Report No. 31193-HN Honduras Drivers of Sustainable Rural Growth and Poverty Reduction in Central America Honduras Case Study (In Two Volumes) Volume II: Background Papers and Technical Appendices December 31,2004 Environmentally and Socially Sustainable Development Latin America and the Caribbean Region Currency andEquivalentUnits CurrencyUnit=Lempiras U S $1 = 18.70Lempiras (at official exchange rate, December2004) FiscalYear January 1to December 31 Abbreviations and Acronyms CA CentralAmerican CAFTA CentralAmerica FreeTrade Agreement CAS Country Assistance Strategy CMAT Modernization andAdministration of Lands(component of PAAR Project) COHDEFOR HondurasForestry Commission DFID Departmentfor International Development ESSD Environmentally and Social SustainableDevelopmentDepartment FA0 Foodand Agriculture Organization FPPL Fundfor ProducersinMountain Slopes (component of P U RProject) GDP Gross Domestic Product GIS Geographic Information Systems GoH Government of Honduras Ha Hectare hh Household HDI HumanDevelopmentIndex HIPC Highly IndebtedPoor Country HYV Highyielding varieties IADB Inter-American Development Bank IFPIU International FoodPolicy ResearchInstitute IICA Inter-AmericanInstitute for Cooperationon Agriculture INA National Agrarian Institute INE National Statistics Institute InfoAgro AgriculturalInformation Department, SAG LAC LatinAmerica andCaribbean LSMS Living Standard Measurement Survey M z Manzana(land measure equalto 0.7 ha) NGO Non-governmentalOrganization OD1 Overseas Development Institute PAAR HondurasRuralLandManagementProject PACTA Access to LandProject PATH LandAdministration Project PESA SpecialProgramfor Food Security (FAO) PROBAP Biodiversity andPriority Areas Project PRSP Poverty Reduction Strategy Paper RNFE RuralNon-farm Employment RUTA RegionalUnitfor Technical Assistance 11 .. SAG Secretariat for Agriculture and Livestock SINIT Sistema Nacional de Informacion Territorial UNDP UnitedNations Development Programme U S A I D UnitedStates Agency for International Development WB World Bank WFP World Food Program Vice President: Pamela Cox Country Director: Jane Armitage Sector Director John Redwood Sector Leader: MartinRaine ... 111 Table of Contents VOLUME I Abbreviations and Acronyms ............................................................................................. .. Acknowledgements .......................................................................................................... vm ... 11 Foreword............................................................................................................................ i x Executive Summay ............................................................................................................. x 1. Introduction................................................................................................................ 1 2. Characterization o f the Honduran Rural Economy and Policy Directions ................8 3. Spatial Analysis o f the Rural Sector in Honduras ................................................... 18 4. Analytical Results.................................................................................................... 41 5. Conclusions and Recommendations ........................................................................ 60 References......................................................................................................................... 68 Boxes 2.1. Defining `Hillsides.. `Hillside Areas' and `Valleys' ............................................... 10 2.2. Land Issues inHonduras.......................................................................................... 12 2.3. Honduras Agricultural Rural Development Strategy 2004...................................... 16 3.1. Household Survey Data ........................................................................................... 23 3.2. Livelihoods Studies.................................................................................................. 24 3.3. Remittances and Hillside Households ..................................................................... 27 4.1. Pitfalls with the Statistical Analyses o f the IFPRI and Wisconsin Household Data.......................................................................................................................... 41 4.2. Gender inthe Hillside Areas.................................................................................... 46 4.3. ESSD Projects Used for the Stocktacking Exercise ................................................ 55 4.4. Community-Led Asset Building:Nuestras Raices Program ................................... 59 Tables 2.1. Select Data for Central American Countries. 2000 ................................................... 8 2.2. GDP Shares and Growth Rates. 1981-2001 .......................................................... 13 2.3. Agricultural Share o f Exports and Agricultural Balance o f Trade .......................... 13 2.4. Select Indicators on the Coffee Sector in Central American Countries. 2000......... 14 3.1. Population Change and Density by Province. 1994.. 2002...................................... 19 3.2. Distribution of Households by Landholding Size and Income................................ 24 3.3. Adoption o f Conservation Practices ........................................................................ 25 3.4. HumanAssets: Education Levels ............................................................................ 26 3.5. Financial Assets ....................................................................................................... 28 3.6. Social Assets: Participation inOrganizations ........................................................... 29 iv 3.7. Access to Public Infrastructure and Services (based on University o f Wisconsin 3.8. Access to Public Infrastructure and Services (basedon IFPRI survey data only)...30 survey data only) ...................................................................................................... 31 3.9. Income-based Indicators of Rural Poverty in Honduras Basedon Survey Data.....31 3.10. Household Typology according to Sources o f Income............................................ 32 4.1. Livelihood Cluster Groups, IFPRI Household Survey ............................................ 42 4.2. Livelihood Cluster Groups, Wisconsin Household Survey..................................... 4.3. Salient HouseholdCharacteristics, by Livelihood Strategy, IFPRIHouseholds.....43 44 4.4. Salient HouseholdCharacteristics, by Livelihood Strategy, Wisconsin Households............................................................................................. 45 4.5. Asset Comparison betweenPoor Households and Non-Poor Households, Wisconsin Sample.................................................................................................... 52 Figures 1.1. The Asset-based Approach ....................................................................................... 7 2.1. Change in Income Distribution in Rural Areas. 1993.. 2003..................................... 9 3.1. Honduras Provinces and Counties ........................................................................... 34 3.2. Honduras Topography ............................................................................................. 35 3.3. Transportation Infrastructure and Population Centers............................................. 36 3.4. Population Densities ................................................................................................ 37 3.5. Change inPopulation Densities. 1988.. 2001 .......................................................... 38 3.6. Road Densities......................................................................................................... 39 3.7. Geographical Coverage of CombinedIFPRI and Wisconsin Household Surveys..40 4.1. Annual Per Capita Income, by Livelihood Strategy, IFPRIHouseholds ................48 4.2. Annual Per Capita Income, by Livelihood Strategy, Wisconsin Households .........48 V VOLUME I1 Abbreviations and Acronyms ............................................................................................. 11 .. Appendixes 1. Using an Asset-Based Approach to Identify Drivers of Sustainable Rural Growth and Poverty Reduction in Central America: Conceptual Framework ....................... 1 2. Problems with MeasuringRuralPoverty inHonduras ............................................ 24 3. Agricultural Sector Information............................................................................... 29 4. Spatial Analysis o f Rural Economic Growth inHonduras ...................................... 32 5. Description of Data from IFPRI and University o f Wisconsin Surveys.................. 66 6. Conceptual Framework and Statistical Methods usedfor Clustering o f Households According to Asset Use........................................................................ 73 7. Summary Report: Honduras Project Stocktaking Exercises.................................... 78 8. Description o f the Livelihood Strategies of the Sample Households .................... 102 9. Results of the MultinomialLogit Models.............................................................. 106 10. Income Regression Results.................................................................................... 119 Boxes 1.1. Sectoral and spatial linkages...................................................................................... 4 1.2. The mystery of assets................................................................................................. 7 1.3. Why the poor are poor: Lack o f assets and low asset productivity .......................... 9 1.4. Defining a livelihood ................................................................................................. 9 4.1. Migration and population change ............................................................................ 35 4.2. Information on household level data sources .......................................................... 37 4.3. Topography, agricultural potential and market access ............................................ 38 7.1. Community-ledasset building:Nuestras Rakes Program ...................................... 87 Tables 1.1. Toward a New RuralDevelopment Strategy for Central America.......................... 21 1.2. Household-Level Assets and Linksto Other Levels................................................ 22 2.1. Comparison betweenincome estimates................................................................... 26 3.1. Shares (%) of Gross Value Added inthe Agricultural Sector ................................. 3.2. Change inPurchasing Power by Agricultural Sub-sector 1978 to 2000 .................30 30 4.1. Populationchange by Province. 1988-2001 ............................................................ 46 4.2. Determinants o f municipio-levelpopulation change. 1988-2001............................ 46 4.3. Assets and changes inrural Honduras ..................................................................... 47 4.4. Assets and changes by year and Province. rural Honduras ..................................... 48 5.1. List of Provinces and Counties covered by the IFPRI and Wisconsin Household 70 Summary of results from regional workshops ......................................................... Surveys..................................................................................................................... 7.1. 91 9.1. Explanatoryvariables usedinmultinomial logit model. by livelihood strategy. IFPRI sample ......................................................................................................... 108 vi 9.2. Explanatory variables usedinmultinomial logit model. by livelihood strategy. Wisconsin sample .................................................................................................. 111 9.3. Determinants of Livelihood Strategies (Full Multinomial Logit Model). IFPRI sample ......................................................................................................... 113 9.4. ReducedForm of the Multinomial Logit Model. IFPRI sample ........................... 115 9.5. Determinants of Livelihood Strategies (FullMultinomial Logit Model). Wisconsin sample .................................................................................................. 116 9.6. ReducedForm o f the Multinomial Logit Model. Wisconsin sample .................... 118 10.1. Determinants of Household Income. IFPRIhouseholds........................................ 121 10.2. Determinants of Household Income. Wisconsin households ................................ 122 Figures 1.1. Schematic presentation o f the asset-based approach: Asset-context-behavior- outcomes .................................................................................................................. 23 4.1. Map o f Honduras ..................................................................................................... 53 4.2. Provinces of Honduras............................................................................................. 54 4.3. Transportation, infrastructure andpopulation centers ............................................. 55 4.4. Population density.................................................................................................... 56 4.5. Percent rural by municipio 2001.............................................................................. 57 4.6. Coffee-production areas o f Honduras...................................................................... 58 4.7. Road density............................................................................................................. 59 4.8. Population change, 1988-2001................................................................................. 60 4.9 Changes inincome distribution (1993 & 2003), all rural Honduras........................ 61 4.10. Percentage o f farmers usingsoil conservation measures......................................... 62 63 4.12. Percent households with basic sanitation................................................................. 4.1 1. Percent households with potable water.................................................................... 64 4.13. Poverty inHonduras ................................................................................................ 65 7.1. Schematic presentation of the Asset-based Approach and facilitation o f the project stocktaking exercise ............................................................................... 90 vii APPENDIX 1 Usingan Asset-Based Approachto IdentifyDrivers of SustainableRural Growth and PovertyReductionin CentralAmerica: ConceptualFramework' Paul B. Siegel (Consultant, World Bank and FAOICP) 1The findings ofthis backgroundpaper areusedas inputsinthe WorldBankresearchproject, Driversof SustainableRuralGrowthandPovertyReductioninCentralAmerica. See Siegel(2005). 1. Introduction The Central American regional study, "Identifying Drivers o f Sustainable Rural Growth and Poverty Reduction," i s part o f ongoing efforts by the Environmentally and Socially Sustainable Development (ESSD) Department and Central American (CA) Department inthe Latin America and Caribbean (LAC) Region o f the World Bank to strengthenanalyses and strategies for rural development, and to address fundamental policy issues and investment priorities. An objective o f this study i s to develop appropriate conceptual and analytical frameworks to understand how broad-based economic growth can be stimulated inrural Central America, and to apply the framework for country case studies. An asset-basedapproachhas beenadopted to guide the conceptual and analytical frameworks used for this study, whereby household assets are considered the "drivers" o f growth. This background paper provides an explanation of the asset-based approach, including a review o f literature. 