Report No. 31857-CA Shocks and Social Protection: Lessons from the Central American Coffee Crisis (In Two Volumes) Volume I: Synthesis of Findings and Implications for Policy December 28, 2005 Poverty Reduction and Economic Management and Human Development Sector Management Units Latin America and the Caribbean Region Document of the World Bank Shocks and Social Protection: Lessons from the CentralAmerican Coffee Crisis Volume I:SynthesisofFindingsandImplicationsfor Policy TABLE OF CONTENTS PREFACE ....................................................................................................................................................... iv EXECUTIVE SUMMARY............................................................................................................................. v CHAPTER 1 INTRODUCTION: SHOCKSAND SOCIAL PROTECTIONINCENTRAL AMERICA....................................................................................................................................................... 1 1 Objectivesof the Report..................................................................................................................... Shocks inthe CentralAmerican Context........................................................................................... 5 Data andAnalysis............................................................................................................................... 5 Organizationof the Report................................................................................................................. 7 CHAPTER 2 THE WELFARE IMPACTSOF THE COFFEE CRISIS: A TALE OF FOUR COUNTRIES................................................................................................................................................... 9 The CentralAmerican coffee crisis: macro trends............................................................................. 9 14 Householdriskmanagementstrategies............................................................................................. The socio-economicimpacts ofthe coffeecrisis .............................................................................. 22 GovernmentResponsesto the Crisis................................................................................................. 29 CHAPTER 3 ELEMENTSOF A STRATEGY TO DEAL WITH SHOCKS............................................. 34 Strengtheningpeople's ex-ante riskmanagement capacity............................................................... 34 Developing appropriate, well targeted safety nets............................................................................. 39 Strengtheningdata, information, andmonitoring systems................................................................ 51 No one-size-fits-allindesigningcrisis responses ............................................................................. 51 CHAPTER 4 IMPLICATIONSFOR POLICY............................................................................................. 53 Key Insightsfrom the Case ofthe Coffee Crisis............................................................................... 53 Implications for Policy - Elementsof a Strategy to dealwith Shocks.............................................. 55 REFERENCES............................................................................................................................................... 59 Annex 1.......................................................................................................................................................... 65 i TABLES 1.1 Data and Analyses Undertaken for this Report ................................................................................... 5 2.1 Coffee production in Central America. 1998-2003 ........................................................................... 10 2.2 Distribution o f coffee farmers and production in 2000 ..................................................................... 11 2.3 Coffee exports revenues from Central America, 1999/2000 and 2000/2001 .................................... 11 2.4 EstimatedCoffee Sector Employment in Central America, 2002..................................................... 12 2.5 EstimatedDecline in Coffee Sector Employment inCentral America, 2000/01 -2001/02 .............13 2.6 Trends inHumanDevelopment Index inGuatemala, El Salvador, Honduras, and Nicaragua, 1995-2002 ....................................................................................................................... 14 2.7 Consumption changes among coffee and non-coffee households, selected municipalities inruralHonduras 2000-2002 ............................................................................................................ 17 2.8 The Impact ofthe Coffee Crisis on Coffee Households' Well-being: Summary ............................. 22 2.9 EstimatedFall inPer Capita Consumption from a 10 Percent decline inHousehold Per Capita Income (inpercent), RuralNicaragua.............................................................................. 23 2.10 Changes inLabor Allocation Toward Non-AgriculturalActivities inEl Salvador, Coffee and Non-Coffee Households, 1995-2001.............................................................................. 25 2.11 HouseholdCoping Mechanisms inRuralNicaragua duringthe Coffee Crisis................................. 27 2.12 Central American Coffee-households' Risk Management Strategies: Summary............................. 29 3.5 A Typology of Safety Nets................................................................................................................ 42 FIGURES 2.1 Composite coffee price index. 1984-2004........................................................................................... 9 2.2 Coffee exports as a share o f total exports.......................................................................................... 2.3 Real per capita income changes. for coffee and non-coffee households. RuralNicaragua...............12 16 2.4 Poverty changes inRuralNicaragua. 1998.2001. coffee and non-coffee households ...................... 18 2.5 19 Incidence o f Stunting. Nicaragua 1998-2001 .................................................................................... Incidence o f Underweight Children. Nicaragua 1998-2001.............................................................. 2.6 19 2.7 Change inPercentage o f Underweight Children inRural El Salvador, by Department and Intensityo f Coffee Cultivation, 1998-2002....................................................................................... 20 3.1 Households that Exitedthe Coffee Sector and Those that Remained: Changes inPer Capita Income and Consumption, RuralNicaragua, 1998-2001....................................................... 35 BOXES 1.1 Shocks. Household"Self.1nsurance". and the Role o f Social Protection............................................ 3 2.1 A Typology o f "Coffee Households" and "Coffee Regions" for Nicaragua..................................... 15 2.2 Exiting the Coffee Sector inNicaragua............................................................................................. 26 2.3 A Sample o f Government Responsesto the Coffee Crisis................................................................ 30 3.1 Market-basedApproaches to ManagingProducer Price Risk........................................................... 37 3.2 Responding to NaturalDisasters: Buildingon the Strengths o f Existing Institutions...................... 40 3.3 Managing CAFTA's Terms-of-Trade Effects ................................................................................... 48 3.4 Redesigningthe NicaraguanRedProtecci6n Social to Address Shocks........................................... 50 .. 11 Volume 11: DetailedCountry Analyses PREFACE.................................................................................................................................................... ii 1. SHOCKS AND COFEE: Lessons from Nicaragua (Renos Vakis, DianaKruger and Andrew D.Mason, World Bank)............................................................................................... 1 2. COPING WITH THE COFFEE CRISIS INCENTRAL AMERICA: The Role ofthe Nicaraguan Social Safety Net Program (John A. Maluccio, IFPRI) .............................................. 58 3. ANALYSIS OF THE POVERTY AND SOCIAL IMPACTS OF THE COFEE CRISIS (Price Shock) IN EL SALVADOR (Alvaro Trigueros and Carolina Avalos, FUSADES)............97 4. COPING WITH THE COFFEECRISIS INCENTRALAMERICA: The Role o f Social Safety Nets inHonduras (David Coady, Pedro Olinto, and Natalia Caldes, IFPRI) .............................................................................................................................. 149 5. THE COFFEE CRISIS: A Short Note on Guatemalan Farmers (Renos Vakis) ......................... 168 Acknowledgements This report was prepared by a team comprising Renos Vakis (HDNSP), Diana Kruger (Pontificia Universidad Catolica de Valparaiso, Chile), Natalia Caldes, David Coady, John Maluccio, and Pedro Olinto (IFPRI), Alvaro Trigueros and Carolina Avalos (FUSADES), Gustavo Arcia, Patricia Monge (Consultants), Nancy Gillespie (LCSHS), Carlos Sobrado (LCSPP), and Kalpana Mehra (PRMPR). The team was led by Andrew D.Mason (LCSHS) under the general supervisionof Helena Ribe (LCSHD) and Felipe Jaramillo (LCC2C). Logistical support, as well as assistance in processing the document, was provided by Claudia Isem and Christina Alquinta (LCSHD). The team greatly appreciated in-country support for the work in Nicaragua from Roberto Bendana, Juan Francisco Gutierrez (MIFIC), Merling Preza (PRODECOOP), Jose de Jesus Rojas (Banco Central de Nicaragua), Luis Villalta (SECEP), Julio Solorzano(MAGFOR), ChristineGoffin, Dulce Mayorga(INEC), and Carlos Arce (Consultant); assistance in processing the LSMS data for Nicaragua was provided by Edelberto Munguia. Anabella Larde de Palomo, Aida Arguello de Morera and Hazel Escrich (Secretaria Tecnica de la Presidencia), Margarita Beneke de Sanfeliu (FUSADES), Claudio Gonzalez-Vega and Jorge Rodriguez-Meza (Ohio State University) all provided valuable inputs and support to the analysis on El Salvador. The team is grateful to our peer reviewers William Maloney (LCRCE), Louise Cord (PRMPR), Peter Lanjouw (DECRG), and Tarcisio Castafieda (Consultant), who provided timely and insightful feedback on the report. The team would also like to thank FlorenciaCastro-Leal(LCSPP), Ariel Fiszbein(LCSHD), Indermit Gill (EASPR), FranciscoPichon (LCSER), Paul Siege1(Consultant) for their comments and support at various stages of the report. The team also benefited enormously from inputs from World Bank country and sector specialists during a series of consultative meetings, including from Jane Armitage, Martin Raine, Laura Rawlings, Manuel Sevilla (LCC2C); Panos Varangis, Bryan Lewin (ARD); and Emanuele Baldacci, Wendy Cunningham, Gillette Hall, BenedicteLeroy de L a Briere, Kathy Lindert, Manuel Salazar, Cornelia Tesliuc and David Warren (LCSHS). Finally, the team would like to thank participants o f seminars in Guatemala (May ZOOS), El Salvador (September 2005), andNicaragua(September 2005) and of the World Bank's Social Risk Management Core Course in Washington, D.C. (June 2004) for useful feedback on earlier versions o f this work. ... 111 PREFACE This study is part o f an ongoing engagement between the World Bank and its counterparts in Central America on social protection, comprising both policy dialogue and operational support to governments to extend basic services to their poorest inhabitants and to protect the most vulnerable from the most deleterious impacts o f shocks. The report was undertaken in response to requests from several Central American governments for support in understanding the welfare impacts o f the coffee crisis - an unprecedented decline in world coffee prices that occurred between 1997/98 and 2001/02 - and its broader lessons for public policy. Work on the coffee crisis began in April 2002, when the World Bank, along with the Inter- American Development Bank (IDB), and USAID, sponsored a regional workshop in Antigua, Guatemala, to assess the situation and identify possible lines o f action. In preparation for the workshop, the donors produced a joint assessment, "Managing the Competitive Transition o f the Coffee Sector in Central America," which identified several areas for action, focusing largely on production-side issues. In follow-up, a World Bank team also produced a more in-depth study, "Dealing with the Coffee Crisis in Central America: Impacts and Strategies," which also provided a preliminary assessment o f the crisis's social impact. All these efforts, however, highlighted the need to understand better the social impacts o f the crisis and to identify avenues for effective government action inresponse. In this context, a major objective of this report is to provide a deeper, more policy relevant understanding o f the welfare impacts o f the coffee crisis - including the effects of the crisis on household income, consumption, poverty, as well as on basic human development outcomes, such as education and child nutrition. To do this, the study has generated a body o f new empirical evidence, drawing from an unusually rich collection o f household survey data from El Salvador, Guatemala, Honduras, and Nicaragua. This includes "panels" o f data from Nicaragua, El Salvador, and Honduras that enable one to track changes in welfare o f the same households over the period o f the crisis. This has helped to provide a more detailed, clearer understanding o f the crisis than has been available to date. Given the prevalence o f both natural and economic shocks in Central America, another key objective o f the study is to draw out the broader policy lessons o f the coffee crisis - to enhance the abilities o f the region's governments to respond to a range o f shocks in a timely and effective manner. To do this, the report draws not only on evidence specific to the coffee crisis, but to other recent analysis on the role and efficacy o f different safety net programs in the face o f different types o f shocks. By learning the lessons o f recent experience, Central American governments, along with their development partners, can be better prepared to deal with a variety o f different shocks in the future. Inpursuing its objectives, the report has been organized into two volumes. Volume Ipresents a synthesis o f the key findings and policy implications, focusing both on the impacts o f the coffee crisis, specifically, and the lessons for government responses to shocks, more generally. Volume I1goes into more detail on the specific impacts o f the coffee crisis, presenting the collection o f background studies commissioned for this report. These background papers provide rich analytical detail on each o f the four study countries for readers who are interested in a more in- depth assessment o f country-level impacts, the unique data sets underlying the analyses, and the methodologies underpinning the analytical work summarized in Volume I. iv Shocks and Social Protection: Lessons from the CentralAmerican Coffee Crisis EXECUTIVESUMMARY Backwound A region vulnerable to natural disastersand economicshocks- Central America is a shock-prone region. Since the mid-l990s, the countries o f Central America have experienced a number o f natural shocks, including Hurricane Mitch (1998), earthquakes (El Salvador, 2001), and a series o f seasonal droughts and floods (often associated with El Niiio and L a Niiia). As small open economies, the Central America countries are also open to a variety o f economic shocks, whether in the form o f external terms-of-trade shocks (e.g., declining coffee prices, rising oil prices) or more generalized slowdowns inthe U.S.and global economies. -including a sharp decline in coffee prices between 1997and 2001- Among the most important economic shocks to hit Central America recently was the "coffee crisis" - a sharp decline in world coffee prices between 1997 and 2001 which had important impacts on the region's economies, and on the families who depend on coffee-sector income. The crisis reflected, in part, an ongoing structural change in the world coffee market; entry into the market o f several new producers (e.g., Vietnam) and dramatic production growth in other parts o f Latin American (e.g., inBrazil) contributed significantly to lower world coffee prices. At the same time, prices reflected considerable year-to-year volatility in the coffee market - the result o f inelastic world demand coupled with periodic supply shocks (e.g., due to weather). By 2001, these combined factors had brought real coffee prices to their lowest levels in more than 50 years. -although the coffee crisis had little effect on other sectors The coffee crisis had several important impacts on the main coffee producing countries o f Central America - Guatemala, Nicaragua, Honduras, El Salvador, and Costa Rica. Between 1999 and 2001, export earnings from coffee declined by 45 percent in Central America as a whole, while demand for permanent and seasonal labor in the coffee sector was estimated to have declined by about one-quarter. Despite these magnitudes, the crisis did not appear to have had significant spill-over effects into other sectors o f the Central American economies. Indeed, annual economic growth in the region continued to average about 4 percent during the crisis period. This growth had an important influence on the nature o f the crisis impacts, as well as on how coffee sector households worked to protect themselves from the worst effects o f the crisis. Initial assessmentsof the impacts of the crisis were based on relatively aggregated data While world coffee prices have since rebounded somewhat, the dramatic price decline associated with the crisis was a cause of great concern throughout the Central American region. Several early assessments o f the coffee crisis raised serious concerns about the poverty and social impacts o f the crisis - especially related to the apparently large declines inthe demand for permanent and seasonal labor in the coffee sector. Yet, these early assessmentswere based largely on aggregate firmand industry-level data. Little informationor analysis existedonthe household-level welfare impacts o f the crisis, making it difficult to know what types o f government responses would be most effective inassisting those most affected by the crisis. V Obiectives of the ReDort Analysis of household data has improved our understandingof the crisis impacts- The report was undertaken in response to requests from several Central American governments for assistance in understanding the welfare impacts o f the coffee crisis, as well as its broader lessons for public responses to shocks. One main objective o f this report is thus to develop a deeper, policy relevant understanding o f the impacts o f the coffee crisis - on household income and consumption, on poverty, and on basic human development outcomes, such as education and nutrition. T o do this, the study has built up a body o f new empirical evidence, drawing from an unusually rich collection o f household surveys from El Salvador, Guatemala, Honduras, and Nicaragua. This includes panel data sets from Nicaragua, El Salvador, and Honduras that enable analysts to track welfare changes inthe same households over the crisis period. This has resulted ina clearer, more detailedunderstanding ofthe crisis than hadbeen available previously. -and will enable us to develop more effective policy responses Given the prevalence o f shocks in Central America, a second main objective o f the study is to draw out the broader policy lessons o f the crisis - to enhance the abilities o f the region's governments to respond to shocks in a timely and effective manner. To do this, the report draws not only on evidence specific to the coffee crisis, but to other recent analysis o f the role and efficacy o f different safety net programs in the face o f different types o f shocks. By learning the lessons o f recent experience, Central American governments, along with their development partners, can be better prepared to deal with a variety o f different shocks inthe future. The Case of the Coffee Crisis-Kev Insights The new analysis o f the coffee crisis has generated several important insights - on the welfare impacts o f the crisis, o n households' strategies and capacity to manage risk, and on the efficacy o f government responses to the shock. Impacts of the coffee crisis. Analysis o f data from Nicaragua, El Salvador, Honduras, and Guatemala paints a consistent picture o f the impacts o f the crisis. Although there are some country-specific differences, the coffee crisis has had significant negative impacts on smallholder coffee farm households throughout Central America - on their per capita income and consumption, on poverty, and on their children's education and nutritional status. Small-scalefarmers in the coffee sector were affected more than workers While the broad areas o f impact are not surprising, there are some important differences between the actual effects o f the crisis and what observers had initially expected. Early assessments o f the crisis suggested that the most devastating effects o f the crisis were occurring through the large- scale loss o f employment by (often landless) laborers inthe coffee sector. Infact, although coffee labor households are consistently among the poorest groups in rural areas, small-scale, sey- employed coffee farmers were unambiguously the hardest hit by crisis. Moreover, these farmers were affected mostly through the effect o f the price decline, rather than through the loss o f employment. Coffee sector workers did experience some increase in unemployment as a result o f the crisis. Nonetheless, in the face o f continued economic growth in Central America and expanding economic opportunities outside the coffee sector, labor households seem to have been able to shield themselves from many o f the negative effects o f the crisis - by shifting into non- agricultural activities and increasingthe labor effort and earnings by other household members. vi Household risk management strategies. Coffee sector households in all four countries used a variety o f ex-ante and ex-post strategies in efforts to prevent, mitigate, or cope with the effects o f the crisis. This involved efforts to diversify household income sources (including exiting the coffee sector all together), migration and remittances, increasing family labor supply, sale o f household assets, andor reliance on informal "insurance" (or "mutual help") networks. I t is better topreparefor potential risks, ex-ante, than to react to their impacts, ex-post- Households that were better prepared for shocks ex-ante - e.g., with higher levels o f education, more diversified incomes, or existing sources o f remittance income - generally did better in protecting themselves from the effects o f the crisis than those who simply coped with the problem, ex-post. Nonetheless, coffee sector households were generally only partially able to "insure" themselves against the effects o f the crisis. This suggests that need for efforts to strengthen andor supplement households' own efforts to deal with risk. -as some ex-post strategies can have negative long-run consequences Moreover, while several o f the strategies households used can be considered "appropriate" ways to manage economic risk, others have potentially harmful long-term effects on household productivity and welfare. InNicaragua and Honduras, for example, coffee households commonly withdrew their children from school and put them to work, as part o f their efforts to maintain and protect family incomes. These negative effects on schooling - coupled with the adverse nutritional effects o f the crisis - raise serious concerns about the long-term effects on children's human capital, including potentially irreversible impacts on children's future economic productivity and welfare. Such effects on children's human capital increase the risk o f poverty being passed on from one generation to another. Regional government responses. In light o f significant welfare impacts o f the crisis - and the fact that even a short-term shock could have negative long-term effects on household welfare and productivity - there was a strong rationale for government interventions to strengthen households' own risk management efforts and to mitigate the impacts o f the crisis. Against this background, initial government responses focused largely on debt restructuring for (mostly large and medium sized) coffee producers. Although some governments in the region subsequently implemented small-scale "workfare" (e.g., food-for-work, cash-for-work) programs to address some o f the employment effects o f the crisis, only a relatively small percentage o f the crisis-response resources went toward safety net type interventions. This raises concerns about the impact o f regional governments' actions and potentially important gaps intheir responses. An initialfocus on debt may not have been appropriate- First,regional governments' focus on debt restructuring inthe face o fthe large structural changes occurring in the world coffee market raises questions about whether the incentives created by such programs were appropriate. While the focus on debt may have reflected, in part, a concern about the broader economic impacts o f a crisis in the coffee sector, it i s possible that debt relief also served to delay the exit o f coffee producers who will not be competitive in the long-run, in the face o f the changing nature o f the world market. This may just defer the need for further sector adjustments - and set the stage for additional "crises" - in the future. -and otherpolicies may have beenpoorly targeted Second, review o f the main government responses to the crisis raises questions about the adequacy o f the response to the human dimensions of the crisis. On one hand, the analysis suggests that the government responses were regressive; the vast majority o f resources benefited large- and medium-scale producers, as opposed to small-scale farmers or laborers. On the other vii hand, the workfare programs that were implemented appear to have had little impact on small- scale, self-employed coffee farmers who have experienced the most serious poverty and welfare impacts o f the crisis. Together, the analysis suggests considerable scope for strengthening government support to adversely affected groups inthe face o f a shock. Lessonsfrom the Coffee Crisis-Elementsof a Stratepvto Dealwith Shocks The long- and short-run dimensions o f the coffee crisis suggest the need for an approach that both facilitates people's economic mobility, ex-ante, and strengthens households' abilities to manage risk and shocks, both ex-ante and ex-post. Inthis context, the evidence argues for an integrated, multi-level approach that includes efforts to: 0 Ensure macroeconomic stability and growth 0 Broaden and strengthen people's ability to manage risk ex-ante through o investments inpeople's long-run economic mobility, and o development o f more effective insurance and market-based risk management mechanisms 0 Develop appropriate, well-targeted safety nets, and 0 Strengthen data, information, and monitoring systems Ensuring macroeconomicstability and growth The evidence suggests there are high returns to prudent macroeconomic policies that facilitate growth. Stable macroeconomic environments, coupled with flexible labor markets, can help to soften the blow o f an economic shock, and facilitate households' adjustment to changing circumstances. Indeed, these factors along with moderate growth environments in Nicaragua and El Salvador appear to have enabled coffee laborers to find alternative sources o f income - something that would not have been possible inunstable or stagnant economies. Strengthening ex-ante risk management instruments Households that engaged in risk management investments and strategies, ex-ante, did better in mitigating the impacts of the coffee crisis than those that did not. Another key element o f a country's strategy to strengthen its risk management capacity is thus to strengthen people's ability to manage risk ex-ante. This includes both investments to enhance people 's economic mobility and the development o f more effective insurance - and other market-based risk management - mechanisms. Indeed such investments, along with suitable ex-ante risk management instruments, can also help to reduce the pressure to introduce measures, such as debt relief, that can create perverse incentives ex-post. 0 Key measures to strengthenpeople's long-run economic mobility include: o Investments ineducation to promote greater mobility, adaptability to changing economic circumstances o Deepening o f rural financial markets to break liquidity constraints to enterprise development o Investments in road and transportation infrastructure to increase people's access to labor and product markets, reduce transactions costs, and raise enterprise profitability viii o Improving information, for example, on prices, trends, and changing market circumstances improves the environment for investments and economic planning among individuals, households, and firms. It is worth noting that these measures coincide quite closely with the core elements o f the "Complementary Agenda" set forth in a forthcoming study by the World Bank on strengthening countries' abilities to benefit from the Central America Free Trade Agreement (CAFTA). Indeed, they operate through many o f the same channels, enabling people to adapt more effectively to changing economic circumstances. 