Report No. 32867-MX Mexico Income Generation and Social Protection for the Poor (In Four Volumes) Volume IV: A Study of Rural Poverty In Mexcio August, 2005 Colombia and Mexico Country Management Unit Poverty Reduction and Economic Management Unit Latin America and the Caribbean Region Document of the World Bank INIFAP Instituto Nacional de InvestigacionesForestales, Agricolas y Pecuarias INTA Instituto Nacional de Tecnologia Agropecuaria LAC LatinAmerica and Caribbean LAG Local Action Group M&E Monitoring and Evaluation MxP Mexican Pesos NAFTA NorthAmerica Free Trade Agreement NBFI Non-Bank Financial Institutions NGO Non Government Organization O&M Organization andManagement OECD Organization for Economic Cooperation andDevelopment PAPIR Programade Apoyo a Proyectos de Inversion PEC ProgramaEspecial Concurrente PET Programade Empleo Temporal PITT Programade Investigacion y Transferenciade Tecnologia PROCAMPO Programade Apoyos Directos a1Campo PROCEDE Programade Titulacion de DerechosEjidales y Certificacion de Solares PRODESCA Programa de Desarrollo de Capacidadesen e l Medio Rural PROFEMOR Programade Fortalecimiento de Empresasy Organizacion Rural PROMAP Programade Modemizacion de laAdministracion Publica PRONAF ProgramaNacional de Agricultura Familiar PSP Prestadorde Servicios Profesionales RD Rural Development RDS RedParael Desarrollo Sostenible de MCxico RNF RuralNon-Farm ROA Roles ofAgriculture SAGARPA Secretariade Agricultura, Ganaderia, Desarrollo Rural, Pescay Alimentacion SCT Secretariade Comunicacionesy Transportes SE Secretariade Economia SECODAM Secretariade la Funcion Publica SEDESOL Secretariade Desarrollo Social SEGOB Secretariade Gobernacion SEMARNAT SecretariadelMedioAmbiente y RecusosNaturales SEP Secretariade Educacion Publica SHCP Secretariade Hacienday CrCdito Publico SIACON SistemaAgropecuario de Consulta SINDER SistemaNacional de ExtensionRural SRA Secretariade ReformaAgraria ST Secretariade Turismo STPS Secretariade Trabajo y Prevision Social TFP Total Factor Productivity UNAM Universidad Nacional Autonoma de MCxico USD UnitedStates Dollars VA Value Added WB World Bank ... 111 Acknowledgements The Rural Poverty report was preparedby ateam comprising JosC Maria Caballero (LCSER, Team Leader), FernandoBarceinas (consultant), Tania Carrasco(LCSHE), Jorge Franco (LCSES), Sara Johansson (consultant), SalomonNahmad (consultant), Fernando Rello (consultant), Mario Torres (consultant), Dorte Verner (LCSEO), andAntonio Yunez-Naude (consultant). The Rural Poverty report i s part ofthe SecondPhase ofthe "Mexico: Programmatic Poverty Work", which i s being carried out bythe World Bank in collaboration with the Government ofMexico. The report benefitedfrom commentsby Benjamin Davis andAlain de Janvry (peer reviewers), Mauricio Bellon (CIMMYT), and Bank staff Mark Cackler (LCSER), Marianne Fay (LCSFP), Ricardo Hernandez (LCSEN), John Kellenberg (LCSES), Gladys Lopez- Acevedo (LCSPP), MiguelLopez (LCC1C), Pedro Olinto (LCSPP), Jaime Saavedra (LCSPP), Emmanuel Skoufias (LCSPP), MichaelWalton (LCRCE), andAnna Wellenstein (LCSFP). Feedbackwas received from Government officials andpoverty analysis experts at a brainstorming meeting heldinMexico City on June 10-112004, and from high-level staff ofthe Secretaria deAgricultura, Ganaderia, Desarrollo Rural, Pescay Alimentacibn (SAGARPA) during apresentationofpreliminary results inAugust 2004. Specialthanks are due Antonio Ruiz Garcia (Subsecreatry SAGARPA), Armando Rios Piter (Subsecretary SRA), JosC Antonio Mendoza Zazueta (Director FIRCO), Ivan Cossio (FAO) and EdgardoMoscardi (IICA). Vice President: PamelaCox Country Director: Isabel Guerrero Sector Director: John Redwood Sector Manager: Mark Cackler Sector Leader: EthelSennhauser Task Manager: JosC Maria Caballero iv Table Of Contents Examining Rural Poverty InMexico: Why And How 1 The Dynamics Of Rural Poverty And Inequality In 1992-2002:An Overview 9 Activities, Employment And Incomes Of The Rural Poor 28 Agriculture, Poverty And The Small Farmer 51 Poverty Friendliness OfRural Policies 78 Issues And Challenges InThe Implementation Of Rural Policies 104 Heterogeneity And Vulnerability Of The Rural Poor 121 Summary OfPolicy Options To FightRural Poverty 152 Annexes 164 Bibliography 224 V 1. EXAMININGRURAL POVERTY INMEXICO:WHY AND HOW Thepurpose of this introductovy chapter is to explain the rationalefor studying rural poverty in Mexico today, discuss the approach, indicate some structural issues that are at the core of Mexicanpoverg, andpresent theplan of the study, data sources and issues. STUDY RATIONALE This studyis part ofthe second phase of a long-termprogrammaticwork onpoverty in Mexico in three phases being carried out by the World Bank at the request of the Government of Mexico (GOM). The first phasehas already concluded, resultingina report with an assessment o f conditions, trends and government strategy on poverty in Mexico (World Bank, 2004). The second phase consists o f three studies: the present one on rural poverty and two companion reports on urban poverty and social protection. The third phase will concentrate on issues related to quality and delivery o f services, especially to the poor in the framework o f decentralization. While the work has benefitedfrom extensive collaboration with government and independent Mexican specialists working on issues related to rural poverty, the views expressed are those o fthe authors alone. Why a study on rural poverty? First, because of the size and intensity of the phenomenon, poverty and inequality in rural Mexico are a matter o f concern not only from the well-being o f the poor point o f view but also from that o f the expansion o f internal market, inclusion o f large sectors o f the population traditionally excluded from the economic and social mainstream, and the political integration and stability o f the country. Poverty incidence in rural areas, in particular extreme poverty, i s much higher than in urban ones. Although most o f the country's moderate poor live inurban areas, most o f the extreme poor are rural, even if the rural population i s only one quarter o f total. Poverty issues have received considerable attention in Mexico over the last decade from academics and policy makers but there i s no recent comprehensive review o f ruralpoverty. Rural poverty differs from urban poverty in many important aspects. There are differences in sources o f income between rural and urban poor. Also, rural environment poses specific constraints for provision o f social infrastructure and services. Furthermore, institutions and culture tend to differ between rural and urban areas. The presence o f indigenous groups i s much larger inrural areas andthe production systems, the economic and other risks faced by rural poor and their coping strategies usually differ from those o f their urban peers. Urban poor are surroundedby services and opportunities -even if they have limited access to them- not available to rural poor, who often need to migrate in order to potentially enjoy those facilities (Warman, 2001: 30). By the opposite, rural poor benefit from safety nets such as subsistence agriculture, access to forest resources, and local community ties not available to the urbanpoor. Mexico needs to move away from a fragmented social protection system to a unified framework which nonetheless tailors different programs to different contexts in rural and urban areas. Consistent and equitable strategies for poverty reduction requires (1) a unified 1 framework for social policy, such as CONTIGO, rather than separate approaches for rural and urban areas, and (2) a unified system for setting policy priorities which consider simultaneously rural and urban needs. Programs, however, and other instruments for income poverty reduction and the provision o f basic infrastructure and social services should be designed and implemented within specific circumstances of the target group in mind. These circumstances usually differ betweenrural and urban areas, although they may overlap inthe semi-urban segment o f the rural space. Even programs targeting similar needs, like education and health, normally require different implementation systems in rural and urban areas. Examples o f differences in rural and urban program needs are provided inBox 1.1. Box 1.1.Differences betweenRural and Urban Program Needs All programs related to agriculture and natural resource management are specifically rural. Micro-finance programs, experience shows, operate differently inrural and urban areas and should be adjusted accordingly. Rural roads programs are rural by definition. Social infrastructure (electricity, drinking water, transport infrastructure, housing) is required in both urban and rural areas, but the specific demands, the cost o f providing the services, the engineering, the O&M systems and the forms o f community participation are usually different. In rural areas, environmental priorities are typically linkedto air pollution, collection and disposal o f domestic and hazardous waste, water scarcity and occupation o f fragilelrisky areas for residential purposes, whereas inurban areas they focus on deforestation, loss o f biodiversity, soil degradation, fertilizer and pesticide contamination o f soil and aquifer, and health hazards in their application. Urban marginality and violence are distinctly urban, linked to family breakdown, drug use and trafficking, degraded neighborhoods, opportunities for specific types o f robbery, close contact betweenthe destitute and the well-off, and tribal youth cultures. Marginality and violence exist in rural areas too but in a different way. Rural marginality i s related to income, employment, geographical constraints and often to ethnic characteristics, and does not carry negative moral implications. There are no "street kids" in rural areas. Rural violence exists but i s typically linked to land conflicts and the desperate fight o f rural organizations for human or economic rights, thus differing from the individual and mob criminality o f the cities. Only domestic violence seems to cut across the rural-urban divide. APPROACH TO POVERTY AND STRUCTURAL ISSUES Approach This study is policy-oriented: it examines poverty inthe rural areas of Mexico in order to examine strategic approaches to rural development, as well as policy options to promote pro- poor rural economic growth. M o r e efficient use of existing resources i s a key aspect. Attention i s paid inter alia to institutional and public administration issues related to the implementation o f rural development programs. There are two reasons for this: first, under conceivable circumstances o f availability o f fiscal resources and public expenditure allocation, significant increases inthe already high federal expenditure budget in rural areas are not likely. Second, there i s room for improvements in the design and implementation o f rural development programs, particularly those related to productive activities. An important result of the first phase of the Programmatic Poverty Study is that while considerableprogress has been made in meetingbasic needs in rural and urban areas over the last two decades, progress has been much lower in income poverty (World Bank, 2004). The present study concentrates, therefore, on the analysis o f income poverty and on 2 examining options for fighting it. Inparticular, we do not discuss issues and programs related to health, education and social infrastructure. This i s not to ignore that there are important complementarities between basic needs and income programs, and hence that both types o f programs shouldbe carried out within an integrated framework particularly at the local level. Rural areas are understood in a broad way, including country towns and considering both farm and non-farm activities. For statistical purposes we use two definition o f rural: a narrower one consisting of disperse populations in localities o f less than 2,500 residents, which i s the definition used by INEGI, and a broader one which includes semi-urban populations in localities between 2,500 and 15,000 residents. Semi-urban areas can be seen as transition regions betweenrural and urban spaces. They are an important part o f the rural system formed by the interaction of countryside and rural towns. In the study we contrast the performance o f different variables under these two statistical definitions o frural. We treat poverty as a multidimensional phenomenon intrinsically relative, with deep cultural aspects, and discuss the merits and limitations of quantifying poverty interms of measurable incomes and income lines. To analyze the incidence and evolution o f rural poverty we take first a more quantitative, income-basedapproach, used inthe first chapters o f the study. This approach is relaxed further down, particularly in chapter 7, in order to examine the heterogeneity o f poverty, the varying sources o f vulnerability, and the coping strategies o f rural poor. The analysis o f poverty friendliness o f agriculture and rural development policies and programs and the policy recommendations are also based on a more multidimensional and nuancedview o fpoverty. Spatial factors are important for rural poverty, which is unevenlydistributed in the Mexican geography. To the extent possible we take into consideration these factors using a space-conscious approach. The main way inwhich this i s done i s by contrasting the performance o f different regions regarding a number of variables. In particular, since the Southern region concentrates a disproportionate number o f rural poor, we often compare the situation o f this region with that o fricher ones. Long-term StructuralIssues Rural poverty in Mexico has strong historical roots resulting from the interplay over time of institutions, political power and market development in a highly varied geographical setting. We try to keep this in focus in the study by examining how poverty, its characteristics and determinants have evolved, and comparing situations at different points in time. We don't intend,however, tracing ahistory ofMexicanruralpoverty, for we look at the last decade only. To put things in perspective, however, it i s useful to mention right from the beginning some long-term structural issues that gravitate on contemporary poverty in Mexico. We do that briefly innext paragraphs. The high incidenceof ruralpoverty, particularly extremepoverty, in marginal areas i s the first structural issue. Combining survey data from ENIGH 2000 and 2002 with municipal level data from the 2000 population census, following a methodology proposed by Bigman et a1 (2000), CIMMYT has prepared a poverty and food security map for Mexico, where the municipal distribution o f poverty i s shown. CIMMYT's map illustrates a "non-uniform distribution o f poverty, i.e. islands o f poverty within Mexico. Poverty i s concentrated in mountainous and indigenous areas, mainly in Central and Southern Mexico, but also inthe mountainous regions o f NW Mexico" (Bellon et al, 2004:20). There is strong correspondence betweenpoor communities and municipalities identified in the poverty map and marginality as defined by the CONAPO marginality index used by SEDESOL.` Extreme rural poverty i s hence prevalent in marginal areas. From a historical perspectivemarginal areas are traditional zonas de refugio (shelter zones) o f indigenous and other destitute populations, characterized by physical isolation and hard topographical and/or agro-climatic conditions (Aguirre Beltran, 1967). Strong incidence o f extreme poverty in these areas i s important from a policy perspective, for much can be done to reduce core poverty by focusing efforts there. In fact, the Mexican government had a program to promote productive development in these areas recently ended -the Marginal Zones Development Program (Proyecto de Desarrollo de Zonas Margina1es)- and has an on-going program, Microrregiones, to promote clustered investmentsin these areas, which i s examined in chapter 5. Incidence o f extreme poverty inmarginal areas with a harsh geographic environment i s also evident in other regions o f the world with large indigenous peasantries, such as the Peruvian Andes (see World Bank, 2002a). Geography not only affects poverty directly by, for instance, making agriculture less profitable and more hazardous, but also and perhaps more, indirectly, by reducing the availability o f public and private assets in marginal areas (Escobar and Torero, 2000). Thus, geography and history combine to produce high incidence o f extreme poverty in certain rural areas. According to CIMMYT's poverty map, poor rural municipalities are dominantly placed inthe mountain. A second issue is the difficulty of raising the productivity of rural labor. The bottom- line o f poverty persistence in contemporary middle income countries with highly dualistic economies, such as Mexico, i s the inability o f the economic system to absorb the labor force engaged in low productivity "refuge" occupations into high-productivity employment. This applies to both urban informal and rural marginal laborers. High productivity employment, capable o f offering returns to labor above the poverty line, i s the only way in which income increases and poverty reduction can be sustained. Only if the economic system i s capable o f offering increasingly more high-productivity employment to low-productivity rural labor can rural development succeed. This was the case in contemporary developed countries with a history o f large populations o f small farmers living inpoor conditions, for example in much o f Southern Europe. Here, rural development took place under a combination o f three circumstances: (1) a strong pull of surplus labor away from agriculture into more productive occupations both within rural areas and outside them; (2) relatively low natural population growth; and (3) fast overall economic growth, which allowed considerable investment in the expansion o f high productivity employment andthe modernization o f rural areas (FAO-WB, 2003). These conditions are not yet in place in Mexico. It i s true that there i s strong rural-to- urban and rural-to-USA migration, but the demographic turning point has not yet been reached: the rural population i s still growing and only expected to stabilize around 2020. Also, most of permanent migrants to urban areas in Mexico seem destinedto swell the ranks o f urban informal sector where labor productivity may be larger than inmarginal rural areas, but remains very low. Second, fertility rates in rural Mexico are falling but are still large. Finally, Mexico's long-term Thus, as explained by Bellonet a1 (2004a: 12-13): "Close to 83 percent (n= 33,752) of the predicted extremely poor rural communities from our model occurred within the priority zones defined by the Mexican government for anti-poverty programs, and of those 33,752 communities 99 percent occurred within either "high" or "very high marginality municipalities. At the municipal level, the model predicted 1,020 municipalities to be below the food poverty line. This compares to 1,314 municipalities classified as of either "high" or "very high marginality by the MexicanMinistry for Social Development (SEDESOL).Out of the 1,020 predicted food poverty municipalities, 89 percent (n=909) coincide with highest marginality rankings of SEDESOL".We do not know, however, if poverty "density", i.e. the number of poor per km2, is higher in these marginal areas than inhighly populated non-marginal ones. 4 economic growth has been disappointingly low, at an annual average of around 0.3% per capita from 1981 to 2003. STUDY PLANAND DATA SOURCES Study Plan This report is organized as follows. Chapter 2 provides an overview o f the evolution o f poverty and inequality inthe rural areas o f Mexico inthe decades between 1992 and 2002. We examine the positive and negative factors influencing that evolution and propose a broad interpretation. The chapter concludes by looking at poverty profiles and comparing the correlates o f poverty in 1992 and 2002. Chapter 3 looks more closely on what happened to rural incomes, employment, labor markets and the characteristics o f rural labor force between 1992 and 2002. Five major themes are analyzed: the evolution o f the demographic and employment characteristics o f the rural labor force; how rural wages have changed and what determines them; sources o f income in rural areas and their evolution over the decade; the nature and growing importance o f non-agricultural rural employment; and the correlates o f the participation o f rural households in non-agricultural occupations. Chapter 4 i s devoted to examine the relation between poverty and the agricultural economy. We discuss first the role that agriculture has in poverty reduction and reviewthe evolution o f agriculture inMexico over the lasttwo decades. Usingsurvey data we examine then the profitability and efficiency o f the small farm sector in Mexico. Chapter 5 reviews the main agriculture, land and rural development policies and programs operating in Mexico, and examines them from the perspective o f poverty- friendliness. Options are examined to increase poverty-friendliness without harm to the primary objectives o f policies or programs, which do not need to be related to poverty. Chapter 6 starts with a theoretical discussion o f the issues and challenges usually faced inthe implementation o f development policies and programs. It moves then to examining concrete institutional problems faced in the implementation o f rural development programs, and to examining options to improve program implementation inMexico. Chapter 7 brings a more multidimensional and qualitative view, looking at how different types o f rural poor can experience their poverty situation, including strategies to survive, manage risk and achieve petty accumulation. We also examine the vulnerability o f the rural poor, and discuss the incidence on this of some government programs. We include political and cultural dimensions and recognize the heterogeneity o f poverty situations. Chapter 8 concludes with a summary o f policy options to reduce rural poverty in Mexico. 5 Data Sources and MethodologicalIssues Four main sources of data are used in the study. The first source is the well-known Encuesta Nacional de Ingresos y Gastos de 10s Hogares (ENIGH) carried out by Geography, Statistics and Surveys National Institute (Instituto Nacional de Estadistica Geografia e Informatica, INEGI) every two years, which contains standard income-expenditure household survey information. ENIGH is used by SEDESOL and the Poverty Measurement Technical Committee (Comite' Te'cnico de Medici6n de la Pobreza, CTMP) to estimate poverty levels in Mexico. The last survey available i s for 2004. ENIGH data for 2004 became available shortly before this report was going to print. It has not been possible, therefore, to incorporate this information fully in the analysis. Also, ENIGH data previous to 1992 i s not entirely comparable with that from 1992 onwards. For these reasons we show poverty figures for 2004 but confine most statistical analyses usingthe ENIGH source to the period 1992-2002. Because o f the recent availability o f the 2004 ENIGH results, it has not been possible to measure poverty based on consumption expenditures and a consumption poverty line. Nor has it been possible to calculate the poverty gap and poverty severity measures (FGT1 and FGT2).2 Hence, the only poverty measure reported for 2004 i s the incidence o f income poverty or FGTO. Questions have been raised as to the comparability o f ENIGH results of 2000 and 2002 because o f some changes in survey design and inthe questionnaire. There have also been some changes in 2004. We discuss this further down. ENIGH i s designed to be representative at the national, rural and urban levels, not at the state or regional levels. We use, however, information at the regional level, but it should be bear in mindthat these are survey results, not necessarily representative of the true situations in the regions. Perhaps more importantly, ENIGH i s not designed to give statistically significant results for the separate sources o f income o f the households, only for the aggregate. However, we use abundantly in our analyses information on the individual sources of income to which the same caveat as before applies. The second source of data is the National Jobs Survey (Encuesta Nacional de Empleo, ENE) carried out by INEGI every three months, which gives information on the characteristics o f the labor force at the individual level, including education, gender, age, occupation, labor status, and other variables. Panel data were included since 2000 following individuals during five quarters. ENE uses a large sample which i s representative at the rural and urban levels, for all states and regions, and for the 48 major cities in the country. Data from ENE surveys previous to 1995 are not entirely comparable with those for the following years. For this reason, when usingENE data we examine the period 1995 to 2003. The third source of data is the Mexican Rural Homes National Survey (Encuesta Nacional de Hogares Rurales de Mixico, ENHRUM), carried out in 2002 in collaboration between INEGI and El Colegio de Mixico. ENHRUM covered 1,782 rural households in localities between 500 and 2,500 residents. ENHRUM provides data on assets, production technology, sources o f income, and socio-demographic aspects including migration. Results are representative for the entire country and INEGI's standard regions (Centro, Sur-Sudeste, Centro- Oeste,Noroeste andNoreste). Finally, we have usedAgriculture, Cattle Farming RuralDevelopmentand Fisheries Ministry (Secretaria deAgricultura, Ganaden'a,Desarrollo Ruraly Pesca, SAGARPA) on-line agricultural data base, the Agriculture Consul System (SistemaAgropecuario de Consulta, These measures are explained inchapter 2. 6 SIACON), for information on crops, area, yields, and prices, and on-line information from INEGI's Economic Information Bank (Banco de Informacibn Econbmica) for other relevant economic and demographic data. For convenience we use point estimates in our tables and text and we make comparisons between them without explicit reference to the underlying confidence intervals.We should bear in mindthat the point estimates used to assess poverty incidence and other relevant measures are just one o f many possible estimates within confidence intervals to which a certain probability i s assigned. We shouldnot attach other value than this to the estimates No definite value should therefore be attached to the figures, more so since in many cases the disaggregate level used for results based on the ENIGH survey makes them not statistically significant under the designo f the survey. Doubts havebeenraised on the comparabilityof poverty incidence and other results from ENIGH surveys for 2000 and 2002 because of some changes inthe questionnaire, larger size o f sample and some change in sample design in 2002. Because o f this, the decrease in poverty incidence between the two dates shown by ENIGH data has been challenged, particularly inview of the fact that this was a period of economic stagnation. The CTMP has expressed its views on the comparability o fboth survey^,^ which can be summarized as f01lows:~ 1. The decrease of rural poverty between 2000 and 2002 indicated by the point estimates from the ENIGH surveys i s statistically significant for the extreme poor (food poverty line) but not for the moderate poor (assets poverty line) and for the capacity poor (capacity poverty line). Even for extreme poverty, where the decrease i s statistically significant, the precise change indicated by the point estimates i s containedwithin confidence intervals and cannot be taken as the true change. 2. The larger sample size in 2002 than 2000 does not seem to introduce sample biases that affect comparability at the aggregate level. 3. Independently of comparability, the changes introduced in 2002 result in an improvement inthe survey and inthe measurement o fpoverty. 4. There are doubts as to the comparability o f specific sources o f income o f rural households because o f the changes in the questionnaire and/or sample size and design. 5. In particular, there are differences in the comparability o f incomes derived from public transfers, where the amount registered in the surveys could be compared with the disbursement figure in the public accounts. Public transfers seem to be better measured in 2002 than in 2000, giving considerably higher estimates for the former year. This i s o fparticular importance for the poorer sections of the population. 6. Changes in the questionnaire create doubts on the comparability o f certain expenditure categories. This i s particularly the case for non-monetary expenditures 3The Comite'is an independent body of academics specialized in poverty analysis created by SEDESOL to provide technical advice on issues related to poverty measurement in Mexico. Some technical staff from SEDESOL participate inthe Committee but without voting capacity. See Cortks (2003), Vences-Rivera (2003), CTMP (2004), World Bank (2004), Scott (2004), and (CTMP 2005). 7 and especially for the substantial increase registered in the amount reported for the imputedvalue ofown housing. The poverty figures shown inthis report for 2004 are those of income poverty of the CTMP, calculated on the basis of ENIGH 2004. The calculation was made according to CTMP's poverty measurement methodology, which i s the one officially adopted in Mexico. In the document reporting the 2004 figures, the Comite' points out that INEGI has indicated the existence o f some changes inthe questionnaire and sampling framework between 2002 and 2004 that were considered by INEGI not to affect poverty estimates. The effect o f these changes has not yet been examined by the Comite'. CTMP carried out, however, tests o f significance o f the variations inpoverty incidence between2002 and 2004. Poverty reduction between the two years i s statistically significant at the 95 percent confidence level for all three poverty lines at the national level and for rural areas. Changes in poverty incidence are not significant for any o f the poverty lines in urban areas, where small increases are reported for extreme and capacities poverty and a slight decrease for moderatepoverty. 