1.1. Background Major economic, political and social changes have taken place in Central America over the past decade. While these changes have ledto some improvements inwell-being and reductions inrural poverty rates, the region i s still characterizedby persistent and stark inequalities in assets and incomes and highnumbers o f poor (Tejo 2000; Morley 2001; Sauma, 2002; Franko 2003; de Ferranti and others 2004). Central American countries have long been characterizedby dualistic agricultural sectors and pervasive rural poverty. A significant share o f the poor can be found in rural areas and rural-urban migration continues to take place (Hereford and Echeverria 2003)*. Broad-based growth is constrained by unequal asset distribution. This inequality i s most evident in terms of landholdings, but other key productive, social and locational assets are also unequally distributed (Attanasio and Szekeley 2001; de Janvry 2002). Policy reforms in the Central American countries ledto new opportunities inthe agricultural sector, especially for production o f export commodities. Infact, agricultural growth inthe 1990s was largely driven by increasing prices o f key agricultural exports and land expansion. Since the late 1990s, prices have fallen dramatically (notably coffee prices) and opportunities for further land expansion are limited. Furthermore, there is not much optimism for sustained commodity price increases inthe near future, with many commodities experiencing negative price trends (FA0 2002). As a result, there has been perception o f a crisis inrural areas. The rural crisis has been compounded by natural disasters such as Hurricane Mitch and recurring droughts which have increased vulnerability in agricultural-dependent rural areas (de Ferranti and others 2000; IADB 2000; Keipi and Tyson 2001; Kiesel2001). Increasedmarket and trade liberalization and decentralization all have the potential to create conditions for growth over time, but these About 40 percent of Central America's total population is rural, but the rural poor constitute an overwhelming majority o f the total poor (World Bank 2002b). Migration has been an important strategy for many rural poor inCentral America, accounting for mucho f the reduction inrural poverty during the 1990s (de Janvry and Sadoulet 2001). However, some of this migration to urban areas has merely contributed to increases in urban poverty, a spatial redistributionofpoverty. 2 changes, includingthe Central American Free Trade Agreement (CAFTA), further contribute to uncertainty in the short-term. Most vulnerable among the rural poor inCentral America are those with small landholdings and landless farm workers living in ecologically fragile areas, such as hillsides and sub-humiddrought-prone areas. Many o f these areas lack basic transport, communication and social infrastructure. Households in such areas have limitedassets and livelihood opportunities. They tend to have lower levels o f education, larger families, and strong communal traditions and cultural values that are not well understood inthe context o f the market economy. They produce for subsistence or the local market and are often net purchasers o f food. Their productivity has not kept pace with other sectors o f the economy, and many see migration as their best opportunity to escape poverty. Traditional policy and market-basedreforms cannot quickly resolve decades of structural limitations facing the rural poor, including highly unequal access to productive and social infrastructure and unequal asset distributions. 1.2. Needfor new rural developmentperspectivesinCentralAmerica Governments and donors seldom understand what drives rural growth and poverty reduction. A s a result they have little guidance on how they can formulate strategies and prioritize investments for the rural sector (Echeverria 2001a). The Latin America and Caribbean (LAC) Regional Rural Development Strategy (World Bank 2002b) and others3 acknowledge that new approaches are needed to conceptualize, analyze and operationalize strategies and investmentsto promote sustainable poverty-reducing economic growth inrural areas o f Central America. A central theme ofthis reappraisal is that agriculture can not serve as the sole engine o f poverty reducing growth inthe rural economy, and that a more balanced and integrated multi-sectoral and spatial approach to rural development is needed. Such an approach should consider supply and demand linkages with non-agricultural activities in rural areas, along with rural-urban linkages and migration. There i s a needto recognize the "pluriactive" nature4o f the rural economy (de Janvry and Sadoulet 2000,200 1;Lanjouw and Feder 2001;Haggblade and others 2002). Furthermore, the heterogeneity o f such factors as agro-ecological conditions, access to infrastructure and services, household assets and livelihood strategies, and formal and informal institutions, points to the need for more attention to sub-national areas and households within geographic areas. This requires improved geographic analyses that consider the heterogeneity o f areas and households within areas5 See table 1.1 for highlights o fthe World Bank's new rural development strategy for Central America. 3 See for example: de Janvry and Sadoulet (2000,2001); Echeverria (2001b); Ashley and Maxwell (2001); IFAD(2001,2002); Valdes and Mistiaen(2001); USAID(2001,2002); IFPRI(2002); Richards andothers (2002); OD1(2003). There is also a World Bank-wide rural development strategy (World Bank 2002a). The pluriactive natureof the rural economy refers to the multiple activities that take place inrural areas. See Jalan and Ravallion (1997); Hentschel and others (1998); Wilcox (1999); Bigman and Fofack (2000); IFPRI(2000); de Janvry (2002); de Janvry and others (2002); Demombynes and others (2002); Davis (2002). 5 Four basic "paths" have been identified for reducing rural poverty in Central America (see de Janvry and Sadoulet 2000; de Janvry and others 2002; USAID2002). A fifth potential strategy has also beenidentified, payments for environmental services (CCAD 2002). 0 Agriculturalpath: increasedproductivity and diversification to higher value enterprises. For both: a) commercially oriented small farmers (primarily household employment and income), and b) larger commercial farmers (owner operated with hired labor). 0 Pluriactivepath: focus on off-farm economic activities (including labor on larger farms), and also attempt to generate basic food staples for home consumption. 0 Social assistancepath: both formal and informal assistance including safety nets, transfers, remittances, special targeted programs. 0 Exitpath: migrationout o f rural areas within country and outside country. 0 Paymentsfor environmentalservices: rural residents would receive payments for activities related to natural resource management and environmental quality.6 Despitechallenges to agriculture and the needto promote non-agricultural activities inrural areas, many policies and investmentsthat support agricultural growth also support growth o f the non-agricultural rural economy. Investments intransport and communication infrastructure, education, health, and improvements infactor and output markets can help stimulate agricultural and non-agricultural activities inrural areas. Thus, it would be a mistake to lessen support for agriculture inthe hope that the non- agricultural rural economy could, in o f itself, be the engine o f poverty reducing rural economic growth (Start 2001; de Janvry and Sadoulet 2001; Reardon and others 2001). Identifyingsynergies between agricultural andnon-agricultural activities is key for rural development (see box 1.1). ~ ~~ ~ ~~ ~~ Box 1.1. Sectoral and spatial linkages Start (2001) points out that besides the obvious productionand consumption linkages, there are other importantsectoral (agricultural and non-agricultural) and spatial (rural-urban) links that have a major impacton economic performance. These include: financial capital linkages (financial flows and investments), human capital linkages (educationand skills can be transferred), labor linkages (part-time employment inruralareas, seasonalandpermanent migration), infrastructure and service linkages (transport, communications, power, water), and social capital linkages (social networks, interest groups). ~~ ~ Linkingnatural resource management with poverty reduction, ruralresidents would receive payments for activities such as watershed protection, management o f protected areas, natural forest management, reforestation activities in erosion-prone areas, conservation o f biodiversity, carbon sequestering, soil and water conservation. This strategy could be important for poor indigenous people living in remote areas that are environmentally sensitive. See Varangis and others 2003 for some examples o f payments for environmental services in coffee growing areas inCentral America 4 2. Asset-basedconceptualframework The asset-based approach focuses attention on the productive, social and locational assets o f households, with the understanding that the quantity, quality and productivity o f their portfolio o f assets determine the potential for long-term growth and poverty reduction (see Siege1and Alwang 1999;Deininger and Olinto 2000). As such, household assets are consideredthe "drivers" of sustainable growth and poverty reduction. The asset-based approach can be usedto explore relationships between assets, context, behavior, and outcomes (see de Janvry and Sadoulet 2001). The assets o f a household are broadly definedto include the productive, social and locational assets that determine the opportunity set o f options for livelihood strategies. These actions, in turn, determine outcomes in terms o f household well-being. Of critical importance i s the context, the policy and institutional milieuand the existence or absence o f risks. The welfare-generating potential o f assets depends on the interface between assets and the context. Thus policy reforms and the buildingo f assets need to be considered intandem, and integrated with riskmanagement strategies (IADB 2000). The asset-based approach is well-suited for understanding and analyzing rural poverty in Central America because o f the highly unequal distribution o f assets, high exposure to natural, economic and social risks, and the ongoing economic, political and institutional reforms. Figure 1.1presents the framework graphically. Using an asset-based approach to understand and analyze the rural situation in Central America frames overall development strategies and specific policy and investment alternatives interms o f households' productive, social, and locational assets; how they complement each other, and the specific interventions that can be taken to strengthen and protect their portfolio of assets in order to improve well-being. Given that agriculture can not serve as the sole engine o f rural growth a more balanced spatial and multi-sectoral approach to rural development i s needed.7 This requires a household-level (microeconomic) orientation toward identifyingdrivers o f growth -- which is providedby the asset-based conceptual framework. The asset-based approach underlies the livelihoods approach (Carney and others 1999) and has increasingly been advocated by numerous development agencies.8 Usingthe asset-based conceptual approach, drivers of sustainable rural growth and poverty reduction are evaluatedby focusing on the assets and combinations o f assets neededby different types o f households in different geographical areas to take advantage o f economic opportunities and improve their well-being over time. This study is not, for example, trying to identifyparticular enterprises such as cut flowers, broccoli, snow peas or sub-sectors that might stimulate growth and poverty reduction. Instead our approach 'Heterogeneity o f agro-ecological zones, access to infrastructure and services, formal and informal institutions, etc. between and within the countries indicates that area- or region-specific approaches are more appropriate. 8 An asset-based approach to welfare reform inthe UnitedStates has beenproposed by Sherraden (1991). 5 can be useful to understandthe type and combination of assets that are requiredby households to take advantage of a particular enterprise or development path. 2.1. Components of the asset-basedapproach Assets A household's assets consist of the stock ofproductive, social, and locational resources usedto generate well-being (see Moser 1998; Siege1and Alwang 1999; Rakodi 1999; Carney and others 1999; de Janvry and Sadoulet2001).9 Householdassets are broadly definedto includetangible and intangible resources drawn from individual, household, community, and national and global levels (see table 1.2). According to the asset-basedapproach. the poor are "asset-poor"; they have limitedassets, hold assets with low returns, and/or are unable to exploit their assets effectively. Householdassets includetangible assets such as landand other natural assets, agro-ecological conditions, equipment and other physical assets, livestock, housing, financial assets, human capital and household composition. Intangible assets are also important, such as social capital andpolitical rights, and the capacity and openness of institutions. Inaddition, community and regional assets such as infrastructure and access to it infrastructure affect livelihood opportunities and returns on other assets. Most economic analyses focus on productive tangible assets andhow they generatereturns. One reason i s that data for tangible assets are more easily collected and available. However, there is growing consensus that both tangible and intangible assets, and their interplay, are important, especially inthe context o frisk management. As noted by Narayan and Pritchett (1997), poverty analysesthat focus exclusively on tangible household assets miss a large part of the "poverty puzzle", by ignoring the community and social context. More attention is now beingplaced on social, institutional and political relationships among households within and outside the community; such as gender relations, social ties andnetworks, social cohesion, empowerment, participation in organizations, and effectiveness of collective action (Moser 1998; World Bank 2002~). Physical and social infrastructure complement other assets and help determine the expectedreturns and the risks o fthese other assets. Inaddition, the location o f infrastructure is considered to be a critical asset, becauseit influences the availability and accessibility o f goods and services (van de Walle 2000a, 2000b). There is considerable heterogeneity of households' assets and livelihood strategies inrural areas o f Central America. Differences inagro-ecological conditions have an important impact on opportunities and constraints for rural households, because they determine the potentialfor agricultural and activities linkedto agriculture (FA0 2001). Inmany cases, agro-ecological zones are also highly correlated with other assets ~~~ ~ ~ Also see WorldBank (2000); Attanasio and Szekeley (2001); Wadsworth(2002); Wintersandothers (2002). 6 at the household, community and regional levels (see table 1.2). Areas with low - agricultural potential and located in environmentally sensitive zones often have relatively higher proportions o f indigenous populations and higherpoverty rates, and/or constraints on income generating potential from assets because o f hightransaction costs related to remoteness (Pichon and Uquillas 1999; FA0 2001;World Bank 2002c; Wadsworth 2002). lo The quantity and quality o f assets determine household well-being and growth potential, for a given context. Certain assets are effective onlv ifcombined with others, and their sequencing can also be critical. For example, access to high-quality land has different implications for well-being depending on its location relative to markets and other infrastructure or on access to credit and high-quality inputs.Education has different implications for welfare generation depending on location, the functioning o f labor markets and related institutions. The existence o f good transport and market infrastructure i s essential for successful adoption o f agricultural technology, by lowering transaction costs and opening new trade opportunities. Other important determinants o f asset productivity include the regulatory and legal systems, and social and political inclusion such as human rights. These characteristics are inexorably tied to the context. Context The distribution o f assets among households and communities and their welfare- generating potential depend on the context (past, present, future), which includes exogenous and endogenous factors. The context consists o f the institutions and policies that define ownership and acceptable use o f assets, along with the risks that affect the welfare-generating potential o f assets. The political, legal and regulatory context determines, to a large extent, how households' assets can be managedto achieve well- being (de Soto 2000; Zezza and Llambi 2002). A major factor affecting the context is how institutions at macro, meso, micro levels function, their degree o f inclusiveness, and how they interact. Issues o f governance are critical, and many new initiatives towards decentralization have been stymied by the lack o f governance capacity and skills. Incompetence and corruption are also widespread. Governance issues are important for area comparative advantage and competitiveness and are receiving more attention by development agencies (World Bank 2002~). The issue o f access to assets and to markets i s also closely linkedto the context, where the "rules-of-the-game" are set (e.g., human and property rights, rules and regulations that relate to social and political inclusion, and environmental quality standards and enforcement).' See box 1.2. 