0 Insurance and other market-based risk management mechanisms o In the case o f shocks to internationally traded commodities, such as coffee, maize, soybeans, sugar, wheat, and some livestock, market-based risk instruments like futures or options markets can play an important role an integrated risk management tool-kit. While these instruments have long been available to large-scale producers, pilot efforts by the World Bank are currently underway in El Salvador and Nicaragua to explore practical ways to provide access to these risk markets to smallholder farmers. o Emerging "index insurance" instruments, such as area-yield and/or weather-based insurance, may also hold promise for helping farmers deal with natural or economic shocks in agriculture (while reducing the problems o f "moral hazard" and "adverse selection" associated with traditional crop insurance). Ongoing applications o f weather- based insurance in Canada, Mexico, and Argentina may provide valuable lessons for future use inCentral America. Developing an effective safety netfor shocks Even with strong ex-ante investments and risk management instruments, households may still require additional support following a shock. As such, the development o f a flexible safety net i s a critical element o f a country's response to shocks. Several principles stand out: Developing an effective safety net response to shocks requires pre-shock preparedness for various contingencies, including which institutions and types o f programs might play the most effective roles inthe face o f different shocks. Ensuring that a program is well-targeted to affected groups is important to assuring program impact, particularly in Central America where governments face tight fiscal constraints. In addition to increasing the efficiency o f interventions, targeting can also serve to minimize the risk o f perverse work incentives to non shock-affected populations. Developing the capacity for counter-cyclical financing and implementation i s also important to ensuring impacts, establishing appropriate incentives, and using fiscal resources effectively and efficiently. Developing counter-cyclical mechanisms requires governments to develop the fiscal discipline and a reliable financing mechanism that can take affect in the face o f a shock. It also requires that governments consider prior to a shock eligibility rules and exit strategies (such as time-limited eligibility) to ensure that participants exit from the program once the shock-related need has passed. Cross-country experience also suggest that responses to shocks are most effective - and easiest to scale up and down in the face o f changing circumstances - if they build on existing program and institutional capacityprior to the shock. ix Finally, designing flexibility into a country's response mechanisms i s paramount. This includes developing contingency procedures into existing social assistance institutions and programs, procedures that will enable agencies to respond quickly and appropriately once the specific nature o f the shock and its impacts have been determined. Which specificprogram is most effective depends on the circumstances Several types o f programs implemented recently in Latin America have - or could be - used to address shocks. Recent experience suggests that different programs have different strengths, however, depending on the nature of the shock and o f affected groups. Different safety nets also pose different administrative and implementation challenges. Specifically: Workfare programs (food-for-work, cash-for-work) are best suited to address employment shocks, where the opportunity cost o f participating i s low among affected groups. Workfare programs are relatively simple to administer and, if well-designed, can be self-targeting. The ability to make workfare programs self-targeting, however, depends on whether the regulatory environment allows authorities to set program wages below the prevailing market wage, so as to attract only those who are truly inneed. Decoupled income support, transfer programs in which payments are de-linked from current (or future) prices or production to minimize economic distortions, are better suited to dealing with price and income shocks faced by rural producers - e.g., from trade reform or other terms-of-trade shocks. Decoupled transfers are more complex to administer than workfare programs. This is due, in part, to the need to identify affected producers and verify their eligibility which, historically, has relied on good cadastral and/or land-use records. Conditional cash transfers (CCTs) provide cash payments to families conditional on their making specified investments in children's human capital. While traditionally CCTs have focused o n reducing structural poverty through investments in human capital, new evidence indicates that they are also effective in protecting families against price and income shocks, as well as against adverse effects on children's education and nutrition. CCTs are relatively complex to administer, however, requiring both the capacity to target eligible groups and to monitor their compliance with program conditions. Temporary fee waivers for school and healthcare services can be used specifically to address the effects o f shocks on people's human capital. By administering benefits through local service providers, this approach uses a relatively straightforward targeting mechanism, although its effectiveness depends on local institutional capacity. Programs based at schools and healthcare facilities are most effective when the shock has a strong geographic dimension and when the impacts on human capital o f an income or employment shock are strong. Strengthening data, information, and monitoring systems The differences between observers' initial expectations about the coffee crisis impacts and its actual impacts highlight the importance o f establishing data and surveillance systems to understand the both the nature o f the crisis, and the key characteristics o f affected groups. In the case o f Nicaragua, for example, the workfare programs implemented by the government were probably less effective than desired, as they were better suited to assist newly unemployed laborers than to farmers who faced a dramatic decline in the value o f their product. In contrast, based on evaluation o f the Red de Proteccibn Social, which addressed both the income and human capital dimensions o f the crisis, some kind o f conditional or unconditional transfer program might have been more effective. Better information on impacted groups earlier on, as X well as explicit monitoring o f program impacts, may have facilitated the fielding of more appropriate programs with greater impact. No one sizefits all Finally, it is important to note that in spite o f the similarities o f coffee crisis impacts across the four study countries, the most effective responses likely differ across the countries, given differences in the structures o f the coffee sector, different geographic distributions o f the shock, differences in existing programs, institutions and institutional capacity, and so on. This reinforces the need for good forward planning and information inensuring effective responses to shocks. xi 1. Introduction: Shocks and Social Protectionin CentralAmerica 1.1 Central America i s a shock-prone region. Since the mid-l990s, the countries o f Central America have experienced a number o f natural shocks, including Hurricane Mitch (1998), earthquakes (El Salvador, 2001), and a series o f seasonal droughts and floods (often associated with ElNiiio and L aNiiia); as small open economies, the Central America countries are also open to a variety of economic shocks, whether in the form o f external terms-of-trade shocks, like the recent "coffee crisis," policy-induced terms o f trade changes, like those expected from the Central America Free Trade Agreement, CAFTA'), or more generalized slowdowns in the US. and global economies. 1.2 This report is about the impact o f shocks on people's welfare in Central America and about public responses to shocks. Whether natural disasters or economic events, shocks can have traumatic effects on people's well-being in the form o f reduced income and consumption and increased poverty. Income and consumption shocks often generate secondary effects on children's nutrition, health, and access to schooling. Even ostensibly short-term shocks can thus have long-term impacts on family well-being, productivity, and economic prospects. For families that are already poor, such impacts could mean the difference between exiting or remaining in poverty for yet one more generation. Well-designed and timely public actions to help families deal with shocks can thus be critical not only to determining how shocks affect people's well- being inthe short-term, but the extent to which they have detrimental long-term impacts. 1.3 This report focuses on shocks and social protection policy in Central America, focusing on the recent coffee crisis - a sharp decline in the world coffee price in the late 1990s and early 2000s - as a case study. The report brings new empirical evidence on the welfare impacts o f the coffee crisis in four Central American countries - El Salvador, Guatemala, Honduras, and Nicaragua. Based on the insights from this analysis, along with other recent evidence on shocks and social protection in the region, the report then draws lessons for the region's governments (and donors) on how to develop effective public strategies to address future shocks. Shocks in the CentralAmerican Context Central American households are exposedto a wide variety of shocks- 1.4 A recent household survey from Guatemala (ENCOVI 2000) sheds light on the prevalence o f shocks in Central America. In 2000, over half o f all households - 53 percent - reported experiencing one or more shocks. Nearly a quarter (23 percent) reported being hit by a natural shock in the past year, 17 percent reported having experienced a "man-made" - or economic - shock, and 13 percent reported having experienced both types o f shocks.* The most common types o f shocks reported were agriculture-related - pest infestations and harvest losses; other commonly reported shocks included job loss, drought, and worsening terms-of-trade. Moreover, the majority o f households reporting shocks in 2000 were hit by more than one. A quarter reported having experienced two shocks that year, while another 25 percent reported three or more shocks. While the reported incidence o f natural shocks was somewhat greater for the ' TherecentlynegotiatedCentralAmericaFreeTrade Agreement (CAFTA) representsan importantgrowthanddevelopment opportunity for the CentralAmerican counrries. Nonetheless,there are concernsabout howwell certainruralhouseholdswill beable to dealwith changingterms oftrade inagricultureassociatedwiththe liberalizationof trade in so-called"sensitive" agricultural commodities,suchas maize, beans,dairy, beef, andpoultry. GuatemalaPoverty Assessment, World Bank, 2004. poor than for the non-poor, the incidence o f multiple shocks was similar across poor and non- poor groups. -including natural disasters- 1.5 The impacts of several recent large natural shocks in Central America have been significant. For example, in late-October, 1998, Hurricane Mitch hit Honduras, Nicaragua, and to a lesser extent, El Salvador and Guatemala. InHonduras, the storm-which followed three weeks of rainthat had already soaked the countryside - pounded the country for three days. Torrential winds, floodwaters, and mudslides destroyed villages, shattered social and economic infrastructure, paralyzed production, and left up to three feet o f mud and debris throughout the country. Communities cut off from economic activity and from basic services faced immediate health concerns and security risks. Some 6,000 people died, 8,000 were missing, 13,000 were wounded, and more than a millionwere homeles~.~ InNicaragua, as many as 3,000 people also died and over 850,000 were by displaced by the storm. Although Mitch didnot appear to have a major effect on Nicaragua's poverty situation at the national level, the hurricane did have a short-term negative effect on poverty in the country's Central rural region, due to Mitch's impact on agriculture and to the region's relatively difficult accessibility and high levels of ~ulnerability.~ 1.6 In January and February, 2001, two powerful earthquakes hit El Salvador, resulting in over 1,000 casualties and leaving more than 300,000 families homeless. As many as 85 percent o f rural households in the main earthquake-affected areas experienced some damage to their homes and as much as 40 percent o f households in affected areas experienced damage severe enough to make their homes ~ninhabitable.~Although the earthquakes did not result in an increase in income poverty at the national level, they did have important short-term impacts in those localities hardest hit by the quakes. About 3 percent o f the population in the most affected departments fell into poverty following the earthquakes.` Moreover, the earthquakes adversely affected basic education outcomes in the most-affected areas, where nearly 4 percent o f children left school, at least temporarily. -to which thepoor are often the most vulnerable- 1.7 Although the geographic location o f a natural shock determines in large part who i s affected and how, several factors lead to particularly highrisks among the poor who are affected. For example, the poor are more likely to live on marginal lands or in high risk geographic areas that are vulnerable to secondary impacts, such as mudslides; and they are more likely to reside in low-quality housing which can collapse during earthquake or other extreme weather events.' Poor households also tend to have fewer savings and liquid assets than the non-poor, meaning that they tend to have fewer "instruments" with which they can manage risks or cope with the impacts o f a shock. Further, destruction o f infrastructure during a natural disaster - e.g., rural roads used by smallholder farmers to sell their production - may particularly affect poor households' access to markets and, thus, their abilities to bounce back after an event. 'Warren (2003). Nicaragua Poverty Assessment, World Bank, 2001, Benekede Sanfeliu and Shi, 2004. 'ElSalvadorKruger, Poverty Assessment, WorldBank,2004 Klugman, andWithers (2004) 2 1.8 While recent natural disasters have had high human tolls, several recent World Bank studies suggest that the impacts could have been worse - and longer lasting - in the absence of timely responses by the region's governments and by the international community.* In Nicaragua, for example, post-Mitch reconstruction and investment spending, supported by large in-flows of donor resources, played an important role in offsetting the potential negative impacts of the hurricane on household income and consumption. Moreover, quick mobilization of resources and programmatic responses by existing institutions, such as the Honduras Social Investment Fund (FHIS) and the Social Investment Fund for Local Development (FISDL) in El Salvador, enabled quick and effective relief and rehabilitation activities that helped to mitigate the most serious impacts o f M i t c h and the 2001 earthquakes, respectively. Indeed, evidence from Nicaragua and El Salvador suggest that the resources and activities contributed not only to the short-term recovery, but to improved poverty and social indicators in affected areas. The experience in Central America is broadly consistent with a body of international evidence on shocks and the role o f social protection (Box 1.1). Box 1.1: Shocks, Household "Self-Insurance", and the Role of Social Protection People in Central America face a number o f risks. Some risks, such as economic recessions, hurricanes, earthquakes, drought or floods can affect entire societies or, at least, large portions o f the population. These are commonly referred to as "covariate" or "aggregate" risks. Other risks, such as the illness o f family members or loss of a household breadwinner's job may affect only individual households or families. These are commonly referredto as "idiosyncratic" risks. Inthe face of such risks-and inthe absence of adequate insurance markets -households generally adopt a range o f private, informal risk management strategies. Common strategies include diversification o f income sources, sale o f assets, migration and remittances, and sending additional family members - including often school-aged children - into the workforce. A growing body o f international evidence indicates that households are only partially successful in "insuring" themselves against risk and the impacts o f shocks, however. Indeed, recent evidence from several Latin American countries, including Mexico, Nicaragua, and Peru, suggests that households in these countries are able to protect (or insure) only between 60 and 75 percent o f their per capita consumption in the face of an income shock. In other words, a 10 percent shock to a household's per capita income translates on average into 2.5-4 percent decline in their per capita consumption. The evidence also indicates that poor households are often less able to "smooth" their consumption in the face o f shocks than non-poor households. A recent study o f income dynamics in rural El Salvador indicates, for example, that poor households take considerably longer to recover from an income shock than non-poor households. Evidence from rural China also indicates that income shocks have deeper impacts on poor households. While an income shock o f 10 percent had only a one percent impact on the consumption o f the wealthiest households, it caused a 4 percent decline inconsumption among the poorest households. While social protection responses to shocks - especially those targeted to the poor - may be justified on welfare grounds, recent evidence suggests that there can also be important efficiency reasons. Indeed, recent studies suggest that too much uninsured risk has signiJicant costs to growth andpoverty reduction, as the poor continue to invest in safe, but low-productivity activities within and outside agriculture. Such low-risk, low-return investments can serve to perpetuate low productivity and, thus, low living standards among the poor. World Bank (2001), WorldBank (2003), SPectrum (2003), World Bank (2004). 3 For both efficiency and householdwelfare-related reasons, therefore, there are rationale for governments to engage in efforts to assist households andindividualsinmanaging risks andshocks. Sources: Glewwe and Hall 1998, Jalan and Ravallion 1999, Klugman, Kruger, and Withers 2003, Ravallion 2002, Rodriguez-Meza and Gonzalez-Vega 2004, Skoufias 2002, Spectrum 2003. For more information on recent studies on household consumution smoothincr and self-insurance. see Annex 1. -and economicshocks, whether caused by externalfactors or domesticpolicy 1.9 Economic shocks are also an integral part o f the Central American landscape. While Central America's macro-economic situation is relatively stable, countries are exposed both to external and policy-induced terms o f trade shocks. In the context o f policy, Costa Rica, El Salvador, Guatemala, Honduras, and Nicaragua - along with the Dominican Republic - recently completed negotiations on a Free Trade Agreement with the United States. Although the vast majority o f Central Americans are expected to benefit from the Agreement, known as CAFTA, there are specific sub-populations - e.g., net producers o f agricultural staple crops - who, in the absence o f specific policy measures, would be exposed to adverse terms-of-trade changes. While CAFTA includes provisions to phase out tariffs and quotas gradually on agricultural staples, providing an extended adjustment period to affected farmers, new empirical analysis o f the likely impacts o f CAFTA finds that nearly 20 percent of rural households inNicaragua and 25 percent o f rural households in Guatemala could experience negative income effects, if the negotiated safety net provisions (or alternative mitigating measures) are not implemented. As with natural shocks, the risk o f experiencing negative impacts tends to be higher among the poor than the non-poor.lo The coffee crisis, a steep decline in world coffee prices, wasparticularly noteworthy 1.10 Among the most important external economic shocks to affect the region recently has been the "coffee crisis" - a sharp decline in world coffee prices - that has had important impacts on the region's families who depend on coffee-sector income. Indeed, structural changes in the world coffee market have significantly affected the coffee sectors in Guatemala, Nicaragua, Honduras, El Salvador, and Costa Rica. The recent entry o f several new producers (most notably Vietnam), as well as dramatic production growth in other parts o f Latin American (most notably Brazil), led to unusually steep declines in international coffee prices during the late 1990s. By 2001, these forces helped pushreal coffee prices to their lowest levels inmore than 50 years, well below long-term trend levels. Coffee prices even fell below the average cost o f production in several producing countries. 1.11 World coffee prices have since rebounded somewhat (due in part to weather-related shocks in Vietnam). Nonetheless, the precipitous drop in coffee prices between 1997 and 2001 was a cause o f great concern throughout Central America. Several early assessments o f the coffee crisis raised serious concerns about the poverty and social impacts o f the crisis -especially related to apparently large declines in the demand for permanent and seasonal labor in the coffee sector following the coffee price declines. These assessments were based largely on aggregate, coffee industry data. And while such information was instructive, it did not enable a detailed understanding o f how the crisis affected the well-being o f coffee sector families; nor did it provide sufficient informationto facilitate the most appropriate and effective policy responses. "Net producers" are definedas householdsthatproducemoreof agiven commodity than they consume, Le., they sell at least some o f their productioninto the market. Inthe absenceo f mitigatingmeasures,netproducersof agricultural staples slatedto be liberalized underCAFTA would standto lose fromthe expectedfall inthe domestic pricesofthose goods. IoMason (2005). 4 Obiectives of the Studv Analyzing the impacts of the coffee crisis- 1.12 This report was commissioned in response to requests from several Central American governments to better understand the impacts o f the coffee crisis on household income and consumption, poverty, and several key related outcomes, such as child education and nutrition. To that end, much o f the remainder o f this report focuses on the case o f the coffee crisis in four coffee-producing countries in Central America - El Salvador, Guatemala, Honduras, and Nicaragua - both to understand its impacts on the people's well-being and its lessons for public actions with respect to shocks. -and drawing lessonsfor public responses to shocks 1.13 The objective o f this study extends beyond simply understanding the impacts o f the coffee crisis or its lessons for future coffee price shocks, however." Rather, based on the findings o f the empirical work commissioned for this report, as well as other recent evidence on shocks and social protection, the objective o f this report i s to present the broader lessons for Central America regarding shocks and social protection. This study i s thus intended to support Central America's policymakers, as well as their development partners, in strengthening their ability to address future shocks, whether within or outside the coffee sector. By learning the lessons from the coffee crisis, as well as from other recent shocks, the region's policymakers can be better prepared to protect their citizens from the most serious and long-lasting effects o f adverse economic or natural events. Data andAnalvsis The report brings together evidencefrom several newly commissionedstudies 1.14 To achieve its objectives, this report draws on several newly commissioned studies o f household panel surveys, on the efficacy o f government responses to the crisis, and on the role o f existing safety net programs in helping to mitigate the crisis impacts. The new empirical evidence presented here draws on an unusually rich collection o f panel data from Nicaragua, El Salvador, and Honduras, as well as on a detailed cross-section o f household survey data from Guatemala (Table 1.1). The report also draws on several recent World Bank Poverty Assessments and related studies on shocks, safety nets, and social protection policy in the Latin America region. Country Data/Analyses Nicaragua - Living StandardsMeasurement Surveys (LSMS), apanel o f the same householdsin 1998 and 2001, just before the outset and inthe midst o f the crisis; analysis o f crisis impacts on household welfare andhousehold risk management strategies - International FoodPolicy ResearchInstitute (IFPRI) panel data from 2000, 2001,2002; collected to evaluatethe households impacts o f the Red de Proteccibn Social (RPS), a Nicaraguan safety net program; analysis o f the impacts of the RPS inmitigating the impacts ofthe coffee. El Salvador - BASIS RuralHousehold Income Surveys, collected by FUSADES in collaboration with Ohio State University; panel data following the same I'As will bediscussedin Chapter2, the internationalcoffeemarketis characterizedby highprice volatility as well as by the long-term secuiardecline incoffee prices. As such, periodic"coffee crisis'' are a predictablepart ofthe world coffee economy. The lessons of this reportshouldthus provideuseful guidance for dealingwith this eventuality. 5 I rural households in 1995, 1997, 1999, and2001, prior to andincluding the crisis period. Analysis o f the impactsof the crisis on householdincomes andhumancapital outcomes. Honduras - InternationalFoodPolicy ResearchInstitute (IFPRI) panel data from 2000 and 2002; collectedto evaluate the householdsimpacts ofthe Programa deAsignacidn Familiar (PRAF), a Honduransafety net program; analysis o f the impacts of the coffee crisis on householdwelfare, as well as o f the impacts of the PRAF inmitigatingthe impacts ofthe coffee. Guatemala - ENCOVIhouseholdsurvey data from 2000 (modeledon LSMS);included special data module on householdresponses to shocks; analysis builds on earlier work on the welfare o f coffee sector householdsinthe Guatemala PovertyAssessment (2003). Thedatafrom Nicaragua and El Salvador areparticularly rich- 1.15 While the report is regional in nature, it should be noted that the relative balance o f country-specific evidence largely reflects the availability o f country-level survey data. In the context o f the coffee crisis, the largest body o f country evidence presented in this report comes from Nicaragua, which has particularly rich sets o f data with which to examine crisis impacts. The "core" analysis for Nicaragua uses panel data from the Nicaragua Living Standards Measurement Study (LSMS) surveys from 1998 and 2001, which include detailed informationon household consumption, income, and a number o f key social indicators and outcomes for the family. The 1998-2001 time period bracket the crisis period almost perfectly, enabling analysis o f changes in household welfare over a period o f a nearly 60 percent decline in the world coffee price. This core analysis is supplemented by an evaluation o f the extent to which the Red de Proteccidn Social (RPS), a Nicaraguan safety net program, was successful in mitigating the negative effects o f the coffee crisis. A separate household panel was used for this assessment, data that was collected by the International FoodPolicy Research Institute (IFPRI) specifically to evaluate the impacts o f the RPS on a range o f householdwelfare indicators. 1.16 The El Salvador country analysis also draws on an unusually rich panel data set, with income data collected from the same rural households in four separate waves - in 1995, 1997, 1999, and 2001. As in the case o f Nicaragua, the data brackets the pre-crisis and crisis periods well (specifically, 1997-2001). In contrast to the Nicaragua LSMS surveys, the El Salvador data sets lack detailed information on household consumption; nevertheless, the surveys do have detailed data on household income and on several other key social indicators at the household level, which facilitate detailed analysis o f the crisis impacts. -while datafrom Honduras enable analysis of impactsfrom a subset of rural areas 1.17 The household surveys usedto analyze the crisis impacts inHonduras are also panels. In contrast to the Nicaragua and El Salvador data sets, however, the Honduras data are not statistically representative o f rural areas. The Honduras surveys, also collected by IFPRI, were collected specifically to evaluate the impact o f the Honduran safety net program, Programa de Asignacidn Familiar (PRAF), and thus focused on enabling an unbiased analysis of the PRAF impacts, rather than on beingrepresentative o f all rural areas.l2 In the case o f Guatemala, only a single cross-section of household survey data was available with which to analyze the welfare of households working in the coffee sector - the ENCOVI 2000 survey, a nationally representative householdsurvey which, like the Nicaragua surveys, is modeled on the LSMS. While Guatemala The survey covers households in 70 (out of 297) of the most disadvantagedmunicipalities in7 departments(Copan, Intibuca, Ocotepeque, F. Morazan, La Paz, Sta. Barbara and Lempira). These municipalities are all located inthe westem part ofHonduras and cover many of the countries major coffee-growing departments. For details, see Coady et a1(2004). 