8 2. THE DYNAMICSOF RURAL POVERTYAND INEQUALITY IN1992-2002: AN OVERVIEW This chapter presents basicfigures about rural poverg and inequalig in Mexico and how they evolved in the decade between 1992and 2002. The main findings fromthe chapter are as follows: Poverty inurban and rural areas, both moderate and extreme, went through a cycle in the decade o f 1992-2002 marked by the 1995 crisis, with a strong increase aroundthe 1995 crisis between 1994 and 1996 and decreasing trend thereafter. Rural poverty decline was strong between 1998 and 2004. Rural poverty and inequality are comparatively high and did not experience long term progress during the 1992-2002 decade. In 2002, 35 percent o f rural dwellers were extreme poorly and 67 percentmoderatelypoor. There are large poverty differences within rural areas, with a gradient o f poverty incidence that increases from urban to semi-urban and to disperse rural areas and from Northern to Southern regions inthe country. The lack of overall progress in long-term poverty reduction in rural areas can be explained mainly by the 1995 economic crisis, the sluggish performance o f agriculture, and stagnant rural wages. These circumstances were compensated to some extent by the increase in private and public transfers, improved targeting o f parts o f public expenditure in rural areas, and a notable expansion o f employment and income inruralnon-farm activities. Rural poverty i s positively associated with lack o f education, living in disperse rural zones and in the South, Gulf o f Mexico and Central regions, young households with small age children andbeingemployed in agriculture. The main areas for policy actionderived fromthese findings are as follows: Macroeconomic stability i s essential for reducing poverty. Maintaining the level o f direct transfers to the poor i s important in view of their significance for their livelihood, but incremental resources at the margin could better beusedto promote income and employment growth. Regions and areas where poverty i s more concentrated requires specific attention. Buildingup critical investment masses to trigger endogenous growth process usinga territorial development approach with participatory planning methods would be more effective than thinly distributinginvestments. 9 Rural education, andtechnical and vocational training for ruralpeople relatedto farm and non-farm activities, are keyto productive development. Young households need specific support to access productive assets to start up independenteconomic activities. EVOLUTIONOF RURALPOVERTYAND INEQUALITY Poverty in 1992-2004' Poverty saw a moderate increasein 1992-1994, a strong increasein 1994-1996, and a decreasing trend in 1996-2004 (Box 2.1). The evolution o f national, urban and rural poverty, both extreme and moderate, between 1992 and 2004 was in fact rather similar (Figures 2.1 and 2.2). Since 1996, with the economic recovery, Mexico has made headway in reducing national poverty, and extreme poverty inparticular (Table 2.1). The incidence o f extreme poverty in 2002 was similar to that in 1992, o f the order o f 20 to 22 percent, with around 20 million Mexican living in extreme poverty, i.e. without sufficient income to buy a minimumbasket of food. The lack o f overall progress registered in the 1990s compounds the unfavorable situation in second part o f the 1980s. In 1984-89 poverty increased sharply, with rural areas beingthe most affected.6 A decade was thus lost for poverty reduction. The evolution between2000, 2002 and 2004 was very positive inrural areas, with a reduction of2.8 millioninthe number of extreme ruralpoor in 2000-2002 and another 2.4 million in 2002-2004. It i s interesting that the falling trend in rural poverty since 1998 has taken place under differentperformances of the macroeconomy, although inthe absenceofan economic crisis. Box 2.1. Poverty indicators This reportuses three poverty indicators developed byFoster, Greer and Thorbecke. The poverty headcount ratio, FGT(O),measures the percentage share of people whose income or consumption falls below the poverty line. The poverty gap index, FGT(l), measures the depth of poverty, i.e. the distance of the actual income or expenditure o f the poor to the poverty line. Unlike FGT(O), FGT(1) can tell us whether the poor are on average very poor, i.e. very far fiom reaching the income or consumption expenditure benchmarked by poverty line, or whether many o f the poor are clustered around that line. FGT(1) can also be interpreted as the cost o f eliminating poverty with perfect targeting, i.e. providing each poor enough income to place him or her above the poverty line. The poverty severity index, FGT(2), measures the relative position of the poor with respect to the poverty line, but distances are squared to give more weight to individuals or 5 Throughout this chapter, unless otherwise indicated, poverty refers to extreme poverty and poor to the extremely poor. The Secretaria de Desarrollo Social (SEDESOL), following the Comite' Te'cnicopara la Medicion de la Pobreza, uses three poverty lines: a "food-based" poverty line (income required to acquire enough food to cover nutritional needs); a "human needs" or "capacities" poverty line, which includes also the income required to acquire basic education, health, housing, dress, footwear, and transportation; and an "assets- based" poverty line, which also includes other needs. The latter corresponds to the usual broad definition of "poverty", which we call "moderate poverty', while the former corresponds to the usual definition of "extreme poverty" (see SEDESOL/CTMP, 2002, and World Bank, 2004, for more information on poverty lines). See Cortks and Rubalcava (1991), Lustig (1992), Szkkely Pardo (1994), and Hernandez Laos (1994) on poverty evolution during the 1980s, and the review of this literature by McKinley and Alarcon (1995), who stress the impact that poverty increases had on rural areas. 10 households most distant from the line. The poverty severity index picks up changes in the distribution o f income within the poor. See Foster, Greek and Thorbecke (1984). Extreme poverty is moreprevalent in rural than in urban areas. Extreme poverty has declined at a faster rate inrural than inurban areas since 1998, and the gap i s hence closing. But poverty incidence in rural areas i s still considerably higher. Thus, in 2004, the poverty count measure for extreme poverty was 27.6 percent for rural areas vs. 11.O percent for urban areas, and the equivalent figures for moderatepoverty were respectively 56.9 and 41.0 percent. Figure 2.1. Mexico: Incidenceof Extreme Poverty FGT(O), 1992-2002 60 , , , , , , 1992 1994 1996 1998 2000 2002 2004 -*- National ..*..Rural +Urban Source: WB staff calculations based on ENIGH. Mexico has steep gradients in the conditions of living from the more developed urban areas through the urban periphery and smaller towns to the more remote rural areas. In 2002, the incidence o f extreme poverty was in fact twice higher in disperse rural localities than in semi-urban areas and close to four times higherthan inurban areas (Table 2.1). Figure 2.2. Mexico: Incidenceof Moderate Poverty FGT(O), 1992-2002 11 10 1992 1994 1996 1998 2000 2002 2004 -a-National-- X- 'Rural +Urban - Source: WB staff calculations based on ENIGH 1992, 1994, 1996, 1998,2000 and 2002 The steep rise in poverty along the urban-rural continuum is also visible in the poverty gap and poverty severity indices. The two other FGT poverty indicators confirm the general picture given by the poverty headcount index. As shown in Table 2.1, the poverty gap increases the smaller and more rural the localities are considered. The same holds for the poverty severity index. Thus, people are, on average, poorer -further below the poverty line- in rural than in urban areas, and there are also more people clustered at the very bottom o f the distribution. Indeed, the average income of the poor decreases as we move from urban to semi- urban and to disperse rural areas, indicating higher depth o f poverty (Figure 2.3). The trends between 1992 and 2002 are also similar, as both the poverty gap and the poverty severity index increase in conjunction with the 1995 crisis, and in spite o f the reduction since 1996, remain around the levels prevalent in 1992. Table 2.1: Extreme poverty inRural and Urban Mexico, 1992-2002 (percent) Headcount Poverty Gap Poverty Severity FGT(0) FGT(1) FGT(2) 1992 1996 2002 1992 1996 2002 1992 1996 2002 National 22.4 37.1 20.3 7.5 14.1 6.3 3.5 33.9 3.2 Urban (localities > 15,000) 13.3 26.5 11.4 3.6 8.3 2.8 1.4 3.7 1.1 Rural (localities with 1-15,000) 35.6 52.4 34.8 13.1 22.5 12.2 6.5 77.5 6.6 Semi-urban (localities with 2,501- 15,000) 17.4 35.6 21.1 6.0 13.1 7.0 3.0 11.0 3.4 Disperse Rural (Localities with 1-2,500) 44.7 60.8 42.1 16.7 27.1 14.9 8.2 11.4 8.3 Source: WB staff calculations based on ENIGH 1992, 1996 and 2002. Table 2.2: Average Monthly Per Capita Incomeof the Extreme Poor and the Non-Poor in Mexico, 1992,1996 and 2002 (2002 Prices) 1992 1996 2002 Region Non-poor Poor Non-poor Poor Non-poor Poor DisperseRural(') 1,277.76 329.29 1,163.56 302.80 1,417.20 315.45 Semi-urbad2) 1,600.69 346.55 1,304.62 349.91 1,521.37 327.74 Urban 2.940.99 520.70 2.470.30 511.12 2.647.41 506.64 (1) Less than 2,500 residents; (2) between 2,500 and 15,000 residents Source: WB staff calculations based on ENIGH 1992, 1996 and 2002. 12 Geographic Factors Poverty is more prevalent in the Southern regions and the gap between Northern and Southern states increased between 1992 and 2002. Regional area appears to be an important determinant o f poverty. There are large variations in income poverty among regions, with a generalized gradient from North to South (Figure 2.3).7 In2002, the poverty rate in rural areas o f less than 15,000 inhabitants in the Norte region was 14 percent, a third o f that inthe Sur region, where 48 percent were poor. Distance among regions may in some cases matter more for rural poverty than the rural-urban difference. The dynamics o f poverty i s also sensitive to region, as poverty fell in some regions but not in others between 1992 and 2002. Thus, as shown in Figure 2.3, the Capital, Centro, and Centro-Norte regions experienced a considerable fall in the headcount poverty rate in 1992-2002, with the share o f poor people falling by more than 10 percentage points in each o f these regions. This compares to the Golfo region where it i s estimated that poverty increased by more than 10 percentage points, and the Sur region where poverty also increased Figure 2.3: Incidence of Extreme RuralPoverty in Mexico by Regionin1992-2002 (Percentage) 0 1 9 9 2 n 0 1 9 9 6 0 2 0 0 2 60 0 I r 1 - Mexico Norte Capital Golfo Pacific0 Centro- Centro Norte Note: Rural defined as localities o f less than 15,000 residents. Source: WB staff estimates based on ENIGH 1992, 1996 and 2002. It is important to note that ENIGH - the basis for the income poverty estimates - is not designed to be representative at a regionallevel. As a result, the regionalnumbers on income poverty should be interpreted with care. The differences between regions here presented are statistically significant. 13 Figure2.4: Mexico: RuralPovertyGap, FGT(l), byRegion in SelectedYears - Sur Centro-Node Centro Note: Rural defined as localities of less than 15,000 residents. Source: WB staff estimates based on ENIGH 1992, 1996 and 2002. Poverty is also deeper in the Southern, Gulf and Central regions. If we look at poverty gap figures by region (Figure 2.4) we can observe large regional differences. Thus, in 2002, FGT(1) was less than 5 percent inthe Norte and Capital regions, and more than 15 percent in the Centro, Golfo and Sur regions. During 1992-1996 the poverty gap increased most in regions where poverty was less deep in 1992, such as the Capital andNorte. As a consequence of the 1995 crisis, the depth ofpoverty was at its highest for all regions inthe middle of the 1990s. Other measures of regional welfare levels corroborate the regional differences in income poverty.Work on census data confirms the higher prevalenceofpoverty inthe Southern part ofthe country as well as its persistence over time (see Araujo, 2003). Inequality Mexico's income distribution did not improve much during the decade 1992-2002 and remains very unequal. In 2002, the Gini coefficient for rural Mexico was 0.517 for household income and 0.492 for household expenditure. As shown in Table 2.3 and Figure 2.5, rural inequality decreased in 1992-96, increased in 1996-00 and decreased again in 2000-02. Altogether, income inequality among rural households did not change over the decade but inequality in household expenditure increased. There were strong variations in both income and expenditure Ginis during the period, but as i s usually the case inequality in expenditure was generally smaller than in income, with the exception of 2000. Inequality has tended to be counter cyclical for both income and expenditure, with the 1995 crisis having no significant influence, the 1996-2000 recovery a disequalizing one, and the 2000-02 stagnation period being equalizing (World Bank 2004). Table2.3. Mexico: Gini Indicesfor IndividualIncomeandExpenditure inRuralAreas 1992-2002 1992 1994 1996 1998 2000 2002 Income 0.47 0.44 0.45 0.49 0.53 0.49 Expenditure 0.41 0.40 0.42 0.44 0.56 0.48 Note: Rural defined as localities of less than 15,000 residents. Source: SEDESOLestimates usingENIGH. 14 Figure2.5: Mexico: IncomeandExpenditureInequalityinRural andUrbanAreas Measuredby Gini coefficients,1992-2002 0.6 National 1 Urban 1 Rural National 1 Urban 1 Rural I " C 0 I l E Expenditure lm1992 1994 o ~ ~ ~ ~ ~ w zooz_l~ ~ s ~ ~ ~ o o o ~ w Source: World Bank, 2004. When we compare rural income and expenditure inequality measured by person and measured by household, inequality is consistently smaller for personal than for householdincome and consistentlylargerfor personalthan for householdexpenditure.*The former follows from the fact that inequality in family size compounds the effect o f inequality in personal income resulting in more variation in the income distribution. The latter seems to indicate that there are scale economies in expenditure by families who are also more successkl than individuals at expenditure smoothing. Income distribution is more dispersed for families with diversified income sources and/or who receive transfers and this inequality has increased over time. Income inequality differs markedly according to the dominant source o f income o f rural families. InTable 2.4, we show Gini coefficients for rural families classified according to their main (more than 50%) source of income. Inequality i s smaller for agricultural wage labor families and larger for families with diversified income sources, and for those depending mostly on transfers. The fall in inequality between 1992 and 2002 infamilies depending on wage labor i s probably a reflection o f the fall in real rural wages. Contrarily, the increase in inequality in transfer dependent and diversified income families reflects the raising importance o f transfers and non-agricultural activities as sources o f income and employment opportunities in rural areas, resulting in more variation in the income distribution for these categories o f families. Interestingly, as discussed in more detail in chapter 3, the new opportunities seem to have both reduced poverty and increased inequality. This assertion is derived from ENIGHdata not shown inthe tables and figures presented here. 15 Table 2.3. Mexico: GiniIncomeCoefficientsby Type ofRuralHouseholds Classifiedbytheir Main (>50%) Source ofIncome,2002 Independ. Non-ag. Agricul. Non-ag. Transfer Diversi- Farming Entrepren. wage wage dependent fied labor labor 1992 0.45 0.41 0.36 0.53 0.44 0.36 2002 0.44 0.48 0.33 0.46 0.54 0.55 Source: WB staff calculations based on ENIGH 1992 and 2002. Figure2.6: Mexico: RuralExpenditureShareby ExpenditureDecile, 1992,1996 and 2000 Expenditure Decile Source: WB staff estimates based on ENIGH 1992, 1996, and 2002. Another way of looking at rural inequality is to focus on how different sectors of the rural population participated in total rural expenditure. We do this in Figure 2.6, where we compare the share o f total expenditure by expenditure decile in three years: 1992 (continuous line), 1996 (broken line), and 2002 (dotted line). We observe that there was no alteration between 1992 and 1996 and a worsening o f the distribution between 1996 and 2002. The reason was the increase inthe share o f expenditure at the top o f the distribution, i.e. by the 10th decile. Whereas in 1992 and 1996 the expenditure ofthe top decile was 32 percent ofthe total, itwas 45 percent in 2002. A similar exercise carried out for agricultural families (both independent farmers and agricultural wage labor families) and non-agricultural families (the rest), not shown inthe figure, indicated that there was no change in the distribution o f the share o f expenditure in agricultural families during the entire period. The change, however, between 1996 and 200, was marked for non-agricultural families. It i s hence the non-agricultural families that explain changes in the distribution o f expenditure between 1996 and 2002 for the rural sector at large. The more unequal income is distributed the less effective economic growth is in reducing poverty. InMexico, even with steady growth, poverty reduction tends to be slow as a consequence o f the country's high income inequality. This i s also the reason why the poverty indicators o f Mexico are worse than those o f other countries with similar per capita income (World Bank 2004). Changes in inequality are typically slow, except during periods o f radical social and institutional change. Where inequality has fallen, it has usually happened inassociation with major expansion and equalization ineducational attainment, as inKorea and Malaysia inthe 1970s and 1980s. Mexico's expansion in education and reduction in education inequalities in rural areas may be too recent and perhaps too segmentedin quality to have a significant effect on the composition o f skills, and occurred during a period in which the overall return to high skill levels was rising and that to basic skills falling. Mexico experienced a decrease in returns to 16 tertiary education since the middle o f the 1990s, but rural areas experienced relatively little o f this fall as very few rural-dwellers hold a universitydegree. EXPLAININGTHE TRENDS INRURALPOVERTYAND INEQUALITY The conclusion from the previous section is that following a period of poverty increase in the second part of the 1980s, rural poverty started a sustained decreasing trend in spite of the lack of progress in rural inequality. Comparing, however, 1992 with 2002, poverty in rural areas remained stagnant. The economic crisis in the mid-l990s, the sluggish performance in the agricultural sector and the reduction in rural wages all had a negative impact on poverty. Such negative developments were somewhat offset by an improvement in non-farm opportunities and an increase inprivate andpublic transfers to ruralhouseholds. These themes are summarizedbelow and developedfurther insubsequent chapters. TheEconomic Crisis of theMid 1990s The so-called Tequila Crisis in 1995 had a tremendous impact on both rural and urban poverty. There were 3.9 million more rural poor in 1996 than in 1994; the poverty incidence had passed from an estimate o f 37.0 percent in 1994 to one of 52.4 percent in 1996. By 2000 rural poverty had not yet fully recovered from the effect o f the crisis, notwithstanding the favorable evolution o f the macro economy from 1996 to 2000, with an increase o f 30 percent in total GDP between these two dates. Not only did the number o f poor increase substantially, but the rural extreme poverty gap almost doubled, passing from an estimate o f 13.5 percent in 1994 to an estimate o f 23.9 percent in 1996. The crisis hit the rural sector mainly through a fall in rural wages and a reduction of private transfers. Although agricultural GDP increased during the crisis, agricultural real prices and the real value o f crop output fell,9 but it i s not clear how much this was the result o f the crisis or o f the evolution o f international prices and the liberalization o f trade. Public expenditure in rural areas fell as well in real terms from 1995 to 1999. From 1992 to 1996 the average monthly income o f the Mexican rural worker had fallen by one third, from MxP 3,029 to MxP 2,03 1in constant 2002 prices." Agricultural families were hardest hit. There i s little information on how the crisis affected different sectors o f the rural population." From the point of view o f spatial distribution, all regions were strongly hit, but the richer and the poorer regions were hit proportionally more (Figure 2.3). Different types of occupations were also hit differently. As shown in Table 2.5, agricultural families, relaying mostly on independent farming or on agricultural wage labor, were the worst hit. Percentage-wise, poverty increased more among the non-agricultural entrepreneurial families but poverty levels are much smaller in this group. The relatively small The index of the real average price of crops (base 1980=100)fell from 77.8 in1993 to 63.5 in1997, while the index of the real value of food crop output fell from 103.2 in 1993 to 90.2 in 1997.Indices are of purchasing power, i.e. deflated by the consumer price index, and were calculated from SAGARPA's SIACON data base. There was no fall, however, inagricultural GDP in1995.See Chapter 4. loEstimated from ENIGH. 11Lack of panel data for rural areas for the crisis period prevents an analysis of the distribution of the impact of the crisis among households according to their profiles. The finding of Maloney, Cunningham and Bosch (2003)that inurban areas the impact was generally distributed ina way similar to the distribution of shocks in more normal periods probably applies to rural areas as well, although of course shock intensity was particularly large duringthe crisis period. 17 change inpoverty inthe non-agricultural wage labor families i s to be noticed, and it i s somewhat surprisinginview o fthe increase inpoverty among wage laborers inurban areas uponthe crisis. Table 2.4. Mexico: IncidenceofExtremePovertyinRuralFamilies ClassifiedAccordingto their Main (>50%) Source ofIncome, 1992 and 1996 Independ. Non-ag. Agricul. Non-ag. Transfer Diversi- Farming Entrepren. wage wage dependent fied labor. labor 1992 49.2 22.6 50.4 32.7 44.4 49.6 1996 65.8 36.0 65.3 38.7 52.0 62.3 Source: WB staff calculations based on ENIGH 1992 and 1996 Unfavorable labor market conditions appear to have resulted in higher out- migration from rural areas. Figuresare not available on the evolution o frural employment, and hence we do not know how important unemployment was as a source of poverty intensification in rural areas duringthe crisis. As further developed in chapter 7, Mckenzie (2003) found that rural workers could not resort to increased labor market participation because o f the depressed state o f the markets. Unemployment rose in urban areas after the crisis (although the bulk o f the adjustment inurban labor markets was via real wages), and this musthave affected the possibility o f finding urban jobs by temporary migrants, as well as that o f coping with the rural crisis by more permanent rural-to-urban migration. ENHRUM figures show, however, an acceleration o f internal migration from rural areas after the crisis. Rural migration to the USA accelerated even more: ruralhouseholdmembers inthe USA doubledbetween 1994 and 1999 (see chapter 7). AgriculturalPerformance The sluggishperformance of agriculture and thefall in the agricultural terms of trade also depressed agricultural incomes.Agricultural growth duringthe decade concentrated on the irrigated and more commercial farming segments, thus deepening the characteristic dualism o f the sector. Between 1992 and 2002 average food crop yields increased 1.9 annually, which i s not much considering the low yield levels and existing yield potential (see chapter 4). The terms o f trade for the food crop sector fell 5.7 percent annually in the decade and the real value o f food crop output decreased 3.5 percent annually. The yield o f cereals, a particularly important crop for the poor, increased from an average o f 2.5 tonhectare in 1991-93 to an average o f 2.8 todhectare in2000-2002, and the realprice of food crops fell at an annualrate of 6.1 percent between1991 and 2002.12Falling food crop prices do not have an unambiguous effect on poverty, however. The fall o f real producer prices o f food crops may translate into cheaper prices o f food to consumers, and thus increase the real income o f the segment of the rural poor who are net buyers o f food. This may compensate for the decrease of income for the segment of the rural poor who are net sellers o f food crops. Rural Wages Real wages in agriculture fell from an already low level, while rural wages in general stagnated. An index o f average real remunerations in agriculture calculated from l2Calculated form SAGARPA's on-line agricultural information system. More information on the evolution of agriculture is provided inchapter 4. 18 INEGI's national accounting data fell from 102.2 in 1992 to 75.6 in2001 (base 1993=100).13 The agriculture hourly wage, calculated from the EncuestaNacional de Empleo (ENE), fell from an average o f MxP 8.3 in 1995 to an average o f 7.4 in 2003, in constant 2002 MxP, while the average rural wage remained stagnant at around MxP This contrasts with average hourly wages in rural areas for 2002 o f MxP 12.3 in manufacturing, 10.5 in commerce and 14.3 in services (see chapter 3). Low agricultural performance and wages have a particularly significant effect on rural poverty because rural poor depend more on agricultural incomes than rural non- poor, and also because o f the positive association between agricultural growth and rural poverty reduction (see chapter 4). Income Diversification Income diversification out of agriculture was very important in the 1992-2002 decade for all rural families in general, poor and non-poor alike. The weight o f agriculture on income passed from one half to one third. Whereas in 1992 agriculture accounted for 51 percent o f family income in disperse rural areas, it only accounted for 34 percent in 2002,15 including incomes from both independent farming and agricultural wage labor. If the wider definition o f rural i s used (localities o f less than 15,000 residents), the corresponding figures are significantly lower: 44 percent, and 17 percent respectively. Conversely, income from non-farm employment inlocalities ofless than 15,000 residents increased from 35 percentin 1992to 49 percentin2002. The ruralpoor and the ruralnon-poor bothbenefitedfrom the expansion o fnon-farm employment opportunities, but it was the non-poor who benefited most from the best opportunities. These changes in employment and income sources mark a deep transformation in the structure o f rural markets and the rural economy duringthe decade, which i s analyzedindetail in chapter 3. Transfers International migration proved an essential coping mechanism for poor rural households. We have estimated the value o f remittances from abroad reaching rural areas,16 which passed from MxP 5,085 million in 1992 to MxP 12,534 million in 2002 in constant 1993 MxP (see Figure 2.7). The main increase started in 1996 reflecting no doubt the acceleration o f rural migration to the USA that took place since 1995-6. As a percentage o f family income in disperse rural areas, remittances passed from 2.7 percent in 1992to 5.9 percent in2002. l3Calculated from INEGI's National Accounts by dividing the sum of all salaries and wages paid in the agricultural sector by the number of all paid workers occupied in agriculture, and deflating with the National Consumer Price Index. l4E M ! figures before 1995 are not strictly comparable with those for 1995 and after. l5Calculated from ENIGH (see Chapter 3). l6Remittance figures from the Banco de Mexico were transformed into MxP using the average exchange rate for the year, and then deflated using the national consumer price index. Of the total remittances inMxP thus calculated a portion was allocated to rural areas according to the percentage of rural remittances in total remittances given by ENIGH for the corresponding year. Percentages for odd years, for which there are no ENIGH surveys, were estimated by interpolation. 19 Figure 2.7: Estimate of Remittances from Abroad to Mexico RuralAreas inMillion MxP at 1993 Prices, 1988-2002 8,000.0 6,000.0 4,000.0 2,000.0 I 1988 1990 1992 1994 1996 1998 2000 2002 Source: WB staff calculations based on Bank o f Mexico remittance figures and the division o f remittances betweenrural and urban areas from ENIGH. Private domestic transfers to rural families did not increase much during the period, its share in disperserural areas passingfrom 4.1 percent of family income in 1992 to 4.4 percent in 2002. The change, however, i s bigger if a wide definition o f rural is used, particularly in certain regions (Goyo, Centro-Norte and Centro),where private transfers reached around 8 percent o f income in 2002. Private domestic transfers are also more important to the income o f the rural poor than to that o f the non-poor. Improved targeting of direct public transfers have partly compensatedfor the fall in rural incomes. Federal public expending in rural areas did not increase in real terms between 1995 and 2003. After falling in 1995-99, the trend and composition changed (see chapter 4). Expenditure in social and labor programs, which decreased in real terms in 1995-99, then increased until 2002, while productive expenditure, which remained constant in 1998-02, increased in 2003 and were budgetedto increase again in2004. Of particular importance for rural poverty have been the Procampo and Progresa/Oportunidadesprograms, which account for the bulk of direct public transfers to rural areas. Altogether, direct public transfers accounted for only 0.2 percent o f family income in disperse rural areas in 1992, when Procampo and Progresa/Oportunidadeswere not yet in operation. In 2002, they accounted for 6.0 percent o f family incomes in disperse rural areas and 16.7 percent for the bottom quintile o fthe rural income distribution. Progresa alone accounted for 11.9 percent o f the income o f the extreme rural poor (see chapter 3). When public and private transfers are added together, they amount to 16.5 percent o f income in disperse rural areas in2002. The proportion i s larger for the bottom quintile o fthe rural population, 23.8 percent, and for the extreme poor, 25.4 percent (see chapter 3). Since only 84 percent o f the income of the extreme poor was cash income,17 direct transfers accounted for 30 percent o f the cash income o f the extreme poor. The large majority o f transfers were not statutory since pensions and other statutory transfers are very small inrural areas. l7The remaining 16percent is self-consumption, imputedhouse rent, and in-kindgifts and donations. 