, see Southgate and Pearce (1988). For a detailed analvsis of the evolution of agriculture in the North during the 1970s, see Brito and Une (1987). 13. See Oz6rio de Almeida (1992, chapter 12) for a review of this literature, and Mesquita and Silva (1987b and c) for an account of the emergence of soybean plantations in the South and its impact on small farmers. 14. For a discussion of recent population movement on the frontier, see Oz6rio de Almeida (1992), Sawyer (1990), and Martine (1992a). 15. The methodology and data for constructing this map are the same as for map 2.1, but with data from the 1980 and 1985 agricultural censuses. 20 Sutstainnable Settlement in the Baszili/an Amazon 16. See Marques (1993) for a description of the accelerating invasion of Brazilians into neighboring Amazonian countries. Although gold prospectors appear to be the perpetrators of recent violence against native peoples within their reser-vations, these incidents are also indicative of intra-Aimiazonian migration pressures against bounda- ries of all kinds, he thev international or protective of reserves. 17. Credit and fiscal incentives that increased the valie of land and the ability of large farmers to usv out small ones inay have accelerated this centuries-old process during the late 1970s and early 1980s. CHAPTER 3 Economic Shifts in the Amazon during the 1980s Since migratory patterns tend to reflect the nature of economic activitv and trends, the boom of the 1970s and the crisis of the 1980s affected migratory movement in the Amazon differently. This chapter concentrates on the impact of the 1980s national economic crisis-the worst in Brazil's modern history-on the economic dynamics of the frontier. The chapter bears important results for the discussion of intraregional migratory movements found in the remaining chapters of the book. The main finding is that general economic crisis and a weakening federal government are being countered in the Amazon by strengthening provincial governments and growing local economies. In response to generalized economic crisis, frontier agents are increasingly channeling resources to local real estate, provoking a speculative run on land. As land markets consolidate in what had been the frontier of the 1970s, the pioneers of that decade set out for new land, with very small return migration. It is possible that if Brazil's economic instability were to end, the economic and population adjustments made in response to the crisis could endure. This holds important implications for the relationship between general economic crisis and Amazon deforestation. Changing Economic Contexts The 1980s crisis eventually dampened the trend of the previous two decades, marked bv steadily increasing interregional migration. Among the numer- ous causes of this earlier (1960s and 1970s) geographic mobility were: * A combination of legal, political, economic, and agricultural factors that pushed small farmers out of established farming areas of the South, Southeast, and Northeast. These included (a) legal and political 21 22 Sustainable Settlement in the Brazilian? Amnzo77 changes in labor relations in agriculture (Mesquita and Silva 1987a); (b) a long peak in the business cycle (the "Brazilian miracle"), driving up real estate prices, mainlv in the South, which in turn financed the spontaneous outmigration of southern familv farmers (illustrated by Rezende 1981 and Brito 1987); (c) the penetration of the soybean crop in the South and Center-West, consolidating small holdings into large ones (Mesquita and Silva 1987b and c); and (d) a severe drought in the Northeast in the early 1970s, expelling the rural population from a vast area. A military dictatorship that promoted a policy of Amazonian occupa- tion during a period of easy access to international finance for large- scale projects.' This regime, motivated bv xenophobia, geopolitics, and development aspirations, catalvzed the forces that promoted oc- cupation. Characterized by concentrated decisionmaking and executive power, the military government created vast resource-using projects in mining, smelting, hydroelectric power, and other industries, at- tracting to the Amazon hordes of workers and farmers from distant regions. Improvements in transport and telecommunications systems were made, facilitating long-distance, interregional migration; and the establishment of agricultural credit and fiscal incentives expanded agroindustrial and commercial agriculture in frontier areas. All of the above, together with official land titling and directed colonization programs, increased the accessibility of the Amazon to large and small farmers and to an increasing variety of agents. Oz6rio de Almeida (1991a and 1992) gives further information on this point. Martine (1992 a) and Sawyer (1990) describe how in the 1980s and earlv 1990s the forces that had earlier contributed to geographic mobility de- clined. Much of the migration-prone population of former share tenants, ex-minizfindidrios, and ex-squatters wvas gradually absorbed at the source, mostly by cities and towns (many as b6ias-fiias) and some by the frontier, reducing the pool of potential interregional migrants to the Amazon. National Economic Context of the 1980s: Shifts in Fiscal Revenues Federal authority for macroeconomic management in Brazil has experi- enced a profound transformation as a result of the institutional changes that culminated in the new federal constitution in October 1988. The constitu- tion provided for greater decentralization of responsibilities, a considerable redistribution of revenues,2 and an increase in the power of subnational governments. 3 Economic Shbfis in the Amazon during the 1980s 23 7The Transition to Democracy and Fiscal Decentralization The constitutional changes have had a somewhat greater impact in the frontier than in Brazil as a whole. Table 3.1 shows the geometric annual rates of growth of locally collected tax income, transfers, and current revenues in the states of Pari and Mato Grosso, as well as for the entire frontier (Amazon plus cerrado), and for Brazil as a whole.4 The first two columns in each section of the table show two periods: 1982-88 and 1989-91. These are the pre- and post-constitutional reform periods, respec- tively. The third column in each section shows decade averages. Differen- tiating between the pre- and post-reform periods sheds light on more recent frontier trends, which otherwise would be lost in decade averages. Column 3 of table 3.1 shows that the total annual tax income on the frontier grew at a much faster rate than in Brazil as a whole throughout the decade. This relatively high growth rate was not unusual, given the increas- ing urbanization (discussed in chapter 2) and the organization of a complex market network in the Amazon during the 1980s. This point will be discussed in detail in the section entitled "Amazonian Economic Context" (page 27). It is to be expected that the tax base would rise as established markets grow and that tax evasion would decrease as enforcement improves. In the state of Para (Eastern Amazonia), tax income over the decade grew at a faster rate than in Mato Grosso (W'estern Amazonia). This difference Table 3.1 Geometric Average Annual Growth Rates of Tax Income, Transfers, and Current Revenues for Para, Mato Grosso, and the North, 1982-91 Tax income Transfers Current revenzues 1 2 3 4 5 6 7 8 9 1982- 1989- 1982- 1982- 1989- 1982- 1982- 1989- 1982- Location 88 91 91 88 91 91 88 91 91 Pari (Eastern Amazonia) 3.46 30.51 11.79 4.41 29.27 12.19 3.89 30.00 11.95 vlato Grosso (Western Amazonia) 9.67 10.47 9.94 1.46 10.41 4.36 6.96 10.45 8.11 Total Northa 7.10 27.21 13.42 6.10 17.60 9.81 6.49 21.57 11.29 Brazil 0.44 9.31 3.32 9.7' 7.79 9.10 2.14 8.96 4.47 a. Comprises all frontier states, not only Mato Grosso and Pard. Thus, averages for total North are averages over all frontier states. Source: Adapted from the SAFuvi database. Based on constant 1991 cruzeiros. 24 Sustainable Settlement in the Brazilian Amazon is probably associated with the structure of commercial activities in each flank of the frontier. Eastern Amazonian colonization is relatively older than that of Western Amazonia, and those who migrated there during the 1970s were mostly poor Northeasterners who established a relatively atomized structure of comrmercial activities. Enforcement was loose in the region until the late 1980s, and taxation was relativelv easv to avoid. For example, while the average annual growth rate of tax income in Pari was 3.46 percent from 1982-1988, it climbed to 30.51 percent after the 1988 reforn. The state's increased monitoring responsibilities and improved enforcement, however, do not account fully for this dramatic increase in the rates of growth of tax income. As will be discussed later in this chapter, the growth of economic activity in the frontier and, consequentlv, the tax base, has contributed to the steep growth rate of tax income in the North. Unfortunately, the two events-improved enforcement and an increase in the tax base-are not separable in table 3.1. W'Vestern Amazonia, in turn, is a younger frontier with a larger proportion of better-off migrants, most from the South and Southeast, the richest regions of Brazil. The type of economic activities conducted on this side of the Amazon contrast strongly with the atomized commercial structure of the East. Here, intense and highly structured trade with the South since the early days of occupation has resulted in larger and more visible estab- lishments that could not as easily escape taxation. As a consequence, even with improved enforcement after constitutional reform, the growth rate of tax income over the decade was negligible, given the region's larger original base and type of economic activities (see first two columns in table 3.1). Thus, the relatively high rates of growth in tax income on the frontier during the 1980s is apparentlv due to two events: an increase in the tax base as a result of local economic growth, especially in Eastern Amazonia, and improved enforcement, which reduced tax evasion. This rapid growth shows an Amazon that is changing from being simply an alternative to penury for floating populations from outside the region, as it was in the 1970s, into a richer and more complex economy with a well-defined and fast-growing market structure. Column 6 of table 3.1 shows that, over the decade, the frontier region kept pace with the rest of Brazil in terms of the rate of growth of transfers. However, after the 1988 constitutional reform, the rate of growth of transfers to Amazon states climbed dramatically, in contrast with the rest of Brazil. This recent increase in the amount of transfers from the federal government to the states of the North and Center-West has further strengthened the economy of the region. Current revenues in both the Eastern and Western flanks of the Amazon grew at much faster rates than throughout the rest of Brazil during the 1989-91 period (see column 9). Such a growth rate was strikingly higher in Economnic Shifts in7 the Amazon dtrin7g the 1980s 25 the 1989-91 period than in the 1982-88 period. Five simultaneous trends have contributed to this growth: escalation of government transfers to frontier states; proliferation of tax legislation empowering state and local governments to increase their shares and incidence in old and new taxes; improvement in enforcement; increase in incomes derived from the opera- tion and staffing of local and state agencies, as well as of NGOs and other institutions; and, finally, increase in the number and incomes of private initiatives, not only in farming, ranching, and commerce, but also in mining, processing, and other industries (see "The merchant frontier," page 27). Growing local economic power is thus adding to increasing local political power (provided by democratization) in determining the use and distribu- tion of local natural resources. Meanwhile, national-level political and economic powers are dwindling in the Amazon. Reduiction in fiscal and c-redit incentives The recent growth of frontier economic activity and the constitutional reform of 1988 inserted the Amazon into the broader national economic context vet allowed the region to maintain its local dvnamics. The fact that most strikingly distinguishes the Amazon of the 1970s from that of today is the capability to generate revenues locally, counterbalancing somewhat the loss of fiscal and credit incentives that occurred during the crisis of the 1980s. Fiscal and credit incentives have been held responsible for much of the acquisition and deforestation of large tracts of land in the Amazon during the 1970s (Oz6rio de Almeida 1992). These incentives began to decline during the early 1980s, as can be seen in table 3.2, and subsidized rural credit in the 1990s is practically nonexistent. During the 1970s, fiscal incentives augmented the demand for farm-, pasture-, and ranchland, thereby increasing deforestation on the frontier of settlement and accelerating the conversion of forest to farmland in already settled areas (Binswanger 1994 and chapter 5 this volume). The tax code, until recently, essentially exempted agriculture, converting it into a tax shelter. Since it was relatively easy to claim any frontier activitv as agricul- tural in nature, this exemption contributed to the run on land bv urban investors and corporations attempting to diversify their asset portfolios. Tax havens, together with credit subsidies, provoked an early boom in specula- tive demand for Amazonian land. Mahar (1989), Binswanger (1994), and Ser6a da Motta 1991 treat this point. In the 1990s, however, frontier agents can no longer exclude most of their agricultural profits from their taxable income.6 This contributes to an increase in local tax revenues and empowers local governments to invest in local physical and social infrastructure. To a certain extent, new revenues are, therefore, compensating for lost subsidies.' 26 Snstaintible Settle7nent in the Br-azilian Amazon Table 3.2 Official Rural Credit in Classic Amazonia, 1970-85 .Aln70z71t .4771011 ear (rSS) lear (USS) 1970 61,.692 19718 7 7,219 1971 89.220 19,'9 1.062,085 1972 15 3,763 1980 1,095,666 1973 178,498 1981 748.273 1974 118.669 1982 506.628 1975 288.321 1983 275.168 1976 52.8,506 1984 115,352 1977 573,674 1985 172,79' Note: Average 1985 exchange rate: Cr$2,1414/TSS. Lnflation in the LTnited States from 1985 to 1991 = 4.78 percent (World Bank data). SoU7ce: Adapted from Mahar 1989. Reduction in other fedlera/l activities The political transition to democracv and newlv decentralized fiscal feder- alism dramatically weakened the decisionmaking power of the federal gov- ernment. In addition, escalating external debt and deteriorating commodity prices made the government more vulnerable to those who opposed expand- ing the agricultural frontier. As pressures from international agencies and NGOs grew, the general public became more interested in, and informed about, Amazonian issues. Dwindling federal budgets killed many federal infrastructure invest- ments, while local urban centers, swelled bv newly urbanized voters with political clout, increasingly defined local priorities. Political and economic opposition to large-scale federal projects, such as hydroelectric dams, grew while state and municipal projects multiplied. Federal investments during the decade of colonization-road construc- tion, titling, settlement, services-that had cost billions of dollars during the 1970s, fell drastically, further discouraging potential migrants during the 1 980s (Sawyer 1990). Although vast tracts of land continued to be auctioned off to private colonization firms, or to forestry-agroindustrial concerns from the South and from abroad, they were no longer outrageous giveaways as they had been during the heyday of land purchasing credit and other fiscal incentives (Oz6rio de Almeida 1991a). Some large-scale projects were initiated and/or continued during the 1 980s in mining and smelting, hydro- electricity, and other industries, although these tended to have local impact onlv (the broadest, by far, being that of Carajas) and to attract miigrants from adjacent areas.9 Elconomic Sbifts in the Amazo7n during the 1 980s 27 Summ'ay As the national economic crisis dragged on throughout the 1 980s, the forces that had promoted interregional mobility during the 1970s weakened, while new ones emerged that may have reduced interregional mobilitv even further. Among these new forces were the transition to democracv, the newly decentralized fiscal federalism, a sharp decline in fiscal and rural credit incentives, and a reduction of public expenditures on infrastructure. Probably the effects of these changes will be enduring. Even if national priorities were to shift back toward deliberate inducements to settle in the Amazon, Brazilian society has learned over the past two decades that farming the Amazon is an arduous undertaking, that risks are high, and that many have failed. The economic and demographic forces of the 1 980s were such that an increase in interregional migration is highly improbable. Despite the strong reduction in credit and fiscal incentives to agriculture, and reductions in other federal initiatives in the Amazon, forest clearing did not correspondingly decline during the late 1980s. This suggests that Amazonian deforestation during this period responded to intra-frontier forces different from those of the 1970s or even those of the earlv or mid-1980s. Although the speculative motive for deforestation mav have begun with incentives from the federal government, it is now being fed by local stimuli. Amazonian Economic Context: Shifts in Economic Activities Merchants comprise a large segment of the new frontier's urban middle class. The remainder of this chapter will discuss the emergence and eco- nomic evolution of these frontier merchants and the implications of their economic behavior for small-farmer intraregional migration and deforesta- tion.'0 Because of space limitations, it has not been possible to deal with all members of the Amazon's rising and highlv diversified urban middle class, including persons associated with mining, hydroelectricitv, manufacturing, processing, and other sectors, or with governmental agencies and NGcOs active in Amazonian affairs. The merchantfirontier Although the Amazon is typically thought of as an agricultural frontier, it has always been more urban than rural, due to the staffing of local public sector agencies and the growth of local commercial activitv (Oz6rio de Almeida 1992). The precocious urbanization of the frontier, discussed in the previous chapter, is not solely demographic."' An economic dimension 28 Sustainable Settlement in the Brazilian7 Ania?Zon of urbanization lies in the presence of a diversified and voluminous tertiary sector: commerce, ser-ices, and the public sector. Although manv urban services, especiallv personal services, are undoubtedly swelled bv the dis- guised unemployment of former farmers expelled from the land, frontier commerce is nonetheless highlv profitable with outstanding rates of accu- mulation (Hebette and Marin 1977). The importance of local merchants can hardly be overemphasized in a discussion of the recent expansion of the Amazonian frontier. These eco- nomic agents practice a varied set of activities, all of which have some role in linking primary frontier activities-farring and extraction-to the rest of the economy through market transactions for agricultural or extractive products, productive inputs, consumer goods, credit, land, and labor. (For a detailed discussion of market transactions between merchants and farmers, see Oz6rio de Almeida 1992.) These local transactions determine the appropriation of income and the rate of return both of primary producers-farmers and extractivists-and of merchants. Rates that are above opportunity costs for merchants should inv ite new entries. However, distance, lack of communications, high trans- port costs, and other factors may constitute barriers to such entries and foster natural monopolies among merchants, especiallv when a frontier is relatively new. Such local monopolies may, in turn, reduce prices paid (or increase prices charged) to farmers and reduce farmers' incomes. l2 Accord- ing to standard long-run microeconomic analysis, rates to return in a given market signal entries, exits, or stability in that market. But in a frontier, where markets are still incipient, the rates of return to merchants become important also in determining entries, exits, or stabilitv among local small farmers. kkhether a frontier consolidates with its pioneering small farmers, or whether pioneers are expelled by newcomers and by land concentration, depends greatly on the operation of frontier markets and their merchants. '3 Merchant accunmulation by type of activity The merchant frontier seems to lag behind the farming frontier. 4 In the 1970s, businesses in frontier towns were mainly associated with primary activities.'5 This pattern, however, changed in the course of the 1980s, yielding a spectrum of tertiarv sector actixities with very high rates of accumulation of physical and financial assets.'6 Overall, the monthlv average rate of accumulation of frontier merchants in 1991 was an impressive 2.56 percent (see table 3.3 and various tables from Penalva Santos 1993).17 This w as in fact low wvhen compared to the average of 14.5 percent ten years earlier.'8 What had been beginners' monopoly power apparentlv eroded during the 1980s, as newv entrants competed for local markets, as the towns themselves grew and diversified, and as local Econonmic Shifts in the Amazon di7ing the 1 980s 29 Table 3.3 Monthly Rates of Accumulation by Type of Operation, 1991 Eastern lIfesrer77 Frootie,- Type of operatioi Amnazoniar Amazonia total Sawmills - 0.94 0.94 Merchants of agricultural products 4.62 2.23 3.55 Suppliers of agricultural inputs 1.82 2.88 2.31 Suppliers of consumption goods 0.42 2.76 1.63 Others 3.92 3.38 3.48 Total 2.44 2.65 2.56 - Not available. Note: The figures above are weighted averages, based on the number of people interviewed. Source: Penalva Santos 1993. agricultural lands were increasinglv turned to pasture (Mattos, Uhl, and Goncalves 1992). The poor performance of sawmills in comparison with most other mer- chant activities by the earlv 1990s was apparently caused bv the migration of loggers and mills out of old frontiers and on to new ones, as unsustainable tree harvesting eradicated valuable species.1( Sawmnills aside, Western Amazonia appears to have had a more homoge- neous merchant economy, exhibiting smaller differences in accumulation rates among different types of merchants, than did Eastern Amazonia. This may be due to strong links to a more homogeneous set of trading partners in the South. The lowest accumulation rates were among merchants of agricultural products (purchasers, transporters, and processors), partlv caused by the decline of agriculture and the spread of pastures in much of the old frontier of Western A.mazonia. The highest accumulation rates in this region were not sustained bv outside markets but bv growing local urban economies served by other business activities (see Others in table 3.3). This category includes bars, pool and gambling houses, hotels, brothels, restau- rants, dentists, doctors, hairdressers, barbers, and others. In Eastern Amazonia, accumulation rates varied more widely among merchant types, with outstandingly high ones for merchants of agricultural products, followed by local commerce.-' This indicates an agricultural economy strong enough to sustain such merchants, possibly due to the maturing of perennials, such as cocoa, pepper, and even sugarcane. Also, long-term, historical, river-based trading monopolies initiated by extrac- tivist merchants may be moving into expanding agricultural product markets and imposing their traditional monopoly power over farmers. Suppliers of industrialized agricultural inputs (equipment, fertilizers. pesticides, mechanical components, and repairs) had higher rates of accu- 30 Suststainnable Settlhment in the Brazilian Amazon mulation-suggesting more industrialized agriculture-in the Western than in the Eastern Amazon. Interestingly, overall, agriculture-related mer- chants fared better than vendors of consumer goods (pharmacies, groceries. clothing, and odds and cnds). Thus, during the 1980s in old Amazonian frontiers, merclhants' accumu- lation rates decreased, their monopoly power over farmers diminished, and thriving urban businesses grew. The most successful among the merchants were those who continued to function as intermediaries in frontier com- merce, linking local farmers to the rest of the economy. Merthant accumiulaition by market range The geographic market range of merchant activity indicates the main regions to which Amazon frontier transactions were being channeled. Table 3.4 displays accumulation data from different Amnazonian regional markets. Variations in accumulation rates were higher in Eastern than in Western Amazonia. Western Amazonia sold predominantly to the South (in this case including the Southeast), while Eastern Ainazonia sold to the South and the North (in this case including the Northeast).2' TIhe highest accumulation rates were among local-North merchants in Eastern Amazonia, who bought goods from local suppliers (mainlhy farmers) and sold them to northern clients.22 The data do not show the incidence of merchants in Eastern Amazonia purchasing from northern suppliers, but rather from the South and Center (the latter usually relaying southern-produced goods transported through the Center). In general, merchants who brought outside (mainly manufac- tured) goods to the frontier had much lower accumulation rates than those who took goods from the frontier to the outside economy. This may be a reflection of the undervaluing of the region's natural resources, w here low land values and relatively open access drive user costs downward. Western Amazonia attracted a far higher proportion of southern migrants than did Eastern Amazonia, and southern migrants tended to be better off than other migrants, who were mostly northeasterners (Oz6rio de Almeida 1992). As indicated in table 3.4, Western communities tended to be more prosperous than Eastern ones, and local-local merchants appear to have had somewhat higher accumulation rates in Western than Eastern Amazonia. Western Amnazonia's markets were more interactive than those of Eastern Amazonia, in the sense that they supplied frontier products to, and con- sumed industrialized goods from, the South. There were more barriers to entry in commerce in Western Amazonia, which is consistent with the fact that suppliers of industrialized inputs and consumer goods had higher rates of accumulation in the West than in the East (Oz6rio de Almeida 1992). In fact, most frontier businesses in Western Amnazonia today are run by southerners, who were encouraged to come to the frontier by the private Economic Sjifts in the Amnazon during the 1980s 31 Table 3.4 Accumulation by Range of Markets, 1991 R,ange of markets (Suppher-consumer) Easte7n Amazonia Westee n Amazomla Frantrer total Local-local l.95 3.42 2.42 Local-North 10.65 - 10.65 Local-Center - - - Local-South 3. 3 3.03 3.08 North-local - 1.45 1.45 Center-local 1.28 1.00 1.20 South-local 1.86 2.17 2. 09 Total 2.44 2. 6 5 2.56 - Not available. NVote: The figures above are weighted averages, based on the number of people intenriewed. Source: Penalva Santos 1993. colonization programs of the 1970s. Industrial headquarters in the South financed the establishment of frontier branches, or established regular supply contracts (for example, mechanical parts. components, and so forth, for direct sale or repair services) with local businesses, with an eve to increasing earnings and future remittances back to the firms of origin.23 Unfortunately, space limitations prevent further examination of the devel- opment of traditional mercantile colonization or of interregional imperial- ism in the Brazilian Amazon. Commercial profits not sent to trading partners outside the Amazon tend to be invested in local land purchases, mostly by southern merchants, as discussed below. Merchant accumulation in land Two different activities in which frontier merchants accumulate are com- merce and agriculture. Many of those who migrated to the Amazon did so because they wanted to become landowners, and commerce mav have been a stepping-stone to that end.24 In 1991 agricultural assets represented almost one-third of merchants' total assets (32.13 percent as a weighted average). The highest average is in Western Amazonia, where 3 3.32 percent of total merchant assets are agricultural. This weighted average indicates that west- ern (originally southern) merchants invested proportionately more in land than did eastern (mostly originally northeastern) merchants (28.8;7 percent) (Penavla Santos 1993). In a frontier, becoming a commercial farmier or rancher requires invest- ments with long gestation periods. Gaining title to land implies mapping, demarcating, litigating, and registering the plot with different, looselv coordinated official entities and authorities, who must be present for con- 32 Sostainable Settlement in the Brazilian Amazon tracts to be enforceable. Surveving, forest cutting, burning, stumpage, and soil correction must all be done before ordinary: farming or ranching begin. and markets for credit, agricultLral products, inputs, and labor must be established for commercial farming to be able to operate. Since commercial farming cannot function outside the standard institutions of a market economy-with local wage-labor scarcity posing a potential problem for those who intend to engage in such activitv in new frontiers-many migrants tend first to operate urban businesses and onlv slowlv, over the vears, to invest in agriculture."2 The consolidation and concentration of land own- ership, the occurrence of rural exodus and urbanization, and the formation of a wage-based labor force all contribute to easing the wav for local commercial agriculture to develop.26 Until this becomes possible, mer- chants bide their time purchasing land. Whether thev will eventuallv be- come productive commercial farmers or hold the iand for speculative purposes will depend on relative returns. Of course, Amnazonian reality is far more complex and varied than would appear from such a schematic rendering. In manv locations, commercial farming precedes and preempts small farmer settlements. In others, there is no attempt to farm whatsoever; only ranching or outright speculative landgrabbing exists. WNIhat is clear is that in olc( frontiers during the inflationarv economv of the 1980s, merchant demand for land for speculative rather than for pro- ductive purposes increased.28 In general, the land these merchants bought was sold to them by poor (although not exclusivelv so) local farmers, who moved on to deforest farther inland. Without agricultural experience or expertise, such new urbanite landowners have tended to eradicate crops. even productive perennials such as coffee and cocoa, and turned much of the land to pasture.'2 The next chapter provides field data on these trends. Summary and Policy Implications: Shifts in Public and Private Resources The crisis of the 1980s brought changes to both the overall Brazilian economy and that of the Amazonian frontier. Factors such as federal fiscal subsidies and credit, as well as expenditures on colonization, land titling, and roads that had pronmoted interregional migration throughout the 1 970s began to disappear in the 1980s. Meanwhile, the transition to democracv. increasingly decentralized fiscal federalism, and a growing urban economv promoted intraregional migration. A frontier middle class, composed of merchants and other groups, such as public servants, private sector emplov- ees, and others, is becomning the ultimate beneficiary of deforestation in the Amazon. Economic Shifts in the Amna.zon during the 1 980s 33 Markets for land, labor credit, consumer goods, and agricultural inputs and products have favored the development of urban private businesses. As landownership has consolidated in areas that had been new frontiers in the 1970s, commercial agriculture and commercial url)an activity have grown. These old frontiers are linked to the rest of the Brazilian economy by merchants who transact with local farmers and extractivists, a role that appears to profit both the merchants and their outside trading partners. The number of agents interested in, and actively involved with, the growth of the Amazonian economy is far greater than those actuallv living in the region. Thus, economic policies that alter frontier trends will affect a large and growing business communitv, inside and outside the Amazon. The growing urban economy of the Amazon is not leading to sustainabil- ity for small farmers. W'hether in response to inflation and the protracted national economic crisis, or for the sake of speculative gains as land markets consolidate, the urban middle class is purchasing more and more land. This further inflates frontier land prices, pushes pioneer farmers out, stimulates intra-regional migration, and causes the spread of unfarmed, deforested land. Any policies aimed at curbing deforestation must take into account who these newcomers are and what their motives are for holding land. Notes 1. A detailed treatment of these issues goes beyond the scope of this book. Due to the complexity of the discussion and the abundant existing literature, only maior references are provided. 2. The 1988 constitution provided for a considerable redistribution of revenues via an increased role for state and municipal participation finds-unconditional bloc transfers that had both a decentralizing and redistributive role. Socolik (1989), for example, argues that in 1988 the municipal participation fund (Fundo de Participa- cao dos Municipios-FPM) represented some 50 percent of current revenues of municipalities in most cities of the North, as opposed to 25 percent of revenues of cities in other regions of Brazil. At the regional level, in 1991 transfers represented 55 percent of current revenues in the North and 37.5 percent in the Center-West (SAFEM.x database). See Bonfim and Shah (1991), Serra and Afonso (1993, page 35), and Shah (1990) on the distributive effect of transfers on each level of government. 3. See Shah (1990) and Bonfim and Shah (1991) for a discussion on the specific changes brought about by the new constitution. 4. Note that for this table, data for the entire frontier had to he used (Amazon plus cer7ado), rather than for just the Amazon frontier, or the North region, because of shifts in data collection units during the 1980s. Regional trends in the North cannot be consistentlv observed during this time because two federal territories in the region graduated into the category of states, beginning to collect taxes only when the decade was already well under way (as opposed to the other states of the North). This fact would tend to overemphasize the regional growth rates of tax income in the later years of the decade. Furthermore, a relatively rich new state, Tocantins, 34 Sustainable Settlement in the Brazilian Amazon was created and transferred from the Center-West to the North. after Goias was divided; this would also tend to overemphasize the growth rates in the North. The states of Para and Mato Grosso have been singled out in the table because thev have the highest figures in the region, avoid the redefinitional problems just cited, are representative of the Eastern and Western Amazonian frontiers in general, and are the locations of the field surveys discussedl in chapter 3 and analyzed in chapters 4, 5, and 6. 5. Aside from improved enforcement, a large variety of tax instruments began to accrue to state and local governments directly, due to the tax reforms of the 1 980s, such as value added tax and service tax. This also increased local tax income. 6. Although the extinction of tax havens that had promoted deforestation did actually reduce the rate at which the Amazon was being deforested bv the late 1 980s, thatt part of defjbrestatioi ctetsed ky smallf Jkmers inci-eased consider-ably. This issue will be treated in detail in chapter 5. 7. For example, according to table 3.1, total official rural credit to Classic Amazonia amounted to almost us$1.1 million in 1980, its largest year. By 1982 rural credit had already begun to decline, reaching us$506,600. Meanwhile, fiscal reve- nues for the region were already much larger, US$360 million (SAFE.M data, using 1991 constant cruzeiros and exchange rate of Cr$963.80/us$). A decade later, in 1991, total fiscal revenues were us$1. I billion, wvhich is a thousand times the value of subsidized credit at its height in 1980. This calculation does not take into account the amount of income brought into the Amazon via tax subsidies and shelters. The consensus in the literature, however, appears to be that the rate of return on such subsidized projects was negative and that very little was gained by them in the region. See Oz6rio de Almeida (1991a) for a discussion of this literature. 8. See Oz6rio de Almeida (1992, chapters 6 through 10) for detailed costs of colonization during the 1970s. On average, federal expenditures were approximately us$750,000 per year. This is more than subsidized credit, except for the peak four years of 1978-81, according to table 3.2. The demise of the federal colonization program was, thus, a significant additional loss in federal resources to the Amazon. It is not yet clear how much state settlement programs compensated for this loss during the 1980s. Also see Sawyer (1990). 9. See Oz6rio de Almeida (199 1a) for a partial listing of economic agents active in the Amazon during the 1980s. 10. This and the following sections draw extensively on Penalva Santos (1993), which was based on field research undertaken by the Institute of Applied Economic Research (IPEA) of the Brazilian Ministry of Economics and Finance in 1991 under the project Colonizacao Sustentivel na Amaz6nia. (The box in chapter 4 gives a detailed description of the field survey.) Results for 1991 have sometimes been compared to those of a 1981 field survey held in the same Amazonian locations, as presented in Oz6rio de Almeida (1992, chapter 15). 11. Precocious in terms of population density in the Northern region. 12. When the frontiers of the 1970s were new, merchants' monopoly power was indeed compressing farmers' incomes, especially in Western Amazonia, as shown in Oz6rio de Almeida (1992, chapters 15 and 18). 13. Chapters 4 and 5 will specify in detail the relationship between rates of remuneration of (and opportunity costs to) productive factors-land, labor, capi- tal-and the sustainability of small farmers, as measured by turnover and deforesta- Etono70miC Shifts il the Amazon; during ti'h' 1 980s 3 5 tion. 'I'he remaining sections of this chapter merely describe the economic condi- tions, as viewed through merchants' perspectives, that bring forth those rates and costs for farmers. Schmink and XV'ood (1992) discuss the historical evolving role of Amrazonian elites and merchants, focusinlg onl their sociopolitical dynamics at the local, state, and national levels. See also 13rowder (1988). 14. XVhat had been the new frontiers (absorbers of net migratory inflows) of the 1970s became the old frontiers (expellers of net migratory outflovs) of the 1980s. Nowadays the old frontier is itself inside the Aimazonl. The term "old frontier" refers to the areas where the inflow of small farmers was greatest durinig the earlv 1970s, and where their expulsion was also greatest durinig the 198(s. according to the maps in chapter 2. 15. The feverish road building of the 1970s brotght on farming and ranching, cleared accessible land of forests, and displaced, sometimes violentlv, traditional forest extraction (rubber tapping, nut gathering, logging, and so forth), which had been based on river transport. This changed the structure of Amnazonian urban commerce, as will be seen in table 3.3. Roads also facilitated the invasion of prospectors. This was especially unfortunate for the environment, as prospecting, especially for gold, is one of the worst polluters in the Amnazon, destroving immense river basins with mercury and other hazards. Prospecting was very favorable to commerce, however. In many road-frontier towns of the 1970s, prospecting sustained urban commerce until farming, ranching, and logging started to pay off. Unfortunately, the authors were unable to interview any prospectors, who are secretive about their affairs. 16. The rate of accumulation measures the rate of increase in real net worth over time, that is, the geometric average rate of increase in real value of physical and financial assets minus debt outstanding from the time of arrival at the frontier to the present. In cross-sections it is a better measurement of performance over time than the rate of return, which requires observations of an income stream at several different moments in time. By disregarding all incomes not reinvested in the business, the rate of accumnutlation can be considered an underestinate of the rate of return. 17. Frontier commerce is extremely diversified, and all merchants transact in a broad range of markets. In fact, market interlinking is a notorious feature of frontiers, whereby one agent concentrates several market functions, such as in the case of debt peonage, when credit-product-labor markets are interlinked. The data in table 3.3 refer to the dominant, though not necessarily (in fact rarely) exclusive, type of operation of merchants. 18. See Oz6rio de .Ameida (1992, table 32). These outstanding rates of accumu- lation in some frontier locations in the early days did not last long and soon lowered to the standard shown in table 3.3 and in appendix A, table A.4. 19. See Oz6rio de Almeida (1992, chapter 15). See also Uhl and others (1991) for insightful discussions of logging activities in the Amazon. The locations from which these data are drawn are described in the box in chapter 4. According to Zweede (1993), newer loggers and mills associated with the farming frontier migrated after the farmers, along the new highways, toward areas where the forest was being cleared for agrictulture or ranching. Older loggers and mills, associated with traditional riverside extractivism, did not move with the farmers and remained where they were, harvesting increasingly lower-grade trees. 