6 i s the only country in the study for which there i s no inter-temporal data, the survey i s unique in that it included a special module on households' experiences with and responses to shocks. Thus, in its own way, the Guatemala survey enabled an extremely interesting and rich analysis of household-level shocks. Organizationof the Report The mainfindings are synthesizedhere, with more details available in a second Volume 1.18 This report is organized into two volumes. This volume (Volume I) a synthesis presents o f the key findings and policy recommendations. Volume I1 presents a collection o f the commissioned background studies to the report. These background papers provide considerable analytical detail on each country for readers who are interested in more in-depth information on the country-level impacts, the unique data sets underlying the analyses, andor the methodologies underpinning the analytical works summarized inthis volume. 1.19 The remainder o f this Volume is structured as follows. Chapter 2 focuses on the impacts o f the coffee crisis. Specifically, the chapter examines: 0 the socio-economic impacts of the coffee crisis inNicaragua, El Salvador, Honduras, and Guatemala; 0 the risk management strategies Central American households used to deal with the crisis; and the main government responses to the coffee crisis. The next chapter shows that small scale coffee farmers were hardest hit byfallingprices- 1.20 The evidence presented in Chapter 2 shows that the coffee crisis indeed had significant negative impacts on smallholder coffee farm households inthe region- on per capita income and consumption, on poverty, and on children's education and nutritional status. Contrary to early expectations, the welfare impacts o f the crisis were most severe on small-scale, self-employed coffee farmers and not on coffee laborers, as the crisis impacts were transmitted mostly through the price and income effects, rather than through increased unemployment. -and that householdswere onlypartly able shield themselvesfrom the effects of the crisis 1.21 In response to the crisis, coffee sector households in all four countries used a variety o f ex-ante and ex-post strategies to try to prevent, mitigate, or cope with the effects o f the crisis. This included diversification o f household income sources, migration and remittances, and increasing family labor supply, among other things. Nonetheless, the evidence shows that coffee sector households were not able to shield themselves fully from the effects o f the crisis. In fact, insome cases, attempts to cope with the crisis involved pullingchildren out of school or reducing household food consumption, which raises concerns about the long-term effects on children's human capital development, including potentially detrimental and irreversible effects on children's future economic productivity and welfare. Moreover, government responses did relatively little toprotect the most affected families 1.22 While the region's governments did initiate some programs inresponse to the crisis, these focused predominantly on debt restructuring for large- and medium-sized coffee producers, with relatively little emphasis being placed on mitigating the welfare impacts on small farmers and coffee sector laborers. Indeed, the evidence suggests that government responses did relatively little to shield small farmers, or laborers, from the worst effects o f the crisis. 7 Chapter 3 outlines key elementsof a strategyfor dealing with shocks- 1.23 Building on this understanding o f the coffee crisis impacts - and on related evidence on shocks inthe region- Chapter 3 outlines several broad lessons for public policy regarding shocks and social protection. Specifically, the chapter highlightsthe importance of: 0 maintaining macroeconomic stability and growth; 0 strengthening people's capacity to manage risk through policies and investments that enhance people's economic mobility and through development o f stronger market-based riskmanagement instruments; 0 developing, in advance, the institutional and policy framework to ensure timely implementation o f well-targeted safety nets following a shock; and 0 strengthening data, information and monitoring systems to support appropriate safety net design and effective program implementation and follow-up. 1.24 Chapter 3 examines the relative strengths and administrative requirements a several safety net programs that have been used recently in Latin America. The analysis shows that the choice o f the most effective response to a shock will depend not only on the specific nature o f the shock impacts but, critically, on existing country-level institutions and programs as well as on local capacity to administer alternative programs. Chapter 4 summarizes the implicationsfor future public responses to shocks 1.25 Chapter 4 concludes this volume with a brief summary of the main findings o f the report and their implications for government strategies to deal with shocks in the future. While many studies focus largely or exclusively on ex-post safety nets implemented by public agencies, one key message o f this report is the importance o f efforts to strengthen households' risk management capacity ex-ante as well. In this context, a second key message i s that there is scope for further developing private market instruments, such as insurance (depending on the type o f the shock), as part o f a broader compliment o f public and private instruments available to manage risk and shocks. 8 2. The Welfare Impactsofthe Coffee Crisis: A Tale of Four Countries 2.1 This chapter: (i) examines new evidence on the socio-economic impacts o f the coffee crisis; (ii) reviews the private risk management strategies that Central American households used to deal with the crisis, highlighting how those strategies differed across countries; and (iii) reviews the main responses to the crisis fielded by the region's governments. The CentralAmerican Coffee Crisis: Macro Trends Structural changes in the world coffee market have depressedprices 2.2 The coffee sector has historically been subject to price fluctuations and uncertainty, mostly driven by weather and other supply-related shocks. Nonetheless, in recent years, the coffee industry has been undergoing a worldwide structural change. On the demand side, there has been slow growth, and over time the coffee sector has experienced a shift in consumer preferences towards more differentiated and higher quality coffee products (e.g., gourmet blends, fair trade coffees, organic, shade-grown varieties, etc.). More importantly, on the supply side, the entry o f a number o f new coffee producers in the 1990s (e.g., Vietnam), along with technological improvements that have led to large increases in coffee production among key Latin American producers (e.g., Brazil) have resulted in a dramatic increase in coffee production and improvement in coffee quality. These forces served to severely depress international coffee prices, most dramatically during the 2001-02 period (Figure 2.1); this, in turn, has reduced the competitiveness o f many small coffee producing countries, like those in Central America. Indeed, at their lowest point, world coffee prices had fallen below the average cost o f production for producers in a number o f countries. l3 Figure2.1: Compositecoffeepriceindex, 1984-2004 250 I 1984198519861987 1988198919901991 199219931994199519961997199819992000200120022003200~ Year Source: IC0 Website. l3Lewin andGiovannucci2003. 9 2.3 While prices have subsequently rebounded, the crisis representedthe culmination o f two characteristic features o f the world coffee market: (i) long-term secular decline in the world the coffee price, and (ii)significant price volatility, often driven by weather-related shocks, that by late 2001 had left real prices at their lowest levels in more than 50 years. As can be seen in Figure 2.1, these price lows followed quite closely upon two short-lived upward price spikes duringthe mid-to-late 1990s.l4 Coffee has been one of the most important cash crops in the region- 2.4 Coffee has long been one o f the most important cash crops for the Central American economies. In fact, together the four countries included in this case study - El Salvador, Guatemala, Honduras and Nicaragua - along with Costa Rica - comprise the second largest coffee producer, after Brazil." Moreover, the coffee sector is an important source o f employment. An International Coffee Organization (ICO) estimate suggests that 200,000 farmers produce coffee and between 2-3 million workers find seasonal or permanent work in the sector. At the same time, the coffee sector in Central America is heterogeneous, characterized by a combination o f small-, medium- and large-scale farmers; the latter may work on their own coffee farms, but also offer an important source o f permanent andor seasonal employment opportunities to agricultural workers. 2.5 In terms of recent production trends, between 1998 and 2001, world coffee production increased from 108 million bags to 124 million (Table 2.1). In contrast, combined coffee production in the four study countries decreasedover the period. Indeed, coffee production in El Salvador, Guatemala and Nicaragua declined by almost 10 percent; only in Honduras did it increase and, then, only slightly. Table 2.1: Coffee productionin CentralAmerica, 1998-2003 (in millions of bags) Crop year El commencing Salvador Guatemala Honduras Nicaragua Brazil World 1998 1.8 4.3 2.5 1.1 35.6 108.5 1999 2.6 4.4 3.1 1.5 30.8 113.4 2000 1.6 4.6 2.8 1.6 34.1 117.7 2001 1.6 3.5 3.1 0.9 35.1 110.9 2002 1.4 3.8 2.7 1.o 51.6 123.9 Source: IC0 Website. -produced by a mix of large fincas, medium-scale, and many small-scalefarms 2.6 In terms of the structure of production, the coffee sectors in El Salvador, Guatemala, Honduras, and Nicaragua are all very heterogeneous. The majority o f the coffee producers are small-scale farmers. In Honduras, for example, an estimated 92 percent o f the producers are considered small-scale, producing less than 100 quintals per year (Table 2.2); only 3 percent o f Honduran coffee farmers are defined as large-scale. On average in the four countries, large-scale While it i s clear that realworld coffeepricescannottrenddownwardindefinitely- i.e., that therewill be point at which adjustments on the supply side will roughly equilibratewith demand- it is not obviousthat long-termprice trendswill reverseinthe short-run, absent amajorshift inthe structureofworld demandfor coffee (e.g., astructuralincrease in demandfor coffee inChina or other large potentialconsumingcountries). For a more detaileddiscussionof supply anddemand forces inthe world coffee market, see Lewin andGiovannucci(2003). '*KO. 10 farms orfincus only comprise about five percent o f all coffee farmers. At the same time, except in Honduras, large- and medium-scale producers dominate, in terms o f the quantities produced. For example in both El Salvador and Guatemala, large farmers produce more than half o f all coffee produced; medium and large-scale coffee farmers represent more than 85 percent o f coffee production in Nicaragua. Accounting for this heterogeneity, and particularly for the large proportion o f small-scale farmers in the sector, will thus be important to understanding the welfare impacts o f the coffee crisis. Table 2.2: Distributionof coffee farmers and productionin 2000 Small a) Medium b, Large % farmers ElSalvador 81 15 4 Guatemala 80 14 6 Honduras 92 5 3 Nicaragua 90 9 1 YOproduction ElSalvador 10 32 58 Guatemala 20 30 50 Honduras 45 44 11 Nicaragua 14 50 36 a)Small-scale: less than 1OOqq; b,Medium:ElSalvador 100-1000 qq, Guatemala 100-2000qq, Honduras 100-1000 qq, Nicaragua 100-1500qq. Source : Adapted from Varangis et al. (2003) Loss of export revenuesdue to theprice drop equaled roughly 1% of GDP 2.7 Due to the severe drop in coffee prices, coffee export revenues in the four countries declined dramatically during the crisis period. Between 1999 and 2001, for example, coffee export revenues declined by an average o f 45 percent across the four countries (Table 2.3). As a consequence, the importance o f coffee exports to total exports has also decreased by about 50 percent in each o f the four countries (Figure 2.2). Losses in export revenues corresponded to approximately 1 percent o f GDP, on average, across the four countries, and put considerable pressure on these countries' balance o f payments during the period.16 Table 2.3: Coffee exports revenuesfrom CentralAmerica, 1999/2000and 2000/2001 (inmillions o fUS$) 1999/2000 2000/2001 Change (%) ElSalvador 276 108 -61 Guatemala 598 400 -3 8 Honduras 345 167 -33 Nicaragua 170 85 -50 Total 1389 760 -45 Varangis et al. (2003) l6World Bank2002a. 11 Figure 2.2: Coffee exportsas ashare of totalexports 28 -s 30 3 3 g 20 0. 8 Q) a 10 u c s0 0 1998 1999 2000 2001 2002 1-Nicaragua ". Honduras -Guatemala -E Salvador 1 Source: IC0 Website. Initial concerns about social impactsfocused on unemployment among coffee workers- 2.8 Early concerns about the social impacts o f the coffee crisis focused on the employment and incomes o f thousands o f permanent and seasonal coffee plantation workers in each country, many o f whom own no cultivable land o f their own. Indeed, as noted above, coffee sector employment is significant in Central America-in the form o f both self-employment on family farms and in permanent and seasonal wage employment on coffee plantations. It was estimated that in 2002, for example, more than 3,000,000 people were involved as laborers in the coffee sector (Table 2.4). This suggests that nearly 30 percent o f the rural labor force in these countries, on average, has some connection to the coffee economy. Table 2.4: EstimatedCoffeeSector Employmentin CentralAmerica, 2002 Labor force incoffee Labor force incoffee as % o frural labor force ElSalvador 936,000 17 Guatemala 700,000 31 Honduras 1,152,000 26 Nicaragua 672,000 42 Source: ECLAC 2002. 2.9 Initial estimates o f the impact o f the crisis on employment suggested significant effects on coffee sector employment. In El Salvador, the Consejo SalvadoreAo del Cafi (Salvadoran Coffee Council) estimated that employment in coffee productionfell from 185,630 individuals in 2000 to just 58,800 in 2003.17 In Guatemala, an estimated 200,000 people were permanently "Consejo SalvadoreiiodelCafk. 12 employed in the coffee industry and another 300,000 were seasonally employed;" o f those workers, approximately 40,000 jobs relatedto coffee production were thought to have been lost in 2002 alone. l9 Similarly, in Nicaragua, initial estimates suggested that 35,000 permanent and more than 100,000 seasonal coffee plantation jobs were lost in 2001.20 Overall, roughly 20 percent o f the seasonal coffee sectorjobs and more than 50 percent o f the permanent coffee sector jobs inCentral America were thought to have beenlost between2000 and 2002 (Table 2.5). Table 2.5: EstimatedDeclinein CoffeeSector Employment in CentralAmerica, 2000/01-2001-02 2000/0 1 2001/02 Change % Seasonal 1,700,000 1,350,000 -2I Permanent 350,000 i60,000 -54 Source: Varannis, et a12003. Note: IncludesCostaRica. -but did not account adequatelyfor laborers' abilities tofind work in other sectors- 2.10 While the estimated job losses associated with the coffee crisis are clearly significant, aggregate data such as these figures need to be interpreted carefully. This is because the estimates do not necessarily reveal the net job losses - or corresponding income or welfare losses -associatedwiththecrisis. Inparticular, theseestimatesrefersolelytothelossofemployment inthe coffee sector, and do not consider that laid-off coffee workers could potentially find work in other sectors. Infact, given that several of the Central America countries were experiencing positive economic growth at the time o f the coffee crisis, it is likely that at least some workers laid off from coffee sector jobs found alternative employment, without extended periods o f unemployment. At least in principle, in such a growth context, coffee sector laborers and their families may not have experienced a significant loss in income or welfare as a result of the crisis,2' Moreover, as will be discussed further below, households often employ risk management strategies inthe face o f expected or actual economic shocks that help them to deal with shocks or other economic changes. Such actions, like sending additional family members into the labor force inthe face o fjob uncertainty or job loss, may help to soften (or perhaps even eliminate) the negative income effects of the shock. 2.11 Indeed, analysis o f panel data from Nicaragua, El Salvador, and Honduras, undertaken for this study, indicates that the main impacts of the crisis were more nuanced than originally envisioned. On one hand, analysis o f household survey data from rural Nicaragua between 1998 and 2001 confirms that there were negative employment effects resulting from the coffee crisis; membersofNicaraguanhouseholds that remained involvedinthe coffee economy over the 1998- 2001 period experienced significant increases in unemployment at a time when unemployment rates in other rural sectors remained largely unchanged.22 Ministry of Agriculture, Guatemala. l9ANACAFE put this figure at 60,000. 2oIbid. "Especiallyinthecaseofseasonallabor,suchworkersaregenerally"footloose" andarelikelyhavemultiplesourcesofincomeof income (bothacross family members and over time) that, with small changes, may enable themto smooththe incomeimpactof a decline inthe demandfor labor incoffee plantations. 22Vakis et a12004. 13 -or for the direct effect ofprices on small-scale coffee farmers' incomes 2.12 On the other hand, the analysis indicates that the most severe impacts o f the crisis in Nicaragua were not transmitted primarily through increased unemployment. Indeed, households whose members worked as coffee laborers in 1998 managed largely to maintain their income and consumption levels in 2001, at least in part, by finding other earnings opportunities outside o f agriculture. In contrast, small-scale, self-employed coffee farmers experienced substantial declines in income and welfare as prices dropped between 1998 and 2001. The Socio-Economic Impacts of the Coffee Crisis Realper capita income growth in CentralAmerica between 1995 and 2002- 2.13 Overall, the 1995-2002 period is characterized by modest-to-high economic growth in Central America. Real GDP growth rates in El Salvador, Nicaragua, Honduras, and Guatemala averaged about 4 percent per year - ranging from 3 percent in Honduras to 5 percent in Nicaragua.23GDP per capita grew at a rate of 1percent per year, on average, inthe four countries over the same period. -translated into a general reduction inpoverty- 2.14 Partially in response to economic growth, overall poverty declined during this period. Between 1998 and 2001, for example, poverty in Nicaragua declined by 4 percent to achieve a headcount o f 46 percent in 2001.24 Poverty rates declined faster in rural areas than in urban areas. Similarly, rural poverty in El Salvador decreased from 65 percent in 1995 to around 52 percent in 2001.25In all four countries, other socioeconomic indicators - such as education and health - improved over this period. This is reflected in the Human Development Index (HDI), which synthesizes health, education and growth dimensions o f development. The HDIimproved inall four countries over the 1995-2002 period(Table 2.6). Table 2.6: Trends in Human Development Index in Guatemala, ElSalvador, Honduras, and Nicaragua, 1995-2002 Guatemala ElSalvador Honduras Nicaragua 1995 0.60 0.68 0.62 0.62 1999 0.63 0.70 0.63 0.64 2002 0.65 0.72 0.67 0.67 Note: A higher number denotes an improvementinhuman development. Source: UNDP (2004). -but not among those in the coffee sector 2.15 Inspite of these broad socio-economic gains, the coffee crisis had significant deleterious impacts on those working in the coffee sector - and particularly on small-scale, self-employed farmers and their families. Indeed, households that earned income in the coffee economy experienced setbacks in a number o f areas - with respect to their income and consumption, poverty, child schooling and child nutrition. Inlight o f early concerns about the crisis impacts on "WDI2004. 24Vakis et al. 2004. For a more detailed analysis o fpoverty trends inNicaragua, using a slightly different sample from the same data set, see Nicaragua Poverty Assessment, World Bank, 2003. For further details on the data definitions used here, see Vakis et al, 2004, Volume I1or this report ''ElSalvador Poverty Assessment Strengthening Social Policy, World Bank, 2004. 14 coffee laborers, in the sections that follow the analysis attempts to distinguish between the impacts o f the crisis on households that are linked to the coffee sector through self-employed farming activities and through labor earnings. To illustrate, the typology o f "coffee households" used in the Nicaragua analysis is shown inBox 2.1. Similar distinctions were made inthe other country case studies. 2.16 Inaddition, because coffee production tends to be concentrated in specific regions within each country, analysis o f the effects o f the crisis was generally also carried out across different sub-regions o f countries, based on the intensity o f coffee production (or lack thereof) in each location. Regional, as well as household, definitions were used to examine possible spill-over effects o f the crisis beyond the coffee sector; they were also usedas a gauge o f the robustness o f the basic findings on coffee crisis impacts (Box 2.1). Box 2.1: A Typology of "Coffee Households" and "Coffee Regions" for Nicaragua The analysis of the coffee crisis inNicaraguarelied on several methods to define a household's affiliation with the coffee sector. This is done to help distinguishthe effects of the crisis across self-employedfarming and wage earning households. Regional, as well as household, definitions were used to help assess the robustnessof the basic results. In general, the household-level definition focuses on household employment activities and classifies a householdas either "coffee" or "non coffee" basedon whether any member of the householdworked inthe coffee sector either as awage earner or as a producer. Coffeehouseholdsare then further definedbasedon their status inthe sector as farmer or laborer. This results ina three-pronged definition of ruralhouseholds: non-coffee, if the householdhadno members involved inany coffee activities inagivenyear; coffee-labor, ifthe householdhadmembersinvolvedincoffee-laboractivities ina givenyear; and coffee-farm, ifthe householdhadmembersinvolvedincoffee farming activities in a given year. Inaddition, the use ofpanel datapermittedidentification ofhouseholds that movedinor out of the coffee sector during the crisis period; for example, a householdthat was categorizedas coffeeproducerinthe first year, but classified as a non-coffee household in the second year was defined as having "exited coffee" over the period. Finally, a geographically based index of the intensity of coffee production was constructedto facilitate analysis of the regional dimensions of the crisis impacts, based on the share of land dedicated to coffee cultivation, and using the 2001 Censo Nacional Agropecuario (Agricultural Census). The three-pronged regional definition - low coffee intensity, medium coffee intensity, and high coffee intensity -was linked to the household survey data and used to test for potential spillover effects between the coffee and non- coffee sectors. It also allowed for assessment of the robustness of the results obtainedusing household- level definitions. Source: Vakis et a12004. Impactson HouseholdIncomes, Consumutionand Poverty Small-scale coffee farming householdswere relatively well-off prior to the crisis- 2.17 Income. According to householdsurvey data from the study countries, small-scale coffee farming households were among the wealthier sub-groups in rural areas prior to the crisis. For example, the data from Nicaragua indicates that, in 1998, real per capita income o f coffee farmers was more than 50 percent higher, on average, than that o f non-coffee households (Figure 2.3). By 15 contrast, coffee labor households were the poorest o f the three rural groups. Similarly, in El Salvador (and to some extent in Guatemala and Honduras), coffee farmers were by far the wealthiest rural sub-group during 1995, when world prices were high, while coffee labor households were the poorest sub-group.26 -but their incomesfell most rapidly aspricesfell- 2.18 This income profile changed considerably over the coffee crisis period, however. In Nicaragua, for example, real per capita income increased by 40 percent for non-coffee households living in rural areas between 1998 and 2001, while per capita income of small-scale coffee farm households declined by more than 38 percent over the period (Figure 2.3). Somewhat surprisingly,per capita incomes among coffee labor households stayed largely the same over the period. InEl Salvador, real per capita incomes o f all 3 groups increased between 1995 and 2001; however, relative income changes followed the same pattern as in Nicaragua. Incomes among coffee farm households grew at the slowest pace, followed by that o f coffee labor households. Per capita incomes o f non-coffee households in rural areas grew the fastest over the period.27 Although panel data are not available for Guatemala, in 2000 coffee farm households were found to be the poorest sub-group inrural areas, a finding that is consistent with the findings on impacts inthe other countries.28 Figure 2.3: Real per capita income changes, for coffee and non-coffee households, Rural Nicaragua, 1998-2001 f 6000 n Es OD m -.-Es $ 4000 E .-12000 n 0 L n al 0 Non-coffee Coffee-labor Coffee-farmers , H 1998 m2001 1 I Source: Vakis et a1(2004). -as did their consumption levels 2.19 Consumption. Declines in per capita consumption were more severe among coffee farm than in coffee labor households or non-coffee households. In Nicaragua, for example, real 26 Trigueros and Avalos (2004), Vakis (2004), Coady et al. (2004). 27 Trigueros and Avalos (2004). These differences in income growth, across coffee and non-coffee households are still seen after controllingfor a range of household and regional characteristics that might also have beenassociated with different levels o f income **growth. For details, see Benekede Sanfelid and Shi (2004). Vakis (2004). 16 consumption per capita declined by more than 25 percent among coffee farm households, while per capita consumption among coffee labor households remained ~nchanged.~'This compares with real per capita consumption increases of 15 percent, on average, among non-coffee households. Relative patterns of consumption changes were similar in Honduras, based on data collected for a subset o f 70 o f the most disadvantaged municipalities inrural Honduras duringthe 2000-2002 period.30 Incontrast to the aggregate-level data for Honduras, which indicates real per capita income growth over the period, per capita consumption among households in the survey areas was found to have declined by an average o f 16 percent in real terms from 2000 to 2002. The decline among coffee farm households was significantly larger than that among non-coffee households, however (Table 2.7). Specifically, real per capita consumption among coffee farm households was found to have declined by 20 percent over the period, compared to declines o f 12 percent among non-coffee households over the same period. Table 2.7: Consumptionchanges amongcoffee and non-coffee households, selected municipalitiesin ruralHonduras 2000-2002 Percent change inper capita consumption Total consumption Food consumption only Coffee farmer -20% -16% Live in"coffee region" -16% -13% Non-Coffee -12% -9% Total -16% -13% Source: Coady et a1 (2004) 2.20 Results o f analyses o f different rural regions, according to their intensity o f coffee cultivation, mirror that o f coffee and non-coffee households. Specifically, households living in regions where coffee production i s important to the local economy experienced larger declines in per capita consumption than households that lived outside coffee growing areas. It should be noted that in both Nicaragua and Honduras, where measuring changes in household consumption over time is possible, a significant proportion o f the declines in total household per capita consumption came from declines in householdfood consumption (see, for example, Table 2.7 for Honduras). This likely contributed to the adverse nutritional outcomes associated with the crisis (and discussed further below). Poverty rose amongst coffee farm households 2.21 Poverty. Changes in poverty among the different sub-groups over the coffee crisis periods are consistent with changes in income and consumption. Poverty analysis similarly indicates relatively large negative impacts on the welfare o f (mostly small-scale) coffee farm households. In Nicaragua, whereas poverty headcount among non-coffee sector households declined by more than 15 percent (from 65 to 55 percent), coffee farm households experienced a 10 percent increase in their incidence o f poverty (Figure 2.4). Again, somewhat surprisingly, poverty rates among coffee labor households declined over the period (by 4 percent) - albeit at a slower pace than among non-coffee households. The findings indicate, moreover, that while coffee labor households were, on average, poorer than coffee farm households, the coffee price shock did not affect wage earners in the coffee sector as significantly as it did farmers. Consistent with the findings on income, coffee farm households in Guatemala were found to be 29Vakis et al. (2004). 