20 A PROFILE OF THE RURALPOOR Who are the rural poor, where do they live, and where do they work? Comparing poverty levels for different categories allows us to examine which population groups are falling behind or are catching up. In the first part of this section we trace the evolution of poverty incidence for various groups during 1992-2002, examining the characteristics o f the rural poor and their evolution in the decade. In the second part we discuss the results from an econometric analysis to study more formally the correlates o f poverty in rural areas. By no means i s the poverty analysis below complete, as many important aspects o f poverty beyond conventional income and non-income indicators are missing (Box 2.2), including safety, voice and participation, as well as the role o f ethnicity in determining poverty (the latter i s discussed in chapter 7). Box 2.2. Other dimensionsof poverty. The poverty analysis provided in this chapter lacks important information identified by the poor themselves in rural Mexico, like safety, peace of mind, good health, sustainable environment, belonging to a community, and freedom o f choice and action. In particular, crime, violence and safety are flagged as important problems and obstacles to well-being in poor communities. Unfortunately, ENIGH surveys do not contain information on these variables or other important ones for poverty analysis like land assets or ethnic background. Other studies on poverty incorporate other variables than those used here. Thus, Finan, Sadoulet, and de Janvry (2002) measure the poverty reduction potential o f land in rural Mexico using 1997 ejido data. They show that for small landholders, an additional hectare o f land increases welfare by 1.3 times the earnings o f an agricultural worker. The marginal welfare value o f land depends much on access to complementary assets and on the context where assets are used. For non-indigenous small farmers with at least primary education and access to a road, the welfare benefit o f additional land i s seven times higher than for small farmers without these attributes. Ethnicity lowers the marginal value o f land, whereas education increases it. Households facing lower transactions costs, measured by access to roads, obtain a return to land two to three times higher than those without access to roads. These findings suggest that land can indeedbe an important element in a poverty reduction strategy, but there are specific conditions that must hold for this to be the case. As mentioned, the ENIGH data set does not contain information on ethnic background. Other studies on poverty inLatin America have shown that ethnicity is an important factor related to poverty, controlling for other characteristics. On-going work at the World Bank on ethnicity and poverty inLatinAmerica, should shedlight on this issue. Poverty Characteristics Table 2.6 presents the poverty profiles for 1992 and 2002 for disperse rural areas, semi-urban areas, and geographical regions. Before commenting on the data we must call attention here again to the fact that the ENIGH survey i s not designed to be representative at the highlydisaggregate levels ofTable 2.6. The results inthis table musthence betaken as indicative only. Time trends and the differences evidenced among population categories are probably correct, but the point estimates provided should not be taken as reflecting true population values. With this caveat inmind, the following poverty characteristics call our attention: 21 Table 2.6: Characteristicsofthe ExtremePoor in 1992 and 2002 1992 2002 Disperse Semi- Disperse Semi- Characteristics Rural Urban Rural Urban Gender (% who are extreme poor) Male 38.2 14.9 36.8 17.0 Female 36.6 4.0 32.1 11.6 Age Cohort (%who are extreme poor) 15 to 25 27.4 16.3 30.5 6.2 26 to 40 44.5 18.1 42.7 16.9 41 to60 37.7 10.3 36.0 15.4 >6 1 31.4 8.2 30.0 15.4 Education (%who are extreme poor) N o education 46.2 25.2 45.0 28.1 Primary incomplete 38.0 11.4 36.2 21.1 Primary complete 31.0 15.0 36.1 13.7 Secondary complete 22.0 7.0 17.9 8.9 Highereducation 0.7 0.4 0.5 0.6 Labor Status (%who are extreme poor) Employed 38.5 14.2 37.1 16.5 Unemployed 56.6 39.5 20.4 18.8 Inactive 32.7 6.4 29.4 12.0 Work Position(%who are extreme poor) Salariedworker 39.3 13.7 35.7 14.3 Self-employed 42.6 14.5 42.2 20.6 Employer 21.3 16.6 17.2 12.3 Family worker 36.3 39.3 Work Sector (%who are extremepoor) Agriculture 43.7 21.3 45.3 36.0 Extraction 26.2 9.8 7.7 1.1 Manufacturing 24.3 6.4 22.1 7.5 Construction 40.3 16.5 35.8 21.6 Utilities 7.7 24.3 Sales 25.5 12.0 23.2 7.0 Hotel-restaurant 1.9 0.2 15.4 6.6 Services 23.6 14.6 23.1 14.1 Education 1.4 0.9 1.8 1.1 Government 23.6 0.1 25.1 2.3 Source: WB staff calculations based on ENIGH 1992 and 2002 Education levels are strongly related to poverty. In rural Mexico, as elsewhere, poverty rates fall with educational attainment. Thus, in 2002, extreme poverty in disperse rural areas was estimated at 45 percent for household whose heads had no education. Extreme poverty fell slightly duringthe decade for households whose head had none or little education in disperse rural areas, whereas it increased slightly for those in semi-urban areas. There is an important premium to having some education, comparedto none, and to having secondary education compared to lower levels. While there i s a large difference in 22 disperse rural areas in the incidence o f extreme poverty betweenhousehold whose heads have no education (45 percent) and those who have incomplete primary education (36 percent), the difference i s very small between household whose heads have incomplete and those who have complete primary education. Secondary education makes a big difference; only 18 percent o f household whose heads have completed secondary education are poor. Extreme poverty for households with heads who are high school graduates decreasedrapidly during 1992-2002, down from 22 percent in 1992. These findings indicate that while education i s a crucial element for poverty reduction in rural Mexico, it i s not a silver bullet, as the more educated also experience poverty, and proportionally more now than a decade ago inthe case o fprimary education. 18 Very young household heads in semi-urban areas are far less likely to be poor than those in disperse rural areas and the difference has increased over time. Of the households headed by a person o f less than 25, six percent are extremely poor in semi-urban areas compared to 31 percent in disperse rural areas. In fact, in semi-urban areas, poverty rates are markedly lower for these households compared to those headed by people above age 25. Furthermore, while poverty among young households has fallen in semi-urban areas, it has increased in disperse rural ones. Instead, poverty has increased among households headed by elderly persons in semi-urban areas and stagnated in rural disperse areas. Fifteen percent o f the households headed by elderly persons (of more than 61 years) were below the food poverty line in 2002 in semi-urban areas-an increase o f 7 percentage points from 1992. In disperse rural areas poverty among this group has stabilized at around 30 percent during 1992-2002. Poverty incidence in households with heads in age groups o f 25 to 40 and 41 to 60 were stable at around 43 and 36 percent over the decade in disperse rural areas, with an increase inpoverty for the latter group in semi-urban areas. Thus, the life-cycle profile o f poverty indicates strong poverty incidence for households in age brackets where children are young and the family i s expanding, which decreases as the children grow up, and decreases again as the size o f the family becomes smaller with old age (Figure2.8). Female-headed households have lower incidence of poverty than male-headed ones in both semi-urban and disperse rural areas. The incidence of poverty among female-headed households decreased in disperse rural areas during the decade, but increased considerably in semi-urban ones. It i s risky, however, to use information on household heads as the basis for wider gender-poverty analysis, since income poverty figures are only part o fthe myriad o f factors that affect a poor woman's well being." 18Skill-biased technical change, changes in the relative supply of and demand for workers with different characteristics, and trade liberalization, have all been mentioned as possible explanations for the changes taking place in the impact of education on poverty in Mexico and Latin America in general (see Blom and Velez, 2001, and Blom, Pavcnik, and Schady, 2001). We will come back to the impact of different levels of education when discussing rural wages and non-farm occupations inthe chapter 3. l9Gender-poverty analysis based on gender of household head is fraught with difficulties. First, female- headed households are often underreported if migrant spouses are still identified as the non-resident head. Second, it is impossible to separate out individual consumption levels of men and women (and boys and girls) inhouseholds headed by either men or women. Yet, women, and perhaps especially girls tend to be disadvantaged in intra-household distribution of goods or investments in human capital. (see e.g. Quisumbing and Maluccio, 1999 and Case and Deaton, 2003) and continue to lag behind in terms of many social indicators. Non-income aspects of poverty, includinglevel of and access to human capital, access to labor markets, health status, etc., are therefore required to properly analyze gender-poverty links. 23 Households with heads employed in agriculture suffer far more incidence of poverty. Thus, the poverty incidence was 36 and 45 percent in agricultural households in semi- urban and disperse rural areas in 2002, compared to 8 and 22 percent in manufacturing, and 14 and 23 percent in service occupations (excluding sales and hotelh-estaurants where it i s much lower). Poverty among households with the head employed in agriculture increased by an astounding 15 percentage points in the last decade in semi-urban areas, while in disperse rural areas the increase was only o f 1.6 percentage point. Figure2.8. Mexico: Povertyincidenceby age of householdhead,in disperse rural areas (left) and semi-urban areas (right), 1992 and2002. 50 45 45 40 40 35 35 30 30 25 25 E x) 5 0 Et025 %to40 Age 41to60 Eto 25 26to40 Age 41to60 Source: WB staff calculations based on ENIGH. Poverty Correlates The poverty profile is corroborated by the analysisof conditionalpoverty correlates. Simple correlations can be misleading since it can be difficult to distinguishthe influence of the individual characteristics. Probit regressions for 1992, 1996 and 2002 provide information on conditional correlation between poverty and the characteristics o f household heads. By running regressions for these three years we also obtain information on the volatility o f the impact o f the attributes on the likelihood that a household experience poverty during the beginning and mid- 1990s and the beginning o f the 2000s. Finally, we obtain information on the groups that are particularly vulnerable, as well as changes inthese groups over the decade. A detailed analysis o f the results together with the probit regression coefficients are provided in Annex 2.A. The main statistically significant results for 2002 and the main trends between 1992 and 2002 are as follows: Households in disperse rural areas were more likely to be poor than those in semi-urban areas. Formal sector workers in rural areas, Le. those contributing to the social security system, were much less likely to be poor than their informal sector counterparts, reflecting an increase in the formal-informal gap since 1992. In contrast to earlier years, households with heads inactive in the labor market were more likely to experience poverty than other households. Surprisingly, among the households with active heads the probability o f being poor did not increase with 24 unemployment. This holds in both 1992 and 2002, but not in 1996, after the crises, when unemployed headswere more likelyto experience poverty than employed ones. Employers are the group with lowest probability of being poor followed by self- employed and salaried workers. In 1992, households with self-employed heads were 4 percentage points less likely to experience poverty than those headed by salaried workers inthe off-farm sector. Household with self-employed heads had the same probability o f being poor than those headed by non-agriculture salaried workers. During 1992-2002, households with heads self-employed in agriculture experienced a dramatic increase in the likelihood o f experiencing poverty. Households headed by salaried workers in agriculture also saw their probability o f being poor increased compared to households in the non-farm sector. A second income -whether from spouses engaged in the off-farm sector or from the household head taking a second job- had an increasingly important effect on reducing the probability of poverty for households. We discuss this more in chapter 3. Educational attainment is the single strongest correlate of poverty among the variables included in the analysis, controlling for other variables. The positive effect o f education on poverty reduction increases with the level o f completed education o f the household head and the spouse. The probability o f falling into poverty o f secondary school graduates decreased between 1992 and 1996 and increased between 1996 and 2002, reaching in 2002 a level not much different from that o f 1992. Having a female householdhead reducesthe probability of poverty. This was not the case in 1992, however, and this finding i s different in some other countries, for example Brazil, where male headed households have been found to have a lower probability o f beingpoor (see Elberset al, 2001), or inEcuador where there seems to be no association betweenpoverty and gender o f the household head (World Bank, 2004~). The dependency ratio of the householdis also an important factor. The presence o f children or youth in the household makes it more poverty prone, but the probability o f being poor falls with increasing child age. Households with children under 5 are more likely to bepoor than childless families, andtheir higherprobability o f being poor has been rather constant over the past decade. Households with children between6 and 11years have also higher probability o f beingpoor than those without children, although the likelihood i s lower than that for families with smaller children, and the probability has increased since 1992. Households with members aged 19 to 25 were significantly less likely to be poor than households with no children. The fact that young members enter the labor market and bring home an income contributes positively to the household's poverty situation. Inthe Sur region, however, households with young members did not experience the same lower probability o f beingpoor in 2002. Finally, the presence o f older members (above 65 years o f age) in the household makes it more poverty prone. In 2002, households with members of old age experienced a higher likelihood of poverty than those without them but the magnitude was lower than in 1992: 0.3 percent in 2002 comparedto 5 percent in 1992. 25 POLICY IMPLICATIONS What policy conclusions can be derivedfrom the above findings? The aggregate and rather general type o f analysis carried out in this chapter does not allow inferringdetailed policy options. We can suggest, however, some general orientations likely to influence poverty positively. More detailed suggestions are made inother chapters. The first policyimplicationrefers to the importanceof maintainingmacroeconomic stability. The effect on poverty of the 1995 crisis confirmed what was already known from the 1982-83 crisis and from the experience o f other countries: that strong macroeconomic shocks can undo in one or two years longtime improvements on poverty reduction, and that poverty levels take time to recover after the shocks. Macroeconomic stability i s hence a necessary element o f any poverty reduction strategy. Another implication refers to the importance of cash transfers for the very poor, since one quarter of their income comes from this source. Transfers have become indispensable for many rural poor families who could not survive without them. Public transfers, particularly those associated with programs like Procampo and Progresa/Oportunidades, are an important part o f all transfers received by the poor. There are evident dangers in a rural economy and society which depends always more on transfer incomes, and there i s need o f revitalizing both agriculture and non-agriculture sources o f employment and income inrural areas. Butpublic transfers to the income-poor inrural areas cannot be discontinued without much sufferingand the probable reversal o f the poverty reduction trend o f the last years. It i s at the margin, however, where policy decisions are usually made. And the decision in this case i s whether to increase at the margin public expenditure in direct cash transfer programs or in employment and income generation ones. Our view i s in favor o f the latter option, butwithout reducing the current level o f cash transfers to the poor, or not at least before new income opportunities become available. This view, however, is conditional on substantially strengthening the efficiency o f economically- oriented programs, which as we will see in chapter 6 could be improved in design and implementation. The decision o f what to do with Procampo when it officially comes to an end in 2008 i s relatedto this policy issue. It is important to focus actions on areas of concentration of poverty to have maximumimpact.The fact that ruralpoverty is particularly prevalent and acute insome regions, in disperse rural areas, and, as indicated in chapter 1, in an conglomerate of refuge zones, has important policy implications. Focusing actions on poor areas i s already the rule in some programs where the allocation o f funds or the selection o f areas i s relatedto the marginality index or other poverty indicators. This i s a good practice which could be extended to other programs. Focusing on poor areas could be combined with an assessment o f the capacity o f specific programs to promote local development there. Different zones have different potentials, and different programs have different capacities to promote local development in those zones and may do it in differentways. This poses trade-offs between the depth o f needs existing in different zones and the cost-efficiency o f programs to attend them under the specific conditions o f the zones. Population in disperse rural areas present a particular challenge because their needs are usually greatest butthe cost-efficiency o f programs tends to be low inthose areas. A good principle in connection to these trade-offs is to concentrate investments in particular localities -inthe way for instance envisaged by the Microrregiones strategy- so as to build up critical masses of infrastructure, services and production-oriented programs to help triggering endogenous processes of local development. The best instrument to priorize investmentslocally and identify strategic areas for local economic growth is that o f the territorial 26 approach to rural development and the use o f participatory planning methods. We come back to this inother chapters ofthe study. Households with children of young age requires special policy attention. Oportunidades offers already a mechanism to support these households, but other mechanisms could also be considered. Given the importance o f secondary income sources, subsidizing rural child care facilities to allow mothers o f young children to participate inthe labor market or carry out on-farm productive work are potential areas for policy action. Some programs operating in rural areas like, for instance, Jornaleros Agricolas, operated by SEDESOL,already include some support o f this type. A problem discussed in other chapters is the young households lack of access to assets that would allow them to undertake independent economic activities. There is in particular very limited or no access to land, technical assistance, training, and loans for farm investments and to start up on -and off-farm economic ventures. Focusing on young households and supporting them to develop productive activities emerges hence as another policy priority. The Secretaria de la Reforma Agraria is starting a program to facilitate the access of young landless workers to land and complementary assets, which i s a welcome step in the mentioned direction. Educational improvements should be part of any effort to reduce poverty in rural areas, for lack o f education gives individuals less opportunities, reduces the returnto their labor, and hence tends to make them poorer. An advantage o f education i s that being a "portable asset" it accompanies the individual ifhe or she decide to migrate. Since progress over the last decades has made primary education now practically universal among young citizens in Mexico rural areas, the focus should be on improving quality, which i s lagging behind, expanding secondary education facilities and enrolment, and strengthening different types o f technical education and vocational training inrural areas relatedbothto farm and non-farm economic activities. 27 3. ACTIVITIES, EMPLOYMENTAND INCOMESOF THE RURAL POOR2' This chapter sheds light on the changes that have taken place in employment and income generation in the rural areas of Mexico. The mainfindings of the chapter are as follows: The characteristics of the rural labor force are changing, with greater participation of women, higher levels o f education, and considerable ageing o f the workforce. Rural employment has fallen, which points to limited employment opportunities and the impact o f rural out-migration. Also, formal employment i s not expanding in rural areas. Informal salaried employment has increased at the expense of unpaid workers, as has ruralnon-farm (RNF) employment at the expense o f farm employment. Rural wages, which fell with the 1995 crisis, had on average just recovered in 2003, butthe average agricultural wage was still below the 1995 level inrealterms. Wage levels are largely determined by education levels, but also by gender, experience/age, level o f formality, and location. The importance of agriculture in rural incomes, including independent farming as well as agricultural wage labor, fell significantly between 1992 and 2002, giving way to rural non-farm occupations. The poor remain more dependent on agricultural income than the non-poor, however, and their access to rural non-farm activities, especially high-return ones, i s more limited. Instead, transfers, especially public transfers from Progresa and Procampo, have developed into a major source o f income for the poor and appear to have lowered rural inequality. Rural non-farm occupations appear to play a key role for sustaining rural incomes, especially for the moderately poor. Access to high-returnRNF activities i s more limited for women and for workers with little education, located in disperse rural areas, and belonging to an indigenous culture. The mainpolicy implications deriving from the above findings are: 2oDetailed statistical tables and other relevant materials are presented inAnnexes 3A to 31. 28 There i s need for a comprehensive view of rural development which include both farm andnon-farm activities. Rural policy may be more effective if focused on the family rather than the farm, moving away from the viable farm concept in order to promote competitiveness through multiple interventions embracing small and large farms, full-andpart-time farmers. Education -both access and quality, and including vocational and technical training - together with improved rural infrastructure, microfinance schemes, and technical and management support systems are all key policy areas for raising rural productivity and increasing access to higher wages and RNF activities. A spatial policy could facilitate the geographical concentration of investments and services for productive development, favor the growth o f rural towns and intermediate cities, and encourage the establishment o f links between these urban centers and their ruralhinterlands. Economic investments to promote the RNF sector could be included in rural development program. This could be done by decentralizing investment decisions through a system o f local participatory planning based on a territorial approach to rural development. EMPLOYMENTAND THE RURAL LABORFORCE Demographic Trends In 2000, 24.8 million of Mexico's 97.5 million people lived in disperse rural areas (Table 3.1). Expanding the definition o f rural to locations with less than 15,000 inhabitants increases this figure to 38.1 million. The national average conceals significant regional difference; however, rural-urban patterns vary throughout the country. For example, in the Sur, 47 percent o f the population lives in disperse rural areas compared to only 12 percent in the Norte. The rural population share has fallen over time, but the rural population is still growing. Between 1990 and 2000, population growth in disperse rural areas was lower than the national average: 0.6 percent annually compared to 1.8 for Mexico as a whole. This was the result o f out-migration since fertility rates in rural areas, though falling, still reached around 4.3 in 1995-2000, compared to 3.1 for cities over one million (Partida Bush, 2004). Thus, in 2000, 25 percent o f Table 3.1. Mexico: Total Population, Ruraland Urban Shares, and Annual Growth Rates, 1990,1995 and 2000 29 Annual Region 1990 1995 2000 Growth(%) Typeof Area 1990- 2000 Mexico Total 81,249,645 91,158,290 97,483,412 1.8 Urban (%) 71.3 73.5 74.6 2.3 Rural (%) 28.7 26.5 25.4 0.6 Norte Total 13,246,99 1 15,242,430 16,642,676 2.3 Urban (%) 84.7 86.5 87.9 2.8 Rural (%) 15.3 13.5 12.1 -0.1 Capital Total 18,05 1,539 20,196,971 21,701,925 1.8 Urban (%) 91.4 91.5 91.6 1.9 Rural (%) 8.6 8.5 8.4 1.5 Golfo Total 10,121,385 11,388,767 12,024,666 1.7 Urban (%) 59.9 62.5 63.7 2.4 Rural (%) 40.1 37.5 36.3 0.7 Pacific0 Total 9,O 77,660 10,177,075 10,745,699 1.7 Urban (%) 75.7 77.4 78.7 2.1 Rural (%) 24.3 22.6 21.3 0.3 Sur Total 12,398,892 13,600,852 14,424,973 1.5 Urban (%) 48.7 52.0 52.9 2.4 Rural (%) 51.3 48.0 47.1 0.6 Centro-Norte Total 10,382,375 11,488,771 12,113,254 1.5 Urban (%) 59.4 62.9 64.8 2.5 Rural (%) 40.6 37.1 35.2 0.1 Centro Total 7,970,803 9,063,424 9,830,219 2.1 Urban (%) 64.0 66.5 67.7 2.8 Rural (%) 36.0 33.5 32.3 Note: Rural defined as locations o f2,500 residents, urban as locations o f 2,500 residents. Source: WB staff calculations based on INEGI. Mexicans lived in disperse rural areas, down from 29 percent one decade earlier and 34 percent two decades earlier. Not all regions followed the same pattern, although in all regions the urban population grew more. Inthe Norte the disperse ruralpopulation actually diminishedby 0.1 percent annually during 1990-2000. In the Capital region, the difference in population growth between disperse rural and urban areas was small, with the rural population expanding at 1.5 percent annually. Population growth in the Sur and Golfo regions followed closely the national average. The demographic turning point has not yet been reached in rural Mexico: the disperse ruralpopulation i s only expectedto stabilize (at around26.8 million) by 2020 (CONAPO, 2004). Changing Characteristicsof theRuralLabor Force The Mexican rural labor force is becoming more feminized, better educated, older, and less dependent on unpaid employment. These trends are summarized in Table 3.2 below and shown in more detail in Annex Tables 3.A.1, 3.A.2 and 3.A.3 of Annex 3.A, based on ENIGH, andAnnex Table 3.B.1. ofAnnex 3.B, basedon ENE. The participation rate of women in the rural labor force increased 45 percent between 1992 and 2002, from a rate of 21.8 percent to one of 31.6 percent. Male migration, lower fertility rates, and the increased importance of non-farm occupations, where women are strongly represented, would seem to be the main reasons for this. The participation of family and 30 unpaid family workers decreased 3.5 points and that o f employers another 3.3 points. The fall in the position o f women as unpaid family workers shown by ENEi s remarkable: from 39.5 percent in 1995to 25.0 percentin2003 (Annex Table 3.B.1). Table3.2: Mexico: Compositionofthe RuralLabor Force in 1992 and2002 (percentages) Concept 1992 2002 Gender Male 78.2 68.4 Female 21.8 31.6 Labor Status Salaried Worker 45.4 48.7 Self-employed 30.9 34.4 Employer 6.8 3.5 Familv and unnaidworker 16.9 13.4 Education Without or incomplete primary 61.8 51.3 Complete primary 26.7 27.3 Complete lower secondary 9.3 15.3 Upper secondary andhigher education 2.2 6.1 AgeLess than 15 4.7 3.5 15 to 25 31.6 25.4 26 to 40 31.7 30.0 41 to 60 24.0 28.1 61 or more 8.0 13.0 Source: WB staff calculations based on ENIGH. The improvements in education of both male and female workers are substantial. The average years o f education o f the 9.3 million people in the rural labor force increased from 4.4 years for both genders in 1995 to 5.1 and 5.4 years for males and females respectively in2003 (Annex Table 3.B.1). The increased level o f education between 1995 and 2003 reported by ENE i s shown in Figure 3.1. The percentage o f male and female workers with complete lower secondary education increased by 62.2 and 83.6 percent, respectively, in the period. The percentage o f workers with complete upper secondary, technical and higher education expanded from 3.6 to 5.3 percent inthe case o f males and from 8.8 to 9.8 percent inthat o f females. The labor force is rapidly ageingin rural Mexico, because of the combined effect of migration of workers in younger age cohorts and the increase in life expectancy. Working men are slightly older than working women; the average age was 37.9 for men and 36.3 for women in 2003, up from 34.8 and 33.6 in 1995 (Annex Table 3.B.1). Rural workers between 15 and 25 years decreased their participation by 6 points in 1992-2002, while mature workers of 41 to 60 and 61 or more increased their participation from 24 to 28 percent and from 8 to 13 percent, respectively, inthe same period(Table 3.2). 