36 Sustainable Settlement in the Brzazilian Amazon 20. See the box in chapter 4 for a description of surveyed communities. 21. Merchants in all parts of the Amazon frontier were in fact trading with partners all over Brazil. The data refer only to the dominant, but not necessarily(in fact rarely) exclusive, trading partners of the frontier merchants interviewed. Trade between the Amazon frontier and the Center-West is well-known, but it is hidden (table 3.4) by the dominant influences of the South and the North. 22. It is interesting to observe that this route follows, by road, the traditional commercial river transport ties of Amazonian extractivism to exporters and proces- sors in Santar6m and Belem. 23. See merchant company histories based on field interviews in 1991 in Penalva Santos (1993). 24. See Ozorio de Almeida (1992, chapters 12 and 15) for a discussion of the interaction between merchants and farmers during different stages of frontier settlement. 25. Numerous efforts are traditionally made to circumvent the labor scarcity problem of the Brazilian Center-West and North regions. These often involve recruiting and transporting workers from impoverished communities, generally in the Northeast; obligating them to a wide variety of debt peonage systems; and establishing some form of violent coercion or enslavement. In 1981, one of the authors witnessed large truck convoys illegally transporting cargoes of indebted peons to commercial farms along the T rans-Amazonian highway. The practice seems to have become even more widespread during the 1980s. See Oz6rio de Almeida (1991b) and Romanoff (1992). 26. See Oz6rio de Almeida (1992, chapters 12 and 15) for a more thorough discussion of frontier stages and merchant-farmer interactions. 27. Merchants' speculative demand for land will be seen to be a very important factor in undermining small farmer unsustainability in the Amazon, as will be discussed in chapters 4 and 5. 28. Thus, contrary to the usual emphasis of environmental economics, the issue is not property rights, or the lack thereof, but what causes property rights to become increasingly concentrated. The social and economic issues that provoked such concentration in former frontiers appear to be the main causes of Amazonian deforestation today. CHAPTER 4 Land Markets and Sustainable Frontier Farming In the 1980s a complex urban economy, commercially linked to the rest of the country, replaced the isolation of pioneer frontiers in the Amazon, with harsh consequences for small farmers and the forest. During this time the sustainability of small farmer settlements declined all over the frontier, as measured by changes in the ratio of private benefits to environmental costs. In low productivity frontiers, sustainabilitv declined because benefits from settlement declined, given stable environmental costs. In high productivity frontiers, benefits from settlement remained stable, but environmental costs increased. The evolution of farming in the 1980s varies widelv throughotut the Amazon in terms of crops, migrant origins, soil quality, physical and institutional infrastructure, and other characteristics. This chapter analyzes a representative sample of old Amazon frontiers, described in detail in the box (next page) and map 4.1. The authors recognize, however, that the basin is so diverse that no single sample can ever do it justice. WVhile only directed colonization projects, not spontaneous colonization frontiers, are included, many of the migrants interviewed were not settled in their plots bv colonizing agencies, whether official or private.' Many squatted on the land for many years before acquiring claim to it, a frequent occurrence in official projects during the 1970s and earlv 1980s. Others purchased the land from earlier colonizers who left, a practice that became increasingiy common in both spontaneous and directed colonization pro- jects in the 1980s. In fact. the difference between the two types of coloniza- tion became more and more blurred as land turned increasingly into private property and the real estate market grew throughout the frontier. Undoubtedly. however, the migrants sampled here-ly being in directed colonization projects-benefitted fron. greater security of tenure and pro- tection from violent social conflict than did most spontaneous migrants.- To the extent that personal and tenure security contributes to successful farming, colonists in these projects had an advantage over those in areas of 37 38 Siortaina/le Sette/meut io the Brazilian .zniazoii Box 4.1 Amazon field locations: size, distribution, and locations of the sample frontiers Eastern Amazonia In the state of Pari, colonies are official projects. Nlain temporary crops are rice, corn, and beans; main permanent crops are cocoa, pepper, and manioc. Pacal is located oni the Trans-Amazoniani Highway, west of Altamira (120,000 inhabitants). A large old city kept alive by forestry extraction and river transport, Pacal sits on a large patch of fertile soils (terras roxas). Sugarcane and cocoa are the main cash crops. Initially, infrastructure, roads, homes, and even land cleared of forest were provided by the colonizing agency (INCRA). Over the past fifteen years, local commerce and incipient processing have grown in the town of Medicilandia (38,000 inhabitants). Migrants come from all over Brazil, but gauich.o communities (people from the southernmiost state of Brazil-Rio Grande do Sul) are particularly notable. Farmners are militant and more than once have interrupted transport along the Trans- .mnazonian Flighway to protest mishandling of the sugar mill on which their earnings depend. 4napit is a tiny roadside stop along the Trans-Amazoniian HFighwav, east of Altamira and west of Pacaja. It is distinguished from the rest of the Altamira colonization project by its large proportion of poor northeastern migrants (mostly maran7henses) and its high incidence of malaria. Pacajd (27,000 inhabitants) began as a spontaneous settlement, emerging in the eastern region of the Altamnira directed colonization project. Residents, mostly northeasterners, are proud of their community. Cocoa and pepper were once the main cash crops, but grazing has largely taken over. There is gold prospecting in nearby rivers. For a long time, little infrastructure was provided by INCRA. and many feeder roads were built by the settlers. .M1onte Alegre (66,000 inhabitants) overlooks the northern bank of the Amazon River. River transport provides the only access to the towin and the colonization project nearby. The colonization project began in the 1920s uith Japanese migrants and benefited from initialJapanese funding. Most settlers today are children or grandchildren of original migranits, mainlv from the northeast. Some southerners came during the earlv 1980s and manvjapanese remain. The cooperative is quite active and sells farming products to Manaus, Santarem, and Belem. Newly opened areas of the project have good soil, but much of the land is of poor quality; many lots lie abandoned by their owners. Farmers use the vast riverside floodplains (vdzeas) as common land for seasonal cattle grazing. Western Amazonia All field locations in the state of Alato Gr osso are private colonization projects. All farmers in these projects are southern, for other migrants cannot afford land in this area, which is considerably more expensive than in official Land larIkets a7d Suistalinable Fr-ontier Farnming 39 colonization projects. Non-souitheriners in private projects are usually either tenants or squatters and are subject to insecure tenure. There is a large difference in capital endowments and in net worth bet-ween settlers in Alato Grosso's private projects and in Pari's official projects. A/ta Flo07esta (66,000 inhabitants) was initially considered the jewel of private directed colonization. Commerce, church, education, and several other insti- tutions were brought in, along with migrants actively recruited fromii Parana. This commercial and institutional infrastructure wvas sustained by gold pros- pecting uitil a violent war vas waged hv the colonizing firm, which expelled prospectors in the early 198Os. Coffee and cocoa were until recently the main cash crops, but they are l)eing replaced by grazing. Soils were not as rich as expected, and the coffee planted was the wrong kind (arabica instead of robusta). Paranaita (I12,000 inhabitants) vas fouLnded lv the same firm that colonized Aita Floresta. The twVo projects are siniilar in terms of migranlts, crops, soils, and problems. Paranaita is switching to grazing at an even greater rate than Alta Floresta. Mutlon (5,000 inhabitants) is a imonoculture of soybeans. It Was founded hv an entrepreneurial cooperative that brought its oxvn members from Rio Grande do Sul. Originally small farmers, these migrants exchanged highly appreciated lots durinig the 1970s-the hevday of real estate appreciation in the south of Brazil-for large lots (40() to 1.000 hectares) on the frontier. Located near a road to market and retaining commillercial, institutional, aud cultuiral ties to the south, the project boomed. Its town is too small to provide the infrastructure demanded bhy its xxealthv colonists, wvho often travel to Cuiabi, the state capital. This is the richest commurutv in the sample. Sio jos6 do Rio Clamo (16,0(00 inlabitanits) is located on a baad road and has been unisuccessful in agricultiire. Coffee failed even thouigh migrants wvere from Parana and specialized in coffee productionl. Rulbber wvas attempted but failed because the rubber credit program (PROBOR) endledl. Agricultuire has declined, grazing has increased, and logging is liecoming the main income source for the community. A large part of the population has urbanized in the local center. I. The results obtained (incomiie, net w(orth, accuiulation, and other performance indicators) show that mlumtum is no longer a small fariming community. This location could thus be remnoved from the sample. which is supposed to target only small colonization projects. Mutum is retained because of its origin. It bad sniall farmer migratory flow, as did the other settlements analyZed. '1he differing evolutions are interesting to investigate. Coiiparison benteen Muituin and other locations idenitifies conditions for successful colonization. 40 Sstahinhable Settlement in the Brazilian Amazon Map 4.1 IBRO 27150 A \ [ENEZUELA C O LO M B IA > v % U B W ,E p AM *_P A _[ ArLANTIC OCEAN ,- AMAZONAS An4Fau(+Pa ,iI MA 0AhAQ CBA A , 5N>tujmb\°°t .briocgol" , -. >, ~ ~ ~ DO top Y ACRE N. Alit 10~~ ''' /N ID z^ 8GE~~~~IYA 5 b s k LA.GE7A/S ~~~RONVt5NI PC 8AIiA .Sd Jo*4 do Itie Ctoro) OA + j-* MATO GROSSO.) JOIAS PA CIFIC 0 CH J . , , MARANANA OCEAN S 2_, ,t,SJ.R CATANA ATLANTIC OCEAN z y ~~~~~RIO GRAUNDE ARGENTINA DO BAD (,I j~c"uL AGEriAB0 500 )W00 KILtOMETERS 30' URUGUAY- 0 200 O0 600 MffS B R A Z I L LOCATION OF SURVEYED COLONIZATION PROJECTS * PRIVATE COLONIZATION PROJECTS * OFFICIAL COLONIZATION PROJECTS L PROJECT AREA lD NATIONAL CAPITAL - -- STATE BOUNDARIES - -INTERNATIONAL BOUNDARIES The boundr,es co-ors deno-nn-t-os ond ony other wh forma., on ,pon Art rnop fo -o, replo. orn te p.or f ThM. World BofI- Groip. cay jJ fgment orl tef le fl status >f a07 ecfrrtorr or o.ry enrdors eoror ar occep!onco oi soch tr. oonore, NOVEMBER I 95 Sorurce: Oz6rio de Aimeida (1992). Land Ailaikets and Sustainable Frontier Far7ning 41 spontaneous settlement.3 In this sense, then, the sample may consist of the elite of frontier migrants. If small farming turns out to be more successful in directed colonization projects than in the Amazon as a whole, then one should considerexpanding those aspects of directed settlement thatcontrib- ute to such successes. The objective of this chapter is to identifv some of the most important determinants of diverse frontier outcomes and to isolate those that most contribute to sustainable frontier farming, specifically, those that decrease farmer turnover on plots. Appendix A presents the descriptive tables discussed throughout the analysis in this chapter. The meaning and precise definition of each variable, a description of the sample, and a discussion of all empirical procedures used to construct the tables are presented in appendix B.) Main Findings During the 1980s, frontier farming apparently covered its opportunitv costs in the labor and financial markets, but not in the land market. Earnings were high for family labor and for family assets, compared to the overall economv, making it worthwhile to continue farming in the frontier. Returns to land, however, were apparently low compared to rapidly appreciating frontier real estate, and farmers were motivated to move on and to reap capital gains successively from each plot of land. Settling a frontier, therefore, requires that land productivity keep up with land prices. Where this occurred, turnover on lots was low and settlement was relatively successful. High productivitv farming, however, turns out to be associated not only with high retention rates (low turnover) of farmers on the land, but also with high rates of deforestation on already occupied plots. Thus, while high productivitv farming indirectly prevents the deforestation of new plots farther inland, it promotes greater levels of deforestation directlv on the occupied plots. In order to escape this paradox, one must first analyze how deforestation rates respond to farm productivity and then devise appropriate economic policies to reduce "productive deforestation." The rapidity of turnover and defor- estation, as discussed in the present chapter, spurs a renewed sense of urgency in the search for policies to curb the frontier advance in the Amazon. Methodological Issues: Costs, Benefits, and Sustainability This study attempts to take as much real world variation into account as possible in explaining the performance of small farmers in the Amazon and identify those policies that could increase the sustainability of frontier small farming. Given the variety of conditions in the Amazon, the evaluation of performance, based on the calculation of benefits and costs, is perhaps as important as the conceptual framework of analysis, that is, how these costs and benefits are defined, measured, and analyzed. 42 Sustainable Settlemenit in the Brazilian Amazon Benefits of settlemennt In order to derive private benefits from settling the frontier, farmers must cover their opportunity costs in some or all factor markets: labor, capital, and land. These opportunity costs are calculated as follows: * Labor mna-ket. As half the Brazilian urban labor force earns less than a minimum wage (in real terms, approximately US$1,000 per year, varying with the annual inflation rate), the minimum wage is a reason- able upper-bound proxy for small farmers' opportunity costs in this market.4 In these terms, small farmers cover such opportunity costs if family labor income is at least one minimum wage per worker. * Capital market. The only financial instrument widely available to small savers is the savings certificate (cader-neta de poripan(a), which paid in 1991 a real interest rate of 0.5 percent per month. If this is used as a proxy for opportunity costs in the capital market, then small farmers cover these opportunity costs if the accumulation rate of all physical and financial assets is at least 0.5 percent per month.5 * Land market. The price per hectare of land is a straightforvard variable. Farmers cover opportunity costs in this market if the rate of return to the use of the land is at least equal to the percentage variation in the price of the land during the same period.6 As long as real wage rates in the economy as a whole are low compared to real earnings on the frontier, and accumulation rates for small capital in the informal sector of the economy are low relative to those of frontier farmers, then Amazon settlers will cover their opportunity costs in the labor and capital markets.7 On the other hand, as long as speculative demand for land hikes up real estate prices, then Amazon settlers will tend not to cover their opportunity costs in the land market. The private benefits of settlement increase as the private returns to labor, capital, and land rise and as the settler retention rate rises (that is, turnover on lots falls). The longer settlers remain on their plots, the longer they will accrue benefits. Thus, policies that increase farm incomes and reduce turnover should increase the benefits of frontier small farming.8 Costs of settlemenit In broad terms, the total costs of settlement include both private and environmental costs. Private costs comprise current and start-up costs, which include conventional fixed costs plus the farm-level costs of defores- tation. Relatively open-access frontier conditions tend to reduce these costs, as access to forested land tends to be cheap relative to established areas. Other private costs of settlement are related to uncertainty and lack of Land larkets and Sussainzn7be Frontier Farmning 43 information, which introduce considerable variance into expected returns on any one piece of land. Environmental costs arise with improper farming and mismanagement of nonrenewable natural resources. These costs include, among others, the loss of biodiversity and the erosion of soil. Although it is difficult to measure environmental costs, one can identify conditions in which they are reduced.9 For example, cost reductions occur per unit of benefit when, for any given deforested area, agricultural intensitv increases, such as when crop yields rise, the ratio of harvested to total cleared area rises, or the number of harvests rises prior to turning land to pasture or fallow. Very often there is an interplay betveen low private and high enxiron- mental costs. By providing cheap access to forested lands, frontier open-access conditions induce a wasteful use of soils (manv times cleared for extensive agriculture and ranching), increasing the environmental costs of settlement. Sulstainability Amazon farming will probably never be sustainable in the sense that it can fully replace all of the natural resources it uses. However hard to define and measure sustainability, when the ratio of private benefits to environmental costs rises, one could say that there has been an increase in sustainabilitv.'0 Using this notion, sustainability can be increased in tropical farming by: (1) reducing the environmental cost per unit of private benefit, for example, by decreasing the turnover or increasing farm income per hectare of deforested land; (2) increasing the private costs of environmental destruction, for example, by internalizing the social cost of deforestation; and (3) decreasing the private cost of alternatives that would cause less environ- mental damage, such as alternative tropical farming technologies. Turnover on Frontier Plots Overall, 64 percent of those interviewed in 1991 had been on the same lots in 1981, but this percentage varied wlidely across subgroups"1 (see appendix table A.1). Retention rates were higher in official colonization (67 percent) than in private colonization (61 percent). Within official colonization, it was higher among southerners (82 percent) than among other migrants (61 percent). Aside from Monte Alegre (81 percent), which is already in its third generation of colonists, the highest rate in Para was in Pacal (74 percent), and in Mato Grosso, the highest rate was in Mutum (68 percent). The lowest rate was among other migrants in Pacaja (48 percent), located close to Pacal via the Trans-Amazonian Highway. 44 Stuctainamle Settlement in tlje Bratzilianz Am,7zon Opportunity Costs of Frontier Farming In light of the variations in retention rates on lots, the following sections will investigate how economic alternatives compared with the opportunity costs of frontier farming. [imzge 7fte In each survey, the opportunity cost in the labor market was taken to be the minimum wage (NIV), Us$1,000 per vear. As mentioned previouslv, since roughly halfthe urban labor force in Brazil earns less than this, a lower value might provide a better comparison. Nonetheless, the minimum wage is the most widely accepted standard in Brazil and was therefore adopted here. Pavments to family labor can be approximated by measuring the value of monetary and nonmonetary yearly consumption per household and dividing this figure by the number of full-time adult workers.'" The result is an estimate of how much families were remunerating their own labor and can be compared to the going wage in the labor market. On average, there appears to have been a large increase in imputed remuneration to family labor over the 1980s, from 1.1 Mw in 1981 to 3.2 MW in 1991 (table A.2). This increase seems to have been largest for southerners and those in private colonization projects. The imputed wage is significantly higher in Pacal (4.01 MWV) and in Mutum (19.54 M\\). The dollar value of the imputed wage also grew (table A.2). Throughout the 1980s, exchange rate devaluation was smaller than the loss in real value of the minimum wage, so the change from 1981-91 is smaller in table A.3 than in table A.2. Differences among locations remain roughly the same. In this case, only Mutum, with its outstanding value (LUS$19,000), differed significantly from the rest. On the average, family workers earned around US$4,000 per year, ap- proximately US$333 per month, which is about four times the vearlv .minimum wage.'3 Even those in the lowest income groups self-remunerated above the minimum wage, albeit narrowlv so. Note in table A. I for example, the "Others" in Monte Alegre, as well as those in the highest turnover locations, Saojose do Rio Claro and Anapu-Pacaji. Thus, although much of the sample was verv poor by international standards, it did better than average by domestic standards, given stringent conditions in the Brazilian labor market.'4 Using this criterion, therefore, frontier farming covered its opportunity costs in the labor market. Interest rate In the financial market, the opportunity cost of frontier farming is equivalent to the interest rate paid on the most widespread saving instrument for small La.nd Markets and Sustainable Frotntier Farming 45 and medium savers in Brazil, the cadern eta de poupanca. In 1991 this certificate paid 0.5 percent per month plus monetary correction. The rate of accumulation is the average monthly percentage increase in net worth since arrival on the lot. It is measured by comparing all real and financial resources brought to the frontier with the value of net worth in 1991.15 This concept is different from that of rate of return, which shows how much net income was earned on farmers' real assets over time. The accumulation rate is an alternative measure of asset appreciation.16 Appendix table A.4 shows that the average monthly real rate of accumu- lation was very high, about 2.3 percent, or more than four times the interest rate. Mutum, once again, had the highest figures. Although there were slight variations across locations and across migrant groups, these were statistically insignificant and small compared to variations in imputed wages. The rate of accumulation, as measured, does not imply net additions to productive capacity, but merely additions to net asset values. Land is an important part of farmers' net worth (about 50 percent, on average), and rising land prices have contributed in a large measure to land appreciation (approximately 80 percent, on average). So capital gains have inflated accumulation rates by approximately 30 percent,17 meaning real rates of accumulation were around 30 percent lower than those shown in table A.4. Even so, these real rates would still be at least double the interest rate. Farming, therefore, was very good business all over the frontier and covered its opportunity cost in the financial market. Price of land In the land market, the opportunity cost of land is the price of land per hectare."8 This can be measured by weighting the microregional prices of forested areas, pemianent or temporary crop areas, pasture areas, and fallow areas bv the number of hectares devoted to each of these uses. Subsequently dividing the result by total land area gives one overall land price per plot.'9 This is the value the farmer would get by selling land under current market conditions. In this way, the change over time in the price of land can be compared to the rate of return gained from farming it.20 During the 1980s, the price of land rose considerably in most of the frontier. As we have seen, tax and credit incentives, large-scale colonization and titling programs, the laying down of physical and social infrastructure. and an inflationary economy all boosted demand for frontier land.2' As people moved into the region, and more and more land became private property, the frontier land market began to operate dynamically, and real estate appreciated far more here than in the rest of the economy.22 According to calculations based on table A.8, during the 1980s average land prices seem to have appreciated much more in the public colonization projects of Para (5.1 percent per year) than in the private ones of Mato 46 Sistainal'le Settlement in the Bi-azilian Amazon Grosso (1.4 percent). The highest rate of increase was in Anapu-Pacaji (I 1.1 percent), where land prices had been the lowest. The next highest were Pacal (6.6 percent) in the state of Para and Mutum (6.1 percent) in the state of Mato Grosso. In Alta Floresta-Paranafta, however, land values declined (-6.0 percent). Across locations, land prices followed no trends. Thev sometimes rose with agricultural productivity, as in Mutum and Pacal, but thev could appreciate despite declining productivity, as in Anapu-Pacaja. Neither did land prices necessarily reflect the amount or quality of infrastructure or government services available. Alta Floresta, known as the jewel of coloni- zation and rated highest for physical and social infrastructure among all surveyed locations, witnessed declining land prices. Anapu-Pacaja, one of the most neglected areas along the T rans-Amazonian highway, saw sharply rising land prices. It appears, therefore, that exogenous factors contribute to variations in land prices in the Amazon frontier. Many determinants in widely varying weights are active in different places, and any generalizations would be simplistic.23 Land prices should be compared to the returns to land in agriculture in each location, to ascertain whether farmers are covering the opportunitv cost of frontier farming in the land market. Unfortunately, it is difficult to measure the return to land, because factor remuneration is mixed with land rent. Open access conditions in a frontier further complicate the issue, as the price of land in relatively remote locations is, theoretically, zero (Cam- pari 1993a and Deacon 1990). An added problem is that landed property rights have traditionally been poorly defined or enforced in the Amazon, with much of the basin having been exploited (under ambiguous titling concessions) by activities that were less focused on the land than on the forest covering it. This, of course, is the case in extraction, whether in forests (for example, rubber tapping, nut gathering, logging) or rivers (prospecting, fishing, and so forth). The patchwork effect of haphazard land titling has only confused matters even further, making it practically impossible to measure directly the real productive returns to land on the frontier. The next section, therefore, will look at indirect indicators of such returns. ProductivitY, land use, and agricultuyral prices The returns to land depend on the fertilitv of the soil, the intensiveness of land use, the productivitv of other factors,24 and prices of agricultural products.25 See appendix tables A.5 to A.9. Total quantum productivity (table A.5) adds together kilograms of all crops and divides by harvested area. This, however, is flawed as an indicator of land productivity, because Pacal's extremely high productivity average, largelv based on its production of sugarcane, an extremely bulky product, distorts total averages. The distri- Lnnd larkets and Sustainable Frontier Fannming 4 7 bution of annual yields of rice, corn, and coffee (table A.7), however, indicates interesting productivity variations. Overall, productivitv in- creased, with the greatest increases in Pacal (especially coffee) and Mutum, although there were some declines in Anapu-Pacaja and Alta Floresta- Paranaita (rice). In several cases, considerable gains occurred over the ten-year period, even in temporarv crops. Such gains mav indicate that, as farmers gain experience, they learn to identify the best soil in their properties and acquire seeds and technologies most suitable for local conditions. Productivity gains are to be expected, of course, for perennial crops, such as coffee or cocoa, as trees mature. The significance of the general increase is considerable. It shows that contrary to widespread belief, productivity has not tended to decline in Amazonian soils.26 The average productivities of this group also compared well with national averages.27 The few cases of constant or slightly declining productivities, such as in Alta Floresta-Paranaita or Anapu-Pacaji, indicate involution from farming, eradication of perennials (mainly coffee in Alta Floresta), and the conversion of most of the deforested land to pastures. Elsewhere, as in Monte Alegre, slash-and-burn shifting agriculture left behind low productivity crops and expanding fallow areas to which farmers rarelv, if ever, returned. Table A.6 shows the low overall percentage of deforested land used for temporary and pemianent crops (32.5 percent). In Para, southerners were significantly more intensive farmers than others (36 percent compared with 26.9 percent). The highest intensity of land use for agriculture was in Mutum (84.7 percent), followed by Pacal (39.8 percent), with the lowest in Monte Alegre (18.4 percent), where many very old plots have already been totally deforested and abandoned by their owners. Table A.8 indicates that agricultural prices during the 1980s were stag- nant for basic temporary crops (rice and corn) and drastically declined for the main perennial (coffee). These figures are less than half those quoted at the international level, reflecting high transport costs to ports and to regional and national markets, as well as monopsonized frontier markets.28 Thus, during the 1980s, rates of return to land in the Amazon seem to have been rather poor in the face of stronglv rising land prices. Agricultural prices were so low that, except for showcase locations, such as Pacal and Mutum, many farmers gave up on agriculture, sold their land, reaped capital gains, and moved on. Some went to other frontiers; others went to urban centers, where they set themselves up in business with the proceeds of their land sales. In fact, many of the most successful colonists in the sample had done exactly the same in the past. They had moved from frontier to frontier, as had their fathers and grandfathers before them, buying and selling land as they went, in a process of "itinerant accumulation."" The less successful were even more likely to sell out, as rising land prices increased the opportunity cost of their low productivity farming. 48 Susrtainable Settlement in the Brazilian7 Amazon Summai-y: labor, fiinancial, and lan)d markets on the fi-ontier- In the 1 980s frontier farming does not seem to have covered its opportunitv cost in the land market, but these costs were covered in the labor and financial markets. The combination of high returns to capital and labor and low returns to land reduced the advantage to a farmer of remaining long on a specific plot of land, but not the advantage of continuing to farm in the frontier area generally. The opportunitv costs to frontier farming apparently contributed to farmer intraregional migration, with high capital gains spurring turnover in old frontiers and financing the opening of new frontiers. Newcomers: the merchant fi-ontier There seems to be a two-stage process of frontier occupation in directed settlement projects that is not unlike the traditional frontier process in the rest of Brazil.30 Original family farmers deforest, leave, and are replaced by newcomers who belong to a higher income bracket and use land for different purposes than did the original settlers. Tables A. 10 A.I 1, and A. 12 suggest that these newcomers, embedde(d in the 1991 total, wvere younger and significantly, wealthier than survivors in high turnover locations such as Anapu-Pacaja, Alta Floresta-Paranaita, and S5o Jose do Rio Claro. Many were probablv not farmers at all, but rather merchants, public servants, and other city dwellers.3' Ihere is still much work to be done to identify who these newcomers were, where thev came from, what their motives were for buying up frontier lands that had already significantly appreciated, and whatever other distinguishing characteristics they may have had.32 WNNhat is clear is that the process that shifts the frontier is accelerating. Oveerall pe7Jornnance of smanllfna11 ier settlenments dur-ing the I 980s There are many more causes for high turnover on frontier lands than those discussed thus far (Mahar 1989).33 Tables A.12 to A.43 describe directed colonization in the Amazon frontier according to the variables defined in appendix B. Time and space limitations prevent a detailed discussion of each table here. Appendix C organizes these variables into a consistent model of frontier farming, and appendix D statisticallv analyzes their impact on deforestation. This section briefly summarizes the principal findings.)4 In general. Amazon colonists did well during the 1 980s, not only in terms of being able to cover opportunity costs in the labor and financial markets, but also in terms of the absolute values of their net worth (table A. 12), which increased significantly from 1981 to 199 1. Most (71 percent) expressed the opinion that they had improved their lot in life (mellorotu de vidla) bv Land Markets and Snstainable Fronztier- Faming 49 migrating to the frontier (table A.13), with the highest ratings occurring in low turnover locations (Mutum, Pacal, and Monte Alegre). An even higher proportion (76 percent) planned to stay on their present plot (table A. 15), though only 41 percent planned to invest further in these plots (table A. 14). This apparent inconsistencv-planning to stay on, but not in agricul- ture-may indicate that in 1991 these farmers did not expect much from agriculture and planned to diversify into other activities. In fact, as table A.8 indicates, the evolution of agricultural prices during the 1980s was not promising. Except where agriculture was most productive (Mutum), gross agricultural income (table A.16) covered less than half of total household expenditures (table A.17).36 Many farmers, therefore, diversified into non- agricultural activities, such as wage employment, businesses, rents, transfers, and so on, to the extent that nonagricultural income (table A. 18) became generally larger than agricultural income. In Alta Floresta and Paranaita, agricultural income practically disappeared by 1991. This represents a drastic change from the previous decade, when nonagricultural income was, on average, 10 percent of total income. Nonmonetary, or subsistence income, was measured by imputing market prices to all goods and services produced for purposes of family consumption or production, agricultural or nonagricultural (table A. 19). This value also grew as a percentage of gross income (table A.20), from roughly 25 percent in 1981 to roughly 35 percent in 1991 (Oz6rio de Almeida 1992, chapter 16). Once current household consumption and productive expenditures were met, current net income levels (table A.2 1) were lower in Para (sometimes negative) than in Mato Grosso.37 Net current income, in this sense, is somewhat meaningless, as it is net not only of productive expenditures, but also of household expenditures. Solvent farmers have positive net current income, and insolvent farmers have negative net current income, which, in turn, indicates whether there is net debt or net investment.38 High net income figures in Mato Grosso, in the face of declining crop area during the 1980s, indicate that these colonists were no longer mainly farmers, another important change relative to the past.39 Tables A.22, A.23, and A.24 show, respectively, that one-quarter of the sample had owned land elsewhere before coming to the present location; all had farmed in at least one place before arrival; and the parents of 86 percent of the colonists had been farmers. Yet, as noted in table A.14, less than half of these farmers still intended to invest in their present lots either because the lots were not sustaining family farming (negative net incomes imply indebtedness) or because the farmers were already thinking of moving on. Investment is difficult to analyze, as it refers to dynamic decisions that go beyond current production. Net investment (table A.25) is defined as investment (all expenditures intended to increase net worth, after the current agricultural year) minus disinvestments (sale of all durables, reduction in stocks of physical and financial assets, and so on).40 Thus measured, invest- 50 Sustainable Settlement in the Brazilian Amazon ment turns out to be negative or small in most of Pari but quite large in Mato Grosso. (This result is consistent with Para's negative and Mato Grosso's positive net incomes.) These differences in investment help explain the pattern of deforestation that was occurring in each location. The major investment cost item in the Amazon frontier was deforestation (table A.26). Since arrival on the lot, by 1991 deforestation had eaten up 54 percent of total land held (table A.27). In the public projects of Pari, southerners had deforested a smaller proportion of total land (46 percent) than others (63 percent), but this is because, more than the decade, south- erners were purchasing new forested plots and expanding their holdings more rapidly than others, especially in Anapu-Pacaji. In Monte Alegre, the oldest colonization project, plots were already over 80 percent deforested. Among the private projects of Mato Grosso, the largest plots and the greatest deforesters were, by far, the colonists of Mutum, who had alreadv cleared 69 percent of their lands bv 1991. The pace of deforestation varied widely from place to place in 1991 alone (table A.28). Overall, it appeared to be approximately 8.9 percent per year: 3 percent in Pari and 18 percent in Mato Grosso. But most of Mato Grosso's projects were deforested at around only 1.5 percent a year, while Mutum, with the largest plots, was going at 29 percent. In sum, colonists with lower net agricultural incomes (as was generally the case in Para) had less left over for investment and deforested less; colonists with higher net agricultural incomes (as was the case in Pacaja, the exception in Para, and in AMlutuin in Mato Grosso) had more left over for investment and deforested more. The source of income apparently matters, as farmers with relatively high nonagricultural income (as was the case in Alta Floresta) deforested less, despite the fact that they did have positive balances left over for investment.41 Interestingly enough, fully 83 percent of Mutum colonists, who defor- ested the most, declared that forest preservation is important (table A.30). Yet only 4 percent of them perceived that loss of soil fertilitv might become a problem (table A.3 1), contrary to the rest of the sample (43 percent), and none of them practiced any kind of conservation technique, such as crop or area rotation, compared to 21 percent overall (table A.32). Those who were least worried about soil fertility and conservation were located in projects with the highest farmer turnover: Anapu-Pacaja in Para and Alta Floresta- Paranaita in Mato Grosso. Thus, low soil fertility to begin with, rather than eventual loss of soil fertility, may have had an important association with the exodus from these locations.42 Market conditions may also have had an important influence on colonists' performance (Repetto 1989). Insufficient storage facilities generally led farmers to sell 82 percent of their product during the first three months after harvest (table A.33). Distance to market (table A.34) and insufficient trans- portfacilitiesled them tosell 53 percent atthe farm gate (table A.35). Mutum and Monte Alegre, however, due to active cooperatives, sold only around LaInd Aarkets and Sustainable Frontier Failning SI 10 percent at the farm gate. For the same reason, agricultural credit was also greatest for Monte Alegre and especiallv for Mutum (table A.36). Loans add to current resources if incoming flows are greater than debt repayment on former loans. If repayment burdens are greater. then current account resources must be used to service debt outstanding. Chronic indebt- edness leads to dependence on local merchants and to various debt-peonage conditions that frequently lead farmers to sell out and creditors to move in (Oz6rio de Almeida 1991b). This important motive for turnover is the hardest to observe empirically, as fanmers are loath to reveal their debts or their creditors. For this reason observed indebtedness is deceivingly small (table A.37). Negative net current incomes are a better indicator of indebt- edness, as they indicate that farmers are covering their current expenditures either through borrowing from themselves (that is, by selling durable goods, such as cattle) or through borrowing from others. In either case the farmer is insolvent.43 The capacity to escape informal sector (usury-mercantile) indebtedness depends mainly on titled property ownership of the land (table A.38).44 One hundred percent of the colonists in Mutum held such title, as did 84 percent in the whole sample, certainly a much higher percentage than is typical of Amazonian small farmers. Although manv farmers had low absolute income levels, thev still earned more than did half the labor force in Brazil. Their net worth, rates of accumulation, access to credit, productivities, and other economic charac- teristics set them off as part of a small farmer elite within the Amazon. 'To the extent that such benefits are attributable to the fact that they were in directed colonization projects, then colonization may deserve reevaluation for its social and distributive impact.45 Yet the environmental destructive- ness of colonization is considerable. It contributes to deforestation directly, in the projects themselves, and indirectly, by allowing for high turnover on plots. Turnover fuels intraregional outmigration, which, in turn, promotes further deforestation. Farmers' Economic Decisionmaking in the 1990s Given the conditions described throughout this chapter, small farmers in the frontier must constantly make decisions on whether to sell or to keep their land and whether to hoard or to farm the soil. Much in the literature deals with the farmers' discount rate as an important determinant in their patterns of resource use.46 This section will examine concrete economic factors upon which farmers base their decisions. It is not the intent here to analyze how expectations arise; rather this section describes how such expectations inform the farmer's decisions to continue or abandon farming his plot. 52 Swstainable Settlement in the Bra7zilian Amazon A frontier farmer in the Amazon faces three choices in a sales decision about his land: he may choose not to sell his farm and instead productively farm it; he may choose not to sell, but leave his plot idle, keeping it only as a store of value-that is, hoard it; or he may choose neither to farm nor to hoard but to sell his land at the going market price. To deal with this varietv of choices, the farmer must be able to discount, at the moment of his decision, the expected payoff of his choice. Thus his decision depends on the net present value of agricultural income (NPVA) and the discounted salvage value of land (NPXL)4 as computed in table A.42.48 By adding both figures, the farmer can determine the overall net present value of his plot (NPV) and compare it to the going market price of land. If the NPV is higher than the price of land, the farmer is likely to keep the plot; othervise, he will probably sell it. Comparing the NPV to the going market price of land, we found that in half the sampled locations farmers would be better off selling the land. Not surprisingly, this occurs only in locations where the NPVi exceeds the NPVA, that is, where returns to agriculture are not keeping up with land prices. Comparing the NPVA to survival rates on plots bears this finding out: survival rates are lower in areas of relatively low NPVA. High turnover on plots is thus positively associated with low returns to farming, as summarized in table A-42. In Mato Grosso, land sales peaked in Alta Floresta, Paranaita, and Sao Jos6 do Rio Claro, all of which are areas of agricultural involution, encroach- ing pastures, rapid urbanization, and rapidly appreciating land prices. In Para, land sales peaked in the region of Anapu-Pacaja. In that particular location the NPVA and NPVL were approximately the same, us$33 and US$3 1, respectively. The average price of land in Anapu-Pacaji was US$59 per hectare, making the NPV only slightly higher than the price of land (US$64 compared with US$59, respectively). As land prices continue to appreciate rapidly in this area, farmers are, in fact, selling out. For those farmers who choose not to sell, that is, those whose NPV exceeds the going price of land, the question is whether to produce. A farmer is likely to produce only if the NPTA exceeds the NPVL, meaningr that his discounted income stream from agriculture must support his decision to forgo the potential gains from land sales. In this case, the farmer has motivation to produce; othervise his plot will remain idle. The results of this section are summarized in the following matrix (refer to appendix table C. I for a supporting and complementary argument): Factors that inform a farnner's decision to retain or relinquish his plot NPV'>P NPV