10The historical context can not be ignored. Household and community assets have been shaped by history. For example, the pervasiveness o f rural poverty inthe Central American countries is largely the result o f historical factors (see for example, de Janvry and Sadoulet 2000; Attanasio and Szekeley 2001; Morley 2001;Franko 2003; Plataforma Agraria 2003; World Bank 2002~). "Anexampleoftheasset-contextinterfaceistheissueofpropertyrightsforindigenouspeopleandtheir access to natural resources. The larger society has historically decided on indigenous property rights, often denying them rights to capitalize on assets. 7 Box 1.2. The mystery of assets Ina book called The Mystery ofcapital, de Soto (2000) explains the "mystery" of the asset-context interface. "Capital, like energy i s also a dormant value. Bringing it to life requires us to go beyond looking at our assets as they are to actively thinking about them as they couldbe. It requires aprocess for fixing an asset's economic potential into a form that can be used to initiate additional production .... [the] key process was not property ownership. ... Although we use these mechanisms a11the time, we do not deliberately set up to create capital but for the more mundane purpose o f protecting realize that they have capital-generating hnctions becausethey do not wear that label. We view them as parts of the systemthat protects property, not as interlocking mechanisms for fixing the economic potential of an asset in such a way that it can be converted into capital. What creates capital inthe West, in other words, i s an implicit process buried in the intricacies o f its formal property systems (p. 45-46)." The risks to which ruralhouseholds and their assets are exposed are also part o f the context. Risks include climatic factors such as drought, flood andhurricanes, price risks (for outputs and inputs), lack of markets, human health risks, plant and livestock diseases and pest infestations, and risks associated with conflicts and crime affecting personal security (de Ferranti and others 2000; IADB 2000; Keipi and Tyson 2002; Kiesel2001). The presence o f risk often invokes a cost o f risk management; this cost can include lower income due to risk avoidance behavior (opportunity costs) and risk- reducing activities (actual costs), and costs associated with coping activities. Risk also induces fluctuations in consumption which, by themselves, lower household well-being. To a large extent, the context is shaped by factors external (or outside the control) o f households. Domestic and international policies, institutions and markets, and forces o f nature shape the context. On the other hand, households can invest inassets, and allocate their assets and select livelihood strategies in a manner that reduces risks associated with the prevailing context. Furthermore, important links between policies and risks exist, because policies and investments can either increase risk and exposure to risk, or help households better manage risk and vulnerability to poverty (Siegel and Alwang 1999; Anderson 2001;Varangis and others 2002; Siegel and others 2003). Inaddition, social protection and safety nets can help households manage risk (Lustig2001; World Bank 2001; Devereux 2001). The general lack o f risk management instruments for the rural poor in Central America constrains their ability to protect their assets and to generate higher returns from their portfolio o f assets. Despite the low asset base o f many rural poor households, there is still potential to assist them by: increasing the efficiency and use o f their existingassets, increasing the productivity o f their existing assets, providing them with additional assets, by protecting the assets, and different combinations o f these options. This requires investments, policy and institutional reforms, and significant capacity building. It also requires time. The temporal dimensions and dynamics o f asset enhancement and expansion need to be 8 carefully considered in project planning. l2A critical part o f this process includes enhancing and expanding human and social capital, whereby individuals, households and communities learn entrepreneurial and management skills to become empowered inthe newly liberalized and decentralized markets and institutions (Siege1 and Alwang 1999; de Janvry and Sadoulet 2001; Attanasio and Szekely 2001). See box 1.3. Box 1.3. Why the poor are poor: Lack of assets and low asset productivity Explainingwhy the poor are poor in Central America, Valdes and Mistiaen (2001, p.12) cite another study that states: "Most basically it is because they have few assets (both human and physical, including social capital) and also because the productivity of their assets i s low. The assets are meager not only inquantity but also in quality (for example, low levels o f schooling are usually combined with poor quality o f schooling). The low productivity o f assets results from a combination o f government failures and imperfect o f incomplete markets." Valdes and Mistiaen(2001) claim that: "This taxonomy helps in guiding the analysis o f rural poverty determinants by distinguishing those factors that contribute or constrain the buildingo f the assets o f the poor (education, demographics, land, and others) from those influencing the productivity o f such assets (the incentive framework, financial policies, overall economic growth, and others). Traditionally the bulk o f the literature on agricultural development and povertv in Latin America has emphasized control over assets (land inparticular) as the key factor in exulainina rural poverty. Why the "low productivity of assets" effect on rural poverty has beenpractically ignored in a region with such a historv o f poor policies i s puzzling." (Emphasis by author) Livelihood strategies The "opportunity set" (options) for households to achieve different levels of well- being depends on the interface between assets and the prevailing context. Strategic management by a household of its asset portfolio to achieve preferredwell-being outcomes defines its behavior or livelihood strategy (Ellis 1998; Carney and others 1999). Livelihood strategies include: landand labor use decisions, investments in education, migration, participation in social capital buildingand other asset allocations. Different economic and social activities require mobilization of different amounts and types o f assets. Asset holdings determine the ability to undertake a given enterprise and the productivity of resources allocated to that enterprise, while the potential returns depend also on the context. See box 1.4. I'For example, it is importantto consider the short-termcash flow needs for projectsthat include investmentsinassets becauseretumscan take time to materialize. Such is the case with investments in fruit orchardsand livestock, for example. 9 Box 1.4. Defining a livelihood According to Chambers and Conway (1992, pp.7-8): "A livelihoodcomprises the capabilities, assets (stores, resources, claims and access) and activities required for a means o f living; a livelihoodi s sustainable when it can cope with and recover from stress and shocks, maintain or enhance its capabilities and assets, provide sustainable livelihood opportunities for the next generation, and contribute net benefits to other 1ivelihoods.at the local and global levels inthe long and short run." The asset-based approach uses a "livelihood focus" inrecognition that rural households hold a portfolio o f assets and allocate these assets among a range o f welfare- generating activities (Chambers and Conway 1992; Camey and others 1999). The asset- based approach helpsus understand why and how households manage assets and risks to "select" certain livelihood strategies to achieve welfare outcomes; inthe face o f specific asset-context interface conditions (Wadsworth 2002). Livelihood strategies o f rural households in Central American include a wide range o f on- and off-farm agricultural and non-agricultural activities as self-employed or laborers, and migration (temporary and permanent). It has been estimated that about 50% or more o f rural households' income inthe Central American countries come from rural non-farm employment (RNFE) (Berdegue and others 2001;Reardon and Berdegue 2002; Corral and Reardon 2001). Many o f the RNFEjobs are low-skill and low-paying jobs that are not obvious paths out o f poverty. More attention needs to be devoted to understanding households' asset portfolios and allocation o f assets, particularly labor. For example, contrary to long heldbeliefs in the need for labor intensive agricultural technologies, many small farmers with limited land assets really would be better served by labor saving agricultural technologies to free up labor for alternative activities (USAID 2001; Reardon and others 2001; Start 2001). The asset-based approach also focuses on the longer-term implications o f short- term decisions about the allocation o f assets. For example, coping strategies usedby poor rural households can lead to the degradation or decapitalization o f assets such as cutting down trees, taking children out o f school and these actions can contribute to a cycle o f poverty. Alternatively, household livelihood strategies can leadto improved asset portfolios such as investments in improved technology, training programs, empowerment in social and political networks that can lead to a virtuous cycle o f sustainable growth. Asset accumulation and changes inlivelihood strategies are important drivers for sustained improvements in well-being, and this study describes patterns o f asset accumulation and livelihood strategies and investigates their causes. Rural diversification, i s not necessarily poverty reducing. Inmost cases there will be winners and losers, and those minimally affected such as remote communities that are semi-subsistent and autarkic). Infact, there is evidence that past diversification initiatives inCentral America have beenbiasedtowardhigher-potential areas andhouseholdswith stronger asset bases (Tabora 1992).Diversification efforts targeted to lower-potential areas with hightransactions costs such as those that are geographically remote and/or are 10 characterized by disadvantageous agro-ecological conditions, lack o f infrastructure, low humancapital might requirerelatively large per-capita investments intangible and intangible assets, with relatively low returns inthe short-term (Start 2001; Reardon and others 2001). This implies that national and local governments needthe need explicit growth and/or poverty reduction objectives to carry out targeted investments effectively. It also implies the needto recognize potential growth-efficiencv-equity trade-offs. Outcomes Ultimately, we are concerned with outcomes that reflect household well-being and prospects for growth over time.I3 Household well-being i s multi-dimensional, and some o f these dimensions are admittedly very difficult to quantify and measure (World Bank 2000). Since income and consumption are more easily measured and generally correlate well to other indicators o f well-being, they tend to be the primary indicators used, especially in the quantitative analyses. Non money-metric approaches can be found, butusually the indicators are tied back to the concept of poverty usinga money-metric baseline (Glewwe and van der Gaag 1988). However, different measures o f well-being can be used to measure outcomes to reflect economic, social and environmental outcomes that can be material and/or non- material innature. More attention is being devoted inthe development literature to multi- dimensional economic, social and environmental measures o f well-being (Moser 1998; Carney and others 1999; Coudouel and Hentschel2000; Narayan and others 2000). Poor rural households are also concerned about food security, health status, vulnerability in general, empowerment and self-esteem, participation incommunity affairs, environmental quality, and hopefulness towards the future (Carney and others 1999; Narayan and others 2000). Such measures o f well-being are not easy to obtain and quantify, necessitating; the use o f participative methods and qualitative analyses. Poverty tends to be a transitory state for many households and there is a tendency to move above and below the poverty line. That is, many households are vulnerable to poverty because o f changing asset-context conditions and livelihood strategies (Siege1 and others 2003; World Bank 2001). Changes in well-being may be concentrated along certain points o f the distribution o f well-being (not just below the poverty line), and public policy should be designedto improve well-being for broad segments of society - notjust those below or above the poverty line. Because o f these factors, it is important to examine levels and changes inwell-being along the entire distribution o frural households; including poor and non-poor households (Alwang and others 2002). The asset-based approach leads us to consider a variety o fmeasures o f household well-being. It also leads us to use both quantitative and qualitative analyses to better understand the complex relationships between assets-context-behavior-outcomes. This i s l3Inaddition to outcomes relatedto householdwell-being, it is possible to use the assetbase approach to consider how the asset-context-livelihoods interface generate other outcomes, such as environmental impacts that are extemal to the household's well-being (e.g., down-stream pollution) but have an impact on social welfare and sustainability. This requires additional types o f data and analysis. 11 because we needto consider tangible and intangible assets and material and non-material measures o f well-being, in addition to subjective perceptions about opportunities and risks and the selection o f multiplelivelihood strategies. More tangible outcomes are measures o f income/consumption, savings, food security, and nutritional and health status. Intangible measures o f well-being are more subjective, and include perceptions o f self-esteem and empowerment, hope towards the future, and leisure and recreation. 3. Conclusion The asset-based approach is an appropriate conceptual framework for organizing thinking about poor rural households in Central America, and for identifying drivers o f poverty-reducing growth. The asset-based approach considers linkages between households' portfolios o f productive, social and location-specific assets, the policy, institutional and risk context, their behavior as expressed intheir livelihood strategies, and outcomes interms o f well-being. For economic growth to be poverty reducing in a sustainable manner, it i s critical to have a better understanding o f household asset portfolios, and how assets interact with the context to influence the selection o f livelihood strategies which, inturn, determine well-being. Policy reforms can change the context and income-generating potential of assets. Investmentscan add new assets or increase the efficiency o f existinghousehold assets, and also improve households' risk management capacity. Investment priorities and project design influence (and are influenced by) the sequencing and complementarity o f changes inthe asset portfolio. After all i s said and done, a household's asset portfolio will determine whether growth and poverty reduction can be achieved, and sustained over time. The asset-based framework i s amenable to different analytical techniques. It i s suggested to combine quantitative and qualitative spatial and household level analyses (and linkedspatial and household level analyses) to deepen understanding o f the complex relationships between assets, context, livelihoods and outcomes. l4Combining quantitative and qualitative analyses can generate especially interesting insights. l4I t i s suggested to undertake the following analyses descriptive statistical and graphical analyses o f the distribution of assets and incomes among households, GIs-typemapping techniques along with some simple regressions of spatial relationships, quantitative household analysis, participatory qualitative analyses of assets and livelihoods, and participatory qualitative assessments of existing CAESSD projects. 