31 Figure3.1: Mexico: EducationAttainment ofthe RuralLabor Force : I 1 01995Male W1995 Female 1 1 02003Male W2003 Female I1 No Complete Eduation I Complete Lower sec Complete upper sec Complete Higher Education Technical Source: WB staff calculation based on ENE surveys. Rural Employment Rural employment has fallen slightly since the 1990s. From 1995 to 2003, Mexico's rural work force decreasedby 0.5 millionreaching around 9.3 million in 2003, o f which 9.2 were employed, 0.3 million down with respect to 1995 (Annex Table 3.C.1.). The main explanation for the shrinking work force seems to be the migration o f the younger age cohorts to urban areas or abroad, which also contributes to the higher age o f workers.21 Notwithstanding their increasing participation, women are still a minor part o f the rural work force; according to ENE there were only 27.0 percentwomen inthe ruralwork force in2003.22 21Rural labor force defined here as individuals above 12years of age inareas<2,500 inhabitants. 22This figure is likely to under-represent the actual female share of the work force, because of the way the question is phrased inthe ENE questionnaire. Women participation in2002, calculated from ENIGH, is close to 32 percent (Table 3.2) 32 Figure3.2: RuralEmploymentby RegioninMexico, 1995-2003 + -c 1800- ;:ll&-/- 12001000 20001 norte capital 3000 pacific0 sur 2oooi 7 1995 1996 1997 1998 1999 2000 2001 2002 2003 1995 1996 1997 1998 1999 2000 2001 2002 2003 Year Graphs by region Source: WB staff calculations based on ENE 1995 to 2003, second quarter. Rural employment followed different paths in the regions as can be seen from Figure 3.2 and Annex Table 3.C.1. The Paczjico and Centro-Norte regions experienced job expansion o f around 200 to 400 thousand workers in 1995-2003, while employment decreased in the Centro, Golfo, Norte, and Sur regions, particularly in the latter region where nearly 400 thousandjobs were lost. This pattern seems to be partly due to the expanded production o f export crops in the Paczjico and Centro-Norte, which increased the demand for labor, with workers migrating from other parts o f Mexico to graspjob opportunities. 33 Agriculture is still the main employer, although its share has fallen since 1995. The distribution o f the working population in disperse rural areas by main sector o f primary occupation i s shown in Table 3.3, and a more detailed breakdown i s provided in Table 3.D.1 o f Annex 3.D. Of the working population o f rural Mexico, 56 percent was engaged in agricultural activities in 2003, the vast majority in cultivation. As shown in Annex Table 3.B.1, males had a much greater participation in agriculture than females; 66.9 o f rural males were in agriculture against only 25.0 percent o f females, who were more involved in services. Agricultural employment has fallen from 63 percent in 1995 to 56 percent in 2003 as the main field o f occupation declaredbyworkers. The share in total occupation of those who declare agricultureas their main area of work is much higher than the share of agriculturalincome in total income. We can think of three related explanations for this. The first i s that being agriculture a cyclical activity many farmers with agriculture as their principal occupation work also in other sectors during the low season. Second, independently o f the seasonality factor, there are many part-time farmers who complement agriculture with other activities on a permanent basis. Finally, there are rural dwellers that come from a farming tradition and continue to see themselves as agriculturalists, and hence declare farming as their principal occupation, even if they currently work little in agriculture and derivemost o f their income from transfers or other sources. Table 3.3. Mexico: RuralWork Force Sharesby PrimaryOccupation, 1995and2003 1995 2003 Urban Rural Urban Rural Agriculture 9.6 62.8 5.4 55.6 Mining & Extraction 0.3 0.7 0.4 0.1 Manufacturing 24.8 11.0 26.3 18.5 Commerce 21.2 11.7 21.5 10.0 Services 44.1 13.8 46.4 15.8 TotalNonAgriculture 90.4 37.2 94.6 44.4 Note: Rural work force defined as individuals above 12years o f age in areas < 2,500 inhabitants. Source: WB staff calculations based on ENE 1995 and 2003. Individuals o f 12 years and more. Formal rural salaried employment is small. In 2003, 8.1 percent of men were employed in the formal sector, and 12.5 percent o f women, many o f whom teachers and other government-paid staff.23Employers and the self-employedjointly amounted to 40.9 percent o f all men and 36.5 of all women employed in rural areas, while the informal salaried and contract workers jointly represented 31.3 percent o f men and 21.7 percent o f women, and unpaid family workers 16.0 and 25.0 percent, respectively (Annex Table 3.B.1). Duringthe period covered, the percentage o f informal salaried workers, both men and women, increased much, while that o f family workers decreased much. It i s interesting to notice the sharp decrease in women family workers and their larger involvement in self-employment activities. On the whole, trends during the periodtended to reduced gender differences inlabor status. 23The figures for formal employment are probably underestimated because the formal sector is here defined for the purpose of quantification according to whether workers contribute to social security. This means that all "formal" work (i.e. where labor is paid an agreed wage and where the worker's labor time and effort are controlled by the employer) inwhich labor laws are not respected is classified as informal. Given weakness of enforcement of labor regulations in Mexico, many workers who work under "formal" conditions are considered informal. 34 Falling labor force participation rates together with higher informality point to a lack of income opportunities in rural areas. The open unemployment rate i s very small, o f the order o f 0.7 percent, but under employment i s estimated to be high.24At the same time, employment has become increasingly informal over the 1990s, as self-employment and informal salaried work has increased (although unpaid family work has fallen). Moreover, while unemployment has fallen since the mid-l990s, the share o f inactive people o f working age has increased (Figures 3.3 and 3.4.). Together with increasing informality, this points to the lack o f rural income opportunities. Figure3.3. Mexico: InformalEmploymentand Self-employmentas a Share of Total EmploymentinRuralAreas, 1995 and 2003 ''6olI 70 1 W InformalSalaried 0Self-employed 50 I 18.7 13.9 ~ 20 36.3 29 10 0 Male 95 Male 03 Female 95 Female 03 Source: Annex 3.B, Table 3.B.1. Figure3.4. Mexico: Labor Force Participationand Unemployment ratesinRuralAreas Laborforce 37 T participation rate T 2.5 ,- 33 0.0 ~ ~ 1995 1996 1997 1998 1999 2000 2001 2002 2003 Source:Annex 3.C, Table 3.C.1. RURALWAGES AND WAGE CORRELATES Level and Trendsin Rural Wages Mean rural wages in all occupations fell since 1995 as a consequence of the 1995 Tequila Crisis, and did not start recovering until after 1999. Mean hourly wages by sector o f occupation in 1995 and 2003 are presented in Table 3.4., and the evolution by sector between 24Oxford Analytical, April 2004, estimates under-employment at 20 percent. 35 these two years i s shown in Figure 3.5. Recovery was slower in agriculture than in industry and services, particularly between 2000 and 2002 (Figure 3.5). As a consequence, the rural agricultural wage was 11percent smaller in real terms in 2003 than in 1995, while real wages in mining, extraction activities, manufacturing and services were slightlyhigher. The average rural wage rate remained stagnant. Hourly wages in the three sectors followed the same downward trend as unemployment from 1995-99, indicating little trade-off between unemployment and wages as adjustment mechanisms inthe rural labor market (Table 3.4).25 Figure 3.5: RuralAverage Hourly Wages by Sector (2002 Pesos), Mexico, 1995-2003 1 6 07 2 0 ~~ ~ t S ervices Source: WB staff calculations based on ENE 1995 to 2003, second quarter. Table 3.4. Mexico: Mean HourlyWagea by Occupation, 1995 and 2003 1995 2003 Urban Ruralb Urban Ruralb Agriculture 12.0 8.3 13.5 7.4 Mining & Extraction 18.6 12.7 36.6 16.6 Manufacturing 17.1 11.8 19.1 12.3 Commerce 16.6 10.8 17.2 10.5 Services 21.7 13.9 23.0 14.1 AVERAGE 18.5 9.8 20.3 9.7 Source: ENE 1995 and 2003; a. 2002 MxP per hour o f work; b. Localities < 2,500 residents. Median wages are higher in the non-agricultural sectors. Considering the whole distribution for the three sectors, median wages are higher in the industry and services sectors than in agriculture (Figure 3.6.). The righttail o f the distribution o f agricultural wages i s also less heavy than those o f industryand service wages, indicating that more people are beingpaid higher wages inthese sectors than inagriculture. 25Wage figures by gender according to labor status, sector of activity, and education are presented inTable 3.E.1. of Annex E, and wage figures for rural and urban areasbroken down by type of activity are provided inAnnex Table 3.D.1. 36 Figure3.6. RuralHourlyWage Distributionby Sector (2002 Pesos), Mexico, 2003 m 0 0 1 I I I I 0 3 0 60 90 1 2 0 Hourly W a g e -----------Agriculture ----- Industry Service s Source: WB staff calculations based on ENE2003,2nd quarter. Figure3.7: Average RuralHourly MaleWages (2002 pesos), byEducationAttainment,Mexico, 1995-2003 3 0 01 2 5 0 / ... 1 0 01 ------ \ +Lower Sec Complete Upper Sec Complete +i+igherEducation H +Technical Source: WB staff calculations based on ENE2003,2nd quarter. There are high returns to education. Unskilled workers with less than complete primary education receive an average hourly wage of MxP 8.2 compared to skilled workers with complete high school education who earn MxP 25.0 or more per hour (Annex Table 3.E.1). Hourly wages increase monotonically with the levels o f education, as can be seen in Figure 3.7. In2003, a male rural worker with complete higher education received on average a wage more than three times higher than a similar worker with no education or incomplete primary. Education, however, did not protect workers from wage falls during the Tequila Crisis. At all levels o f education, realwages o f rural workers fell between 1995 and 1999. 37 Wages are also related to labor status, with self-employed and informal salaried workers showing lower returns to their labor than employers and formal sector employees, two categories with similar returns (Annex Table 3.E.1). Figure 3.8. shows the hourly wage distribution of formal and informal sector rural workers in 2003. Formal workers are defined as those who contribute to social security, and therefore are protected. Median wages are higher in the formal sector and this remained unchanged during 1995-2003. Not only do formal sector workers receive benefits and other social services, but their hourly wages are also larger than those o f unprotected informal workers. Figure3.8: Distributionof RuralHourlyWages by Social SecurityCondition,(2002 Pesos), Mexico, 2003 "1 --_---------___ I I I I I 0 3 0 H o u rly6 0W a g e 9 0 1 2 0 I n f o r m a1 ----- F o r m a1 Source: WB staff calculations based on ENE2003,2nd quarter. Factors Explaining Rural Wages This section looks at factors correlated to rural wages and investigates the characteristics that differentiate low and high paid workers. We saw in chapter 2 the importance o f education and other personal and context characteristics as correlates o f poverty. We focus here instead on the impact o f these characteristics on rural wages. The results are o f course largely connected, since wages are an important component o f rural incomes. We compare workers located at different points in the wage distribution to analyze this issue, using quantile regression based on the ENE survey from 2003 (2"d quarter). Details o f this exercise are discussed in Annex 3.F, with the main results presented in Annex Table 3.F. 1. Wages are compared across workers grouped by gender, education, experience, labor status, and location. Findingsindicate that wages are by no means determined inthe same way for highand low paidworkers. For example, female workers are paidmuch less than males workers inthe high end ofthe wage distribution relative to their peers inthe low end of the distribution, and returns to lower levels o f education are far smaller in the upper income quantiles than in the lower ones. While the detailed results are discussed in Annex 3.F, the main findings are as follows: Having completed primary education contributesto better wages, and the premium increases rapidly with the level of education attainment. Better-educated individuals in rural Mexico earn much higher wages than their less-educated counterparts. In 2003, the association with the wage level of primary, lower secondary, upper secondary, tertiary, and technical education relative to no or incomplete primary educationwas positive at all quantiles, controlling 38 for other individual characteristics. Compared to the wages o f non-educated workers and those with incomplete primary, median wages of workers with complete tertiary education were 168 percent higher; the comparable premiumfor secondary schooling was 58 percent.26Workers with complete technical education received a 71 percent higher return compared to peers with no complete education. Returns across the wage distribution do not vary much for workers with complete upper secondary and tertiary education, Le. workers in the low end of the income distribution are not being paid comparatively less than their peers in the high end. This would seem to indicate that: (1) there i s no wide heterogeneity in the quality o f education in rural areas across the wage distribution, and (2) the capacity o f workers to convert their educational capital into higher earnings through labor market networks i s similar for poorer and richer workers. Hence, poor people with education seem to benefit from good labor market connections to the same degree as richer people. Workers with complete primary and lower secondary education face decreasing returns across the wage distribution, however: those at the low end are paid proportionally more than those at the high end, indicating that workers with the same level o f education are not compensated equally. The very poorest (1Oth quantile) receive a wage premiumfrom completing primary education o f 29 percent, while the rich (90thquantile) receive only a 19 percent premium. Inthe case oflower secondary schooling, workers inthe low end(1Oth quantile) obtain apremium o f 52 percent, while workers in the top end (90thquantile), obtain only 30 percent. One possible explanation i s that social networks that facilitate labor market connections operate better among the poorer than the richer segments o f the rural labor force. Another i s that these levels o f schooling are more relevant for employers hiring workers at the low than at the high end o f the wage distribution. General experience-here proxied by the age of the worker- increaseswages. We use two variables, age and age squared, to take into account possible non-linearities. We investigate two questions: (1) i s experience important to explain wages? and (2) are returns to experience homogeneous across the population? The answer i s yes to the first question and no to the second one. The impact o f experience on wages i s positive and increases untilworkers reach 49 years o f age. Thereafter, the returns fall in all quantiles. One explanation may be that older workers adapt less easily than younger ones to new technologies or they are simply less productive because o f their age. Returnsto experience tendto fall as we move upthe wage distribution, butthe variation i s not large. Workers inthe informal sector obtain a significantly lower pay after controlling for other variables. The negative impact o f informality increases across the wage distribution; a worker in the loth quantile has an 11 percent wage discount because o f informality, whereas a worker in the 75thquantile has a discount o f 16 percent. The informal sector generally provides lower quality jobs than the formal one. Since higher quality jobs may require more skills, the informal sector variable may be capturing skill differences not signaled by other variables included in the regression. The wage gap may also be due to lower productivity in the informal sector relative to the formal one not captured by education and experience. The labor status of workers is another important determinant of wages. Looking at the median o f the distribution, employers obtain the highest return: 66 percent, relative to the 26 The percentage return is calculated as (exp (coefficient estimate) - 1) 100. All figures presented in the following paragraphs are percentage premiums thus calculated from the marginal coefficients inTable 3.F.1. 39 group "other workers". For the richest (90th quantile), the premium gap i s even larger: 103 percent. The self-employed, informal salaried and contract workers are systematically worse off. Itis interestingthat inthe case of self-employed and contract workers the negative gap decreases sharply as we move upthe wage distribution. Thus, richer self-employed and contract workers are not as penalized with respect to "other workers" as poorer ones. The opposite i s the case with informal salariedworkers. Large inequalities persist in rural areas between men and women. Female wages are significantly differentfrom male wages at all quantiles. Results also suggest that the gender gap i s homogeneous across quantiles for women without children (married and single), but heterogeneous across quantiles for women with children (married and single). Married women with children experience the largest wage gap at the low end of the distribution; they obtain 36 percent lower wages than their male peers in the lothquantile, with the gap narrowing along the distribution to reach 28 percent at the 90thquantile. Workers in disperse rural areas are paid significantly less than workers in semi- urban rural areas, after controlling for other factors. The semi-urban -disperse rural wage gap i s significantly different from zero for all quantiles and varies across the wage distribution. It increases from the lothto the 50thquantile and declines from the 50thto the 90th.The semi-urban premiumis 12percent for the medianworker. Onepossible explanation for this gap is that prices, for example that o f urban land, are higher in semi-urban areas, and hence the higher wage i s a compensation for this - a reflection o f the fact that semi-urban workers have a labor supply curve above that o f disperse rural workers. Another possible explanation i s that work opportunities, i.e. labor demand, are higher insemi-urban areas, pushingup wages. All regions with the exception of the Golf0 and Sur enjoy a wage premium with respect to the Centro, and this is consistent across the whole distribution(except for the Sur). The Norte, Capital and Pacijko regions have the highest premiums. Workers in the Sur region have an advantage over their peers in the Centro (but not over those in other regions) in the bottompart o f the distribution, which they soon lose as we move up to higher quantiles. THE RURAL NON-FARM ECONOMY The GrowingImportance of theRural Non-Farm Economy The rural non-farm (RNF) sector is extremely important for income, employment and poverty reduction in Mexico. The significance of RNF activities is being increasingly recognized in the development literature and in applied programs.27Inthe traditional view o f the growth process, the sector i s expected to shrink with economic development, as urban manufacturedgoods and service provision substitute for low quality traditional local products and services in the clothing industry, shoe manufacturing, housewares, construction materials, local finance, local communication services and many others. With economic development, the countryside was supposed to specialize in primary activities where its comparative advantage lied. The opposite, however, has been the case. Today's advanced countries have highly diversified rural areas, with agriculture as only one o f many economic activities. Also, RNF incomes and employment have expanded rapidly in low and middle income countries. In Latin America, in the mid to late 199Os, RNF incomes accounted on average for some 40 percent o f rural incomes (Reardon, BerdeguC and Escobar, 2001). 27See Lanjouw and Lanjow (2001) and Reardon, Berdeguk and Escobar (2001) for two recent surveys. 40 Four reasons why the RNF sector should be given policy attention have been put forward by Lanjouw and Lanjow (2001): first, because o f its potential to absorb a growing rural labor force; second, because it can slow down rural-urban migration; third, because o f its contribution to national economic growth; and finally because it can promote a more equitable distribution o f income. Yet, it must be remembered that the RNF sector i s very complex and heterogeneous and offers very different opportunities depending on its structure (Box 3.1). Inthis section, we first examine available data on the importance o f RNF employment and income in Mexico, and move then to examine the correlates o f the participation o f Mexican rural household inRNFactivities, andthe type ofactivities inwhich they participate. Box 3.1. HeterogeneousRural Non-Farm Activities T h e RNF sector i s a complex and heterogeneous one. This i s logical given its negative definition, which embraces all economic activities taking place in rural areas different fiom agriculture. There are, hence, many different RNF occupations with varying productivities and returns to labor. There are also different barriers to entry to RNF activities in the form o f the education, skills, and financial or other assets that may be required. Easy to access activities, like petty commerce, are much more competitive and have much lower returns than better-protected ones. For empirical purposes, some authors distinguishtwo broad types o f activities, which we can refer to as "low return" and "high return" occupations (see for instance Lanjouw, 1999, for the case o f Ecuador, and Ferreira and Lanjouw, 2001, for Northeast Brazil). Poor households in poor areas are normally involved in "low return" RNF occupations, which are in a way "the equivalent o f `subsistence farming' -low productivity, low wage, unstable, with low growth potential" (Reardon, Berdegue and Escobar, 2001: 396). These occupations are a "rehge" for poor families with few or no farming assets. They do not offer a route to escape poverty but serve to complement income and make productive use o f little tradable family labor, thus alleviating poverty and diversifyingrisks. They are, hence, useful to the poor. In our analysis of ENIGH income data by occupations, we have distinguishedbetween low return and high return occupations, using the assets poverty line as the cut off. Occupations providing average earnings below the poverty line are classified as "low return", those above as "high return". EMPLOYMENT AND INCOME INTHE MEXICANRNFSECTOR RNF Employment RNF activitiesnow account for an important share of employmentinrural areas. In 2003, about 44 percent o f the rural working population declared non-agricultural activities as their primary source of employment. As already mentioned, these figures are likely to be highly conservative estimates o f the importance o f RNF activities because they do not take into account seasonality and do not consider secondary occupations. Also, the figures refer to a definition o f rural as disperse areas. If we include semi-urban areas, the share o f workers declaring non-farm activities as their primary occupation rises to 55 percent. Growth in RNF activities from 1995 to 2003 has been rather general affecting most occupations but in particular construction, food processing and clothing in manufacturing, and within the services sector personal services, hotels and restaurants, and education (Annext Table 3.D.1). There are however some exceptions (beverages, tobacco products, footware, printing, chemicals, plastic and rubber goods, metal goods, electronic goods, transport, communications, 41 and financial services), which are revealing because are occupations that tend to be o f relatively good quality and demanding in skills. RNF Income We should make a distinction between income coming from RNF occupations in the form of wages or entrepreneurial earnings, and income from public and private transfers. Both are non-farm incomes and both have increased much in the rural areas o f Mexico at the expense o f farm incomes. The implications, however, are different: while the former can indicate a certain dynamism o f the rural economy, the latter points to an expansion o f private and public social protection systems inrural areas. Table 3.5. IncomeShares inRuralMexicoa. 1992 and 2002 1992 2002 Income Shares All Extreme All Extreme Households Poor Households Poor IndependentFarming 38.5 38.1 12.6 16.8 AgriculturalWage Labor 12.3 19.6 11.3 21.9 Sub-totalAgriculture 50.8 57.7 23.8 38.7 IndependentNon-FarmActivities 8.1 4.8 5.7 6.8 Non-FarmWage Labor 20.4 15.9 36.1 17.2 Highreturn 4.9 1.3 23.8 4.4 Lowreturn 15.5 14.6 12.3 12.8 Transfers 8.0 6.0 16.5 25.4 Other Sources 12.6 15.5 17.8 11.9 Sub-totalNon-Agriculture 49.2 42.3 76.2 61.3 a Ruraldefinedas localities ofless than 2,500 residents. Source:WB staff calculations basedon ENIGH. We look first at the composition of rural incomes in 2002. Our results, derived from ENIGH data, are rather consistent with those obtained by other authors from other data sources, e.g. Taylor, Yunez-Naude and Ceron (2004), and de Janvry and Sadoulet (2001). The main findings can be summarized as follows: Independent farming has little importance for the average rural household but somewhat more importance for the extreme poor, amounting respectively to 12.6 and 16.8 percent o f income (Table 3.5.). While crop income and self-consumption are more relevantfor the poor, other farming activities (mostly livestock) are more relevant for the better off (Annex Table 3.G.6). Agricultural wage labor is a significant source of income (22 percent) for the extreme poor, second only inimportance to transfers (Annex Table 3.G.6) On average, agricultureaccounts for 23.8 percent ofincome,but its weight grows as we move down from the top to the bottom quintile where it accounts for 46.6 percent o f income (Annex Table 3.G.6). Income from RNF occupations is 41 percent of the total for the average rural household, but considerably less (24 percent) for the poor. RNF entrepreneurial incomes are small comparedto RNF employment incomes. 42 High returnoccupationsare a major source of incomein general, more than low returnones, butthe latter are ofmore significance to the poor. Wage income (from farm and non-farm occupations) accounts for 47.4 percent o f the income o f the average household and 39.1 percent o f the income o f the extreme poor. The figure for the extreme poor i s lower because o f their higher dependence on transfers and independent farm income, and because the figure for the average rural household i s inflated by the highwage income earned by comparatively richer households inRNFhighreturnoccupations. Transfers, both public and private, are a significant source of income in general in rural areas, and they are crucial for the poor, amounting to 25 percent of their earnings. Progresa and Procampo alone account for 15 percent o f the income o f the extreme ruralpoor. The significance of remittances grows as we move up the distribution ladder, whereas that o f other private transfers diminishes slightly. Ingeneral, remittances are more relevant for the non-poor while other private transfers are more relevant for the poor. Even among landowning households, the greater share of income is derived from non-agriculturalactivities. The 1994 and 1997 ejido surveys havebeenusedto calculate income shares by several authors. They have the advantage o f allowing the breaking down income composition figures by size o f family holding (reduced to rain-fed equivalents). In examining these figures we must bear in mindthat they refer only to ejido households, i.e. households who by definition have access to land, given to them by the agrarian reform. As seen in Table 3.6, even inthis sample o f land-endowed households, the share o f agricultural income for the average household i s less than half, and i s less than 30 percent for households with 5 hectare or 1ess.The importance o f agricultural incomes increases with the size o f holdings. Wage income accounts for more than 30 percent o f income for all households with less than 10 hectare, and its importance decreaseswith land size. Table 3.6. Mexico: Sources of Income in the Ejido Sector by F a r m Size, 1997 FarmSize" All <2 2-5 5-10 10-18 >18 TotalIncome in Pesos 25,953 12,474 17,314 28,368 30,564 44,255 Income Shares (%) Total FarmIncome 45.1 22.9 28.1 41.8 50.3 62.0 Total off farm income 54.9 77.1 71.9 58.2 49.7 38.0 Wages 24.6 40.3 36.9 30.4 18.2 11.1 Agricultural wages 4.8 10.0 7.5 4.2 5.7 1.2 Non-agricultural wages 19.9 30.3 29.4 26.2 12.5 9.9 Self-employment 9.4 17.1 14.2 4.6 12.1 6.8 Remittances 6.5 2.6 5.4 8.9 6.0 6.0 Other 14.4 17.1 15.3 14.3 13.3 14.1 Sourcee: D e Janvry and Sadoulet (2001: 469), based on the 1997 ejido survey. Size in rain-fed equivalent a hectares. As expected, the wider the definition of rural area, the lesser the importance of agriculturalincomes.We show this inFigure3.9. where the evolution ofthe share of agriculture intotal income from 1992 to 2002 is shown for our two definitions of rural. Both shares follow the same pattern, but that corresponding to a wider definition o f rural i s always below. In 2002, 43 the share o f agriculture using a wide definition o f rural was 17.1 percent (including both wage and independent farming incomes), 6.7 points below the share corresponding to the more restricted definition. Figure3.9. ParticipationofAgricultureinFamilyIncomeAccordingto the Definitionof Rural 60 ~ aJ c) 2 5 0 - c .