NTXi Case 1: keep the land in prodnction Case 2: sell the land NPVA < NPNT Case 3: keep the land out of production Case 4: sell the land (hoard) Land Markets anid Sastainable Frontier Fa7ning 53 These findings confirm that farming in the Amazon is unlikely to be sustainable unless land productivity and, consequently, agricultural income keep up with land prices. In high turnover locations, the NPVA is not sufficiently high to retain farmers in their plots and motivate production. In other words, the price of land is still higher than the expected income stream from agriculture. For the 1990s, therefore, the rational behavior of many frontier farmers in the Amazon would be to sell the land at the going market price. IThe implication of this for regional development policies in the Amazon is evident. Policies should aim at keeping the net present value of agricultural income level with that of land prices (see chapter 6 for recommended policy types). If the gap between agricultural incomes and land prices becomes too large due to rising land prices, the NPVL is likely to rise beyond the NPVA, in which case it would pay not to produce at all but to keep land only as a store of value (that is, in areas where NPV> P). This would further fuel the speculation under way in the frontier land market.49 C'ost to the global economy of red7cing deforestation Standard microeconomic theory dictates that the price of an asset is equal to the present value of the discounted stream of earnings that it can generate. In the case of agricultural land, these earnings are, generally, the stream of net incomes derived from agricultural production and discounted in time. Despite wide variations across frontier locations, monetary agricultural incomes tend to be low in absolute terms. Among small farmers in the Amazon, the average NPVA per hectare is only us$129.60 (appendix table A.42). If the standard theoretical construct were to prevail, then the cost per hectare to the global economy of buying Amazon land to reduce deforesta- tion could be set as low as us$129.60. Such costs, however, reflect only the productive potential of the land and not the capital gains that can be derived from speculation.50 Yet the cost to the Brazilian economy as a whole of reducing deforestation is actually much larger than that of net present value of agricultural income, or the NPVA forgone (Young 1994). This is because monetary agricultural income has been declining relative to total income (monetary and nonmone- tary, agricultural and nonagricultural)-at least among frontier farmers. Total income and consumption are relatively high among the same farmers, when compared to economic alternatives in the rest of the economy. Furthermore, frontier land prices are inflated beyond the INTVA, so part of income forgone in the sale of land is the capital gain it affords. The farmer, therefore, would not want to sell at a price equivalent only to the NPVA. The Amazonian farmer is not the only one who forgoes income when he stops planting. Frontier farming feeds a large and growing commercial and processing sector inside and outside the Amazon where income would also 54 Sustainable Settlement in the Brazilian Amazon be forgone if frontier agriculture were to be reduced. For all of these reasons, the cost per hectare to the global economv for containing deforestation in the Amazon is higher than that immediateiy apparent from direct measure- ment of the NPVA or even hand prices. Conclusion: Land Markets and Sustainable Frontier Farming Much of the literature on the shifting frontier emphasizes penury as a cause of high turnover (Mink 1993). Penury is triggered by the following factors: ecological-technological problems (nutrient mining) of shifting agri- culture; * social violence in competition for frontier lands; * exclusion of many small farmers from official land titling and other formal types of property legitimation procedures; and, finally, * economic push factors, such as indebtedness and debt peonage. Although relevant, however, penury is not all that drives small farmers deeper into the Amazon forest. In colonization projects originally intended for the poor, many small farmers leave because they can accrue capital gains by selling their lands. Those who buy the land and move in are not poor at all. Although farmers may cover opportunity costs in the labor and capital markets and consume at a level three or four times that of the minimum wage, this does not mean that the majority of them are well off in any absolute sense. Houses are rustic, health care is dismal, schooling is minimal, and transport and communications are sorely insufficient. Even if they are better off than the majority of the Brazilian labor force at the same skill level, most farmers are still poor and will only settle down in their current plots if it becomes worthwhile to do so, that is, if they cover their opportunity cost in the land market. Given the pressure of inflated real estate markets on the frontier, however, covering such costs becomes possible only if land becomes highly productive. During the 1980s, high productivity frontiers had high survival rates and low turnover on plots. Conversely, low and/or declining productivity frontiers had low survival rates, and turnover was high. Even some colonists with relatively high incomes and net worth withdrew from agriculture and converted their lands into pastures or even sold their plots and moved away. Newcomers were not eager to farm the deforested land but mostly held it for speculative purposes. Frontier land markets fueled high turnover among all but the best farming communities. In high turnover locations, those who moved in, and those who remained behind, became disenchanted with agriculture. Poor prices, poor marketing, poor credit facilities, and weak institutions led them to diversify into pasturing and nonfarming activities. Thev showed concern about soil fer- tility, professed to practice some forms of conservation techniques, planted Land Markets and Suistainable Frontier Fa-rming 55 little, and deforested little (about 2 percent of their holdings per year). In low turnover locations, agriculture boomed, with little worry about conser- vation or loss of soil fertility. Lots expanded by incorporation of new areas, credit was available, and defbrestation proceeded at around 20 percent per year. Thus, on one the hand, high turnover seems to be associated with agricultural involution and nonproductive use of already deforested lands. On the other hand, low turnover seems to be associated with agricultural expansion and rapid deforestation. In 1991, the price of land in many locations was higher than the overall net present value (NPV) of potential earnings in any given plot, prompting farmers to sell. Where the NPV was higher than land prices, but returns to land (NPVL) were still higher than returns to agricultural production (NPVA), a farmer kept his plot but did not farm it, holding it instead as a store of value. Land sales (which occur when the NPV is lower than the price of land) and speculation (which occurs when the NPV is higher than the price of land, but the NPVA is lower than the NPVL) constitute the heart of high turnover rates in Amazon settlements. Thus, private benefits of settlements (as indicated by high intensiveness and falling turnover) per unit of environmental cost (as indicated by defor- ested area) evolved differently throughout the frontier. The outcome ap- pears to have been a general decline in sustainability. In low productivity frontiers, this was due to a decline in private benefits from settlement, as intensiveness decreased and turnover increased. Here, the environmental cost of settlement increased slowly, as deforestation rates were relatively low. In high productivity frontiers, on the other hand, the rise in environ- mental costs was the main cause of decreased sustainability. Here, the private benefits of settlement were high, because intensiveness was relatively high and turnover relatively low. Environmental costs rose rapidly as deforesta- tion increased, sometimes at alarming rates. This chapter analyzed one factor that would increase the sustainability of frontier farming: reducing turnover on plots. Policies that would increase agricultural prices and/or agricultural productivity on the frontier would help farmers resist giving into the pressure of frontier land markets, thus diminishing the rate at which they moved out to deforest new frontiers. The problem is that high productivity farming is not only associated with low farmer turnover but also with high rates of deforestation on current plots. Thus, while improved farming may reduce deforestation farther inland, it accelerates deforestation in the current location. The shifting frontier may always have been a feature of Brazilian agricul- ture. But the evidence in this chapter indicates an extraordinarily rapid rate of turnover in some locations, suggesting an acceleration in the general rate of frontier advance in recent times. The massive migration to the Amazon in the 1970s increased the stock of small farmers to relay forward from old 56 .Srstainable .Settlenmenit in the 8r1azilian Aniazon frontiers to new. So, despite the decline in interregional migrations over the 1980s, the flow of intraregional migrations may have sped up. increasing the urgency of finding ways to improve farming without accelerating defores- tation on already occupied plots. Notes 1. Directed colonization projects are public (official) or private. 'Vhile official projects were promoted by the government, private settlement companies developed their own ty-pe of project. For a description of the two types, see Oz6rio de Almeida (1992, chapter 8). This source also contains a description of the directed colonization program in Brazil and detailed information on empirical procedures. For a treatment of spontaneous colonization see Barbira-Scazzocchio (1980), Moran (1982), Musumecci (1988), Porto Tavares (1972), Velho (1981), and Wesche and Bruneau (1990). 2. See Feder and Feeny (1991). 3. A preliminarv test for the relationship between tenure securitv and deforesta- tion is presented in chapter 5 and in Pearce and Warford (1993, chapter 1). Goodland and Daly (1993), Lopez (1992), and Lopez and Niklitschek (1991) explain the linkage among secure tenure, poverty, and environmental degradation. 4. The going minimum wage rate is an overestimate of the opportunity cost to small-farmer labor. It does not take into account the probability of earning less (for example, 50 percent in Brazil overall), which should be discounted. So the compari- son made in this chapter is unfavorable to the frontier small farming alternative. 5. In Portuguese, capital market (me-rcado de capital) refers to financial markets in general; in the United States, capital market sometimes refers only to long-term financing. In this text no distinction is made between long- and short-term capital markets. Also, see the formula for the calculation of the accumulation rates in appendix C.1.3, expression 5. 6. The price of land was observed independent of construction and improvements (which were counted as a separate part of networth) and of use. The price of different types of land-forested lands, cleared lands, lands covered with permanent crops, fallow lands, and lands considered not suitable for agriculture-were valued sepa- rately, then the microregional gradient of prices was applied to the proportion of different uses in sampled plots. This method minimizes the tendency for land price to become correlated with productivity variations and preserves the exogenous components of land price wvith respect to land productivity and behavioral variables. This point is crucial to the analysis of turnover on plots and speculative versus productive deforestation (the subject of chapter 5 and appendix C). 7. Although the rate of return to capital may be high among large firms, it is compressed to a minimum among small firms, which represent the main occupa- tional possibilities for rural-urban migrants. Thus, small farmers do not have access to the high rates reported for Amazon merchants in the previous chapter but can expect the low rates typical of microenterprises, especially those in the informal sector. See Oz6rio de Almeida and Graham (1994) for a discussion of the rates of return to capital in the informal sector. Land Alarkets and SIsMtainable Frontier Fa7rming 57 8. The benefit of small-farmer settlement for nonfarmers, such as users of roads, and those employed by surveying. zoning, extension, and titling agencies, as well as in plisical and legal infrastructure projects, located outside settlement projects, will not be taken into account nor will general increases throughout the economv in opportunity costs of Amazon farming. This means omitting discussion of improve- ments in employment, incomes, income distribution, small firm earnings, interest rates on savings certificates, and other factors in the economv as a whole that would reduce the demand for frontier settlement. 'Fhese issues are much broader than that of Amazon settlement itself, involving a discussion of the use of policies to combat recession, promote development, and so on. The policies investigated here are onlv those that directly benefit small farmers already settled in the Amazon. One important point will be taken up. however-that of the relationship between the economic crisis in Brazil and deforestation. As the crisis is resolved, the oppor- tuniry cost of frontier farming should rise and the rate of deforestation fall. This suggests that the decline in interregional migration to the Amazon during the 1980s was not caused by the economic crisis in and of itself, but by overall structural shifts in the population and economy. This is a significant point because it contradicts dominant expectations in the literature. 9. The issue of valuation of natural resources is discussed from differing points of view by several authors, such as Ahmad, El Serafy, and Lutz (1989), Pearce and Markandya (1989), Pearce and Warford (1993), and Stirling (1993). 10. Important attempts at a rigorous definition of sustainability are in Goodland, Dalv, and El Seraf* (1992). See also Pezzey (1992) for a general review on different concepts of sustainability, as well as Pearce and Atkinson (1993). 1 1. This and subsequent sections of this chapter rely heavily on the sampled locations described in box 4.1 and in appendix tables A. l-A.43. Prior to reading this section, one should read box 4.1, which gives the names and main features of each sampled location. 12. Consumption figures, rather than income, are used to in order to avoid the problem of double counting with respect to total factor productivity. A household's total income is derived from the productivitv not only of labor, but also of capital and land. To divide total income by labor input is to inflate the returns to labor. Total household income, however, was more than twice the value of goods and senices purchased and/or produced for the purpose of consumption by household members. This illustrates the importance of farming (non-household oriented) expenses in farmers' budgets. Variations in income were much wider than variations in consumption; that is, richer farmers spent more and poorer farmers spent less on their units of production but tended toward similar levels of expense on consumption per capita. The decision to produce is thus, apparently, a very different one from the decision to allocate resources to family consumption. In this sense, farmers appear to be truly remunerating household labor, by whatever criteria, not merely distrib- uting total income among them. (We thank the participants at the FsP Labor Market Seminar at the World Bank for discussing the issue of total factor productivity with us.) Note that, to the extent that labor's contribution to total income is not fullv accounted for, the estimates of labor remuneration presented in this section are underestimates of the true imputed wage in Amazon frontier farming. 13. Schneider (1993), in reviewing the first draft of this text and comparing it to an FAO/tJNDP/N1ARA 1992 smidy, comments: "In terms of income, settlements in the 58 Sustainable Settlement in the Blrazilian Amazon North generated incomes four times Brazil's minimum wage-larger than those in any of the other regions other than the South. The abilitv of settlers to accumulate household durable goods and productive capital... reflects the incomes shown above, with the northern region again surpassed only by the South. Northern settlers more than tripled their assets (increased by 222 percent)" (page 3). 14. Although the farmers in the sample are poor, they are better off in the frontier than they would be elsewhere. 15. Net worth was calculated by including the value of land and that of all fixtures and constructions on the land as of 1981, as well as equipment, other durables, and financial assets. Calculations were in constant 1991 dollars, as seen in appendix table A.2. 16. See the discussion in chapter 3 on the use of the rate of accumulation rather than the rate of return. Note that the use of the rate of accumulation avoids the problem of double counting that would occur if one were measuring the return to capital. The issue here is not how to distribute total factor productivity among the different factors but merely to ascertain how much net worth has appreciated over time. 17. Productive land appreciation can be factored out from total accumulation rates. This is done by using constant prices to recalculate changes in the value of land over a given time period (for example, 1981-91) in patches that are forestland (or otherwise not useful for agriculture) or devoted to crops (permanent or tempo- rarn) or pasture, or are fallow. This result (productive land appreciation) is then subtracted from the total variation in land price over the same period, yielding an estimate of the value of speculative land appreciation. See Oz6rio de Almeida (1991 b) for this methodology. 18. As noted in the previous paragraph, there is some double counting in dealing with land separately from net worth. Land is, on average, half of farmers' asset value, and the rate of accumulation is influenced by the rate of increase in the price of land. This means that land is important for farmers' decisionmaking in complex ways and influences his (idealized) calculations in different ways. In terms of deciding where a farmer should put his savings, land price is a factor in calculating the opportunity cost of capital-that is, the accumulation rate. In terms of deciding whether to farm the land cr to speculate with it, land price in and of itself constitutes the opportunity cost of farming. 19. This estimate of the price of land does not take into consideration structures and other fixtures on the land. These components are part of total net worth and are included in the calculation of the rate of accumulation. The price of land used here is a microregional average for each type of land use; it is proportionate to the area each plot devotes to different uses but not to the specific productivity of each plot of land. This reduces the estimated variation in land price in each location but does not alter the value of the mean. The estimated price of land rises with agricultural intensity, increasing as land use is shifted from forest, to temporary crops, or to perennials, and declines as land use shifts to pastLre and to fallow. Other work has shown that productive land use accounted for approximately 50 percent of the increase in the price of plots during the 1980s, while increases in the price per land-use-nwpe accounted for the other 50 percent. Variations across locations were great. See Oz6rio de Almeida (1990 a and b). 20. The reasons for variations over time in the price of frontier land are complex. Rental values become capitalized in the land price, according to a distance-to-market Land iVInmkets and Sustainable Frontier Fa7ming 59 gradient, in a "Von Thunen Ring" effect, as noted hy Schneider (1993). Such gradients are, in turn, altered by the changing locus of economic activity and the evolution of governmental, institutional, and other convenience factors. As urban economies evolve differently, so too do their impacts on land prices in various hinterlands. Speculative demand for land also interferes with land prices, influenced by the evolution of local conditions as well as those of the overall economy, especially inflationary experience and expectations. The objective in this work is not to analyze the evolution in the price of land per se but to observe this evolution directly and to analyze its impact on small farmers' decisions to stay or move on. 21. See Oz6rio de Almeida (1992) for a description and analysis of these programs during the 1970s and 1980s. Also see Brandao and Rezende (1992) for an interesting model of the effect of credit and inflation on land prices in Brazil. 22. See Ozorio de Almeida (1990a and b) for comparisons between real estate appreciation on the frontier and in the rest of the economy. See also chapter 3. 23. Exogenous land prices will be an important feature of the econometric model presented in appendix C and estimated in appendix D. 24. This section does not calculate the specific contribution of individual factors of production to total factor productivity. Total agricultural income (derived not only from the land but also from labor and capital) is entirely attributed to the land, causing an overestimate of the returns to land. \'When calculating the imputed wage rate, the problem of double counting was avoided by using household consumption rather than income. In the case of the accumulation rate, the problem was avoided by taking into account only changes over time in net worth, not in income flows. In this section, however, the problem of double counting could not be avoided. This overestimate of the returns to land weights the case in favor of productive, rather than speculative, demand for land. Even so, the speculative demand for land predominated. So the conclusions of this section are strengthened, not weakened, by the imprecision in its methods. 25. See Renkow (1993) for a comparative view on the issue. 26. In support of this, scientific studies have found only recently that most cleared soils in the Amazon do not lose their fertility for agriculture. It has been widely believed that most soils in the Amazon region were of an aluminum-and-iron-rich type called lathyritic. When deforested and exposed to weathering, such soils quickly lose fertility and harden. According to the new scientific evidence, however, only about 30 percent of the region has such soil, and most of that is deep within the rain forest. See Rensberger (1993). 27. According to the 1980-85 agricultural census, average rice vields (kilo- grams/hectare) in the Amazon and ce77ado frontier areas were: 1980 1985 Rice 1,266 1,3 15 Corn 944 1,183 Coffee 904 1,050 Comparing these vields with those of appendix table A.7 is instructive. In 1981, sample vields were slightly lower than regional yields in the case of rice and corn and much lower in the case of coffee. In 1991, sample averages were higher than regional averages for rice and corn but much lower in the case of coffee. Except for 60 Sustainable Settlement in tbe Br-azilian Amazonz coffee., productivity growth in crops appears to have been higher in the sample than in the regional average. 28. See chapter 3 on frontier markets; also Oz6rio de Almeida (1992, chapters 15 and 18). These findings imply that improvements in roads, marketing facilities, and whatever else might contribute to improving frontier agricultural prices wiould also contribute to the sustainabilitv of frontier farming and to reducing turnover. Thus, contrary to general expectations, more infrastructure would contribute to less deforestation, not more. 29. See Oz6rio de Almeida (1992, chapter 21) for an analysis of the process of itinerant accumulation in the Brazilian frontier. 30. See Oz6rio de Almeida (1992, chapter 12) for an analysis of the two-staged frontier of Brazil in general and of the Amazon in particular. 31. See chapter 3 for a summary description of these newcomers, mostly mer- chants and other citv dwellers. 32. There is still much research to be done on who left, who stayed, and who came into the Amazon frontier small farming communities analvzed here. Much of the data collected in the 1981 and 1991 surveys have not yet been coded or analyzed. This project's research team at IPEA performed the painstaking work of manually correcting and retrieving data from questionnaires and of returning to the field to make additional corrections to the 1991 sample. 33. This section explores the evidence of the 1981 and 1991 samples only and will not attempt to give a general historical overview of the various colonization efforts in the Amazon. For that purpose, see Bunker (1985). Hemming (1985a and b), Lisansky (1990), Oz6rio de Almeida (1992), and Smith (1982). 34. All variables mentioned in this section are used in the regression analysis performed in appendix D. Acronyms in parentheses refer to the definition of variables in appendix B. to descriptive tables in appendix A, and to regression tables in appendix C. 35. Net worth is the 1991 U.S. dollar valuc of all assets: land (percent of the total), constructions, equipment, stocks, other durables, and financial holdings. Note that the variation among locations sampled is extreme, going from less than $10,000 among survivors in Anapu-Pacaja (Para) to more than $500,000 in Mutum (Mato Grosso). In fact, the accumulation of plots during the 1980s in Mutum was so great that owners can no longer be considered small farmers, although they began as such before they emigrated from Rio Grande do Su] in the late 1970s and early 1980s. This case illustrates the variety of Amazonian experiences and the peril of generali- zation based on an unrepresentative sample. 36. Since small farmers make consumption and production decisions simultane- ously. it is impossible to differentiate between these two microeconomic categories in practice. They are, therefore, lumped together into total household expenditure figures in appendix table A.17. 37. Negative net current income is not a contradiction in terms but a common condition whereby current expenditures (household and productive) are larger than current gross income, occasioning flows from the financial account whether in the form of borrowving or dissaving. 38. See chapter 5 and appendix C for a microeconomiiic model relating income, expenditures, investment, and other variables to turnover and deforestation. Land Markets and Sustainable Frontier Fal7ning 61 39. The difference in deforestation between solvent and insolvent farmers is important and is analyzed in chapter 5 and appendix C, especially with respect to the typology set up in appendix table C. 1. 40. Maintenance and depreciation costs are counted as part of current expendi- tures (E). See appendixes B and C (sections C-1.2, C-1.3) for a definition of this variable. 41. Possibly many of those with high non-agricultural incomes who deforested less were recently arrived merchants and other city dwellers who held land not for farming but for speculative purposes. See chapter 3. 42. As was seen in appendix table A.7, there is no evidence of loss of soil fertility during the 1980s, but there is evidence of low fertility in what are also high turnover locations. 43. Chapter 5 and appendixes C and D show the importance of farmer insolvency in explaining turnover and deforestation. 44. The regressions presented in appendix D show the importance of including definitive property title to land in explaining small farmer deforestation in the current year and since arrival. 45. Similar findings by FAO/INCRA, based on a much larger sample of agrarian reform projects all over Brazil, indicate that the cost/benefit ratio for small-farmer settlements may be much more favorable than supposed. See Oz6rio de Almeida (1992, page 618) for a rough approximation of the cost-benefit ratio (1.5) for directed colonization in the Amazon during the 1970s. 46. See, for example, Schneider (1993) on the imediatisno that affects small-farmer behavior in the Brazilian Amazon. 47. The net present value of agricultural income (NPVA) is the value today of the future income stream that can be generated by agricultural production, discounted according to the interest rate. The discounted salvage value of land (NPXL) is the value today of the land price per hectare at some future date, discounted according to the interest rate. Farmers take past experience into account in their expectations regarding the evolution of land prices for the 1990s by applying to the 1991 land prices the same yearly rate of increase that occurred during the 1980s. 48. The NPVA and NPVL were computed for a ten-year period-the 1980s-by discounting current agricultural income and current land prices at the relevant rate. In this case the opportunity cost of capital (the interest rate on savings certificates) was taken to be the appropriate discount rate. 49. This trend is evident in our sample. In the state of Mato Grosso, where most of the land sales are taking place, the NPVL is already greater than the NPVA. Speculation in that area is triggered by the presence of a large class of urban ranchers. See chapter 3 for details. 50. With respect to the cost for the global economy of stopping Amazon deforestation, see Pearce and Warford (1993) and Pearce and Atkinson (1993). CHAPTER 5 Productive and Speculative Deforestation by Frontier Small Farmers In any frontier, farmers' deforestation activities may be taken as a form of demand for land. Such demand may be for agricultural purposes, which would imply productive deforestation, or for speculative purposes, which would imply speculative deforestation. This chapter, and appendixes C and D, test whether deforestation in the Amazon's old frontiers is mostly productive or speculative. The main finding is thatfrmners who deforestfor the piurpose of agiiclturzal production are sensitive to price and income variations. Therefore, by managing agricultural prices, levying taxes on agricultural income, and imposing penalties on deforestation, governments could induce productive farmers to deforest less and make better use of already deforested land. Economic policv could thus be an important instrument for increasing the sustainabil- ity of small farmer settlements. On the other hand, farmers who deforest in order to hold land as a store of value respond inversely to variations in prices and incomes. Many of the economic policies intended for agriculturallv produc- tive farmers would probably only reinforce speculative farmers' motives for holding land. During the 1980s in old Amazon frontiers, speculative motives for defor- estation grew more powerful than productive ones. 1 Thus, standard income, price, and fiscal policies aimed at agricultural production lost their capacity to discourage deforestation. To reverse, or at least dampen, speculative deforestation, policies that punish land hoarding-such as a tax on capital gains-and promote good agricultural practices must be enforced. If prop- erly administered, these policies can increase the productivity and solvency of good farmers and punish those with speculative tendencies.2 Increasing the number of good farmers in the Amazon would strengthen farmers' sensitivity to economic policies that reduce turnover and (leforestation and raise the sustainability of Amazon small farming in general.3 This policy 63 64 Suistainable Settlement in the Brazilian Amazon implication is controversial. Good farming in an Amazon location, it is argued, may have a demonstration effect and attract migrants from else- where. The influx may pressure resources and swamp the sustainabilitv achieved. Chapter 6 argues against such a possibility. The evolution during the 1980s in deforestation motives can be seen by comparing total deforestation (since arrival on the current plot) to current deforestation (during the most recent year). Total deforestation appears to have been most directly related to characteristics of farmer background. The most significant indicator was the extent of a farmer's itinerancv, that is, how many places he had farmed before. Also important were farmers' place of origin, whether their parents had been farmers, how much they brought with them, and so on. Deforestation in the current year, however, was much more influenced by conditions of the present location, such as access to credit (the greatest cause of current deforestation), local institutions, mar- keting systems, and so on. Apparently, it takes time for migrants to overcome past experiences and become responsive to current local conditions. In order to analyze the economic determination of small farmers' defor- estation, appendix C specifies, and appendix D tests, an econometric model of how small farmers' deforestation responds to policies that alter income and land prices. Appendix C proposes that the demand for land may be static, when there are only current income and expenditures, or dynamic, when there is net borrowing and/or net investment. In the dynamic case, the effect of total income and land price on demand for land is complex. The effect of income depends, in its own right, on the relationship between farmers' portfolio management and agricultural productivity. Among solvent farmers, if productivity keeps up with rising land prices, then demand for deforestation should fall as the price of land rises, according to conventional demand for a factor of production (land).4 However, if productivity fails to keep up with land prices, then the demand for defores- tation mav become speculative, increasing with rising land price, according to a portfolio demand for an asset (land). Among insolvent farmers, inverse relationships hold (see appendix C, section C.3.2). Irproductivity keeps up with rising land prices, demand for deforestation should rise as the price of land rises, as farmers try to pay off debts by increasing agricultural produc- tion and demanding more land. If productivity fails to keep up with land prices, then demand for deforestation should fall as the price of land rises. However, this is not because ofconventional demand for a productive factor (land), but occurs as farmers give up on agriculture, sell out, and move away. Appendix D uses econometric methods to test for deforestation as a demand for land, based on the model specified in appendix C and adapted to the sample of Amazon frontier small farmers described in chapter 4. Appendix D introduces several new variables-origin, destination, and Productive and Speculative Deforestation by Fronitier Small Farmers 66 individual characteristics of farmers-and specifies a four-stage estimation procedure for hypothesis testing. Background A frontier is a place very different from the remainder of the established economy. Since migrants are, literallv, starting from scratch, growth of productive capacity, or accumulation, is their objective. To grow, however, farmers must invest, which mostly means carving agricultural land out of one of the most dense and inhospitable forests in the world. Thus, settlers deforest. The issues are how much or how little they must deforest and how to minimize this environmental cost. As a frontier develops, markets form and consolidate (see the more extensive discussion of this topic in chapter 3). As transport conditions improve, product, input, and capital markets become more competitive. As they become less informal, interlinked, and concentrated, usury-mercantile ties to farmers are broken and the debt peonage typical of frontier farming weakens.5 Rural and urban economies grow, and labor markets emerge. As open-access conditions end and private property becomes more widespread, the land market becomes more active. Newcomers-merchants, officials, commercial farmers-who operate in the market economv, bid up local land prices, offering substantial capital gains to pioneers. Farmers must then choose whether to remain where they are or to sell their land and decide what to do next. As pioneers sell, part of the benefit of deforestation accrues to newcomers, who take over the land in old frontiers. Settlement is a discontinuous process with respect to life outside the frontier. The decision to migrate is based on the opportunities perceived and the resources available at that time. The decision to stay depends on the opportunities perceived and the resources that can be generated at that location. Pioneer farmers typically farm on several different plots, moving with the shifting frontier, before arriving at their current location. Their relationship to the land, thus, resembles that toward variable capital rather than fixed capital. As long as they believe that there is still land available to rmove onto, many will choose to do so.6 Sustainable settlement implies continuous farming in one place since arrival on the frontier, and this sustainability depends on how agriculture stands up to alternatives in land, labor, and capital markets of the overall economy. If frontier agriculture covers its opportunity costs, farmers will want to continue where thev are and will tend to conserve the land they have deforested. If frontier agricul- ture does not cover its opportunity costs, farmers will transfer their labor and resources elsewhere. They may seek outside employment, they may sell their land and buy land in another part of the frontier, they may remain in the location and utilize the proceeds from their sale to invest in other 66 Stutainable Settlement in the Brazilian Amazon activities, or they may revert to agricultural involution, perhaps withdrawing from crops and leaving the land fallow or in low productivity pasture. The most immediate issue is not how well to conserve a farmer's propertv, but how long a farmer remains on the land. Ultimately, sustainable settle- ment will depend on migrants' performance as farmners. To the extent that settlers remain on their land, thev do not open up or deforest new frontiers. To the extent that such settlers deforest for productive purposes, such deforestation in their current location can be discouraged by policies that reduce their