12 References Alwang, J., B. F. Mills,and N.Taruvinga. 2002. Why Has Poverty Increased in Zimbabwe? Washington, D.C.: The World Bank. Anderson, J. 2001. "Risk Management inRuralDevelopment: A Review." Rural Strategy BackgroundPaper 7. RuralDevelopment Department. Washington, D.C.: The World Bank. Ashley, C., and S. Maxwell. 2001. "Rethinking RuralDevelopment." Development Policy Review 19 (4): 395--425. Attanasio, O., and M.Szekeley, eds. 2001. Portrait of the Poor: An Assets-Based Approach. Latin America ResearchNetwork. Washington, D.C.: Inter-American Development Bank: Berdegue, J. A., T. Reardon, and G.Escobar. 2001. "The Increasing Importance of Nonagricultural Rural Employment andIncome." In. R. Echeverria, ed., Development of Rural Economies. Washington, D.C.: Inter-American Development Bank. Bigman, D., and H.Fofack. 2000. "Geographical Targeting for Poverty Alleviation: An Introductionto the Special Issue." The WorldBank Economic Review 14 (1): 129- -45. Camey, D.,M.Drinkwater, T. Rusinow, K.Neefjes, S. Wanamali, and N. Singh. 1999. "Livelihood Approaches Compared: A BriefComparison of the Livelihoods Approaches of DFID, CARE, Oxfam, and UNDP." London: Department for Intemational Development (DFID). CCAD (Central American Development and Environmental Commission). 2002. Nature, People and Well-Being: Mesoamerica Fact Book. Antiguo Cuscatlan, ElSalvador: CCAD. Chambers, R., and G. Conway. 1992. "Sustainable Rural Livelihoods: Practical Concepts for the 2IStCentury." IDSDiscussion Paper 296. Brighton, UK:Institute for Development Studies. Corral, L., and T. Reardon. 2001. "Nonfarm Incomes inNicaragua." WorldDevelopment, 29 (3): 427--42. Coudouel, A., and J. Hentschel. 2000. "Poverty Data and Measurement.'' Preliminary draft for A Sourcebook on Poverty Reduction Strategies. Washington, D.C.: The World Bank: Davis, B.2002. "Is it Possible to Avoid a Lemon? Reflection on Choosing Poverty Mapping Method." Rome: Food and Agricultural Organization. 13 Deininger,K.,and J. S. Chamorro. 2002. "Investments and Income Effects ofLand Regularization: The Case o fNicaragua." Policy Research Working Paper 2752. Washington, D.C.: The World Bank. Deininger, K.,and P. Olinto. 2000. "Asset Distribution, Inequality, and Growth." Policy ResearchWorking Paper 2375. Washington, D.C.: The World Bank. Deininger, K.,E.Zegarra, andI.Lavandez. 2003. "Determinants andImpacts of Rural LandMarket Activity: Evidence from Nicaragua." WorldDevelopment, 31 (8): 1385--1404. DeFerranti, D., G. E.Perry, I.S. Gill, andL.Serven. 2000. Securing our Fulure in a Global Economy. Washington, D.C.: The World Bank. D e Ferranti, D., G. E. Perry, F. H. G. Ferreira, and M.Walton. 2004. Inequality in Latin America: Breaking with History? World Bank LatinAmerican and Caribbean Studies. Washington, D.C.: The World Bank. D e Janvry, A. 2002. "Geography of Poverty, Territorial Growth and Rural Development." Paper presentedto USAID/LAC RuralEconomy Workshop: PromotingRuralProsperity, February 4, Washington, D.C. See www.ruralprnsperi tv.com/presentations.cfm. DeJanvry, A., andE. Sadoulet.2000. "Rural Poverty inLatinAmerica: Determinants and Exit Paths." FoodPolicy 25(4): 389--409. ---------. 2001. "Investing inRuralDevelopment i s GoodBusiness." InR.Echeverria, ed., Development of Rural Economies. Washington, D.C.: Inter-American Development Bank. D e Janvry, A., E. Sadoulet, and C. Araujo. 2002. "Geography of Poverty, Territorial Growth and Rural Development" Paper presented at Annual World Bank Conference on Developing Economies Economists' Forum. April 30, 2002. The World Bank, Washington, D.C. De Soto, Hernando. 2000. TheMystey of Capital: Why Capitalism Triumphs in the West and Fails Eveywhere Else. New York: Basic Books. Demombynes, G., C. Elbers,,J. Lanjouw, P. Lanjouw, J. Mistiaen, and B. Olzer. 2002. "Producing an Improved Geographic Profile of Poverty: Methodology and Evidence from Three Developing Countries." WIDER Discussion Paper 2002/39. Helsinki:World Institute for Development Economics Research. Devereux, S. 2001. "Livelihood Insecurity and Social Protection: A Re-emerging Issue in Rural Development." Development Policy Review 19 (4): 507-- 19. 14 Echeverria, R. 2001a. "Options for Investing inthe Rural Economy" InR. Echeverria, ed., Development of Rural Econoinies. Washington, D.C.: Inter-American Development Bank. Echeverria, R., ed. 2001b. Development of Rural Economies. Washington, D.C.: Inter- American Development Bank. Ellis,F. 1998. "Household Strategies and Rural Livelihood Diversification." Journal of Development Studies 35 (1): 1--38. FA0 (Food and Agricultural Organization). 2001, Farming SystemsandPoverty: Improving Farmers 'Livelihoods in a Changing World. Produced with The World Bank. Rome: Food andAgricultural Organization. ---------. 2002. "Committee on Commodity Problems: Report of the Consultation on Agricultural Commodity Price Problems" March 18-21, Rome. Franko, P. 2003. ThePuzzle of Latin America Economic Development. Rowman & Littlefield, Inc.: Blue Ridge Summit, PA, USA. Glewwe, P., and J. van der Gaag. 1988. "Confronting Poverty inDeveloping Countries: Definitions, Information and Policies." Living Standards Measurement Study WorkingPaper 48. Washington, D.C.: The World Bank. Haggblade, S., P. Hazell, and T. Reardon. 2002. "Strategies for StimulatingPoverty- Alleviating Growth inthe RuralNon-farm Economy inDeveloping Countries." Environment and Production Technology Division, EPTD Discussion Paper 93. Washington, D.C.: International FoodPolicy Institute. Hentschel, J., J. 0.Lanjouw, P. Lanjouw, and J. Poggi. 1998. "Combining Census and Survey Data to Study Spatial Dimensions of Poverty." Policy ResearchWorking Paper 1928. Washington, D.C.: The World Bank. Hereford, R., and R. Echeverria. 2003. "Pobreza Ruralen Centroamerica." Report RUR- 03-102. Washington, D.C.: Inter-American DevelopmentBank. IDB(Inter-American Development Bank). 2000. SocialProtectionfor Equity and Growth. Baltimore: Johns Hopkins Press. IFAD (International Fundand Agricultural Development). 2001.Rural Poverty Report 2001: The Challenge of Ending Rural Poverty. Rome. ---------.2002. Enabling the Rural Poor to Overcome their Poverty: Strategic Framework 2002-2006. Rome. 15 IFPRI (IntemationalFood and Policy ResearchInstitute). 2000. "Promoting Sustainable Development in Less-Favored Areas." Focus 4. Washington, D.C.: IFPRI. ---------.2002. "Mesoamerican Representatives andIFPRl Board Draw Up TO-DOList for the Region." Washington, D.C. Jalan, J., and M.Ravallion. 1997. "Spatial Poverty Traps?" Policy ResearchWorking Paper 1862. Washington, D.C.: The World Bank. Keipi, K., andJ. Tyson. 2001. "Planning andFinancial Protection to Survive Disasters." Sustainable Development Department, Technical Studies Series No. ENV-139. Washington, D.C.: Inter-AmericanDevelopment Bank. Kiesel, C. 2001. GuiaPara La Gestibn Del Riesgo: En Proyectos de Desarrollo Rural. Series de Publiciones RUTA. Working Series Document No. 4. San Jose, Costa Rica: Regional Unit for Technical Assistance (RUTA). Lanjouw, P. 2001. "Nonfarm Employment and Poverty inRural El Salvador." World Development 29 (3):529--47. Lanjouw, P., and G.Feder. 2001. "Rural Non-Farm Activities and RuralDevelopment: From Experience Towards Strategy." Rural Strategy BackgroundPaper 4, Rural Development Department. Washington, D.C.: The World Bank. Lustig, Nora. 2001. Shielding the Poor: Social Protection in the Developing World. Washington, D.C.: Inter-AmericanDevelopment Bank and Brookings Institution Press. Morley, S. A. 2001. TheIncome Distribution Problem in Latin America and the Caribbean. Santiago, Chile: Economic Commission for LatinAmerica and the Caribbean (ECLACKEPAL). Moser, C. 1998. "The Asset VulnerabilityFramework: ReassessingUrban Poverty Reduction Strategies." WorldDevelopment 26 (1): 1--19. Narayan, D., and L.Pritchett (1997) "Cents and Sociability: Household Income and Social Capital inRural Tanzania." Policy ResearchWorking Paper No. 1796. Washington, D.C.: The World Bank: Narayan, D., R. Patel, K. Schafft, A. Rademacher, and S. Koch-Schulte. 2000. Voices of the Poor: CanAnyone Hear Us? New York: Oxford University Press for the World Bank. OD1(Overseas Development Institute). 2003. "Options for Rural Poverty Reductionin Central America." OD1BriefingPaper. London: Overseas Development Institute. See www.odi.org.uk. 16 Pichon, F., and J. Uquillas. 1999. "Rural Poverty Reduction and Improved Natural Resource Management Through Participatory Technology Development inLatin America's Risk-Prone Areas." InF. Pichon, J. Uquillas, and J. Frechione, eds., Traditional and Modern Natural Resource Management in Latin America. Pittsburgh: University of Pittsburgh Press. Plataforma Agraria. 2003. Abriendo Brecha: UnaPropuesta de Desarrollo Rural. Guatemala City. Rakodi, C. 1999. "A Capital Assets Framework for Analyzing Household Livelihood Strategies: Implications for Policy." Development Policy Review. 17 (3):3 15-42. Reardon, T., and J. Berdeque.2002. "The RapidRise o f Supermarkets inLatinAmerica: Challenges and Opportunities for Development." Development Policy Review 20 (4): 371--88. Reardon, T., J. Berdegue, and G.Escobar. 2001. "Rural Nonfarm Employment and lncomes inLatinAmerica: Overview and Policy Implications." World Development 29 (3):395--409. Richards, M.,E. Baumeister, I. Colindres, M.Laforge, M.Lopez, S. Maxwell, H.Noe Pino, P. Sauma, and I. Walker. 2002. "Exploring the Options for Rural Poverty Alleviation in Central America." DFID-RUTA-ODI, London. Processed. RubenR., andM.van den Berg. 2001. "Nonfarm Employment andPoverty Alleviation o f Rural Farm Households inHonduras." WorldDevelopment 29 (3): 549--60. Sauma, P. 2002. "La Pobreza en Centroamerica en 10s Noventa." Publicationprepared for Regional Unit for Technical Assistance (RUTA), San Jose, Costa Rica. Sherraden, M. 1991.Assets and the Poor: A New American Welfare Policy. New York: M.E.Sharpe, Inc. Siegel, P.B. 2005. "Using an Asset Base Approach to IdentifyDrivers o f Sustainable Rural Growth and Poverty Reductionin Central America: Conceptual Framework." Agriculture and Rural Development Department, Latin America and the Caribbean Regional Office. Policy Research Working Paper Series No. WPS 3475. Washington, D.C. : The World Bank. See www.worldbank.org/research. Siegel, P. B., and J. Alwang. 1999. "An Asset-Based Approach to Social Risk Management: A Concept Approach." Social ProtectionDiscussion Paper 9926. Social Protection Unit,Human Development Network. Washington, D.C.: The World Bank. See www.worldbank.org/sp 17 Siegel, P.B, J. Alwang, and S. Jorgensen. 2003. "Rediscovering Vulnerability Through a Risk Chain: Views from DifferentDisciplines." Quarterly Journal of International Agriculture 42 (3):35 1--70. Start, D. 2001. "The Rise and Fall of the RuralNon-Farm Economy: Poverty Impacts and Policy Options." Development Policy Review 19 (4): 491--505. Tabora, Jr., P. C. 1992. "Central America and South America's Pacific RimCountries: Experience with Export Diversification." InS. Barghouti, L.Garbus, and D. Umali, eds., Trends in Agricultural Diversification: Regional Perspectives. World Bank Technical Paper 180. Washington, D.C.: The World Bank. Tejo, P. 2000. La Pobreza Rural UnaPreocupacidn Permanente en Pensamiento de la CEPAL. Santiago, Chile: Economic Commission for Latin America and the Caribbean (ECLACICEPAL). USAID(U.S. Agency for International Development). 2001. "USAID Bureaufor Latin America and the Caribbean Strategic Concept Paper: Rethinking the Rural Economy inLAC.'' Concept paper. Washington, D.C. Processed. ---------.2002. "Rural Prosperity inthe Latin American and Caribbean Region: A USAID/LAC White Paper." Paper presented to USAIDiLAC Rural Economy Workshop: PromotingRural Prosperity. February 4, Washington, D.C. See www.ruralprosperitv.com/presentations.cfm. Valdes, A., and J. A. Mistiaen. 2001. "Rural Poverty inLatin America: Recent Trends and New Challenges." InK. Stamoulis, ed., Food, Agriculture andRural Development: Current and Emerging Issuesfor Economic Analysis and Policy Research. Rome: Food and Agriculture Organization. Van de Walle, D.2000a. "Are Returnsto Investment Lower for the Poor?Human and Physical Capital Interactions inRuralVietnam." Policy ResearchWorking Paper 2425. Washington, D.C.: The World Bank. -_-____-_ .2000b. "Choosing Rural RoadInvestments to Help Reduce Poverty." Policy Research Working Paper 2458. Washington, D.C.: The World Bank. Varangis, P., D.Larson, and J. Anderson. 2002. "Agricultural Markets and Risks: Management of the Latter, Not the Former." Policy ResearchWorking Paper 2793. Washington, D.C.: The World Bank. Varangis, P., P. Siegel, D. Giovanucci, and B.Lewin. 2003. "Dealing with the Central America Region Coffee Crisis: Impacts and Strategies." Policy ResearchWorking Paper 2993. Development ResearchGroup. Washington, D.C.: The World Bank. 18 Vreugdenhil, Daan, Jan Meerman, Alain Meyrat, LuisDiego Gomez, and Douglas J. Graham. 2002. "Map of the Ecosystemso f Central America: Final Report." Washington, D.C.: World Bank. Wadsworth, J. 2002. "Rural Livelihoods inCentral America: Findings of a Dialogue in Nicaragua and Honduras." Presentation at the World Bank, October 16. Wilcox, B. 1999. "Rural Development and Indigenous Resources: Toward a Geographic- Based Assessment Framework.'' InF. Pichon, J. Uquillas, and J. Frechione, eds., Traditional and Modern Natural Resource Management in Latin America. Pittsburgh: University of Pittsburgh Press. Winters, P., B.Davis, and L.Corral. 2002. "Assets, Activities and Income Generation in Rural Mexico: Factoring in Social and Public Capital." Agricultural Economics 27(2): 139--56. World Bank. 2000. WorldDevelopment Report 2000/1: Attacking Poverty. Washington, D.C.: The World Bank. ---------. 20001. SocialProtection Sector Strategy: From Safety Net to SpringBoard. Washington D.C. ---------. 2002a. Reaching the Rural Poor. Washington, D.C. ---------.2002b. Reaching the Rural Poor in the Latin America and Caribbean Region. Washington, D.C. ---------. 2002c. WorldDevelopment Report 2003: Sustainable Development in a Dynamic World: Transforming Institutions, Growth and Quality of Life. Overview Draft Washington, D.C. ---------.2002d. "Nicaragua: PromotingCompetitiveness and Stimulating Broad-based Growth inAgriculture." Report 25115-NI. Environmentally and Socially Sustainable Development Unit,Washington, D.C. ---------.2002e. LaVentanaNewsletter. Volume 1. Latin American and Caribbean Social Development Unit,Washington D.C. See www.worldbank.orn/afrolatin. _____c---.2003. "Poverty inGuatemala." Report 24221-GU. LatinAmerica andCaribbean Region. Washington, D.C. World Bank and CCAD (Comision Centroamericana de Ambiente y Desarrollo). 2001. "Ecosystems of Central America (ArcView RegionalMap Files at 1:250,000)." World Bank, CCAD, World Institute for Conservation and Environment (WICE), and the Centro Agronomic0 Tropical de Investigaciony Enseiianza (CATIE), Washington, D.C. See http://www.worldbank.org/ca-env. 19 Zezza, A., and L. Llambi.2002. "Meso-Economic Filters Along the PolicyChain: Understandingthe LinksBetweenPolicy Reforms andRuralPovertyinLatin America." WorldDevelopment 30 (11): 1865-44. 