* 2 4 0 - 30 ~ W O 20 gE 1 0 - 0 I I I I I 1992 1994 1996 1998 2000 2002 Rural<2,500 -w- Rural<15,000 Source: WB staff calculations based on ENIGH 1992-2002. Regional differences matter too (see Annex Tables 3.G.7 and 3.G.8). Thus, agriculture i s much more important in the Sur (41.2 percent) than in other regions, including both wage employment and independent farming, and less important in the Paczjico (12.4 percent). Domestic private transfers are similar across regions but remittances are much more important in the Sur and Centro-Norte. There i s a difference however between these two regions. The relevance o f remittances in the former region i s due to the highmigration to the USA o f Centro- Norte population. Instead, migration from the Sur to the USA i s not as large as from other regions, but the impact o f remittances on income i s significant in view o f the low incomes prevailing in that region. RNF enterprises are important in the Sur, Centro-Norte and Paczjico, and highreturnRNF occupations are particularly important inthe Paczjico. We look now at the trends in income shares between 1992 and 2002, which are shown in Table 3.6, Annex 3.G. and Figures 3.9 and 3.10. The main findings can be summarized as follows: A substantial increase in the weight of the RNF sector in relation to the farm sector. Thus, the share of agricultural income fell from 50.8 to 38.7 percent between 1992 and 2002 (43.9 to 23.8 percent for a wide definition o f rural). The increase was also significant but less marked for the ruralpoor. A substantial increase of wage incomes (farm and non-farm) relative to independent or entrepreneurial incomes (farm and non-farm). The share o f wage income increased from 32.7 in 1992 to 47.4 percent in2002 (34.8 to 48.2 percent for a wide definition o f rural). The increase was not as important for the poor because they participate little inthe opening ofhighreturnRNFwage earning opportunities. 44 A substantialincreaseof public and private transfers relative to earned incomes. The share o f all transfers increased from 8.0 to 16.5 percent (6.6 to 15.6 percent for a wide definition o f rural). The increase was particularly marked for the rural poor for whom transfers increased more than fourfold. A substantialincreaseinwage and salary earningsfromhigh return occupations relative to those from low return ones. The share o f earnings from high return RNF occupations increased from 4.9 to 23.8 percent. This expansion hardly reached the poor, however, for whom the share passed from 1.3 percent in 1992 to 4.4 percent in 2002.28 Figure3.10.: RuralIncomeCompositionby Quintiles inMexico, 1992-2002 100 Ho 0 0 q?rd q4h qiih arm Income1Income Off-farm Source: WB staff calculations based on ENIGH 1992 to 2002. Shares do not tell the whole story of income changes. We also need to look at absolute values. We do that in Table 3.7, where we present both absolute values and shares o f differentsources o frural family incomes broken down by consumption quintiles. We observe that while the share o f independent farming income falls rapidly as we move from the lstthe 5th to quintile, the value of income increases. Hence, on average, rich rural dwellers obtain more income from independent farming than poor ones, but independent farming earnings mean much less to them as a proportion o f all their operations. Agricultural wage labor incomes do not change a great deal across quintiles but the share falls rapidly as we move up the distribution ladder. Income from RNF low return occupations has similar importance for the poor and non- poor, being most important for the 3rdand 4th quintiles. The share, but not the absolute amount, falls to less that half for the 5thquintile. Finally, both the income and the share from high return RNF occupations grow much as we move to the upper quintiles but income grows more rapidly than the corresponding share. 28 Itis not possible from ENIGH data to establish if richer households have a higher percentage of high return RNF income because these occupations are monopolized by them or because access of poor households to these occupations made them able to graduate from poverty. Absence of suitable panel data prevents establishing this. Both things are probably true, but barriers to entry to high return occupations would lead to believe that it was the richer households who benefited most from them. 45 The development of the RNF sector opens economic opportunities to the rural population. But who grasps those opportunities, and do they have an equalizing effect or not?29Onthe basis o f the ejido sample, de Janvry and Sadoulet (2001) conclude that agriculture i s the main source o f inequality inthe ejidos, and that access to RNF employment has an equalizing impact. Reardon et a1(2000) suggest that the effect o f the RNF economy on equity depends much on the circumstances, andthat ingeneral inLatin America it has an equalizing effect, whereas the opposite i s the case in Africa. Araujo (2003: 3rd essay) using municipal-level manufacture and service data, along with poverty indicators drawn from the 1990 and 2000 population censuses, finds that rural manufacture and service development tended to decrease Mexico rural poverty in the 1990s. Table3.7. RuralFarmand Off-farmOccupationIncome by ConsumptionQuintileand PovertyCondition, Mexico, 2002 All ConsumptionQuintiles Extreme Non Househ. ~ ~ zndt t 3rd ~ 4th~ 5th Poor Poor TotalIncome (2002 MxP/year) 37,263 21,181 18,136 25,256 33,837 97,990 11,884 51,256 INDEPENDENTFARMING IncomeWxP) 4,695 2,068 2,684 3,536 4,331 10,779 1,997 6,048 Share (%) 12.6 18.5 14.8 14.0 12.8 11.0 16.8 11.8 Agriculture WageLabor Income Income (MxP) 4,211 3,142 4,026 5,177 5,177 3,724 2,603 4,921 Share (%) 11.3 28.1 22.2 20.5 15.3 3.8 21.9 9.6 RNF High Return WageIncome Income (MxP) 8,869 313 1,505 2,803 4,771 34,688 523 13,839 Share (%) 23.8 2.8 8.3 11.1 14.1 35.4 4.4 27.0 RNF Low Return WageIncome Income (MxP) 4,583 1,286 2,612 4,243 6,429 8,427 1,521 6,253 Share (%) 12.3 11.5 14.4 16.8 19.0 8.6 12.8 12.2 RNFEntrepreneurial Income Income (MxP) 2,124 481 1,088 1,919 2,267 4,900 808 2,819 Share (%) 5.7 4.3 6.0 7.6 6.7 5.0 6.8 5.5 Otherlncome Sources Income (MxP) 12,781 3,891 6,22 1 7,577 10,862 35,472 4,433 17,376 Share (%) 34.3 34.8 34.3 30.0 32.1 36.2 37.3 33.9 Source: WN staff calculations from ENIGH2002. What can we say on this subject in the light of thefigurespresented above? Rural poor families seem to have benefited from the opportunities opened by the RNF economy. We do not know, however, if these opportunities were taken up because of the fall in other sources o f income, thus substitutingfor them, or were an addition to these sources. If they were not additional, the conclusion is that they did not serve to reduce poverty. Looking, 29 To give a proper answer we would nee to carry out a counterfactual analysis of what would have happened to rural poverty and equity inthe absence of RNF growth or with a different type of BNF growth. This type of counterfactual analysis was carried out by Paes e Barros, de Carvalho, and Franco (2004) for Brazil, but we will not attempt it here. 46 however, at the figures and in the absence o f a detailed analysis, it i s reasonable to assume that there has been additionally, andthat the impact on poverty has therefore been favorable. High return RNF occupations seem to have been mostly taken up by the comparativelybetter off, although the poor have also participatedin them to some extent. Hence, it i s likely that the impact has not been equalizing, and the RNF sector has contributed in some measure to the worsening o f the rural income distribution. Public and private transfers are a different case; they have definitely helpedthe poor more than other groups, and have therefore had an equalizing impact. WhatExplains Participation in RuralNon-farm Employment? We have carried out an exercise on the basis of the ENE 2003, 2"dquarter survey, using a Probit model, to determine the probability of individual involvement in non-farm activities as primary occupation, conditional on a range of personal, household and geographical characteristics. Regression results are discussed in detail in Annex 3.H. Because o f limitations in the ENE survey, some important variables cannot be considered, including access to land, ethnicity, social networks, and physical infrastructure. The results from other studies using these variables are discussed inAnnex 3.H, however. The main findings are summarized below. Women have considerable higher probability than men to participate in RNF activities, controlling for all other variables. This result holds for married and single women, with and without children, with marginal effects that are not very different among these groups o f women. However, women have more limited access to high return occupations and remain confined mostly to low return ones. This does not change with marital status or having children. Age, a proxy for experience, increases the probability of employment in high- return non-agriculturaljobs. The association is negative for low return ones. N o evidence was found o f the association declining at a certain age. Since this result i s at odds with findings in other studies (see Annex 3.H), more research i s needed to understandthe age factor. Involvement in the non-farm sector is significantlyrelated to education levels. As education levels rise, so does the probabilityofbeing employed inlow return and highreturnoccupations. The exception is university and technical education, which, not surprisingly,diminishthe probability o f engagement in low return RNF activities butincreases that ofparticipating inhighreturnones.3o Workers living in disperse rural areas are less likely to be employed in RNF occupationsthanthose livinginsemi-urban areas. Workers in highpoverty regions are less likely to participatein RNF activities. Regional differences in poverty levels are mirrored in regional patterns for RNF activities. Relative to those living inthe Centro region, workers inthe Norte, Capital, Golfo, and Centro-Norte regions are more likely to be employed in RNF activities. Instead, workers in the Paczjico and Sur are less likely to participate in the RNF economy than their peers inthe Centro. 3o Itshould be acknowledged that the exogeneity of education inthese models can be questioned. 47 CONCLUSIONSAND POLICY IMPLICATIONS The rural labor market is undergoing a rapid transformation. Labor force characteristics are changing as a result o f better infrastructure, education and health services in rural areas, and a general process o f social modernization. Agriculture i s rapidly losing importance relative to the RNF sector. Wages are becoming the primary source o f income for rural dwellers relative to independent entrepreneurial incomes. Public and private transfers have increased dramatically, having become crucial to the survival o f the rural poor. Many new RNF occupations have become available in rural areas, although the best o f these opportunities can be hypothesizedto have been seized by the comparatively better Yet, productivity, employment, wages and income levels remain more or less stagnant. Inspite o f this structural transformation, the rural economy i s not sufficiently dynamic. There i s a cleavage betweenthe dynamism o f socio-demographic change in rural areas and their limited economic development. Rural employment has fallen by around half a million over the last eight years, notwithstanding population growth and higher women participation in the labor force -a worrying symptom o f the effects o f the migration process and the paucity o f employment opportunities. The latter i s also evidencedby the stagnation o f rural salaries over the same period. The rural poor, hence, increasingly rely on outside assistance in the form o f private and public cash transfers to cover many o ftheir basic needs, The demographicchanges and those inthe characteristicsof the labor force reflect a rural society in profound transformation -a view reinforced by the increasing extent o f migration, and the modification o f rural income composition with a substantial rise in non-farm earnings. We have to put these trends in the context o f the little encouraging figures on the incidence o f rural poverty, the modest performance o f agriculture, and the increase in transfer incomes. The picture that emerges i s one o f a rural society which experiences the impact o f social modernization and market exposure but has not yet found a firm way to sustainable economic development. A comprehensive vision of rural development beyond sectoral approaches is forming in Mexico. The ruralworld i s not an agricultural world, and the rural economy i s not an agricultural economy; indeed, the combination o f economic activities i s the dominant characteristic o f rural households and communities in contemporary Mexico. Fortunately, there i s a trend in Mexico, both in the public sphere and in civil society, in favor o f a more comprehensive view o f the rural world, evidenced for instance in the Ley de Desarrollo Rural Sustentable. The challenge i s how to transform this into full policy reality, including appropriate changes inthe institutional set up. Rural policy could benefit greatly by moving from the farm to the family as the unit of analysis for rural development and the receiver of rural policy interventions. Given the various occupations o f rural households and their combined sources o f income, the family economy becomes a clearer choice o f focus than the farm economy. This would also remove the 31 Itis worth repeating here the warning regarding causality problems. We do not know if better off households seized these opportunities or households lucky to profit from them became better off. Two considerations offer some comfort, however, for believing that the dominant causation runs from income position to high return RNF employment. The first is that variables like education, age, location, and ethnicity, positively correlated with high return RNF employment, are negatively correlated with poverty. The second is that it is reasonable to expect comparatively better off households to have access to assets that help overcoming the barriers to entry into highreturnoccupations. 48 dichotomy between "viable farms", typically the focus o f production-oriented programs, and "non-viable farms", generally supposed to be targeted by "social" policy and relief programs. Instead, more modern visions look at the continuum o f farms going from the very small to the large, recognize common interests and possibilities o f economic and corporate co-operation betweenfarms of differentsizes through contract systems, chain arrangements, joint ventures and other means, and acknowledge that small farmers can reap most o f the relevant economies o f scale from association and the use o f competitive rental markets for indivisible inputs. This approach helps placing economic factors and programs and not only "social" and relief programs at the core o f poverty reduction. Secondary education has been repeatedly found to be strongly linked to participation in the RNF economy, and also to enhance the income obtainedfrom a variety o f occupations. This i s one more reason to expand the coverage and quality o f secondary education inrural areas, inparticular for those who are falling behind, like indigenous groups and residents o f remote areas. Quality i s probably as important or more than coverage if we want to increase the impact o f education. Raising education i s not enough, however; it i s the synergy o f knowledge with other productive assets that raises incomes (see D e Ferranti et al, 2002). Infrastructure and location characteristics are other important correlates of RNF participation where policy can intervene.Roadconnections, communications, and energy have been shown to be important for the development o f the RNF economy, and o f manufactures in particular. Disperse rural areas are systematically associated with lower incomes and employment opportunities, and proximity to urban centers favors RNF development. This i s another reason why raising the low levels of investment in rural infrastructure (see chapter 4) could pay off. Population dispersion and proximity to urban centers are long-term undertakings from a policy perspective. But much could be gained from a spatial policy that favor the concentration o f investment and services, the development of rural towns and intermediate cities, and the establishment o f links between these and their rural hinterlands. The decision for instance of the Microrregiones strategy to concentrate investments in some localities to help them become centers o f local development, i s an encouraging step inthat direction. Rural poor workers need help to overcome barriers to entry to RNF occupations, especially lack of skills and financial assets. Vocational training i s a need which could receive more attention. There i s a tradition o f technical agricultural schools in Mexico, like the CBTAs, but there are not enough comprehensive medium-levelpolytechnic training centers inrural areas with a modern vision for the provision of vocational training. This is a field with important potential for policy intervention. The promotion o f rural finance i s recommended in several parts o f this study, which i s logical in view o f the multiple functions o f finance in rural development and poverty reduction. Micro-finance institutions have proven inmany countries their capacity to promote the growth o f micro and small enterprises and help the poor getting started in business. Promoting RNF activities i s hence one more reason to advocate for policies to expand the rural finance system and make it friendly to the poor. Finally, the inherent heterogeneityof the RNF sector would favor a policy approach focused on decentralization through territorial development. A territorial approach to rural development would simplify the implementation o f policies to stimulate the RNF economy and would make the specific programs more relevant to the local context. Like in the case o f rural finance, we advocate in several parts o f this study in favor o f a territorial approach to rural development. The reason i s that this approach i s a policy framework where the many facets o f rural development can best be tackled. We summarize in Annex 3.1 the characteristics o f the territorial approach. 49 Pursuing RNF growth should not be seen as an impediment or an alternative to pursuing agricultural development; there are strong synergies between the farm and non- farm sectors. In signaling the importance of the RNF sector and advocating more policy focus on the non-farm economy, we do not intendto diminishthe importance o f agricultural development. There i s no contradiction between the development o f the farm and non-farm sectors (see Lanjouw and Lanjouw, 2001, and Reardon, BerdeguC and Escobar, 2001). The synergies are examined in chapter 4. What i s needed i s a comprehensive rural development policy where farm and non-farm can find their place andtheir connections berecognized. 50 4. AGRICULTURE, POVERTY AND THE SMALL FARMER This chapter takes a closer look at the agricultural sector to examine what has been and what could be its role in reducing ruralpoverg. The mainfindings of the chapter can be summarized as follows: There i s evidence for the case of Mexico that agricultural growth i s pro-poor. The impact is stronger for the worst poverty conditions and is mostly confined to ruralpoverty. Agriculture experienced modest growth in the 1980s and 1990s of around 1.5 percent per year in each decade, below national growth and population growth. Outputof food crops grew more, at around 2.2 percent in 1980-02, mostly as a result o f some improvement in the yield o f individual crops and a change o f crop mix in favor o f higher value crops. Aggregate land expansion did not contribute to output growth due to the exhaustion of the crop land frontier. Agricultural growth was higher inthe northern states where agriculture is more commercial, in irrigated lands where commercial farming concentrates, and in the more commercial crops. Land and labor productivity rose in the 1990s at a rate above 2 percent, and total factor productivity also grew. However, by international standards, land and labor productivity are low in Mexico, and the gap to the nonagricultural sector i s high. Federal government expenditure in rural development is high by international standards. Social and productive development are the major expenditure areas, while infrastructure and environmental expending are small. Value added per hectare does not increase with farm size, suggesting that land is more productive insmall farms. Access to variable capital, which includes seeds, fertilizer and chemical inputs, i s the main factor explaining output inthe small farm sector, while labor contribution to output i s comparatively small at the margin. Credit restrictions; however, appear to preventsmall farmers from usingoptimum quantities of inputs. There i s no evidence of economies of scale in farm production but there i s evidence o f substantial inefficiency infarming. Farmers producing more commercial crops are comparatively more efficient than others. 51 The main policyimplications deriving from the abovefindings are: The exhaustionof the land frontier and the comparatively low levels ofland and labor productivity in agriculture point to agricultural intensification as the thrust o f agricultural policy to increase output and incomes. This puts focus on how to improve varieties, switch crops, increase yields and reduce the incidence o f natural shocks. Poorer farmers need special assistanceto move from low to high value crops. An important element would be the presence o f extensive and well functioning research and extension and rural finance systems. These systems are also essential to raise crop yields, which i s the other component o f intensification. Small farmers tend to face more market failures and need to have services like research, extension, and rural finance tailored to their needs. While federal government expenditure on agriculture is high, reflecting the importance traditionally given by Mexican governments to rural areas, its efficiency and effectiveness i s called into question by the weak historical performance o f the sector. THE ROLEOFAGRICULTURE INPOVERTYREDUCTIONINMEXICO Agricultural growth can affect poverty through several mechanisms: (i)higher output of poor farmers, (ii) wages for unskilled labor, (iii) higher indirect demand for rural non- farm activities, (iv) lower food prices, and (v) inter-industry linkages, both upstream, e.g. fertilizers and machines, and downstream, e.g. food-processing industries. Whether agricultural growth i s actually poverty reducing will depend on how and where growth takes place.32This i s discussed inBox 4.1. Box 4.1. How Can Agricultural Growth Help Alleviate Poverty? Agricultural growth can affect poverty in both urban and rural areas through a wage effect, a production effect, a RNF multipliereffect, a food price effect, and an inter-industry linkages effect (Lopez, 2002). The wage effect is due to the fact that, becauseofthe high labor mobility observedin many countries, agricultural labor markets are strong determinants o f the wage rate for unskilled labor throughout the economy. Agricultural development i s likely to raise agricultural wages, and this will translate into higher wages in other parts ofthe economy. The impact will dependon the distribution o f agricultural assets, the existence or not o f surplus labor in agriculture, and the way growth came about. If growth resulted from improvements in average labor productivity in the semi-subsistence sector, the wage rate effect may be strong. If, instead, growth originated in the plantation sector and was not accompanied by increased labor demand (or surplus agricultural labor was enough to meet the increased demand), the impact would be small or nil. We can even imagine a negative wage effect, for instance if agricultural growth was the result o f widespread labor saving innovations (see, e.g. Arrighi, 1967,who discusses the case o f colonial Rhodesia). 32The role of agriculture in economic development and the connection between agricultural growth and poverty reduction have been discussed for a long time inthe literature. See, for instance, Rao and Caballero (1990)for the first theme and Mellor and Desai (1985)for the second. A recent review of issues and literature can be seen inMellor (2000). 52 The direct production effect is simply the result of growth itself. Ifgrowth originated inthe small farm sector, through for instance irrigation, improvedtechnology, better prices or new crops, the outcome will be increasedby real incomes inthis sector with a direct effect on poverty. If,instead, larger farmerswere themainactors involved, thepoverty impactmaybe small or nil.It could be negative, if increased output generated by large farmers displaced from markets that o f poor farmers. The RNF multiplier effect consists of the impact of agricultural growth in the RNF economy (see e.g. Mellor, 1976). The relation refers not only to intermediate links, like the increase in farmers' demand for local transport and marketing services when production grows, but also to final consumption and investment demand, like the increase in the demand for entertainment services or housing or agricultural works when there are bumper crops or particularly good prices. To the extent that the rural poor are engaged in the production o f RNF goods and services, they will benefit from expanded demand. The impact, however, depends on whether the incremental demand concentrates on tradable or non-tradable items, on the existence o f spare capacity in the RNF economy, and on the degree and type o f involvement o f the poor in the RNF economy. Differences in these conditions have been shown to result in different outcomes (e.g. research on Indian villages by Foster and Rosenzweig, 2003) The food price effect results from the fall of food prices that may follow increased agricultural production. Lower food prices raise wages and other incomes inreal terms throughout the economy, thus reducing poverty. The effect, however, will depend on the openness of the economy, and on whether production increases center on tradable or non-tradable agricultural items. It will also depend on demand elasticities and on how badly hit are poor farmers by price falls. Falling prices could deteriorate poverty conditions in the rural sector if the increased output i s generated by rich farmers, who could collect higher revenues even with lower prices, while poor ones receive lower prices for their stagnant output. We finally have the inter-industry linkage effect, which refers to the incentive that agricultural growth generates in upstream and downstream-connected industries. The incentive operates via prices and quantities, and depends on how open i s the economy, how tradable are the sectors where incremental demand concentrates, ifthere is or not excess capacity in those sectors, and how good i s economic coordination along the value chains' The impact on poverty o f this effect will mostly work through the labor market, and will depend on what happens to labor demand and the wage rate o f unskilled workers when production expands in the industries stimulated by agricultural growth. Hence, although there is a good possibility that agricultural growth reduce rural and urban poverty, there i s no certainty that this will occur. In countries with a very unequal distribution o f agricultural assets, especially land ownership, agricultural growth is most likely to concentrate on the middle and large farming sectors without benefitto the poor who may even end up being worst off. Indeed, research suggests that an agriculture-biased growth pattern directly reduces poverty but indirectly raises inequality due to the unequal distribution o f land assets (De Janvry and Sadoulet, 1996).33In a dual agricultural economy the question o f which sector i s the main actor o f agricultural growth i s o f great importance for poverty reduction. The direct production effect and the wage effect will be bigger if the small farm sector i s the leading actor. The RNFmultipliereffect is also likelyto bebigger, for it is reasonable to assume that an increase in incomes would translate into larger local demand if it originated inthe small farm sector. The fall in food prices that may accompany expanded production would be less consequential to small producers ifthey were the ones to generate the incremental output. Inter-industrialeffects may not be very different if agricultural expansion is ledby small or large farmers except that if ledby the 33De Janvryand Sadoulet (1996)examine the evolution of Latin American countries between 1970 an 1994 showing that an agriculture-biased growth pattern directly reduces poverty but indirectly raises inequality due to the unequal distribution of land assets. This may lead to an increase inrural poverty as a combined result of both effects. 53 latter there may be more leakages to imports because of the type of technology they use and the markets inwhich they operate. The impact of agricultural growth on poverty has been empirically examined in Mexico by Soloaga and Torres (2003) in the framework of the FA0 ROA project.34 Following the approach used for Ravallion and Datt (1996) for India, and using ENIGH data for 1992 through 2000, the authors analyze the impact o f agricultural and non-agricultural growth on total, urban and rural poverty. They estimate poverty elasticities to growth using state- and region-level data for both the extreme and moderate poor as well as the impact o f both types o f growth on distributional equity inrural andurban areas. UsingENE data the authors also estimate the impact o f agricultural and non-agricultural growth on labor demand to see if there was some wage effect. They finally check for the existence o f a food price effect.35 Both agricultural and non-agricultural growth have a substantial and statistically significant impact on the reduction of total poverty, both extreme and moderate, but the effect o f agricultural growth i s stronger (Table 4.1). The impact on total poverty o f the two types o f growth occurs through their separate effects on urban and rural poverty, since the cross effects are not significant. Thus, agricultural growth promotes poverty reduction in rural areas while non- agricultural growth promotes poverty reduction in urban areas.36A one percent increase o f agricultural GDP decreases extreme rural poverty by 1.5 percent and moderate rural poverty by 0.8 percent. Similarly, an increase o f 1 percent in non-agricultural GDP reduces extreme urban poverty by 1.6 percent and moderate urban poverty by 0.7 percent. The elasticities shown in the table refer to the poverty headcount or FGT(0). Other measures o f poverty, like the poverty gap FGT(1) or the squared poverty gap FGT(2), not shown in the table but also examined by the authors, verify that agricultural growth has higher impact than non-agricultural growth on the poorest population sectors.37There i s variation in the regional impact o f agricultural growth, which i s relatedto the share o f population inrural areas. Thus, the more rural regions (Sur, Golfo, Centro-Norte and Centro) have higher elasticities than the more urban ones (Norte, Capital and Pacijko). 