30For details, see Coady, Olinto, and Caldes(2004), Volume I1ofthis report 17 the poorest among the different categories o f rural households in 2000. While lack o f panel data make it impossible to observe changes in poverty status directly, qualitative evidence from Guatemala suggests that as in the other countries, this reflects a change in the welfare status o f coffee households resulting from the dramatic decline in world coffee prices.31 Figure2.4: Poverty changes inRuralNicaragua, 1998-2001, coffeeand non-coffeehouseholds Source: Vakis et a1(2004). Impacts on ChildNutrition and Education General gains in nutrition were not shared by children in intensivecoffee-growing regions 2.22 Nutrition. The impacts o f the coffee crisis extended beyond its effect on family income, consumption, and poverty. Evidence from the study countries indicates that the crisis also affected family investments in children and, thus, children's human capital outcomes. In Nicaragua, for example, despite the fact that the different measures o f malnutrition (stunting, wasting, and underweight) all showed improvement during the period - national declines o f 35, 11, and 73 percent, respectively -these gains were not shared by children from households living inintensive coffee-growing regions.32 As can be seen inFigures 2.5 and 2.6, inthe Central Rural region, where more than 80 percent of Nicaragua's coffee production is concentrated, the incidence o f underweight children changed very little while chronic malnutrition (stunting) actually increased slightly. Comparing these outcomes to national trends, the evidence suggests 3'Vakis (2004). 32Stunting (height-for-age)reflects chronic malnutrition, which results from years of retardedskeletal growth and is associatedwith poor economic conditions; wasting (weight-for-height) captures deficiencies in fat tissue and indicates food loss from a short-term, emergency situation; and underweight (weight-for-age)combines the previous two measures and reflects total malnutrition.A child (ofusually 5 years or less) is considered"stunted", "wasted" or "underweight" ifhisher correspondinganthropometricmeasure is two or more standarddeviations below the mediano f the internationally recognizedreferencepopulation. Also see Marini and Gragnolati 2002, andChawla2001. 18 that the coffee crisis had an adverse impact on the nutritional status o f children under 5 years, at least inrelative, ifnot in absolute terms. Figure 2.5: Incidenceof UnderweightChildren, Nicaragua 1998-2001 Managua Pacific Pacific Rural Central Central Rural Atlantic Atlantic Urban Urban Urban Rural 1 0 1 9 9 8 m2001 I Sources: Nicaragua LSMS 1998 and 2001. Figure 2.6: Incidenceof Stunting, Nicaragua 1998- 2001 40 Managua Pacific Urban Pacific Rural Central Central Rural Atlantic Atlantic Urban Urban Rural 1 0 1 9 9 8 m2001 I Sources: Nicaragua LSMS 1998 and 2001. 19 2.23 Similar patterns are observed in El Salvador. Between 1998 and 2002, the percentage o f underweight children in rural El Salvador declined by almost 50 percent.33 Nonetheless, regions with greater involvement in the coffee sector tended to experience smaller declines in malnutrition than those with less involvement. For example, administrative Departments with higher intensity o f coffee production, measured as the proportion o flandunder coffee cultivation, tended to experience smaller declines in the incidence o f underweight children between 1998 and 2002 (Figure 2.7). In fact, the Department with the highest coffee production intensity, Ahuachapan, actually experienced an absolute increase inthe percentage o f underweight children that coincided with the timingo f the coffee crisis. Figure2.7: Change in Percentageof Underweight Childrenin Rural ElSalvador, by Departmentand Intensityof Coffee Cultivation, 1998-2002 Percentage changein underweight incidence aAHUACHAPAN lSAN SALVADOR .CUSCATLAN @CABANAS mCHALATENANGO -5 .LA PA2 I L A UNION mMORAZAN -10- W USULUTAN 2.24 While establishing a causal relationship between the coffee crisis and malnutrition i s difficult, given existing data constraints, the descriptive findings appear robust. Similar patterns are found in El Salvador when changes in the incidence o f underweight children are examined across Departments according to the extent o f labor force involvement in the coffee sector.34 In addition, analysis o f the correlation between changes in malnutrition and several other variables, including changes in real per capita income, female literacy rates, and intensity o f coffee production, reveal the strongest statistical association between changes in malnutrition over the period and the level o f intensity of coffee cultivation in the D e ~ a r t m e n t . ~ ~ Lower school enrolment ratesfor coffee households in Nicaragua and,possibly, Guatemala 2.25 Education. With respect to education, all four study countries achieved increases in average enrollment rates during the period, both at the primary and secondary levels. N o t all 33FESAL2003. 34 ''TriguerosandAvalos (2004) hid. 20 groups o f households were able to benefit from the broader advances in schooling, however. Children among coffee households in Nicaragua did not fair as well as those in non-coffee households, for example. In particular, while primary net enrollment rates increased from 78 to 86 percent among children non-coffee households, they declined from 77 to 72 percent among children in households involved inthe coffee sector.36 Similarly, while secondary net enrollment rates almost doubled - to 40 percent - among children in non-coffee households, they remained essentially unchanged among children in coffee-sector households. School enrolment in Guatemala among children in coffee farm households was significantly lower compared to non- coffee ones.37 While it is not possible to attribute such differences solely to the coffee crisis, given data limitations, it is likely that these patterns reflect, at least in part, harmful coping strategies on the part o f coffee households. This issue is discussed in greater detail below. 2.26 In contrast to Nicaragua and (what appears to be the case in) Guatemala, there seems to have been n o systematic, negative impact o f the coffee crisis on primary and secondary schooling rates among children residing in coffee households in El Salvador - although there is some evidence suggesting schooling effects in specific locations. The results from El Salvador suggest that the absence o f obvious negative schooling effects may be attributable, at least in part, to active Government o f El Salvador programs to increase school enrollments in rural areas - for example, EDUCO and Escuelus Suludubles (Healthy Schools) which provides food supplements to students, the timing o f which overlapped with the timing o f the coffee 2.27 To summarize, household survey data from Nicaragua, El Salvador, Honduras, and Guatemala indicates that households involved in coffee activities were not able to benefit fully from what was otherwise a period o f growth and socio-economic progress in the four study countries. In fact, many key socio-economic indicators worsened for coffee sector households duringthe crisis. While accurately quantifying the impact o f the coffee crisis is challenging, the large differences between coffee and non-coffee households across the four countries cast little doubt that the coffee crisis had a strong impact on small-scale coffee farm households and, contrary to expectations, had a relatively smaller effect on coffee labor households. The major impacts o f the coffee crisis on household welfare in the four countries are summarized in Table 2.8. 36Vakis et al. (2004) 37Vakis (2004). 38Trigueros and Avalos 2004. 21 Table 2.8: The Impact of the CoffeeCrisis on CoffeeHouseholds'Well-being: Summary Outcomes Relative to Non-Coffee Households over the Crisis Period Income, Consumption and Poverty Nutrition Education Coffee Labor Households ElSalvador 0 Guatemala* n.a. n.a. n.a. Honduras n.a. n.a. Nicaragua Coffee Farm Households ElSalvador 0 Guatemala* Honduras * No paneldata exist for Guatemala. Comparisonsare basedon cross sectionaldata during 2000. Source: El Salvador: Trigueros et al. (2004), Guatemala:Vakis (2004), Honduras:Coady et a1(2004), Nicaragua:Vakis et al. 2004). HouseholdRisk Management Stratepies Limited ability of households to counteract impacts of crisis 2.28 Household survey evidence indicates that the ability o f coffee households in Central America to protect themselves against the impacts o f the crisis was somewhat limited. This is consistent with the findings o f a broader empirical literature on consumption smoothing, which finds, ingeneral, that households inpoor countries are only partially capable to insurethemselves against income shocks. Recent evidence from several Latin American countries - Nicaragua, Mexico, and Peru - suggests that, on average, households are able to protect between 60 and 75 percent o f their consumption in the face on an income Inthe case of Nicaragua, new empirical tests o f the extent to which negative shocks to household income translate into shocks to consumption (commissioned for this study) indicate that coffee sector households are significantly less able to "self-insure" against shocks than non-coffee households (Table 2.9). As can be seen in the table, a 10 percent decline in household per capita consumption among non- coffee households in rural Nicaragua translates into a 1.4 percent decline in per capita consumption, This compares with a 2.2 percent decline in consumption among coffee labor households and a 2.0 percent decline in consumption among coffee farm households as a result o f a same sized income shock. Households adjust consumptionpatterns-although in different ways 2.29 There are also differences in the way that shocks affect household consumption among those working in the coffee sector: for the coffee labor households income shocks appear more 39See Box 1.1 for a summary o f findings from severalrecent studies of householdconsumptionsmoothinginLatinAmericaand beyond, see Annex 1. 22 likely to affect food consumption, while for coffee farm households, shocks appear more likely to affect non-food expenditures. More specifically, a 10 percent shock in per capita income results in a 4.3 percent decline in food consumption among coffee labor households, but only a 1.2 percent decline in food consumption (that is not statistically significant) among coffee farm households. In contrast, the same sized income shock leads to a 3.4 percent decline in per capita consumption among coffee farm households, but only a 0.8 percent decline (that i s not statistically significant) among coffee labor households. Table 2.9: EstimatedFallinPer CapitaConsumptionfrom a 10 PercentDeclinein HouseholdPer CapitaIncome(in percent),RuralNicaragua Total Food Non-Food All rural Households 1.4** 1.4** 1.3** Non-Coffee Households 1.4** 1.4** 1.3** Coffee Labor Households (1998-2001) 2.2* 4.3* 0.8 Coffee Farm Households (1998-2001) 2.0** 1.2 3.4** Notes: * statistically significant at 10%level; ** statistically significant at 5% level or above. Estimates basedon a fixed effects model of (log) changes in consumption per capita regressed on changes in income per capita andhousehold size changes for the corresponding coffee classification. Bothregressorstreated as exogenous. Municipal-level fixed effects arejointly significant for all the specifications. Source: Vakis et a12004. Households use a variety of coping mechanisms- 2.30 Although they were not completely successful in protecting themselves against the impacts o f the crisis, coffee households did use a variety o f actions inan attempt to mitigate, cope or prevent its worst impacts. These included efforts to diversify their income sources, migration and remittances, adjustments in family labor supply including increased dependence on child labor, sale o f assets, and reliance on informal, community-based insurance networks. Understanding which risk management strategies and mechanisms households use, how they affect family welfare in the short- and long-terms, and how they vary across locations and/or economic circumstances, is critical to designing effective public policies to address shocks. -that reflect both ex-ante and ex-poststrategies 2.3 1 Inexamining how householdmanage risk, it is useful to distinguish between ex-ante and ex-post ~trategies.~'Ex-ante mechanisms address what households (and to some extent, public and private instruments) can do to reduce or prevent the occurrence o f risks and mitigate the impact o f a shock, should an adverse event occur. Examples o f ex-ante mechanisms include the purchase of crop insurance or commodity futures, where such instruments are commercially available, income diversification to spread production risk, and/or migration, which can be used to spatially diversify ones income sources in advance of a shock. In contrast, ex-post risk management mechanisms focus on households' abilities to respond and cope with a shock, once the event has been realized: for example, drawing down savings, selling assets, adjusting family labor supply, including removing children from school to help generate income, and finding alternative or additional income sources. 4oHolzmannandJorgensen(2000). 23 2.32 This section examines, in turn, several o f the most important risk management strategies used by coffee households during the crisis period. Ex-ante Risk Manavement Stratevies Income diversiflcation is an effective way toprevent and mitigate adverse impacts of shocks- 2.33 Agricultural insurance mechanisms are not widely available in Central America, especially for small-scale, family-based farmers. Households thus look for other ways to manage risk ex-ante. In this context, household income diversification in agricultural and non- agricultural activities is perhaps one o f the most effective ways to prevent and mitigate adverse impacts o f shocks as it spreads household production risk and can promote overall income growth and upward mobility. For example, in Nicaragua, non-coffee households, which were more income diversified (as measured by the number o f different agricultural and non-agricultural income sources in the household), were more likely to experience growth in per capita consumption between 1998 and 2001.41 In contrast, coffee sector households (both labor and farm) were less diversified and, as such, did not experience the same level o f consumption growth. In addition, while non-coffee households were typically diversified in both agricultural and non-agricultural activities, coffee households were at best "diversified" only within the agricultural sector. These patterns suggest that access to non-agricultural activities and incomes may have been particularly important for mitigating risk and increasing incomes and consumption. -as are migration and remittances 2.34 In addition, evidence from El Salvador and Nicaragua suggests that migration and remittances have also been important in helping coffee households mitigate the impact o f the coffee crisis. In El Salvador, coffee households that received remittances during the pre-crisis years tended to experience greater income growth during the crisis years than those who did not.42 Although it is not tested inthe data directly, it may be that receipts o f remittances provided liquidity that aided household income diversification before (or during) the crisis period. Similarly, in Nicaragua, coffee households that received remittances in 1998 were more likely to experience growth innon-food consumption between 1998 and 2001 than those who didnot. 43 Ex-post RiskManavement Strategies 2.35 Although Central American households do adopt ex-ante risk management strategies, the evidence suggests that the ability o f coffee households - especially the poorer, labor households - to use ex-ante instruments effectively to address coffee sector risks was rather limited. As such, many families turned to ex-post risk management - and particularly risk coping - mechanisms. Indeed, the use by coffee households o f ex-post strategies to deal with the coffee crisis was widespread, with a diverse set o f ex-post risk management instruments being used. This diversity reflected both a diversity o f household situations and country economic contexts. Increasing non-agricultural incomes- 2.36 One common strategy involves families making household labor and income "portfolio" adjustments. Analysis o f the Nicaragua panel data (1998-2001) shows that coffee households commonly sought to increase their non-agricultural incomes - both wage and self- employed activities - as a way to cope with the impacts o f the crisis. In fact, coffee households 4'Vakis et al. (2004). This is consistentwith Beneke and Gonzalez-Vega (2000) and Beneke de Sanfeliu and Shi (2004) who find positive effects of income diversification on income growth inEl Salvador, 42Trigueros and Avalos (2004). Vakis et al. (2004). 24 that were successful in increasing their non-agricultural incomes over the crisis period fared better than those who were not.44 Similar patterns are seen inEl Salvador as well. In general, the late 1990s-early 2000s was a period in which rural households shifted their labor allocation increasingly toward non-agricultural activities. But coffee labor and coffee farm households made more dramatic shifts than non-coffee households intheir labor allocation and thus incomes, as a result o f the crisis (Table 2.10). In El Salvador, adjustments in coffee households' income portfolios involved not only shifts o f existing labor out o f coffee and into non-agricultural enterprises, but also increases in total hours worked by households. An important element o f this was the widespread adoption of household-based, non-agricultural, self-employed enterprises (often micro-enterprises) over the period.45 Table 2.10: Changesin Labor Allocation TowardNon-AgriculturalActivities in ElSalvador, Coffee andNon-Coffee Households, 1995-2001 (YOof total householdlabor supply) 1995 1997 1999 2001 Coffee farmers 27.6 30.3 42.3 48.8 Coffee laborers 27.0 22.1 35.7 46.4 Non-coffee 36.9 38.9 42.4 45.3 All 34.2 34.7 40.9 45.8 Source: Trigueros and Avalos (2004). -or leaving the coffee sector altogether- 2.37 A particularly dramatic form o fincome portfolio adjustment involves complete exit from an activity - in this case from coffee. Indeed, the data indicate that a significant number o f coffee households, both farmers and laborers, exited the coffee sector duringthis period. Inboth Nicaragua and El Salvador, this "exit option" was higher among coffee laborers. This i s partially explained by the fact that (landless) laborers may be more mobile in the event o f a shock - especially when, as was the case in Central America, there was concurrent economic growth and thus market or self-employment opportunities available elsewhere inthe economy. The fact that laborers in Nicaragua and El Salvador appear to have been more mobile than farmers also highlights the short-run inability o f coffee farmers to exit the coffee sector due to the perennial or multi-year nature o f coffee production and related investments. -but not all households can easily make the transition 2.38 While empirically distinguishing between choosing to exit coffee from beingforced to exit is challenging, the analysis highlights the critical importance o f education, assets, and access to markets in facilitating households' upward mobility and coping capacity.46 Such findings reinforce the importance o f strengthening households' risk management capacity ex-ante so that households are better equipped to adjust and respond in the face o f a shock to minimize the impact of an adverse event (Box 2.2). 44Vakis et al. (2004). Benekede Sanfeliu and Shi (2004); Rodriguez-Meza and Gonzalez-Vega (2004). 46See Vakis et a1(2004) for Nicaragua, and Beneke de Sanfeliu and Shi (2004) for El Salvador. 25 Box 2.2: Exiting the Coffee Sector in Nicaragua InNicaragua, households that exited completely from the coffee sector didbetter overall interms of socio- economic outcomes than those who did not (or could not). Analysis o f households' "decisions" to exit coffee provides a number o f insights. Controlling for income, coffee farm households were less likely than coffee labor households to exit the coffee sector. This is consistent with the idea that farmers are tied to their land by the perennial nature o f coffee investments and productionprocesses. Among households that did exit coffee between 1998 and 2001, the main sources o f income growth were related to increases in non-agricultural income. Interestingly, coffee labor households that exited the coffee sector tended to shift their effort to non-agricultural labor (wage) activities, while coffee farm households tended to shift their labor to non-agricultural enterprises (self-employment). Wealth, greater asset holdings (including land), and more diversified income sources were all positively associated with households' abilities to exit the coffee sector. These findings reinforce the notion that poorer households tend to face greater barriers to economic mobility, all other things being equal. They also highlight the importance o f assets inenabling households to engage innew activities. Similarly, access to credit was found to be associated with a higher probability o f exiting the coffee sector. This suggests that credit (and other forms o f liquidity) can be critical in helping households mitigate the impact for shocks, not only by assisting them in smoothing their consumption in the face o f an income shock, but through enabling them to diversify their economic activities. Several geographic features were also linkedto households' abilities to exit coffee. For example, living closer to Managua andor living outside a coffee region were both positively correlated with the probability o f exiting the coffee sector. Both o f these attributes capture greater access to markets, economic activity, and opportunities outside the coffee sector. Conversely, living in a region that had been affectedby Hurricane Mitch decreased the probability of exiting the coffee sector, suggesting that households facing multiple shocks may face even greater barriers to economic mobility. Source: Vakis et a1(2004). 2.39 While analysis o f the Nicaragua panel data suggests that migration did not play an important role in ex-post coping among coffee households, it does appear to have been a widely used strategy in El Salvador. In fact, Salvadoran coffee households that received remittances after the start of the crisis generally experienced faster income growth (or smaller welfare declines) over the crisis period.47 While it is generally believed that in El Salvador most remittances are used for consumption, rather than investment, the evidence does suggest that the liquidity provided by remittances has enabled many rural Salvadoran households to adjust or diversify their income portfolios, ex-post, in the face of shocks. People with assets can sell them to smooth consumption 2.40 Selling assets was another strategy used by some coffee households to cope with the crisis. In Nicaragua, for example, coffee farmers (who during 1998 were among the wealthier households in terms of asset holdings) used assets, like land or livestock, to cope with the crisis. By distinguishing between poor and non-poor households, the analysis firther suggests that poor 47Benekede Sanfelid and Shi (2004) for El Salvador. Earlier work by Beneke de Sanfelid and Gonzalez-Vega (2000), also find that the presence of international migrants within a family was correlated with higher income growth during a downturn in agricultural production inEl Salvador. For Nicaragua, although formal data analysis didnot identify migrationas an important coping strategy, its importancewas highlighted during various informal interviews inruralNicaragua. 26 coffee farmers were 14 percent less likely to sell land and 9 percent less likely to sell (or consume) cattle than non-poor coffee farm households (Table 2.11). As expected, given their low (long-term) income levels, coffee labor households were the most asset-poor among all households in rural Nicaragua; as a result, their ability to use assets to cope with shocks was relatively limited. In Guatemala, data collected on coping mechanisms also indicates that coffee farmers sold assets to cope with the crisis. Overall, the data indicate the importance o f asset ownership in the context of risk management, and highlight the limited capacity among poorer households to use physical assets as part o f their coping strategy.48 Table 2.11: HouseholdCopingMechanismsin RuralNicaraguaduring the Coffee Crisis (according to pre-crisis standing inthe coffee sector) Householdexperienced: out- Increases in Increases in Migration Decreases in Decreases in child labor adult labor o fmember(s) landowned cattle owned Coffee labor (in 1998) -0.08 0.18 0.01 -0.04 -0.04 Coffee farmer (in 1998) 0.21** 0.02 -0.02 0.25** 0.26** Poor in 1998 0.16** Oslo** -0.01 0.06** 0.02 Coffee labor * Poor (in 1998) 0.10 -0.23** -0.01 -0.13 -0.07 Coffee farmer * Poor (in 1998) -0.06 -0.16* 0.02 -0.14** - 0.09** Affected by Hurricane Mitch -0.02 0.03 0.00 -0.00 -0.04** Notes: * statistically significant at 10% level; ** statistically significant at 5% level or above. Source: Vakis et a12004. Family and social networks cans act as informal insurance 2.41 Informal"insurance" networksare another instrumentthat households use to deal with shocks. For example, informal social networks established through memberships in civic, religious, or neighborhood organizations can provide households with an alternative source o f resources and support in the event o f an adverse shock. In addition, continuing ties with out- migrant household members may ensure support in the form o f additional transfers, gifts, or emergency loans during crises. The empirical analysis shows that the role o f family networks has been important, at least to some extent. In Nicaragua, for example, remittances (which can be thought o f as a proxy for the existence o f a family network) were positively correlated with non- food consumption growth for both coffee labor and labor households. The impact seems to be stronger for coffee labor households implying that other informal coping mechanisms may be more important for the poorer coffee household^.^^ But children can suffer from ex-post coping strategies- 2.42 While many o f the risk management activities undertaken by coffee households may be considered appropriate short- and long-term responses to risk (whether adjusting income portfolios, household labor supply, or drawing down assets), there is evidence that some 48 This finding is similarto results inConning, Olinto andTrigueros (2000) for El Salvador, who find that households owning landor other productiveassetswere betterable to protecttheir incomeduringeconomic downturns. 49 Intheir analysisof the impactsofHurricaneMitch, Klugman, Kruger, andWithers (2004) find that Nicaraguanhouseholdsoften reportparticipatingin informalinsurancenetworks via churchgroups, producers'associations,out-migrants, andthrough emergency loanarrangements with shopkeepers, and so on. At the same time, the authorsdid not find that suchnetworkshada statistically significant impactonhouseholds' ability to smoothconsumptioninthe face of shocks. 27 households engaged in harmful coping strategies. For example, some coffee households increased their use of child labor, withdrawing children from school. In Nicaragua, for example, child labor among coffee households increased by 24 percent between 1998 and 2001.50 Interestingly, the use of child labor as a coping strategy was more prevalent among coffee farm than coffee labor households (Table 2.1 l)?Similar outcomes were found in Honduras. Moreover, in 2000, the incidence o f child labor in Guatemala was significantly higher among coffee farmers than non-coffee households. While the existing data do not permit direct observation of changes in child labor in Guatemala, observed outcomes are similar to those seen inNicaragua and Honduras following the crisis. Moreover, the evidence is consistent with other information o n coping mechanisms among Guatemalan households that indicates that child labor i s among the most important instruments households use to deal with shocks.'* -particularly in the long term ifthey are withdrawnfrom school 2.43 While the use o f children as part o f the household labor pool i s common in developing countries, child labor that involves withdrawing children from school can have particularly adverse consequences in the long-term. What starts as a short-term coping response commonly evolves into the long-term loss o f human capital and productivity, raising the risk o f poverty for the next generation. As such, these patterns o f withdrawing children from school and the use o f child labor as a coping mechanism raise serious issues about the need o f policy interventions to protect children's human capital inthe face o f adverse shocks.53 2.44 Insum, as elsewhere, Central American households use a variety o f actions inan attempt to prevent, mitigate, and cope with risk and shocks. This is evident in the context o f the coffee crisis, in which coffee households used a variety o f approaches to try to deal with the effects o f the coffee price shock, including efforts to diversify their incomes, migration and remittances, adjustments in family labor supply, sale o f assets, and increased dependence on child labor. While the evidence suggests a number o f similarities in household risk management strategies across Central America, it also highlights several differences across countries, the result o f a number o f complex factors, including people's access to alternative economic opportunities. These similarities and differences are summarized in Table 2.12. Vakis et al. (2004). "Table2.11indicatesthatchildlaborisalsoprevalentamongthepoorinruralNicaraguaingeneral. 