20 Table 1.1. Toward a New Rural Development Strategy for Central America A recent World Bank report on rural development inCentral America highlights the following strategies and actions to achieve poverty reducing growth in Central America (World Bank, 2002b): Need for "right" mix o f policies, institutions and support programs to improve the investment climate Macroeconomic and trade policies, sectoral policies and programs, and good governance are key to competitiveness Agricultural growth i s critical for rural development, but there are many non- agricultural rural activities that are critical for both agricultural and rural development Rural-urban dynamics and linkages, including migration needto be considered Needfor a regional development perspective (Le., "rural space approach") based on a "new institutionality" with decentralization andmore localparticipation, and the inclusion o fmarginalized groups Increasedproductivity, competitiveness and private sector development are (or at least should be) key engines of growth Need for efficient and equitable access to product and output markets Need to buildhuman and social capital by expanding delivery o f education, health and nutrition services targeted to the rural poor Needto broadenand strengthen risk management and to provide social assistance and safety nets for the poor and vulnerable Need to manage natural resources in a sustainable manner through better incentives, regulations, and enforcement 21 . Table 1.2. Household-Level Assets and Linksto Other Levels Micro Level Meso Level Macro Level Regional, National, International Level "Private" land, pasture, "Common" land, pasture, National and Global forests, fisheries, water: forests, fisheries, water commons, rivers and quality and quantity watersheds, lakes, seas, oceans, air Human HHcomposition and size Labor pool Labor markets Health and nutritional status Education and skills Physical Productive assets (tools, Productive assets Productive assets (rental equipment, work animals) (communal and private) markets) Household assets (e.g. Stocks (e.g., livestock, Stocks (e.g., buffer stocks) housing, household goods food) and utensils) Stocks (e.g., livestock, food, jewelry) Financial Cash, savings, access to Cash, savings, access to Finance and insurance credit, and insurance markets credit and insurance systems markets Access to international finance Social HHsocial ties andnetworks Community social ties and Extra-community social ties Intra-household dynamics networks and networks 1Location and Proximity and access to Water and sanitation, Distance to markets, Infrastructur water and sanitation, schools, health centers, transportation, e education and health, marketplace, storage communication, informati.on marketplace, storage, roads facilities, roads systems Proximity to transport and Health and education communication infrastructure infrastructure Political and Participation inhousehold Participation incommunity Political stability Institutional decisionmaking (including decision-making Political participation power relationships related to Governance Effectiveness o f collective gender and age) Security o f person and action property Governance Human rights and security o f person and property From: Siege1 andAlwang (1999). 22 * 2 c) -: 0 0 G O G O O *e > Lcd 3 i 3 9o 5 v1 2 Y E 4- 0 cd E c cd 5 0 0 0 0 0 0 0 0 0 0 0 0 0 c, Q) m + c, 0 3 E =3 c, 0 k P 0 P 0 0 0 0 aaJ c, 2 0 0 0 0 3 0 0 0 0 0 0 0 0 APPENDIX 2 Problemswith MeasuringRuralPoverty in HonduraslS Hans G.P. Jansen (IFPRI-RUTA) l5 Prepared for the World Bank research project, Drivers o f Sustainable Rural Growth and Poverty Reduction in Central America, Honduras Country Case Study. The National Statistical Institute (INE) carries out periodic national surveys o f households. These surveys (called Encuestu Permunente de Hogures de Propdsitos Multiples) contain only limited information from the households (the survey instruments are generally less than 5-7 pages in length) but have wide coverage o f households. Rural households tend to be under-represented in these national surveys and the fact that own- consumption i s not part o f total income o f these households almost certainly leads to underreporting o f rural incomes. INE recently produced a municipality-level poverty map based on an application o f a statistical relationship betweenhousehold income and a number of welfare indicators obtained from the September 2003 version o f the national survey and applied this relationship to the 2001 population census data (INE 2003a). Although the methodology usedto imputerural poverty is sound, the level o f accuracy o f the poverty map is likely to be less than optimal due to the above mentionedproblems with rural income data. Inthe analysis ofthe IFPRIand Wisconsin survey data, we defined total household income as the sum of the net value o f crop and livestock production (revenues minus costs), off-farm salaried work (either farm or non-farm), own business and transfers. Own production, whether consumed by the household or sold, i s included in the calculation o f household income inthe IFPRIand Wisconsin surveys. Because it does not include own-consumption and i s likely to miss important components o f income inrural areas, their estimates are substantially below our own. They have a much narrower range as well (see table 2.1). While the IFPRI and Wisconsin household survey data allow for a more precise and generally more reliable calculation o f household income, their disadvantage i s their partial coverage that does not allow the construction o f a representative poverty map. The national level statistically representative LSMS household-level survey, which i s scheduled to start in the second half of 2004, will generate more accurate information on rural poverty. See World Bank (1994) for additional details about problems measuring poverty inHonduras. 25 Table 2.1. Comparison between income estimates (USD/capita/day) Comayagua S. Jose del Potrero 0.19 26 St Barbara Azacualpa 1.65 0.11 Yoro Yorito 0.28 0.12 Sources: Own income estlmates based on IFPRIand Wisconsin household survey data; and INE (2003). 27 References INE (NationalStatistical Institute).2003a.Estimucion de Indicadores de Pobrezay Desigualdad u Nivel Municipal en Honduras. Tegucigalpa, Honduras:INE. World Bank. 1994."Honduras Country EconomicMemorandum/PovertyAssessment." Report 13317-HO. LatinAmerica and CaribbeanRegion.Washington, D.C.: The WorldBank. 28 APPENDIX 3 Agricultural Sector Information16 Hans G.P. Jansen (IFPRI-RUTA) l6 Prepared for the World Bank research project, Drivers of Sustainable Rural Growth and Poverty Reduction inCentral America, Honduras Country Case Study. Table 3.1. Shares (YO)of GrossValue Added in the Agricultural Sector Source: Own calculations based on SAG (2002) for 1980 and 1985, and ECLAC (2004). Table 3.2. Change in PurchasingPower by Agricultural Sub-sector 1978 to 2000 Crop production Rice 29.6 6.7 - 77 18 - 96 * (6) =[((4) + 100) (5)/100] - 100 * Sources: (2) and (3): Cotty et al. (2001); (5): Walker and Ptno (2002); (4) and (6): own calculations. 30 References Cotty, D., M.Garcia, I.Estrada, and E. Anchundia.2001. "Indicadores Basicos Sobre el Desempeiio Agropecuario 1971--2000." Proyecto de Informacion Agricola y Analisis de Politicas (Zamorano -USAID).Tegucigalpa, Honduras : Escuela Agricola Panamericana (Zamorano) y Instituto Nacional de Estadistica (INE). ECLAC (Economic Commission for Latin America and the Caribbean). 2004. Informacidn basica del sector agropecuario. Subregion Norte de Amirica Latina y el Caribe, 1990--2002. Mexico D.F.,Mexico: ECLAC. SAG (Secretaria de Agricultura y Ganaderia). 2002. Compendio Estadistico Agropecuario 2001. Unidad de Planeamiento y Evaluacion de Gestion (UPEG). Tegucigalpa, Honduras: SAG Walker, I., H.N.Pino. 2002. "Desarrollo Ruraly Pobreza enHonduras y Nicaragua: queand sigue? Estudio de Caso: Politicas Estrategias y Acciones en Desarrollo Ruraly Reduccion de Pobrezaen Honduras." Paperpresentedat the DFID workshop, Taller Sobre Politicas y Reduccion de la PobrezaRural-Enfocando el Dialogo Sobre Experiencias en Nicaragua y Honduras, May 29--30, Tegucigalpa, Honduras. 31 APPENDIX 4 SpatialAnalysis of RuralEconomicGrowthin Hondurasl7 Jeffrey Alwang (Professor, Virgina Tech) DavidWoodall-Gainey (Consultant, Virgina Tech) Paul Siege1(Consultant,World Bank andFAO-CP) Hans G.P. Jansen (IFPRI-RUTA) ~~ ~ 17The findings o f this background paper are used as inputs inthe World Bank research project, Driverso f Sustainable Rural Growth and Poverty Reduction in Central America, Honduras Country Case Study. The authors thank the staff o f the SistemaNacional deInjormacion Territorial(SINIT) at the Honduras Rural Land Management Project and InfoAgro, the Geographic Information Systems Unit at the Secretariade Agriculturay Ganaderia(SAG) for access to the maps and data, andinsightful comments and suggestions. 1. Introduction Because o f varied topography, limitedroad coverage and historical investment patterns that favored certain areas, economic opportunities inrural Honduras are heavily influenced by geographic space. As with other countries in Central America, growth and development have been concentratedin areas where agro-ecological conditions are favorable for exporting agricultural products (the so-called "T o f Development").'* Economic potential and area comparative advantage are, to a large extent, determined by spatial assets. As a result, the spatial distribution o f natural, human and physical assets i s o f interest for policy design and investment strategies. If location-specific and infrastructure assets contribute to economic growth, then growth-oriented investments should be targeted toward areas with more favorable natural assets. On the other hand, if poverty rates are higher inmore isolated areas with less infrastructure and longer travel times to markets and population centers, then growth-oriented investmentsmightbypass such areas, exacerbate regional inequalities, and increase social problems. This background paper is designed to address these issues and set the stage for a more in-depth analysis o f determinants o f growth and poverty reduction inrural areas o f Honduras. It i s descriptive innature and i s intendedto provide an overview o f the spatial organization of the rural Honduran economy. We beginby showing the spatial distribution o f population compared to the distribution o f transportation infrastructure. As expected, more densely populated areas are also those areas with better road and other forms o f infrastructure such as sanitation. We compare these distributions with the spatial distribution o f agricultural potential and derive zones o f economic growth potential. We examine the spatial distribution of outcomes interms o f poverty and food insecurity to understand the relationship between population density, growth potential and these outcomes. 2. Data and Methods Data for this spatial overview come from a variety of sources, with the major ones being the SistemaNacional deInformacidn Territorial (SINIT), which was set up under the World Bank's PAARProject andInfoAgro (The Secretariat of Agriculture's geographic information systems unit). These data are supplemented with data from the 1988 and 2001 population censuses, and maps from the vulnerability assessment conducted by GoH together with the World Food Program (WFP). Insome cases, new maps were created using standard GIS techniques. Maps were created by overlaying variables, or by generating new variables. An example o f a map created by overlaying variables i s the map generated by overlaying transportation infrastructure and population centers. An example o f a new variable is the map o f population changes by county (municipio), for which we calculated the population changes from 1988 to 2001 and then mapped the growth rates by county. Insome cases, we usedexisting maps such as the GoH/WFP maps o f vulnerability to food insecurity and for climatic risks. ''Hondurasis not unique inthis respect: Roadnetworks andother infrastructure have been built to support production incoffee and banana-producing areas o f Guatemala, in coffee and cattle-producing areas of Nicaragua and in other high export-potential areas around the region. 33 3. Overview of Rural Space in Honduras Honduras i s the second-largest country in Central America, comprising about 112,000 square kilometers. Except for the far Eastern region o f Gracias a Dios, Honduras i s almost entirely mountainous. About 80% o f the country's western land area consists of interior highlands or hillside areas, with the remaining 20% classified as lowland valleys (IICA 1999). Within the interior highlands, numerous flat-floored valleys are mainly used for extensive livestock operations. Hillside areas are dominated by subsistence agriculture, staple food production and characterized by small landholdings, low levels of technology, and low productivity." Flat-floored valleys, 300 to 900 meters inelevation, are scattered throughout the interior highlands. The valleys provide sufficient grass, shrubs, and dry woodland to support livestock and, insome cases, commercial agriculture. Subsistence agriculture has been relegated to the slopes, with small-sized holdings, low levels o f technology, and low productivity. Villages and towns, including the capital, Tegucigalpa, are tucked inthe larger valleys (see http://countrystudies.us/honduras/34.htm). Honduras i s politically dividedinto provinces (departamentos), which in turn consist of a number o f counties2' (figures 4.1 and 4.2). The country i s unique among Central American countries in that a substantialportion o f its urban population i s shared among two large urban centers: the capital, Tegucigalpa, located inthe south-central part o f the country, and San Pedro Sula inthe northwest. These two citiesjointly account for more than 50% o f the urban population. Other population centers with more than 100,000 inhabitants are scattered around the country (figure4.3). Ingeneral, Honduras has a relatively low population density of less than 60 peopleikm". However, given the mountainous nature o f the country, the number or people per unit of arable landtends to be much higher2'.The most densely populated areas are those surrounding Tegucigalpa and San Pedro Sula, along the Western border with Guatemala and the South-Western border with El Salvador (figure 4.4). Much of the rural population resides inthe fertile valleys, so that intra-municipio variability inpopulation density (and location-specific assets) is high. The most densely populated lowland area i s the Rio Ulua valley inthe province of Cortes (on the bank of the Rio Ulua). Incontrast, Gracias a Dios inthe eastern region o f the country i s much less densely populated. l9 The westernmountains have the highest peaks, with the Pic0 Congolon at an elevation o f 2,500 meters and the Cerro de Las Minas at 2,850 meters. These mountains are covered mainly with pine forests and merge with the Guatemalan mountains at the western border. The eastemranges, which eventually merge with the Nicaraguan border, possess some highpeaks, such as the Montaiia de la Flor, ElBoqueron, and Pic0 Bonito. 20Theaggregatenature ofthe spatial analysis can mask substantialintra-regiondifferences. We use the administrative regions here; later we examine differences by agricultural regions. The latter cross administrative borders, but are useful means o f grouping by productionpatterns, livelihoodstrategies, and agricultural systems. 21 Of the 11.2 million hectaresof land, only 1.7 million (about 15%) is well-suited for agriculture. The Honduran government and two banana companies -- Chiquita Brands Internationaland Dole Food Company -- owned approximately 60 percent of Honduras's cultivable land in 1993. 