current demand for land. On the other hand, frontier settlers deforest not only for agricultural purposes but also for reasons of tenure insecurity, speculation, and so on. Deforestation fueled by other than productive purposes is less sensitive to policies that alter the opportunity cost of poor farming practices. The effectiveness of such policies, therefore, in curbing deforestation in frontiers that are already occupied, though not yet deforested, depends on the strength of farmers' productive motives compared to other motives for deforestation. Comparing these motives requires that they be modeled and tested against empirical evidence. The econometric model of small farmers' deforestation, presented in appendixes C and D, takes a large number of variables into account. It includes not only static, conventional prices and costs for current agricul- tural inputs and outputs, but also opportunity costs for farmers in the product, input, land, labor, and capital markets. Additionally, the model incorporates frontier-specific dynamic conditions, such as accumulation, variety of economic activities, start-up costs, and balance on financial account to show whether net investment or net indebtedness occurred. Finally, the model takes into account how migrants with different economic, cultural, social, and personal experiences, as well as different attitudes and expectations, responded differently to the diverse conditions of the frontier. Space and time limitations preclude a discussion of several other impor- tant issues, such as risk, information, contracts, technology, market inter- linking, life cycle, subjective equilibrium, production function specification, and peasant motivations.7 The role of property rights and tenure security in the Amazon is very important and is the subject of an innovative and growing body of literature. However, it is beyond the scope of this book.' Productive and Speculative Responses to Land Price Variation In established economies, land price tends to vary with productivity, which becomes capitalized into land rent and real estate value. T his is not so in the Amazonian frontier, however, where many other factors affect the forma- tion and evolution of land values.9 During the 1970s, the difference between land prices for lots with similar soils but in different frontier locations was in the order of 100 to 1; Productive and Specldative Deforestation by Frontier Smnall Farnmerv 67 differences were even greater between locations in the Amazon and those in the established parts of Brazil. During the 1980s, land prices rose in some places and fell in others, seemingly unrelated to the evolution of productivity, though perhaps related to which crops were dominant in that location and to the evolution of product prices."1 Real land prices are quite low in the Amazon, with hectares still selling at prices equivalent to thirtv bags of rice.'' The set of factors that determines land values can be considered exoge- nous to the decisionmaking of any individual farmer; that is, frontier farmers are price takers with respect to the land market.'2 An individual farmer compares the productive potential with the speculative potential of his lot in deciding whether it should be farmed, hoarded, or sold. His response to land price variations will be productive or speculative depending on the evolution of land productivity and his portfolio management.'3 Productive and speculative response The productivity of a farmer's land is an important factor in determining whether he will sell the land when its price rises. When capital gains become greater than the expected discounted income stream from the agricultural use of his land, the farmer will want to sell. The decision thus depends on how real land prices and agricultural productivity (an important influence on the rate of return on investments in land) evolve compared to each other. In unproductive frontie,-s agricultural productivity is low or declining relative to real land prices. Reduced deforestation and farming involution conditions, such as decreased cropped area and increased pasture and fallow areas, typify unproductive frontiers. Alta Floresta and Paranaita in Mato Grosso and Anapu and Pacaja in Para are examples of such frontiers. In productivefrontiers, agricultural productivity is high or rising relative to real land prices, meaning land is held for the future income to be derived from farming it, not for the capital gain from its future sale. Increased deforestation and agricultural intensification, such as increased cropped area, reduced pasture and fallow areas, and low turnover of farmers in plots, generally characterize productive frontiers. Pacal in Para and Mutum in Mato Grosso are examples of such frontiers. The other important factor determining whether a farmer's responses are productive or speculative is his portfolio management. This is the composition of his investments and disinvestments and his borrowings and repayments, which jointly determine whether his balance on financial account is positive or negative.'4 Among insolvent farmers, net borrowing means that future repayment will reduce their capacity to cover current expenditures in subsequent years. For the poorest family farmers, such borrowing frequently takes the form of crop-lien (when payment is in kind, out of the next harvest) or of 68 Sustainable Settlement in the Brazilian Amazan debt-peonage.15 Net disinvestment reduces productive capacitv in subsequent years, for example, when cattle is sold to pav for current expenditures. Among solvent farmers, net repayment reduces current income, but there are still sufficient resources to cover current expenses and reduce debt outstanding. Net investment also reduces current income but adds to productive capacity in subsequent years.6 In sum, the following are four cases for productive and speculative responses: A productive response occurs whenever a rise in income augments demand for land and an increase in land price reduces demand for land. A productive response can be expected in two instances: * high productivity lots/solvent farmers, whose high yields and high investment rates lead to a process of accumulation (case I in appendix table C.1) * low productivity lots/insolvent farmers, whose debts and low yields lock them into a debt-peonage condition (case 4 in appendix table C. 1). A speculative response occurs when an increase in income reduces demand for land and an increase in land price increases demand for land. This can happen in two instances: * high productivity lots/insolvent farmers, whose debts are discouraging them from persisting in agriculture (case 3 in appendix table C. l) * low productivity lots/solvent farmers, whose low yields are driving them out of agriculture (case 2 in appendix table C. I). Given frontier conditions where all land is originally forested, a farmer's productive demand for land is necessarily a demand for deforestation. Despite wide variations in deforestation rates in early years, and although managed forestry alternatives are important alternative sources of income, there is no way to cultivate the soil without removing the forest cover17 (see appendix table A.29). As time passes and a higher proportion of forest is cleared, demand for additional land is increasingly met bv purchases of already deforested lands and/or agricultural intensification on such land. Deforestation, then, should fit a model of demand for land better on newer rather than older frontiers. The empirical equations specified in the econometric model of demand for land were thus tested on total deforested area (see appendix D). Deforestation is the main form of investment performed on a frontier. Clearing land of its original forest cover considerably raises its value, even if nothing else is done to it.'8 Deforestation is dynamic in that it increases agricultural potential; therefore, it may be expected to fit the dynamic version of the econometric model better than the static version. 19 Deforestation was analyzed according to two different measurements. Deforestation since arrival (F) refers to total deforested area of a given plot; ProdutctiVe a7nd .SpeculatiV'e Deffil-estatioz by Frontier Small Far-meri 69 deforestation in current year (F9 1 ) refers to areas deforested during the year 1991. Total deforestation is relatively more sensitive to long-run, earlv determinants whereas current deforestation is relativelv more sensitive to short-term, recent determinants. Summary of Regression Results-Deforestation: Past and Current Trends Despite very large variations in Amazon settlement and the many determi- nants of frontier farmers' deforestation, a simple model of productive demand for land accounted for most of the observe(d variation in deforesta- tion. The largest, most significant, and most consistent positive influence on deforestation is agricultural income; as long as farmers are productive, a rise in such income stimulates deforestation (see appendix tables D. 1 and D.2). During the 1 970s, agriculture was the major activity of colonists, so the overall income effect for deforestation since arrival (F) was positive (see table D.1).20 During the 1980s, however, more and more frontier farmers began to diversify, moving out of agriculture into pasturing and investing in other activities.21 Wealthier, higher-income, nonfarming newcomers purchased colonization lots for portfolio management, to add to their net worth, as a hedge against inflation, and as speculative real estate investments in the face of rising frontier land prices. Therefore, in 1991, higher total income was associated with less agriculture.22 The less productive the land, the greater the movement away from agriculture. The overall income effect, both static and dynamic, for current deforestation (F91) was negative. This process of agricultural involution provokes a "Giffen" positive price effect. Falling land prices reflect diminishing demand for agricultural land, thus reducing deforestation; rising land prices increase speculative and portfolio demand for land. So the overall effect of the price of land on deforestation, past and present, is positive. Migrant characteristics, especially a history of itinerancy, had a significant positive effect on deforestation since arrival. Present location characteristics, however, had a more important impact on current deforestation: access to credit and local institutions stimulates agriculture and, therefore, defores- tation; poor marketing and poor storage conditions hinder agriculture and inhibit deforestation. This implies that improving local conditions for agriculture for those who have been in the location for a while fosters defores- tation. Recent migrants, however, tend to be less sensitive to such conditions and behave more according to what they have experienced in the past. In general, the regressions capture the dominant trend in the sample, that is, the takeover of lots by nonfarming newcomers. So the estimated income and price effects reflect mainly speculative and portfolio management behavior rather than productive behavior. To the extent that agriculture wvas 70 Surstainable Settlement in the Brazilianl Amazon practiced, however, it had a strong positive impact on deforestation (see appendix table C.1).23 Conclusion: Turnover and Deforestation Chapter 4 indicated that successful agriculture reduced the turnover of colonists on their present lots, thereby reducing their migration to new frontiers. However, although successful agriculture prevented farrners from deforesting elsewhere, it also resulted in more deforestation of the successful lots. Econometric results of this chapter and appendixes C and D reaffirm the findings of chapter 4 that successful farmers deforest more and poorer farmers deforest less. Additionallv, however, this chapter finds that frontier farmers were sensitive to income and price variations, especially those who were productive and solvent. This indicates that economic policies could be used to increase the sustainabilitv of Amazonian small farmers. Comparing results for deforestation since arrival (F) with deforestation in the current year (F91), one finds that many colonists diversified out of agriculture and that total family income on plots settled in the 1970s became increasingly nonagricultural. Thus, estimated income effects on current deforestation were negative, which reflects the increasing tendencv of farmers to transfer resources, both current and financial, out of agriculture. Rising land prices, for their part, did not reduce the demand for current deforestation; rather, they increased it, revealing speculative motives for deforestation. Estimated land price effects on current deforestation, there- fore, were positive. Thus, during the 1980s, speculative motivesfor deforestation increasingly predominated over prodactive ones. This implies that conventional income and pricin?g policies also increasingl/ lost their power to reduce deforestation byfi-ontier colonists. Raising the price of land, by fiscal or other means, would only stimulate more speculative deforesta- tion. Taxing agricultural incomes would result in increasing withdrawal from agriculture and holding of unproductive land as a hedge against inflation. Thus, this set of economic policies, taken alone, would be self-defeating. If colonists are to deforest less, they will have to be motivated to do so by a complementary set of policy alternatives aimed at promoting good farming in the Amazon. On the one hand, policies should be pursued that reward small-farmer agriculture, that is, those that increase farmers' agricultural productivitv and financial solvency. If this were to be achieved, farmers might become less disenchanted with, and more willing to invest in, agri- culture, rather than choosing to sell out and move or stay behind and involute. On the other hand, policies aimed at directly punishing defores- tation should simultaneously be pursued. As productive motives strengthen and turnover declines, appropriate fiscal, pricing, and other economic Productive an7d Speculartive Dejbrestation b.y Fronztier- Small Fairmers 71 policies should be used to monitor farmers and curb deforestation in their current plots. Notes 1. See chapter 3 on the changing economic context of the Amazon frontier during the 1980s. 2. See Campari (1994) for a model linking the liquidity condition of frontier farmers to the rates at which deforestation occurs. 3. See the description of this process of agricultural involution in chapter 4. 4. See chapter 4; in appendix A discussion of appendix tables A.2 1. A.25, A.37; and appendix C, section C. 1.2 and table C. I. 5. Debt peonage is a situation in which farmers overwork their families and land to pay off debts. Also see Ozorio de Almeida (1991 a and b; and 1992, chapter 12) for a discussion of the credit market interlinking with land, labor, and product markets and implications for debt peonage and deforestation on the Amazon frontier. 6. Moran (1989) finds that poverty and mobilitv are strongly related. Previous mobility is a strong predictor of future mobility, and the historv of crop yields bears a strong negative relationship to the owner's number of previous migrations. 7. For an analysis of risk see Binswanger (1980) and Fearnside (1986). Also, see Oz6rio de Almeida (1992, chapter 14) for a discussion of this literature and the model's omissions. 8. For a discussion on the role of property rights in the Brazilian Amazon, see Alston, Libecap, and Schneider (1994); Binswanger. Deininger, and Feder (1995); Lin and Nugent (1994); Besley (1993); Fearnside (1993); Feder and Feeny (1991); Aubertain (1990); Eggertsson (1990); Prosterman (1990); North (1990); Feeny and others (1990); Thomson, Feeny, and Oakerson (1992); Field (1989); Feder and others (1988); and Jessup and Peluso (1986). For an empirical and descriptive treatment ofthe subject, see Oz6rio de Almeida (1992), especially tables 60, 61, and 62 (chapter 20) and chapter 21, which offers a good description of farmers' relative rates of success in achieving definitive title to property according to their origins. See also tables D.3 to D.10 of appendix D (summarv of regression results) in this volume and Ianni (1979). 9. See Oz6rio de Almeida (1992) for the different tvpes of colonization agencies in the Amazon during the 1970s. During that time, land prices were determined, among other factors, by type of colonization (public, private, spontaneous); income level of potential colonists (southerners, others); period when lots became available (early to late 1970s, early to late 1980s, early 1990s); whether the project was crisis related (rural violence relief, drought relief); distance to market; road conditions; tenure security (titling agency, time required for gaining title, and so on) and risk of expulsion due to such factors as local violence or proximity to large-scale projects; and access to social infrastnicture (education, health services, extension, credit, amenities). 10. According to the field survey analyzed in chapter 4, the most spectacular fall in land prices occurred where coffee growers gave up on agriculture and turned most of their land to pasture, as coffee prices plummeted from an average of us$0.76 to 72 Sustainable Settlement in the Brazilian? Amazonz an average of us$0.23 during the 1980s (see Alta Floresta-Paranaita in appendix table A.8). The largest percentage increase in land prices occurred where produc- tivity was lowest, where original price was also lowest (see Anapu-Pacaja in appendix table A.7). and where most agricultural land was also turned to pasture. Thus, the evolution of land prices depends on several factors, such as whether land price was low or high to begin with, given the institutional process of price setting during the early days of directed settlement. I1. According to Ozorio de Almlleida (1992) and appendix table A.8, in 1981 a standard (60 kg) sack of rice cost, on average, us$9.60 while one hectare of land was valued, on average, at us$190; in 1991 a standard (60 kg)sackofrice cost, on average, us$7.20 while one hectare of land was valued, on average, at us$219. 12. Soil quality comes in patches that underlie farm plots and affect the land price of many farmers in a given location, some of which fall into a given sample. A notorious problem in every analysis of cross-sectional data is the tendency for many variables to becomiie collinear. Stratified sampling-as done in the present data set-helps reduce the problem. If the decisionmaker is the individual, as in this studv, and not a local average, then there is a better case for the exogenous land price, an important issue in specifying and testing the econometric model presented in appendices C and D. 13. The section entitled "Farmers' Economic Decisiorunaking in the 1990s" in chapter 4 discusses the empirical evidence provided in appendix table A.42 on farmers' decisions to farm, to hoard, or to sell land, according to net present value of agricultLral income and land prices in different frontier locations. Due to space limitations, it is not possible here to explain the conditions that generate the four cases cited in chapter 4. They are carefilly specified in appendix C and table C. I. 14. See appendix C, section C.1.2, for a more detailed discussion of frontier farmers' portfolio managemenieiit. 15. See Oz6rio de Almeida (1991b) for literature on, and modeling of, debt peonage in the Amazon frontier. 16. Chapter 4 showed that frontier locations with the highest rates of accumula- tion (high k) were those where agriculture was most productive and where agricul- tural income had the highest share of total income. 17. Most migrant farmers in the Amazon today brought with them agricultural traditions that rely principally on mechanical and chemical modification of the soil, not on extractivism of pre-existing forest. Over time, as farmers become acclimated to the specific ecologies of the Amazon, they may learn to derive increasing proportions of their incomes from sustainable forestry management, which should reduce deforestation. Conventional farming practices, however, dominated in the 1970s and 1980s and will soLrely continue to do so in the 1990s. 18. Field observation, based on the sample described in appendix A and box 4.1, indicates that clearing forest typically doubles price per hectare. 19. See appendix C, section C.3 and results for regressions 3 and 4 in appendix tables D.l and D.2. Also see section C.2 and results for regressions I and 2 in appendix tables D. l and D.2. 20. Ozorio de Almneida (1992, chapter 16) shows that in 1981 nonagricultural income averaged 10 percent of total income of farmers in the same lots as sampled here. Pr oductive and Speculative Defbre.tation by Fron7tier Small Farmwery - 3 21. Chapter 4 and appendix A show that, on average, nonagricultuiral income was almost three times greater than agricultural income in 1991. 22. For instance, in 1981. gross agricultural incoine accounted for 40 percent of gross total income (Oz6rio de Almeida 1992. table 39c). But in 1991, gross agricultural income accounted for 14 percent of gross total income, with extreme variations across locations, as can be seen by comparing appendix tables A.16 andl A.20. 23. See signs and significance levels of coefficients for agricultiral income in regressions in appendix tables D.3 to D.10. CHAPTER 6 Policy Implications: Institutional Improvement for Sustainable Settlement in the Amazon The population and economic shifts that occurred in the Amazon during the 1980s brought about a speculative boom in local land markets that drove small farmers out of the frontiers they had deforested in the 1 970s and into the frontiers they are deforesting in the 1990s. Amazon deforestation will onlv be curbed if sound economic policies reward farmers for staving where they are and reverse incentives to move on. This chapter proposes that international organizations, NGOs, and the World Bank could assist federal and local Amazonian governments in creating local long-term economic conditions that would make good farming in the frontier compatible with environmental conservation. Summary of Results Small farmers in the Brazilian Amazon have typically farmed along many migratory steps. Manv were squatters and tenants who moved from farm to farm, living off lands that were never theirs. Some were landowners who bought, deforested, sold, and profited from successive plots. Relatively high returns to household labor and capital, and relatively low returns to land, have kept small farmers moving, shifting the frontier forward. In recent years, frontier land prices rose beyond the growth of agricultural productivity mainly because of an inflationary economy in general and a prospering urban frontier economy in particular. Physical and social infra- structure increased, while local public and private sectors grew rapidly. Rising land prices set off different reactions among farmers. Many original colonists reaped capital gains and moved from old to new frontiers within the Amazon, which they proceeded to deforest and quit once again. Others 7S 76 Sustainnble Settlement in the Brazilian Amazon held on to their lands but diversified out of agriculture. The more urban groups held on to land mainly for speculative motives. Finally, some highly success- ful farmers neither moved out of their lands nor abandoned agriculture. On the contrary, they expanded their holdings and increased their agricultural production. They also deforested their land at an accelerated pace.' Thus, small farmers on the Amazon frontier tend to deforest: more successful farmers continue to deforest where they are; less successful farmers deforest wherever they relocate. Meanwhile, land, originally cleared of forest for farming, is being added to the net worth of a nonfarming and mostly urban local middle class. lMlarket, institutional, and individual contributions to deforestation Frontier institutions and market structures have seemed to stimulate defor- estation and discourage sustainable farming. Tenure-related institutions have not proven capable of handling titling needs. Marketing and transport facilities have been poor, markets concentrated, and agricultural prices depressed. Availability of credit, although declining, still has made forest clearing less costly, because the value of land rises with such clearing, meaning that collateral subsequently increases. Taxation on deforestation, or on the capital gains derived from land speculation, is practically nonex- istent. Therefore, the economic conditions necessary for sustainable settle- ment are not there. Yet settlers have proven to be sensitive to economic conditions; their response to income and prices accounts for most of the variation in deforestation in directed colonization projects. Total deforestation (since arrival on plot) depends significantly on where farmers came from, what they originally brought with them, whether their parents were also farmers, whether they were landowners before, and how many migratory stops they have made. Current deforestation (during 199 1) depends significantly on the characteristics of farmers' present location. Important factors include the type of project they are in, public or private; how distant they are to markets; and what their marketing and storage conditions are. Individual variations, such as age, family size, time on lot, attitudes, and plans and expectations, seem to matter much less, in terms of total or current deforested area, than origin and destination characteristics. Thus, policies that reduce deforestation on the frontier would do so differ- ently among different migrant groups, according to their prior experience and current local conditions. Productivityfactors and deforestation In the Amazon during the 1980s, low productivity frontiers witnessed high turnover, though deforestation was relatively low. Only in high productivity Poliiy Implications 77 frontiers was turnover relatively low, but deforestation was rapid. There are different underlying motives for deforestation that change in different circumstances. In productive frontiers, farmers (both solvent and insolvent) cover their opportunity costs in land, labor, capital, and other markets. Increasing incomes in productive frontiers, then, should increase deforesta- tion; increasing land price should decrease deforestation. In speculative frontiers, farmers (both solvent and insolvent) do not cover such opportunity costs, with inverse policy implications. Increasing income in speculative frontiers would decrease deforestation; increasing land price would increase deforestation. More generally, there does not seem to be a trade-off between the achievement of national macroeconomic stabilization and accelerated Ama- zonian deforestation. While a growing economy does help to finance settlement of the frontier, a stable one reduces land speculation and slows down the migration of small farmers from old frontiers to new. Moreover, the higher probability of finding a job in a booming economy increases the opportunity cost of frontier agriculture and offers inducements to small farmers to move to alternative activities. Thus, recovery from economic crisis will not necessarily bring about more deforestation. A Policy Prescription The appropriate policy package for increasing the sustainability of small farmers in each Amazon location will vary according to local circumstances. Whatever combination of policy instruments the locality chooses to use, the simultaneous goals must be to promote good farming and penalize specula- tion and deforestation. This is difficult, as it implies an apparent contradic- tion between means and objective: taxing agricultural income to stimulate sustainable farming. If this can be achieved, however, it will contain intra- Amazonian migrations and deforestation both in new and old frontiers.2 The following are some of the alternatives available. First, there is an important role for zoning and for promoting settlement only where the land is sufficiently productive to sustain farming income. However, as zoning cannot by itself control the intraregional forces that promote deforestation, since it is the more productive farmers who deforest most, it should be accompanied by a redesign of regional fiscal policy by federal and state governments. Second, fiscal policy should tax net worth, capital gains from land sales, and deforestation itself. A tax on net worth would increase the cost of deforestation undertaken merely to legitimize property. A tax on capital gains from land sales would increase the cost of speculating in the land market and improve the farming alternative. A tax on deforestation itself, 78 Sustlaiinable Settlement in the Bra,zilian Ama7zon such as a stumpage tax, for instance, would increase the effective price of land, without adding fuel to overheated land markets. TIhird, there is a role for the generation and diffusion of appropriate technology. An ample stock of research findings exists in Amazon research institutions, such as at the Goeldi Institute and INTPA, as well as in other tropical agriculture research centers in Brazil and around the world, that shows the competitiveness in the Amazon of innovative agroforestry, using combinations of conventional and sophisticated biotechnological tech- niques. Rural extension must incorporate such results, some of which are widely practiced elsewhere, are experimental, or are still being researched. Application of existing and developing technologies could greatlv increase the productivity of Amazon soils. This policv approach does not exclude the use of caboclo (indigenous technologies) as feasible and valid wavs to achieve a sustainable frontier.3 Although such technologies tend to be overly roman- ticized, their use, singly or in combination with biotechnological and other high-tech alternatives, expands the set of tools available to deal with the large varietv of local conditions. Fourth, there is a role for policies that improve market systems, storage, transport, and roads, thus helping to break typical frontier monopsonies and monopolies and improve farmers' bargaining power in principal markets. Marketing conditions must be improved to help small farmers-rather than merchants and large landowners-to appropriate the benefits from produc- tivity increases. Fifth, the operation of credit and other institutions must be revised to eliminate strong incentives to deforest. Sixth, there must be conditionality clauses on federal government transfers. Seventh, all of the above should be developed as locally enforceable instruments, some at the state and some at municipal levels. Democratiza- tion combined with fiscal decentralization has distributed proportionately more revenues than responsibilities. The federal government is simply incapable of dealing with the complex set of policies necessary to ensure environmental conservation throughout the countrv. Local authorities, on the other hand, are now politically and economicallv capable of taking on this task. This is especially true in the Amazon, where there is the greatest varietv of local circumstances and where centralized environmental policv has not been successful. Local-level governance, however, must confront local elites (that is, the same urban middle class that speculates with land) who have become not only an economic, but also a political, force in the region. This political force is likelv to offer resistance to the policies recommended here, unless the interest of small farmers continues to receive support from NGOs, the scientific communitv, and from multilateral organi- zations such as the World Bank.4 Poli-y Implications 79 Last, in areas where the above set of policies proves insufficient for retaining small farmers, directed settlement projects could be instituted. Such settlements would be different from those of the past; that is, instead of bringing migrants in, their aim would be to prevent migrants from movi ng out and deforesting another plot. Unfortunatelv, proposals for such projects have met strong resistance because of the failure of large-scale colonization projects directed by the federal government in the 1970s. This is unfortu- nate, since failures of the past offer lessons for the future, the main one for today being to avoid merely setting up the projects with mapping, roads, and tiding, without creating the policies that would make them sustainable. The main contribution of a 1 990s diluted directed settlement would be the provision of tenure security, or property title, an important influence in curbing both turnover and deforestation (see appendix D).5 These eight policies aim to reduce deforestation by promoting good farming frontiers that reward migrants for staving in the lots they have already cleared, weaken their motives for speculative and unproductive deforestation of their current location, and reduce economic incentives to move on to other frontiers deeper into the Amazon. Certain conditions must be met for such policies to be successful. As discussed earlier, one such condition is that these policies generate sufficient local revenues to pay for their local enforcement as well as for the expenses of direct conservation measures. The other is that successful locations not be overrun bv interregional migrants in a perverse demonstration effect. The evidence in chapters 2 and 3 indicates that this is not likely. If and when the Brazilian economic crisis ever ends, the financial and demographic adjustments that it caused will probably have become irreversible. Thus, it is highly improbable that sustainable frontier farming would trigger another Amazon-bound migra- tion similar to that of the 1970s. The establishment of regional equilibrium, therefore, would not be invalidated bv a national equilibrium adjustment. In addition, there are obstacles to the above policies that must be over- come. Establishing zoning, rural extension, and improvements in marketing and credit svstems will be a major undertaking. Long-term political resis- tance to taxes on capital gains will not disappear merely because such taxation will support conservation. Stumpage taxes have been difficult enough to levy on large-scale logging, let alone on masses of small, medium, and large farmers. The failure of earlier directed settlement projects has generated widespread resistance to their revival in any form. Finally, in the absence of national macroeconomic stabilitv, it is questionable whether local policies could have an impact on Amazonian deforestation. Although all of these issues are relevant, it is not possible to address them in this chapter. The objective here is merely to make explicit what policies 80 Suistainable Settlement in the Brazilian Amazon are needed to curb Amazonian deforestation, not to provide a blueprint for their implementation. Economic-environmental policies for c7urbing deforestation: a role for international, nationial, state, and local authorities Incentives for Amazonian deforestation in the 1 990s are different from those of the 1970s and 1980s. Rather than being federally instituted, they are locally generated. Reversing them has become much more complicated than merely repealing central tax and credit legislation. Environmental- economic policy is now the key to achiering sustainabilitv in Amazonian small farming. VV/hat is needed is the political will and the competence to design and enforce the policy instruments necessary to encompass a basin noted for its divergent development. There is a natural role for the federal government in encouraging, monitoring, and coordinating Amazonian economic policy for reducing deforestation. State and local authorities are subject to local opinion and political pressure favorable to raising revenues from economic activity that includes deforestation. The federal government, on the other hand, is subject to international and national public opinion and political pressure in favor of reducing deforestation. The problem is that federal environ- mental policy relies heaVily on command-and-control instruments, which, due to the ongoing economic crisis, tend to be underfinanced and unevenly enforced.6 Another problem is that economic policies with indirect environ- mental impacts are not ordinarily seen to be the responsibilitv of environ- mental authorities, but of economic authorities.7 The levying of a stumpage tax to increase the cost of deforestation, for instance, is not a decision that can be taken bv an environmental ministry or secretariat, but rather a financial authority. The environmental-fiscal policv field is still in its infancy in Brazil and its practice incipient.8 There is a need to develop an innovative svstem of institutional coopera- tion for applying economic-environmental policies. Although international institutions, national and international NGOs, and private enterprises could play an important role toward this end, these agencies have tended to focus only on protecting specific locations. Although such direct conservation measures are necessary, they are not sufficient. The broad forces that are impelling small farmers out of old and toward new frontiers in the Amazon will eventually put pressure op these forest reserves, no matter how well implemented. The conservation of specific remote areas will be short lived, therefore, unless the large Ama- zonian rural population attains sustainability in old frontiers. The following two sections outline some of the forest conservation programs supported by the World Bank and NGOs.9 Policy Implications 81 Rain forest pilot program The Pilot Program to Conserve the Brazilian Rain Forest was created to support an integrated set of projects targeted to reduce deforestation in a manner that is consistent with the sustainable development of the area's natural and human resources."0 The pilot program supports innovative grassroots initiatives aimed at harmonizing environmental objectives with improving the standards of living of the more than 10 million people who inhabit the Brazilian Amazon.'1 The program was launched at the request of the Group of Seven (G-7) countries who, along with the Commission of European Communities and the Netherlands, give it financial support. Total financial and technical assistance pledged to the pilot program is US$280 million, of which US$59 million is committed to the Rain Forest Trust Fund, under the World Bank trusteeship.12 The pilot program is coordinated by the World Bank, in accordance with the agreements reached by its participants, that is, the donors and Brazil. The participants of the pilot program attempt to coordinate their activities with the activities of other organizations committed to combating the underlying causes of deforestation in the region."3 Through direct governmen- tal actions and through measures that encourage other agents-individuals, enterprises, and private organizations-to conserve, the pilot program attempts to promote patterns of behavior that favor sustainable development. Although the pilot program recognizes that unsustainable small farmer settlement is one of the major causes of deforestation in the region, it has no specific instruments for dealing with this issue. The subproject on national forests, for example, states that the "resolution of land tenure issues" is important but does not detail ways in which the World Bank and other pilot program participants, including the Brazilian government, would address the problem or at least have the incentive to do so.14 Not unlike the national forests project, almost all other projects of the pilot program allege concern for the shifting frontier. However, actual implementation of tenure-related policies is considered to be the job of the Brazilian Environmental Institute (IBAMA) and the National Institute for Colonization and Agrarian Reform (INCRA). Both of these organizations have overlapping, uncoordinated, and spotty records in the treatment of land tenure.15 Since the 1980s economic crisis, furthermore, INCRA's insti- tutional capacity to do justice to the problem has suffered severely.'6 The areas selected by the Brazilian government for the implementation of several pilot projects avoid the settlement issues, as these places are not near land under dispute or areas where the frontier is advancing.17 Conse- quently, the projects are likely to be successful in the short term. The issue is how long it will take until the frontier arrives and how these projects will fend off encroachment when it does. 82 Sustainable Settlement in the Brazilian Amazt1on In a(ldition, the pilot program has also created the Indigenous Reserves Project in an effort to protect the people who depend on the forest for survival."8 Although such areas are important to the sunival of indigenous Amazon populations, their success also depends on whether their lands can be protected from encroachment. The intrusion of small farmers, gold miners, ranchers, and others on Indian and extractive reserves has been a recurring problem, contributing to the overall violence in the region. 