34 The ROA (Roles of Agriculture) project is a study undertaken by F A 0 on the socioeconomic roles of agriculture in developing countries. It covered 11countries, Mexico among them, and 5 areas of potential importance of agriculture outside the direct production of food and other agricultural goods. These areas are: environmental management, poverty reduction, buffering economic cycles and shocks, facilitating the viability of ruralCnmmiinitiQs 2nd rnntrihiltinatn rillhlr21 traditions. The study was carried out between 2001and 2003 (See n) 35Soloaga and Torres' approach consist of a reduced form equation with change inpoverty rates inthe left hand side and agriculture and non-agriculture growth rates on the right hand side. They use poverty rates and growth rates for the 32 Mexican states and 7 geographic regions as observations, and apply both OLS and instrumental variables. They use current consumption rather than income as the welfare variable to measure poverty. They use two-in-two year rates of growth from 1992to 2000 from the national accounts for all the states and INEGI regions for ag and non-ag sectors from INEGI national accounts, and run these against the changes inpoverty inthe corresponding two years registered calculated from the ENIGHs. 36 Unfortunately, available data does not allow separating growth in the rural and urban non-ag sector. Hence, Soloaga and Torres investigate the impact on poverty of all non ag growth. Given the very large dominance of the urban sector in non-ag growth, it is not surprising that the elasticity of rural poverty to non-ag growth is not significant. It the elasticity of rural poverty to rural non-ag growth could be computed itwould probably besignificant and possibly even larger inabsolute value than that for aggrowth. 37 Thus, the value of the elasticities of the extreme poverty gap FGT(1) to rural growth is -1.7 for total poverty and -2.1 for rural poverty. The corresponding elasticities for the squared poverty gap FGT(2) are -2.1 and -1.1 54 Table 4.1 Poverty Elasticities ofAgricultural and Non-Agricultural Growth inMexico Poverty and Growth Sector Total Urban Rural Extreme Poverty Agricultural Growth -1.3 n.s. -1.5 Non-Agricultural Growth -0.9 -1.6 n.s. ModeratePoverty Agricultural Growth -0.6 n.s. -0.8 Non-Anricultural Growth -0.5 -0.7 n.s Only instrumental variables results reported.. n.s indicates no significance. Source: Soloaga and Torres (2003: Table 4). Growth in rural consumption reduces inequality at the national level and also in urban areas but has no effect on rural inequality. The positive effect on urban inequality i s probably linked to the positive impact o f agricultural growth on the demand for unskilled labor. Soloaga and Torres run regression equations for the demand o f skilled and unskilled labor with the growth o f agricultural and non-agricultural output as arguments, controlling for wage levels and the rental price o f capital. They obtain elasticities o f unskilled labor demand o f 0.6 for the growth of non-agricultural output and 0.2 for that o f agricultural output. The elasticity o f demand for skilled labor to non-agricultural growth i s 0.9 while that to agricultural growth i s not significant. The real exchange rate has an impact on domestic food prices while agricultural growth does not, signaling the openness of the Mexican food economy. Soloaga and Torres check for the presence o f a food price effect by regressing food prices on the growth rates o f the agricultural and non agricultural sectors, controlling for variations in the real exchange rate. The results indicate that there i s no food price effect. Instead, the real exchange rate shows a significant impact on domestic food prices. In conclusion, we can say that there is good evidence of a positive effect of agricultural growth on poverty in Mexico. The pro-poor impact is stronger for the worst poverty conditions and i s mostly confined to the rural sector itself. In view o f the impact on poverty, the next question i s how i s Mexico's agricultural sector performing. RECENT AGRICULTURAL PERFORMANCE INMEXICO Agricultural ValueAdded and Food Crop Output ValueAdded Agriculture has experienced modest growth over the past two decades, below that of the national average. In 1980-2003, agricultural growth rates have hovered around 1.5 percent per year, close to one percentage point less than national GDP. Inper capita terms agricultural growth was negative, although it was positive but less than one percent if computed per capita o f the ruralpopulation (Table 4.2). Table 4.2. Mexico: Growth Rates ofAgriculture and Total GDP Growth Rates (real GDP) Periods GDP Ag GDP GDPp.c. A g GDPp.c. 1980-91 2.08 1.45 0.00 -0.62 1991-03 2.67 1.55 1.oo -0.10 55 1980-03 2.39 1.50 0.52 -0.35 Source: WB staff calculations based on INEGI'sNational Accounts. Figure 4.1. Indexesof Evolution of Total and Agriculture GDP inRealTerms (1988=100), Mexico, 1988-2002 200 ' aM 150 $100 d= 3 50 0 +GDPTotal GDPAgriculture Source: WB staff calculations based on INEGI'sNational Accounts. The agricultural and non-agricultural sectors grew in parallel until 1989, when the non-agricultural economy started a faster growth trend (Figure 4.1). Performance was uneven, with "good years" like 1990, 93, 96 and 2001, when growth was above 3 percent, and "bad years", like 89, 92, 94, 97, 2000 and 2002, when growth was close to zero or negative. The performance o f the national economy was somewhat less volatile,38 with a major event, the Tequila Crisis in 1995, and a less dramatic one, the 2002 recession (Figure 4.2). There was no correspondence betweenagricultural and non agricultural which was evident duringthe 1995 crisis, when agriculture grew at 2 percent. 38The coefficients of variation for the annual growth series are 123% for Ag GDP and 103% for GDP. 39The correlation coefficient of the two series is -0.1. 56 Figure 4.2. Mexico: RealAnnual Growth Ratesof Total andAgriculture GDP, Mexico 1981-2003 a 8 Source: WB staff calculations based on INEGI'sNational Accounts. Food Crops An increase in crop yields, partly due to a shift towards higher value crops compensated somewhat for a dramatic fall in real prices for food crop output. The harvested area increased little inthe 80s andnothing inthe 90s (Figure4.3). This points to the exhaustion o f the crop frontier in Mexico, a situation which conditions agricultural growth prospects. Real prices fell dramatically, especially in the 90s, largely as a consequence o f the opening o f the economy. Yields, however, increased; first at a modest rate o f 1.4 percent during the 8Os, and thenat a stronger rate of2.5 percentinthe 90s, which maybeinterpreted as apositive response to the tighteningo f international competition. Yield increases were the result o f modest and uneven improvements in the yields o f individual crops, and o f shifts from low to high value crops, particularly into vegetables and fruits, away from cereals and oil crops, with the consequent improvement o fthe crop mix. M o r e commercialcrops and the irrigated sector provided most of the dynamism to the rural economy while rain-fed farming fell behind. The groups that experienced larger area expansion were vegetables and fruits, whereas oil crops and tubers decreased in area, especially the latter, and cereals remained stable (Table 4.3). Yield expansion was different, however, in irrigated and rain-fed areas, as shown in Figure 4.4. With some ups and downs, food crop yields in rain-fed areas increased in the 1990s somewhat, after being stagnant in the 80s. Yields in irrigated areas increased both inthe 80s and 90s through the early 2000s with a particularly strong trend inthe latter period. Thus, from 1991to 2001 the yield index for irrigatedproduction moved from 120 to 180 -a 50% increase, comparedto a 23 percent increase for rainfed areas. 57 Figure 4.3. Agricultural Production,Yields, Area, Value of FoodCrop Output, and RealPrices, Mexico 1980-2002. Indexes, 1980=100 180 160 140 - -Production Yield- -%-- GrossValueOutput *RealPrices -Area Harvested Source: WB staff calculations based on SAGARPA agricultural database. Table 4.3. Mexico: HarvestedArea in2002, andArea and Yield Growth Rates ofFoodCrop Groups in 1980-2002 Cereals Oil Veget- Fruits Tubers Legumes Crops ables Area in2002 (000 ha) 7,848 209 548 1,222 66 2,238 Growth Rates Area (%) 1980-1991 0.5 -3.7 3.0 1.7 -0.8 2.1 1991-2002 -0.3 -9.1 2.4 2.6 1.4 0.5 1980-2002 0.1 -6.4 2,7 2.1 -1.0 1.3 Growth Rates Yields (%) 1980-1991 1.2 3.3 2.2 0.3 1.99 0.9 1991-2002 2.1 -3.3 1.2 0.9 3.3 0.9 1980-2002 1.6 0.0 1.7 0.6 2.6 0.9 Source: WB staff calculations based on SAGARPA's agricultural database, SIACON. Land productivity stagnated in the 1990s in poorer regions in the South Pacific. Figure 4.5 contrasts the evolution of land productivity in the North Western region (Baja California Norte, Baja California Sur, Sinaloa and Sonora states) and inthe South Pacific region (Guerrero, Oaxaca and Chiapas). Land productivity increased in the North Western region, characterized by its modern commercial agriculture, while it remained stagnant in the South Pacific region, dominated by a more traditional and peasant type farming. Table 4.4 shows that growth rates for all types o f crops were larger in irrigated lands. Not surprisingly, the highest growth rates corresponded to horticultural and fruits crops, which are characteristic o f modern commercial farmers. 58 Figure 4.4. Mexico: Evolution of Food Crop Yields in Irrigatedand Rain-fed Areas. Indexes, 1980=100 -Irrigated Area -Ram-fed Area Source: WB staff calculations based on SAGARPA's agricultural dataset. Figure 4. 5. Mexico: Evolution of a LandProductivity Index in the North Western and South Pacific Regions, 1992=100 1.h i i 0.6 V A 0.4 0.2 0.0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 -North-Westem South-Padc Source: WB staff calculations based on SAGARPA's agricultural database, SIACON. No data is available on agricultural growth by type of farm that would allow us to show how different type of farming sectors performed during the decade examined. However, the evidence presented -higher agricultural growth in the northern states where agriculture i s more commercial, inirrigated lands where commercial farming concentrates, and in the comparatively more commercial crops- points clearly to an uneven type o f agricultural development. Thus, agricultural growth was not only modest but concentrated also mostly in the more commercial farming sector. The evidence points hence to an increase during the decade o f the dualism characteristic o f Mexican agriculture and corroborates the findings in chapter 3. Table 4.4. Mexico: Annual Growth Rates of Output 1991-02 of Various Types of Crops inIrrigatedLands and All Lands GrowthRate Output 91-02 59 Type of Crop IrrigatedLands All Lands Cereal Crops 2.1 1.9 Oil Crops -18.7 -11.8 Horticultural Crops 3.7 3.5 Fruit Crops 4.5 0.5 Tubers 2.6 2.0 Legumes 2.9 1.4 Source: WB staff calculations basedon SAGARPA's agricultural database, SIACON. Agricultural Productivity Land and labor productivityimprovedin the 1990s. Inboth cases, growth rates inthe 1990s were above 2 percent, which can be considered a reasonable performance. As can be seen from Figure 4.6, the increase in the productivity o f land was approximately one third larger than that o f labor, indicating a certain land-bias inthe technical change taking place. Figure 4.6. Mexico: Evolutionof Land and Labor Productivity inAgriculture, 1988to 2001 2001 0 c / , ,/'020000" 1998. 0 1 9 9 7 R0 * I 9 9 9 0,' 1996 , OyO /` 01995 / /MO 0 1 9 9 3 0 1 9 9 2 ,/ 01991 ,0' 01994 0 /' 0 1 9 8 9 m 0 1 9 8 8 I I I I 100 110 120 130 A g ricu Itu raI L and Prod uctivity Source: WB staff calculation basedon FAO's AGROSTAT. Land productivity in Mexico is modest in relation to that of comparable countries, however. With the exception of some crops, notably wheat, land productivity, i.e. yields, are fairly low inMexico. As can be seen from Table 4.6 and Figures4.7 and 4.8, Mexico i s below the L A C yield average in maize and it i s also below in the whole cereal group, notwithstanding its advantage inrice and specially wheat. Yields are also below the L A C average in coffee and citrus fruits, and above in sugarcane, cotton and vegetables. L A C averages are a modest standard, however, for a middle income country with a good proportion o f land under irrigation, a long farming tradition, and, as we will see, plenty public investment in agriculture. Comparison with Argentina, Chile and Brazil i s more appropriate and i s not favorable to Mexico, particularly inthe case o f the first two countries. Table 4.5 Crop Yields inSelected Countries,Average 2000-2002, (todhectare) Mexico Areentina Chile Brazil LAC USA EU India China 60 Cereals 2.8 3.4 4.9 2.9 2.9 5.8 5.6 2.3 4.8 Maize 2.6 5.8 9.8 3.0 3.0 8.5 9.1 1.9 4.7 Wheat 4.9 2.3 4.1 1.6 2.4 2.6 5.7 2.7 3.8 Rice 4.3 5.3 5.1 3.2 3.8 7.2 6.4 2.9 6.3 Sugar Cane 74.1 65.4 _ _ 69.6 64.9 77.2 _ _ 67.3 61.3 Cotton (Seed) 3.3 1.3 2.7 2.1 1.9 3.3 0.6 3.3 Coffee (Green) 0.4 _ _ __ __ 0.9 0.8 _ _ _ _ 1.o _ _ Citrus 12.4 20.1 15.4 22.0 17.0 34.7 18.3 17.8 8.2 Pulses 0.8 1.1 1.6 0.7 0.8 1.9 2.7 0.6 1.4 Vegetables 16.5 17.2 25.6 17.9 14.9 27.1 26.7 12.9 19.2 Wheat is the only field crop where Mexico has shown to be able to produce at top technical levels, surpassing Chile, a country with plenty irrigation, good technology, and a strong wheat producing tradition. Italso surpassesthe USA andcomes closeto the EUyield. Source: WB staff calculations basedon FAO's AGROSTAT. Figure 4.7 Comparative Cereal Yields as Percentageof LAC Average, Average Yields 2000-02 m 3 150 P Y e s o U 0 Mexico Argentma Chile Brazil LAC USA EU India Chma Source: WB staff calculations basedon FAO's AGROSTAT. Labor productivity levels in Mexican agricultural sector have fallen behind and are below the average for LAC countries. Labor productivity i s compared across L A C in Figures 4.9, 4.10 and 4.11. Not surprisingly, land-abundant countries like Argentina and Uruguay show the highest productivity, but other countries like Costa Rica and Chile, which are land-scarce, come next (Figure 4.9). Of the 20 countries reported, Mexico i s in position 13, with a labor productivity of USD 2,265, which i s one third below the L A C average of USD 3,368. Moreover, Mexico i s not catching up vis-&vis other countries in the 199Os, while countries like Peru and Nicaragua who have also low labor productivity in agriculture are closing the gap, and in countries like Costa Rica, Chile and Brazil, where labor productivity i s already high, its growth rate i s above the L A C trend (Figure4.10). 61 Figure4.8. ComparativeYields as PercentageofLAC Average, Various Crops, Average Yields2000-02 n Mexico Argentina Chile Brazil LAC USA EU India China BSugar Cane $Cotton (Seed) OCoffee (Green) OCitrus Ol'ulses OVegetables Source: WB staff calculations based on FAO's AGROSTAT. Figure4.9. AgriculturalLaborProductivityl/inLAC Countriesin2002 Measuredin 1995USDper Worker 9,000 %000 7,000 6,000 d000 4,MM 3,MM 2,000 1,000 0 Hat Bo1 Hon E m ElSal Gua Per Meu Nic Pan Par RD. Col Cub Bra Ven Chi C.R Um Arg. LAC Source: WB staff calculations based on CEPAL (for ag value added) and FA0 (for ag labor force) data assembled by Dirven (2004). l/ Definedas agricultural value added dividedby the agricultural labor force and measured inU S Dollars o f 1995. 62 Figure 4.10. Labor Productivity inAgriculture in2002 Measured in1995 USD Der Worker, and its Growth Rate in 1990-2000 0 2 0 0 0 4 0 0 0 6 0 0 0 8 0 0 0 1 0 0 0 0 A g r i c u Itu r a l L a b o r P r o d u c t i v i t y Source: WB staff calculations based on CEPAL (for ag value added) and FA0 (for ag labor force) assembled by Dirven (2004). The gap between agricultural and non-agricultural labor productivity is particularly high in Mexico. In Figure 4.11, we compare labor productivity in agriculture and outside agriculture. In general, it i s clear that all countries with the exception o f Nicaragua have lower productivity in agriculture. In some countries, like Argentina and Uruguay, labor productivity i s high in both sectors, although smaller in agriculture, whereas Mexico shows a marked difference between the high productivity o f labor outside agriculture and its low productivity in agriculture. Figure 4.11. Labor Productivity inAgriculture and Non-Agriculture in2002, Measured in 1995USD per Worker Source: WB staff calculation prepared with data fi-om CEPAL (for ag value added) and FA0 (for ag labor force) assembled by Dirven (2004). Mexico has lost momentum in agricultural efficiency. An increase in total factor productivity (TFP) signals efficiency improvements in the use o f inputs (land, labor and capital) 63 due to better technology, better entrepreneurship or both. Using FA0 data for 1961-2001, Avila and Evenson (2004) compute rates o f change o f TFP in crop and livestock production. The inputs considered for crops are cropland, labor, fertilizer, animal power, andmachine services. Those for livestock are pasture land, labor, fertilizer, animal capital, and feed. The results, reported in Table 4.6, indicate that agricultural efficiency gains have slowed down in Mexico, passing from a TFP growth performance in 1961-80 which was 63 percent above the L A C average to one 35 percent below the average in 1980-01. The major fall came about in the livestock sector, which performed strongly inthe first periodfalling to more modest levels inthe second one. Table 4.6 Growth Ratesof Total Factor Productivity inLAC Countriesin 1961/80and 1981/2001 Regions and Crops Livestock Aggregate Countries 1961/80 1980/01 1961/80 1980/01 1961/80 1980/01 Average Southern Cone 1.49 3.14 0.72 2.51 1.02 2.81 1.92 Argentina 3.08 3.93 0.90 0.43 1.83 2.35 2.09 Brazil 0.38 3.00 0.71 3.61 0.49 3.22 1.86 Chile 1.os 2.22 0.24 1.87 0.69 2.05 1.37 Paraguay 3.97 -1.01 -0.36 1.29 2.63 -0.30 1.17 Uruguay 1.29 2.02 -0.32 0.53 0.01 0.87 0.44 Andean Region 1.11 1.71 1.73 1.92 1.41 1.81 1.61 Bolivia 1.73 3.14 2.81 1.39 2.30 2.33 2.3 1 Colombia 2.01 1.27 0.49 2.24 1.37 1.73 1.55 Ecuador -0.74 2.24 0.98 2.5 1 -0.16 2.34 1.09 Peru -0.83 1.86 1.86 2.14 0.36 1.98 1.17 Venezuela 2.42 0.87 3.41 1.07 3.03 0.99 2.01 Central and North 1.65 1.05 2.77 1.53 2.17 1.32 1.74 America CostaRica 2.86 2.09 1.10 0.75 1.74 1.19 1.47 ElSalvador 1.22 -0.87 1.99 1.00 1.77 0.32 1.05 Guatemala 3.31 0.53 0.90 -0.28 1.38 -0.08 0.65 Honduras 1.54 -0.39 2.07 1.91 1.91 1.25 1.58 Mkxico 1.53 1.43 3.02 1.63 2.26 1.51 1.89 Nicaragua 1.33 -0.70 2.94 1.92 2.25 0.99 1.62 Panama 2.29 -1.33 1.61 1.49 1.93 0.02 0.97 Caribbean 0.66 -0.89 2.60 2.06 2.03 0.90 1.47 Dominican Republic 0.99 -1.15 1.88 2.60 1.59 1.28 1.43 Haiti 0.60 -1.04 3.44 1.80 2.60 0.50 1.55 Jamaica -0.65 1.32 3.28 -0.35 2.3 1 0.12 1.22 LAC 1.46 2.40 1.42 2.21 1.39 2.31 1.85 AVERAGE Source: Adapted from Avila and Evenson (2004). Public Expenditure in Agricultural and Rural Development The amount spent by the Mexican government on rural development is truly remarkable. It i s a fiscal effort without parallel in Latin America, especially when the low tax incidence and fiscal revenues o f Mexico are considered. A Programa Especial Concurrente (PEC) mandated by the Ley de Desarrollo Rural Sustentable, where all federal spending in rural development i s lumpedtogether, was assembled for the first time in 2003. PEC was not a joint inter-secretaries programming exercise, but was a positive step in that direction. Thanks to PEC we have a betterknowledge o f federal spending inrural areas and o fwho i s responsiblefor itthan used to be the case in the past. We present in Table 4.7 the total amount budgeted for federal spending in rural development in 2003 broken down according to the major expenditure areas 64 consideredin PEC. InTable 4.8 we show how federal secretarias involved inrural development contribute to this effort. PEC amounts to some 30 percent of agricultural GDP.The total budgetary figure for PEC i s MxP 117.1bn (Table 4.7), equivalent to some USD 10.4 bn, which comparesto a GDP in agriculture in 2003 of the order USD 33 bn. Not all of this, however, i s spent in agriculture, because there are social, infrastructure, and other programs includedin PEC. On the other hand, this is only federal spending. Since state and municipal governments also spend in rural development -we do not know how much-, the actual amount of rural public expenditure is more than that reported under PEC. Table 4.7. Mexico: Federal SpendinginRural Development by Major Areas in2003 (MillionMxP) Number Budget Major Rural Development o f (Million % ProgramAreas Programs MxP) Social Conditions 20 45,343 38.7 Labor Conditions 4 2,874 2.5 Land Tenure 2 3,257 2.8 Productive Activities 22 50,687 43.3 Basic and Productive Infiastruc. 6 10,030 8.5 Environment 5 4,905 4.2 TOTAL 59 117,096 100.0 Source: WB staff calculations based on SAGARPA (2004). Table 4.8. Mexico: Contribution of Different Secretariasto Federal SpendinginRural Developmentin2003, (MillionMxP) Programmable Rural Secretaries Budget o/n Development o/n Economia 5,403 2.1 576 0.5 SAGARPA 41,783 15.9 40,583 34.6 Salud 20,867 7.9 6,829 5.8 SCT 23,124 8.8 1,092 0.9 SEDESOL 18,978 7.2 15,574 13.3 SEMARNAT 17,404 6.6 8,977 7.7 SEP 106,355 40.4 17,554 15.0 SHCP 21,785 8.3 10,310 8.8 S R A 2,759 1.o 3,566 3.0 STPS 3,151 1.2 866 0.7 Turismo 1,459 0.6 16 0.0 Others 11,153 9.5 Total 263,068 100.0 117,096 100.0 Source: SAGARPA (2004). Table 4.8 shows the total programmable budget of the secretariasinvolved in rural development and thus participating in PEC, alongside with the amount actually spent in rural development. Of the MxP 263.1 bn that the secretarias have available, MxP 117.1 bn or 44.5 percent i s devoted to rural development. If programmable expenditures of secretarias not involved in rural development are also considered, which come to another MxP 83.1 bn, PEC amounts to 33.8 percent of all federal programmable expenditure. 65 The two major expenditure areas in PEC are social conditions and productive development, which together account for more than 80 percent ofthe total. Infrastructure is a comparatively small expenditure area compared to its importance and need, and so i s the environment. Ofthe secretarias, that which contributes most i s SAGARPA, followed by SEP and SEDESOL. Figure4.12 PublicExpenditureinAgriculture perAgriculturalWorker inLAC Countries, Average 1996-2000 `hnl Source: WB staff calculations based on data from Kerrigan (2001). Mexico spending on agriculture i s very high compared to other LAC countries, whether measured per agriculturalworker or as share of total public expending. Mexico is the country in Latin America with the smallest share o f agriculture in GDP, making the high levels o f public spending on agriculture all the more remarkable. Figure 4.12 looks at public expending in agriculture per agricultural worker in L A C countries for the average o f 1996-2000. Figuresrefer to expenditure in production-related programs for agriculture only, not to all rural development expenditures. They are in current U S dollars. For some countries the average i s for 1996-1999 due to lack o f data for 2000. Unfortunately it i s not possible to tell from available data how this expenditure brakes down into the provision o f public goods, farm modernization incentives, and other subsidies. The dispersion i s big, with some countries spending twenty times more per agricultural worker than others. Chile i s first, with expenditures per agricultural worker close to USD 400, followed by Mexico with USD 303 per worker. Ifwe consider instead public expenditure in agriculture relative to total public expenditure, Mexico comes out first, followed by Bolivia, Nicaragua and Dominican Republic (Figure4.13). In contrast to Mexico, these three countries are highly agriculturally oriented, with shares o f agriculture in GDP o f the order o f 14, 35 and 11percentrespectivelyby the end o fthe 90s. Figure 4.14 shows the distributionof LAC countries according to an "agricultural orientation" index, which is built by dividing the share of agricultural spending in total spendingby the share of agriculture in GDP. The index measures the intensity of the fiscal effort in agriculture relative to the economic importance of the sector. Mexico is the only country of those included inthe study with an agriculture orientation index o f more than one, which i s in fact more than four times larger than that of the next countries, Bolivia, Chile and Dominican Republic. Mexico, hence, practices public expenditure discrimination in favor o f its agriculture. 66 Figure4.13. PublicExpenditureinAgricultureas Percentageof Total PublicExpenditureinLAC Countries,Average 1996-2000 9 Source: WB staff calculations based on data from Kerrigan (2001). Figure4.14. AgricultureOrientationIndex inLAC Countries, Average 1996-2000 2.0 s Y Source: WB staff calculations based on data from Kerrigan (2001). Rural development expenditures have picked up since the end of the 1990s, due mainlyto an increase in productive expenditures. Between 1996 and 1999, while the country was recovering from the crisis, rural development expenditure was stagnant in current pesos and strongly decreasing in constant pesos. 40 Real expenditure recovered w e l l from 1999 onwards but not completely, resulting ina fall of close to 12percent between the two extremes ofthe period. 40A public expenditure series from 1995 to 2004 (figures for 2004 are programmed) with information from the SHCP was prepared for this purpose. We have aggregated programs inthe same areas as PEC, buthave combined social and labor aspects, and land tenure and productive aspects. We are grateful to Oscar Diaz Santos, experienced civil servant and graduate student at UNAM, for compiling these figures for us. Since there have been many changes inprograms and subprograms, and inthe allocations, names, and location of programs in the public accounting system, it is difficult if not impossible to trace with exactitude the 67 Figure 4.15. PublicExpenditureinAgriculture and RuralDevelopment inMexico, 1995-2004, in Current Prices (MillionMxP) .- TotalARD - - - Social--k -Infrastructure *Environment &ProductmnandLand Source: Annex 4.A. Figure 4.16. PublicExpenditureinAgriculture and RuralDevelopment inMexico, 1995-2004, in Constant 1993 Prices (MillionMxP) 40,000 3S,000 311,000 10,000 S,000 Total ARD Social--k- Infrastmchre *Environment -Production and Land Source: Annex 4.A. T o be noticed are the decrease in social expenditure between 2002 and 2004, and the parallel increase in productive e~penditure.~~ Also to be noticed are the low amounts spent in infrastructure and environmental programs relative to other spending areas. Detailed expenditures byprograms are presented inAnnex 4.A. evolution of expenditure. A reassuring fact was the correspondence between our own estimate of federal publicexpenditure for 2003of MxP 116.7 bnwith the MxP117.1inthe PEC. 41 The classification is largely arbitrary because many programs have characteristics or components that could lead to their classification under different categories. Also, the concepts themselves of social, productive, and environmental spending are not free from ambiguity. We have followed as much as possible the classification used in the PEC. For the evolution of specific programs consult Annex Tables 4.A.1 to 4.A.6. 68 PROFITABILITY AND EFFICIENCY OF THE SMALL FARM SECTOR ProJta bility How profitable is the small farm sector in Mexico today? To answer this question we use 2002 data from the Encuesta Nacional de Hogares Rurales de Mkxico (ENHRUM) survey which, as described in chapter 1, focused on disperse rural areas and i s representative at the regional level. We use a subset o f the data covering 661 households o f agriculturalists dedicated mostly to crop production. Farmprofitability i s difficult to estimate (Box 4.2). Thus, for example, we do not investigate the profitability o f livestock operations because o f the difficulty of establishing for small farms the amount o f time dedicated by the householdto looking after their animals, andthe amount o f own inputsusedto feed them. Box 4.2. Estimating the profitability of small farms Two main problems arise in estimating the profitability o f small firms. First, small farms are usually part o f larger household operations, and it i s artificial to cut farming away fiom these operations. This i s why it i s usually better to use "household models" instead of "farm models". W e do not follow this advice here, however, because in the present context w e are interesting in examining the profitability of agriculture and decision o f rural families to participate in different occupations and the income derived from them. W e are interested in particular in the income derived from crop production, not that o f the entire family economy. Later on, in Chapter 7, we examine the process whereby rural households select different occupations and complement income from various sources. The second problem i s the choice o f profitability indicator. Small farm behavior in Mexico i s generally that o f the campesino or peasant producer. As i s known, peasant micro-economic behavior i s different from that o f commercial farmers, inthe sense that the objective function and often too the constraints are different.42The reasons are the lack o f a contractual link between employer and labor, since the own labor o f the farmer and hidher family are used, and the overlap of the production and consumption unit. T o measure profitability we use several indicators. The first on is the Gross Profit margin (GP), defined as the total value o f output, including the estimated value o f self- consumption and farm re-employments, minus all direct production costs consisting o f labor, bought inputs, re-employments (for instance own seed), and machinery or draft animal services. Managerial costs, depreciation o f assets, the rent value o f land, and the actual or imputedvalue o f financial services are not included. We use two specifications o f GP. In one o f them, we include the estimated cost of family labor valued usingthe local wage rate, while inthe other we consider only the cost o f hiredlabor and not that o f family labor. The latter specification is better suited to the peasant farm, which does not look to family labor as a cost inthe same way that it does with hiredlabor. A second indicator is Value Added (VA), defined as the value o f output minus direct production costs, without including labor (either family or hired). The difference between VA and the second specification of GP is that GP includes hiredlabor. Reemployments incrop production (for instance seed) are included as part o f output and as part o f costs in both measures, thus canceling out. The smallest farmers generally have negative or very low profits when family labor put in the farm is valued at the wage rate prevailing in the area. There are 385 farms with negative GP when family labor i s valued and subtracted, against 276 with positive GP (Figure 4.17). This i s not surprising; negative profitability o f peasant farms measured with commercial *A classical discussion of this is that of Nicolas Georgescu-Roegen (1960). 