34 Public investments inhuman and physical assets in Honduras have been skewed towards the 55 counties inthe "T o f Development." These counties also have relatively good natural assets, so investmentsthere have been based on growth potential. Outside the T, public investments (particularly road networks and other infrastructure) have been concentrated where agro-ecological conditions are favorable for export agriculture such as coffee (concentrated on small and medium-sized farms inthe west) and bananas (mostly on large plantations innorthern valleys). Most other rural areas, the hillsides in particular, are excluded from the T o f Development. While this policy may have been rational to the extent that investments inphysical assets were neededto exploit comparative advantages inareas with favorable natural assets, it has also resulted in poverty beinghighest and deepest outside the T. Road networks stretch inmany directions out Tegucigalpa with the most extensive stretch o f road following the valleys betweenthe capital and San Pedro Sula. Another major roadnetwork heads south out o f Tegucigalpa to the Gulf o f Fonseca near Choluteca. Another major roadnetwork runs eastward through the coffee producing areas near El Paraiso22.The roadnetwork running parallel to the Guatemala border between San Pedro Sula and Santa Rosa de Copan serves the densest area o f coffee production in the country (figure 4.6). The north coast, which contains significant agricultural productive potential, i s served by a major road as well. Areas of low populationdensity inthe east are, as expected, underserved by transportation infrastructure; Gracias a Dios has no major highways. Roaddensity23is a measure o f access to markets that may be associatedwith economic growth potential. InHonduras, road access i s highest in areas around Tegucigalpa, near the coffee-producing areas along the Guatemalan border and south toward the Gulf o f Fonseca (figure 4.7). Other highroad-density areas correspond to productive agricultural regions (e.g. the stretch along the Caribbean coast). Incontrast, the easternparts ofthe country have very low densities, inpart due to their low, population densities. An important contrast emerges in figure 4.6 where the eastern half of Honduras is seen to have very low road densities while the western halfhas uniformly higherdensities. This result mirrors the distributionofpopulationwherethe easternhalf has much lower population densities than the south (figure 4.4). I t also shows a constraint to growth inthe east due to lack o f infrastructure. Population change between the 1988 and 2001 censuses didnot follow a regular spatial pattern (see figure 4.8 and table 4.1). Areas near Tegucigalpa, San Pedro Sula and near the Rio Ulua grew faster than the nationalaverage. This fast growth is due to their proximity to urban centers and the highly productive valley agriculture inthe latter region. Eastern areas also grew at rates muchfaster than the nationalaverage, mostly due to their low base o f population in 1988 and increasing in-migration from other areas. Coffee-producing areas near ElParaiso and Morazan and near the Guatemalanborder grew more slowly than the national average, probably due to impacts o f the coffee crisis. 22A more detailed description o f agriculturalproduction regions is found below. 23Kilometers ofclass Ior I1highways divided by the land area measured inkm2. 35 Box 4.1. Migration and population change Rural-to-urban migration has been a strong source o f demographic change in Honduras. Much o f the rural-to-urbanpopulation shift was the result o fmigration froin the southwestem provinces (Ocotepeque, Lempira, Intibuca, La Paz, and Valle) to cities inthe provinces on or near the Caribbean coast (Cortes, Yoro, Atlantida, and Colon) and to Tegucigalpa. Duringthe early twentieth century, employment opportunities in newly established banana plantations attracted many people from southem and westem Honduras to the Caribbean coast. Cities on the banks o f the Rio Ulua, especially ElProgreso, experienced impressive growth as a result o f this migrationfrom the south. Migration from the mountainous southwest sparked development in San Pedro Sula. The search for employment also led many to move to Tegucigalpa, even though the capital has neverbeen a center for industryor agriculture. Inthe second halfo f the twentiethcentury, Tegucigalpa experienced major population increases. These increases have placed strains on already fragile infrastructure. Housing i s inadequate, and many residents either lack runningwater or receive inadequate amounts at irregular intervals. Theft o f electricity service i s common, as i s squatting. Between 1950 and 1980, San Pedro Sula had a population growth rate that exceeded that of Tegucigalpa. Inthe 1980s,the annual growth rate dropped somewhat and was less than that o f Tegucigalpa (3.7 and 4.4%, respectively). San Pedro Sula has dealt more successfully with its population growth, but it i s still challenged to meet the housing, services, and employment needs o f new inhabitants. Other urban centers experiencing a highpopulation growth rate are L a Ceiba, on the Caribbean coast, and El Progreso, inthe agricultural valley o f the Rio Ulua. Most male migrants gravitate toward agricultural areas, especially the Caribbean coast. Women traditionally have a more limited choice o f employment, and their occupational skills are similarly limited. On the other hand, there i s a tradition o f families migrating to seek agricultural labor, with wives and children also working. Many women migrants seek domestic employment or work as street vendors in urban areas. Since the early 1990s, an increasing number o f women have been seeking employment in the maquiladoras, or textile assembly factories. Male urban migrants seekjobs in artisan shops, with merchants, and as laborers. Source: htta://countrvstudies.us/honduras/34.htm A simple regression ofthe determinants ofpopulation change sheds some light on the sources of growth (table 4.2). A higher level o furbanizationin 1988 was associated with higher rates o fpopulationgrowth between 1988 and 2001. Populationdensity in 1988 had a negative impact on 1988-2001 growth, but the effect i s not statistically significant. Once we control for the degree o f urbanization inthe county, access to roads does not help explain growth. However, interestingrelationships between access to land and agricultural opportunity emerge inthe data. Increased landlessness and higher proportions of smallholder family farms with more than 7 ha both contribute to slower 36 rates o f growth at the county level. People are actually migrating out o f such areas (see box 4.1). Areas with higher proportions of people engaged inthe agricultural labor market grew at a higher rate. Areas with lower literacy also grew more quickly. Thus, migration i s taking landless people away from the more crowded agricultural lands, and migrants are moving toward areas where agricultural work i s more abundant. Counties in the Northern Agricultural Region (more details on this below) exhibited higher rates o f population growth, holding other factors constant. Coffee-producing areas grew more slowly or lost population over the period. Changes in rural households over time An investigation o fchanges inhousehold well-being inrural Honduras between 1993 and 2003 reveals little, ifany change inhuman assets (table 4.3). Household size, dependency and headship show no significant change across the ten-year period in question. Rural households average more than five people in size, and approximately 80% o f them are headed by males. Literacy rates have increased slightly, to 63%, but the difference between 1993 and 2003 i s not statistically significant. The proportion o f households headed by someone who completed primary school has increased by about five percentage points, while the proportion with secondary school education has declined. Inthe latter case, the decline occurred between 1997 and 2003 and might indicate out-migration o f well educated rural residents due to the rural crisis. Box 4.2. Information on household level data sources The Encuesta Permanente de Hogares (see http://www.ine-hn.org for details) is a nationally representative sample survey conductedevery year since 1990 by the National Statistical Institute(ME).Its objectives are to: a) obtain information about general characteristics of the population; b) obtain information by residence(urban and rural); c) estimate rates o f employment; and d) obtaininformationon sources of income. We use these data to generate information on changes inhouseholdcharacteristics over time. Inaddition, we can use the survey to comparepatterns o f changes inhouseholdincome per capita. Some indicators o f housing quality remained stable inthe ten-year period, while others showed deterioration. The proportion o f households with good flooring and electricity access remained relatively constant (at about 50 and 36%, respectively), while the proportion with access to potable water declined by more than ten percentage points. Aggregate measures mask some o f the heterogeneity across provinces inrural areas (table 4.4). Household size varies from about 4.9 members in Cortes to more than 6.2 people inIntibuca. Headship also varies significantly by province, from a high of 87% male in Intibuca to a low of about 70% in Valle. Literacy and schooling show significant variability across space, as does housing quality. Household incomes have shown a subtle shift between 1993 and 2003, although the mean income has not seen a statistically significant shift. The distribution has 37 widened, with proportionally more households at both lower and higher ends o f the income distribution (figure 4.9). 24 This widening indicates growing inequality, a problem that might be of concern to policy makers. Lack of improvement of the mean (from a low base in 1993) should also be o f concern. Agriculture and infrastructure, economicpotential andpoverty Information on agricultural potential can be combined with information on access to infrastructure to create a map o f economic potential (Pender and Hazel12000). Due to the importance o f agriculture to growth potential in rural areas, we begin by investigating how agricultural potential varies over space. We continue by looking at the distribution o f infrastructure. Then we look at the spatial distribution of economic potential and of poverty (box 4.3). 1 Box 4.3. Topography, agricultural potential and market access I A recent study o fwestern Hondurasconcludes (Munroeet al. 2002, p.367): "The most significant finding inthis study i s that the probability of stable agricultural productioni s significantly greater at lower elevations, flatter slopes, and inlarger patches than areas offorest cover inthe region. Stable agriculture also tends to be found in areas that are relatively more accessible to local markets, but less accessible to regional markets." Agricultural regions andpotential Many hillside areas are used for food staple production usingunsustainable technologies that have ledto increasing degradation of natural resources, ingeneral, and soil, forest, and water resources inparticular (Kok 2001;Pender, Scherr and Duron 2001; Jansen et al. 2003). Inits 1998 Human Development Report, UNDP (1998) divided Honduras into seven agricultural macro regions (or zones), each containing characteristics (physical and others) that lead to certain patterns o f production. These agro-ecological and land use zones can be useful for conceptualizing agricultural potential and appropriate actions to promote broad-based growth in agriculture.*' Hillside areas correspond mainly to zones 111,IV, V and parts o f VII. The seven zones are: Zone I:Agricultural Frontier: represents the east and southeastern regions o f the country. This area has very low population densities and is underserved by roads. It contains most o f the country's biological reserves and i s potentially an important area for 24Non-parametric Kernel density smoothing techniques are used to create these density differences (see Alwang, Mills and Taruvinga 2001). The density differences shown in figure 4.9 represent the difference in the density (roughly, the proportion o fhouseholds at each point along the distribution o f well-being) between the entire rural sample and each rural region. Above the horizontal line proportionally more households are at the corresponding level of well-being compared to the region o f comparison. 25The UNDP (1998) report also mentions the problems o f land distribution (the Gini for landholdings i s reported to be about .76 -see p. 89), environmental problems associated with intensive agricultural production and with campesino agriculture. 38 tourism development. People tend to be very poor (rural poverty i s deep) but relatively few (rural poverty is not dense). Zone 11: Northern Agro-industrial area: comprises the northern coastal areas and the valley around San Pedro Sula with high population densities and relatively large landholdings. Landuse i s dominated by plantation agriculture and extensive livestock operations. The maquila industry attracts many migrants from rural areas elsewhere in the country. Ruralpoverty is less deep, but quite dense because o fthe highpopulation densities. Zone 111: Mountains and Valleys in the CentralInterior: are areas where small-scale peasant agriculture predominates inthis region. Roughly 30% of the land is devoted to crops and 40% to intensive livestock (small pastures). Both agricultural potential and infrastructure vary considerably inthis zone. Rural poverty i s deep and dense, particularly in areas with highpopulation density. Zone IV: Western Coffee-growing area along the Guatemala border: includes most o f Honduras' coffee production, mostly produced on relatively small farms (usually < 3.5 ha) that co-exist with larger farms. Dueto fairly good road infrastructure, producers have relatively good market access, despite the difficult terrain. Tourism potential, particularly inthe Copan area, has not been fully exploited. Dueto relatively highpopulation densities, rural poverty i s dense and has become deeper as a result o f the coffee crisis. Zone V: Mountains and Steep-slope Campesinosof the South: i s an area largely comprised o f very poor small-scale producers o f basic grains and small-scale livestock. Poverty and small-scale agriculture are associated with environmental degradation that assumes particular importance as the region contains parts o f the four major watersheds. These include Ulua, Chamelecon, Lempa and Choluteca. Rural poverty i s both deep and dense. Zone VI: Southern BusinessArea: includes industrial producers o f various scales of production located on the coast o f the Gulf o f Fonseca who focus on export products such as melons and shrimp. The area, which i s relatively urbanized, has access to infrastructure and social-support institutions. As a result, rural poverty i s relatively less deep and less dense. Zone VII: Central Latifundio: i s an area located inthe Central Valleys toward the western part o f the Nicaraguan border. This area includes Tegucigalpa and occupies about 16% o f the land area. It i s characterized by high geographic diversity. Historically, this area was characterized by extensive livestock operations on large holdings. More recent changes have included coffee expansion on mountainous slopes and increased horticulture closer to Tegucigalpa. Landless agricultural workers and producers and smallholdings are found among large-scale landholdings. Ruralpoverty is both deep and dense. - 39 Infrastructure, economicpotential andpoverty Potable water i s fairly widespread inrural Honduras, with most counties reporting more than 60% coverage o f potable water (figure 4.11). INEfigures suggest a coverage o f 70% for potable water. Basic sanitation is, however, not nearly as widespread (figure 4.12). Areas near San Pedro Sula and along the Northern Coast are relatively well served, as are areas around Tegucigalpa, but the in the remainder o f the country, coverage averages less than 40%. Although Honduras does not have a detailed nationally representative household expendituresurvey, which is the preferredmeans o f measuring poverty, the Encuesta Permanente deHogares has been usedby INE to create a poverty map (INE 2003a and figure 4.13 but see appendix 2 for pitfalls with INE's poverty calculations). Poverty rates are only weakly associated with lack o f access to transport infrastructure. Highest poverty rates are found along the southwestem border with El Salvador, the far-western Copan area, along the southeastern border with Nicaragua, inthe east, and inthe north-central region. Many o f these areas are relatively well served by transportation, especially Copan and Intibuca, which produce significant quantities o f coffee. The east i s clearly underservedby transport infrastructure, but while poverty rates are highthere, the very low density o f population ensures that poverty densities, or poor people per square kilometer, are low. Highpoverty areas along the Guatemalan and Salvadoranborders inwestem and southwestem Honduras represent somewhat o f a conundrum. These areas have relatively highaccess to infrastructure and agricultural potential (e.g., incoffee producing areas), but also relatively highrates o fpoverty. These areas might be targeted for highlabor demand investments (inside and out o f agriculture). Inparticular, the Copan area has substantial tourism potential, but despite good "locational" conditions, measures o f well- beingare lagging far behind potential. As is noted, the mountains andvaried topography throughout the country leave many people almost completely isolated even though transportation access is, on average, quite good. Intra-regional heterogeneity is masked by the county-level analysis. Population densities tend to decline as we move away from the major urban centers o f Tegucigalpa and San Pedro Sula. Pattems inchanges inpoverty rates are not quite uniform. As noted incompanion reports for Nicaragua and Guatemala, differences may exist between highpoverty rate areas and highpoverty density areas, which have implications for targeting of investments. InNicaragua, for example, areas with high poverty rates are distant from Managua, in areas where population densities are low. Targeting such areas, although minimizing leakages to the non-poor, would bypass most o f the poor, since most o f them live near Managua (poverty densities there are high).A differentpattem exists inGuatemala, where poverty rates and population densities are highest inthe Western Altiplano, leading to spatial coincidence between highpoverty rate and highpoverty densities. 