9 Thus, demarcating and protecting specific areas-be thev indigenous or ecological reserves or areas for ecotourism-are necessarv but not sufficient to guarantee long-term protection from deforestation The WVorld Bank is increasingly recognizing the shortcomings of tradi- tional top-down involvement in development projects and has begun to seek ways in which to involve the community in the projects it finances.20 Toward this end, the pilot program has set up an international advisory group, composed of a broad range of agents and experts. Strangely, however, economists are absent from this group, and their absence may hamper the pilot program's ability to incorporate economic-environmental policies into its agenda.2' Local economic programs for environmental conservation, if appropriately designed and enforced with pilot program support, would be a powerful complement. Although the pilot program is still in its infancv, a review of its design is critical in order to ensure its long-term success. Other natural resource management projects witb international funding: Alato Grosso anti Rond6nia For the past several years the World Bank's countrv assistance strategy for Brazil has increasingly emphasized environmental issues.22 Despite the serious problems of the earlier Polonoroeste projects, the Bank is much better placed today to assist in a more rational development of the North- west region because of the gradual strengthening of local institutions, improved technical knowledge, and a growing commitment within Brazil to environmentai concerns.23 The Rond6nia and Mato Grosso projects are based on a broad under- standing of frontier dynamics and regional issues. Their objectives-im- proved frontier farming and environmental protection-are not mutuallv exclusive; on the contrary, issues such as biodiversity conservation, inte- grated farming systems, and agroforestry are all given equal weight. Mindful of the risk of relying solely upon short-tenn conservation measures, such as the establishment of conservation units and reserves, these projects target the reform of economic and financial incentives or disincentives to migration. Their activities are designed to intensify land use in suitable areas and develop sustainable extractive activities elsewhere. Unfortunately, however, despite good design andi understanding of Ama- zonian dynamics, which were largely missing from the Polonoroeste opera- Poliry Inplications 83 tions, the Rond6nia and Mato Grosso projects are not accomplishing their goals.24 Role of NGOs Sustainable settlement requires agents who are sufficiently flexible to adapt to a highly heterogeneous ecology and society. Large institutions, such as central government agencies and international organizations, are not well equipped to deal with varied local conditions-conditions leaving a role for local governments, private enterprise, and NGOs. NGOs are burgeoning in Brazil as elsewhere in the world (Goodland 1992). The World Bank is increasingly consulting with them in different stages of project cycles, not without some controversv. Hundreds of NGOs located in Brazil are involved in conservation, education, and social pro- grams in the Amazon.25 There are at least eighty-three environmental NGOs operating in Brazil with which the Bank maintains contact. Besides raising vast sums of money for conservation, these NGOs have also had an impact on the alleviation of poverty.26 In the Amazon, different settlement and tenure schemes are being tested by NGOs in collaboration with local authori- ties and international organizations, and many of these schemes are quite successful. Although NGOs can mobilize large sums of money, their main contribu- tion is organizational, not financial (Cernea 1988). These organizations are good at mobilizing people into organized structures of voluntary groups whose goals are the achievement of self-reliance and self-development. NGOs could assist small farmers in implementing the best available tech- nologies for the sustainable use of natural resources. NGOs in the Amazon are already forming collaborative groups, such as the Amazon Working Group (GTA) established in 1991, to assist programs targeting conservation of the forest, including projects growing out of the pilot program.27 Summary and Conclusion Given the demographic and economic shifts that occurred in the Armazon during the 1980s, increasing local speculative demand for land has become the main cause of Amazonian deforestation today, driving small farmers out of old frontiers and into new ones. Local policies that first promote productive agriculture and then tax agricultural income, punish speculation, and penalize deforestation directly could contain such intra-Amazonian migrations. Substantial institutional development is needed. Environmental institu- tions must learn to use economic policy instruments to achieve environ- mental conservation. Economic institutions must learn to wield fiscal and 84 Sustainable Settlement in the BTaizilian Amazon pricing instruments to Fulfill environmental objectives. Local governments must take on executive responsibilities previously reserved for federal gov- ernments. International organizations must broaden their objectives beyond estal)lishing protected areas, and they must contribute to improving the design and enforcement capacity, of local authorities and help to reverse local preferences for deforesting. Xkhat is needed is imaginative conservationist economic policies that generate alternative revenues at the local level. Amazon deforestation will only be curbed when local economic policies make good farming compatible with environmental conservation. It is increasingly urgent to do so in order to give poor frontier farmers an alternative to invading today's forest reserves. Notes 1. This is true for small farmers only. Although the end of tax havens and subsidized rural credit during the late 1980s considerably reduced the overall rate of Amazoni deforestation. that part of deforestation attributable to small farmers increased. Note the case of Mutum and, to a lesser extent. Pacal in the sets of tables presented in appendix A. 2. This approach is similar to thatofother authors who also proposed that Amazon deforestation could be reduced by repealing economic policies embedded in general or regional fiscal and creditincentives that rewarded deforestation. See Mahar (1989) and Binswanger (1994). 3. See Davis (1991) for a discussion on different approaches to production and resource sustainability practiced by indigenous people. 4. Accumulated World Bank experience on the institutional aspects of forest managemenit and settlement in other countries, particularly Indonesia and Thailand, would be of invaluable importance in designing policies for the Brazilian Amazon. See World Bank (1994a and b); U'orld Bank (1993a); and D'Silva and Appenah (1993). See also Feeny (1988), Ascher and f lealy (1990), and Jessup and Peluso (1986). 5. See Ozorio de Almeida (1992) and Oz6rio de Almeida and others (1992) for an evaluation of the performance of directed colonization in the Amazon during the 1970s and the 1980s. 6. See UtFR/IJUSP (1992) and ongoing programs for zoning, remote sensing, vigilance (Projeto stV.MM), and health, among other things. Various state-level environmental programs also are being set up. Baumol and Oates (1988) also present a standard theoretical treatment on the relative inefficient economic outcomes brought about by such policy approaches. 7. See, for instance, the many alternative/complementary suggestions contained in Schneider (1992, chapter 3, section C, especially pages 92-97). 8. Environmental pressure for the repeal of fiscal and tax incentives for agriculture is a good example of economic environmental policy in action in Brazil. See also Margulis (1992). 9. The folloviing section discusses only two World Bank projects, which were selected on the basis of their relevance to this study in the Amazon region. For a broader discussion of other efforts in the region, such as those that involve the Global Poliv Implications 85 Environment Facility (GEF) and other groups concernedl specifically wvith hiodiver- sity conservation, refer to WN`orld Bank (1 993c) and World Bank (1994). 10. See World Bank (19931.). 11. This excludes the cerrado frontier. 12. See World Bank (1993h)). 13. Technical contacts on environmental initiatives are frequent between the World Bank and other multilateral agencies. including the Inter-American Devel- opment Bank, UNDP, UNEP, and FAO. The Wkorld Bank has framework cofinancing agreements with many of the rain forest pilot program participants, and the proce- dures for cofinancing are flexible and well understood. 14. An analysis of the whole pilot program is not possible here, as its unique and highly complex structure would require a lengthy explanation of the high degree of coordination involved in its various projects and subprograms. 15. See Oz6rio de Almeida (1992), especially chapter 8. 16. See Ozorio de Almeida and others (1992) especially chapter 5. 17. 'I'hree projects, namely "Environmental Nlonitoring and Surveillance." "En- vironmental Enforcement and Control," and "Ecological and Economic Zoning," will be tested in three areas, one of which encomiipasses the portion of the lower Tapajos basin contained withini the state of Pari. This is an area in the process of conversion from forest to ranching and agriculture, with widespread goldrnining. Activities carried out in these areas will test, subject to a variety of conditions and levels of threat of deforestation, the integrated methodolopy that will later be expanded to other parts of the Amazoniian region. 18. See World Bank (1993b). 19. The killing of Chico Mendes in 1988 gained world infamy as an illustration of the conflict between forest reserves and ranching. More recently, more than forty Yanomami Indians, includingchildren, were massacred, possibly by gold prospectors encroaching on the tribe's traditional territory. This violence occurred only eighteen months after the former president of Brazil bowed to international pressure and staked out a reservation the size of the state of Indiania. In 1992, approximately 600 gold prospectors were expelled from that reservation by federal police and their airstrips destroyed. Prospectors. however, long locked in xiolent conflict with Indians, soon returned. See Blount (1993). 20. See Binswanger, Deininger, and Feder (1993) and Serageldin (1993). 21. In March 1993, the World Bank announced the initial membership of the International Advisory Group ([AG) of outside experts "selected for their scientific and technical knowledge" in areas addressed by the pilot program. The LAG is supposed to provide technical opinions on and analysis of individual projects, as well as technical reviews of projects before appraisal and annual reviews. The LAG is composed of twelve members, of whom four are geographers, two anthropologists. two ecologists, one sociologist, one forester, one biologist, and one businessman. 22. LTnless otherwise noted, data for this entire section wvere extracted from internal World Bank staff appraisal reports. 23. POLONOROESTE was a program launched in 19( and supported by five World Bank loans. It was aimed at resolving problems that had been ongoing since the 1970s. Still under debate is whether the environmental costs of the POLONOROESTE projects outweighed the benefits. 86 .Sustainable Settlement in the Btew7zilia,Amazon 24. Even wvith World Bank involvement (personal interviews with Bank staffin 1993). 25. See World Bank (1993b). 26. Goodland and Daly (1993) cite examples in which the annual budget ofeertain NGOs exceeded the budget of UNEP because of tax-deductible contributions. Fur- thermore. they state that "it is ironic that NGOs are free of the problems besetting governmental organizations. The reality is that the latter are not directly account- able, and that they pursue short-term self-interest (for example, tied-aid) to a certain extent" (page 19). 2 7. The GTA has been activelv involved in the preparation of the pilot program's demonstration project, which is a small-grants project designed to fund local community efforts in conservation and sustainable development. APPENDIX A Descriptive Tables This appendix presents information from surveys taken in 1981 and 1991 in the Amazon frontier locations described in box 4.1. The variables for the tables presented here are described in appendix B. The foot of each table contains the reference to the particular variable in appendix B. The 1981 data have been previously analyzed and published (see Ozorio de Almeida 1994). The 1991 sample was based on the same plots of land sampled in 1981 regardless of whether the same farmers were still on the land. Analysis of the 1991 survey is still under way, as is construction of a panel from the 1981 and 1991 data. Although both data sets contain much the same information, the 1991 survey was enhanced and incorporates variables (mainly institutional ones) that had not been present at the time the first (1981) survey was conducted.' All tables separate migrants into "southern" migrants-those from the South, a more developed part of the country (Rio Grande do Sul, Santa Catarina, Parana, and Sao Paulo)-and 'other" migrants, from all other parts of the country (mainly northeasterners). For each group, the tables contain 1981 and 1991 conditions. The 1991 tables further distinguish those who were interviewed in both surveys ("survivors") from the 1991 total. The tables also distinguish between public (official) colonization projects- located in the state of Para and run by INCRA-and private colonization projects, which are located in the state of Mato Grosso and run by different private firms and cooperatives. Within these project categories, the tables indicate the means observed in specific projects or locations. Statistical tests consider the variances among these means to determine if there are significant differences (at the 10 percent level) invariable means among subgroups of colonists. Differences between public and private projects reflect differences in the types of policy settlement adopted. Dif- ferences among locations refer to differences in Amazonian habitat and social environments. Differences between southern and other migrant groups refer to differences in origin. Deviations from the mean within groups indicate individual variations. Differences between "survivors" and other inhabitants reveal migration characteristics. 87 88 Susrtainable Settlement in the Brazilian Amazon The structure of the statistical tests is shown at the foot of each table. A difference is significant if it passes the t-test at the 10 percent level. Other- wise, the text in chapter 4 refers to seeming or apparent differences. XVhen statistical tests were not performed, this is indicated at the foot of the table. XTVhen differences are statistically insignificant, there is nothing at the foot of the table. Tables A. I Survivors [SURV] A.2 Imputed Wage [LMPW] A.3 Wage Rate ['(V A.4 Monthlv Accumulation [k or ACCUM] A.5 Agricultural Productivity [QH] A.6 Deforested Land for Agriculture [AGES] A.7 Annual Yields (CROPG] A.8 Prices for Crops and Land [CROPP and LP] A.9 Rice ProductiVity [RICE] A.10 Time on Plot [T(LiE] A. 11 Age [AGE] A.12 Net Worth [K] A.13 Qualitv of Life [LN1PR] A.14 Plan to Invest [PLAN] A.15 Plan to Stay [FIX] A. 16 Agricultural Gross Income [A] A.17 Total Expenditures [E] A.18 Net Nonagricultural Income [N] A.19 Net Subsistence Income [S] A.20 Gross Income [GY] A.21 Net Income [Y] A.22 Former Landowners [FORNM] A.23 Itinerancy [I-nN] A.24 Farming Background [PARNIT] A.25 Net Investment [I] A.26 Deforestation Since Arrival [F] A.2 7 Total Land [11] A.28 Deforestation in 1991 [F91] A.29 Deforestation in First Three Years [F3] A.30 Forest Preservation [OPINION and A/VANTT] A.31 Soil Fertility [FERT] A.32 Environmental Conservation [AGCON] A.33 Sales after First Harvest [STOR] A.34 Distance to Nearest Market [DISTI] A.3 5 Sales at Farm Gate [GATE] Appendix A 89 A.36 Agricultural Credit [CREDIT] A.37 Net Debt [D] A.38 Title [TD] A.39 Family Workers [WKRS] A.40 Agricultural Land [CROPLj A.41 Agricultural Price Index [PINDEX] A.42 Net Present Values of Agricultural Income and Land [NPV A and NTPVL A.43 Distribution of Initial Capital [K(] Note 1. N'hen the 1981 survey was done. there was no expectation that a follow-up would occur. Data storage was inadequate and retrieval is slow, By the time of the 1991 survey, the 1981 research team had disbanded. The 1991 team had different qualifications, understanding of research objectives and methods, and knowledge of what had been done before. In 1993 this new team lost its financing from IPEA, where all previous research was undertaken. Changed priorities in research demand intro- ducing new topics, removing old ones, and reducing compatibility between the two samples even further. The information presented here is an indication of possibilities for further analysis. No discussion of the specific methodological problems of dealing with panel data is made here. 90 Suistainable Settlement in the Brazilian2 Amazon Table A.1 Distribution of Survivors, 1991 [SURV]* Southen7ters Oihery Totwl Survivorr Tota7l Sumvivors Total Survivors Total No. . No. % Nvo. % No. % No1\O. % IVO. % Para: Official colonization Pacal 39 83 47 100 33 66 50 100 72 74 97 100 Anapu-Pacaja 8 67 12 100 24 44 55 100 32 48 67 100 Monte Alegre 6 100 6 100 29 78 37 100 35 81 43 100 Total 53 82 65 100 86 61 142 100 139 67 207 100 Mato Grosso: Private colonization Alta Floresta 48 63 76 100 48 63 76 100 Paranaita 20 56 36 100 20 56 36 100 Mutum 17 68 25 100 17 68 25 100 S5oJos6 do Rio Claro 10 53 19 100 10 53 19 100 Total 95 61 156 100 95 61 156 100 Total 148 67 221 100 86 61 142 100 234 64 363 100 Note: Percentages are averages. 'See appendix B.2.2(1). Appendix A 91 Table A.2 Distribution of Imputed Wage, 1981 and 1991 [IMPWl** (in minimum wages) Southerners Others All 1981 1991 1981 1991 1981 1991 Para: Official colonization Pacal 4.01b 3.22 3.54b Anapu-Pacaja 1.82 b 2.74 2.49b Monte Alegre 1 98b 1.09 1 19b Average 1.1 3.22 0.9 2.55 1.0 2.76 Mato Grosso: Private colonization Alta Floresta 1.69b 1.69b Paranafta 1.28b 1.28b Mutum 19.54a l9.542 SaoJose do Rio Claro 1.12b 1 b Average 1.1 3.83 3.83 Total 1.1 3.66 .9 2.35 1.1 3.23 Note: 1981 differences were not tested. Lowzercase letters: The presence of a common letter superscript indicates that there is nlo difference at the 10 percent level of significance in the following cases: Vertically (within columnm-lowercase letters; a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). *See appendix B.2.3(1). 92 Sn.rtainable Settlement in tie Brazilian Amlazon Table A.3 Average Annual Wage for Family Worker, 1981 and 1991 [W)* (thotisands of Ut.S. dollars) Sontlhel 7cI-r Otberes All 1981 1991 1981 1991 1981 1991 Para: Official colonization Pacal 31). 9.5 7.oab Anapu-Pacaja 1.61) 2.4 2.21) Monte Alegre 1.7' 1.0 1.01) Average 1.8 2.8 1.4 4.9 1.7 4.3 Mato Grosso: Private colonization Alta loresta 1 5b 1.5b Paranaita 1.3b 1.3b Mutom f9.0a 19.02 Siojos6 do Rio Claro 1.01' 1.01) Average 4.1 3.5 3.5 Total 1.8 3.3 1.4 5.0 1.7 3.9 IVote: 1981 differences were not tested. Lowercase letters: The presence of a common letter superscript indicates that there is 7oo difference at the 10 percent level of significance in the following cases: Verticallv (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). *See appendix B.2.3(1). Appendix A 93 Table A.4 Distribution of Monthly Accumulation, 1981 and 1991 [k or ACCUMI* (percent) .Southemners Others .ll 1981 1991 1981 1991 1981 199) Para: Official colonization Pacal 2.7 2.1 2.4 Anapu-Pacaja 1.2 2.7 2.4 Monte Alegre 1.5 2.0 1.9 Average 2.1 2.3 2.0 2.3 2.0 2.3 Mato Grosso: Private colonization Alta Floresta 2.5 2.5 Paranafta 1.6 1.5 Mutum 2.8 2.8 SaoJos6 do Rio Claro 2.0 2.0 Average 1.8 2.3 2.3 Total 1.7 2.3 2.0 2.3 Note: 1981 differences not tested. No differences found at the 10 percent level. 'See appendix B.2.3(1). 94 Su/stainable Settlement in the Brazilian Amazon Table A.5 Distribution of Total Agricultural Productivity, 1991 [QHI* (thousands of kilograms per hectare) Southerners Others .4A1 Para: Official colonization Pacal 62.4 16.0 36.7 Anapu-Pacaja 0.6 0.6 0.6 Monte Alegre 0.9 18.0 15.4 Average 45.2 10.6 22.1 Mato Grosso: Private colonization Alta Floresta 0.3 0.3 Paranaita 1.5 1.5 Mutim 1.8 1.8 SaoJose do Rio Claro 0.2 0.2 Average 0.8 0.8 Total 16.6 10.6 14.1 Niote: This table reflects productivity per hectare regardless of product. Values reflect output for various crops. For example, Pacal averages are larger than other averages because of widespread sugarcane production. *See appendix B.2.3(4). Appendir A 95 Table A.6 Deforested Land Used for Agriculture, 1991 [AGH1* (percentage) Soutlee?71e? Otber-s All Para: Official colonization Pacal 41.2b 384a 39.8b Anapu-Pacaja 29.5bc 21.8b 23.2bc Monte Alegre 11.2c 19.6b 18.4c Average 36. C 26.9c 29.8x Mato Grosso: Private colonization Alta Floresta 26i7bc 26.7be Paranaita 24.7bc 24.7bc Mutum 84.7a 84.7a SaoJose do Rio Claro 33.1bc 33.lbc Average 36.2 36.2y Total 36.IC 26.9c 32.5 Uppercase C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southerners compared with Others). Lowercase letters: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). x,y-Between states (Parn compared with Mato Grosso). *See appendix B.2.3(l). 96 Sustainabkle Settlemen7t in tbe Brazilian A7imasZon Table A.7 Distribution of Annual Yields, 1981 and 1991 [CROPG]* (kilograms per hectare) Ri-c Co0177 C'offJe 1981 1991 1981 1991 1981 1991 Para: Official colonization Pacal 1,097 1,765 859 1,500 200 2975 Anapu-Pacaja 1,068 1,007 704 902 415 658 Monte Alegre 1,010 1,369 941 1.500 Average 1,077 1,402 844 1,301 307 1,816 Mato Grosso: Private colonization Alta Floresta 1,573 696 933 Paranaita** 1,683 Mutum 1.244 1,861 1,230 2,400 485 SaoJose do Rio Claro 807 1,092 1,500 599 549 Average 1,352 1,671 1,255 1,824 550 741 Total 1,166 1,502 1,005 1,563 396 1,278 *See appendix B.2.3(1). -*Paranaita means are included in Alta Floresta means. Appendixc A 97 Table A.8 Prices for Crops and Land, 1981 and 1991 [CROPP and LP]* (U.S. dollars per kilogram) Land (US. dol//av per Rice C*0171 Coffee hectmre) 191i 1991 1981 1991 1981 1991 1981 1991 Para: Official colonization Pacal 0.14 0.14 0.09 0.14 0.16 0.17 130 248 Anapu-Pacaja 0.13 0.12 0.09 0.14 0.90 0.17 20 59 MonteAlegre 0.14 0.10 70 87 Average 0.13 0.12 0.09 0.14 1.13 0.17 80 131 Mato Grosso: Private colonization Alta Floresta 0.13 0 10 0.10 0.09 0.38 0_29 460 258 Paranaita** 0.14 Mutum 0.14 0.11 0.13 0.08 220 398 Saojose do Rio Claro 0.11 0.15 0.00 0.11 0.39 200 267 Average 0.13 0.11 0.10 0.09 0.38 0.29 290 308 Total 0.16 0.12 0.10 0.12 0.76 0.23 190 219 'See appendix B.2.3(2). *^With one exception, all Paranaita means are included in Alta Floresta means. 98 Sustainable Settlement in the Brazilian Amazonz Table A.9 Rice Productivity, 1991 [RICEI* (kilograms per hectare) All All Southerners Otbiers 1981 total 1991 total Pari: Official colonization Pacal 1,662 2,178 1.097 1,765 Anapu-Pacaja 1,100 990 1,068 1,007 MonteAlegre 3,125 492 1,010 1,369 Average 1,854 1,019 1.077 1,402 Mato Grosso: Private colonization Alta Floresta 1,390 1,541 1,390 Paranaita 1,683 ** 1,683 Mutum 1,861 1,244 1,861 SaoJose do Rio Claro 807 Average 1,671 1,3 52 1,671 Total 175IC I0 19C 1,166 1,502 Uppe-case C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southemers compared with Others). *See appendix B.2.3(1). **This mean is incorporated into the Alta Floresta mean. Appendir A 99 Table A.10 Time on Plot, 1981 and 1991 [TIME]* (years) Soutbhen7er Others Total 1991 1991 1991 1981 SUt- 1981 Sur- 1981 Szur- total vivors Total total v',vors Total total viZors Total Para: Official colonization Pacal 6.1 b 16bc 14b 6 8b h7b 14h 6.4Db 16b 14Db Anapu-Pacaji 3.8Dbcd 14bc 14Db 3.7c 13b 8.5c 3.7Dc 13b 9.3D- Mont Alegre 19Da 29' 28Da 16a 33a 25a 17Da 32Ea 25DEa Average 6.6 17B 16D 7.8D 21BE 15DE 7.4D 19E 15DE Mato Grosso: Private colonization Alta Floresta 2.8cd I3bc lob 2.8Dc 13b 1oDbc Paranaita 1.5d 12bc 9.5b 1.5Dc 12b 9.5Dbc Mutum 1.5d 12c 10b l. 51a 12h 1 oDbc SaoJose do Rio Claro 42bc 16b 13b 4.2bc 16b 13Dbc Average 2.6 13 1OD 2.6D 13E 1ODE Total 3.7AD 14BE 12CDE7.8D 21BE 15CDE5.3D 16E 13DE UTppercase letters: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance in the following cases: Horizontally between subgroups (Soutberners, Others, Total): A-Southerners 1981 compared with Others 1981. B-Southemers-Survivors compared with Others-Survivors. C-Southerners 1991 Total compared with Others 1991 Total. Horizontally witbin subgrouips (Souther-ners, Others, Total): D-1981 compared with 1991 Total. E-Survivors compared with 1991 Total. Lowercase letters: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b, c, d-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). Survivors: farmers who were interviewed in both 1981 and 1991 on the same plot of land. ' See appendix B.2.3(4). 100 Sustainable Settlement in the Brazilian Amazon Table A. 11 Age of Head of Household [AGEI* (years) Soutberniers Othei-s Total 1991 1991 1991 1981 Sin- 1981 Stir- 1981 Sur- total Iivors Total total vivory Total total vii ors Tota7l Para: Official colonization Pacal 44 50 50 44 55 52 44d 52' 51' Anapu-Pacaja 43 49 49 46 53 47 450 52' 47ab Monte Alegre 44 52 52 48 55 52 48a 540 52a Average 44D 50 50 46 54 50 45 53 50 Mato Grosso: Private colonization Alta Floresta 42 50 48 42ah 50ab 48ab Paranafta 45 53 47 45' 530 47ab Mutim 36 41 42 36b 41h 42b Sao,Jos6 do Rio C(laro 46 53 52 46a 53' 52a Average 42D 49 4 42 49Y 47D Total 43 50B 48 46D 54BE 50DE 44D 51E 49DE U.bperrcase lerterc The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance in the following cases: HorizontaIy betvween subgroups (Southerners, Others, Total): A-Southerners 1981 compared with Others 1981. B-Southerners-Survivors compared with Others-Survivors. C-Southerners 1991 Total compared with Others 1991 Total. Hozontally within subgroups (Southbern7er, Others. Total): D-1981 conipared with 1991 Total. E-Survivors compared with 1991 Total. Lowercrase letters: The presence of a commion letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, l-Between locations (for exaniple. Pacal compared with Mutum compared with Monte Alegre). Survivors: farmers who were interviewed in both 1981 and 1991 on the same plot of land. *See appendix B.2.3(4). Appendix A 101 Table A.12 Distribution of Total Net Worth (K)* (thousands of U.S. dollars) Southernoe-s Other-s Total 1991 1991 1991 1981 Sur- 1981 .u- 1981 Sur- total vizors Total total zivors Total total vivers Total Para: Official colonization Pacal 38.7 38.9 50.4 37.4 60.4' 63.1I 38.1 50.1 57.7c Anapu-Pacaj 13.2d 9.9C 79.4h 15.2h 17 1'h 29 2h 15 0 15 .3c 36.6- Monte Aiegre 172.9Ah 168.7Bh 168.7cb 19.7Ab 11.3B 11.6 37.0 42.8'c 33.5c Average 43.9x 51.4' 68.00 2 3.3 34.6 37.5D 29.5 40.7' 45.8x Mato Grosso: Private colonization Alt Fl 54 ODcd 6 hc 10 Dh 54.0 '64.1h 106.3Dh Alta Floresta 54.0 Dd64.1 10 I6.3 Dh 5.OI 641hs 0 Dbc Paranaita 39.3Dcd 63.4hc 1060oDh 393Dcd 63.41K 106.0 D Mutum 264.2Da 564.1' 513.00D 264.2Da 564.13 513 0OD SaoJose do Rio Claro 80.4k 124.8 141.2 804h 124.8 141.2 Average 78.0 . 164.5 176.5Dy 78.0 164.5> 176.50> Total 68.9 127.6 147.20 23.3D 34.6 37.5 51.60 93.7 101.4 Lppercase letters: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance in the following cases: Horizon tally hetween subgroups (Southen2ers, Others. Total): A-Southerners 1981 compared with Others 1981. B-Southerners-Survivors compared with Others-Sunrivors. C-Southerners 1991 Total compared with Others 1991 Total. Horizontally witbin .uhlgroups (Southerners, Others, Total): D-1981 compared with 1991 Total. E-Survivors compared with 1991 Total. Lowzeease letters: The presence of a common letter stuperscript indicates that there is no difference at the 10 percent level of significance in the following cases: V ertically (within column)-lowercase letters. a, h, c, d-Betxveen locations (for example, Pacal compared to Mutum compared to Monte Alegre). x, y-Between states (Para compared to Mato Grosso). Survivors: farmers who were interviewed in both 1981 and 1991 on the same plot of land. *See appendix B.2.3(1). 102 Sutstainable Settlement i?7 the Brazilian Amazon Table A.13 Perceived Improvement in Quality of Life, 1991 [LMPR]* (percentage) Sou tberiiers Otlbei Total Para: Official colonization Pacal 76ab 70' ab Anapu-Pacaja 67ab 73ah 72ab Monte Alegre 100a 88a 90a Average 76 74 755 Mato Grosso: Private colonization Alta Floresta 63ab 63ab Paranafta 50h 50b Murum r1 ab 91a S5oJos6 do Rio Claro 72 / Average 66 66Y Total 69 74 71 Lo-wercase letters: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase lerters. a, h-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). x, y-Between states (Para compared wvith Mato Grosso). *See appendix B.2.3(4). Appenzdix A 103 Table A.14 Planned Investment on Plot, 1991 [PLAN]* (percentage) Southeroers Others Total Para: Official colonization Pacal 44 3 9ab 39 Anapu-Pacaja 65 56a 58 Monte Alegre 30c 26 Average 44 42 43 Mato Grosso: Private colonization Alta Floresta 33 33 Paranaita 37 37 Mutum 39 39 SaoJos6 do Rio Claro 55 55 Average 37 37 Total 40 42 41 Lowercase letters: The presence of a common letter superscript indicates that there is 10 difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b, c-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). *See appendix B.2.3(4). 104 Sustainable Settlement in the Braziliani Amazon Table A.15 Inhabitants Who Plan to Remain on Current Plot, 1991 [FLX]* (percentage) Southernersv Otbn s Total Para: Official colonization Pacal 82 70b 76ab Anapu-Pacaja 82 82ab 82ab Monte Alegre 83 94a 93a Average 82x 81 81x Mato Grosso: Private colonization Alta Floresta 65 65b Paranaita 63 63b Mutum 96 96a SaoJos6 do Rio Claro 65 65b Average 69Y 69)? Total 73C 81C 76 Uppercade C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southerners compared with Others). Lowercase letters: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). x, y-Between states (Para compared with Mato Grosso). *See appendix B.2.3(4). Appendi.r A 105 Table A.16 Agricultural Gross Income [A]* (thousands of U.S. dollars) Soiitheo77e?r Otler7-Y Total Pari: Official colonization Pacal 4.3h 3.2 3.8h Anapu-Pacaja 5.3b 1.0 1.8 1 Monte Alegre 6.3b 2.1 2.7b Average 4.7 2.0 2.9t Mato Grosso: Private colonization Alta Floresta 0.9b 0.91) Paranaita 0.4b 0.41) Mutumn 43.3a 43.3a SaoJose do Rio Claro 6.2 b 6.2b Average 8.1 8.1 y Total 7.lC 25.C 5.1 Uppercase C. The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups, (Southemers compared with Others). Lowe7-case letters: The presence of a common letter superscript indicates that there is 7oo difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). x, y-Berween states (Para compared with Mato Grosso). *See appendix B.2.3(l). 106 Ststainiaable Settlement in the Brazilian Amazon Table A. 17 Total Expenditures, 1991 [E]* (thousands of U.S. dollars) Southerner-s Others Total Para: Official colonization Pacal 14.9b 3.7 9.5b Anapu-Pacaja 2.2it 2.9 2.8h Monte Alegre 14.8b 17.3 16.9b Average 13.0 6.9 8.9 Mato Grosso: Private colonization Alta Floresta 9 sb 9.5b Paranaita 9.2b 9.2b Murum 47.2a 47.2a Saojos6 do Rio Claro 8.1b 8.1 b Average 15.3 15.3 Average 14.6c 6.9C 11.7 Ul pper Case C. The presence ofa common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southerners compared with Others). Lowercase lettery: The presence of a common letter superscript indicates that there is no difference at the It percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Betneen locations (for example. Pacal compared with Mutum compared with Monte Al egre). *See Appendix B.2.3(3). Appendix A 107 Table A.18 Net Nonagricultural Income, 1991 [NI* (thousands of U.S. dollars) Soutber'ners Otb'erv Total Para: Official colonization Pacal 4.1 6.9 5.5 Anapu-Pacaja -0.3 4.2 3.5 Monte Alegre 11.9 1.5 3.1 Average 4.2 4.4 4.4x Mato Grosso: Private colonization Alta Floresta 13.0 13.0 Paranafta 34.9 34.9 Mutum 71.8 71.8 SaoJose do Rio Claro 14.9 14.9 Average 2 7.8 27.8y Total 20.9 4.4 14.7 Lowercase letters: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. x, v-Between states (Para compared with Mato Grosso). *See appendix B.2.3(l). 108 Sustainable Settlement in tbe Brazilian ,-znazon Table A.19 Net Subsistence Income, 1991 [S]* (thousands of U.S. dollars) Souther,7ers- Orbei-s Total Para: Official colonization Pacal 0.4 1.9 1.1 Anapu-Pacaii 0.5 -0.2 -0.05 Monte Ale-re -0.4 -0.3 -0.8 Average 0.05 0.5 0.4 Mato Grosso: Private colonization Alta Floresta 23.8 2 3.8 Paranaita -2.0 -2.1 Mutum 4.4 4.4 SaoJose do Rio Claro -1.6 -1.6 Average 11.5 11.5 Total 8.4 0.5 5.5 NAote: Statistical tests were performed at the 10 percent level of significance between origin groups, and no difference was found. This is indicative of the large degree of variance this variable possesses. See appendix B.2.3(l). Appendi.r A 109 Table A.20 Gross Income, 1991 [GY]* (thousands of U.S. dollars) Southerners Others Totarl Para: Official colonization Pacal 12.7 17.8 1S.2b Anapu-Pacaji 15.0 8.5 9.41) Monte Alegre 31.0 5.7 9.6 Average 14.8X 11.0 12.2x Mato Grosso: Private colonization Alta Floresta 57.5 57 5ab Paranaita 46.0 46.0 Mutum 156.5 156.52 SaoJose do Rio Claro 29.1 29.1 Average 67.C ' 67.1 ' Total 52.1c l lOC 36.5 ULppercase C: The presence of a conmnon letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southerners compared with Others). L.owsercase letterv: The presence of a common letter superscript indicates that there is 0io difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Betwveen locations (for example, Pacal compared with Mutumn compared with Monte Alegre). x, y-Between states (Para compared with Mato Grosso). *See appendix B.2.3(l). 110 Slltainable Settlemen?t in the Braziliani Amazon Table A.21 Net Income, 1991 [xl* (thousands of U.S. dollars) Southernery Othe s Total Pari: Official colonization Pacal -6.3 8.5 0.8 Anapu-Pacaja 1.8 2.5 2.5 Monte Alegre 0.2 -13.8 -12.0 Average -4.6 0.2 Mato Grosso: Private colonization Alta Floresta 28.2 28.2 Paranaita 24.1 24.1 Mutum 72.3 7 2.3 Saojose do Rio 11.3 11.3 Claro Average 32.2 32.2y Total 22.0 0.2 14.0 Uppercase C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southerners compared with Others). Lowercasve lettern: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Verticallv (within column)-lowercase letters. x, y-Berween states (Pars compared with Mato Grosso). *See appendix B.2.3(4). Appendix A 111 Table A.22 Formnal Land Owners in the Past, 1991 [FORMI]* (percentage) Soutbenrers Otlers Total Para: Official colonization Pacal 19 2jah Anapu-Pacaji 33 35a 35a Monte Alegre 0 8h 7b Average 20 23 22k Mato Grosso: Private colonization Alta Floresta 34 34ab Paranaita 31 31 ah Mutum 25 25ab SaoJos6 do Rio Claro 22 22ab Average 30 30Y Total 27 23 26 Lowercase letters: The presence of a common letter superscript indicates that there is 7o difference at the 10 percent level of signuficance in the following cases: Vertically (within column-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). x, y-Between states (Para compared with Mato Grosso). *See appendix B.2.3(4). 112 Sstwainable Settlement in the Brazilian Amazon Table A.23 Itinerancy [ITINM* (number of stops prior to arriving at current plot) Southeinerus Otbe r. Total Para: Official colonization Pacal 1 6ab 0.9 13 bc Nnapu-Pacaji 2.5' 1.6 173bc Monte Alegre 0.7 ) 0.9 0.8C Average 17Cx l )C 1.4x Mato Grosso: Private colonization Alta Floresta 2.3a 2.3a Paranaita 2a 2 lab Mutum 1.6ab 1.6abc SaoJose do Rio Claro 2.3' 2.3a Average 2.2y Total 2. c 1.2 1.7 Utppercase C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southerners compared with Others). Loulercase letters: The presence of a common letter superscript indicates that there is nio difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b, c-Between locations (for example, Pacal compared with Mutumn compared with Monte Alegre). x, 3-Between states (Para compared with Mato Grosso). *See appendix B.2.3(4). ,Appendix A 113 Table A.24 Inhabitants Whose Parents were Farmers, 1991 [PARNT]* (percent) Southeriewn Otbe7r Total Pari: Official colonization Pacal 91a 79b 85 Anapu-Pacaja 8,* 9011) 89 Monte Alegre 1 7Cb 95Ca 84 Average 83 88 86 Mato Grosso: Private colonization Alta Floresta 82' 82 Paranaita 92a 92 Mutumn 960 96 Sao 1ose do Rio Claro 67' 67 Average 85 85 Total 84 88 86 Upperca.re C: The presence of a common letter stiperscript indicates that there ir a difference at the 10 percent level of significance betwveen origin groups (Southerners compared with Others). Lowercise letters: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). 'See appendix B.2.3(4). 114 Sustainable Settlement in the Brazilian Amazon Table A.25 Net Investment, 1991 [11* (thousands of U.S. dollars) Southeer. Othe rs Total Para: Official colonization Pacal -6.2 8.5 0.9 Anapu-Pacaja 2.1 2. 5 2. S Monte Alegre 1.7 -13.8 -11.8 Average -4.4 0.2 -1.3X Mato Grosso: Private colonization Alta Floresta 28.4 28.4 Paranaita 24.1 24.2 Mutum 72.8 72.8 SaoJose do Rio Claro 11.3 11.3 Average 32.4 32.4~' Average 22.2 0.2 14.1 Lowercase letters: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. x, y-Berween states (Para compared with Mato Grosso). *See appendix B.2.3(3). Appendix,4 115 Table A.26 Deforestation Since Farrners' Arrival on the Frontier [F]* (hectares) Sourtheere Others Total Para: Official colonization Pacal 71.78 ' 74.89 73 0b Anapu-Pacaja 410.25ab 164.62 206.19b Monte Alegre 530.75Ca 60.70c 129.49b Average 177.96 108.09 130.29 Mato Grosso: Private colonization Alta Floresta 133.18b 133.18b Paranaita 92.89b 92.89b Mutum 581.71 581.71a SaoJose do Rio Claro 203.1 1ab 203.1 1b Average 198.46 198.46 Total 192.24c 108.09c 159.03 Note: Deforestation occurred not necessarily on current plots. Uppercase C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southemers compared with Others). Lowercase letters: The presence of a common letter superscript indicates that there is ns difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). *See appendix B.2.3(2). 116 Sustainable Settlement im the Brazilian Amazon Table A.27 Total Land, 1991 [HI* (hectares) Soulthe7ners Others Total/ Para: Official colonization Pacal 147.61 169.98a) 1 U4 Anapu-Pacaja 1,194.65Ca 240.( 57Ci 402.03'b MonteAAlegre 635. 33'i 6770 b 50,77b Average 384.40 171.75 239.31 Mato Grosso: Private colonization Alta Floresta 274.08'b 274.08b Paranaita 193.35b 193 5b Mutum 845.62ab 845.62a SioJose do Rio Claro 440.01'b 440.0l'b Average 358.56 358.56 Total 366.40c 171.75c 289.60 Ulppercase C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southemers compared with Others). Lo-werc7se lettet-c The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within colunin)-lowercase letters. a, b-Betwveen locations (for example, Pacal compared with Mutum compared with Monte Alegre). 'See appendix B.2.3(4). Appendix A 117 Table A.28 Deforestation in Current Year, 1991 [F913* (hectares) Southermers Other-s Total Para: Official colonization Pacal 17.49b 1.47 9.75b Anapu-Pacaja I I lb 4.55 3.88b Monte Alegre 800 b 3.03 3.72b Average 14.3x 2.92 6.93x Mato Grosso: Private colonizafion Alta Floresta 3.95b 3.95b Paranaita 3.46b 3.46b Mutum 244.05a 244.05a SaoJose do Rio Claro 7.80b 7.80b Average 64.1 SY 64.15Y Total 43.18C 2.92C 25.63 UppercaFe C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southerners compared with Others). Lowercase letters: The presence of a common letter superscript indicates that there is n1o difference at the 10 percent level of significance in the following cases: Verticallv (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). x, y-Between states (Pari compared with Mato Grosso). *See appendix B.2.3(2). 118 Sustainable Settlement in the Brazilian Amazon Table A.29 Deforestation in First Three Years Since Farmners' Arrival at the Frontier [F3]* (hectares) Southen ero Others Total Pari: Official colonization Pacal 17.54 b 19.32 18.44b Anapu-Pacaja 19.46b 24.52 23.57b Monte Alegre 2.67b 4.99 4.61b Average 16.52 17.99 17.51X Mato Grosso: Private colonization Alta Floresta 11.52b 1152b Paranaita 21.27b 21.27b Mutum 135.292 13 5.292 SaoJos6 do Rio Claro 7.02b 7.02b Average 3 3.56 33.56v Total 28.34 17.99 24.39 Note: Deforestation occurred not necessarily on current plots. Lowercase letters: The presence of a common letter superscript indicates that there is 77o difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). x., y-Between states (Para compared with NIato Grosso). See appendix B.2.3(2). Appendix A 119 Table A.30 Inhabitants Who Think Forest Preservation Is a Good Idea [OPINION and FWANT]* (percent) Southerners Others Total Para: Official colonization Pacal 66ab 62b 64ab Anapu-Pacaja 752b 832 82ab Monte Alegre 50b 57b 56b Average 66x 69 68x Mato Grosso: Private colonization Alta Floresta 76ab 76ab Paranafta 892 89a Mutum 83ab 832 SaoJose do Rio Claro 89a 89a Average 82Y 82Y Total 77C 69C 74 Uppercase C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southemers compared with Others). Lo-wercase ltters: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). x, y-Between states (Para compared with Mato Grosso). 'See appendix B.2.3(2). 120 Susrtainable Settlementt in the Brazilian Amazon Table A.31 Inhabitants Who Perceive a Loss of Soil Fertility, 1991 [FERT]* (percentage) Souther7oerr Otbers Total Para: Official colonization Pacal 36ab 4 351) Anapu-Pacaja 42ah 28 31be Monte Alegre 33ab 41 40ab Average 37X 34 35X Mato Grosso: Private colonization Alta Floresta 67a 67a Paranafta 60a 60;1b Mut-um 4b 4c Sao Jose do Rio Claro 56' 56ah Average i4Y 54Y Average 49c 34C 43 LUppercase C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southerners compared with Others). Losenr-ave letterr: The presence of a common letter superscript indicates that there is lio difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b. c-Between locations (for example, Pacal coinpared with Mlutum compared with Monte Alegre). x, y-Berween states (Para compared with Mato Grosso). *See appendix B .23(4). Appendix A 121 Table A.32 Inhabitants Who Employ Environmental Conservation Agricultural Techniques, 1991 [AGCON* (percentage) Southervere Others Total Para: Official colonization Pacal 23ab 23 23 bc Anapu-Pacaja 17b 13 14bc Monte Alegre 33ah 22 23bc Average 23 19 20 Mato Grosso: Private coloinization Alta Floresta 166b 1 bc Paranaita 25ah 25b Mutum 0b oc SaoJose do Rio Claro 56a 56a Average 20 20 Total 21 19 21 Lowercase letters: The presence of a common letter superscript indicates that there is 110 difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b, c-Between locations (forexample, Pacal compared with Alutum compared with Monte Alegre). *See appendix B.2.3(4). 122 Sustainable Settlement in the Brazilian Amazon Table A.33 Product Sold after First Harvest, 1991 [STOR]* (percentage) Sowtb'enrer.r Others TotMl Par-a: Official colonization Pacal 78 93 86 Anapu-Pacaja 90 77 79 Monte Alegre 61 72 68 Average 78 85 82 Mato Grosso: Private colonization Alta Floresta 83 83 Paranaita 78 78 Mutum 77 77 Saojose do Rio Claro 89 89 Average 81 81 Total 80 85 82 Note: No statistical differences were found at the 10 percent level. 'See appendix B.2.3(4). AppendixA 123 Table A.34 Distance to Nearest Market Center, 1991 [DISTII* (minutes) Southerners Others Total Para: Official colonization Pacal 28.3'b 20.7b 24.5bc Anapu-Pacaja 35.8' 32.8a 33.