52 GuatemalaPoverty Assessment, World Bank, 2004. 53 Ofthe countrieswith panel data, only inEl Salvadordoes children's education not appear to have been significantly affected by the crisis. The fact that: (1) children's school enrollments inElSalvador do not seem to have been adversely affected inthe face of the crisis, and(2) inEl Salvador, coffee householdsappear to have relied more an income diversificationstrategies, poseinteresting contrasts to what happenedinthe other Central American countries, andpoint the way to promisingapproachesto dealwith developmentin shock-pronecountries. As noted above, a recentstudy o fthe coffee crisis inEl Salvador (Trigueros andAvalos 2004) suggests that the GovernmentofEl Salvador's active policies to promote rural education, such as the EDUCO program andEscuelus Suludubles(Healthy Schools) which provides food supplementsto students, worked to offset any potential declines in children's school enrollments. Interms o f diversification as a risk management strategy, recent progress in education and in developing rural road infrastructure that better connects people to markets,may haveprovided a stronger basis ex-ante for economic mobility in the face ofthe coffee shock (El SalvadorPoverty Assessment, World Bank 2004). Receipto f remittances may have substituted for credit inthe absenceofdeep ruralfinancial markets. Togetherwith flexible labor markets, these factors together may have enabled relatively greater economic mobility among coffee households, bothsmall-scale farmers and laborers, during the crisis. 28 Table 2.12: CentralAmerican Coffee-households' RiskManagement Strategies: Summary Nicaragua ElSalvador Honduras Guatemala Ex-ante strategies Income diversijkation Agricultural income X Non-agriculturalincome X Migration X X Ex-post strategies Income diversijkation Agricultural income Non-agriculturalincome X X Migration X Exit coffee X X Consumptionpatterns Reduce food consumption X X X Reduce non food expenditure X X Adjustment of labor supply Increase work hours X X Child labor X X X Dissaving and asset sales X X Informal networks X Government ResDonses to the Crisis A lack of counter-cyclicalsafety netprograms 2.45 Although the Governments of Nicaragua, Honduras, Guatemala, and El Salvador all implement social assistance programs for vulnerable and at-risk groups, none o f these countries have in place counter-cyclical safety net programs designed to support those affected by shocks such as the coffee crisis. Duringthe crisis period, most o f the existing safety net programs were designed to address important structuralrisks facing the poor, such as child malnutrition and poor access to basic health and education services, such as school feeding programs, scholarships to promote school attendance, and/or fee waivers for basic education and health care.54 Honduras and Nicaragua also had - and continue to have - conditional cash transfer programs that are targeted to the poor and provide eligible households with cash in return for households sending their children to school and undertaking specific investment's in their children's health and nutrition.55 In addition, all four countries have social investment funds to support the development o f socio-economic infrastructure in local communities. While the Central American 54For recentassessmentsof safetynet programsinNicaragua, Honduras,Guatemala, andElSalvador, see WorldBank 2001b, World Bank 2002a, WorldBank2002b, andWorld Bank 2003a. Marques(2003)provides asynthesis ofthe safety net situationinCentral America. 55InOctober2005, the Government ofEl Salvadorbeganimplementingasimilar program,the RedSolidaria, thatprovides conditional transfersto the extremepoor inthe 100poorestmunicipalitiesinthat country. 29 countries have some incipient capacity to address natural disasters in the wake o f recent shocks like Hurricane Mitch and the 2001 earthquakes, they still do not have systems in place to deal with an externaleconomic shock like the coffee crisis. A post-crisisfocus on debt and employment- 2.46 In this context, specific responses to the coffee crisis did not begin until the crisis had become acute - perhaps due to a presumption among policy makers that declining coffee prices simply reflected "normal" price volatility inthe world coffee market. For whatever reason, inthe early stages o f the world price decline (1998-2000), there was little crisis-related action on the part o f the region's governments. Once governments did begin to focus on the situation - in part a response to pressure from large coffee producers' associations - concerns focused o n two main issues: first, whether the crisis would precipitate a debt crisis in the coffee sector, especially among large coffee producers (estates); and, second, whether it would lead to widespread unemployment among (often poor, landless) labor working on coffee estates. Indeed, inresponse to concerns about a debt crisis in the coffee sector, governments' responses tended to focus disproportionately on programs to alleviate the debt burden o f coffee farmers, particularly that o f large-scale producers (Box 2.3).56 Box 2.3: A Sample of Government Responsesto the Coffee Crisis Nicaragua. The debt situation was o f rimary concern in Nicaragua. Indeed, by 2002, coffee-farmers' debts totaled approximately US$105 million.' As coffee farmers' ability to repay these loans declined, concerns increased about a potential crisis in the country's already stressed financial system. The Government thus intervenedby promoting, coordinating, and providing financing for different debt-restructuring programs. These programs varied according to the type o f debt held by a coffee producer, with the following main restructuring categories being created: (i) debts to solvent commercial banks (US$55 million - 684 cases); (ii) debts to bankrupt commercial banks (US$32 million - 665 cases); (iii) to micro-finance organizations debts (US$6 million - 7,520 cases); and (iv) debts to exporting firms (US$12 million - 2,300 cases). The first two categories targeted mainly medium and large coffee farmers (with farms sizes o f at least 20 manzanas), the third focused on small farmers (5 manzanas or less) while the final category did not distinguish based on farm size. It is important to note that the majority o f the government restructuring schemes (more than 80 percent) - and the bulk o f the government financial support -focused on large coffee farmers. The government o fNicaragua didfield a couple o f safety net programs, butthey were o f much smaller scale than the debt restructuringprograms. For example, in2001, the Government o f Nicaragua financed and implemented a small-scale workfare program, which benefiting 300 coffee workers (representing about 1,000 family members). In addition, in 2002, the Government o f Nicaragua implemented a "Food-for-Work on Coffee Farms" program through the Ministry o f Agriculture (MAGFOR). With a budget o f US$574,336, this program was implemented in 21 coffee municipalities and provided family food rations to 8,212 households; 6,317 of these were small coffee farm owners (6 manzanas or less), while 1,895 were coffee farm workers. Participating households received the food complement inexchange for working on a variety o f activities on coffee farms.58 El Salvador. ElSalvador was the only one o fthe four study countries that hadestablished an Emergency Fund for Coffee in 1992 in response to the 1989 price declines.59The fund, which focused on debt relief, began with 56 These debt reliefand restructuringprograms were launched, at least inpart, in the face o f Central American governments' concern5 about the broader effects on their countries' banking and financial sectors of widespread default by large-scale coffee estates. Inthat sense, it can be argued that debt reliefto predominantly large coffee farmers had a wider economic rationale. Nonetheless, the policy approach adopted by the region's governments entailed important trade-offs - both in terms o f limiting (or delaying) necessary structural changes in the coffee industry as a result o f structural changes in the world coffee market and in terms o f drawing policy attention away from the social and poverty impacts of the crisis, which fell most squarely on small-scale coffee producers. "Nicaraguan Coordination and Strategy Secretariat o f the Presidency (SECEP). 58 Prior to this program, the Government financed a small-scale workfare program benefiting 300 coffee workers (representing about 1,000 family members) in2001. s9 Costa Rica had a similar fund. 30 $45 million obtainedthrough an internationalloan, andwas intendedto be distributedto all farmers, irrespective of their debt or size. By 2000, however, the funddid not have sufficient money to satisfy demand. Infact, even with asecondemergency fund of $80 million establishedin2000, an additional $250millionwas neededto fully address the coffee farm sector debts. By the endof 2001, athird program(Programa de Rescate Cafetalero) was created, which aimedat restructuringsome of the coffee producers' debt, providing supplemental credit to farmers, andeliminating some o fthe lowest quality coffee. Inaddition to the debt relief and restructureprograms, in 2000 a broader programaimed at addressing some of the long-term dimensions of the coffee crisis was established by the Ministry of Agriculture and the Banco Multisectorial de Znversiones (BMI). Financedthrough the banking system, it made available $100 million for the renovationof coffee farms. The programwas orientedtoward financially solvent producers with more than 25 manzanas o f land, situatedin areas 800 meters or higher. The basic idea of this programwas to allow coffee farmers upgrade their farms by improving farm productivity. While this programdid address some of the longer- term dimensions of the coffee crisis, the objective of the programandrelatedtechnicalassistancewas focused on enhancing coffee production practices, as opposedas considering alternative crop production and/or production diversification. Honduras. In early 2000, Honduras created the National Coffee Fund (Fondo Cafetero Nacional) with approximately $20 million available to provide price supports to coffee farmers during the crisis. The program aimed at providing a support o f $6 per quintal of green coffee produced. In addition, in 2002 the Government proposed a $70 million fund aimed to help coffee farmers reschedule bank debt. These efforts were complemented by assistancefrom the World Foodprogram, which agreed in mid-2002 to provide 1,500 tons of food aid families living indrought strickenareas as well as to coffee farmers inneedof food. Guatemala. In 2001, a trust fund was establishedin Guatemalato finance coffee farmers undertakinga number of different activities, including: (i) enhancing agro-processing; (ii) improving marketing; (iii) restructuring; debt and/or (iv) exiting from coffee production. The Trust Fund, administeredby the Bank o f Rural Development (BANRURAL), was established through bonds offered in the domestic market, bearing an interest rate of 8.5 percent. Out o f the $100 million that was initially raised, only $40 million were targeted toward the estimated 50,000 small farmers, while the remaining $60 million went to the 12,000 medium and large-scale farmers. This corresponds to an average of $800 per small farmer, as comparedto an average of $5,000 per medium and large scale farmers. -reduced policymakers 'effectiveness in addressingthe social impacts of the crisis 2.47 Governments' focus on producer debt (on one hand) and on employment losses (on the other) may have been reinforced - perhaps inadvertently - by donor focus and attention to the problem. In April 2002, when the world coffee price was at its depths, the World Bank, along with the Inter-American Development Bank (IDB), and USAID, co-sponsored a regional workshop in Antigua, Guatemala, to assess the situation and identify possible lines o f public action. In preparation for the workshop, the donors produced a joint assessment that identified several areas for action focusing largely on production-side issues - e.g., on improving the competitiveness o f coffee farmers, where feasible, and on promoting farmer diversification out o f coffee, in situations where improvedcompetitiveness seemed unlikely.60 2.48 At about the same time, CEPAL publishedits own assessmentof the economic and social impacts o f the coffee crisis featuring, among other things, some gross estimates o f the expected employment losses associated with the crisis.61 A World Bank team also produced a more in- depth study o f the crisis impacts in follow-up to the regional workshop.62 That study, like the workshop report, focused largely on the production side o f the crisis - although it did provide a preliminary assessment o f the crisis's social impacts, again based on broad estimates o f employment. The World Bank study also provideda note o f caution that it was not clear how the ~~ ''"Floreset Inter-AmericanDevelopmentBank, UnitedStatesAgency for InternationalDevelopment,WorldBank(2002). al (2002). 62 Varangiset a1(2003). 31 aggregate employment estimates translated into changes o f welfare among those involved in the coffee sector. Yet, a detailed assessment o f the welfare impacts o f the coffee crisis was not available at the height o f the coffee crisis. Indeed, each o f these efforts highlighted the need to understand better the social impacts o f the crisis to enable effective government action. 2.49 Insum, the dominant focus on producer-side issues and, inparticular, on debt, combined with a limited understanding o f the household-level impacts o f the crisis, hampered the ability o f policymakers to address the social impacts o f the crisis in a suitable and effective way. Indeed, review o f the programs Central American governments had in place and/or that were implemented inresponse to the coffee crisis indicates that: 0 The governments o f Central America did not have programs in place ex-ante that were poised to respond to the social impacts o f the coffee crisis (or a comparable sector-wide or aggregate economic shock); nor was there in place a particular mechanism for "counter-cyclical'' financing inthe face o f a shock. 0 The majority o f government responses to the coffee crisis focused on "short run" dimensions o f the crisis and, most commonly, on assisting producers with dealing with their debt burdens in the face o f dramatically lower world coffee prices. Relatively few o f the programs (or resources) were dedicated toward enabling farmers or workers to adjust to long-term structural changes inthe world coffee market. 0 The interventions that were implemented were frequently oriented toward larger scale farmers and estates, while only a relatively small share o f the resources going towards smaller farmers who comprise the majority o f the actors inthe coffee sectors inthese four countries. 0 An even smaller share o f resources allocated in response to the coffee crisis seems to have been dedicated toward social protection type interventions to support either laid o f f workers or poor farmers. 2.50 In general, the government responses to the coffee crisis appear to have been designed with little information about the crisis impacts (beyond its impact on producer debt) and with little regard for their incentive compatibility - or incompatibility - with the longer-run structural transformation o f the world coffee market and o f the rural economies in Central America. Indeed, to date, no evaluation has been done regarding the impact or effectiveness o f these debt- restructuring schemes - either from the perspective o f the economics o f coffee or from a social protection perspective. A need to distinguish between the long- and short-run dimensions of the crisis 2.51 Indesigning programs to protect those adversely affected by crises, like the coffee crisis, a more comprehensive approach i s needed to enable policy makers to distinguish between the long- and short-run dimensions o f the problem. While some o f the short-term dimensions may have been addressed, the longer-term issues related to structural changes in the coffee market - efforts that would have involved increasing coffee farmers' competitiveness or supporting agricultural diversification and/or movement out o f the coffee sector - were largely unaddressed. The outcome may, unwittingly, have been to keep some producers and workers attached to the coffee sector when economic circumstances dictate otherwise. In doing so, they may be making these producers and workers even more vulnerable to future coffee-related shocks. 32 Conclusions 2.52 Although the specific impacts differed across the four countries - and across groups within countries - the evidence reveals significant effects o f the crisis on several dimensions o f household and individual welfare, especially among small-scale, coffee farming households. The fact that small-scale coffee farmers appear to have been the worst affected by the crisis, while not surprising in retrospect, defies early expectations that coffee laborers would bear the brunt o f the shock through the employment effects o f the crisis. As importantly, in addition to income effects o f the crisis, the evidence suggests strong pass-through effects from the income shock to investments inchildren's human capital (education and/or nutrition) inthe study countries. 2.53 The evidence also shows that coffee households used a mixture o f ex-ante and ex-post risk management strategies intheir attempt to soften the impacts o f the crisis. While many o f the coping mechanisms households used could be considered appropriate - e g , income diversification, household labor adjustments, exit from coffee, migration and remittances - there i s evidence that households also often employed harmful coping mechanisms, such as withdrawing children from school to contribute to family earning activities, which may have important long-term costs to family productivity and welfare. 2.54 Thus, in spite of their private, informal risk management strategies, households involved inthe coffee sector inCentral America were not able to protect themselves fully from the effects of the sharp decline in the coffee price. This i s consistent with a broader body o f policy research on consumption smoothing and household self-insurance. It is possible that government programs launched in response to the crisis may have helped to soften the blow to affected households. But to the extent they did, the support was indirect via debt relief to mostly large coffee producers. Moreover, even if public responses served to partially offset some o f the adverse impacts, programs appear not to have targeted effectively those who faced the most serious welfare impacts o f the crisis; nor did most government responses focus on ensuring incentives that were compatible with both the short- and long-term dimensions o f the coffee crisis. 33 3. Elementsof a Strategy to Dealwith Shocks 3.1 Effective, country-level strategies to deal with shocks require an integrated approach that includes efforts to: ensure macroeconomic stability and growth, strengthen people's ability to manage risk ex-ante, develop appropriate, well-targeted safety nets, and strengthen data, information, and monitoring systems for crisis management. High returns to a stable macroeconomic environment 3.2 Although it is outside the realm o f what i s generally considered social protection, a critical element of managing shocks effectively is prudent economic management - policies that ensure macroeconomic stability and growth. Terms-of-trade shocks like the coffee crisis may be exogenous, but economic management is not. And it is clear that there are highreturns to sound economic policies. Stable economic environments, coupled with flexible labor markets, can help to soften the blow o f an economic shock, and facilitate households' adjustment to changing economic circumstances. Indeed, in El Salvador and Nicaragua, stable economies and moderate growth during the coffee crisis were critical to coffee laborers' abilities to find alternative sources o f employment and income, which enabled them to mitigate the impacts o f the crisis. Such labor force adjustments would not have been easy -perhaps not even possible - inunstable or stagnant economies. 3.3 From a strict social protection perspective, the evidence from the coffee crisis highlights the importance o f focusing both on strengthening people's ability to manage risk ex-ante and on providing targeted assistance ex-post to those who are unable to protect themselves adequately from a shock. The evidence also indicates that to design effective safety nets, it i s important to understand the precise nature o f the shock and who is affected. This requires the development o f data, information and monitoring systems to assure that programs appropriately address the most pressing challenges generated by the shocks and that ex-post interventions can be adequately targeted. This chapter focuses on these latter three elements o f a strategy to deal with shocks, drawing both on the lessons from the coffee crisis and on other recent experiences with shocks and social protection inthe region. StrenPtheninPPeoule'sEx-ante RiskManapementCauacitv Ex-antestrategiescan beparticularly effective in enablinghouseholds to deal with crises 3.4 The evidence indicates that certain households were better prepared to deal with the coffee crisis than others. In Nicaragua, for example, coffee households that used ex-ante risk management strategies (e.g., had diversified income portfolios, had out-migrant members and received remittances) appear to have done better in mitigating the negative impacts o f the crisis and protecting themselves from poverty than those that solely relied on ex-post coping instruments (e.g., selling assets, taking children out o f school, employing child labor) to deal with the income shock.63 63Vakis et al. 2004. Similarly, Beneke de Saneliu and Shi (2004) find that households that had a greater number o f non-agricultural enterprises and that receivedremittances were less likely to fall into poverty during the coffee crisis period than those that did not. 34 3.5 In addition, some households possessed characteristics that made them more economically mobile and helped them make the necessary economic adjustments in the face o f the shock. This economic mobility i s perhaps most exemplified by the households who exited the coffee sector duringthe crisis periodto diversify their income portfolios. Indeed, households that exited the coffee sector were not only able to distance themselves from the effects o f the crisis, but experienced highgrowth inboth income and consumption, based largely on increases innon- agricultural incomes (Figure 3.1, for the case o f Nicaragua). Figure 3.1: Householdsthat Exitedthe Coffee Sector and Those that Remained: Changesin Per Capita Income and Consumption, RuralNicaragua1998-2001 25.0 15.0 5.0 Percent -5.0 -15.0 -25.0 Consumption Income 1 Source: Vakis et al. (2004), 3.6 But what characteristics or factors influenced certain households' success indealing with the coffee crisis? As noted in Chapter 2, households' abilities to exit from the coffee sector in Nicaragua was positively associated with several key factors, including family education levels, control o f physical assets (including land), access to credit, and access to markets and alternative (non-coffee) market opportunities. The importance o f these factors is corroborated by several other analyses o f poverty, vulnerability and economic mobility in the region. In rural El Salvador, households with higher levels o f education, greater asset holdings, and greater proximity to markets were not only less likely to fall into poverty during the coffee crisis, but they tended to experience faster real income growth over the period.64 Education helps to enhancepeople's economic mobility- 3.7 Similar analysis for Nicaragua also shows that higher levels o f education significantly increased growth o f household per capita consumption, reducing vulnerability to poverty and, indeed, increasing the likelihood o f exiting poverty during the crisis period. Access to alternative (non-coffee) economic opportunities, as proxied by residence outside a coffee growing region, 64Benekede Saneliuand Shi (2004). Similar householdcharacteristics andstrategies were found to be important to mitigatingthe impacts of the El Salvadorearthquakes. See Benekede Sanfeliuand Shi (2004) andEl SalvadorPoverty Assessment, World Bank (2004) for additional details. 35 also translated into a lower probability of experiencing a consumption shock and a greater likelihoodo f exiting from poverty. -as does investment in infrastructure & access to ruralfinancial markets 3.8 The collection of evidence thus suggests that public policies and investments to strengthen households' risk management capacities and facilitate greater economic mobility ex- ante, including investments in: 0 education, 0 socio-economic infrastructure, e.g., roads, transportation and communication infrastructure that reduce transactions costs and improve people's access to markets and opportunities, and 0 deepening o f rural financial markets. are all key components of a strategy to strengthenhouseholds' abilities, ex-ante, to deal with risk, shocks, and economic change.65 While not part of the statistical analyses underpinning this report, qualitative evidence further suggests that improving availability of market information, including on prices, price trends, and changing market circumstances, improves people's capacity to make rationale economic decisions inthe face o f changing economic circumstances. Market-based instruments can be important to managing risk- 3.9 There are also several market-based risk management instruments, including insurance mechanisms that show promise with respect to strengthening producers' ability to mitigate (before the fact) the impacts o f risk and shocks.66 For example, in the case o f internationally traded commodities, including coffee, maize, soybeans, sugar, wheat, and some livestock, market-based risk instruments like futures or options markets already exist and can play an important role as part o f a strategy to help producers manage price risk and shocks. These instrumentshave long been available and used extensively in developed countries, as well as by large producers in many developing countries. A key challenge remains, however, with respect to making such instruments available to small-scale producers in rural areas o f developing countries -like those adversely affected by the coffee crisis inCentral America. -but are not often available to small-scaleproducers 3.10 Indeed, there are still many obstacles associated with small producers' access to such "price risk markets", given that contracts in these markets are generally associated with large underlying volumes and that futures and options trading involves large fixed costs. Several recent efforts have beenmade, however, to provide access o f smaller-scale producers to price risk markets by "aggregating" local demand via some large intermediary, whether through specially designedprograms, through cooperatives or other producer associations. InMexico, for example, the Agricultural Products Options Program (Programa de Coberturu de Precios de Productos Agricolas) offers subsidized dollar-denominated futures options to maize, cotton, sorghum, and soybean farmer^.^' In Guatemala, the National Coffee Association, Asociacion Nacional de Thesefindings are consistent with recent work by the WorldBank on ensuringthat the poor canbenefitfrom regionaltrade benefits like NAFTA or the recentCentralAmericaFree Trade Agreement, CAFTA. This work highlightthe importanceof a "complementary" policy agendathat includesinvestments ineducation, infrastructure, and financialsector strengtheningto ensure that poor andremoterural householdsare able to adjust appropriately to changingterms of trade and relativeprices in the economies. See Lederman,Maloney,and Serven(2003) and World Bank (2005, forthcoming). 66For amore comprehensivereviewof these instruments, see Larsonet a1(2004). For further discussioninthe contextof the Central Americancoffeecrisis, see Varangiset al (2003). 67Varangis andLarson(1996). 36 Cafb, facilitates price-hedged credit for smallholder producers.68 In El Salvador and Nicaragua, pilot efforts by the World Bank are currently being undertaken to identify practical ways to provide access to these risk markets to smallholder farmers in those countries. 