69 parameters i s a well know fact.43It means that family labor used inthe farm gets a return smaller than the going wage rate. But there are various reasons why the wage rate i s not necessarily a good measure o f the opportunity cost o f family labor. They include the possible presence o f involuntary unemployment, the fact that much o f family labor may not be tradable, and the possible effect o f risk management considerations that may make farmers prefer to use their labor in their own farm even if at a smaller return. As seen in Table 4.9, the smallest farms have on average negative profits. Figure 4.17. Number of Farms with Positive and Negative Gross Profit in a Sample of Small Farms, Mexico 2002 -500 OnlyPaidLabor Included AllLabor Included Source: Calculated from ENHRUM. See for instance the analysis of Witold Kula (1970) for the Polish peasant economy during the "refeudalization" period. 70 Table 4.9. Gross Profit and Value Added per Hectare of a Sample of 661 Small Farms, Mexico 2002 (2002 MxP) GROSS PROFIT PER HA (MXP) w/o Subtracting Subtracting Value Added Family Labor Family Labor Per Ha (MxP) Size Class 1/ 0 -2 hectares 2,415.2 (440.7) 3,752.1 2 -5 hectares 2,079.3 819.6 2,740.1 5 -10 hectares 2,480.4 1,750.7 2,820.6 10 and more hectares 2,210.6 1,784.6 2,354.3 Type of Ownership Ejido 1,822.7 657.4 2,578.1 Comunidad (32.1) (1,9 18.4) 583.0 Private 3,614.4 1,38 1.2 4,494.8 Mixed 2,275.9 1,141.O 2,48 1.7 Average sizes for size groups are as follows: 0- 2 hectares: 0.9 hectare; 2-5 hectare: 3.0 hectare; 5-10 hectare: 6.6 hectare; more than 10 Has- 29.4 hectare; all farms: 7.7 hectare. Source: Calculated from ENHRUM. Value-added per ha does not increasewith farm size, suggesting that land is in fact more productive in smaller farms. Gross profits per ha increase by farm size, but value added per ha is actually larger in farms of less than 2 hectare than inbigger farm categories, and farms inthe more than 10 hectare category have the lowest VA per hectare (Table 4.9). Hence, under existing conditions, from the point o f view o f generating income, land i s more productive if it i s insmallthan inlarge farms. The low profitability of small crop production can help explain high poverty ratesin the rural sector, and why rural non-farm activities play a key role in securing income. We can compare our value added figures with the rural poverty lines for 2002, o f MxP 5,937 per person per year for food poverty and MxP 11,363 per person per year for moderate poverty. One hectare in our sample generates a value added from crop production somewhere between 40 and 60 percent o f the income required for a person not to be food poor, and between 20 percent and one third o f that requirednot to be poor at all. Put in another way, a family o f four would require on average between 6 and 10 hectare not to be food poor, and between 12 and 19 hectare not to be poor, if crop farming were the only source of family income. But, as examined in chapter 7, farming families would usually seek other occupations, starting with some livestock, among which they decide whether and how much to participate and from which they obtain income. The above figures point, however, to the low profitability shown on average by small crop production in Mexico, and help explaining why farming families try to access other income sources. There are o f course wide variations, and much depends on the quality o f land, the type o f crops, and context variables such as road availability and access to marketing infrastructure and channels. For instance, small farms with irrigation producing highvalue crops can have a much larger value added, and three or four hectares may be enough to keep a family out o f poverty without engaginginother activities. 71 Private farms are also more profitablethan farms in ejidos or comunidades (Table 4.9.). The latter are particularly little ~ r o f i t a b l eValue added in crop production per hectare in a . ~ ~ private farm i s nearly eight times higher than in a comunidad, and 1.7 times higher than in an ejido. Few inputs are used in small farms. If we compare the gross value o f output (GVO) and the GP without subtracting family labor, the difference i s the cost o f the inputs used in production, including seed, fertilizer, water, chemicals, hired labor, and the cost o f mechanical and animal services. The small difference betweenthe two figures shown in Table 9, which on average i s o f 10 percent, points to the few inputs bought in small crop production. A curious feature i s that the difference between GVO and GP i s proportionally higher in farms o f less than 2 hectare. There may be measurement errors, but a possible reason for this i s that the part o f output that i s re-employed within the farm i s more used in crop production in the smaller farms and more in animal production in larger farms. Since we are examining crop operations only, crop reemployments used in the farm's own animals (eg. maize to feed pigs) i s accounted as part o f output but not as part o f costs. Small Farm Efficiency How efficient are small farmers?Again using ENHRUMdata, we have carried out an exercise to determine the economic efficiency o f small farmers. Details o f the econometric approach are explained in Annex 4.B. In summary, our method allows to test (i) what explains output, (ii) usingthis information, which farmers deviate from the efficiency frontier, andto what extent, and (iii) what variables may explain why some farmers deviate from the efficiency frontier. Table 4.10 reports the results from estimates to explain output. The dependent variable i s the gross value o f crop output o f farms, and the explanatory variables are: variable capital which includes seed, fertilizer and chemicals, fixed capital, which includes the value o f machinery and draft animal services, labor, which includes all labor used during the entire production process, and land, which enters in the equation in a standardized form to account for differentquality o f land(and not only size o flandholdings). Table4.10 Value ofElasticities inthe ProductionFunction RegressionEquationfor all ENHRUM Sample of Crop Farms Parameter Variable Coefficient S.E. P>z bi Variable 0.430 0.0354 0.000 capital b2 Fixed 0.159 0.0353 0.000 Capital b3 Labor 0.093 0.3809 0.015 b4 Land 0.232 0.0302 0.000 bo Constant 3.259 0.3666 0.000 Note: Using a Cobb-Douglas Production Function. The logarithmic specification implies that coefficients can be interpreted as elasticities. Source: Staff estimates based on ENHRUM 44 Proper analysis would require controlling for farm size and other characteristics that may vary between the private and social sectors, using econometric methods. That work is currently in process and has not been possible to include here. 72 Variable capital is key in explaining output, with a big difference to other factors, followed by land. A one percent increase in variable capital increases output by 0.43 percent. This may be the result of the underutilization of fertilizer and chemicals due to the lack of access o f small farmers to seasonal credit (see chapter 5 on this). Lack o f credit to buy inputs prevents farmers from using them optimally, i.e. up to the point when the marginal contribution to production equals the cost to the farmer. A one percent increase in land holdings increases output by 0.23 percent. Labor has little weight, signaling the probable presence of surplus labor inmany farms. The sum o f elasticities i s less than one, which in principle would indicate the presence o f diseconomies o f scale. However, the confidence intervals o f the coefficients are sufficiently large not to reject the null hypotheses that the sum o f the coefficients i s one. Hence, there i s no statistical evidence o f diseconomies o f scale, but there i s little likelihood o f economies o f scale for the entire sample.45 Which type of farms are less efficient? Table 4.11 groups inefficiency residuals by categories o f farms, where the average inefficiency o f each category o f farms i s measured as the distance to the efficiency frontier. Thus, for instance, at the national level an effort would be required to increase production by 91 percent with existing factors to reach efficiency, i.e. production would almost needto double. Farmers that experiencednatural shocks, maize and beans farmers, and farmers in the Sur-Sureste and Centro regions, are the least efficient. Producers o f coffee, other perennial crops and vegetables, and producers in the Noreste and Noroeste regions are the most efficient. While farms in the Centro tend to be less efficient, the value o f land in this region i s the highest (see Appendix 4.B2 in Annex 4.B). The reason could be that farms in the Centro are better communicated and closer to markets than in other regions, and population pressure i s also high, pushing up the price of land. Another surprising result is the large number of farmers that suffered from natural disasters, and its strong impact on e f f i ~ i e n c y .We~ discuss this more in ~ chapter 7 but we must notice here that this i s a frequently overlooked element with important repercussions on efficiency. Table 4.11. Distribution of the Inefficiency Error Term by Category of Farms Average Variables No. of farms Inefficiency S.D. Sur-Sureste Region 211 0.94 0.3809 Centro Region 233 1.03 0.3980 Centro-Occidente Region 122 0.78 0.2391 Noroeste Region 36 0.69 0.4104 Noreste Regions 64 0.73 0.4012 Maize andBeans farmers 456 0.98 0.3845 Coffee Farmers 43 0.65 0.2273 Vegetable Farmers 33 0.69 0.2467 Perennial Crops Farmers 79 0.75 0.3419 Oilseeds and Other Grain Far. 55 0.86 0.4288 Farmers with natural shocks 295 1.05 0.448 1 Farmers without natural shocks 371 0.80 0.2862 45 Notice that economies of scale in this context do not refer to increased farm size, which is the common meaning in usual parlance, but to the simultaneous increase of ALL factors in the production equation, whichis the technical meaning. 46 This was defined as farmers who reported having suffered from rains, hurricanes, droughts, frosts or pests and diseases, and whose output was less than 50 percent that of a good year. 73 All Farmers 666 0.91 0.3870 Sourcce: calculated from ENHRUM. There are substantial changes in the elasticities of production factors for different types of producers, although that for variable capital remains always the highest. The stochastic production function regression exercise was carried out separately for various categories o f farmers.47The elasticities are shown in Table 4.12. For maize and beans producers the importance o f fixed capital i s higher than for the entire sample o f farmers, while that o f labor remains low. This indicates that more use o f animal power and/or tractor services for these producers would have a large effect on output. Contrarily, reducing the amount o f labor put in these crops would not have a large impact on output. Land elasticity in maize and beans production i s surprising low, less than half o f that for all farms and for farms without natural shocks. Increasing output in maize and beans production depends hence more on improved technology embodied invariable and fixed capital than on increasing the area. This i s good news for small peasant farmers who are the main producers of maize and beans, for it means that they could boost output in their small farms if they had access to better technology. Under present conditions, however, shiftingland from maize andbeans to other crops would raise total output. Farmers who did not experienceshocks represent"normal" Le. shocks free, farming condition^,^' and their elasticity coefficients and efficiency levels are illustrative of this situation. In the case o f farmers who experienced natural shocks, labor is much more relevant than for all farmers in general, and there i s no evident explanation for this. Instead, for farmers who did not experience shocks, the importance o f land i s the highest. Hence, under "normal" conditions land i s more relevant than under "abnormal" ones. The presence o f a large proportion o f maize and beans farmers in the sample, nearly 70 percent, decreases the elasticity o f land since, as we have seen, this elasticity i s low for these farmers. In farms, therefore, producing other crops under "normal" conditions land must be much more important. An interestingresult for the "farmers without shocks" sub-sample i s that the inefficiency test failed to reject the hypothesis o f no inefficiency. This does not necessarily mean that these producers are all efficient, but points to the strong link between"normality" in production conditions and farming efficiency. Table 4.12 Production Function Elasticitiesfor Different Crop Farmers inthe ENHRUMSample Elasticities Variables All Farms Maize and With Without Beans Shocks Shocks Kvar 0.430 0.443 0.399 0.391 K f x 0.159 0.265 0.273 0.183 Labor 0.093 0.107 0.276 0.207 Sland 0.232 0.132 0.171 0.249 Sum Elasticities 0914 0.947 1.119 1.030 Constant 3.259 3.669 2.658 2.226 All coefficients significant at 95% level. Source: calculated from ENHRUM. 47We only carried out a separate analysis for these categories of farmers because sample sizes were too small for the other categories. But we can see in Table 4.11 that this "normal" condition is not so normal: 295 out of 666 farmers, 44 percent, experiencednatural shocks. 74 Private transfers tend to be associated with less efficiency. One could expect the opposite, thinking that transfers would release liquidity constraints o f farmers to buy inputs or fixed capital services or hire labor. But this i s already controlled for in the regression carried out on the residuals, and there i s no reason why transfers should increase efficiency inthis way since the existing utilization o f factors i s given. A possible interpretation i s that the more transfers received by farmers the lesser their dependence on farming for survival and hence the lower their interest in farming operations and good crop husbandry. M o r e work i s needed to understand the causes of inefficiency. Many o f the explanatory variables included in the regressions on efficiency, like gender, age, education, existence o f services in the community (measured by a services index), land tenure, farm size, and government transfers, were not statistically significant. The reason seems to be that these variables influence more the choice o f technology, i.e. the combination o f inputs in the production function, than the efficiency o f production given a combination o f inputs, which i s what we investigate by regressing the efficiency residuals on these variables. HOW CANPRO-POOR AGRICULTURAL GROWTHBE STIMULATED? Agricultural growth has a strong potential for poverty reduction in Mexico. As shown, agricultural growth reaches the extreme poor, reduces poverty intensity, and lowers income inequality in society at large. Resolving the challenges faced by the agricultural sector, including increasing labor productivity, and ensuring that smaller farms and the rain-fed sector become more competitive, i s therefore essential to rural poverty alleviation. Federal expending in agriculture and rural development i s very substantial in Mexico, and a true mark o f the importance that Mexican governments have traditionally given to rural areas. We discuss here in general terms how policy could respond to the trends and conditions depicted so as to promote agricultural growth in a way that i s friendly to the poor. More specific policy options are discussed in chapters 5, 6 and 7. We organize our discussion around four areas: agricultural intensification; the small farm economy; competitive conditions; and public programs. We should note, however, that land scarcity and surplus labor require also policies to promote the development o f non-farm activities in rural areas. This could create a virtuous circle: by raising incomes, intensification stimulates demand for RNF activities, and conversely RNF development absorbs agricultural labor surpluses, improving factor proportions inagriculture. Agricultural intensification is the best option in view of the exhaustion of the crop frontier and the low average levels of land and labor productivity in agriculture. Labor productivity in agriculture i s low inMexico, and under existing conditions substantialamounts o f land are needed on average for a family to pull out o f poverty by means of crop farming alone, more than that usually available to small farmers. At the same time, the labor absorption capacity o f non-agricultural activities i s limited, the rural population i s still growing, and the crop frontier i s closed. Thus, average farm sizes cannot be increased by enough farmers changing occupation and abandoning their lands, which could then be rented or bought by other farmers. Intensification i s hence the best option to increase land and also labor productivity in agriculture, and thus agricultural incomes. Intensification needs to proceed along two ways: changing the crop mix in favor of higher value crops, and raising crop yields. The poorer farmers will need policy assistance to achieve this. Changing the crop mix in line with relative prices and market opportunities i s a difficult and long-term process. There i s much rigidity in the farming and marketing systems that hinder crop changes. After nearly 20 years ofliberalizationpolicies andrelative price change, itis reasonable to assume that the "easy" part o f substitution has been achieved. The uncertainties and 75 the transaction and other costs o f substitution are probably increasing at the margin. As we have seen, commercial crop farming i s more efficient than maize and beans, and there should be market incentives to move from the latter into the former. But there are barriers to that process that many farmers willing to substitute cannot overcome. These barriers are technological, financial, commercial, in knowledge, and in the capacity to bear risk. These are all areas where policy can intervene, and in fact Mexico has a variety o f programs to this effect, some o f which we examine in the next chapter. Better off farmers, endowed with more agricultural assets, are better equipped to break these barriers ifthey have not done it already. It is the poorly endowed farmers who face the major difficulties and needmore assistance. This will require a series of programs -research, technical assistance, extension, rural finance, infrastructure, and market information- that jointly create the conditions enabling farmers to carry out this movement by themselves. Some specific interventions may perhaps be requiredto accelerate the process, for instance in export marketing, the identification o f market opportunities, and the linkingo f small farmers to new major sources o f demand such as supermarkets. Work may also be neededinspecific crops and production chains. Butthe essential thing is the presence of extensive and well functioning research & extension and rural finance systems, for both crop diversification and yield increases. The small farm economy encounters most marketfailures and must be at the center of a poverty-friendlyagriculturalgrowth strategy. Although enabling conditions will often be similar for small and larger farmers, markets tend to fail more for small than for large farmers. Giventhat there is no evidence of economies of scale, the small-farm sector is a potential carrier o f agricultural growth. A "diffuse" intensification strategy should embrace all type o f farms, irrigated or rain-fed, small or large, and instrumentslike research and extension and rural finance need to be specially calibrated to be able to reach small farmers. Raising value added in small farms would not only reduce poverty directly; it would also have a general equilibriumimpact on rural and urban poverty through the wage effect. Although we do not advocate a separate agricultural growth program for small farmers, there are interesting experiences o f government programs for small farmers in L A C countries, like PRONAF in Brazil and INDAP's programs in Chile, which could serve as inspiration for an enhanced policy effort in favor o f small agriculture inMexico. Rentalmarkets for lumpyinputs are important.We saw that inthe case of maize and beans, typical small farm crops in Mexico, fixed capital i s an important element to increase production, more than land and labor, and second to variable capital. This points to the importance for small farmers o f promoting the development o f rental markets for lumpy inputs such as machinery and draft animals. How can the impact of natural shocks be mitigated? Our research shows that efficiency, and thus competitiveness, i s hampered by shocks. Technologies less vulnerable to prevalent risks inparticular regions can be developed and promoted through appropriate research and extension, by promoting for instance varieties more resistant to water stress or to pests or maturing at a suitable time according to the local weather calendar. Another possible intervention i s facilitating farmers' change to crops less vulnerable to recurrent shocks in the particular location. Pest control and sanitary measures ingeneral are also means to reduce natural shocks. Infrastructure is another area of policy intervention to enhance competitiveness. The importance o f infrastructure on competitiveness i s well known, and we will not expand on it here. The World Bank study o f Mexico's Southern States shows how the lack o f communication infrastructure hampers the development possibilities o f this part o f Mexico (World Bank, 2003a). 76 We have seen that investment in rural infrastructure i s a minor part o f public spending in rural areas, and the same i s pointed out with respect to infrastructure in general in the Public Expenditure review o f Mexico carried out by the World Bank (World Bank, 2004a). 77 5. POVERTYFRIENDLINESSOF RURAL POLICIES In this chapter we examine policies and programs relevant for agriculture and rural development. By design, many policies and programs are not directly addressed to poverg reduction but have other legitimate objectives. Yet, a country with high poverty rates and limited fiscal resources like Mexico may need to focus its public resources more clearly on programs which can help remove the dualism of the agricultural sector and target the ruralpoor. Thus, we look at rural development programs fFom the perspective of the rural poor, asking if there are ways to make themfriendlier to the ruralpoor without prejudice to theirprimary objectives. The mainfindings and policy implicationsofthe chapter are summarizedbelow. Liberalization policies embarked upon in the late 1990s and carried out throughout the 90s tended notto favor the small-scalefarming sector, The 1992 land reforms brought significant benefits to the rural poor, including more security of tenure. The impact, however, on land productivity and benefiting mostly instead the more commercial and export oriented farming sector. Since there was little support for poor farmers to reconvert to more promising crops and take advantage o f new market opportunities, liberalization was not, overall, pro-poor. This points to the need for liberalizationpolicies to be accompaniedby support measures needed for seizing export market opportunities from which small farmers could benefit.Important programs like Procampo andAserca were adopted in connection to the opening of the economy but they were compensatory programs not programs to enhance agricultural investmentsand hence competitiveness. Alianza para el Campo was introduced in 1996 but its subsidies didnot follow arestructuring strategy and were only limitedly oriented to small farmers, particularly during the early years. Nor was the economic opening accompanied by the development o f a sound financial system for rural areas and the promotion o f an agricultural knowledge system friendlyto the poor farmers' incomes was small. A major concern i s the difficulty faced by young ejido residents in accessing land. Agricultural programs have had varied success and are by and large not targeted to the poor. The most important program, Procampo, is comparatively friendly to the poor because of its extensive coverage, but (i) farmers absorb a large majority o f the resources, (ii)landless farmers and agricultural workers are not included, and (iii) the program will be discontinued in a few years, and it i s not clear what policy initiatives will replace it. Nonagriculturalprograms could generally be considered to have been more pro- poor. Programs like Microrregiones, Microcuencas and the decentralization of infrastructure investments have in general favored a territorial and multi-sector approach, making them more successkl in reaching marginal areas and focusing on local priorities. 78 POVERTYFRIENDLINESSOFAGRICULTUREAND LANDPOLICIES Commercial Policies Economic liberalization, including that of the agricultural sector, was a main thrust of economic policy in Mexico since the mid-eighties through the late nineties. Agricultural liberalization was stimulated by GATT, which Mexico joined in 1986, and a number of bilateral treaties inthe 1990s (Japan, EU, some L A C countries and, o f course, in 1994, NAFTA). Most o f the conversion o f quantitative restrictions to tariffs and o f tariff reduction was accomplished before NAFTA. Economic opening was accompanied by internal liberalization o f agricultural prices and marketing, the elimination o f most subsidies, the phasing out o f the state marketing company, Conasupo, and the introduction o f compensatory programs like Procampo and Aserca Price and Marketing Supports, and a farm modernization program, Alianzapara el Campo. What has beenthe effect of NAFTA on the agricultural sector?The main conclusions are summarized inthe following paragraph^.^^ There was a large expansion in agricultural trade with a trend of the agricultural balance of trade to deteriorate (Figure 5.1). Imports, particularly those o f feed grains, increased much, from US$ 3.4 billion in 1994 to 5.9 billion in 2003. As a proportion o f total apparent demand of agricultural goods, imports increased from 23 percent in 1994 to 39 percent in2001. Inturn, exports rose from US$2.7 billion in 1994to 4.7 billion in2003, particularly non traditional exports o f fruits and vegetables (which rose from US$ 1.6 billion in 1994 to US$ 3.3 billion in 2003, see Figure 5.2). The agricultural balance o f trade deteriorated and the agro-food balance (including agro-industrialproducts) deterioratedeven more. Agricultural prices fell markedly in real terms (Figure, 5.6), following rather closely innominalterms the evolution ofinternationalprices. Production of wheat, soy beans and rice fell but not that of maize, contrary to what had been anticipated (Figures 5.4 and 5.5). This was probably due to the fact that (1) maize i s largely a subsistence crop, with a good part o f output being self-consumed or circulating in local markets only, and i s the base o f the rural diet, (2) the stimulus to maize production provided by Procampo and Aserca subsidies, and (3) the fact that maize i s a comparably low risk crop, adapted to most parts o f the country, well known to farmers, requiring simple technology, with low production costs andwell established marketing channels. It i s hence difficult to substitute. Overall production and yields of food crops increased, contrary to expectation, although modestly so (Figure 5.5). As discussed in chapter 4, most o f the increase in production and yields took place in irrigated agriculture, commercial farms, richer regions and export- oriented crops, while most o f the increase in surface occurred inrain-fed areas. There was a clear rise in fruit and vegetable areas, a significant reduction in that o f oil crops, and no obvious difference in cereals, pulses and tubers. Changes were not sufficiently large to constitute a major transformation o f the overall crop mix. As anticipated, agricultural employment decreased (from 8.1 million in 1993 to 6.8 million in 2002), but there was no sign of factor price convergence with NAFTA partners, particularly inwages. 49The conclusions should be considered as a first approximation. More definite conclusions would require decomposition analysis to isolate the impact of commercial policies from that of other variables. 79 Figure 5. 1. Agricultural Trade. Mexico 1980-2003 (Millions Dollars) 6000 ~ 5000 ~ 4000 ~ 3000 ~ m - - 2000 ~ 1000 ~ 0 - 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 Source: World Bank staff calculations basedon SIACOM. Figure 5.2 Exports. Non traditional Agricultural Products. Mexico 1991-2003. Millions of Dollars 3500 3000 ~ 2500 ~ 2000 ~ 1500 ~ 1000+ 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Year +Fruits and Vegetables ~ ~ Source: World Bank staff calculations basedon SIACOM. 