-ab MonteAlegre 40.8a 47.6b 46.70 Average 30.9 32.6 32.00 Mato Grosso: Private colonization Alta Floresta 24.3ab 24.3bc Paranaita 27.9ab 27.9bc Mutum 3 1l Oab 30o9abc SaoJose do Rio Claro 15.9b 15.9c Average 25.2 25.2y Total 26.9c 32.6C 29.2 Upperc-ase C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southerners compared with Others). Lowercase letters: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b, c-Between locations (for example, Pacal comparedwith Mutun compared with Monte Alegre). x, y-Between states (Para compared with Mato Grosso). 'See appendix B.2.3(4). 124 Sustainable Settlemenit in tbe Bazi/lian Amazon Table A.35 Product Sold at Farm Gate, 1991 [GATE]* (percentage) .Sutherne,iev Ots'e-rs Total Par.: Official colonization Pacal 77a 75a 76' Anapu-Pacaji 4-lbc 5 a 3 a Monte.4legre oc 14 b 11 b Average 65x 54 58" Mato Grosso: Private colonization .lta Floresta 60ah 60' Paranaita 67a 67' Mutum 9bc 9b Sao !ose do Rio Claro 56ab 56' Average 44 44' Total 53 54 53 Low-ercase letterv: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: .Vertically (within column)-lowercase letters. a, b, c-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). x, y-Between states (Pars compared with Mato Grosso). *See appendix B.2.3(4). Appendix, A 125 Table A.36 Inhabitants Who Have Received Agricultural Credit, 1991 [CREDITI* (percentage) Sothberners Others Total Para: Official colonization Pacal 4C 2 Anapu-Pacaja Oc 0 Oc Monte Alegre 33Cab 37bc Average 6 1 3 Mato Grosso: Private colonization Alta Floresta IC IC Paranaita 3c 3C Mutum S2' 52a Sojos6 do Rio Claro 17bc 17b Average I I I I y Total 1 OC IC 7 Uppercase C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southemers compared with Others). Lowercase letter.r: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b, c-Between locations (for example, Pacal compared with Munum compared with hlonte Alegre). x, y-Between states (Para compared with Mato Grosso). *See appendix B.2.3(4). 126 Sustainable Settlement in the Brazilian Amazon Table A.37 Net Debt, 1991 [D]* (thousands of U.S. dollars) Sorutherners Others Total Pari: Official colonization Pacal 0.1 0.02 Anapu-Pacaja 0.2 0.03 MonteAlegre 1.3c 0.02c 0.2 Average 0.3 0.006 0.08 Mato Grosso: Private colonization Alta Floresta 0.2 0.2 Paranaita 0.007 0.007 Mutum 0.5 0.5 SaoJose do Rio Claro Average 0.2 0.2 Total 0.2c 0.006C 0.1 Upperrase C: The presence of a common letter superscript indicates that there ir a difference at the 10 percent level of significance between origin groups (Southerners compared with Others). *See appendix B.2.3(2). Appendix A 127 Table A.38 Inhabitants with Title to the Plot, 1991 r[I]* (percentage) Southerners Others Total Para: Official colonization Pacal 87 85 86b Anapu-Pacaja 92 72 75ab Monte Alegre 100 ** ** Average 88 78 82 Mato Grosso: Private colonization Alta Floresta 82 82ab Paranaita 86 862b Mutum 100 1foo SaoJose do Rio Claro 94 94ab Average 87 87 Total 87C 78C 84 Uppercase C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southemers compared with Others). Lowercase ktters: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within columnn-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). *See appendix B.2.3(4). 'Data not yet available. 128 Su7stainable Settlement in the Brazilian Amazon Table A.39 Family Workers Older than 13 Years, 1991 [WKRS]* Southerners Other-s Total Para: Official colonization Pacal 3.6 3. 1 3.3 Anapu-Pacaja 3.4 30ab 3.2 Monte Alegre 45 1.9b 2.2 Average 3.6Cx ?.8C 3.0X Mato Grosso: Private colonization Alta Floresta 2.7 2.7 Paranaita 2.4 2.4 Mutum 2.1 2.1 SaoJose do Rio Claro 1.7 1.7 Average 2.5y 2.5y Total 2.8 2.8 2.8 Upperca;.e C. The presence of a common letter superscript indicates that there ir a difference at the 10 percent level of significance between origin groups (Southerners compared with Others). Lowuercase letters: The presence of a common letter superscript indicates that there is n7o difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). x, y-Between states (Para compared with Mato Grosso). *See appendix B.2.3(4). .4ppendix A 129 Table A.40 Agricultural Land, 1991 [CROPL]* (hectares) Southerners Otbev-s Total Para: Official colonization Pacal 25.63b 31.36a 28.49b Anapu-Pacaja 57.79ch 10.24 18.88b Monte Alegre 33.42b 6.13b 10.03b Average 32.28x 16.41 21.52x Mato Grosso: Private colonization Alta Floresta 39.40b 39.40b Paranaita 6 )b 6.52b Mutum 373.79a 373.79a Saojose do Rio Claro 53.33b 5333 b Average 86.28Y 86.28Y Total 7O.03C 16.4lc 49.22 Uppercase C: The presence of a common letter superscript indicates that there is a difference at the 10 percent level of significance between origin groups (Southemers compared with Others). Lowercase letters: The presence of a common letter superscript indicates that there is no difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Mutum compared with Monte Alegre). x, v-Between states (Para compared with Mato Grosso). *See appendix B.2.3(2). 130 Suftainable Settlement in the Brazilian Amazon Table A.41 Agricultural Price Index, 1991 [PINDEX1 (U.S. dollars per lilogram) Southenrers Otber-s Total Pari: Official colonization Pacal 0.2 ab o29ao Anapu-Pacaja 0O 17h . 26a 0.24b Monte Alegre 0 09b 0 09b 0 09b Average 0.22 0.23 0.23 Mato Grosso: Private colonization Alta Floresta 0.23hb 0.23b Paranaita 0.1 7b 0.1 7h Mutum 0.08b 0.08b SaoJose do Rio Claro 0.52a 0.52a Average 0.20 0.20 Total 0.21 0.23 0.22 Lowercase letters: The presence of a common letter superscript indicates that there is 110 difference at the 10 percent level of significance in the following cases: Vertically (within column)-lowercase letters. a, b-Between locations (for example, Pacal compared with Murum compared with Monte Alegre). Appendix A 131 Table A.42 Net Present Values of Agricultural Income and Land [NPVA and NPVL] IVPI' Strn it7al (NP! A + rates IVPT' ~~~~~~~~~~~~h cPe pVlq IVPVIL- NPTIT) Landprices _berenet) Pari: Official colonization Pacal 176.4 130.6 307.0 248.0 74 Anapu-Pacaja 33.0 31.1 64.1 59.0 48 Monte Alegre 131.7 45.8 177.5 87.0 81 Average 89.2 69.0 158.2 131.0 67 Mato Grosso: Private colonization Alta Floresta 24.1 135.9 160.0 258.0 63 Paranaita* Mutum 376.8 209.6 586.4 398.0 68 SaoJose do Rio Claro 103.8 140.6 244.4 267.0 53 Average 145.0 162.2 307.2 308.0 61 Total 129.6 115.3 244.2 219.0 64 a. \TPVA and NPrI were calculated using data from tables A.8 and A.27. The NPVA calculation [A]/[HI (tables A.16 and A.27) obtains current value of agricultural income givennondeclining land productivity throughout the period under consideration. The NTPV was obtained by dividing that ratio by (I + r)t for each year over the 1981-91 time span. The discount rate used v.was 0.06 (the yearly interest rate on the cadenieta depoupanfa, farmers' opportunity cost in the financial market). This period was considered because it is approximately the mean time a farmer spends on each lot (see table A. 10) and because the study focuses on a decade-long trend. -pxt calculation: \'Tv is the salvage value of the plot at the end of the period under consideration. From table A.8 one obtains (Price 91/Price 81)1/t.100-100 = g, the average geometric growth rate of land prices during the 1980s. Then, NPXL = Price 91.g/(l + r)t. b. Obtained from table A.8. c. Obtained from table A. . *Mean land prices in Paranaita were included with mean land prices in Alta Floresta. 132 Sustainable Settlement in the Brazilian Amazon Table A.43 Distribution of Initial Capital (Ko)* (thousands of U.S. dollars) Southemners Otbers Totnl 1991 1991 1991 1981 Sur- 1981 Sur- 1981 S7r- total vivory Total total vivors Total total vivors Total Pari: Official colonization Pacal 9.0 16.2 10.4 19.1 9.8 17.7 Anapu-Pacaja 2.7 2.7 8.0 28.9 6.2 24.8 Monte Alegre Average 6.4 7.6 13.9 4.3 7.2 17.9 5.0 7.4 16.8 Mato Grosso: Private colonization Alta Floresta 14.0 23.0 Paranafta 18.4 23.2 Mutum 37.7 73.0 SaoJose do Rio Claro 42.6 27.7 Average 28.2 21.7 31.6 Total 22.5 18.4 28.1 16.5 14.9 24.6 *See appendix B.2.3(4). APPENDIX B Empirical Procedures B.l Data Collection The data for this analysis were collected from colonization projects in the Brazilian Amazon in 1981 and 1991. Field work was conducted in Para and Mato Grosso by staff from IPEA from September to December 1991. Farmers, merchants, and institutions were interviewed to establish a broad picture of the economic, social, political, and institutional conditions of directed colonization in the Amazon. The 1981 sample covered 498 farmers, 50 merchants, and 100 institu- tions. The 1991 sample covered 372 farmers, 128 merchants, and 85 institutions. The years 1981 and 1991 were chosen for several reasons. They were near census years, which provide general information on the universe being sampled. They were normal years agriculturally in the Amazon in general and in the sample locations in particular, with no extraordinary positive or negative features. The ten-year comparison, therefore, is not biased, and this is an important issue in a panel with few time-series.' Finally, these years covered the 1980s, a period when the economic and political landscape of Brazil changed drastically. This decade is important in evalu- ating demand for settlement programs during the 1990s. The locations sampled in 1991 were the same as those in 1981. All were directed colonization projects, some official and some private. Official colonization projects were all IN'CRA projects in Para-Anapu, Monte Alegre, Pacaja, and Pacal. Private colonization projects were conducted by different private firms and cooperatives in Mato Grosso-Alta Floresta, Mutum, Paranaita, and SaoJose do Rio Claro. These locations were chosen based on multiple criteria, including the lack of comparable field studies. Projects in Rond6nia were not included in this research because of the large amount of data already collected there. Locations visited in 1981 were representative of the Amazon frontier. Ten years later, these same plots were no longer frontier areas. By returning to them, the team observed how frontier conditions affect locations in the long run as these areas become incorporated into the broader economy. This 133 134 Sustainable Settlemenzr in the Brazilian Amazon methodology allowed for the observation (and quantification) of the eco- nomic motives that trigger farm turnover-an unprecedented accomplish- ment in the economic literature on the deforestation of the Brazilian Amazon. A random stratified sample was designed in each location, based on population size, access to market, origin of migrants, and time on plot. In 1981, 498 farmers were interviewed, covering 7 percent of the local popu- lation in corresponding census districts. In 1991, 356 farmers were visited, all in the same plots of land visited in 1981. The percentage of the universe sampled in 1991 cannot be calculated for lack of complete 1991 census data. It is certain, however, that the percentage is smaller than ten years pre- viously-the sample is smaller and the areas have grown. Many farmers were not reached in 1991 because of farm consolidation and the absence of interviewees in the sample plot. Also, because of financial limitations on the 1991 research, 10 percent of the Trans-Amazon sample was not revisited.2 Farmers interviewed in both 1981 and 1991 are called survivors. Those interviewed in 1981 but not in 1991, the old-timers, are compiled with survivors in the 1981 total. Those interviewed in 1991 hut not in 1981, the newcomers, are compiled with survivors in the 1991 total. The tables in appendix A present these three totals separately (1981 total, 1991 survivors, and 1991 total), and the differences among them are tested statistically. Questionnaires measured a large set of microeconomic, agricultural, and other variables of the farming household. All production and consumption activities are considered part of a simultaneous decisionmaking process. WA'hen more than one family inhabited a plot of land, a decisionmaker for the entire plot (sometimes a father or a son) was identified. Values for the other households were then added to those of the decisionmaker. Social informa- tion (such as origin) was recorded only with respect to the decisionmaker. Alicroeconomic variables. The relationship among these variables was es- tablished according to standard accounting procedures referring to the most recentlv completed agricultural year (1980 and 1990). Current account refers to net monetary and nonmonetary income measured with respect to agri- cultural and nonagricultural activities. Financial account refers to net mone- tary and nonmonetary investment and indebtedness-inflows and outflows of financial resources (loans and repavments) and physical investment. Net uworth refers to the value of all assets at the end of the reference agricultural year June 30). Initial capital refers to net worth brought to the land on arrival. Time on plot refers to the number of years spent on the plot. Imputed wage refers to the value of all goods and services bought or produced for family consumption, divided by the number of familv workers. Accumulation rate is the average monthlv increase in the value of net worth since arrival. Unit prices and unit costs of main product inputs were also recorded. Agric7ltural variables. These variables refer to various indicators of farm- ing performance-productivity, land use, deforestation, and so on. Appendix B 135 Other variables. These variables consider the origins, destinations, and individual characteristics of migrant farmers. Origin characteristics refer to previous farming location, prior itinerancy, whether parents were farmers, and land ownership. Destination characteristics refer to type of colonization project the farmer chose to settle in, land qualitv, and distance to market. Individuial characteristics are tenure, age, education, family size, storage facilities, marketing outlet, intentions, and perceptions concerning conser- vation issues, benefits, fixation, and so on. Data correction and retrieval are still under wav. As information becomes available, more issues will be analyzed than have been possible to address in this publication. B.2 Data Analysis B.2.1 Subsamples 1) Total 1981: farmers intenriewed in 1981; 2) Total 1991: farmers inteniewed in 1991 (but not necessarily in 1981); 3) 1991 Survivors: farmers inteniewed in 1981 and who were still on the same plots in 1991. B.2.2 Statistical procedures 1) Descriptive tables for: i) "Total 1981," "Total 1991," and "Sunrivors 1991;" ii) First differences: "Total 1981-91" and "Survivors" means and standard deviations; iii) Tests for differences of means by origin of migrants: "Southerners" compared with "Others"; by type of colonization project: "Official" (Para) compared with Private (Mato Grosso); by location: (1) in Pari: Pacal, Anapu, Pacaji, Monte Alegre; (2) in Mato Grosso: Alta Floresta. Paranaita, Mutum, Sao Jose do Rio Claro; by date of interview: 1981 compared with 1991; by duration: "Survivors" compared with "Total 1991." 2) Correlation, regression, and covariance analysis for: i) "Total 1981," "Total 1991," and "Sunivors 1991;" ii) First differences: "Total 1981-91" and "Survivors." B. 2.3 Description of variables 1) Indicators of private benefits: Monetary values in 1991 U.S. dollars per year per family: 136 Sutstainable Settlement in the Brazilian Amazon w wage rate per family worker per year. IMPW imputed wage (in minimum wages) = (monetary + nomnonetary family consumption)/familv workers. GY gross income = all monetary and nonmonetary incomes. Y net income = GY - E (see E in (3), next page). A gross agricultural income = monetary income from all crops. N net nonagricultural income = monetarv income from interest pay- ments + rents + wage employment of family workers + transfers (official and personal) + businesses - nonagricultural monetary expenditures. s net subsistence income = all income in kind and barter - all expen- ditures in kind or barter. K net worth: value of all assets (financial and physical) = stocks of products and inputs + constructions + land + durables + equipment - debt outstanding. 2) Indicators of environmental costs: AGH deforested land used for agriculture. F total deforestation since arrival (in hectares). F3 deforestation during first three years on lot (in hectares). F91 deforestation in 1991 (in hectares). FwANT desire to preserve (I = yes; 0 = no). CROP intensiveness (in hectares) = land in perennial + temporary crops. FERT loss of soil fertility since arrival (I = yes; 0 = no). CROPP distribution of prices for crops and land (U.S. dollars per kilogram and U.S. dollars per hectare). 3) Indicators of private costs: E current family farming expenditures = monetary agricultural expen- ditures + monetary family expenditures. The prices associated with such expenditures are: p prices of agricultural products. w wage rate paid out. h cost of land. c cost index of inputs. 4) Colonization characteristics: Destination va?iables: STATE type of project in each state (I = official; 0 = private). DIST distance to market = time (in minutes) it took to reach plot from local center. Three different measurements were taken: DIST 1-distance from plot to nearest market center. DIST 2-distance from plot to main regional market center. DIST 3-travel time from local center to plot (in minutes). Appendix B 137 QH general agricultural productivity index (kilograms per hectares): quantity (kilograms)/area (hectares) harvested. Temporary crops (rice, soy, corn, manioc, beans, others) + perennials (cocoa, coffee, rubber, pepper, sugarcane, others). STOR sales during first post-harvest quarter (percentage). GATE sales at farm gate (percentage). CREDIT whether farmer receives credit (I = yes; 2 = no). TNST number of local institutions in which the farmer participates- rural extension, cooperatives, unions, associations, and church. Origin variables: ORIG previous farming experience (I = southern; 0 = other). FORM former owner of landed property (I = yes; 0 = no). ITIN number of migratory stops prior to this location. PARNIT parents were farmers (I = yes; 0 = no). Individual variables: H farm size (hectares)-total area = cleared plus forested land. WIKRS total number of family workers older than 13. AGE age of head of household. Ko initial resources on arrival at plot. AGCON whether farmer practices area or crop rotation (I = yes; 0 = no). FERT whether farmer perceives loss of soil fertility now or in the future (I = yes; 0 = no). TIME average time on plot (years). TD whether farmer has title to property (1 = yes; 2 = no). PLAN whether farmer plans to invest more in agriculture. FiX whether farmer intends to remain on the plot (1 = yes; 2 = no). IMPR whether farmer perceives improvement compared with previous condition (1 = yes; 2 = no). B.3 Exchange Rates The exchange rates below were used to standardize all monetary values to 1991 U.S. dollars. The exchange rate devalues the Brazilian cruzeiro beyond the average exchange rate to reach what is considered appropriate parity.3 This correction may be exaggerated and may underestimate 1991 dollar values. This places strict conditions on the analysis of changes over time. Some discussants of this work consider these values and growth rates too high, so analysis should be considered carefully. June 1991 exchange rate-Cr$346.3 1/US$. Monetary correction:4 Alta Floresta and Paranaita: 1991 prices/1.9. Mutum and SaoJos6 do Rio Claro: 1991 prices/1.5. 138 Sustainable Settlenent in the Brazilian Amazon Monte Alegre: 1991 prices/2.39. All others: 1991 prices/I. 1981 exchange rate (in constant 1991 cruzeiros): Cr$93 = US$ 1. Correction factor used: 1.5 x (1981 US$) = 1991 U.S. dollars. Notes 1. For standard procedures in panel data analysis., see Hsiao (1986) and Maddala (1983). 2. This gap is being bridged by supplementary field work. Data are still being processed and will be the subject of future analysis. 3. According to [PEA. 4. Since locations were surveyed at different times, monetary adjustment is needed to correct for inflation. APPENDIX C A Model of Amazonian Deforestation C.1 The General Model C. l.1 Cumrent account Agricultural product is measured over a one-year period, from July 1 to June 30 of the following calendar year. Production occurs according to a conventional production function (Q), which is continuous and twice dif- ferentiable in land (H), labor (L), and inputs (C), such that: [1] Q = Q (L,C,H); Qi > 0; Qi < 0; for i = L,C,H. In family farming it is difficult to separate the use of fuel, motors, transport, equipment, buildings, and family labor into productive and household purposes. It takes long-term live-in observation to separate the use of time and goods into consumption and production functions. Inter- views that attempt to differentiate these activities yield inconsistent answers because respondents do not use these microeconomic concepts in consider- ing their activities. The family farming account, therefore, lumps together both household and farming expenditures (E). These include the cost of land (hH), labor (wL), and other current costs (cC). The cost of land held by the farmer comprises purchase, rental, and deforestation costs, weighted and averaged into a unit price per hectare (h). A farmer covers opportunity costs of land if the expected future income from farming is greater than or equal to (h). Given the inflationary Brazilian economy, there is a general tendency for real estate appreciation because land is an important store of value. In a frontier-where population and physical and social infrastructure are increasing rapidly relative to the established regions of the country-real estate tends to appreciate much 139 140 Sustainable Settlement in the Brazilian Amazon faster than elsewhere. This raises the rate of return necessary for frontier farming to cover its opportunity cost in the land market. The cost of labor comprises consumption by household workers (self- remuneration) and payments to hired workers, weighted and averaged into a mean wage rate (w). A frontier farmer covers opportunity costs of family labor if (w) is greater than or equal to the average wage in the labor market, at the farmer's skill level. Since half the urban labor force in Brazil earns less than an official minimum wage (MXV), the opportunity cost of labor in frontier farming can be measured by this minimum wage discounted bv the probability of getting a job at this rate. Recession, unemployment, and underemployment in the 1980s reduced this probability from the level of the 1 970s. A decline in the purchasing power of the minimum wage during the 1980s lowered the average real wage in the economy as a whole. Both tendencies have reduced the opportunity cost of frontier farming in the labor market. I The cu7rent cost of all other inputs, such as depreciation, maintenance, and repairs, is weighted and averaged into a unit cost index (c). These costs are measured in local markets and include transport costs from industrial centers in the south of the country to the frontier. In recent years, roads have deteriorated, energy and fuel subsidies have decreased, and interest rates on credit for the purchase of agricultural inputs have risen. Frontier farming production costs, therefore, have risen compared with those paid by farmers in the rest of the economy. These tendencies have increased the opportunity cost of frontier farming in input markets. Annual monetary (cutrent) expenditures for a family farm are therefore: [2] E=wL+cC+hH. Agricultural income comes from the sale of a part of the farm product (q) at an average unit price (p), so gross monetary farm income is A = p-q. Nonmarketed output (s) is valued at the same market price, so total gross farm income (monetary and nonmonetary) is p (q + s) = p.Q (GY in appendix table A.20 and in the regressions of appendix D). Net nonmonetary (or subsistence) farm income is S = p.(s - e), where (e) corresponds to nonmone- tary expenditures, such as use of farm resources for building (lumber), fuel (firewood), inputs (seed), and so on. Net nonagricultural income (N) comes from hiring out family members, rents, private and official transfers, non- farm businesses, and interest payments on loans extended to others minus interest payments on debts incurred previously (r*DT-1). DT-I is debt outstanding and (r) is the going annual interest rate. The sum of net farming income (A - E), net nonagricultural income (N), and net subsistence income (S) comprises total net current income: Appendix C 141 [3] Y=A+N+S-E whereA,E>O; N,SŽOor 0) or deficit (Y < 0) in current account. C. 1.2 Financial account: portfolio management Total net income (TY) is given by net current income (Y) plus the balance on financial account (B): [41 TY=Y+B where B = D - I. Financial account includes two important components. One is net indebt- edness (D)-loans received in the current year (DT) minus principal repay- ment from previous loans (DT-1). The other is net investment (I), which is gross investment minus depreciation. Net investment has several parts: savings minus dissavings; purchase minus sale of such durables as land, equipment, and cattle; net additions to stocks of products and inputs; construction and manufacture of such durables and benfeitorias as fences, wells, and roads; and land-use upgrading, including deforesting, destump- ing, pasturing, planting temporary crops, and planting permanent crops. A farmer's portfolio is determined by the combination of investments and disinvestments, borrowings, and repayments that determine whether the financial account balance is positive or negative. Positive balances in financial account (B = D - I > 0) indicate net borrowing (D > 0), disinvestment (I < 0), or both. Net borrowing means that future repayment will reduce capacity to cover current expenditures in subsequent years. For the poorest familv farmers this often takes the form of crop liens (when payment will be in kind, from the next harvest) or debt peonage (when the farmer ovenvorks his family and land to pay off debts).2 Net disinvestment reduces productive capacity in subsequent vears. It is frequently used to cover current expenditures such as selling cattle to cover 142 Sustainable Settlemnent in the Braziliani Amazon deficitary current accounts. Positive balances in financial account implv insolvency; in the long run they mav prevent accumulation. Negative balances in financial account (B = D - I < 0) indicate net principal repayment (D < 0), investment (I > 0), or both. Net repavment reduces current income, but it indicates there are sufficient resources to cover current expenses and to reduce debt outstanding. Net investment also reduces current income, but it adds to productive capacity in subsequent years. Negative balances in financial account imply solvency; in the long run they may lead to a process of accumulation. C. 1.3 Investment and accumulation The decision to invest is not simple and depends on several intertemporal criteria, including the expectation of future income from alternative uses of capital. This decision will not be modeled in this section, as the objective here is not to project current behavior onto the future but to observe the results of past behavior on the present. Past investments determine the net worth of farmers in the current agricultural year and the rate of accumula- tion they have experienced since arrival on their plots. These variables are indicative of economic performance on the frontier over time. The rate of accumulation is the average annual rate of increase in net worth since arrival (k). Net worth in year (T) is (KT), equal to the value of total assets minus total debt outstanding. Household and productive assets are lumped together in this total because it is difficult to separate them in family farming. Initial capital-the real net value of all financial and physical assets brought to the frontier by the migrant-is Ko. Given that real interest payments on savings certificates (caderneta de poupan7(a) were fixed at 0.5 percent a month in 1991, a frontier farmer was covering the opportunitv costs in the use of his capital if his accumulation rate (k) was at least 0.5 percent a month. Thus, current net worth is the result of investments since arrival. Rates of accumulation (k) and net worth (KT) can be computed as follows: Frr [5] a) k = (KT/KO) T100 - 100 given KO > 0. T b) KT=lK%(l+k)T+yX1,I1(+k)t r=0 The rate of accumulation and current net worth are thus functions of initial capital and time on the plot.3 A farmer's portfolio determines whether and how much he can invest and accumulate. Farmers with positive financial balances (B > 0) are not capable of investing in agriculture and should have relatively low rates of accumu- Appendix C 143 lation (low k). These farmers will be called insolvent farmers. Farmers with negative financial balances (B < 0) are investing productively in agriculture and should have relatively high rates of accumulation (high k). These farn-mers will be called so/vent.4 C. 1.4 Mlaximizatioin For equilibrium, farmers maximize agricultural output Q = Q(L,C,IH) each year, subject to the yearly total net income constraint (TY = Y + B), which leads to first order conditions: [6] a) Qi -. £(Ei) = 0 where Qi = dQ/di; Ei=dE/di; i = L,C,H; £ = Lagrangean multiplier. b) Y+B=0 The marginal effect of factors (L,C,H) on the total income constraint reduces to their effect on current expenditures (TYi = El). Net current income (X-wand its components, agricultural income (A), nonagricultural income (N), and subsistence income (S)-are assumed to be predetermined, since they are given from the previous year. Therefore, they are not affected by the use of labor, land, and other inputs during the current agricultural year: [7] dY/di = dA/di = dN/di = dS/di = 0 where i = H,L,C. According to equation [2] (page 140), farming expenditures in the current year (E) are affected by current use of inputs, according to unit costs (w, h, and c) The balance on financial account, in turn, is affected by variations in input use (Bi > 0), because debt repayment and investment both require increasing use of inputs even if they do not add to net current income. The model has four equations, five exogenous variables (£.h,w,c,p), and six endogenous variables (L,H,C,Q,D,I). It is, therefore, not exactly defined. C.2 The Static Model The model has a static and a dvnamic version. For the static version, assume that there is no balance on current account: [81 B=I=D=0 144 Sustainable Settlem,ent in the BraZilia n Amazoon From equation [5], the rate of accumulation would be zero (k = 0), and net worth in year T would be the same as in the initial year (KT = Ko). Under such steady-state conditions, equations [6a] and [61)] reduce to a system of four equations in four dependent variables (L,C,H,Q) and four independent variables (£, w/p, c/p, h/p) with product price (p) as numeraire. Thus: [91 a) Q1 - £Ej =O where Qj = dQ/di is the marginal productivity of factor i = L,C,H; Ei = w/p,c/p,h/p. b) Y=O In equilibrium. equation [9] is exactly satisfied, and each productive input (L,C,H) is used up to the point where its marginal productivitv (Qi) is equal to its opportunitv cost, which is its real market price (w/p, c/p, hI/p). If second-order conditions for a maximum are satisfied, these equations jointlv determine the demand for labor, inputs, and land as well as the supplv of agricultural product. Equation [9a] represents the first-order condition in which marginal productivities are proportional to real input costs. In par- ticular, the marginal productivity of land should be proportional to the real price of land: 110] dQ/dlI=Y£h/p where £ is a positive constant. Demand equations for agricultural inputs L,C,H are derived from equa- tion [9] as functions of real input costs and net income. In particular, the derived demand for land (H) is: [11] H=H(w/p,c/p,h/p,Y). Given the numerous factors that determine variation in farmer behav- ior, equation [11] can be considered stochastic and to include, on first approximation, a normally distributed random error term (v). As an initial approach to econometric analysis, the empirical steady-state deforestation equation is: [12] H=ao+al.h/p+a,.w/p+a,.c/p+a4.Y+v where a(1 is the intercept, a (j = 1,...,4) are the coefficients, and v is the error term to be estimated. A variant of equation [12] that takes into account different sources of net income is: [13] H' = ao' + a,'.h/p + a2'.w/p + a,'c/p + a4'A + a,'.N + a6'-S + v' Appendix C 145 where ao' is the intercept, a,' (j = 1,...,6) are the coefficients for prices and for each income component, and v' is the error term to be estimated. C. 2.1 Static income and price effects If there are few technical discontinuities in agriculture and if economies of scale are not significant, it is expected that demand for land should rise with income. That is, there should be a positive income effect for net income (Hy = a4 > 0) and its components: agricultural, nonagricultural, and subsis- tence incomes (HA'= a4' > 0, HN' =a5' > 0, H5' = a6' > 0). An increase in land prices, however, has both a substitution and an income effect. The substi- tution effect is always negative-an increase in the price of land decreases the demand for land-other things held constant. This effect should be enhanced by the positive income effect, since a decrease in income (pro- voked by the increase in prices) decreases the demand for land. The total own-price effect is thus expected to be negative. Cross-price effects depend on whether land substitutes for or comple- ments other factors. If it is a substitute the cross-price effect should be positive (a2', a3' > 0). An increase in the wage rate, for example, should decrease demand for labor and increase demand for land. But given the opportunity cost of family labor in the outside labor market, an increase in the wage rate beyond the marginal value product of labor on the farm may discourage household labor, promote exit from farming, and reduce the demand for land. In this case the cross-price effect of wages on demand for land will be negative. Cross-price effects cannot be predicted because they generally depend on internal rates of technical substitution between factors and on farmers' alternatives in the market. C.3 The Dynamic Model: Indebtedness and Investment Since equilibrium is rarely attained even under normal circumstances, the instability of frontier conditions should lead not to equalities (equation [9]), but to Kuhn-Tucker inequalities. These inequalities, in turn, require slack variables to satisfy first-order conditions. By relaxing conditions (equation [8]), surplus or deficit arises in current accounts, with the balance in financial account (B) picking up the slack, (B = D - I) becomes the slack variable in the model. Thus, equation [9] becomes: [14] a) Qi - (E, - Bi)= 0 where Q, = dQ/di is the marginal productivity of factor i = L,C,H, as in [9]; E, = dE/di is the marginal expenditure on factor i; B, = Di - Ii = 0 is the effect of factor i on balance of financial account; £ is a Lagrangean multiplier. b) Y+B=O. 146 Sustainable Settlemenit in the Brzazilian77 Amazon The farmer no longer maximizes output during the current agricultural year, but maximizes over a longer period. In this dynamic system, both indebtedness and investment occur (D,I 0). Demand for current agricul- tural inputs (L,C,H) is derived from equation [141 as functions of marginal expenditure (Ei = w/p, h/p, c/p), as before, plus marginal balance on financial account (Bi = Di -1I). For the purpose of exploratory estimation, one may single out from [14a] an equation for dynamic empirical demand for land: [15] H = b0 + blh/p + b, w/p + b1.c/p + b4TY + u where (TY = Y+B) is the total income variable, bo is the intercept, b, (j = 1, . . .,4) are the coefficients, and u is the error term to be estimated. An alternative specification for the above equation is to separate total income into its components in current account (Y) and financial account (B), so that [161 H'=b ' +b 1.h/p + b,'.w/p + b,'.c/p + b4'Y + b'B+u' where b)' is the intercept; b1' j = 1,....5) are coefficients for prices, net income, and financial balance; and u' is the error term to be estimated. C. S. 1 Dynamic income effrcts In dynamic equations the current income effect on demand for land is replaced by a total income effect. In equation [14a], equating the marginal expenditure on land to the real price of land (Eh=h/p) provides [14a'] Qh - £ (Eh - dB/dH) =O. Rearranging these terms yields: [171 dH/dB = -1/ (Qh/£- Iip) given that £ > 0, and Qh/£ > h/p. This poitjolio/productivit} con(dition has important implications for analyz- ing farmers' demand for land. It states that financial balance (B) has its own impact on demand for land (H) and that this impact (dH/dB) varies with a farmer's productivity and portfolio. If farmers are insolvent (dB > 0), they are net borrowers, net disinvestors, or both. If productivity keeps up with land prices (Qh/N > h/p), the balance effect will be negative (dH/dB < 0). If productivity does not keep up with land prices (Qh/y < h/p). the balance effect will be positive (dH/dB > 0). If farmers are solvent (dB < 0), they are net savers, net repayers, or both. If productivitv keeps up with land prices (Qh/C > h/p), the balance effect will Appendi.v C 147 be positive (dH/dB > 0). If procluctivitv does not keep up with land prices (Qh/t < h/p), the balance effect will be negative (dH/dB < 0). Table C. I presents a summary of these cases. Table C.1 Dynamic Income Effects: Productivity and Portfolio Poofio*t Product,zvty Solvent: dB < 0 Insolvent: dB > 0 Productive: Qh > h/p dH/dB > 0 dH/dB < 0 Unproductive: Qh < h/p dE-/dB < 0 dH/dB > 0 C(ase 1. Among solvent farmers (dB < 0) whose productivity is keeping up with real land prices (Qh > h/p). the dynamic effect should be positive, adding to the static income effect. The total income effect should be relatively large and positive. Cose 2. Among solvent farmers (dB < 0) whose productivity is not keeping up with real land prices (Qh < h/p), the dynamic effect should be negative, reducing the static income effect. The total income effect may be positive, though relatively small, or even negative. Increasing incomes are directed out of agriculture and land is held as a store of value, rather than as a factor of production. Case 3. Among insolvent farmers (dB > 0) whose productivity is keeping up with real land prices (Qh > h/p), the dynamic effect should be negative, reducing the static income effect. The total income effect may be positive, though relatively small, or even negative. Agricultural product is being used to pay off debt, so increasing incomes are directed out of agricultuire. Land is held as a store of value rather than as a factor of production. Cnase 4. Among insolvent farmers (dB > 0) w%hose productivitv is not keeping up with real land prices (Qh < h/p), the dynamic effect should be positive, adding to the static income effect. The total income effect will be relatively large and positive. This is an example of debt peonage, where farmers over- work their own families and land to pay off debt. C. 3.2 Dynamic price effects The effect of own-price variation on demand for land is expected to contain two components: a negative substitution effect (dH/d[h/p] < 0) and an income effect, which may be static or dynamic. The static income effect (dH/dY > 0) is expected to be positive. The dynamic effect may be larger or smaller than the static income effect (dHI/dTY < dH/dY) according to 148 Suistainable Settlemenzt in the Brazilian Amazon portfolio and productivity (see table C.1). The total income effect can be negative under extreme conditions (special cases of 2 and 3, above). If the income effect is positive, the total own-price effect should be negative-as prices rise, demand falls. But if the income effect is negative, it dampens the price effect and mav provoke a positive own-price effect. A positive own-price effect on the demand for land would arise if productive farmers become increasingly insolvent or if the productivity of solvent farmers does not keep up with rising land prices. The overall effect of land price on demand for land cannot be predicted. A negative overall own-price effect may be a solvent/productive (case 1) or an insolvent/unproductive (case 4) response to land price. Either can be called a produictive response to land price variation. A perverse, positive, overall own- price effect may be an insolvent/productive (case 2) or a solvent/unproductive (case 3) response to land price. Either of the twvo mav be called a speculative response to land price variation. C.4 Empirical Equations The empirical equations derived from the model regress total deforested area (F) against real prices for land (h/p), labor (w/p), other costs (c/p), and an income variable. This income variable is specified differentlv according to each version of the model. The static model uses current income (Y) or its income components-agricultural (A), nonagricultural (N), or subsis- tence (S). The dynamic model uses total income (TY) or total income components: net indebtedness (D) and net investment (I). The four regressions (estimated in appendix D) are derived from equa- tions [12] and [16], as follows: Static model: [1] F=ao+aa.h/p+a22w/p+a,c/p +a4Y+v. 2] F = a + a1 h/p + a2' w/p a3' c/p+ a4'Aa'N a6'S+ v'. Dynamic model: 3] F = bo + b1 h/p + b2.w/p + b3.c/p + b4TY + u. [4] F = b0' b'Ih/p + bw/p+ b3'c/p + b4'Y+ b'B +u' where ao and bo are intercepts; ai (j = 1,...,4), ai'(j = 1,...,6), bi (j = 1,...,4), and b,' (j = 1,...,5) are coefficients; and v, v', u, and u' are the error terms to be estimated. Appendix C 149 If the income effect is positive, income coefficient estimates are expected to be positive: Static model: [1] a4>O. [2] a,' > 0 (j = 4,5,6). Dynamic model: [3] b4 >O [4] b'> 0 (j=4,5). The total own-price effect on demand for land should be negative, and all land-price coefficient estimates are expected to be negative: Static model: [1] a, < 0. [2] a' < 0. Dynamic model: [3] b1 < 0. [4] b '<0. If the income effect is negative, it may provoke a positive own-price effect on the demand for land. Expected coefficient estimates then are opposite in sign from those above. According to table C.l, this would be the case if productive farmers become increasingly indebted or if the productivity of solvent farmers does not keep up with rising land prices. The overall effect of land price variation on demand for land cannot be predicted. A negative overall own-price effect mav be called a productive response to land price variation. A positive overall own-price effect may be called a speculative response to land price variation. C.5 Summary and Policy Implications The previous sections propose that frontier farmers deforest differently in response to changes in income and land prices according to whether they I SO Sut- tainiable Settlement in the Brazilian Amazon are productive in agriculture (productive response) or speculating in the land market (speculative response). A productive response occurs when farmers are solvent and land is produc- tive (table C. I, case 1) or when farmers are insolvent and land is unproduc- tive (table C. 1, case 4). In the first case, farmers with relativelv high rates of accumulation respond to rising incomes by increasing demand for land, and they respond to rising land prices by decreasing demand for land. In the second case, debt peonage occurs. Farmers also respond to rising incomes by increasing their demand for land. Their response to rising prices is a decreased demand for land. A specuiLative response occurs when farmers are solvent and land is unpro- ductive (table C.l, case 2) or when farmers are insolvent and land is productive (table C.l, case 3). In case 2, low yields drive farmers out of agriculture so that rising incomes may decrease demand for land while rising land prices may increase demand for land. In case 3, it is debt that drives farmers out of agriculture, but the effects of rising incomes on demand for land are the same as in case 2. In agricultural frontier economies, deforestation can be taken as a form of demand for land. Such demand may be for productive purposes-which would imply productive deforestation-or for speculative purposes, which would imply speculative deforestation. Farmers who are deforesting for the purpose of agricultural production are sensitive to price and income vari- ations. On the other hand, farmers wvho deforest in order to hold land as a store of value respond inverselv to variations in prices and incomes. Eco- nomic policies intended for agriculturally productive farmers would onlv reinforce speculative fanners' motives for holding land. Policies aimed at raising the level of sustainability of small Amazonian farming must, there- fore, determine which motive-that is, production or speculation-prevails in a specific frontier. This model's empirical equations are estimated according to the statistical procedures outlined in appendix D. Notes 1. This procedure disregards a vast discussion in the literature on how to analvze the cost of household labor in family farming or peasant agriculture, referred to in FAG/fN1CRA (1992). 