40 InHonduras, most hillside areas are found that have both highrates o fpoverty and highpopulation densities (leading to highpoverty density). But also inthese areas conditions are not uniform. For example, the western areas around Copan, the southern areas inValle and Choluteca, and the Province o f Comayagua have both highpoverty rates and highpoverty densities. By targeting these areas, significant proportions o f the rural poor should be reached26.The problem o f leakages to the non-poor is minimized because poverty rates there are high. The geographic correspondence between high poverty rates and highpoverty density means that there is little or no trade-off in targeting highpoverty areas for poverty-reducing interventions. But several o f these areas are also blessed with good-quality infrastructure and access to markets. They would be good candidates for poverty-reducing investments. The combination ofhighpopulation densities, relatively good infrastructure and good soil quality shows that these areas have economic potential. Persistent highrates o f poverty, however, show that this potential i s not beingtapped by households and that to the extent that it is being tapped, the poor are not able to participate. Despite highroad densities, internal distances within these topographically complex counties may limit participation inthe market economy and cause highlevels of poverty. Alternatively, asset bases o f the poor inthis region (land, capital, human assets) most probably need to be strengthened before they can benefit from growth-related spillovers. This descriptive analysis suggests that more needs to be known about human and other assets inorder to design appropriate interventions. L o w population densities inthe eastern part o f the country lead to much lower poverty densities. Here we witness a tradeoff between poverty rates and poverty densities. Because o f the highpoverty rates in some o f these counties, a project or investment neednot have a complicated explicit targeting mechanism; leakages to the non-poor are reduced inareas with higher rates o f poverty. On the other hand, because population densities are low, the projects should be spatially targeted to specific population clusters, or the investment/program should be selected based on low per unit costs o f delivery over space. For example, investments like health clinics should obviously be targeted to population clusters. Others, such as schools should be located to guarantee a reasonable degree o f access, even in low population density areas; our analysis indicates that the number o f schools per capita tends to be highest inthe most remote areas. Agricultural development programs such as land titling and distance delivery o f technical services might also be appropriate in low-density areas because they can be delivered across space at a minimal cost. Such programs might also be appropriate due to good agricultural potential and infrastructure inthese areas. 26Because of the absence of high-quality representative consumption data, it i s really impossible to provide accurateestimates o f the numberofpoor people inany area. 41 Vulnerability tofood insecurity The World Food Program (WFP), together with the Government o f Honduras, has recently publisheda set o f maps reflecting different dimensions o f food insecurity (GoH/WFP 2003). The maps, which were produced usingGIS techniques that include overlaying different factors on a single map, reflect a number o f agro-ecological and socio-economic factors that influence the availability and affordability o f food for households throughout the country. Vulnerabilityto food insecurityas a functionof economic access: WFP combines 5 factors that influence economic access and the ability o f households to deal with risks. The factors and their respective weights are: per capita income (22%), education level (22%), household dependency ratio (22%), road density (16%), gender o f household head (14%), and land area under permanent crops (3%). The T of Development can be appreciated quite clearly as areas represented by low vulnerability to food security and, conversely, higheconomic access. Hillside areas inthe western and central part o f the country, and eastern areas are mostly classified as very highly or highly vulnerable to food insecurity because o f a combination o f lower capita income, low education, highdependency, and low road density. Climatic risks: WFP combines 5 factors o f climatic risk.The factors and their respective weights are: vegetative cover (l6%), erosion potential from rain (30%), desertification index (25%), share o f population facing drought risk CO YO), share o f population facing flood risk (19%), and share o f population using soil conservation practices (6%). The major parts o f this climatic risk index are related to drought, but it also includes floods, and factors such as vegetative cover and conservation practices that can lower vulnerability to climatic risks. Most areas o f Honduras face very high or high climatic risks. A notable exception i s the central part o f the country, much o f it corresponding to the T o f Development. Hillside areas inthe west and south, where poverty density i s relatively highcontain most o f the very highrisk areas. 4. Summary of key findings Rural poverty i s deep and widespread throughout much o f Honduras. However, it i s still possible to identify a number o f geographically more or less well-delimited areas where both poverty rates and population densities are particularly high, despite relatively good infrastructure and access to markets. These areas are good candidates for poverty- reducing investments and include the western and south-western areas in Copan, Intibuca, Lempira and L a Paz; the southern areas inValle and Choluteca, and the central Province o f Comayagua. Population changes are largely driven by economic potential with movement towards urban areas, the Northern Coastal area, and movement away from extensive agricultural areas, particularly those with highproportions o f landless. Some areas o f highagricultural potential are actually growing much more slowly than the national 42 average, suggestingthat access to productive resources inthese areas may limit opportunity and stimulate migration out o f these regions. Economic potential has a strong spatial pattern, with highpotential areas close to the main cities, along the Northern Coast and inmountainous areas with favorable soil and road conditions. Due to the high density o f roads and favorable agro-climatic conditions in the western part o f the country, growth potential there i s high. Eastern areas have less access to infrastructure and growth strategies might be more inward looking than in the better endowed western areas. This finding underscores the importance o f a spatially differentiated rural development strategy for Honduras. 43 References Alwang, J., B. F. Mills,and N.Taruvinga. 2002. Why Has Poverty Increased in Zimbabwe? Washington, D.C.: The World Bank. CIAT (International Center for Tropical Agriculture). 2001. Atlas de Honduras (con datos Mitch). Cali, Colombia: CIAT. See http://www.ciat.cp;iar.orp;/catalogo/listado.jsp?tema=LAND USE&offset=5 GoH/WFP (Government o f Honduras and World FoodProgram). 2003. Ancilisis y Carlografia de la Vulnerabilidad a la InseguridadAlimentaria y Nutricional en Honduras. Tegucigalpa, Honduras.: World FoodProgram. IICA (Inter-American Institute for Cooperation inAgriculture). 1999. Desarrollo Institucionalpara la ProduccidnAgricola Sostenible en las Laderas de Am&rica Central. Segundafase Z998-2002. San Salvador, El Salvador and Tegucigalpa, Honduras: IICA. INE(National Statistical Institute). 2002. CensoNacional dePoblaciony Vivienda. Tegucigalpa, Honduras: INE. ____-__--2003a. EstimaciBn de Indicadores de Pobrezay Desigualdad a Nivel Municipal en Honduras. Tegucigalpa, Honduras. _--------2003b. Vigksimo Octava Encuesta de Hogares de Propositos Multiples. Document found on http://www.ine-hn.org/. Jansen, H.G.P., A. Rodriguez, A. Damon, andJ. Pender. 2003. "Determinantes de Estrategias Comunitarias para Ganarse la Vida y e l us0 de Practicas de Produccion Agricola Conservacionistas en las Zonas de Ladera en Honduras." EPTD Discussion Paper No. 104, Washington, D.C.: InternationalFoodPolicy ResearchInstitute (IFPRI), Environment and Production Technology Division (EPTD) . Jansen, H.G.P., A. Rodriguez, A. Damon, J. Pender, J. Chenier and R. Schipper. 2004. "Determinants of Income-earning Strategies and Adoption o f Conservation Practices inHillsideCommunities inHonduras." Agricultural Systems (inprint). Kok,K.2001. "Spatial Determinants ofHonduran LandUse: Empirical Evidence for Malthus' Theory." Chapter 3 in:Scaling the Land Use System: A Modeling Approach with Case Studiesfor Central America. PhD. thesis. The Netherlands: Wageningen University. Munroe, D.K., J. Southworth, and C.M. Tucker. 2002. "The Dynamics of Land-cover Change inWestern Honduras: Exploring Spatial and Temporal Geography." Agricultural Economics, 27:355-369. 44 Pender, J. and P. Hazell. 2000. "Promoting Sustainable Development inLess-Favored Areas." FOCUS 4, Brief 1. Washington, D.C.: IFPRI. Pender, J., S. Scherr, and G. Duron. 2001. "Pathways of Development inthe Hillside Areas of Honduras: Causes and Implications for Agricultural Production, Poverty, and Sustainable Resource Use." In:David R.Lee and Christopher B. Barrett eds., Tradeoffi or Synergies? Agricultural Intensification, Economic Development and the Environment. Wallingford, U.K.: CAB International. UNDP (UnitedNations Development Program). 1998. Inforine Sobre Desarrollo Humano. p 232.'Tegucigalpa, Honduras: UNDP. 45 Table 4.1. Populationchangeby Province,1988-2001 Province Atlantida Choluteca Colon 48.7 Comayagua 31.5 Copan 25.7 Cortes 61.0 ElParaiso 22.7 Francisco Morazan 25.1 Gracias a Dios -4.7 Intibuca Islas de la Bahia 33.7 La Paz 32.1 Lempira 30.4 Ocotepeque 40.0 Olancho 29.0 Santa Barbara 15.2 Valle 13.4 Source: Census ofPopulation, 1988 & 2001 Table 4.2. Determinantsof municipio-levelpopulationchange, 1988-2001 (dependentvariable=percentage changeinpopulation) -3.71 -2.28 0.000212 -1.28 0.075662 -5.13 0.068913 -2.36 0.008 144 0.02708 Northern Agricultural region agcafe Dummyvariable= 1iflocatedin -0.0309 0.021811 I -1.42 Coffe- roducin A ricultural re ion Interce t 0.615243 0.204 46 Table 4.3. Assets and changes in ruralHonduras % of HH 0.05527 0.228518 0.069119 0.253702 0.069196 0.25388 heads with secondary schooling % o f H H 0.508865 0.499948 0.478184 0.499613 0.481378 0.499838 with a good floor % ofHH 0.724369 0.446856 0.912256 0.282973 0.850744 0.356472 with good water access % of HH 0.361435 0.480442 0.384185 0.486489 0.363437 0.481167 with access to electricity Meanper 735.6182 1468.674 606.3528 1057.513 753.47 2725.723 capita income per vem i--- I I I I I I 47 Table 4.4. Assets and changesby year and Province,ruralHonduras 48 49 50 51 11 % of literate HH heads I 0.656009 I 0.475337 I 0.550942 I 0.499538 11 52 Figure 4.1. Map of Honduras -Intsmatlonal Honduras boundary Depsrfamento boundary - National capital B Oepertamenlo capital -Railroad Road 0 25 M 73 IWKilomafan 0 25 52 i s tbM 1000meters above sea level), farming relatively less fertile soils. Market access3*and education are average for these households. These farmers rely on basic grains for their subsistence needs: they use about one-third o f their farm area and more than one-quarter o f their household labor to produce basic grains. The income o f coffee producing households was U S D 0.33, just over half o f livestock farmers. However, the survey was taken duringthe period when coffee prices collapsed (falling in2000-2002 to about half the price o f previous years). Households incluster #3 are the poorest among all livelihood groups, earning an average o f only U S D 0.15 per person per day. The explanation lies inthe fact that these households rely nearly exclusively on basic grains production that has low profitability (partially caused by limitednatural assets interms o f quantity and quality) and is relatively low-value. These households have small farms (2 ha on average), tend to be located at high elevations and/or steep slopes, and have little interms o f other productive assets. Inaddition, they are the most geographically isolated households, severely limiting off-farm opportunities. The probability of a female head is highest for this cluster. The livelihoods strategy o f households incluster #4includes basic grains and off- farm employment. This i s the 2"dlargest livelihood group inthe sample with almost one- quarter of all households. These households have the smallest landholdings, with less than 2 ha o f farmland o f which less than 20% is owned, on average. Thus, they need to rent land, but overall land access is limited. By working off-farm they are able to earn more than double (USD 0.41 per person per day) the income o f cluster #3 households, despite an above-average dependency ratio and below-average education. It seems that 38This result is somewhat surprisinginview of the investmentsinroads madeby the HonduranCoffee Institute (IHCAFE) incoffee growing areas. 103 limited access to land "pushes" these households to be more entrepreneurial and seek out alternative employment opportunities, in or out of agriculture. Cluster #5, the largest livelihoods group, accounts for 30% o f the total sample. On average they have over 10 ha o f land, o fwhich nearly two-thirds is kept either fallow or under forest. Their livelihood strategy i s similar to households in cluster #4 but with considerably more land, so they hire (rather than sell) labor and devote more time to livestock activities. However, their average daily per capita income o f USD 0.29 i s about 30% less than that o f households incluster #4, but higher than households in cluster #3, which just produce basic grains. Apparently by working on-farm, these households have lower incomes than those seeking off-farm employment. On the other hand, these households may be less vulnerable to risks than those in cluster #4, since they have greater wealth and more diversified income sources. Education i s slightly above average for this cluster, whereas both physical and natural assets are about average. Cluster #6 represents a small group o fpermanent crop producers with small landholdings (2.4 ha on average) whom devote most o f their land and labor to intensive tree crop production such as fruits, oil palm etc. These households have the highest average incomes inthe sample (USD 0.66 per capita per day). They have smaller than average household sizes and are located infavorable agro-ecological areas with high population densities, highrainfall and good access to paved roads and public transportation, all o f which are important for diversification into higher-value permanent crop production. Finally, most households incluster #7 are vegetable producers who allocate most o f their labor to working on their own farms. Despite being far from a paved road in areas with low population densities, these households are close to a non-paved road, which gives them a sufficient degree of market access to specialize invegetable production. Somewhat surprisingly i s the fact that their average daily income during the survey year (USD 0.38 per capita) was only slightly above average despite an average farm size o f about 4.5 ha, good market access and the relatively high educational level o f the household heads. 