69 3.1 1 Several other market-based approaches to managing agricultural producers' price risk also exist - or are being explored - including using inventory management, using long-term "guarantee" contracts, andor linking price insurance to loan agreements. As with futures and options instruments, producer size has tended to be a barrier to access; nonetheless, efforts are underway in different contexts to aggregate demand for these instruments as well, through intermediate organizations and institutions (Box 3.1). Box 3.1: Market-based Approaches to Managing Producer Price Risk While several tools to manage price volatility already exist, small and medium-size agricultural producers in developing countries have tended not to access them, even when they might be appropriate, due to high transaction costs, and limited understanding o f the instrumentsor how they work. Moreover, there tend to be few actors or institutions at present that help to aggregate demand for such instruments among small producers - although cooperatives andor other producer organizations and associations can potentially play this critical role. A recent World Bank study outlines market-based risk instrumentsthat exist - or are being explored - that could potentially help coffee (and other) farmers manage price risks, appropriate linkscanbe facilitated betweenriskmarkets andsmallproducers. These include: , Using inventory management. Small-scale producers (via cooperatives or other producer organizations) may not wish to sell all their product immediately after harvest, but hold back some or all o f their product in storage until price conditions are more favorable. Indeed, by providing greater flexibility in the time o f sale, this type o f price risk management can allow producers (whether of coffee or other crops) to protect the value o f their inventories from unexpected price declines within a limitedtime duration (e.g., during a crop year). . Using guarantee contracts. There are several types o f price-contracting arrangements inplace between farmer organizations and downstream users (e.g., processors, traders, inthe case o f coffee) that provide price protection to farmers. Long-term, guarantee contracts, Fair Trade, i s one such instrument. Fair Trade contracts are particularly attractive as they guarantee a price to farmers that is not only higher then the prevailing market price, but also fixed over time. Indeed, Fair Trade contracts, when they could be negotiated, have been an effective way to provide particularly strong price protection to coffee producers (e.g., some contracts provided for $1.20 - to $1.30 per pound lb. even when market prices declined to $0.50 - $0.60 per lb.). Linking price insurance to loan agreements. In principle, a farmer with price insurance who borrows should be a better credit risk than one who borrows without it. From the perspective o f the lender, a portfolio o f debt that is insured should strengthen the lending institution. From the perspective of the borrower, ongoing access to credit can improve their ability to income smooth over difficult periods. I t should also improve the flow o f credit for farmers who agree to buy price insurance. This arrangement may be usefbl to countries seeking to improve the flow o f credit to coffee (and other agricultural) sectors. The efficacy o fthese instruments depends on anumber o f factors, includingthe critical challenge o f linking small and medium-sized producers to the appropriate instrument. In this context, there is scope for continued experimentation and piloting o f programs that work to aggregate demand for risk management instruments, along with technical assistance to small farmers to strengthen their understanding o f the role of market-based risk management instruments. Monitoring and evaluation o f experience will also be critical to identify and, then, replicate particularly effective program models. Source: Adaptedfrom Varangisel al(2003). 68Larson, Anderson, and Varangis (2004). 69Varangiset a1(2003). 37 While moral hazard and adverseselection can make traditional crop insuranceproblematic- 3.12 Several insurance instruments may also hold promise with respect to helping rural producers manage risk and shocks, ex-ante - particularly yield risks associated with adverse weather, pests, andor diseases. 'O Formal agricultural insurance institutions are generally better suited to address covariate yield risks than are informal ("mutual") insurance mechanisms. Nonetheless, the track record o f traditional crop insurance schemes has been problematic; as such schemes have often been plagued by serious moral hazardand adverse selection problems. Moral hazard arises when farmers are able to take actions that affect insurance payouts; adverse selection arises when those purchasing insurance have better information about risks than the insurance provider. Inboth cases o f moral hazard and adverse selection, the self-selected pool o f insurance participants will have above-average risklpayouts, adversely affecting the viability o f the scheme. 3.13 To address the problems associated with moral hazard and adverse selection, several emergent insurance products, broadly termed "index insurance" have been developed. These insurance products attempt to eliminate moral hazard and adverse selection by using an insurance trigger linked to farm losses that cannot be manipulated by the insured party, such as weather or average product yield over a defined group o f producers. Area-yield crop insurance is a contingent contract that pays out when average yield for a group, usually geographically defined, falls below a specified trigger, while weather-based index insurance pays out based on the occurrence o f a weather event, rather than on actual crop losses. -emerging instruments,such as weather-based or area-yieldinsurance, holdpromise 3.14 Although experience with these types o f insurance products, particularly weather-based insurance, is relatively limited, several recent applications are being watched with interest to see if they might have broader applicability in developing countries. For example, weather indices are currently being used for agricultural insurance in Ontario and Alberta, Canada.'l In Argentina, inlight o f the positive correlation between rainfall and milk yields, a rainfall insurance contract is being used by a milk-producing cooperative. And in Mexico, AGROSEMEX, the agricultural reinsurance company recently used weather derivatives to protect part o f its crop reinsurance portfolio exposed to weather risks. Indeed, there is growing interest intheir promise to help rural producers deal ex-ante with common production risks in developing countries. In this context, in2001, the International Finance Corporation (the private-sector lending arm o f the World Bank Group) joined with private investors to create a US$SO million facility to reinsure weather risks in developing areas. 3.15 In sum, analysis o f the effectiveness o f various risk management instruments to mitigate the impact o f the coffee crisis, enhance upward mobility and reduce vulnerability to poverty suggests that households that use ex-ante rather than ex-post coping mechanisms did better along those welfare dimensions. This suggests that strengthening ex-ante risk management strategies - including long-term investments in education, socio-economic infrastructure, and financial intermediation -that enhance economic mobility -may be among the most powerful instruments for dealing with shocks. Indeed, long-term efforts to address shocks and risk exposure should include consideration o f ex-ante mechanisms as among the key instruments for managing risk. Moreover, although insurance markets or other market-based risk management instruments are not that well developed at the moment in Central America, serious consideration should be given to the development o f such instruments to strengthen the risk management capacity o f 70This sectiononinsuranceinstrumentsdrawsheavily onLarsonet a1(2004). 7'Larsonet a1(2004). 38 households, especially in rural areas, as part o f a comprehensive strategy to deal with risk and shocks.'* This would include further piloting o f promising approaches, such as those aimed at reducing the (transaction) costs associated with small farmers accessing market-based risk management instruments. Developinv Appropriate,Well-TarpetedSafetv Nets 3.16 Even if ex-ante risk management instruments are strengthened and investments are made to enhance people's ability to adapt to changing circumstances, the development of flexible, well- targeted, counter-cyclical safety net instruments is crucial, especially in the event o f covariate shocks like the coffee crisis. Key principles of safety net responses: Recent experience highlights the importance o f several key principals: Developing an effective safety net response to a shock requires pre-shock preparation for a variety o f contingencies, including which institutions and types o f programs can play the most effective roles in the face o f different shocks and differentially affected groups. As noted in Chapter 1, social investment funds played an important role in responding to Hurricane Mitch and the El Salvador earthquakes, but different types o f shocks require different institutional andor programmatic responses. Developing an efficient safety net response also requires effective targeting o f shock- affected groups. Ensuring that the safety net is well-targeted i s important for several reasons. First, in Central America, fiscal resources are generally scarce. So, to be feasible, let alone effective, targeting i s critical to conserving scarce public financial resources. Second, targeting is important because even the most incentive compatible safety net programs may create perverse production and labor force incentives for those who have not been affected by the shock, but who receive program benefits. Developing the capacity for counter-cyclical financing and implementation is also important to ensuring impact, establishing appropriate incentives, and using fiscal resources effectively and efficiently. Developing counter-cyclical mechanisms requires that governments develop the fiscal discipline, as well as reliable financing mechanisms that can take effect inthe face o f a shock. Developing effective counter-cyclical safety nets also requires that governments prepare -priorto ashock-eligibility rules andexitstrategies(such astime-limited eligibility) to ensure that participants exit from the program once the shock-related need has passed. This is again critical to minimize perverse incentive effects (or create aid dependency), as well as to ensure effective use o f scarce government resources. '*While a detailedanalysisof macroeconomicpolicy is beyondthe scope of this report, it is critical to highlight the importance o f soundmacroeconomicpolicies, includinga soundfiscal position, as part of acomprehensivestrategyto for dealing with shocks, ex ante, A stable macroeconomicenvironment is important, for example, to minimizing the likelihoodthat a sector or location-specific shock grows into an economywide-crisis. As noted above, the fact that the Central American countries continued to grow inthe face o f the coffee crisis was critical to coffee laborers' ability to find new economic oppomnities in other sectors when demandfor coffee labor declined. Moreover, soundfiscal management is anecessaryprerequisite to undertaking the types o f counter-cyclical spending on interventions that are often neededto respondappropriately and forcefully to a shock. See de Ferranti et a1(2001) for an in-depth discussionofthe macroeconomicelements o f managing crisis and shocks, as well as for a discussionof the rationale for counter- cyclical financing for safety nets. "Lesson 3: Developing an Appropriate, Well-targeted Safety Net," below, discusses further the importanceofcounter-cyclical safety nets as part of acomprehensivestrategy for dealingwith shocks in Central America. 39 Cross-country experience also indicates that responses to shocks are most effective - and easiest to scale up and down in the face o f changing circumstances - if they build on existing program and institutional capacity prior to the shock. In general, waiting to establish a program and/or response institution until the time o f a shock or crisis is too late, and results in slow and inefficient action. Recent experience from the Latin America region indicates that establishing a new safety net program from scratch can be a multi- year enterprise. 0 Finally, designing flexibility into safety net responses and institutions is important for several reasons. As noted above, to be effective, the program needs to be able to respond to the specific nature o f the shock, its geographic location, and the characteristics of the affected individuals or households. As importantly, with advanced planning, existing programs and institutions can be designed, ex-ante, to be agile in responding to shocks. 3.17 Recent experience with natural disasters inthe region, and the relatively successful public responses, demonstrate the importance o f being able to build on existing programs and institutions when it comes to responding to shocks. Indeed, quick mobilization o f existing country-level institutions - like the Honduras Social Investment Fund (FHIS), the Nicaragua Emergency Social Fund (FISE), and the Social Fund for Local Development (FISDL) in El Salvador - enabled quick and effective responses (Box 3.2). The experiences with Hurricane Mitch and the El Salvador earthquakes also highlight the importance o f building in flexibility to selected implementing agencies' procedures - e.g., establishing contingency procedures and agreeing on them in advance with financing agencies -to ensure appropriate responses to shocks. Indeed, in the wake o f recent natural disasters in the region, many social investment funds now have such "contingency manuals" that enable them to adapt and move quickly in the event o f a Box 3.2: Responding to Natural Disasters: Building on the Strengths of Existing Institutions Honduras after Mitch. The Honduras Social Investment Fund(FHIS) played a pivotal role in the government's response to Mitch, responding to requests from both local and central levels to help rebuild the country's critical infrastructure. Already engaged in developing social infrastructure in local communities across Honduras, and with built-in operational flexibility and a relatively lean structure, the fund was able to respond to the crisis immediately. To confront the challenges in a quick and efficient manner, the FHIS established 11temporary regional offices, and delegated responsibilities and resources to senior staff appointed as regional directors. Its technical experts were in affected areas within hours o f receiving the news o f hurricane's impact to assess damage. Regional offices worked closely with community members and municipal representatives to assess clean-up needs and to repair or replace water and sanitation systems, access roads, bridges, health centers, and schools. Recognizingthe need for quick action, FHIS simplified its subproject cycle, reducing the required number o f steps from 50 to 8, and increased its use o f standardized subprojects and simplified procurement methods. It also established safeguards in each o f the regional offices to ensure accountability and transparency. Within 100 days, FHIS approved 2,100 projects with a total value o f $40 million, with critical support from IDA, the International Development Agency, in securing funds. By the end o f 1999, about 3,400 emergency subprojects had been financed. FHIS's immediate focus on restoring economic activity and basic social services prevented the emergency from aggravating poverty. To this end, the FHIS financed highly labor-intensive projects - labor accounted for up to 70 percent o f some clean-up activities - generating temporary employment in precisely the communities where productive activities had been disrupted. Indeed, FHIS created about 73Warren (2003). 40 100,000 person-months of employment during the first three months-on a par with workfare programs in other countries, such as Argentina's Trabajar. FHIS demonstrated that a social fund can play a vital role as part o f a social safety net in times o f natural disaster. Inmany ways, social funds are ideally suited to meet the combination o f a sudden need for employment and rehabilitationo f community infrastructure. Indeed, some o f the same qualities that FHIS had acquired as the principal government agency responsible for small scale civil projects proved invaluable in an emergency - its unique presence in communities throughout the country; its existing partnerships with municipal governments and other local entities; its track record in financing and overseeing implementation of large numbers o f small- scale projects quickly, effectively, and with operational transparency; and its ability to adapt procedures to new circumstances. El Salvador after the earthquakes. Following the earthquakes, the Government of El Salvador charged the Social Investment Fund for the Local Development (FISDL) with developing an Emergency Program to assist the most adversely affected families and areas, drawing on FISDL's unique experience in working with municipal governments and communities on local development projects. The program focused on two key sets o f activities: Debris removal. Families whose homes were destroyed were provided with incentives to remove debris and place it near the public highways, where municipal governments could then dispose o f it. Specifically, affected families were given vouchers and a set o f tools, made available through municipal governments; additional resources were made available to municipalities to support debris disposal. This program enabled families to clear their living spaces o f debris, facilitated timely support by municipal governments, and helped to catalyze resumption o f local economic activities, as materials and transport services were purchased directly in earthquake-affected areas. Temporary housing reconstruction. Affected families were provided with construction materials for easy assembly o f provisional housing, to ensure that that they had basic shelter and to reduce the risk o f disease after the earthquake. As with debris removal, FISDL channeled needed tools and materials to families through the municipalities. Inthe case o f temporary housing, the FISDL also engaged in direct distribution o f materials and, on the basis o f information prepared by affected municipalities, coordinated delivery o f provisional housing by the Salvadoran Army and several NGOs. The first deliveries o f goods began three days after the earthquake, and in less than three months, over 150,000 aid packages, with vouchers and tools, had been distributed. Materials for building more than 210,000 provisional houses were also moved, with local coordination and participation, contributing to the reconstruction o f significant amounts o f housing and social infrastructure over the following two years. Even with FISDL's experience inworking with municipalities and local communities, implementing the process often involved a steep learning curve - in e.g., small towns that were not accustomed to standard (transparent) accounting systems or had no experience in implementing large community projects. To support the process, therefore, the central government undertook promotional campaigns, clarifying the objectives o f the Program and encouraging municipalities to deliver resources quickly and efficiently under the Emergency Plan. ISources: Warren (2003); FISDL (2004). The effectiveness of specificprograms depends on the nature of the shock and those affected 3.18 But how do governments (and donors) identify the type of safety net program that would be most appropriate in the context o f events like the coffee crisis? Table 3.5 highlights four different safety net - or potential safety net - instruments that have been used recently in Latin American and Caribbean countries, including: (i) workfare programs, (ii) "decoupled" income support programs, (iii)conditional cash transfers, and (iv) fee waivers and related support provided by education and health service providers to prevent adverse impacts on human capital investment. The table summarizes the key features o f each type o f program; including the nature (or type) o f event or shock they are particularly responsive to, the program's main beneficiary group, their administrative and targeting requirementsand their incentive effects. 41 Y 98 8 c-' 8 m c e, R l m ff ci E .I 8e W 2 Y v1 Y h 2 m cd c 0 B 0 3 E3xa 0 x.. 4 0 - The table also notes the link each program has (or not) to the coffee crisis, and where the instrument has been implemented inthe region. Workfare programs can be effective in addressingemployment shocks- 3.19 As can be seen inthe table, workfare programs (e.g., cash-for-work, food-for-work) are well-suited to address employment shocks, whether due to seasonal factors (say, inagricultural) or to specific economic or natural events. They are particularly effective in addressing open unemployment, situations in which the opportunity cost o f participating in and benefiting from the program i s low. From a comparative perspective, administrating workfare programs i s relatively simple, which may account in part for why they have been used across developing countries to address employment shocks.74 If designed properly, workfare programs have the benefit o f being "self-targeting". For example, by setting the benefit payment (at least slightly) below the prevailing market wage, workfare programs tend to attract only those with few or n o alternative opportunities; this simplifies the administrative requirements for targeting and minimizes disincentives to participants seeking regular market employment when it is available. -but may provide little relief to self-employed coffee (or otherperennial crop)farmers 3.20 Although workfare programs or some variant have been work-horses o f employment crises in Latin America (e.g., TrabajarlJefes de Hogares in Argentina) and beyond, it is not clear that they were the most appropriate choice o f safety net inthe context o f the coffee crisis. While it is possible that the programs fielded in Nicaragua had some impact on the welfare o f laid-off coffee laborers there, the size o f the programs were so small that it is unlikely that they had much impact on crisis outcomes in aggregate. More importantly, given the nature o f the coffee crisis, it i s unlikely that workfare programs could do much to mitigate the impacts o n the small-scale coffee farmers who were hardest hit by the crisis. Specifically, since self-employed coffee farmers needed to dedicate at least part o f their labor time to protecting their investments in this perennial crop (inthe expectation that prices would eventually rise again), the opportunity cost o f participating in a workfare program was relatively high. This made it difficult for coffee farmers to participate inworkfare programs. Decoupled income transfers can buffer producersfrom the effects ofprice and income shocks 3.21 Inthis context, some other form of social assistance inthe face of the coffee crisis may have been more appropriate and more effective. Indeed, other types o f safety net programs could potentially be used to address price and income shocks - the most salient features o f the coffee crisis. These include different types o f transfer programs, such as: (i)"decoupled" income support payments, and (ii) conditional cash transfers. So-called "decoupled" income payments are agricultural support programs to farmers that base payments on clearly defined and fixed historical measures rather than linking them to current or future prices or production. This de- linkingo fpayment from prices andproduction helps to minimize the distortionary effects o fthese transfers on economic decision making.75 Decoupled transfers have been used to compensate farmers for the loss o f trade protection and the concomitant terms o f trade "shock". 3.22 This is the case in Mexico, for example, where the PROCAMPO program provides income support to Mexican farmers who cultivated one or more o f the agricultural commodities liberalized under NAFTA. While, in principle, decoupled income supports could be designed as 74 Workfareprogramshaveimplementedwidely indevelopingcountriesoutsideof LatinAmerica andthe Caribbean, includingina numberof countriesinAfrica andAsia. Among the mostwell-knownworkfareprogramsoutside Latin America is the Maharastra EmploymentGuarantee Scheme (for further detailsonworkfare programs, their designandeffectiveness, see Subbarao 2003). "In principle, decoupledincometransfersavoidcreatingthe economic distortionscausedby many traditionalagriculturalsupport programsthroughtheir influenceondomestic prices, input use, technologychoice, or current or future productiondecisions. By not distorting productionand, inturn, trade, properly designeddecoupledtransfer programsalso fall into the ``Green Box categoryof incomesupportprogramsas agreedunderWTO rules (see Baffes and de Gorter2003; CasteAeda2004; Mason2005). 44 an unconditional transfer to those working in a shock-affected sector, inpractice, they have often included restrictions. In Mexico, for example, PROCAMPO is targeted solely to cultivators - farm owners or operators - and not to wage laborers; and while the program does not link the transfer to production o f specific crops, it does incorporate some restrictions on land use. Conditionalcash transfers may helpprotect children's human capital during crisis- 3.23 Conditional cash transfers - such as the Red de Proteccidn Social (RPS) in Nicaragua, the Programa de Asignacidn Familia (PRAF) in Honduras, the Red Solidaria in El Salvador, and PROGRESNOportunidades in Mexico - provide a cash transfer to households conditional on families sending their children to school, visiting health facilities andor undertaking training in preventative health care and nutrition. Traditionally these programs have been used to address structural poverty and to promote family investments in children's human capital development, rather than as responses to shocks. But, there is a growing body o f evidence that conditional cash transfers (CCTs) also can play an important safety net role, protecting household consumption during an income shock and helping to mitigate the negative pass-through effects o f a shock on children's human capital investments. -although transfers can be relatively complex to administer 3.24 Both categories o f transfer programs presented in table 3.5 are more administratively complex than workfare programs, particularly with respect to targeting beneficiaries and verifying their eligibility. Both categories o f programs require the capacity to identify and target beneficiaries via an information-intensive targeting mechanism. Inthe case o f decoupled income supports, effective targeting o f beneficiaries generally requires the existence o f good cadastral andor land use records or some related method o f verifying program eligibility. In the case o f CCTs, programs have tended to use some combination o f geographic and household-level targeting mechanisms; CCTs also require some mechanism for monitoring and enforcing beneficiary households' compliance with the conditions. Even with good targeting capacity, transfer programs risk some incidence o f targeting errors - excluding those affected by the shock as well as including those who were not affected. In contrast to well-designed workfare programs, transfer programs can also generate perverse incentives for work, especially if they are not designed and implemented with clearly defined (and communicated) time limitations andor exit strategies. 3.25 There are several considerations that might guide policymakers' choices among transfer programs - whether within or outside the context o f the coffee crisis. If, for example, temporary income replacement to affected farming households is considered sufficient to address the negative impacts o f the crisis - including the pass-through effects o f the shock o n investments in children's human capital - then a decoupled income support program might be appropriate. The need to have and maintain good cadastral or land-use records might make this approach difficult or impossible to implement in some countries, however, at least inthe ~ h o r t - t e r m .In ~ ~principle, eligibility for decoupled income support programs could be broadened to included shock-affected labor as well as self-employed farm households. In such a case, a different targeting mechanism would be needed - e.g., geographic or household-level targeting or some combination - depending on the geographic dispersion o f the shock. '`One original expectation o f decoupled income support programs, like PROCAMPO, is that they would facilitate households' economic adjustment inthe face o f terms-of-trade changes. This rationale might be appealing inthe context o f a terms-of-trade shock like the coffee crisis, given the long-term secular decline inworld coffee prices. Nonetheless, while a recent impact evaluation o f PROCAMPO shows that the program has had income multiplier effects - especially among medium-to-large scale farmers -there is little evidence that the program has helpedto generate significant structural change inthe economy o f rural Mexico (Sadoulet de Janvry, andDavis 2001). Indeed, Mexicanofficials acknowledge that from the perspective o f promoting economic adjustment inrural areas, PROCAMPO has not met expectations (SAGARPA, verbal communications, October 2004). 