2. See Oz6rio de Almeida (1991b) for literature on, and modeling of, debt peonage in the Amazon frontier. 3. The rate of accumulation, time on plot, and initial capital are included in the econometric model tested in appendix D. 4. Chapter 4 confirmed these predictions by showing that the frontier locations with the highest rates of accumulation were those where agriculture was most productive and where agricultural income accounted for the highest shares of total income. APPENDIX D Econometric Results Estimation of the Model Theoretical zvariables Given the data base described in appendix B, translating the theoretical model in appendix C into quantifiable variables requires adjustments. The principal variables were observed in many different ways, and more empiri- cal tests were performed than reported. Deforestation was observed since farmers' arrival (F) and during the current agricultural year (F9 1).' Income was measured as current total gross income (GY) or its components-gross agricultural (A), net nonagricultural (N), and net subsistence (S) incomes. Net income (Y) is gross income minus current agricultural production expenditures (E). Total income (TY) is gross income (GY) plus balance on financial account (B), which is composed of net indebtedness (D) minus net investment (1). Price of land (h/p) is measured as a weighted average of the value of areas-forested, temporary crop, perennial crop, pasture, fallow, and unproductive. Real wages (w) are measured as an average of daily wages paid to outside workers. Other costs-total (C) and per unit (c/p)-were omitted from the regressions because the composition of inputs varied too much by location and by farmer to permit a viable index.2 The output price was the average sale price of a kilogram of rice.3 Theoretical equations The model's four empirical equations lead to eight regressions: 151 1 52 Suestainable Settlement in the Briaziliani Amazon Reg7ession: 1 2 3 4 Empiial eqNationi: [12] [13] 1 15] [16] Dependent variables: F, F91 Constants: ao ao bo bo' Error terms: v v u u Independf7et variables Coefficients Prices: h/p,w/p (i = 1,2) aj ai' bi bi' Income: GY a4 A, N, S ai (i = 4,5,6) Total rY income b4 GY, B (i = 4,5) bi' For each dependent variable (F, F91), regression I estimates demand for deforestation based on current income (Y) and real price effects. Regression 2 distinguishes agricultural (A) and other activities (N,S) in the current income effect. Regression 3 estimates demand for deforestation based on the total income effect (TY) and price effects. Regression 4 distinguishes indebtedness and investment (B = D - I) in the total income effect. Shifter variables If the model was perfectly specified, the above relationships would account for variations in deforestation demand (F, F91) among frontier farmers. But there are many influences on migrant farmer behavior-based on origin, destination, and individual characteristics-that have not been taken into account in this model. Given the lack of information on Amazon coloniza- tion, it is important to explore a wider spectrum of possibilities than can be dealt with in a rmicroeconomic hypothesis. To bring the model into the Amazonian context, several variables were added to regressions I to 4. Listed in appendix B, these variables are grouped according to the origin, destination, and individual characteristics of frontier farmers. Origin characteristics relate to the past: where a farmer comes from (ORIG); whether he was a landowner before (FORM); how itinerant he was (ITINT); whether his parents were farmers (PARNT). Destination characteristics pertain to conditions since arrival at the fron- tier, such as: the kind of a project the farmer participates in: 0 = public (Para); I = private (Mato Grosso) (STAT). Appendix D 153 marketing conditions: distance to market (DIST),4 how much the farmer sells at the farm gate (GATE), and how much he sells during the three months after harvest (STOR). the productivity of the land (QH). tenure: squatter (SQUAT), sharecropper (SCROP), owner (OWN), or titled owner (TD). access to credit (CREDIT) and participation in local institutions-rural extension, cooperatives, unions, associations, and church (mITST). Individual characteristics may be personal and family characteristics, or they may indicate attitudes, perceptions, and expectations. Among the variables tested are: age (AGE), time on plot (TIME), number of family workers (wVKRS), practice of crop or area rotation (AGCON), perception that loss of soil fertility may be a problem now or in the future (FERT), belief that living standard has improved since arrival (iMPR), intention to remain on plot (FIX), and plans to invest in agriculture (PLAN). These variables contribute to a farmer's performance. Specifically, a farmer's deforestation responses to price and income variations may shift according to scales of origin, destination, or individuality. If so, controlling these shifts will improve estimation of the productive deforestation model. Given the complexity of how such influences affect behavior, they are investigated in a preliminary manner, as indicated below. Empirical results for the deforestation model are summarized in tables D. 1 and D.2. These tables do not present the estimates themselves (which are in tables D.3 to D. 0)-only the signs of estimates of income and price coefficients. The tables also show the percentage contribution and significance of four estimation steps that reduce residual variation in deforestation. Step I is the deforestation model, step 2 is the origin variables, step 3 is the destination variables, and step 4 is the individual variables. A letter exponent indicates the significance of each estimate according to the F-test; if there is no letter exponent, the estimate was not significant at the 10 percent level. Tables D.3 to D.10 are final (step 4) regressions, with all shifter variables included for each dependent variable: F: deforestation since arrival (tables D.3 to D.6); F91: deforestation during 1991 (tables D.7 to D.10). Analysis of Cross-section Data Ordinary least squares regressions for deforestation since arrival (F) and during 1991 (F91) are estimated in the following sequence: 1 54 Sus-tainable Settlement in the Brazilian Amnazon Step I estimates coefficients for regressions I to 4 (empirical equations [12], [133, [15], and [16]) and tests for the significance of each independent variable. This step also calculates the significance of the deforestation model in reducing the residual variance of dependent variables. All the regressions are included in step 1. Step 2 also estimates coefficients for variables in step 1, adding origin variables to the regressions. It tests the significance of each origin variable and the joint contribution of the origin variables in reducing residual variance. Step 3 estimates coefficients for the variables in step 2, adding destination variables to the regressions. It tests the significance of each destination variable and the joint contribution of the destination variables in reducing residual variance. Step 4 estimates coefficients for the variables in step 3, adding individual variables to the regressions. It tests the significance of each individual variable and the joint contribution of the individual variables in reducing residual variance. There are thirty-two regressions-eight in each of the four steps. This procedure controls which variables enter each regression and does not leave the statistical package to determine which variables will be ana- lyzed. The part of the analysis that corresponds to the theoretical model is in step 1. The part of the analvsis that is not based on microeconomic modeling is in steps 2, 3, and 4. Steps 2, 3, and 4 introduce shifter- variables, which change the model's n-space, correcting distortions and omissions not adequately predicted in the theory. In this way, one is rigorous where allowed for and adventurous where possible. These procedures contain several limitations. No attempt is made to specify the functional form in which the shifter variables enter the equations, linearity being assumed throughout. Steps 2 to 4 are prone to heteroschedas- ticity, colinearity, excluded variable bias, wrong-specification bias, errors in variables, simultaneity, problems in the use of dummy variables, and the difficulties of regressions based on cross sections.' These problems are caused by the exploratory nature of the exercise and the need to attain a set of descriptive results before investing in more sophisticated modeling and statistical analysis. Empirical Results This section analyzes regression results for deforestation (F). These regres- sions correspond to step I (model), step 2 (model + origin), step 3 (model + origin + destination), and step 4 (model + origin + destination + individual variables). Step 4 coefficients have signs that are closer to theoretical expectations than step I coefficients but tend to display less statistical significance than those of step 1. Appendix D 1 S Deforestation responds to a large number of influences not accounted for in any single microeconomic model. Excluded variable problems tend to bias theoretical coefficient estimates. Because the signs of these coefficients are crucial to interpreting empirical results, excluded variables are an im- portant problem in the model. When the inclusion of a variable changes coefficient estimates, this indicates that previous estimates were biased by the omission of important variables. The inclusion is beneficial to the interpretation of results. When the inclusion of a variable does not alter coefficient estimates, estimates of standard errors become upwardly biased. This reduces the significance of theoretical variables. Prior modeling provides expectations regarding the signs of theoretical coefficients. Thus, to draw implications, maximizing statistical significance is less important than minimizing coefficient biases. In other words, the cost of exclusion is high in terms of biased coefficient estimates; the cost of inclusion is low in terms of statistical significance. In this case, too many variables are better than too few. Two other costs of inclusion should be mentioned. First, missing values are different across observations for each variable; so the larger the number of variables analyzed, the smaller the portion of the total sample used in the regressions. This further reduces the statistical significance of the estimates. Second, simultaneities among many destination and individual variables (steps 3 and 4) and theoretical variables (step 1) provoke an upward bias in R2 estimates for steps 3 and 4 relative to steps I and 2. For expediency, shorthand terms are used to describe regression results. An estimate is significant if it passes the t-rest (for an indi'Vidual coefficient) or the F-test (for the group of coefficients estimated in a particular step) at the 10 percent level. The term "explanatory power" is used in reference to reductions in the sum of squared residuals. A coefficient is referred to as the impact of an independent variable on a dependent variable, though causality cannot be inferred from correlation. Other liberties are taken w ith statistical terminology to facilitate exposition. Because of space limitations, the coefficients estimated in steps 2 and 3 are omitted from tables D.3 to D.10. Only those obtained in steps I and 4 are presented. The analysis of covariance at the foot of each table indicates the statistical significance of each step (1, 2, 3, and 4) in reducing the residual variance in deforestation and displays their respective contributions to R2. Analysis of deforestation since arrival The economic model of deforestation (F-step 1) contributes more than half the reduction in residual variation of deforestation (table D.l). All income effects are significant and positive. Of these, agricultural income is the largest deforester (table D.4). Balance on financial account has a positive income effect (table D.6). This means that solvent, high-productivity farm- 156 Sustainable Settleme77t in the Braiziliani Amazon Table D.1 Deforestation Since Arrival Dependent variable: F Sta,i7 mlodiel Dyvnamic rmodel Reg e.nioi, / 2 4 Empirical equaition /12/ [131 [151 /161 Independent variables Signs of coefficients Prices h/p + + + + w/p + + + +c Income (GY + A + N + S + Total income Tn + GE' + B +d R2 Step 1: Model 0.42a 0.72a 0. 3a 0.61a Step 2: Origin 0. 1 3 0. 14a 0. 14a 014b Step 3: Destination 0.15b 0.03 0.04 0.07 Step 4: Individual 0.01 0.01 0.01 0.06 Total 0.71 0.90 0.72 0.81 Statistical significance tests: a-I percent; b-5 percent; c-10 percent.The absence of a superscript indicates that the variable is not significant at the 10 percent level. ers are deforesting to invest in agriculture, while insolvent, low-productivity farmers are deforesting to repay debt (table D. I). Cross-price effects are positive and generally significant. Contrary to expectation, the effect of the price of land is positive and insignificant. This means that deforestation is not behaving according to a model demand for land. A positive own-price effect was expected only in the case of negative income effects. But all income effects turned out to be positive and signifi- cant for total deforestation (F). So the positive impact of land price on deforestation is inconsistent with a productive, current demand response. Farmers are deforesting not to increase agricultural production but to add to their net worth. Therefore, the higher the price of land, the more surrounding lots they purchase and deforest. The shifter variables are too colinear to attain statistical significance individually. In groups, only origin characteristics (step 2) have a consider- able and significant contribution-about 14 percent-to reducing residual variance in deforestation. Of these, prior itinerancy ([TIN) has the most positive effect. Although destination variables are generallv insignificant, distance to market (DIST), access to credit (CREDIT), and institutional participation (INST) seem to have positive impacts on deforestation. Appendix D 157 Table D.2 Deforestation in 1991 Dependent variable: F91 Static Nlode DyrnmZic A -odel Reression 1 2i 4 Empirical equation /121 /1 11 [/1 116] Independent variables Signs of coefficients Prices h/p + + + w/p Income (;y A +a N Total income TY GY B R2 Step 1: Model 0.33a 0.730 0.3 1 0.3 5.1 Step 2: Origin 0. 14 0. 14 0. 14a 0.15b Step 3: Destinaton 0.24' 0.04c 0.27a 0.23' Step 4: Individual 0.01 0.01 0.01 0.01 Total: 0.74 0.92 0.73 0.74 Statistical significance tests: a-I percent; b-. percent; c-l0 percent. The absence of a superscript indicates that the variable is not significant at the 10 percent level. Table D.2 indicates that the economic model (step I) contributed be- tween one-third and two-thirds of the residual variance reduction in defor- estation in current year (1991). Agriculture is the most significant deforester-its separation from total income raises the explanatory power of the regression considerably (table D.8), as it did in the case of total deforestation (F). Except for the impact of agricultural income (A), which continues to be positive, all other coefficients change signs. These regres- sions indicate an important difference in the economic determination of current deforestation (F91) compared with deforestation since arrival (F). Income effects are negative but generally insignificant. This is the oppo- site of the finding regarding deforestation since arrival (F) (tables D.3 to D.6), where all income effects were positive and significant. The impact of balance on financial account also switches sign, from positive and significant to negative and insignificant (regression 4, table D.1O). This implies that insolvent, high-productivity farmers have large debt burdens and, therefore. have insufficient funds to invest in agriculture. Solvent, low-productivity farmers are diversifying and investing outside of agriculture. Cross-price effects also switch signs relative to regressions reported in table D.l, becoming negative though insignificant. The effect of land I 58 Sutstainable Settlement in the Brazilian Amazon pricing, however, is positive and becomes generally significant. Given the negative income effects for F91, this positive own-price response for F91 is consistent with the economic model of deforestation. It implies that farmers are moving out of agriculture and that a reduction in land price wvill speed the movement away from agricultural production. This agricultural involu- tion is compounded bv dynamic income effects (regressions 3 and 4, tables D.9 and D.1O), which are also negative. Thus, farmers who are investing and those that are servicing large debts will move out of agriculture even more quickly when land prices fall. As before, shifter variables are too colinear to attain statistical significance individually. However, their group contribution is significant in reducing the unexplained variance in current deforestation. After the modeling vari- ables (step 1), destination variables (step 3) are the most important group, contributing about one-quarter of the explanatorv power of the regressions. Of these, access to credit is the largest and most significant deforester, followed by participation in local institutions. Being in a public project, selling at the farm gate, and lacking storage have significantly negative impacts. These are probablv capturing disguised income effects, since poorer farmers deforest less. The other two groups of variables-origin and individual-do not attain statistical significance individually. Origin variables are significant as a group and contribute about the same-14 percent-to explaining current deforestation (F91) as they did to deforestation since arrival (F). Thus, total deforestation responds more to origin influences, while current deforesta- tion responds more to destination conditions. Over time, the impact of the past weakens and that of the present grows. It may take time to influence colonists' behavior. Policies do influence migrants, but colonists, particu- larly recent arrivals, may be insensitive to current stimuli and react to past experiences.6 Appendix D 159 Regression Results Table D.3 Regression 1: Deforestation (F), Static Model, Aggregate Income Parameter Standard Variable estimate error F Prob > F Intercept 1,991.899 5,286.368 0.14 0.708 Group: Model and origin 8.06 0.000 GY 0.000 0.000 27.65 0.000 h/p 0.113 0.225 0.26 0.615 w/p 1.811 0.767 5.57 0.023 PARNT 29.266 81.761 0.13 0.722 rrIN 23.109 12.952 3.18 0.082 FORM 1.275 53.375 0.00 0.981 ORIG 1.233 62.371 0.00 0.984 Group: Destination 2.28 0.047 STAT -39.397 86.027 0.21 0.649 DIST3 1.753 0.964 3.30 0.076 QH 0.000 0.000 0.00 0.993 STOR 31.403 68.104 0.21 0.647 GATE -95.562 63.069 2.30 0.137 INST 46.339 22.295 4.32 0.044 CREDIT 214.816 118.185 3.30 0.076 Group: Individual 0.23 0.963 TIMIE 0.060 0.509 0.01 0.905 K0 0.000 0.001 0.10 0.748 AGE -1.663 2.046 0.66 0.421 TD -49.337 86.323 0.33 0.570 AGCON 25.591 82.459 0.10 0.757 FERT -34.276 57.032 0.36 0.551 Sununary of Stepwise Procedure for Dependent Variable F Partial Model Step Group entered Number R2 R2 C(p) F Prob > F I Model 3 0.4200 2 Origin 7 0.1300 0.5504 18.0856 9.27 0.001 3 Destination 14 0.1546 0.7050 10.3902 3.4445 0.0048 4 Individual 20 0.0099 0.7149 21.0000 0.2317 0.9638 160 Sstdainable Settlement in the Brazziliain Amazon Table D.4 Regression 2: Deforestation (F), Static Model, Disaggregated Income Parameter Standard I-ariable efti,mate er; or F Pr-ob > F Intercept -24.088 3,446.480 0.0( 0.994 Group: Model and origin 23.85 0.000 A 0.006 0.001 15.19 0.000 N 0.(00 0.000 10.00 0.003 s 0.001 0.001 1.32 0.257 h/p 0.059 0.155 0.14 0.706 w/p 0.237 0.510 0.22 0.644 PARNT 9.871 50.617 0.04 0.846 ITtN 4.737 8.318 0.32 0. 572 FORM 4.977 33.024 0.02 0.881 ORIG 26.158 39.162 0.45 0.508 Group: Destination 0.99 0.454 STAT -8.753 56.442 0.02 0.877 DIST3 0.982 0.603 2.65 0.112 OH 0.000 0.000 0.41 0.526 STOR -2.935 42.348 0.00 0.945 GATE -36.915 40.125 0.85 0.363 INST 16.501 14.839 1.25 0.270 CREDIT -61.426 125.582 0.24 0.627 Group: Individual 0.86 0.529 TLNIE -0.071 0.314 0.05 0.822 K0 0.001 0.000 1.91 0.175 AGE -0.070 1.295 0.00 0.956 TD -93.889 53.638 3.06 0.088 AGCON 5.453 51.057 0.01 0.915 FERT -3.651 35.458 0.01 0.918 Summary of Stepwise Procedure for Dependent Variable F PNrtial .lodel Step Group entered Nzumber R- C(p) F Prob > F I Model 5 0.7200 2 Origin 9 0.1400 0.8487 14.6066 31.81 0.0001 3 Destination 16 0.0338 0.8826 16.1840 1.8081 0.1097 4 Individual 22 0.0141 0.8967 23.0000 0.8641 0.5299 Appendix D 161 Table D.5 Regression 3: Deforestation (F), Dynamic Model, Aggregate Income Parameter Standard VZariable estimate errar F Piaob > F Intercept -4,873.952 5,242.006 0.86 0.358 Group: Model and origin 0.52 0.000 T'Y 0.002 0.000 29.95 0.000 h/p 0.034 0.222 (.02 0.876 w/p 1.567 0.766 4.18 0.047 P.RNT -43.633 80.875 0.29 0.592 ITIN 16.401 12.943 1.61 0.212 FORA;I 30.414 51.822 0.34 0.560 ORIG 58.381 62.305 0.88 0.354 Group: Destination 0.67 0.697 STAT 72.588 84.933 0.73 0.397 DIST3 -0.429 1.050 0.17 0.684 QH 0.000 0.000 0.26 0.614 STOR 39.459 66.834 0.35 0.558 GATE -25.038 62.353 0.16 0.690 INST 10.701 22.586 0.22 0.638 CREDIT 83.746 118.223 0.50 0.482 Group: Individual 0.32 0.923 TILME -0.212 0.503 0.18 0.675 K 0 0.000 0.001 0.08 0.775 AGE 0.145 2.056 0.01 0.943 TD -49.485 84.893 0.34 0.563 AGCON -34.712 81.824 0.18 0.673 FERT -30.324 55.940 0.47 0.497 Sunnnary of Stepwise Procedure for Dependent Variable F Partial MXiodel Step Group entered Number R2 R C(p) F Piob > F I Model 3 0.5300 2 Origin 7 0.1400 0.6627 3.9311 14.04 0.001 3 Destination 14 0.0484 0.7111 10.9084 1.1013 0.3784 4 Individual 20 0.0132 0.7243 21.0000 0.3181 0.9237 162 Sustainable Settlement in the Brazilian Amazon Table D.6 Regression 4: Deforestation (F), Dynamic Model, Disaggregated Income Parameter SWl,zdal-rd Variable e."timate elrror F Pr'ob > F Intercept -1,950.979 4,495.922 0.19 0.666 Group: Model and origin 12.50 0.000 GY 0.001 0.000 30.71 0.000 B 0.001 0.000 18.68 0.000 h/p 0.015 0.188 0.01 0.933 w/p 1.180 0.655 3.24 0.079 PARNT -14.772 68.847 0.05 0.831 rEIN 14.438 10.971 1.73 0.195 FORM 1.426 44.449 0.00 0.974 ORTG 44.032 52.877 0.69 0.410 Group: Destination 1.12 0.371 STAT 21.277 73.004 0.08 0.772 DIS13 0.032 0.896 0.00 0.971 QH 0.000 0.000 0.17 0.682 STOR 21.823 56.759 0.15 0.702 GATE -5 5.666 53.327 1.09 0.303 INST 22.763 19.352 1.38 0.246 CREDIT 124.658 100.608 1.54 0.222 Group: Individual 0.20 0.975 T I E -0.141 0.426 0.11 0.741 K 0 0.000 0.001 0.08 0.775 AGE -0.095 1.742 0.00 0.956 TD -55.937 71.905 0.61 0.441 AGCON -18.032 69.408 0.07 0.796 FERT -15.948 47.684 0.11 0.739 Summary of Stepwise Procedure for Dependent Variable F Partial AlModel Step Group entered .Numd be R R C(p) F Prob > F I Model 4 0.6100 2 Origin 8 0.1400 0.7365 10.2986 12.50 0.0001 3 Destination 15 0.0648 0.8014 11.1854 2.0978 0.0632 4 Individual 21 0.0059 0.8072 22.0000 0.1976 0.9756 Appendix D 163 Table D.7 Regression 1: Deforestation (F91), Static Model, Aggregate Income Pa-rameter Standaird Variable eatlmate error- F Pr-ob > F Intercept 11,347.177 4,302.804 6.95 0.013 Group: Model and origin 3.88 0.004 GY 0.000 0.000 0.28 0.602 hIp 0.869 0.172 25.54 0.000 xv/p -0.594 0.546 1.18 0.285 PARNT -10.427 58.335 0.03 0.859 ITIN 3.296 9.809 0.11 0.739 FORiM 27.314 47.306 0.33 0.568 ORIG -12.588 45.054 0.08 0.781 Group: Destination 3.96 0.003 STAT -218.502 72.523 93.08 0.005 DIST3 0.282 0.695 0.17 0.687 QH 0.000 0.000 0.2 5 0.620 STOR -66.228 55.975 1.40 0.246 GATE -93.543 51.582 3.29 0.080 7NST 36.592 16.271 5.06 0.032 CREDIT 275.072 83.574 10.83 0.002 Group: Individual 0.21 0.971 TLME 0.227 0.362 0.39 0.534 K 0 0.000 0.001 0.12 0.729 AGE 0.263 1.552 0.03 0.866 TD 19.559 67.813 0.08 0.775 AGCON 47.100 73.834 0.41 0.528 FERT -7.827 46.002 0.03 0.866 Summary of Stepwise Procedure for Dependent Variable F91 Partiail Model Step Group enttered Number R2 C(p() F Prob > F I Model 3 0.3300 2 Origin 7 0.1400 0.4799 24.0960 5.54 0.0001 3 Destination 14 0.2493 0.7293 10.2422 4.6051 0.0010 4 Individual 20 0.0111 0.7404 21.0000 0.2070 0.9718 164 Sustainable Settlemeent in tbe Brazilian Amazon Table D.8 Regression 2: Deforestation (F91), Static Model, Disaggregated Income Pir-ameter Stanldar-d 1 air-iabl'e entimate error F Pr-ob > F Intercept 777.196 2,893.684 0.07 0.790 Group: Model and origin 15.96 0.000 A 0.007 0.001 48.54 0.000 N 0.000 0.00( 0.20 0.656 s -0.008 0.001 61.62 0.000 h/p 0.187 0.135 1.93 0.176 W/p -0.099 0.324 0.09 0.7,61 PARNT 22.277 34.008 0.43 0.518 IT\IN1 -1.298 6.017 0.05 0.830 FORM 23.215 27.193 0.73 0.400 ORIG 29.589 26.930 1.21 0.281 Group: Destination 1.51 0.205 STAT -30.018 49.274 0.37 0.547 DIST3 0.404 0.399 1.02 0.320 QH 0.000 0.000 0.32 0.574 STOR -12.483 32.907 0.14 0.707 GATE -0.807 34.682 0.00 0.981 INST 8.502 10.445 0.66 0.422 CREDIT -130.716 71.716 3.32 0.079 Group: Individual 0.43 0.854 TILIE 0.077 0.210 0.13 0.717 K 0 0.000 0.000 0.94 0.340 AGE 0.597 0.954 0.39 0.536 TD -9.743 39.820 0.06 0.808 AGCON 13.128 43.061 0.09 0.762 FERT 27.126 27.668 0.96 0.335 Summary of Stepwise Procedure for Dependent Variable F91 Partial Model Step Group e7itered Number R2 R- C(p) F Prob > F 1 Model 5 0.7300 2 Origin 9 0.1400 0.8724 13.3096 30.3969 0.0001 3 Destination 16 0.0405 0.9129 13.5574 2.1934 0.0606 4 Individual 22 0.0075 0.9205 23.0000 0.4262 0.8549 Appendix D 165 Table D.9 Regression 3: Deforestation (F91), Dynamic Model, Aggregate Income Parameter Standard Variable estAmate error F Prolb > F Intercept 12,117.644 4,287.165 7.99 0.008 Group: Model and origin 3.86 0.004 TY 0.000 0.000 0.20 0.661 h/p 0.873 0.174 24.99 0.000 Nv/p -0.612 0.544 1.27 0.269 PARNT -4.721 58.735 0.01 0.936 ITIN 3.301 9.898 0.11 0.741 FORMI 20.524 44.255 0.22 0.646 ORIG -17.952 46.035 0.15 0.699 Group: Destination 4.30 0.002 STAT -231.297 71.788 10.38 0.003 DIST3 0.391 0.762 0.26 0.611 QH 0.000 0.000 0.34 0.566 STOR -66.969 55.994 1.43 0.241 GATE -101.131 51.515 3.85 0.059 IN'ST 39.611 16.803 5.56 0.025 CREDIT 281.581 84.341 11.15 0.002 Group: Individual 0.23 0.961 TLME 0.244 0.367 0.44 0.510 K0 0.000 0.001 0.16 0.696 AGE 0.115 1.615 0.01 0.943 TD 17.772 67.728 0.07 0.794 AGCON 56.544 74.323 0.58 0.452 FERT -11.055 46.349 0.06 0.813 Sunmnary of Stepwise Procedure for Dependent Variable F91 Partial .1odel Step Grotup entered Nutmber ' R2 C(p) F Prol! > F I Model 3 0.3100 2 Origin 7 0.1400 0.4535 26.8697 4.98 0.0004 3 Destination 14 0.2735 0.7271 10.4037 5.0103 0.0005 4 Individual 20 0.0126 0.7379 21.0000 0.2339 0.9619 166 Stustainable Settlement in the Brazilian Amazon Table D.10 Regression 4: Deforestation (F91), Dynamic Model, Disaggregated Income Paramete7- Standard ["ariable estimate enro1 F Pr-o' > F Intercept 11,645.663 4,525.022 6.62 0.015 Group: Model and origin 3.29 0.008 GY 0.000 0.000 0.14 0.71 5 B 0.000 0.000 0.07 0.798 h/p 0.877 0.177 24.39 0.000 w/p -0.570 0.563 1.03 0.319 PARNT -7.803 60.168 0.02 0.897 ITrN 3.723 10.108 0.14 0.715 FORNl 27.065 48.096 0.332 0.578 ORIG -15.497 47.172 0.11 0.745 Group: Destination 3.63 0.006 STAT -223.179 75.927 8.64 (0.006 DIST3 0.365 0.777 0.22 0.641 QH 0.000 0.000 0.27 0.605 STOR -65.471 56.974 1.32 0.260 GATE -96.508 53.684 3.23 0.083 N'ST 38.082 17.523 4.72 0.038 CREDIT 278.504 85.994 10.49 0.003 Group: Individual 0.21 0.972 TL\IE 0.242 0.372 0.42 0.520 K 0.)000 0.001 0.12 0.735 AGE 0.147 1.641 0.01 0.929 TD 19.690 68.934 0.08 0.777 AGCON 51.210 76.734 0.45 0.510 FERT -9.518 47.221 0.04 0.841 Summary of Stepwise Procedure for Dependent Variable F91 Par,7tial i1lodel Step Grouip entered Nuimber- R- R- C (p) F Prob > F I Model 4 0.3500 2 Origin 8 0.1500 0.4948 22.6148 5.02 0.0002 3 Destination 15 0.2348 0.7296 11.2325 4.2174 0.0019 4 Indiridual 21 0.0114 0.7410 22.0000 0.2054 0.9722 Appendix D 167 Notes 1. Many more dependenit variables than those reported here were tested: defor- estation during first three years (F3); demand for total land (H); demand for harvested area (FA); time on plot (T); survival from the first survey (in 1981) to the second (in 1991) on the same plot of land (SURV), among other things. Time and resource constraints prevent presentation of the full set of estimates performed. The ones included summarize the main findings and contain the most important implications. 2. An effort was made to include one or two representative cost items, such as fuel. but the cost of doing so, in terms of the number of observations sacrificed, was too great relative to the benefit of including an additional variable in the regressions. 3. Rice is the only crop produced by all farmers in all parts of the frontier. So it was the only possible numeraire. See Oz6rio de Almeida (1992b, chapter 17) for further evidence of its predominance in the Amazon frontier. 4. Three different measurements of distance were taken (see appendix B). Travel time from local center to plot (DisT3) appeared to be the most sensitive to the estimation procedure. 5. See Oz6rio de Almeida (1992, chapter 20) for a use of similar methodology and a discussion of the relevant econometric literature. 6. A similar finding is reported in Oz6rio de Almeida (1992b, chapter 21). Bibliography Ahmad, Yusuf J., Salah El Serafy, and Ernst Lutz (eds.). 1989. Environmental Accotuntinlgfob Sistainable Development. Washington, D.C.: World Bank. Aklwabi-Amevaw. 1990. "The Political Economy of Agricultural Resettlement and Rural Developmenit in Zimbabwe: The Performance of Family Farms and Pro- ducer Cooperatives." Human Organization 49(4): 320-338. Alicbusan-Schwab, Adelaida, andJohn F. Wilson. 1991. "Development Policies and Health: Farmers, Goldminers. and Slums in the Brazilian Amazon." Environment Department Working Paper 1991-18. World Bank. Washington, D.C. Alston, Lee J., Gary Libecap, and Robert Schneider. Forthcoming. "The Determi- nants and Impact of Propertv Rights: Land Titles on the Brazilian Frontier." .7ornwal of Law, Economics, annd Organization. Anderson, Anthony B. 1987. "Forest Management Issues in the Brazilian Amazon." Department of Botany, Museu Paraense EmTlio Goeldi, Para, Brazil. ed. 1990. Alternatives to DefJrestation: Steps Toward Sustainabile Use of the Amazon Rain Forest. New York: Columbia University Press. Anderson, Anthony, and Darrel A. Posey. 1989. "Reflorestamento Indigena." Ciincia Hoje 6(31): 29-50. Anderson, Jock R. 1992. "Difficulties in African Agricultural Systems Enhance- ment-Ten Hypotheses." Agr7iultur-alSystetms, 38: 387-409. Arrhenius, Eric A., and Thomas WV. Waltz. 1990. "The Greenhouse Effect: Impli- cations for Economic Development." Environment Department Discussion Pa- per 78. World Bank, Washington, D.C. Ascher, William, and Robert Healy. 1990. Natural Resource Poligcmaking in Develop- ing Countries. Durham, North Carolina: Duke University Press. Aubertin, C. 1990. "Population Movements and Economic Change in Central Western Brazil." Cahie7s Orstom, Serie Sciences Humanes 26(3): 327-42. Bacha, Edmar, and Herbert Klein, eds. 1986. A Transi(do Incompleta: Brasil desde 1945, Vols. I and 2. Sao Paulo: Paz e Terra. Bakx, Keith. 1987. "Planning Agrarian Reform: Amazonian Settlement Projects, 1970-86." Development and Change 18(4): 533-555. Bank, G., and K. Kooning, eds. 1988. Social Change in Contempora7y Brazil. Amster- dam: CEDLA. Barbier, Edward B., Joanne C. Burgess, and others. 1991. "The Economics of Tropical Deforestation." AA1BlO 20(2): 55-58. 169 1 70 Sustainable Settlement in the Brazilian Amazon Barbira-Schazzochio, Francoise, ed. 1980. Land, People and Plannin7g in Contemporaiy Amazonia. Occasional Publication 3. Cambridge, U.K.: Centre for Latin Ameri- can Studies, Cambridge University. Bardhan, Pranab K. 1984. Land, Labo;, and Rutral Povert0y: Esavs in Development Economics. New York: Columbia University Press. Barnes, Douglas F., andJose Olivares. 1988. "Sustainable Resource Management in Agriculture and Rural Development Projects: A Review of Bank Policies, Proce- dures, and Results." Environment Department Working Paper 5. World Bank, WVashington, D.C. Baumol, William, and Wallace Oates. 1988. The Theoiyof Enzviro07nmental Policy. New York: Cambridge University Press. Beckerman, Wilfred. 1992. "Economic Growth and the Environment: lVhose Growth? Whose Environment?" WVorld Development 20(4): 481-496. Besley, T. 1993. Propety Rights and Investment Incentives: Tbeoiy and Micro-evidence from Ghana. Discussion Paper 170. Research Program in Development Studies. Princeton, N.J.: Princeton University Press. Bhadra, Dipasis, and Ant6nio Salazar P. Brandao. 1993. Urbanization, Agricultural Development, and LandAllocation. World Bank Discussion Paper 201. Washington, D.C. Bhattarcharva, Rina. 1990. "Common Property Externalities: Isolation, Assurance and Resource Depletion in a Traditional Grazing Context." Environment De- partment Working Paper 1990-10. World Bank, Washington, D.C. Bigot, Yves, Hans Binswanger, and others. 1987. Ag7ricultural M11echanization and the Evolution of Fanming Systems in Sutb-Shaba-an Africa. Baltimore, Md.: Johns Hop- kins University Press. Binswanger, Hans. 1980. "Attitudes Toward Risk: Experimental Measurement in Rural India." American J'ornal oj Agr7icultural Economics (62)3: 395-407. - . 1994. "Brazilian Policies that Encourage Deforestation in the Amazon." 1World Development 19(7): 821-829. Binswager, Hans, andJohn Mclntire. 1987. "Behavioral and Material Determinants of Production Relations in Land-Abundant Tropical Agriculture." Economic De- velopment and Cultural Change 3 6(1): 7 3-99. Binswanger, Hans P., and Miranda Elgin. 1989. "Quais sao as Perspectivas para a Reforma Agraria?" Pesquisa e Planejlamento Econ6mico 19(1). Binswanger, Hans, Klaus Deininger, and Gershon Feder. 1993. "Power, Distor- tions, Revolt, and Reform in Agricultural Land Relations." In T.N. Srinivasan andJere Behrman, eds. Handbook of Development Economics, Vol. IIIB. Amsterdam: North Holland. Blount,jeb. 1993. "40 Yanomami Indians Slain On Brazilian Reservation." ffnash- ington Pot, August 21. Bonfim, Antulio N., and Anwar Shah. 1991. "Macroeconomic Management and the Division of Powers in Brazil: Perspectives for the Nineties." Working Paper 567. World Bank, Washington, D.C. Brandao, Ant6nio Salazar P., and Ger-vasio Castro de Rezende. 1992. "Credit Subsidies, Inflation and the Land Market in Brazil: A Theoretical and Empirical Analysis." Fundacao Getulio Vargas Graduate School of Economics, Rio de Janeiro. Bibliog7-aphy 171 Brito, Maria Socorro, and Mitiko Yanaga lUne. 1987. "A Evolucao da Agricultura Brasileira na Regiao Norte na Decada de 70." Rezista Brasileira de Geografia 49 (January/March) 11-46. Brito, Maristella de Azevedo. 1987. "Questoes Associadas a Evolucao Recente da Agricultura Brasileira," Revista Brasileira de Geog7afia 49 (January/March): 139- 161. Brokensha, David, and Alfonso Peter Castro. 1984. Fnelnwood, Agro-Foresny, and Natunral Resource Mlannagement-Utilization7 Syotems. New York: Institute for Devel- opment Anthropology. Brooks, E. 1974. "Frontiers of Ethnic Conflict in the Brazilian Amazon." Intelwa- tional journal of Environmental Studies 7: 63-74. Browder, John. 1988. "Public Policy and Deforestation in the Brazilian Amazon." In Robert Repetto and Malcolm Gillis. eds. Publilic Policies anid the Misuse of Forest Resources. Cambridge, U.K.: Cambridge Universitv Press. 1989. Fragile Lands of 'Latinz America: Strategies fo7 Suistainiable Development. Boulder, Colo.: Westview Press. Brown, Lawrence, and Douglas Southgate. 1991. "The Causes of Tropical Defor- estation in Ecuador." World Development 19(9): 1145-1151. Brown, Nigel. 1990. "Mercury Pollution xsith Specific Reference to the Amazon Basin." University of London. Bruneau, Thomas, and Rolf Wesche. 1990. Integration aid Change in Brazil's Middle Amazon. Ottawa: University of Ottawa Press. Bunker, Stephen. 1985. Underdeveloping the Amazoni: Eixtraction, Unequal Exchn7zge and the Failiure of'the MIodern State. Chicago, Ill.: University of Illinois Press. Campari, Joao. 1994. "Ill-Defined and Poorly Enforced Propertv Rights: A Role for Multilateral Organizations and Local Governments." University of Texas at Austin, Department of Economics, Austin, Texas. . 1993a. "Fiirst- and Second-Best Policies to Control Deforestation in the Brazilian Amazon." University of Texas at Austin, Department of Economics, Austin, Texas. 1993b. "Liquidity-Constrained Agents and the Sub-Optimal Allocation of Natural Resources." University of Texas at Austin, Department of Economics. Austin, Texas. Cano, Wilson. 1985. Desequildwrios Regiotnais e Concentra(io Inldulstrial no Brnasil, 1930-1970. Campinas: Ed. UNICANIP. de Carvalho, Magno, Jose Alberto, and Charles Wood. 1988. The Demog7aphy o' Inequality. Cambridge, U.K.: Cambridge University Press. Castro, Antonio. 1980. Ensaios soblre a Eco7tomia Brasileira. Rio de Janeiro: Forense- Universitiria. Castro de Rezende, Gervisio, and Antonio Salazar Brandao. 1992. "Credit Subsi- dies, Inflation and the Land Miarket in Brazil: A Theoretical and Empirical Analysis." Agricultural and Natural Resources Department, World Bank, Wash- ington, D.C. Cernea, Michael. 1988. Nongovernmental Organization5s and Local Development. World Bank Discussion Paper 40. Washington, DC. Chalamwong, Yongyuth, Gershon Feder, and others. 1988. Lanid Polic-ies and Fa7m Productivity in Thailanid. Baltimore, Md.: Johns Hopkins University Press. 172 Sustainable Settlement in the Brazilian Amazon Clearv, D. 1991. "The Brazilian Rainforest: Politics, Finance, Mining, and the Environment." Economist 2100: 66. Collins, Jane. 1981. "Kinship and Seasonal Migration Among the Aymara of Southern Peru: Human Adaptation to Energy Scarcitv." Unpublished disserta- tion. University of Florida, Gainesville. . 1986. "Smallholder Settlement of Tropical South America: The Social Causes of Ecological Destruction." Human Organization 45: 1-10. Dalfelt, Arne. 1991. "Ecological Constraints to Sustainable Management of the Tropical Moist Forest." Environment Department XVorking Paper 1991-25. World Bank, Washington. DC. Dalrymple, J. B., and D. M. Robison. 1989. "A Soil-Based Assessment of the Sustainabilitv of a Zero-Input Alternative to Shifting Cultivation in Boliia." Interciencia 14(6):329-40. Daly, Hlerman. 1991. "Ecological Economics and Sustainable Development: From Concept to Policy." Environment Department Working Paper 1991-24. World Bank, Washington, DC. Daly, Herman, Robert Goodland, and others, eds. 1991. "Environmentally Sustain- able Economic Development-Building on Brundtland." Environment Depart- ment Working Paper 46. World Bank, Washington, DC. Davis, Shelton H., ed. 1991. Indigenous ViewsofLand and the Environment. Discussion Paper 188. World Bank, WVashington, D.C. Davis, Shelton, and Alaka Wali. 1991. "Land Regularization in Special Amerindian Components of Bank-Funded Projects in Lowland South America." Latin Amer- ica and Caribbean Department, World Bank, Washington, D.C. Deacon, Robert. 1990. "Government Policy and Environmental Quality in Devel- oping Countries: Complements or Substitutes?" Department of Economics, University of California, Santa Barbara, Calif. de Freitas, Davies, and Maria de Lourdes. 1989. "Desmatamento na Amazonia: Causas, Efeitos e Soluc,oes." Interciencia 14(6):298-303. Denevan, W. 1978. "The Causes and Consequences of Shifting Cultivation in Relation to Tropical Forest Survival." In Denevan, W. (ed), The Role of Geographi- cal Researlc in Latin America. Publication 7. Conference of Latin Americanist Geographers. Dewees, Peter A. 1992. "Tree Planting and Household Decision Making Process Amongst Smallholders in Kenya." Food Studies Group, Oxford University. Dickenson, J. P. 1989. "Development in Brazilian Amazonia: Background to New Frontiers." Revista Geogra[fica 109: 141-155. D'Silva, Emmanuel, and Simmathiri Appenah. 1993. Forest7y A-1nnagement for Suls- tainable Development. EDI Seminar Report No. 32. Washington, D.C.: World Bank Dunham, David. 