2. Wisconsin households The livelihood o f households incluster #Iaccounted for about one-quarter o f the sample and i s the most diversified. These households have diversified farm operations and work outside their own farm inboth agricultural and non-agricultural occupations. They have relatively highquantities o f some productive assets (average farm size i s 30 ha, along with sizeable amounts o f livestock), but average endowments o f other assets. These households obtain about 40% o f their total income from the own farm (basic grains, some coffee and livestock); the remainder o f their income i s from off-farm work (65% from agricultural labor on other farms and 35% from labor outside the agricultural sector). About 10% o f their income i s from remittances. Average per capita income i s with USD 1.22 the secondhighestinthe sample. 104 Households in cluster #2 are subsistence-type farmers with very little land (average farm size i sjust above 1 ha and most land i s without title), virtually all o f which i s used for basic grains production for household food security. They have very little other physical capital, low human capital, highdependency ratio, poor access to credit and low social capital. These households generate most o f their cash income from off- farm work (about 70% from working on other farms and 30% from work inthe non- agricultural sector). Only about 6% o f their income consists o f conditional transfer payments and remittances. These households belong to the poorest o f the poor with an average per capita income o f U S D 0.42 per day. These households resemble the ones in cluster # 4 o f the IFPRIsample. Cluster #3 consists o f medium-size livestock farmers (average farm size 17.2 ha 50% o f which is titled) who do very little off-farm. Their endowment o f other assets is average. Most o f their land i s used for basic grains and pastures. These households generate virtually all o f their income on their own farms and also are the richest households inthe sample (income i s USD 1.32/day/person). Coffee farmers who make up cluster #4 on average farm 8.1 ha o f land and also work a considerable proportion o f their time o f f their own farm (though not as much as households in cluster # 2). These households tend to have their own means o f transportation but very little livestock. They are located at higher altitudes. Their human capital i s average but their social capital (participation inorganizations) i s above average, They have average savings but above average credit access. Off-farm work (mostly on other farms) generates nearly 40% o f their total income. At just U S D 0.79, average per capita daily i s low but higher than coffee farmers inthe IFPRI sample. Cluster #5 consists o fhouseholds with relatively large landholdings (26.6 ha on average, a third o f which i s titled) but average for physical assets. However, they do not depend on their farm for most o f their income, because they have their own businesses (shops, trade etc). Education is above average. These households also have relatively highamounts o f financial assets (highsavings) and above average social capital. This group represents only 7% o f the sample but has lower-than-expected average income and higher-than-expected poverty rate. On the other hand, our calculation o f average income for this cluster (USD 0.7 1per day per person) is probably an underestimate because our calculated income for these households from farm operations consistently resulted in negative values. Finally, households in cluster #6 live mostly o f f remittances. Despite the fact that these households have 8.4 ha o f land on average, they have very little other physical capital and belong to the poorest inthe sample (average income only U S D 0.43/day/person). They often have a female household head and work very little outside their own farms. 105 APPENDIX 9 Resultsof the MultinomialLogit Models39 Hans G.P. Jansen (IFPRI-RUTA) 39Preparedfor the World Bank researchproject, Drivers o f SustainableRural Growth and Poverty Reduction inCentral America, Honduras Country Case Study. 1. MultinomialAnalysis The objective o f the multinomial analysis was to identify the main determinants o f household livelihood strategies. The analysis was conducted separately for the IFPRI and the Wisconsin household samples. Together the explanatory variables used in the multinomial logit models constitute a fair representationo f the households' asset base (see tables 9.1 and 9.2). We present the results o f both the full and the reduced form models (see tables 9.3 and 9.4 for the IFPRIsample, and tables 9.5 and 9.6 for the Wisconsin sample). Whereas the full models investigate the influence o f all types o f capital on households' livelihood strategy decisions, the reduced forms only take natural, human and physical capital into account, based on the possibly endogeneity and/or reverse causality o f locational, financial and social capital-related explanatory variables. For example, a household's participation in organizations (social capital) may be partially determined by the particular type o f livelihood strategy pursuedby that household, rather than the other way around. IFPRI households The estimated coefficients are relative to livelihood strategy #3 (basic grains producers) and significant variables should be interpreted as increasing or decreasing the probability that a givenhousehold follows the corresponding strategy rather than strategy # 3. That is, the magnitude o fthe coefficients has no clear interpretation, but their direction does. Because o f limited number o f observations, livelihood strategies #6 and #7 do not form part o fthe models. Wisconsinhouseholds The estimated coefficients are relative to livelihoodstrategy #2 (basic grains producers/farm workers). Just like inthe IFPRImodels, significant variables should be interpreted as increasing or decreasing the probability that a given Wisconsin household follows the corresponding strategy rather than the comparison strategy (strategy #2). Again the reduced form (table 9.6) leaves out all potentially endogenous variables whose influence may often stem from reverse causality. 107 N - C m 3 d m o q r - - W m 0 r - m s 2 O W % m m N d v i d 221 N Dc Lo 2 0 3 - d vi W N u r- G. vi 0. 8 "c N r` -* vi G. m r- m X 2 m 0 d N 0. 2 2 r- 2 d 0 8 2 t- W 0 0 0 m 0 2 0 LA 2 W r- 0 LA 9 W 0 8 2r- WLA 0 * Not part o f the multinomial logit regression. ') Inmmfor the period May-September. Own calculations based on information from CIAT (1998). ') Measured as maize water deficit for the period Oct-January in m (average for sampled plots inmm).The IFPRI survey collected detailed biophysical data and soil samples on a (randomly drawn) sample o f two plots on each farm. These samples were analyzed in a local soil laboratory resulting in data regarding pH, nutrient content, organic matter content and texture. These data were mainly usedfor the calculation o f water availability and soil fertility. Water deficits were calculated on the basis o f data for monthly temperature, effective rainfall (taking runoff into account as determined mainly by slope, slope direction, contour curvature, profile curvature and position on slope), evapotranspiration, and soil characteristics including depth, texture and organic matter content. Only moisture availability for the second season (postrera) was included in the model since the data indicated very few cases o f main season (primera) water deficits. 3, Average altitude o f sampled plots infeet above sea level (data from household survey). 4' Soil fertility was approximated by potential maize yields (nutrient-limited but not water-limited) as determined by the QUEFTS (Quantitative Evaluation o f soil Fertility and response To Fertilizers) model (Janssen 1990), taking into account nitrogen content, pH, and available potassium and phosphorous. Units are kgs. j) Inmanzanas (1 M z = 0.7 ha), data from household survey. `) Value o f machinery, equipment & transportation inLps, data from household survey. ') Inyears, data from household survey. ') Total number o f man-months spent outside the household by household members.Data from household survey. 9, Median years o f schooling o fhousehold members older than 7 years, data from household survey. lo) Dummy variable (=IHHhas received training, 0 ifnot) ,data from household survey. if `I) Dummyvariable (=1 ifHHhas received extension visits, 0 ifnot) ,data from household survey. I') Ratio defined as follows: (# o f HHmembers < 12 and > 70 yrs) / (# o f HHmembers between 12 and 70 yrs) ,data from household survey. 13) I= headofhousehold,datafromhouseholdsurvey. female 14) Females > 12 yrs o f age as a % of total household size, data from household survey. 15' # o f persons per km' in the community, data from SINITICIAT. Average travel time to the nearest input market from the homestead, data from household survey. Logit models use an ordinal market access index variable that takes into account geographical distance, road quality and slope. Index developed by CIAT. 17) Kmo froads/km2inthe community, data from SINITKIAT. 18) Dummy variable (1 = HHhas access to any form of credit, 0 otherwise), data from household survey. 1 9 )Dummyvariable (1 =householdparticipates inthe organization, 0 =household does not participate in the organization). Includes agricultural cooperatives, producer associations, unions and private enterprises engaged in agricultural services etc. Data from household survey. 20) Dummy variable (1 = householdparticipates inthe organization, 0 =household does not participate in the organization). Includes villagers' association, parent organization, ethnic council, water users' group, religious organizations, women's organizations. Data from household survey. 21) Dummyvariable (1 = householdparticipates inthe organization, 0 =household does not participate in the organization). Includes savings and loans type operations. 22) Dummyvariable (1 = householdparticipates, 0 = household does not participate). Includes NGOs delivering mainly technical assistance in the areas o f agricultural production andor marketing. Data from household survey. 23) Percentage o f HHin the community that is involved in any type o f organization. 110 Table 9.2. Explanatoryvariables used in multinomiallogit model, by livelihood strategy, Wisconsinsample znde~endentvariablcs Mean error Mean error Mean error Mean error Mean error Mean error Mean Std error Altitude" 929.5 18.6 857.0 37.8 918.3 61.3 881.0 60.7 1072.0 25.3 803.5 63.3 850.2 54.1 Area of landowned" 22.1 2.8 42.8 9.3 1.9 0.2 24.7 4.7 11.6 1.4 38.0 10.2 12.0 2.5 Valuc of other non-land physical assets" 5823 2230 9671 6343 0 0 0 0 9330 4799 4704 3571 1454 1352 HUMAN CAPITAL LOCATIONCAPITAL Roaddensity Agricultural organizations'5' 10.2 1.1 9.0 2.6 1.5 10.2 3.1 14.1 2.2 3.0 Community organizations:6' 36.9 1.7 35.6 3.2 39.1 4.6 34.7 4.8 39.7 3.2 37.9 6.4 31.9 4.9 111 ''In mmfor the period May-September. Own calculations basedon information from SINITICIAT. Inmeters above sea level; own calculations based on information from SINITICIAT. 3, Inmanzanas (1 Mz = 0.7 ha). 4, Value o f machinery, equipment & transportation inLps, data from household survey. j) Inyears, data from householdsurvey. Total number o f man-months spent outside the household by household members. Data from household survey. ') Medianyears o f schooling o f household members older than 7 years, data from household survey. 8' Ratio defined as follows: (# o f HHmembers < 12 and > 70 yrs) I (# o f HHmembers between 12 and 70 yrs). Data from household survey. 9, 1= female head o f household, data from household survey. `O' Females > 12 yrs o f age as a % o f total household size, data from household survey. # of persons per km' in the community, data from SINITICIAT. I?) Km o froadslkm' in the community, data from SINIT/CIAT. 13) Inkm, data from household survey. 14) Dummy variable (1 = HHhas access to any form ofcredit, 0 otherwise). Data from household survey. 15) Dummy variable (1 = household participates in the organization, 0 = household does not participate in the organization). Includes agricultural cooperatives and producer associations. Data from household survey. Dummy variable (1 = household participates in the organization, 0 = household does not participate in the organization). Includes villagers' association, parent organization, ethnic council, water users' group, religious organizations, political organizations, women's organizations. Data from household survey. '') Dummy variable (1 = household participates in the organization, 0 = household does not participate in the organization). Includes community banks etc. I*) Dummy variable (1 = household participates, 0 = household does not participate). Includes NGOs and other projects. Data from household survey. 112 \o 00 W Ici 8 9 4 x N 0 v, m N v) 00 0 8 8 9 0 Q\ z 0 N o\ \o x 0 r- m 8 2 v , - m 3m m c1 W 2 xm m 2 N vi 0 00 2 0 00 m 2 2d 00 m N r- m \ o N W 0 0 2 8 2 8 8 0 8si N v, 0 00 3 9 3 0 8r- In M 8 00 vi 9 9 -In 9 0 0 8 0 APPENDIX 10 IncomeRegressionResults4' Hans G.P. Jansen (IFPRI-RUTA) 40Preparedfor the World Bank researchproject, Drivers of Sustainable Rural Growth and Poverty Reduction inCentral America, Honduras Country Case Study. Inorder to find out to what extent livelihood strategies and asset endowments impact on household income, two OLS (Ordinary Least Squares) income regressions were run, one each for the IFPRI and the Wisconsin household samples. The dependent variable inboth income regressions i s the natural logarithm o f total annual household income, measured in Lps. The explanatory variables include the following: 0 livelihood strategies (except one to avoid perfect collinearity); 0 natural assets: summer rainfall, second season rainfall deficit (IFPRI households only), annual rainfall (Wisconsin households only), soil fertility (IFPRIhouseholds only), elevation, amount o fowned land, share o f land holdings that has title; 0 location assets: population density, market access; 0 physical assets: value o f livestock holdings, value o f machinery, equipment and transportation; human assets: household size, dependency ratio, age o f household head, sex o f the householdhead, median education, migration, % o f female adults, training received (IFPRI households only), extension received (IFPRI households only); financial assets: access to credit; 0 social assets: participation in community organizations, credit organizations, external organizations, participation in any other organization; interaction variables: (owned landx credit), (farm size x education), (farm size x market access), (education x market access), and, only for the IFPRI households, (owned landx soil fertility). The decision whether or not to use the log transformation o f the independent variables depended on their distributions across the sample. The results, shown intables 10.1and 10.2, are discussed in section 4.3 o fthe main report. 120 Table 10.1. Determinantsof HouseholdIncome,IFPRI households') stiinated coefficient" Migration 1.188* 0.365* -0.249 Owned land credit dumm * Only householdswith positive total income. *)*,** indicate statistical significance at the 5%, and 1% level, respectively.3, See table 9.1 inappendix 9 for an explanation of the variables. 4,Predictedvalues from the multinomial logit regression were used. 121 Table 10.2. Determinantsof HouseholdIncome, Wisconsin households'' Constant 15.110 .29 2,*,** indicate statisticaisignificance at the 5%, and 1% level, respectively. 3,See table 9.2 in appendix 9 for an explanation of the variables. 4,Predicted values from the multinomial logit regression were used. 122 123