45 3.26 Among the most important features o f household income shocks, including the coffee crisis, is that their impacts are often passedon to children in significant ways - through increased school drop-outs, reduced enrollments, increased child labor, and adverse impacts on child nutrition. These impacts are important because if they are not addressed, even ostensibly short- term impacts on children's human capital can result in long-term losses o f human capacity, productivity, and family welfare. For this reason, governments may see as a priority not only to provide temporary income support to shock-affected households, but to ensure the protection (and further development) of children's human capital. In this context, conditional cash transfer programs may be appropriate and more effective than either unconditional or decoupled transfers, as they directly address both salient dimensions o f a shock's impact - the effect on household income/consumption and the effect on investments inchildren's human capital. Temporaryfee waivers may be effective when the shock has a strong geographic dimension 3.27 Finally, for a variety o f reasons, policymakers may prefer to focus solely on the protection o f children's human capital during an income or employment shock, rather than providing direct income support or short-term employment. In such cases, a reasonable alternative might be to provide assistance to families in affected areas in the form o f temporary fee waiversfor school and healthcare services." By targeting transfers to local service providers and allowing them to administer benefits, this approach has simpler administrative requirements than household-based transfers at the central government level - although there may be compensating issues of administrative capacity and accountability at the local level. Fee waivers and other facility-based interventions are most effective when the shock has a strong geographic dimension (as is the case in the coffee crisis). If the impacts of the shock are regionally dispersed, then the benefits o f simpler geographic targeting mechanisms may be (more than) offset by leakage o f program benefits. Fee waivers for school were effective in maintaining children's education inparts o f Indonesiaduringthe East Asian crisis o f the late 1990s. Choice of a safety net should account institutional capacity and constraints 3.28 Beyond the particular strengths and administrative requirementso f a given safety net, the choice and effectiveness o f a given instrument in the event o f a shock will be dictated, at least in part, by the broader institutional environment prevailingin a country at the time o f the shock. For example: e Information baseshargeting capacity. As suggested above, the presence or absence o f cadastral records or other information on which to base household-level targeting will dictate what type o f targeting mechanisms are feasible duringa shock response. Labor regulations. A country's minimum wage laws can affect the ability of a government or donor to design a workfare program that is self-targeting. Specifically, if the labor code prohibits the government from paying workers below the minimumwage, and if the legislated minimumwage is "binding", then it might be impossible to design a workfare program that i s self-targeting in practice, since paying at or above the minimum wage might serve to attract those not affected as well as those affected by the shock. This would serve to eliminate one o f the important administrative and targeting advantages o f the workfare approach, all other factors being equal. Financial sector regulations. Laws that enable - or prohibit - households from using cash transfers as collateral for loans also help to determine the extent to which the 77While not associated with a particular shock, this approach has been adopted recently inEl Salvador to promote greater education and health service access among the poor in that country. 46 transfers can be used to support new investment activities by the household or to which they are used mainly to address consumption shortfalls as a result o f a shock. 3.29 These considerations will be important as the Central American governments examine their options for future responses to shocks. To illustrate: in the context o f the Central America Free Trade Agreement, CAFTA, the region's policymakers will be faced with choices about how best to assist households, most notably farm households, who will face changing terms o f trade for their products as a result o f the Agreement. CAFTA allows for a phasing-out o f trade protections as well as other safeguards, but implementing rapid liberalization coupled with targeted transfers to affected groups may be preferable from an economic perspective (World Bank 2005, forthcoming). Nonetheless, administrative and institutional factors, as well as fiscal issues, are likely to determine which approach is adopted inpractice (Box 3.3). 47 Box 3.3: ManagingCAFTA's Terms-of-TradeEffects Although the Central American economies are already relatively open, due to unilateral trade liberalization efforts undertaken inthe 199Os, a handful o f "sensitive" agricultural commodities (e.g., maize, beans, dairy, and poultry) still have significant levels o f protection. Much o f the current trade protection on these items will be reduced or eliminated as a result of CAFTA. Given current levels o f protection in each country, liberalization o f the sensitive agricultural commodities can be expected to lead to lower domestic prices for these goods in each country - in some cases significantly lower. In this context, if not managed, the liberalization process could, potentially, result in employment and income losses among net producers o f those goods, as least in the short-term, as they make the economic transition to new and more remunerative economic opportunities arising from the Agreement. For this reason, CAFTA was negotiated to include a wide range o f provisions for dealing with the liberalization o f sensitive goods, including grace periods for initiating liberalization, extended phase-out periods for tariffs, interim quotas and/or phase-downs o f TRQs, as well as special safeguard measures to protect local farmers from undue harm. Indeed, the Agreement includes generous timetables for reducing protection on sensitive goods. For some commodities, phase-out periods are as long as 20 years and, at least for a few countries, white maize (an important staple crop producedby the poor) was exempted from liberalizing. These provisions represent important protections for producers o f sensitive crops, giving them an extended timeframe over which to undertake the necessary economic adjustments. While CAFTA has a number o f important safety net features built into it, the negotiated approach is not without its economic costs. Although phasing o f reforms provides producers an extended period to make the necessary economic adjustments, it also deprives consumers for that same extended time period o f the benefits associated with lower prices for important agricultural staples. A new World Bank study on the impacts o f the Agreement (World Bank 2005, forthcoming) suggests that under the right circumstances, the governments o f Central America could even do better. A more economically efficient approach would involve rapid liberalization o f the sensitive commodities, coupled with transfers targeted to adversely affected groups. In principle, this approach is more efficient economically than phased liberalization, as consumers do not have to wait up to 20 years to reap the full benefits o f lower prices. Coupling well- targeted transfer programs with quick liberalization would enhance households' welfare in the short-term on the consumption side while providing producers with a reasonable period of support to make the economic transition. Nonetheless, to be effective, the targeted transfer approach would require certain conditions to hold in- country. For example, fielding a program o f targeted transfers would require the government's to make available sufficient fiscal resources to finance the transfers. The World Bank study finds that, should Central American countries choose to pursue rapid liberalization coupled with targeted transfers, two categories o f transfer programs hold particular promise: (i)"decoupled" income supports; and (ii) conditional cash transfers (CCTs). The choice o f the most appropriate transfer programwould depend on a number o f country-specific factors, however, including the country's institutional capacity vis-a-vis safety nets (Nicaragua, Honduras and El Salvador all have operating CCT programs), the country's administrative capacity to target and verify program eligibility, and the country's desire to build on existing programs and institutional structures or to use the opportunity to create new ones. Whilenot originallydesignedas such, evidencesuggests that CCTscanplay a safety net role 3.30 While not designed as a traditional safety net program in the sense o f reacting or adjusting to crises or shocks, CCTs have received a fair bit of attention recently with respect to whether or not they can play a safety net as well as a poverty reduction function. Indeed, two CCT programs in Central America - the Red de Proteccidn Social in Nicaragua and the Programa de Asignacidn Familiar in Honduras appear to have performed like safety nets in the face of the coffee crisis. Recent impact evaluations of these two programs indicate that these CCTs have had a significant protective effect on coffee households enrolled in the program- on a 48 number o f socio-economic indicators - relative to coffee households that didnot participate inthe p r ~ g r a m . ~This raises a question about whether such interventions could be developed more ' systematically to address risk exposure and shocks - strengthening both ex-ante and ex-post risk management instruments available to poor households. 3.31 In Nicaragua, the Red de Proteccidn Social (RPS) had been established to target poor households and help them by providing income transfers for long-run investments in children's human capital and health. The incidental participation by poor coffee households in the program enabled the evaluation o f the impact o f the program on mitigating the adverse effects o f the coffee crisis. 3.32 The evaluation o f the RPS inNicaragua did indeed show that the RPS has performed like a crisis safety net, with larger estimated program effects for those who were more affected by the coffee crisis.79 It protected coffee households against declines in per capita expenditures in the face o f the crisis. It also served to protect household investments in child human capital, as measured by school enrollment rates, decreased child labor, and improved anthropometric outcomes. And while there was no evidence that participation in the program significantly depressed household labor supply relative to before the program, the RPS seems to have muted additional labor effort for beneficiaries in coffee growing areas (relative to similar coffee sector households outside the program). 3.33 The evidence i s more mixed, however, as to whether RPS enabled households to reallocate their resources in a fashion consistent with the historical downward trends in coffee prices. Beneficiaries, who worked in the coffee industry, as laborers, were more likely to exit the coffee industry, whereas those who participated as producers were less likely to have exited. At the same time, beneficiaries residing in coffee growing areas, while working fewer total hours in agriculture, increased the role o f agriculture in their portfolio-to the detriment o f non- agricultural activities. Overall, the RPS appears to have played an important part in the risk coping o f households, a conclusion also supported by separate analysis o f individual household idiosyncratic shocks. 3.34 In Honduras, a similar analysis was undertaken to evaluate how a similarly styled program, the Programa de Asignacion Familiar (PRAF), has helped poor coffee households to protect their welfare in the face o f the coffee shock. The analysis indicates that the cash transfers given out by PRAF, which are conditional on school enrollment, has significantly affected the labor allocation decision o f credit-constrained coffee farmers.' OSpecifically, the additional liquidity provided by the transfers seems to have allowed families to maintain their children in school, while increasing the time dedicated by adults in the household to tending and maintaining coffee trees. That is, the fact that the transfers have been conditioned o n investments in child education seems to have ensured that higher on-farm investments and labor activities have not come at the expense o f investments in children's human capital. 3.35 Similar findings are found elsewhere in Latin America. A recent study of the PROGRESA conditional cash transfer program in Mexico evaluates the role o f the Program in mitigating the impact o f shocks (Sadoulet et al. 2004). The study highlights several interesting and relevant findings. First, the study finds that children who are taken out o f school (partly due to shocks) are less likely to return to school subsequently, reinforcing the notion that there are 78See Maluccio (2004) andCoady et a1(2004), presented in Volume I1o f this report. 79Maluccio (2004). Coady et al. (2004). 49 long-term consequences from short-term decisions. PROGRESA appears to mitigate these long- term effects, however. Second, a number o f shocks - including unemployment or illness o f the household head or younger children, drought, a natural disaster in the community, and loss of land, harvest, or animals - all have strong effects on children's schooling attainment, indicating that that children are employed by their families as risk-coping instruments. While this generates benefits for the household in terms o f short-run consumption smoothing, such a coping strategy also risks long-term losses o f human capital for children, given the reduced probability of children returning to school. Again, the impact evaluation shows that conditional transfers from PROGRESA compensate for these shocks, protecting children's schooling from a range of adverse events. Finally, the shocks that households report facing also seem to induce children to work; this is particularly true for girls and for children o f farm workers when their parents are affected by unemployment. The evaluation o f PROGRESA's impact, however, suggests that transfers conditional on children's school attendance serve to deter the use o f child labor as a risk coping strategy. 3.36 While none o f these programs was designed as a "risk management" or "emergency" intervention, such results show that they can play an important safety net role, protecting child human capital investments from a range of idiosyncratic and covariate shocks. This raises a question about how, it practice, it might be possible inthe future to adapt these or similar existing policy instruments to address shocks more deliberately. Possible ways to address this issue would include incorporating risk exposure and/or shock incidence criteria in the design o f such programs' eligibility rules, or by building into the programs flexibility, in terms o f scaling up or down such interventions, to address large covariate (or idiosyncratic) shocks. This area requires further work and piloting, but based on the findings presented here appears to be a potentially worthwhile direction for and use o f such programs (Box 3.4). Box 3.4: Redesigningthe Nicaraguan Red de Proteccidn Social to Address Shocks Nicaragua i s a country where natural disastersare frequent and the impact of Hurricane Mitch is still visible in some parts of the country. Inaddition, the coffee crisis and the findings inthis report also indicate that adding flexibility to already existing interventions may be a useful and effective approach to allow poor households to smooth consumption and mitigate adverse impacts of shocks like decreases in school attainment or increases in child labor. Still, issues like how flexible these interventions should be in the case of an emergency or where they fit or compare with other potentially appropriate instruments are generally unexplored. As such, learning from these extensions can be useful in designing policies targeted at the extremepoor andvulnerable to various risks. The Ministry of MIFAMILIA in Nicaragua i s currently designing a number o f pilot programs that use conditional cash transfers (CCT) to reach extreme poor households that are vulnerable to various risks (natural, economic). This pilot builds on and expands a successful CCT program (the Redde Proteccidn Social), which has been operating for about 4 years. The basic structure of the pilot has 4 phases: (i) construct an information database with a set o f aledvulnerability indicators that can easily monitor and verify various covariate and idiosyncratic shocks (both natural and economic); (ii)select extreme poor households affected by large shocks; and (iii)implement the CCT for a limited length of time. The pilot will incorporate a number of variations in terms of the conditionality rules, the type of shocks and the length ofthe transfer. A formal impact evaluation will also be conducted. This pilot i s also part of a larger effort of the Ministry to reorganize its various programs using a "Social Risk Management" model with the aim of exploring both ex-ante and ex-post instruments using an umbrella of interlinkedand coordinated interventions. 50 Strenvtheninv Data. Information, and Monitorinp Svstems Policy responses were hindered by weak understanding of the crisis and its impacts- 3.37 Policymakers' abilities to respond to the crisis in an effective and timely manner were hindered, in part, by lack o f an adequate understanding o f the nature o f the coffee crisis and its impacts. Indeed, there were considerable differences between policymakers' initial expectations about the impacts o f the crisis and its ultimate effects. This divergence underscores the importance o f strengthening data, information, and monitoring systems for crisis management. In the case Nicaragua, for example, workfare programs implemented by the government were probably less effective than hoped, given that such programs are better-suited to assist newly unemployed laborers than to support farmers who faced a dramatic decline in the value o f their product. Indeed, the evidence suggests in retrospect that programs oriented towards mitigating the price and income effects o f the crisis - as well as the effects on children's human capital - would have been more effective. Better, timelier, information on the nature o f the crisis impacts and the worst-affected groups, may have helped to ensure more effective programmatic responses; better monitoring o f the crisis and program impacts may also have facilitated mid- course corrections to make the governments' responses more effective. -highlighting the need to improve capacity togather information locally 3.38 The ability to gather information locally and to identify emergency needs as perceived by local populations is an important element in strengthening policymakers' ability to response quickly and appropriately in the face o f a crisis. In this respect, efforts to strengthen local capacities for information gathering, planning and implementation - and to strengthen local institutions by building linkages between central and local governments and between local governments and communities - can all have important impacts on countries' capacities to respond to crises. These features o f capacity and institution-building were important to the abilities o f the region's social funds to respond effectively to recent natural disasters, such as Hurricane Mitch or the 2001 earthquakes inEl Salvador.8' No one-size-fits-all in designing crisis responses 3.39 While there are a number o f important similarities in the patterns o f impacts o f the coffee crisis across countries, there are also a number o f important differences between the countries that suggest that one size does not fit all in designing effective responses to shocks. 3.40 For example, the household survey data indicate that approximately 50 percent o f the Nicaraguan coffee households are attached to the coffee sector through wage employment, while the other 50 percent are attached to the sector through direct coffee production. Contrast this with El Salvador, where over 80 percent of the households involved in the coffee sector are involved via labor. This may help, in part, to explain the weaker impacts o f the coffee crisis in El Salvador, as well as the apparently greater economic mobility among coffee sector households during the crisis. While CCTs or other types o f transfers may be appropriate to support self- employed coffee farmers in Nicaragua, more traditional workfare type programs might be more effective to reach the high concentration o f workers in El Salvador, should the employment impacts have been deemed severe enough to warrant public intervention there. As highlightedinthis chapter, social funds haveplayedimportant roles inrespondingto recentnaturaldisasters inCentral America. While it is not clear that countrieswould want to establishsocial funds simply as a response to shocks, ifthey have a productive role to play inincreasingaccess amongpoor communities to basic infrastructure, then they canalso play avaluable role in dealing with shocks, whether ingeneratinglocal employment or inmobilizing communities to identify their needs. 51 3.41 As noted above, the choice o f the most cost-effective form o f the transfer or safety net may depend, at least in part, on the types o f programs that are already operating in the country and how much institutional capacity they have. In Nicaragua, for example, building a shock response on the foundation provided by the RPS may make a lot o f sense. Given its functioning institutional infrastructure and a demonstrated program impact, it may be worthwhile to extend and adapt the institution to deal with crises, rather than to create a new crisis response apparatus. Incontrast, inElSalvador, where its safety net program is currently under development, there are more degrees o f freedom to develop an institutional response "from the ground up"; in that case, there may be other institutional arrangements and programmatic designs that would greater impact and cost effectiveness. 3.42 In sum, even in the case of a similar shock across countries - differences incountries' socio-economic structure - as well as institutional and fiscal capacity, the existing set o f social protection programs and institutions, and information base - will all influence the choice o f the most appropriate and effective instrument to use inthe event o f a shock. 52 4. Implicationsfor Policy CentralAmerica is vulnerable to a range of shocks 4.1 Central America is a shock-prone region. Since the mid-l990s, the countries o f Central America have experienced a number o f natural shocks, including Hurricane Mitch (1998), earthquakes (El Salvador, 2001), and a series o f seasonal droughts and floods (often associated with ElNiiio and L aNiiia); as small open economies, the Central America countries are also open to a variety o f economic shocks, whether in the form o f external terms-of-trade shocks (e.g., declining coffee prices, rising oil prices) or more generalized slowdowns in the U.S. and global economies. 4.2 Among the most important economic shocks to hit Central America recently was the "coffee crisis" - a sharp decline inworld coffee prices between 1997 and 2001 that had important impacts o n the region's economies and on the families who depend on coffee-sector income. While world coffee prices have since rebounded somewhat, the dramatic price decline associated with the crisis was a cause o f great concern throughout the Central American region. Several early assessments o f the coffee crisis raised serious concerns about the poverty and social impacts of the crisis - especially relatedto the apparently large declines in the demand for permanent and seasonal labor in the coffee sector. Yet, these early assessments were based largely o n aggregate firm and industry-level data -providing little informationon the household-level impacts o f the crisis or guidance on the types o f government responses that would be most effective in protecting individuals and families from the worst effects of the crisis. 4.3 This report was undertaken in response to requests from several Central American governments to better understand the impacts o f the coffee crisis on people's welfare - and the lessons o f the crisis for public responses to shocks. To that end, much o f the report has focused on the case study o f the coffee crisis in four countries - Nicaragua, El Salvador, Honduras, and Guatemala. Indeed, the study has brought together a rich set o f new empirical evidence o n the coffee crisis - analyses that have enabled a more detailed understanding o f the crisis impacts than had been available previously. Based on the findings on the crisis impacts - and related evidence from other shocks in the region - this report has outlined a broader set o f lessons for Central America regarding shocks and social protection, so that the region's policymakers and their development partners are better prepared to deal with a range o f shocks inthe future. 4.4 This chapter summarizes the key insights emerging from analysis o f the coffee crisis and outlines the emerging lessons for crafting effective responses to future shocks. Key Insightsfromthe Case of the Coffee Crisis 4.5 Impacts of the Coffee Crisis. Detailed new analysis o f household survey data from Nicaragua, El Salvador, Honduras, and Guatemala paints a consistent picture o f the impacts o f the crisis. Although there are some country-specific differences, the coffee crisis has had significant negative impacts on small coffee farm households throughout Central America - on their per capita income and consumption, o n poverty, and on their children's education and nutritional status. Small-scalefarmers were the hardest hit 4.6 While the broad areas o f impact are not surprising, there are some important differences between the actual effects o f the crisis and what observers had initially expected. Early assessments o f the crisis suggested that the most devastating effects o f the crisis were occurring 53 through the large-scale loss o f employment by (often landless) laborers in the coffee sector. In fact, although coffee labor households are consistently among the poorest groups in rural areas, small-scale, self-employed coffee farmers were unambiguously the hardest hit by crisis. Moreover, these farmers were affected mostly through the effect o f the price decline, rather than through the loss o f employment. 4.7 Coffee sector workers did experience some increase in unemployment as a result o f the crisis. Nonetheless, in the face o f continued economic growth in Central America and expanding economic opportunities outside the coffee sector, labor households seem to have been able to shield themselves from many o f the negative effects o f the crisis - by shifting into non- agricultural activities andincreasing the labor effort and earnings by other household members. 4.8 Household Risk Management Strategies. Coffee sector households in all four countries used a variety o f ex-ante and ex-post strategies in efforts to prevent, mitigate, or cope with the effects o f the crisis. This involved efforts to diversify household income sources (including exiting the coffee sector all together), migration and remittances, increasing family labor supply, sale o f household assets, and reliance on informal "insurance" (or "mutual help") networks. Ex-ante crisis response strategies are generally more effective than ex-post- 4.9 Households that were better prepared for shocks ex-ante - e.g., with higher levels o f education, more diversified incomes, or existing sources o f remittance income - generally did better inprotecting themselves from the effects of the crisis than those who simply coped with the problem, ex-post. Nonetheless, coffee sector households were generally only partially able to "insure" themselves against the effects o f the crisis. This suggests the need for efforts to strengthen andor supplement households' own efforts to deal with risk. -particularly as some ex-post strategies can have adverse long-term impacts 4.10 Moreover, while several o f the strategies households used can be considered "appropriate" ways to manage economic risk, others have potentially harmful long-term effects on household productivity and welfare. In Nicaragua and Honduras, for example, coffee households commonly withdrew their children from school and put them to work, as part o f their efforts to maintain and protect family incomes. These negative effects on schooling - coupled with the adverse nutritional effects o f the crisis - raise serious concerns about the long-term effects on children's human capital, includingpotentially irreversible impacts on children's future economic productivity and welfare. Such effects on children's human capital increase the risk o f poverty being passed on from one generation to another. 4.11 Regional Government Responses. In light o f significant welfare impacts o f the crisis - and the fact that even a short-term shock could have negative long-term effects on household welfare and productivity - there was a strong rationale for government interventions to strengthen households' own risk management efforts and to mitigate the impacts o f the crisis. The initialfocus on debt and employment- 4.12 Against this background, initial government responses focused largely on debt restructuring for (mostly large and medium sized) coffee producers. Although some governments in the region subsequently implemented small-scale "workfare" (e.g., food-for-work, cash-for- work) programs to address some o f the employment effects o f the crisis, only a relatively small percentage o f the crisis-response resources went toward safety net type interventions. This raises concerns about the impact of regional governments' actions and potentially important gaps in their responses. 