1989. "Politics and Land Settlement Schemes: The Case of Sri Lanka." Developlenzt and Change 13(1): 43. Eastwood, D. A., and H. J. Pollard. 1985. "The Development of Colonization in Lowland Bolivia: Objectives and Evaluation." Boletin de Eatudios Latinoamericanos y del Caribe 38: 61. Eden, Michael. 1988. "Crop Diversitv in Tropical Swidden Cultivation: Compara- tive Data From Colombia and Papua New Guinea." Agriculture, Ecosystems and Environment 20: 127-136. Bibliiogapbv 1 73 1990. Ecology and Land M1anagemnent in AImazon7ia. New York: Bellhaven Press. Eggertsson, Thrainn. 1990. Economic Behaviorof 'institttionis. New York: Cambridge University Press. Ehrlich, Anine H., and Paul R. Ehrlich. 1992. "The V'alue of Biodiversity." A.l BlO 21(3): 219-243. Ehui, Simeon K. 1990. "Forest Resource Depletion, Soil Dynamics. and Agricultural Productivity in the Tropics." J7oural of Environmental Economics and Allanagement 18: 136-154. Eskeland, Gunnar S., and Emmanuel Jimenez. 1992. "Policy Instruments for Pollution Control in Developing Countries." World Bank Obrei-ver 7(2): 145-69. Ewell, Peter T., and Thomas T. Poleman. 1982. "Uxpanapa: Resettlement and Agricultural Development in the Mexican Tropics." A.E. Research 82-7.Cor- nell/International Agricultural Economics Studv, Ithaca, N.Y. Fearnside, Phillip. 1986. HumaanZ Cna7ying Capacity oJ'the Brazilian Rain tbrest. New York: Columbia University Press. . 1989. "Deforestation and Agricultural Development in Brazilian Ama- zonia." Intercienc-ia 14(6): 291-97. . 1993. "Deforestation in Brazilian Amazonia: The Effects of Population and Land Tenure." A2X1BIO 22(8): 537-45. Feder, Gershon, and David Feeny. 1991. "Land Tenure and Property Rights: Theory and Implications for Development Policy." World Bank Eco7nomic Review 5 (anuary): 135-53. Feder, Gershon, Tongroi Onchan, Yongyuth Chalamwong, and Chira Honglacharon. 1988. Land Policies and Farwm Productivity in Thailand. Baltimore, Md.: Johns Hopkins University Press. Feeny, David. 1988. "Agricultural Expansion and Forest Depletion in Thailand, 1900-1975." In John F. Richards and Richard P. Tucker. Wliorld Deforestation in the Twentieth Centuiny. Durham, N.C.: Duke Universitv Press. Feeny, David, Filkret Berkes, Bonniej. McCay. and lames M. Acheson. 1990. "The Tragedy of the Commons: Twenty-Two Years Later." Hutman Etcology 18(1): 1-19. Field, B.C. 1989. "The Evolution of Property Rights." Kyklos 42: 319-45. Ford Foundation. 1989. Alternativas ao Des77atamen7to na 4mazonia: Conselrva(do dos Recnuisos Nan/urais. 2 d edition. New York. Foweraker,Joe. 1981. The Strz ggleforLand: A Political Economy of the Pioneer Fro7itier in Brazilfro0n 1930 to the Presenit Day. New York: Cambridge University Press. Franca Ribeiro. Ueliton, and Paulo Rodolfo Leopoldo. 1989. "Coloniza,a-o ao Longo da Transamaz6nica: Trecho Km 930-1035." Inte-rciencia 14(6): 311-16. Furze, Brian. 1992. "Ecologically Sustainable Rural Development and the Difficultv of Social Change." Environmental Values 1: 141-55. Garcia, Afranio Raul Jr. 1990. 0 Sul: Caminho do Rofado. Emtrate'gias de Reprodilqio Camponesa e Tranfrrma(jo Social. Brasilia: Editora Universidade de Brasilia. Geertz, Clifford. 1963. AgriTcultural Involution: The Process of 'Ecological Change in Indonesia. Berkeley, Calif.: University of California Press. Gillis, Malcolm, and Robert Repetto, eds. 1988. Pu1blic Policies and the Misuse oqfForest Resoulrces. New York: Cambridge University Press. 174 Sustainable Settlement in the Brazilian Amazon Goodland, Robert. 1991. "Tropical Deforestation Solutions, Ethics and Religions." Environment Department Working Paper 43. World Bank, Washington, D.C. -______ 1992. "Environmental Priorities for Financing Institutions." Enviironmental Conservation 19(1). Goodland, Robert, and Herman Daly. 1993. "Poverty Alleviation is Essential for Environmental Sustainability." Working Paper 1993-42. Environment Depart- ment, World Bank, Washington, D.C. Goodland, Robert, Herman Daly, and Salah El Serafy. 1992. "The Urgent Need for a Rapid Transition to Global Environmental Sustainability." Environment Department, World Bank, Washington, D.C. Goodman, David, and Anthony Hall, eds. 1990. The Futtur7e of'Amazonia: Destrtction or Suistainable Development? New York: St. Martin's Press. Gradwohl, Judith, and Russel Greenberg. 1988. Saving the Tropical Forests. Wash- ington, D.C.: Island Press. Gregersen, Hans, Peter Oram, and others, eds. Pr ior itiesfor Forestry andAg-ofrest7y Poligy Researlc: Repo?t of'an International Workshop. Washington, D.C.: Interna- tional Food Policy Research Institite. Gross, Anthony. 1990. "Amazonia in the Nineties: Sustainable Development or Another Decade of Destruction?" Third IWI'orld Quarterly 12: 3-4. Guzman, Rolando M., and Eustaquio J. Reis. 1992. "An Econometric Model of the Amazon Deforestation." Discussion Paper 265. Institude de Pesquisa Economica Aplicada, Rio deJaneiro. Hall, Anthony. 1989. DevelopingAmazonia: Deforestation and Social Conflict in Brazil's Carajas Programme. New York: Manchester University Press. Hansen, Stein. 1988. "Debt for Nature Swaps: Overview and Discussion of Key Issues." Environment Department Working Paper 1. World Bank, Washington, D.C. Harley, J. L., and F. Le Tacon. 1990. " Deforestation in the Tropics and Proposals to Arrest It." AAIBIO 19(8): 3 72-78. Hebette, Jean, ed. 1991. 0 Cerco EFta se Fechando. Petr6polis/Rio de Janeiro: Editora Vozes. Hebette, Jean, and Rosa E.A. Marin. 1977. Colon?iza (do espontanea, politica ag7aria e gr7upossociais. Belem, Para: University of Pari. Hecht, Susanna. 1981. "Deforestation in the Amazon Basin: Magnitude, Dynamics and Social Resource Effects." Stu1dies in Third 1W7or-ld Societies 13: 61. Hecht, Susanna B., Richard Nirgaard, and others. 1988. "TFhe Economics of Cattle Ranching in Eastern Amazonia." Interciencia 13(5): 233-40. Helmsing, Bert. 1982. "Agricultural Production in the Periphery: Settlements Schemes Reconsidered." Development and Change 13(1): 401. 1983. "Colonos, Agricultural Colonization and Production in Andean Countries." WVorking Paper Series 10, Institute of Social Studies. Hemming, John, ed. 1985a. Change in the Amazonz Basin, Voltuzme 1: Alan's Impact on Forests and Rivers. New York: Mlanchester University Press. 1985b. Change in theAmazon Basin, Voluime II: The Frontier-ftera Decade of Colonization. Symposium held at the 44th International Congress of Americanists, Manchester, U.K., September 1982. New York: Manchester University Press. Bibliograplhy 175 Hernandez, Alejandro. 1984. "Migraci6n de Colonos en Darien." In Alberto lcKav and Stanley Heckadon Morefio, eds., Colonzmaci6n v Deserlucci6y de Bosq.es en Panamd. Panama: Asociacion Panamefia de Antropologia. Horowitz, Mlichael, and Peter D. Little, eds. 1987. Lands at Risk in the Third lloo1 d: Local-Level Peispectives. Boulder, Colo.: Westview Press. Hsiao, Cheng. 1986. Alnalysis of Panel Data. Econometric Society 'INonographs. Cambridge, U.K.: Cambridge University Press. lanni, Octamio. 1979. "Colonizacao e Contra-Reforma Agraria na Amazonia." Colecao Sociologia Brasileira 11. Petropolis, Brazil: Editora Vozes. Inter-American Development Bank. 1970. "Evaluation of Tropical Colonization Projects in Latin America." Papers on Agricultural Development No. 7. WVash- ington, D.C. Janz, Klaus. Jean-Paul Lanly, and Deo Singh. 1991. "FAO's 1990 Reassessment of Tropical Forest Cover." Natur-e and Resources 27(2): 21-26. Jessup, Timothy C., and Nancy Lee Peluso. 1986. "Mlinor Forest Products as Common Property Resources in East Kalimantan. Indonesia." Proceedin7gs of tlhe Confe1ence o07 Common Property Resource .1lanagement. Washington, D.C.: National Academv Press. Jones, Donald, Virginia Dale, John Beauchamp, Marcos Pedlowski, and Rober O'Neill. 1992. "Farming in Rond6nia." U.S Department of Energy, W'ashington, D.C. Jones,Jeffrey. 1990. Colonization and Environment: Land Settlement Projects in Centnal America. Tokyo: United Nations University Press. Junk, WolfgangJ. 1989. "The Use ofAmazonian Floodplains under an Ecological Perspective." Iintercienicia 14(6): 317-22. Katzman, Martin. 1977. Cities and Fr-on tier s in Brazil: Regional Dimensions of Ecoionoic Development. Cambridge, Mass.: Harvard University Press. Kelly, Brian, and Mark London. 1983. Amazon7. New York: Holt, Rinehart, and Winston. Kneerim, Jill. 1980. Village lfomen Organize: The 17a1771 Bus Service. New York: SEEDS. Kolhepp, Gerd, and Achim Schrader, eds. 1987. Homen e iVNatieza na Aannzonia. Simp6sio Internacional e Interdisciplinar. Blaubeuren, 1986. Tubinger Geographisches Stidien, Blaubeureni. Lal, R. 1989. "Potential of Agroforestrv as a Sustainable Alternative to Shifting Cultivation: Concluding Remarks." A4grofivrestiy Systems 8: 239-42. Lamb, David. 1991. "Combining Traditional and Commercial Uses of Rain For- ests." Nature and Resol7rces 2 7(2): 3-1 1. Lavinas, Lena, ed. 1987._A hibaniza(do da Fronteira. Volunmes I and 2. Rio deJaneiro: Publipur. Levi, Yair. 1984. "Settlers' Organizations and the Settlement Community: Relationships Emerging in New Land Settlement Projects." j7ournl of Rural Cooperation 12(1-2): 25. Lin, Justin Yifu, and Jeffrey Nugent. 1994. "Institutions and Economic Develop- ment." In Jere Behrman and T. N. Srinivasan, eds., Handbook of Development Econ7omics. Amsterdam: North Holland. 176 Suvtainable Settlement in the Brazilian Amazon Lisansky, Elizabeth. 1990. Aligrants to Amazonia: .SponMtaeous Colonization in the B7-azilian Fr-ontier. Boulder, Colo.: Westview Press. Livernash, Robert. 1992. "The Growing Influence of NGOs in the Developing World." Environment 34(5): 13-43. Lopez, Ramon. 1992. "Environmental Degradation and Economic Openness in LDCs: The Poverty Linkage." 4me?ican jour7nal ofj'Agricultaral7z Economics Decem- ber: 1138-1143. Lopez, Ramon, and Mario Niklitschek. 1991. "Dual Economic Growth in Poor Tropical Areas." joun77al of 'Dezvelopme7nt Economicsv 36: 189-211. Loureiro, Violeta. 1985. Os Parceiros do Ilar: Natureza e Conflito Social na Pesca da Amazonia Bel6,n. Pard. Brazil: Coselho Nacional de Desenvolvimento Cientifico e Tecnol6gico. - 1987. M-ise'ria da Ascensdo Social: Capitalismo e Pequena Produ(do naAmazonia. Sao Paulo: Editora Marco Zero. Lovejoy, Thomas, and Ghillean Prance, eds. 1985. Amazonia: Key Environments. New York: Pergamon Press. Luoma, Jon. 1992. "New Government Plan for National Forests Generates a Debate." The Nev York Times, June 30, page C4. Lutz, Ernst, and Mohan Munasinghe. 1991. "Environmental-Economic Evaluation of Projects and Policies for Sustainable Development." Environment Department Working Paper 42. World Bank, Washington, D.C. Maddala, G. F. 1983. Limited-Dependent and Qualitative Variables in Econometrics. Econometrics Society Monographs. Cambridge, U.K.: Cambridge University Press. Magrath, William. 1989. "The Challenge of the Commons: The Allocation of Nonexclusive Resources." Environment Departnent Working Paper 14. World Bank, Washington, D.C. Mahar, Dennis. 1979. Frontier Development Policy in Brazil: A Study of Amazonia. New York: Praeger Publishers. . 1981. Brazil: Integrated Development of'the Northb!est Frontier-. World Bank Country Study. Washington, D.C. 1989. Governmen7t Policies and Dej7restation in Brazil'sAmazon Region. Wash- ington, D.C.: World Bank. Mahony, Rhona. 1992. "Debt-For-Nature-Swaps, Who Really Benefits?" The Ecolo- gist 22(3): 99-103. Manshard, Walther, and Willian B. Morgan, eds. 1988. Agricultur-al ELpansion and Pioneer- Settlements in Humid Tropics. Tokyo: United Nations University Press. Margolis, Maxine. 1973. The Moving Frontier: Social Economic Change in a Southern B1raziliani Community. Gainesville, : University of Florida Press. Markandya, Anil, and David Pearce. 1988. "Environmental Considerations and the Choice Discount Rate in Developing Countries." Environment Discussion Paper 3. World Bank, Washington, D.C. Marques, Fabricio. 1993. "Limite Imaginario." Veja 26(22). Martine, George, ed. 1985. Biotecnologia e Sociedade: 0 Cas.o Brasileir-o. Sao Paulo: Editora da UNICkMP. . 1992a. "Processos Recentes de Concentracao e Desconcentracao Urbana no Brazil: Determinantes e Implicacoes." Brasilia: Instituto SPN. Biblj1iography 17. . 1992b. "The Recent Trends Toward Deconcentration and Demetropoli- zation in Brazil." Paper prepared for the 1993 IussP General Conference, Mont- real, Canada. . 1992c. "Fases e Faces da AIoderniza co Agricola Brasileira." Planejamento e Politicas Priblicas 3: 3 -44. Martine, George, and Liscio Camargo. 1983*. Crescimento e Distribui(do da Populasao Brasileira: Tendencias Recentes. Discussioln Paper 5/82. Brasilia: IPE/PNTUD. Martine, George, and Vilmar Faria. 1988. "Impacts of Social Research on Policy Formulation: Lessons from the Brazilian Experience in the Population Field." Thye ouly7ial oJ Developing Ar-eas 2 3: 43-62. Mattos, M. M., C. Uhl, and D. A. Goncalves. 1992. Economnic and Ecological Perspec- tines on Ranching in the Eastern Amazon in the 1990s. Belem, Brazil: Imazon. McMillan, Della. Thomas Painter, and others. 1990. "Settlement Experiences and Development Strategies in the Onchocerciasis Controlled Areas of West Africa- Final Report." Land Settlemen7t Reviev. New York: Institute for Development Anthropology. Mesquita, Olindina Vianna, and Solange Tietzman Silva. 1987a. "A Evolu,co da Agricultura Brasileira na Decada de 70." Revista Brasileira de Geogiafia 1(49): 3-10. . 1987b. "A Evoluc,o da Agricultura Brasileira na Regiao Sul na Decada de 70." Revista Brasileira de Geografia 1(49): 159-95. 1987c. "Da Policultura Regional a Hegemonia da Soia-O Oeste do Parani em Questao." Revista Brasileira de Geografia 105 (january/june): 155-70. Millikan, Brent. 1992. "Tropical Deforestation, Land Degradation, and Society: Lessons from Rond6nia." Latin Americani Perspectives 19(1): 45-72. Mink, Stephen D. 1993. Poverty, Population, and the Environment. Discussion Paper 189. Washington, D.C.: World Bank. Monteiro de Barros, Henrique 0. 1991. "Small Commodity Agriculture in North- east Brazil: The Case of Horticultural Farming in Pernambuco Brejos." Ph.D. dissertation, University of London. Moran, Emilio. 1975. "Pioneer Farmers of the Transaniazon Highway: Adaptation and Agricultural Production in the Lowland Tropics." Ph.D. Dissertation, Uni- versity of Florida, Gainesville, Fl. 1982. "Ecological, Anthropological and Agronomic Research in the Ama- zon Basin." Latin Amelican7 Research Review 17(1): 3-41. . 1983. The Dilemma oJfAmazonian Developnent. Boulder, Colo.: Westview Press. ,ed. 1989. The Dilemma ofAmazonian Dev'elopment. Boulder, Colo.: Westiew Press. Mueller, Charles C. 1992. "Centro-Oeste: Evolu,co, Situac,o Atual e Perspectivas de Desenvolvimento Sustentavel." A Ecologia e o Novo Padrao de Desenvolhimnento no Brasil 89-129. Munasinghe, Mohan. 1992. "Environmental Economics and Valuation in Develop- ment Decisionmaking." Environment Department Working Paper 51. World Bank, Washington, D.C. Murgueitio, Enrique. 1990. "Intensive Sustainable Livestock Production: An Alter- native to Tropical Deforestation.",AMBIO 19(8): 397-400. 1 78 Sus.ainiable Settlement in the Brazilian Amazon Musumecci. L. 1988. 0 mito da te77r liberta; coloinizaciio espontdnea, campesinato, e patrondgem na Amaz6nia Oriental. Sao Paulo: ANPOCS/Vertice. Mvers, Norman. 1988. "Natural Resource System and Human Exploitation Sys- tems: Phvsiobiotic and Ecological Linkages." Environment Department Work- ing Paper 12. World Bank, Washingron, D.C. Nelson, Michael. 1973. The Development of Tropical LaEnds: Policy Issules in Latin America. Baltimore, Md.: Johns Hopkins University Press. Nicholaides,jj., and others. 1985. "Agricultural Alternatives for the Amazon Basin." Bioscience 35(1): 279-85. North, Douglass. 1990. Institutions, Institutional Change, and Economic Pe7fot7mance. New York: Cambridge University Press. Oberai, A. S., ed. 1988. Land Settlemen7t Policies and Population Redistributtion in Developing Countries: Achievements, Problens, and Prospects. New York: Praeger Publishers. Oliveira, Francisco de. 1981. Elegia para umna Religiao. Rio deJaneiro: Ed. Paz e Terra. Oz6rio de Almeida, Anna Luiza, ed. 1984. Biotern0ologia eAgricultura: Perspectivaspara o Caso Brasileiro. Rio dejaneiro: Editora Vozes. . 1985. "Atribulac6es de uma Economista na Arnaz6nia." Cie^ncia Holje 3(16): 67-74. . 1987. "Os Comerciantes da Fronteira." Texto para discussao 149. Instituto de Economia Industrial. Rio dejaneiro: UFRI. . 1989. "Colonizing the Amazon." Hemisphere 1(3): 24-25. 1 1990a. A Colonizafiao Oficial na Amaz6nia nos Anos 80. Texto para Discussao 207. Rio de janeiro-Brasilia: IPEA. . 1990b. A Colonizgado Particn1lar na Amazonia nos Anos 80. Texto para Discussao 208. Rio de Janeiro-Brasilia: IPEA. . 1990c. "A Biotecnologia no Brasil: Desenvolvimentos Recentes." Ciincia e Te-niologia 83(5): 545-70. ed. 1990d. "Biotecnologia: Situarao e Perspectivas. Resultados Prelimi- nares." Cadei77os de Economia 2. Rio de Janeiro/Brasilia: IPEA. . 1991a. "Colonizacao na Amazonia: Reforma Agraria numa Fronteira Inter- nacional." Peispectivas da Economia Brasileir a -1992. Rio de Janeiro-Brasilia: IPEA. . 1991b. "Debt Peonage and Over-Deforestation in the Brazilian Amazon." From the proceedings of the meeting of the International Agricultural Economics Association, August. Tokyo. 1992. The Colonization of tbe Amazon. Austin: University of Texas at Austin Press. Oz6rio de Almeida, Anna Luiza, and Carlos Eduardo Rebello de Mendon,a. 1989. "A Pequena Produ,co: Uma Visao Unificada." Estndos Econ6micos 19: 9-23. Oz6rio de Almeida, Anna Luiza, Angela Moulin Penalva Santos, Adriana F. Alves, and Maria da Piedade Morais. 1992. A Colonizafao Suseentd'el na Amaz6nia. Texto para Discussao 266. Rio deJaneiro-Brasilia: IPEA. Oz6rio de Almeida, Anna Luiza, and Charley F. Velloso dos Santos. 1990a. "A Colonizacio Oficial na Amazonia nos Anos 80." Texto para Discussao 207. Rio de J aneiro/Brasilia: IPEA. 1990b. "A Colonizacao Particular na Amazonia nos Anos 80." Textos para Discussao 208. Rio de Janeiro/Brasilia: IPEA. Bibliography 1 79 Oz6rio de Almeida, Anna Luiza, and S. Graham. 1994. "Regulatorv Costs and Employment in the Informal Sector in Argentina." Education and Social Policy Discussion Paper 45. World Bank, Washington, D.C. Padoch, C. 1980. "The Environmental and Demographic Effects of Alternative Cash-Producing Activities Among Shifting Cltiivators in Sarawak." Tropical Ecologv and Development, Department of Botany, Banaras Hindu University, Vanarasi, India. Padoch, C., and Andrew P. Vayda. 1983. "Patterns of Resource Use and Human Settlement in Tropical Forests." In F. B. Golley, ed., Tropital Rain Forest Eco.ys- tems: Str-lnttre and Fnnction. Lausanne: Elsevier Scientific Publishing Company. Padoch, C., J. Chota, and others. 1985. "Amazonian Agroforestrv: A Market-Ori- ented System in Peru." .4groJ7-re.tity Sys-tems 3: 47-58. Palm, Ola, and Klas Sandell. 1989. "Sustainable Agriculture and Nitrogen Supply in Sri Lanka: Farmers' and Scientists' Perspective." AMBIO 18(8): 442-48. Paris, R., and I. Ruzicka. 1991. "Barking Up the Wrong Tree: The Role of Rent Appropriation in Sustainable Tropical Forest Management." Asian Development Bank, Environment Office Occasional Paper 1. Manila. Partridge, William, and Debra Schumman. 1989. The Human Ecology of Tr opicalLan7d Settlement in Latin America. Boulder, Colo.: Westview Press. Pearce, David, and Giles Atkinson. 1993. "Measuring Sustainable Development." Ecodecision 9 (June). Pearce, David, and Anil Markandya. 1989. "Marginal Opportunity Cost as a Plan- ning Concept in Natural Resources Mlanagement." In Gunter Schramm and Jeremy Warford, eds., Envir-onmental Management and Economic Development. Baltimore, Md.: Johns Hopkins University Press. Pearce, David, and Douglas Southgate. 1988. "Agricultural Colonization and Envi- ronmental Degradation in Frontier Developing Economies." Environment De- partment Working Paper 9. World Bank, Washington, D.C. Pearce, David, and Jeremy J. Warford. 1993. WVorld Without End: Econzomics, En'i- ronment, and Sustainable Dez'elopment. New York: Oxford University Press. Penalva Santos, Angela Moulin S. 1993. "Comercio Fronteira de Negocios na Colonizacao da Amazonia." Rio de Janeiro: IPEAJPNUD-BRA. Peuker, Axel. 1992. "Public Policies and Deforestation: A Case Studv of Costa Rica." Report 14. Latin America and the Caribbean Technical Department, World Bank, Washington, D.C. Pezzey, John. 1989. "Economic Analysis of Sustainable Growth and Sustainable Development." Environment Department Working Paper 15. World Bank, Washington, D.C. . 1992. Sustainable Development Concepts: An Economic Analysis. WVorld Bank Environment Paper No. 2. Washington, D.C. Poelhekke, Fabio G. Ml. N. 1986. "Fences in the Jungle: Cattle Raising and the Economic and Social Integration of the Amazon Region in Brazil." Revista Geogra7fica (Instituto Pan-Americano de Geografia e Hist6ria) 104: 33-43. Porto Tavares, Vania, and others. 1972. Coloniza(do Dir-igida no7 Brasil: Suas Possibili- dlades na Regiao Anmazonica. Rio dejaneiro: IPEA. Prosterman, R.L., ed. 1990. Agrarian ReJbrm and Grassroots Development: Ten Case St7ndies. Seattle: University of Washington. 180 Sustainable Settlement in the Brzilian Amazon Redclift, Michael. 1986. "Sustainability and the Market: Survival Strategies on the Bolivian Frontier." TheJour7nal of Development Studies 23(1): 93-105. .1 987. Sustainable Developme77t: Exploring the Contr adictions. New York: Methuen. dos Reis Velloso, Joao Paulo, ed. 1992. A Ecologia e o Novo Padr-io de Desenvolvimento no Brasil. Sao Paulo: Nobel. Renkow, Mitch. 1993. "Land Prices, Land Rents, and Technological Change: Evidence from Pakistan." WFoi-ld Development 21(5). Rensberger, Boyce. 1993. "Environment: Tropical Deforestation's Other Side." The Washington Post. September 20. Repetto, Robert. 1989. "Economic Incentives for Sustainable Production." In Gunter Schramm and Jeremy Warford, eds., Envir0onmental Aranagement and Economic Development. Baltimore. Md.: Johns Hopkins University Press. Rezende, Gervasio Castro de. 1981. "Credito Rural Subsidiado e Preco da Terra no Brasil." Texto para Discussao Interna 41. Rio dejaneiro: IPEA/INTES. Romanoff, Steven. 1992. "Food and Debt Among Rubber Tappers in the Bolivian Amazon." Huiman Organization 5 1(2): 122-35. Sadan, E. 1985. "Settlement, Rehabilitation and Reform: Dilemmas in Agricultural Development Projects in Latin America." Jotnual of Rural Coope7ation 13(2): 135-48. Sanchez, Pedro A. 1991. "Alternatives to Slash and Burn: A Pragmatic Approach to Mitigate Tropical Deforestation." Paper presented at a World Bank conference on "Agricultural Technology for Sustainable Economic Development in the New Century: Policy Issues for the International Community." Airlie House, Virginia. October 23. Serageldin, Ismail. 1993. Development Parme-s: Aid and Cooperation in the 1990s. Washington, D.C.: SIDA. Savvyer, Donald. 1990. "Migration and Urban Development in the Amazon." Paper prepared for the World Bank as part of a critical literature review on Amazonian migrations. World Bank, Washington, D.C. Schmidt, R. 1987. "Tropical Rain Forest Management." TNASYVA 39(2): 2-17. Schmink, Marianne, and Charles Wood. 1979. "Blaming the Victim: Small Farmer Production in an Amazonian Colonization Project." Studiesin Third WVorld Societies 7(77). , eds. 1984. Frontier- Expansion in Amazonia. Gainesville: The University of Florida Press. 1992. Contested Frontieis in Amazonia. New York: Columbia University Press. Schneider, Robert R. 1991. "An Analysis of Environmental Problems and Policies in the Amazon." Paper presented at the Seminario sobre Politicas y Practicas para Desarrollo Sostenible en los Paises Miembros del Tratado de Cooperacion Amazonica. Caracas, Venezuela, October 21-25. 1993. "Land Abandonment, Property Rights, and Agricultural Sustainabil- ityin the Amazon." Latin America and the Caribbean Department, Dissemination Note No. III. World Bank, Washington, D.C. Schonkwiler,J. Scott, and Timothy G. Taylor. 1986. "Alternative Stochastic Speci- fications of the Frontier Production Function in the Analysis of Agricultural Credit Programs and Technical Efficiency." Jou7nal of Development Economics 2 1: 149-60. Bibliog7-aphv 181 Ser6a da Motta, Ronaldo. 1991. "Recent Evolution of Environmental Management in the Brazilian Public Sector: Issues and Recommendations." In D. Erocal, ed., Environmentatl Ilainagemnenzt in Developing Cozrntries. Paris: OECD. . 1992a. "Perdas e Servicos Ambientais do Recurso Agua para Uso Domestico." IPEA, Rio de Janeiro. . 1992b. "Some Comments on Depletion and Degradation Costs on Income Measurement." IPEA, Rio deJaneiro. Ser6a da Motta, Ronaldo, and Peter Herman May. 1992. "Loss in Forest Resource Values due to Agricultural Land Conversion in Brazil." IPEA. Rio deJaneiro. Ser6a da Motta, Ronaldo, and Eustaquio Reis. 1994. "The Application of Economic Instruments in Environmetal Policy: The Brazilian Case." OECD/UN\TEP workshop paper. Paris. May 26-27. Ser6a da Motta, Ronaldo, and Carlos Eduardo Frickman Young. 1991. "Natural Resources and National Accounts: Sustainable Income from Mineral Extraction in Brazil." Rio dejaneiro: IPEA Serra, Jose, and Jose Roberto R. Afonso. 1993. "Financas Publicas Municipais II: Trajetoria e Mitos." Conjazntura Economica. November/December: 35-43. Shah, Anwar. 1990. "The New Fiscal Federalism in Brazil." Working Paper 557. World Bank, Washington, D.C. Sikking, Kathryn. 1991. Ideas and Institutions: Developmentalism in Brazil and Arge7z- tina. Ithaca, N.Y: Cornell University Press. Sioli, Harold. 1989. "Introducao ao Simp6sio 'Amaz6nia: Deflorestamento e Possiveis Efeitos'." Interciencia 14(6): 286-0. Smith, Nigel J. H. 1982. Rain brest Cor7idors: The Transamazon Colonization Scheme. Berkeley: University of California Press. Socolik, Helio. 1989. "O Fundo de Participacao dos Municipios." Coqjuntllra Economica December: 51-58. Southgate, Douglas. 1990. "The Causes of Land Degradation along 'Spontaneously' Expanding Agricultural Frontiers in the Third World." Land Economics 66(1): 93-101. . 1991. "Tropical Deforestation and Agricultural Development in Latin America." Environment Department Working Paper 1991-20. World Bank, Washington, D.C. 1992. "Policies Contributing to Agricultural Colonization of Latin Amer- ica's Tropical Forests." In Narendra Sharma, ed., Managin7g the World's Forests: Looking for Balance Between Conservation and Development. Dubuque, Iowa: Ken- dall/Hunt Publishing Company. Southgate, Douglas, and David Pearce. 1988. "Agricultural Colonization and Envi- ronmental Degradation in Frontier Developing Economies." Environment De- partment WVorking Paper 9. World Bank, Washington, D.C. Spears,John. 1988. "Containing Tropical Deforestation: A Review of Priority Areas for Technological and Policy Research." Environment Department Working Paper 10, World Bank, Washington, D.C. Speth,James Gustave. 1992. "On the Road to Rio and to Sustainability." Environ- mental Science and Technology 26(6): 1075-76. Stirling, Andrew. 1993. "Environmental Valuation: How Much is the Emperor Wearing?" The Ecologist 2 3(3). 182 Sustainable Settlement in the Brazilian Amazon Strong, Maurice F. 1992. "The Promises and Challenges of UNCED '92." Ocean and Coastal AManagement 18: 5-14. Stng., Woonki, and Rosaria Troia. 1992. Developments in Debt Conversion Prog7ams and Conv-esion Activities. World Bank Technical Paper 170. Washington, D.C. Thapa, Gopal. 1990. "Actors and Factors of Deforestation in 'Tropical Asia'." Environmental Conservation 17(1): 19-27. Thapa, Gopal, and Karl E. Weber. 1989. "Land Settlement in Tropical Asia." Habitat International 13(4): 147-60. Thiele, Graham. 1990. A7e Peasant Far7ning Systems in the Amazo-n Sutstainable? Discussion Paper 282. Brighton, England: Institute of Development Studies. Thomson, James T., David Feeny, and Ronald Oakerson. 1992. "Institutional Dynamics: The Evolution and Dissolution of Common Property Resource Man- agement." In Bromley, Daniel, David Feeny, and others, eds. Mlaking the Commons WVork. San Francisco, Calif.: ICS Press. Tisdell, Clem. 1988. "Sustainable Development: Different Perspectives of Ecolo- gists and Economists, and Relevance to LDCs." Wforld Development 16(3): 373-84. Tulchin, Joseph S., ed. 1991. Economic Development & Environmental Protection in Latin America. Boulder, Colo.: Lynne Rienner Publishers. Uhl, C., A. Verissimo, P. Barreto, and R. Tarifa. 1992. "A Evoluc,o da Fronteira Amazonica: Oportunidades para um Desenvolvimento Sustentavel." P1ra Desen- volvimento June: 13-31. Uhl, C., A. Verissimo, M. M. Mattos, Z. Brandino, and 1. C. G. Vieira. 1991. "Social, Economic, and Ecological Consequences of Selective Logging in an Amazon Frontier: The Case of Tailandia." Forest Ecology and Mlanagement 46: 243-73. Van Vliet, Klaas Geert. 1989. "Eco and Ethnodevelopment, Geopolitics and Situ- ational Planning: Perspectives for Alternative Discourse and Praxis in the Amazon Basin." Interciencia 14(6): 304-10. Velho, 0. 1981. Frentes de Exrpans4o e Esturtt17a Agraria: Estudo do Processo de Penetra(ido nutma Area da Transamazica. Rio de Janeiro: Zahar. Verissimo, A., P. Barreto, M. M. Mattos, and R. Tarifa. 1992. Impactos da Atividade AMade7 eira e Perspectivas para o Mllaniejo Sustentavel da Floresta numna Velha Fronteir-a da Amazonia: o Caso de Paragominas. Belem, Brazil: Imazon. Villalobos, Arturo. 1982. "Small Farmers' Production Systems and the Improve- ment of Agriculture in Central America." Ag7icultural System 8:209-17. Vosti, Stephen A. 1992. "Constraints to Improved Food Securitv: Linkages .Anong Agriculture, Environment, and Poverty." Paper prepared for the IFPRI Food Policy Workshop on Famine and Drought Mitigation in Ethiopia in the 1990s. Addis Ababa, Ethiopia, July 2-3. Ministry of Economic Planning. Vosti, Stephen A., Thomas Reardon, and others. 1991. Ag7icultwral StistainZability, Growth, and Poverty Alleviation: Issues and Policies. Washington, D.C.: Food and Agriculture Development Centre. Wachter, Daniel. 1992. "Land Tilling for Land Conservation in Developing Coun- tries?" Working Paper 1992-8. World Bank, Washington, D.C. Weil, Connie, andJim Weil. 1983. "Government, Campesinos, and Business in the Bolivian Chapare: A Case Study of Ainazonian Occupation." Inter-American Economic Afairs 3 6(4): 29-62. Wells, Michael. 1991. "Trust Funds and Endowments as a Biodiversity Conserva- tion Tool." Environment Department Working Paper 1991-26. World Bank, Washington, D.C. Bibliog-napby 183 Wesche, Rolf, and Thomas Bruneau. 1990. Integration and Channge in Brazil's liddle Amazon. Ottawa: lUniversitv of Ottawa Press. W7irth, John, and others. 1987. State and Societv in Brazil: Continuity and Chn77ge. Boulder, Colo.: WVestiew Press. WVood, William B. 1990. "Tropical Deforestation." Glolbal Environmental Change 1(1): 23-41. World Bank. 1978. Agricultural Land Settlement. A World Bank Issues Paper. W,Vashington, D.C. 1989. Renewable Resoulrce MVanagement in AgriCutIur.e A World Bank Opera- tions Evaluation Study. Washington, D.C. 1990. Evalzuationi Results fir 1988. A World Bank Operations Evaluation Study. Washington, D.C. 1991a. Enviro-0777ne7tal Assess.nent Sotrcebook. 7olumne 1: Policies, Procedur-es, and Cr-oss-Secetoral lcisues. Technical Paper 139. Washington, D.C. 1991 b. Environmnental Assessment Sourcebook. 1Volume 3: Gutidelines J7r Envi- ronimenital Assesssnent oqfEnergv and Industry Projects. Technical Paper 154. Wash- ington, D.C. .1991c. "Forest Policy Paper."Agricultural and Rural Development Depart- ment, World Bank, Washington, D.C. 1991d The Fores-t Sector. A World Bank Policy Paper. Washington, D.C. 1991e. The 'orld BBank and the Environment: A Progress Report, Fiscal 1991. Washington, D.C. . 1992a. "Environment and Development in Latin America and the Carib- bean-The Role of the World Bank." Latin American and Caribbean Depart- ment, W orld Bank, Washinigtoni, D.C. . 1992b. Population and the IW 'orld Bank: 17nplications From Eight Case Studies. Washington, D.C.: World Bank. . 1992c. U-'o0ld Development Report 1992: Development and the Environment. Nex York: Oxford University Press. 1 992d. UWorld Bank and the Environment 1992. WVashington, D.C. 1993a. "Early Experience with Involunitary Resettlement: Overview." XVorld Bank, Operations Evaluation Department, Washingtoni. D.C. . 1993h. "Pilot Program to Conserve the Brazilian Rain Forest, Background Materials for the Press." 1993c. "Rain Forest Pilot Program Update." .1994a. "Indonesia Transmigration Program: A Review of Five Bank-Supported Projects." Operations and Evaluation Department. . 1994b. Making Development Susetinalble: The fJ orld Bank Grou7p and the Environment. Fiscal 1 994. . 1995. A linistreanfing the Environ7nent: The M-or0ld Bank Group and the EmVironment since the Rio Ea7th Smnmnit. Fis-cal 1995. Young, Carlos E. F. 1994. "Mr. Keynes and the Environment: An Application of the Use-Cost Concept to the Deforestation Problem." University College of London. Zweede,Johan. 1993. "Sustained Yield Forest Nlanagement (SMTM) in the Eastern Amazon." Seminar paper, XNVorld Bank, August 17. Index Accumulation, rates of, 28-32. 45, 58 Credit policies and institutions, 7, 20 n16, 72 n16, 93, 134, 142 n17, 34 n7, 78, 125 Agricultural extension, 7 Crop liens, 67, 141 Agricultural incomes, 2, 7, 63, 69-70, Crop yields, 59-60 n27, 96 77, 131 Current account, 134. 139-41 Agricultural involution, 7, 10 nl, 55. 66, 69 Agricultural prices, 47-48, 63, 97, 130 Data collection procedures, 13 3-3 5 Amazon VVorkingGroup (GTA), 83 Debt peonage, 35 nl7, 36nI 5,51,68, Appropriate technology, 78 71 n5, 141 Decentralization, Fiscal, 2, 5-6, 23-25, 78 Brazilian Environmental Institute Deforestation (TBAMA), 81 causes of. 6-7, 63-71, 76. See also Deforestation, economic model of Cadernetas de poupinwa, 42, 45, 142 cost to global economy of, 3, 8, Capital gains taxation, 2, 77 53-54 Cerre-ado region, 13, 15, 18, 19 n5 as demand for land, 63, 64-65 Coffee cultivation, 47, 59-60 n27, 72 direct taxation of, 7, 77 nrlO, 97 extent and rates of, 50 Colonizacao Sustentavel na Amaz6nia fiscal and credit incentives en- project, 34 nlO couraging, 25-27 Colonization, directed, 4. 37. 48, 56 impact of land prices on. 69. 70 nl, 79 as intraregional issue, 11 C,onservation (preservation), mi- investment and, 49-50, 68 grants' views on, 50, 119 land turnover and, S5 Conservation projects. 80-83, 84 n4, 121 location characteristics and, 69 ConstitLtion of 1988, 4, 22, 33 n2 macroeconomic conditions and, Cooperation, interinstitutional, 80, 2-3, 6, 8, 21-27, 57 n8, 77 85 n13 marketing and storage facilities Corn cultivation. 47. 59-60 n27. 97 and, 69 Credit and fiscal incentives to defor- policies to reduce, 1-2. 19 nl, est, 25-27 80-83 185 186 Sustainable Settlement in the Brazilian Amazon Deforestation (continued) FPi%M (Municipal Participation Fund), productivity and, 41, 64, 76-77 33 n2 total versus current, 7, 64, 69, 70, Frontier 76, 115, 117-18, 155-56 cost of farming on, 8, 42, 44-S I Deforestation, economic model of emergence of middle class on, I, dynamic. 145-48 5. 27, 78 empirical equations in, 148-49 formation of markets in, 65-66 general, 139-43 geographical shift of, 14-15, 35 income effects in, 146-47, 155-58 n14 price effects in, 145, 147-48, historical pattern of colonization 156-58 of. 3-4, 24, 48 results of, 151-66 "hollowing" of, 15-18 shifter variables in, 152-53, 157 impact of tax incentives on, 23-25 static, 143-45 population trends on, 5, 11-18 Democratization, 5, 23-25, 78 productive versus unproductive, 67 Demonstration effect, 2, 64, 79 shifts in dominant economic ac- Depreciation and maintenance costs, tivitv on, 5, 27-32 61 n40 urban character of, 13-14, 27 Directed colonization, 4, 34 n8, 37, violence on, 4, 20 n16 48, 56n1, 79, 133 Geertz, Clifford, 10 Eastern Amazonia, 24, 29, 38 Global economy, cost of deforestation Economic crisis. See Macroeconomic to, 3, 8, 53-54 conditions Gold prospecting, 20 n16, 35 nlS, 85 Environmental costs, 43, 55, 136 n19 Environmental-economic policy, 80- GTA (Amazon Working Group), 83 83 Exchange rates, 44, 137 Extension, agricultural, 7 "Hollowing" of frontier, IS-18 Farmers. See Migrants IBkUAM (Brazilian Environmental Insti- Federal government, role of, 2, 26- tiite), 81 27, 80 Income Fertility, humnan, 11-13 agricultural, 7, 63, 69-70, 77, 105 Fertility, soil, 9, 50, 59 n26, 72 n12, current account, 134, 139-41 120 effects on deforestation of, 63, Financial account, 134, 141 69-70, 77, 145-47, 155-57 Fiscal and credit incentives to defor- interregional differences in, 49, est, 25-27. See ailso Taxation 57 n13, 107-10 Fiscal decentralization, 2, 5-6, 23-25, nonagricultural, 49, 107 78 proxied by consumption, 57 n1 2 Food and Agriculture Organization INCPA (National Institute for Coloni- (FAO), 61 n45, 85 n13 zation and Agrarian Reform), 3, 81 Forest conservation programs, 80-83, Indebtedness of migrants, 51, 126, 141 84 n4 Indigenous Reserves Project, 82 Index 187 Infrastructure provision and policies, and private benefits of settle- 26-27, 35 nlS, 60 n28, 78 ment, 42, 55 Initial capital, 132, 134, 142 productivity and, 54, 67, 76-77 Institute for Applied Economic provision of rural infrastructure Research (IPEA), 34 nlO, 60 n32, and, 60 n28 89, 133 returns to farming and, 52 Inter-American Development Bank, speculation and, 54 85 n13 Local governments, role of, 6, 7, 21 Interest rates, 42, 44-45 Location characteristics, 69 International Advisory Group (LAG), 85 n21 Investment, 49-50, 68, 103, 114, 141, Mlacroeconomic conditions, impact 142-43 on deforestation of, 2-3, 6, 8. 21- "Itinerant" accumulation, 47 27, 57 n8, 77 Alaintenance and depreciation costs, 61 n40 Land hlarketing of agricultural output, 7, deforestation as demand for, 63, 78, 122-24 64-65 Mendes, Chico, 85 n19 discounted salvage (net present) MNIerchants, role in frontier expansion value of, 52, 53, 55, 61 n47, 131 of, 28-32 intensity of use of, 47, 95 Methodological issues, 41-43. See also merchants' accumulation of, 31- Deforestation, economic model of 32 Middle class, emergence on frontier opportunity cost of, 42, 58 nl8 of, 1, 5, 27, 78 rates of return to, 46-48 Migrants static versus dynamic demand age of, 128 for, 64 destination characteristics of, 76, title to, 31-32, 46, 51, 61 n44, 135, 152 79, 127 diversification into nonagricul- Land prices tural activities by, 49 calculation of, 56 n6, 58 n19 economic decisionmaking of, 5 1- compared with net present value, 54 52, 55 indebtedness of, 51 economic conditions and, 6 intentions to remain on current impact on deforestation of, 69, plot of, 104 70 itinerancy of, 64, 69, 112 productive and speculative re- origin characteristics of, 7, 64, sponses to variation in, 66-69, 69, 76, 152 148-5 0 overall performance of, 48-51 productivity and, 46, 71 nlO past land ownership by, 111 reasons for variations over time portfolio management by, 3, 67 in, 58 n20 size of plots farmed by, 129 Land turnover time on plot of, 99, 134 causes of, 48, 54 views on conservation of, 50, 119 deforestation and, 55 Migration migrant indebtedness and, 51 beyond Brazil, 20 nl6 188 Sustainable Settlement in the Br-azilian Amazon Migration (con7tin71ea) Productivity causes of, 21-22, 48, 65 deforestation and, 41, 64, 76-77 fertility rates and, 11-13 interregional differences in, 94 "hollowing" of frontier and, 15-18 land prices and, 46, 72 nlO intraregional, 5, 9, 13, 15,'30 land turnover and, 54, 67, 76-77 Military regime, colonization of fron- migration patterns and, 6 tier under, 22 returns to land and, 46-48 Minimum wage, 42, 44, 56 n4, 140 in rice cultivation, 98 Ministr-v of Agriculture and Agrarian role of local governments in pro- Reform (mIuRAD), 3 moting, 7 Municipal Participation Fund (FP.\), sustainability of frontier farming 33 n2 and, 55 Property rights, 46, 71 n8 Prospecting, 20 n16, 35 nlS, 85 n19 National forests project, 81 Public colonization. See Directed colo- National Institute for Colonization nization, 46 and Agrarian Reform (tINCRA), 3, 81 Net present value of agricultural in- come (N'PVA), 52, 53, 54, 61 n47 Quality of life, perceived, 49, 102 Net worth, 58 nlS, 60 n35, 77, 101, 134, 142 Nonagricultural income, 49, 107 Rain Forest Trust Fund, 81 Nongovernmental organizations, 2, Revenues, government. 22-27, 34 n5 83 Rice cultivation, 47, 59-60 n27, 72 n I1, 97-98 Road construction, 35 nIS Official colonization. See Directed colonization Opportunity costs of frontier farming, Salvage value of land, 61 n47 8, 42, 44-51, 58 n18 Savings certificates (cadeinetas depou- Origin characteristics of migrants, 7, panf7a), 42, 45, 142 69, 152 Saxwmills, 29 Small farmers. See Migrants Soil fertility, 9,50, 59 n26, 72 n12, 120 Participation funds, state and local, 33 Speculation n2 impact on deforestation of, 7,63, 70 Pilot Program to Conserve the Brazil- land turnover and, 55 ian Rain Forest, 81-82 macroeconomic conditions and, 3 Polonoroeste project, 82-83, 85 n23 by middle class, 1, 32 Population trends, 5, 11-18 tax incentives favoring, 25 Portfolio management by migrants, 3, use of taxation to discourage, 7 67, 141 State and municipal participation "Precocious" urbanization, 14, 19 n9 funds, 33 n2 Price of land. See Land prices Storage of agricultural output, 50-5 1, Private costs of settlement, 42-43 69, 122 Inidex 189 Stumpage tax, 2, 77, 80 United Nations agencies, 85 nl3 "Survivors," 90, 134 Urbanization, 13-14, 19 n9, 27 Sustainabilitv of frontier farming conditions necessary for, 6, 65 as intra-Amazonian matter, 9 Violence, frontier, 4, 20 n16 measurement isstxes regarding, 8, 43 Von Thunen ring effect, 58 n20 productivity and, 55 Wages, 42, 44, 56 n4, 91-92, 134, 140 Tapajos basin project, 85 n16 WVestern Amazonia, 24, 29, 30-32. Taxation 38-39 of agricultural income, 2, 7 WVorld Bank capital gains, 2, 77 consultation with nongovermuneni- direct, 7 tal organizations by, 83 incentives created by, 23-25 cooperation with other multilat- of net worth, 77 eral agencies by, 85 n 13 proposals for reform of, 77 experience with forest manage- stumpage, 2, 78, 79, 80 ment of, 84 n4 Technical assistance, World Bank, 2 role in pilot program of, 81 Title to land, 3 1-32, 46, 51, 61 n44, technical assistance provided by, 2 79, 127 Transfers, fiscal, 24. 33 n2, 78 Turnover. See Land turnover Zoning, 7, 77 #~~~~~~~~~~~~~~~~~~~ -01rnF7 -~~~~~~~'IIII l r m l- ?,~~~~~~~~~~;i- e il- =E~~~~~~~~~~~~~~ S, 8 N, , -I 9- 2 110 - 9