54 -may have impeded necessay adjustmentsin the coffee sector- 4.13 First, regional governments' focus on debt restructuring in the face o f the large structural changes occurring in the world coffee market, raises questions about whether the incentives created by such programs were appropriate. While the focus on debt may have reflected, in part, a concern about the broader economic impacts o f a crisis in the coffee sector, it is possible that debt relief also served to delay the exit o f coffee producers who will not be competitive in the long-run, in the face o f the changing nature o f the world market. This may just defer the need for further sector adjustments -and set the stage for additional "crises" -inthe future. -and led to inadequateprotection of the hardest hit households 4.14 Second, review o f the main government responses to the crisis raises questions about the adequacy o f the response to the human dimensions o f the crisis. On one hand, the analysis suggests that the government responses were regressive; the vast majority o f resources benefited large- and medium-scale producers, as opposed to small-scale farmers or laborers. On the other hand, the workfare programs that were implemented appear to have had little impact on small- scale, self-employed coffee farmers who have experienced the most serious poverty and welfare impacts o f the crisis. Together, the analysis suggests considerable scope for strengthening government support to adversely affected groups inthe face o f a shock. Implicationsfor Policv-Elementsof a Stratepy to Dealwith Shocks 4.15 The long- and short-run dimensions o f the coffee crisis suggest the need for a policy approach - both within and outside the context o f the coffee crisis - that: (i) facilitates people's economic mobility, ex-ante, and (ii) strengthens households' abilities to manage risk and shocks, both ex-ante and ex-post. In this context, the evidence argues for an integrated, multi-level approach that includes efforts to: 0 Ensure macroeconomic stability and growth 0 Broaden and strengthen people's ability to manage risk ex-ante through o investments inpeople's long-run economic mobility, and o development o f more effective insurance and market-based risk management mechanisms 0 Develop appropriate, well-targeted safety nets, and 0 Strengthen data, information, and monitoring systems Ensure macroeconomic stability and growth 4.16 The evidence suggests there are high returns to a prudent macroeconomic policy that facilitate growth. Stable macroeconomic environments, coupled with flexible labor markets, can help to soften the blow o f an economic shock, and facilitate households' adjustment to changing circumstances. Indeed, these factors along with moderate growth environments inNicaragua and El Salvador appear to have enabled coffee laborers to find alternative sources o f income - something that would not have been possible inunstable or stagnant economies. Broaden and strengthen ex-ante risk management instruments- 4.17 Households that engaged in risk management investments and strategies, ex-ante, did better in mitigating the impacts o f the coffee crisis than those that did not. Another key element o f a country's strategy to strengthen its risk management capacity is thus to strengthen people's ability to manage risk ex-ante. This includes both investments to enhance people's economic 55 mobility and the development o f more effective insurance - and other market-based risk management - mechanisms. Indeed such investments, along with suitable ex-ante risk management instruments, can also help to reduce the pressure to introduce measures, such as debt relief, that can create perverse incentives ex-post. -by enhancing economicmobility- Key measures to strengthen people 's long-run economic mobility include: o Investments in education to promote greater mobility, adaptability to changing economic circumstances o Deepening o f rural Jinancial markets to break liquidity constraints to enterprise development o Investments inroad and transportation infrastructure to increase people's access to labor and product markets, reduce transactions costs, and raise enterprise profitability o Improving information, for example, on prices, trends, and changing market circumstances improves the environment for investments and economic planning among individuals, households, and firms. 4.18 It is useful to note these measures coincide quite closely with the core elements o f the "Complementary Agenda" set forth in recent World Bank work on strengthening countries' abilities to benefit from the Central America Free Trade Agreement (CAFTA). Indeed, they operate through many o f the same channels, enabling people to adapt better to changing economic circumstances. -and strengthening insurance and other market-based mechanisms Insurance and other market-based risk management mechanisms o In the case of shocks to internationally traded commodities, such as coffee, maize, soybeans, sugar, wheat, and some livestock, market-based risk instruments likefutures or options markets can play an important role an integrated risk management tool-kit. While these instruments have long been available to large-scale producers, pilot efforts by the World Bank are currently underway in El Salvador and Nicaragua to explore practical ways provide access to these risk markets to smallholder farmers. o Emerging "index insurance" instruments, such as area-yield and/or weather-based insurance, may also hold promise for helping farmers deal with natural or economic shocks in agriculture (while reducing the problems o f "moral hazard" and "adverse selection" associated with traditional crop insurance). Ongoing applications o f weather- based insurance in Canada, Mexico, and Argentina may provide valuable lessons for future use inCentral America. Develop appropriate, well-targeted safety nets- 4.19 Even with strong ex-ante investments and risk management instruments, households will often still require additional support following a shock. Shocks take a myriad o f forms, whether natural shocks, such as hurricanes, earthquakes, droughts or floods, or economic shocks, such as the coffee crisis or other terms o f trade shocks. As such, the development o f a flexible safety net i s a critical element o f a country's response to shocks. Several key principles stand out: Developing an effective safety net response to shocks requires pre-shock preparedness for various contingencies, including which institutions and types o f programs might play the most effective roles inthe face o f different shocks. 56 Ensuring that a program is well-targeted to affected groups is important to ensuring program impact, particularly in Central America where governments face tight fiscal constraints. In addition to increasing the efficiency o f interventions, targeting can also serve to minimize the risk o fperverse work incentives to non shock-affected populations. Developing the capacity for counter-cyclical financing and implementation is also important to ensuring impacts, establishing appropriate incentives, and using fiscal resources effectively and efficiently. Developing counter-cyclical mechanisms requires governments develop the fiscal discipline and a reliable financing mechanismthat can take affect inthe face o f a shock. It also requires that governments consider prior to a shock eligibility rules and exit strategies (such as time-limited eligibility) to ensure that participants exit from the program once the shock-related needhas passed. Cross-country experience also suggest that responses to shocks are most effective - and easiest to scale up and down in the face o f changing circumstances - if they build on existing program and institutional capacityprior to the shock. Finally, designing flexibility into a country's response mechanisms i s paramount. This includes developing contingency procedures into existing social assistance institutions and programs, procedures that will enable agencies to respond quickly and appropriately once the specific nature o f the shock and its impacts have been determined. -that reflect the nature of the shock and existing institutionsand capacity 4.20 Several different types o f programs that have been implemented recently in Latin America have - or could be - used to address covariate economic shocks like the coffee crisis. Recent experience indicates, however, that different programs have different strengths, depending on the nature o f the shock; different programs also different administrative and implementation challenges. For example: Worvare programs (e.g., food-for-work, cash-for-work) are particularly well-suited to address employment shocks, circumstances in which the opportunity cost o f participation is low among affected groups. Workfare programs are relatively simple to administer and, if well-designed, they can be self-targeting. The ability to make workfare programs self- targeting depends on whether the legal and regulatory environment permits authorities to set the program wage below the prevailing market wage, so as to attract only those who are truly inneed. Decoupled income payments, in which payments are de-linked from current (or future) prices or production to minimize distortions in economic decision-making, are better suited to addressing price and income shocks faced by rural producers - say, from trade liberalization or an external terms-of-trade shock. Decoupled transfers are more complex to administer than workfare programs; this is due in large part to the requirements for identifying affected producers and verifying their eligibility which, historically, have relied o n the existence o f good cadastral and/or land-use records. In principle, decoupled transfers could be extended to both affected producers and laborer; however, this would require other mechanisms to identify beneficiaries, whether through geographic or household-level targeting, or some combination. Conditional cash transfers (CCTs) provide cash payments to families in return for their making specified investments in children's human capital. While traditionally CCTs have 57 been designed to reduce poverty through promoting human capital development among the poor, recent evidence indicates that they are effective inprotecting families against price and income shocks, as well as the negative pass-through effects on children's education, health, and nutrition. CCTs are relatively complex to administer, however, requiring both the capacity to target affected groups and to monitor compliance with program conditions. Incontrast to well-designed workfare programs, both decoupled transfers and CCTs also can create perverse labor force incentives, especially if there are no clearly defined time limitations and/or exit strategies. Temporavy fee waiversfor school and healthcare services specifically address the possible negative impacts o f shocks on people's human capital, rather than focusing on providing direct income support or short-term employment. By administering benefits through local service providers, this approach uses a relatively simple targeting mechanism, although its effectiveness depends on local institutional capacity and accountability. Fee waivers and other facility-based interventions are most effective when the shock has a strong geographic dimension; ifimpacts are regionally dispersed, the benefits o f targeting facilities can be offset by considerable leakage o fprogrambenefits. Strengthen data, information, and monitoring systems 4.2 1 The differences between observers' initial expectations about the coffee crisis impacts and its actual impacts highlight the importance o f establishing data and surveillance systems to understand the both the nature o f the crisis, and the key characteristics of affected groups. Inthe case Nicaragua, for example, the workfare programs implemented by the government were probably less effective than desired, as they were better suited to assist newly unemployed laborers than to farmers who faced a dramatic decline in the value o f their product. In contrast, based on evaluation o f the Red de Proteccihn Social, which addressed both the income and human capital dimensions o f the crisis, some kind o f conditional or unconditional transfer program might have been more effective. Better information o n impacted groups earlier on, as well as explicit monitoring o f program impacts, may have facilitated the fielding o f more appropriate programs with greater impact. * * * * * * 4.22 Finally, it is important to note that in spite o f the similarities o f coffee crisis impacts across the four study countries, no one size fits all when it comes to designing a public risk management system that can effectively respond to shocks. The most effective responses are likely to differ across countries, given: (i) differences in the structures o f the coffee (or other affected) sector and the economy as a whole; (ii) different geographic distributions o f poor and vulnerable groups and the specific geographic impacts o f the shock, including across rural vs. urban areas; and (iii)differences in existing programs, public and non-governmental institutions and the government's capacity to target and administer programs. Indeed, the presence o f significant country-level differences reinforces the need for good forward planning and informationin ensuring effective responses to shocks. 58 References Alderman, H., J.R. Behrman, J.R., H-P. Kohler, J.A. Maluccio, and S. Cotts Watkins, 2001, Attrition inlongitudinal household survey data: Some tests for three developing country simples, Demographic Research, 5 (4), November 13,2001: 77-124. Arcia, G., 1999, "Proyecto de Red de Proteccidn Social: Focalizacidn de lafasepiloto. ",Report to the Inter-American Development Bank, Washington, D.C. Behrman, Jere R. and Ani1B.Deolalikar, 1988, HealthandNutrition. Handbook o f Development Economics, Volume 1, Hollis Chenery and T. N.Srinivasan editors, Elsevier. Beneke de Sanfeliu, Margarita, 2000, "Dinhmica del Ingreso de las Familias Rurales: Estudio de panel 1995-1997," Documento BASIS No. 1. Beneke de Sanfeliu, Margarita, 2001, "El papel de las microempresas en el ingreso de 10s hogares salvadoreiios ". BoletinEcon6mico y Social No. 188, FUSADES. Beneke de Sanfeliu, Margarita and Claudio Gonzalez-Vega, 2000, "Dynamics of Rural Household Income in El Salvador: 1995-1997Panel Results, ''Mimeo. Beneke de Sanfeliu, Margarita y Mauricio Shi, 2004, "Dinhmica del ingreso rural en El Salvador, borrador$nal, FUSADES, San Salvador, El Salvador Birdsall, Nancy and Carol Graham, Editors, 2000, "New Markets, New Opportunities? Economic and Social Mobility in a Changing World." Brookings InstitutionPress, Washington, D.C. Briones, Carlos y Andrade-Eekhof, Katharine (2000), "Participaci6n en 10s Mercados Laborales de 10s Residentes en las keas Rurales. Limitaciones y Desafios". Documento BASIS No. 2. Chawla, Mukes, 2001, "Malnutrition Among Pre-School Children." Background paper for Nicaragua Poverty Assessment, Vol. 11,Report No. 20488-NI, The World Bank, Washington, D.C. Coady, D.,P. Olinto, and N.CaldCs, 2003, Coping with the coffee crisis in Central America: The role o f social safety nets inHonduras, IFPRI, Mimeo Conning, Jonathan, Pedro Olinto, and Alvaro Trigueros, 2001, "Managing Economic Insecurity in Rural El Salvador: The role of asset ownership and labor market adjustments," Williams University-Economics Department Working paper Series, Williamstown, Ma. Conning, Jonathan; Pedro Olinto; and Alvaro Trigueros Argiiello, 2000, "Land and Labor adjustment strategies during an economic downturn in rural El Salvador."Technical Report, BASIS. Davis, Benjamin and Marco Stampini, 2002, "Pathways towards prosperity inrural Nicaragua; or why households drop in and out o f poverty, and some policy suggestions on how to keep them out," FA0 and Scuola Sant'Anna, Pisa. 59 Deaton, Angus, 1997, "The Analysis o f Household Surveys: A Micro-econometric Approach to Development Policy." Johns Hopkins University Press. Dunn, Elizabeth y Arbuckle Jr., J. Gordon, 2001, The Impacts ofmicro-credit: a case study from Peru, AIMS. FA0-Food and Agriculture Organization ofthe UnitedNations, at httu:/latxx.fao.ordsubscribcr Ferranti, David; Guillenno E. Perry, IndennitGill; and Luis Serven, 2000, "Asegurando el futuro en una economia globalizada". The World Bank, Washington, D.C. Fields, Gary S., 2001, "Distribution and Development: A new look at the developing world". The M I T Press, 2001. Flores Margarita, Adrian Bratescu, JosC Octavio Martinez, Jorge A. Oviedo, and Alicia Acosta. 2002. "CentroamCrica: El impact0 de la caida de 10s precios del cafi," Estudios y Perspectivas, Sede Subregional de la CEPAL en MCxico, Unidad de Desarrollo Agricola y Unidad de Desarrollo Econ6mico, CEPAL, Mtxico, D.F. Fofack, Hippolyte, Monga, Celestin and Tuluy, Hasan, 2001, Household Welfare and Poverty Dynamics in Burkina Faso. Empirical Evidence from Household Surveys. Policy Research Working Paper, No. 2590, The World Bank, Washington, D.C. Glewwe, P.; and P. Olinto, 2004, "Evaluating the impact of conditional cash transfers on education: An experimental analysis o f Honduras'PRAF program," Final Report for USAID, January, unpublished. Glewwe, Paul and Gillette Hall, 1998, "Are some groups more vulnerable to macroeconomic shocks than others? Hypothesis tests based on panel data from Peru." Journal of Development Economics. Vol. 56, pp. 181-206. Glewwe, Paul and Hall, Gillette, 1995, "Who is most vulnerable to macro-economic shocks? Hypotheses tests using panel data from Peru." Living Standards measurement Study Working Paper No. 117, The World Bank, Washington, D.C. Grimard, F., 1997, "Household Consumption Smoothing Through Ethnic Ties: Evidence from Cote d'Ivoire," Journal of Development Economics, 53, August 1997: pp. 391-422. Halliday, Timothy, 2003, Migration and risk in rural El Salvador, University o f Princeton, Mimeo. Hearne, R.; B. Barbier; and J. M. Gonzalez, 2002, "Development o f a minimumcost, incentive- based plan for the implementation o f a technology standard for coffee processing in Honduras." Paperpresentedat the 2002 AAEA Annual Meetings,Long Beach, CA. Holzmann, R. and Steen Jorgensen, 2000, "Social Risk Management: a New conceptual Framework for Social Protection and Beyond." Social Protection Discussion Series No. 0006, The World Bank, Washington, D.C. Hopkins, Jeffrey; Southgate, Douglas; and Gonzalez-Vega, 1999, Rural Poverty and Land Degradation inEl Salvador, Programa BASIS, Mimeo. 60 I C 0 - International Coffee Organization, at htto:iiwww.ico.org IFPRI, 2001, "Evaluation design for the pilot phase of the NicaraguanRed de Proteccidn Social." Report to the Red de Proteccidn Social. IFPRI, Washington, D.C. IFPRI, 2002, "Sistema de evaluacidn de la fase piloto de la Red de Proteccidn Social de Nicaragua: Evaluacidn de focalizacidn." Report to the Red de Proteccidn Social. IFPRI, Washington, D.C. Inter-American Development Bank, 2001, Transicidn Competitiva para el Cafk Centroamericano: Crisis Internacional del Cafk y su Impact0 en Nicaragua, Mimeo, Washington, D.C. Inter-American Development Bank, UnitedStates Agency for International Development, and the World Bank. 2002. "Managing the Competitive Transition o f the Coffee Sector in Central America,' Discussion Document, Prepared for the Regional Workshop "The Coffee Crisis and its Impact inCentral America: Situation and Lines of Action, Antigua, Guatemala, April 3-5,2002. Jalan, Jyotsna and Martin Ravallion, 1999, "Are the poor less well insured? Evidence on vulnerability to income risk inrural China," Journal of Development Economics, Vol. 58, pp. 61- 81. Jalan, Jyotsna and Ravallion, Martin, 1996, Transient Poverty in Rural China, Policy Research Working Paper No. 1616, The World Bank, Washington, D.C. Jalan, Jyotsna and Ravallion, Martin, 1998, Determinants of Transient and Chronic Poverty: Evidence from Rural China, Policy Research Working Paper No. 1936, The World Bank, Washington, D.C. Jalan, Jyotsna and Ravallion, Martin, 2001,"Income Dynamics in Rural China." Policy Research Working Paper No. 2706, The World Bank, Washington, D.C Klugman, Jeni, Diana Kruger and Kate Withers, 2002, "Consumption Risk and Smoothing DuringDisasters: The Case ofHurricaneMitchinNicaragua." The WorldBank, Mimeo. Kruger, D., A. Mason, and R. Vakis, 2003, "Shocks and Coffee: Lessons from Nicaragua, The World Bank," Mimeo. Lard6 de Palomo, Anabella y Argiiello de Morera, Aida, 2000, Integraci6n a 10s Mercados de 10s Hogares Rurales y Generaci6n de Ingresos, Documento BASISNo. 3. Lard6 de Palomo, Anabella, Claudio Gonzalez-Vega, and Aida Argiiello de Morera, 2000, "Household Integration to the market as a Determinant of Rural Incomes in El Salvador," Woking Paper No. OOP15, Ohio State University, Department of Agricultural, Environmental, and Development Economics, BASIS Research. Larson, Donald F., Jock R. Anderson, and Panos Varangis. 2004. "Policies on Managing Risk in Agricultural Markets," WorldBank Research Observer, (19) 2: 199-230. Fall. 61 Lederrnan, Daniel; William F. Maloney and Luis Serven, 2003. "Lessons from Nafta for Latin America and the Caribbean Countries: A Summary o f Research Findings", The World Bank, Washington, D.C. Lewin, Bryan; and Giovannucci, Daniele, 2002, Global Supply and Demand: New Paradigms in the Coffee Markets, The World Bank, Washington, D.C. Lokshim, Michael and Ravallion, Martin, 2000, "Short-lived Shocks with Long-lived Impacts?", Policy Research Working Paper No. 2459, The World Bank, Washington, D.C. Maluccio and Flores, 2003, Impact evaluation o f a conditional cash transfer program: The Nicaraguan Red de Proteccibn Social. Food Consumption and Nutrition Division Discussion Paper (forthcoming), IFPRI, Washington, D.C. Maluccio, J., 2003, "Coping with the Coffee Crisis: in Central America: the Role of the Nicaraguan Social Safety Nets Program", IFPRI, Mimeo. Marini, Alessandra, and Michele Gragnolati, 2002, "Malnutrition and Poverty in Guatemala." Policy Research Working Paper No. 2967, The World Bank, Washington, D.C. Marques, Josd SilvCrio. 2003. "Social Safety Net Assessments from Central America: Cross- Country Review o f Principal Findings," Social Protection Discussion Paper Series, No. 0316. Social Protection Unit,HumanDevelopment Network, World Bank, Washington, D.C. Mason, Andrew D. 2005. "Ensuring that the Poor Can Benefit from CAFTA: Policy Approaches to Managing the Economic Transition," background paper to the World Bank regional study on the impacts o f CAFTA (forthcoming), Washington, D.C. McCarthy N., and Y. Sun, 2003, "The potential role o f price insurance to improve welfare o f smallholder coffee producers: The case o f Honduras," IFPRI, Mimeo. Ministerio Agropecuario y Forestal, 2003, "Estrategia para reconversih y la Diversificacih Competitiva de la Caficultura en Nicaragua." Mimeo. Morley, S., and D. Coady, 2003, From social assistance to social development: Targeted education subsidies in developing countries, Washington, D.C., Center for Global Development and International FoodPolicy Research Institute. Morris, S.; 0.Neidecker-Gonzales, C. Carletto; M.Munguia; J. M.medina; and Q. Wodon, 2002, "Hurricane Mitchand the livelihoods o f the rural poor inHonduras," World Development, 30 (l), pp. 49-60. Narayan, Deepa, 1999, "Bonds and Bridges: Social Capital and Poverty." The World Bank, Poverty Group PREM. Okrasa, Wlodzimierz, 1999, "Who avoids and who escapes from poverty during the transition: evidence from polish panel 1993-96." Policy Research Working Paper No. 2218, The World Bank, Washington, D.C. Oxfam International. 2002. MUGGED: Poverty in Your Coffee Cup. http:llwwwmaketradefair.com. 62 Partners - Partners o f the Americas, 2002, "Small scale coffee production in Honduras," Washington, D.C., at www.partners.net Ptrez, Guillermo, 2003, BoletinEconbmico y Social, FUSADES. Pritchett, Lant; Suryahadi, Asep; Sumarto, Sudarno; 2000, Quantifying Vulnerability to Poverty. A proposedmeasure, applied to Indonesia. Policy ResearchWorking Paper No. 2437, The World Bank, Washington, D.C. Ravallion, M., 2001, The mystery o f the vanishing benefits: An introduction to impact evaluation, The WorldBank Economic Review 15 (1): 115-40. Ravallion, M., and S. Chaudhuri, 1997, "Risk and Insurance in Village India: Comment," Econornetrica, 65 (l), 1997: pp. 171-184. January Rodriguez-Meza, Jorge, and Claudio Gonzalez-Vega, 2004, "Household Income Dynamics and Poverty Traps in El Salvador," drafl, Ohio State University, Columbus, Ohio. Rodriguez-Meza, J; Southgate, D; and Gonzalez-Vega, C.; 2003, Rural Poverty, Household Responses to Adverse Shocks, and Agricultural Land Use: Panel Results for El Salvador, Mimeo, Ohio State University. Sadoulet, Elisabeth; Frederico Finan; Alain de Janvry and Renos Vakis, 2004, "Can Conditional Cash Transfer Programs Improve Social Risk Management? Lessons for Education and Child Labor Outcomes." Social Protection Discussion Series No. 0420, The World Bank, Washington, D.C. Skoufias, Emmanuel and Agnes R. Quisumbing, 2002, "Consumption Insurance and Vulnerability to Poverty: A Synthesis o f the Evidence from Bangladesh, Ethiopia, Mali, Mexico, and Russia." Mimeo. Stata Corporation, 2001, Stata statistical software: Release 8.0, college Station, TX Stata Corporation Stiglitz, Joseph, 2000, "Reflections on Mobility and Social Justice, Economic Efficiency, and Individual Responsibility." Chapter inBirdsall and Graham, 2000. Subbarao, Kalanidhi. 2003. "Systemic Shocks and Social Protection: Role and Effectiveness o f Public Works Programs," Social Protection Discussion Paper No. 0302, World Bank, Washington, D.C. Thomas, Duncan; Elizabeth Frankenberg; and James P. Smith, 2001, Lost but not forgotten: Attrition and follow-up in the Indonesia Family Life Survey, Journal of Human Resources, 36 (3): 556-92. Townsend, Robert M., 1994, "Risk and Insurance in Village India," Econornetrica, 62, May 1994, pp. 539-91. Trigueros, Alvaro, 2002, "The Economics o f Schooling and Child Labor for Boys and Girls in Rural Households in El Salvador: 1995-1999." Ph.D. Dissertation, Vanderbilt University, Nashville, Tennessee. 63 Trigueros, Alvaro and Carolina Avalos 2004, "Analysis of the Poverty and Social Impacts of the Coffee Crisis (Price Shock) inEl Salvador," Volume I1of this report. Varangis, P., P. Siegel, D. Giovannucci; and B. Lewin, 2003, "Dealing with the coffee crisis in Central America: Impacts and strategies," Policy ResearchWorking Paper No. 2993, The World Bank, Washington, D.C. World Bank, 1997, WorldDevelopment Report 1997: The state in a changing world. The World Bank, Washington, D.C. World Bank, 2001a, Honduras: Poverty Diagnostic 2000, Report No. 20531-HO, The World Bank, Washington, D.C. World Bank, 2001b, Nicaragua: Towards a Social Protection Strategy for Nicaragua: A Review of SelectedPrograms inthe PRSPPortfolio, The World Bank, Washington, D.C. World Bank, 2001c, Nicaragua Poverty Assessment: Challenges and opportunities for poverty reduction, Report No. 20488-NI, The World Bank, Washington, D.C. World Bank, 2002a, El Salvador: Social Safety Net Assessment, The World Bank, Washington, D.C. World Bank, 2002b, Honduras: Vulnerability, Risks and Poverty, The World Bank, Washington, D.C. World Bank, 2003a, Guatemala:Poverty inGuatemala, The World Bank, Washington, D.C. World Bank, 2003b, Nicaragua Poverty Assessment: Raising Welfare and Reducing Vulnerability, Report No. 26128-NI, The World Bank, Washington, D.C. World Bank, 2005, forthcoming, "Challenges of the Central America-US FreeTrade Agreement'', Carlos F. Jaramillo andDaniel Lederman(eds), The World Bank, Washington, D.C. Yang, Dean, 2003, "Risk, Migration and Rural Financial Markets: Evidence from Earthquakes in ElSalvador," University of Michigan, Mimeo. 64 Annex 1: A number o f recent empirical studies -within and outside LatinAmerica -have tried to measure how effectively households smooth their consumption - or "self-insure" - in the face o f adverse income shocks. While the specific findings differ from country to country, these studies find that households are partially - but not fully - effective at mitigating the impacts o f shocks to household income. Overall, the evidence suggests that households, on average, are able to protect between 60 and 90 percent o f their consumption per capita in the face o f changes in income (Table 3). That is, a 10 percent "shock" to householdper capita income translates into a roughly 1to 4 percent change inper capita consumption. Ingeneral, poor households seem to have fewer instruments available -and are less successful - ininsuringthemselves against risk than non-poor households. In China, for example, the wealthiest households only experienced a 1 percent decline in per capita consumption in the face o f a 10 percent decline in per capita income; in contrast, the poorest households experienced a 4 percent decline in consumption response to the same decline in income (Table 3). Annex Table 1.1: HouseholdConsumption Smoothing: A Summary of Recent Evidence from DevelopingCountries: Changein HouseholdPer Capita Consumption Resulting from a 10 PercentChangein Country per capita Income(Percent) Source Mexico (rural) 3.7 Skoufias (2002) Nicaragua (all country) 2.5 Klugman, Kruger, andWithers (2004) Peru(Urban) 3.O-3.6 Glewwe and Hall (1998) China (Rural) I Jalan and Ravallion (1999) Poorest 4.0 Richest 1.o India(Rural) IIII 1.2-4.6 III Ravallion andChaudhuri (1997) 65 66