80 Figure 5.3. Grain Production Mexico 1980-2002 (Thousand Tons) 1,200 1,000 800 600 400 200 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 &Rice +Soybeans +Oat Source: World Bank staff calculations basedon SIACOM. Figure 5.4. Grain Production Mexico 1980-2002 (Thousand Tons) ..- 5,000 ~~ I ~-.`.`~-.-C*..-~~-C1...)~.. . - I 0 - d M a l r e- -fl- Wheat - - Beans Source: WB staff calculations basedon SAGARPA's agricultural database, SIACON. 81 Figure 5.5. Agricultural Indices Mexico 1980-2002 (1980=100) 180 1M 140 120 ~ I 80 60 40 20 0 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 19992000 2001 2002 -Production Yield - Area Harvested Source: WB staff calculations basedon SAGARPA's agricultural database, SIACON. Figure 5.6. Agricultural RealPrice Index Mexico 1980-2002 (1980=100) 120 100 80 60 40 20 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 Source: WB staff calculations basedon SAGARPA's agricultural database, SIACON. In summary, there is no evidence that liberalization policies had the catastrophic effects on agriculture that many feared nor is there evidence that they served to bring the big leap forward that others hoped for. As shown inthe previous chapter, the evidence points to an unequal distribution o f benefits and losses which generally tended to increase the distance between irrigated and rain-fed areas, between rich and poor agricultural regions, and between commercial entrepreneurs (particularly export oriented) andpoor small farmers. The direct effects o f commercial policy do not seem therefore to have been particularly friendly to the rural poor. There may be indirect, general-equilibrium type effects through other markets, particularly the labor market, linked to trade liberalization that could partly compensate for this. No systematic assessment i s available o f these effects, but the evidence on rural poverty and its evolution presented above seems to indicate that ifthey existedthey were weak. 82 Stronger support measurescould help ensure that small farmers can take advantage of liberalization. The outcomes o f commercial policy must however be examined in conjunction with what happened to other policies that affect them. The opening of the economy was accompanied by important programs like Procampo, Aserca and Alianza para el Campo, but these were not sufficient (or indeed intended) to enhance the agricultural competitiveness o f smaller farmers. Important elements like the development o f a sound financial system for rural areas and the promotion of an agricultural knowledge system friendly to the poor were also lacking. These areas are discussed individually below. A new economic order and regulatory system in rural markets seems to have emerged. Some authors have stressed that deregulation and the withdrawal o f the state from the type of heavy handed rural intervention of the 70s and first part of the 80s (powerful institutions for input provision, output marketing and product support, farmers' organizations co-opted by government, and a large, clientelistic extension system), did not result in an institutional vacuum or a complete retraction o f the state from the countryside, as could be expected. There was a process o f "reregulation" o f markets and occupation o f positions o f economic and social command by new actors (Snyder 200l), which took different characteristics in different regions and economic arenas and resulted in diverse outcomes for the rural poor.50 The federal government itself maintains new forms o f intervention in the rural economy through "new agencies that attempt to reach producers as individuals rather than as members o f organized groups, as in the past. (...) This shift completely changed the terrain o f bargaining between the state and peasant movements, since agencies that had been the main targets for producer movements withdrew from their role as shock absorbers betweenpeasants and the market" (Fox, 1995). Rural Finance5' Mexico's financial markets are shallow compared to other middle-income LAC countries'and the situation is particularly acute inrural areas.52An extremely limited supply o f credit results in credit rationing, with many farmers willing to obtain loans at the going interest rate unable to get them. The 2002 rural finance survey showed that only 6 percent o f farmers and 5 percent of ruralmicro-entrepreneursreceived loans from formal banking institutions. Agriculture accounts for a very small part of bank lending (only 4.5 percent in 2001)' and practically no commercial bank lending goes to small farmers. Lending to agriculture from public and private sources has fallen in real terms since 1996 as can be seen in Table 5.1. Due to the fiscal crisis and to a large default portfolio, lending by BANRURAL, the government bank for agriculture, decreasedmuch inthe last part o f the 90s and early 2000s. Also, the decision was made that BANRURAL would not continue lending to small farmers whose 50 Snyder (2001) illustrates this with a detailed study of the reregulation of the coffee economy that took place indifferent growing areasof Mexico. 5 lMuchof this sectionis based on a study carried out in2002-2003by the World Bank inassociation with FIRA on financial markets and the rural economy (World Bank, 2003). As part of the study a survey was carried out in2002 of credit needs and use of financial services of 1,825 individual farmers, 3,301 individual micro-entrepreneurs inrural towns (of 2,500 to 50,000 residents), 954 agricultural enterprises, and 1,073 non- agricultural enterprises. The survey is representative at the national and regionallevel. S2 Thus, the ratio credit/GDP was 19 percent in Mexico in 2000 compared to 34 percent in Argentina, 46 percent in Brazil and 74 percent in Chile. Financial depth has fallen since the Tequila crisis. Also, private banks do not have branches in 74 percent of Mexican municipalities accounting for 22 percent of the country's population. 83 credit needs would be addressed through special programs such as Opciones Productivas. Commercial banks increased their share o f agricultural lendingupon the reduction o fBANRURAL operations but within a decreasing total (Table 5.1). FIRA's support to private lending through insurance and rediscount schemes o f agricultural loans was not enough to attract major participation o f commercial banks, partly because o f their little specialization in this line o f business and due also to the uncertain situation in the banking sector after the 1995 macroeconomic crisis. Table 5.1. Evolution of Private and Public Lending to Agriculture, 1983-2000 inMillionMxP of 1995 Public Private Total Amount (annual averages) 83-90 23,193 19,130 42,323 90-94 14,736 40,690 55,426 94-96 14,825 43,806 58,63 1 96-00 8,674 26,260 34,934 Distribution (%) 83-90 54.8 45.2 100 90-94 26.6 73.4 100 94-96 25.3 74.7 100 96-00 24.8 75.2 100 Source: Anexo Estadistico del V Informe de Gobierno. 1993 p.374; 1990-2000, e INEGI. Table prepared byA. Yiinez y F. Barceinas Instead, small farmers resort to seeking credit from various saving and loan institutions -informal, government, micro-finance, and private money-lenders. Smaller farmers turn to more informal institutions like Cooperativas de Ahorro y Crkdito, Cajas Solidarias, Uniones de Crkdito, and other civil associations (some o f which are regulated and some o f which are not), to government programs, such as SEDESOL's Opciones Productivas and the Fondos Regionales o f former INI,to various micro-finance institutions, and to private money lenders. Finance availability from these sources i s extremely limited compared to need, however (see World Bank 2003). The distribution o f loans by type o f lender i s shown in Table 5.2. The importance o f trade credit (loans from commercial partners like input suppliers, traders and coyotes) i s evident for both farmers and rural micro-enterprises: it accounts for 42 percent o f all loans to farmers and over 60 percent o f all loans to micro-enterprises. The primary source o f loans for farmers i s thus commercial partners, followed by friends and relatives (measured by number of loans, not amounts) and unregulated non-bank financial institutions (NBFIs). The limited access to private and development bank credit for this clientele is also evident from the table. 84 Table 5.2. Mexico: Distribution of Loans Receivedby Individual Farmers and Rural Micro-entrepreneurs by Lending Source (percentages) Farmers Micro-enterprises Type ofLender Numberof Amount Numberof Amount Loans Disbursed Loans Disbursed Formal Lenders 22.8 52.3 10.5 24.2 Private Banks 2.0 8.7 1.5 7.9 Develop. Banks 5.5 7.2 0.5 1.9 NBFIsRegulated 3.1 21.3 1.8 6.0 NBFIsUnregulated 12.2 15.1 6.7 8.4 Informal Lenders 35.3 15.1 25.7 17.7 Friends& Relatives 28.8 8.4 21.0 15.1 Others 6.5 6.7 4.7 2.6 Trade Credit 41.9 32.7 63.8 58.I NBFIs=Non-bankfinancial intermediaries. Source:WorldBank, 2003: Table 5A.2. Individual farmers and rural micro-entrepreneurs face many credit restrictions, illustrated in Table 5.3. Of all individual farmers interviewed52.4 percent declared to be willing to take loans from formal sources and 50 percent from informal sources. Yet, only 7.1 and 9.8 percent, respectively, applied for a loan. Iti s interestingthat most o f those who applied received a loan (although not always for the full amount requested). One reason why so many potential demanders do not apply i s probably because they anticipate being rejected. Transaction costs are high and may also deter loan application. Lack of information of credit sources and eligibility criteria and lack of local credit facilities may be other reasons for not applying. Research suggests that credit restrictions are depressing investment. Moreover, as i s often the case, credit restrictions appear to apply most to those who are in greatest need, including people facing adverse shocks and who have few assets (Box 5.1). Table 5.3. Mexico: Credit Demand of IndividualFarmers and Rural Micro-entrepreneurs by Type of Source (percentages) Farmers Micro-enterprises Demand Formal Informal Formal Informal N o Demand 46.6 50.0 52.7 52.9 DemandbutnotApplied 46.3 40.2 41.2 35.3 Applied(*) 7.1 9.9 6.1 11.8 Applied and Rejected(**) 16.3 1.1 14.3 1.6 ReceivedLoan(*) 5.9 9.8 5.2 11.6 (*) % oftotal. (**) % ofthose applying. Source:WorldBank, 2003: Table 7A.1. There have been positive policy developments in the last years with respect to rural finance, including regulatory reform. One important step is the recent substitution o f BANRURAL for Financiera Rural. Financiera Rural is designed as a 2"d level lending institution, although authorized to operate at 1St level. Other positive steps were the strengthening o f the regulatory framework through the 2001 Ley de Ahorro y Prkstamo, and the approval in 2003 of norms facilitating the use o f movable collateral. There i s finally the creation o f the Banco de Ahorro Nacional y Sewicios Financieros (BANSEFI) as a support institution for the micro- finance sector, whose focus i s primarily rural. I t is illusory to assume that the ordinary commercial banking system will provide the financial services needed by small farmers, rural micro-entrepreneurs and the rural 85 poor at large. A different system i s required. The new system must move from agriculture to rural credit, and should be able to carry out savings mobilizationandprovide the type o f financial services required by the rural poor (saving instruments, personal loans, insurance, money transfers, etc.). Box 5.1. Causes and effects of credit restrictions, econometric evidence On the basis o f the 2002 rural finance survey an econometric exercise was carried out to investigate the impact on investment o f removing credit constraints. The conclusion was that had there been no credit constraints during the period covered, the percentage o f micro-entrepreneurs and farmers making investments would have been about 12 to 41 and 34 to 41 percent higher respectively. The characteristics o f farmers associated with a greater probability of being credit constrained vis-a-vis formal or informal loans were also examined using econometric analysis. The results were as follows: receiving remittances, selling goods through informaltraders, selling outside the municipality, having irrigated land, having few assets, and being affected by adverse shocks, were factors lowering access to both formal and informal credit. Being a wageworker, selling on credit, and planning to improve one's business were factors which specifically restricted access to formal loans, while having a large family and not having formal savings restricted access to informal loans. Of course, some o f these characteristics correspond to being a poor credit subject (like being able to offer no collateral) while others correspond to strong incentives to demand loans (like having irrigatedlands or suffering adverse shocks). Source: World Bank, 2003. Micro-finance institutions are proving useful in the provision of financial services to the poor, both in Mexico and in other Latin American countries (Bolivia, Ecuador, Peru, El Salvador, Costa Rica, Brazil).53 Micro-finance institutions have faced some difficulties in the provision o f credit to agriculture as a specialized sector, because conventional micro-credit methodologies tend to rely on short-term loans with frequent repayments, which do not fit well the strongly seasonal requirements o f agriculture. There are however microfinance institutions in different countries o f Latin America which have successfdly expanded into agricultural lending, and whose experiences the Mexican government might choose to consider for policy reform (see CGAP, 2004). Based on this experience, expanding micro-finance services in the rural areas of Mexico would be facilitated by the following innovations. First, introducing flexibility in disbursement and repayments to adjust to seasonal requirements. Second, introducing flexible collateral requirements, such as personal and group guarantees, movable assets (like animals, farm machinery and stored crops) andhouseholdgoods, adapted to what rural dwellers may offer. Third, introducing technological innovations that can reduce infrastructure and employee costs, such as smart and debit cards and information technology. Fourth, use to the maximum extent possible existing delivery mechanisms, such as rural post offices, retail stores, rented rooms in schools, government offices, rural clinics, and office sharing with other institutions. This can be combined with the use o f mobile staff and mobile credit/deposit units. Fifth, portfolio diversification across clients, activities and rural communities, using a variety o f lending instruments adapted to different clients so as to reduce risks associated with systemic shocks. Finally, introducing other financial products such as saving deposit, money transfer facilities, and 53 See issues involved and successful experiences in building micro-finance systems in Robinson (2001), Drake and Rhyne (2002), Otero and Rhyne (1994), Ledgerwood (1999), and Rhyne and Rotbalatt (1994). 86 insurance schemes. Judicious government incentives to micro-finance institutions could promote the introduction o f these innovations. The Mexican governmentis making a valuable effort to support the development of rural micro-finance services through BANSEFI, but the repressed demand for financial services in rural Mexico remains large. Isolated rural credit programs from federal agencies such as those under SEDESOL's Opciones Productivas umbrella program are clearly o f great benefitto those who have access to them, mostly capital constrainedruralpoor. Butthese ad hoc credit programs cannot substitute for more comprehensive measures vis-&vis the entire system. The option couldbe considered o f reallocating funds investedinthese programs to provide more global support to the development o f a broader rural micro-finance system, through BANSEFI andFinanciera Rural. The new Financiera Rural could take an active role in promoting a robust and healthy system of rural financial services based on micro-finance institutions. There are successful experiences in other countries (like Indonesia, Thailand, Chile and Brazil) o f development banks setting up micro-finance branches.54Jointly with BANSEFI, the Financiera Rural could also help widening the system o f micro-finance institutions operating in rural areas and providing second level funding support. Medium-terminvestmentfinancing i s a crucial area for rural development where micro-finance institutions often face difficulties, because the structure o f their liabilities makes it difficult for them to tie funds for long periods, and because their usually high interest rates discourage demand. This is an area where the Financiera Rural could prove particularly useful. The specific regulatory needs o f rural micro-finance type institutions should also be assessed, and the possibility should be studied o f adjusting the rules to make them more flexible without harming financial soundness. TheKnowledgeSystem The agricultural research and technical assistance system has seen important progress in recent years invarious areas. Competitive funds have been introduced as a way to allocate research funding, a system to which the government agricultural research institute, INIFAP, adaptedrapidly (Roseboon, 2004). More producer friendly formulas for applied research and dissemination, with the involvement o f producer groups, such as the Grupos Ganaderos de Validacibn y Transferencia de Tecnologia (GGAVATT) have been tried out with success. Unfortunately, no similar experience exists in agriculture (with the exception o f the Clubes de Productores active in some parts o f the Northeastern region) probably because of lack o f institutional leadership (FAO, 2002). There was also progress in changing the structure and corporate culture o f INIFAP in order to reduce its administrative heaviness and bring it closer to the requirements o f a scientific institution, and also to open it up more to collaboration with other research outfits, national and international. Finally, the Ley de Desarrollo Rural Sustentable created a Sistema Nacional de Investigacibn y Transferencia de Tecnologiapara el Desarrollo Rural Sustentable. On the whole, however, little priority and resources were given to agricultural research during the last decade. Thus, for instance, the number o f INIFAP's research staff fell from 2,160 full-time equivalent researchers in 1986 to 1,365 in 1996 and 962 in December 2003 (Roseboom, 2004). Mexico spends in agricultural research some 0.4 to 0.5 percent o f its agricultural GDP, which i s far from the 1percent usually considered as a satisfactory benchmark BRI's Desa Units in Indonesia, BAAC's microcredit unit in Thailand, Banestado Microempresas of Banco del Estado inChile, and Credito Amigo of Banco do Nordeste inBrazil. 87 (IICA 2003). Also, there are no clear and consistent research priorities and guiding policies. In general, there i s insufficient recognition o f the importance of the knowledge economy, the complementarities o f knowledge with other assets, and the high returns o f agricultural research investments (De Ferranti et al, 2002), and there i s little progress in making agricultural research more poverty friendly. The focus of technology transfer policy in the last decade was the shift from the public to the private domain. While middle and large farmers were basically left to procure and pay for their own technical assistance, there was recognition o f the need o f an element o f subsidy for poor farmers. A Sistema Nacional de Extensibn Rural (SINDER) was created in 1996 to provide technical assistance particularly to small farmers. The system met with difficulties due to administrative shortcomings in the selection o f extension staff and their economic incentives, inadequate supervision and technical back up, and little client accountability, and was discontinued by the present government. It has not been replaced by an alternative system. Instead, technical assistance functions and budgetary resources are distributed in different programs, particularly in the rural development sub-program o f Alianza para el Campo. Assistance i s given by private providers (the Prestadores de Sewicios Profesionales, PSPs) certified by SAGARPA, andpaidby the programs. Systems of privately providedtechnical assistanceto smallfarmers havebeen shown to work well in different parts of Latin America but have been less successful in the Mexican context. This is the case for instance in Chile with INDAP, and in the Southern Peruvian Highlands with the Marenass project and its predecessors. In Mexico, however, initiatives were less successful due to (1) institutional and administrative aspects, which have resulted in inadequate incentives, many uncertainties and lack of continuity on the part o f providers, and (2) insufficient accountability to the beneficiary/client. Also, technical assistancei s usually conceived in the limited sense o f providing specialized agricultural expertise to farmers, without connection to managerial, organizational and marketing aspects, and to the multisectoral character o f the small farmer economy. Private providers are not organized in networks, do not enjoy a support system and do not have systematic connections to research outfits. The M&E system o f technical assistance outcomes and impact i s weak. Because o f these problems and the limitation o f funds, the extent to which poor small farmers benefit from technical assistance i s small. Because of budgetary constraints and lack of a comprehensive strategy, INIFAP has not been able to respond to the needs of small farmers through stronger client orientation and a farming systems approach. Nor have environmental concerns been addressed sufficiently, in particular those related to mountainous and other fragile environments. Also, INIFAP has shown difficulties inresponding to farmers' growing demands for innovations in the areas o fmarketing and agro-industrialtransformation (see Roseboom, 2004 and IICA, 2003). The Fundaciones Produce, an initiative to strengthen regional knowledge systems, lack sufficient links with smaller farmers.55Unfortunately, with some exceptions (mostly the 55 This initiative was started in 1996. There are 32 Fundaciones Produce one per state, supported by federal and state funds. They are civil society organizations led by local farmers, and have the objective of promoting and fundingresearch projects demanded by farmers, and acting as centers for the dissemination of relevant agricultural knowledge. The total budget for the Programa de Investigacion y Transferencia de Tecnologia(PITT) whichis the administrative umbrella and funding source of the Fundaciones,was some US$ 30 million in2002 of which25 percent from state governments and 75 percent from the federal government. There is hardly any farmers' participation inthe funding. 88 Fundaciones o f the Northern states, such as those o f Sonora and Sinaloa), the Fundaciones have had limited impact, mostly because o f limited resources, frequent disagreements with the Secretarias de Agricultura o f the states, and insufficient leadership from producer organizations and farmer leaders (IICA, 2003). Also, the Fundaciones rely mainly on INIFAP as provider o f scientific services, and have not enough access to scientists from universities and other research centers who prefer to work in research projects funded by CONACYT. Since INIFAP has insufficient coverage o f innovation in areas o f growing demand from farmers like marketing and processing, there i s some mismatch between farmers' demands and what Fundaciones can offer (FAO, 2003). The research projects financed by Fundaciones have been criticized for their short duration, normally one year, and delays in the disbursement o f funds, both things related to administrative and budgetary problems to be examined in chapter 6. Territorial organizations and organizations consisting mostly o f poor farmers rarely participate inthe Fundaciones, nor do the Prestadores de Sewicios Profesionales, who could be closer to understanding the needs o f small farmers. A proposal for Fundaciones Produce. FA0 (2003) has made an interestingproposal in connection with the Fundaciones Produce. It consists o f converting them into Antenas Tecnolbgicas o f the respective states in charge o f the identification, adaptation, and massive dissemination o f tacit knowledge (for opposition to formal or codified knowledge) that already exists and i s being used by producer organization, agro-enterprises, research institutions and others. This reorientation o f the work agenda o f the Fundaciones would be welcome, particularly if accompanied by organizational changes and the institutional representation of small farmers' interests, ifthe Fundaciones are going to be o f service to the ruralpoor. Experience in Mexico as well as internationally could serve as guidance for policy design in addressing shortcomings in the agricultural knowledge generation and transfer systems. There are successful examples o f promoting applied research oriented to small farmers through competitive funds (for instance in Colombia, Peru and Ecuador), as well as successful experiences with publicly funded butprivately provided technical assistance, which couldbe used for inspiration. Best practices exist also in Mexico whose replicability should be analyzed, for instance with some technical assistance programs carried out by FIRCO in the past, experiences carried out today with Alianza funding, such as that o f the GGAVATTs quoted before, and valuable experiments from NGOs and Sociedades de Produccibn Rurales in differentparts o f the country. Even if the assistance i s given by private suppliers, there i s an important role for government, not only in the funding but also in helping developing or adapting technology transfer models appropriate for differentzones andtype of producers, assisting inthe organization o f technical networks o f providers, and facilitating the link with research outfits. Land Policy Major land policy reformswere carried out in the 1990s. The key land right reforms are discussed in Box 5.2. The reforms were exclusively oriented to the social sector -ejidos and comunidades- and were o f great importance given the size o f this sector, which covers more than half o f the country's agricultural land and nearly four million landed families.56With important 56 According to the 2001 Ejido Census, there were 30,305 nticleos agrarios of which 27,786 ejidos and 2,519 comunidades, with an average of 3,467 ha and 127.8 members with full rights, making a total of 105,067,435 ha and 3,872,979 landed members. In addition, there were 957,638 occupants who were not ejidatarios but whose land rights were recognized by PROCEDE. According to the 1991 agricultural census, ejidos had 51% of the country's agricultural land. The labor force inthe ejido sector was 2.3 times that of the private sector, 89 exceptions, ejidos are land poor (in quantity and quality) and have low farm technology, widespread poverty and an aging population. Rural poverty in Mexico i s mostly housed in ejidos and comunidades. Legal and administrative reforms in the social sector have brought benefits to a large portion of the rural poor. Inparticular, they: (1) gave more security o f tenure and freedom to decide on their lands to the majority o f the small farming population, with Certificados de Posesi6n beingissuedby PROCEDE to more than 3 millionhouseholds; (2) gave also security o f tenure to close to one million land occupants (Posesionarios); (3) improved conflict resolution and social peace in rural areas; and (4) improved the functioning o f land markets in the social sector (World Bank, 2001a). Also, the application o f PROCEDE was accompanied in many ejidos by the division o f part or all collective lands and their distribution to ejidatarios as individual plots. It has been shown (see Muiioz-Piiia, de Janvry and Sadoulet, 2003) that this allocation was equalizing since it benefitedproportionally more ejidatarios with smaller holdings as well as those of indigenous ethnic origin. Division o f collective lands was also usedto allocate agricultural plots to landless ejido residents and include them as e j i d a t a r i o ~ . ~ ~ Box 5.2. Land Policy Reformin the 1990s The reform of land rightsin 1992 centered on the change o f article 27 o fthe Constitution and the land law that followed. The main changes were: (1) the agrarian reform was formally closed, thus ending the possibility o f land being expropriated for this purpose; (2) land rights in the social sector (ejidos and comunidades) were improved and made more transparent; (3) a judiciary system o f specialized courts (tribunales agrarios) was set up to rule on land disputes in the social sector (previously dealt with by govemment); and (4) a sort of ombudsman institution was created, the Procuraduria Agraria, to defend the rights and serve the legal needs o f ejidatarios, comuneros and other small farmers. Other than this, a land titling and registration program o f social sector lands, PROCEDE, was launched and has beenvery active, and a national registry for these lands, the Registro Nacional Agrario, was created. There are three types o f land rights in the ejido: (1) homestead plots with full, unrestricted ownership rights; (2) farming plots individually owned by the ejidatarios, which have some restrictions; and (3) collective lands (usually forest and grazing areas) for communal use, without separate rights. Farming plots (average 5 ha per ejidatario) can be rented in and out without restriction, freely sold to other ejidatarios, transmitted as inheritance, and used to constitute joint ventures with private capital. Farming plots, however, have three legal restrictions: (i) cannotbesoldtonon-ejidatarioswithoutpermissionfromtheejido'sgoverningbody(the they members' assembly), (ii) be parceled up (upon inheritance or otherwise), and (iii) cannot cannot exceed certain size limits.Ejidos can change to a full private ownership system if 213 o f members so decides. So far, less than 1% o f ejidos, mostly periurban ones, have usedthis legal provision. Land policy issues are complicated by the fact that the ejido (like the comunidadj` is both a land tenure system and a form o f social organization. As a land tenure system it consists o f a large tract o f land with a collective title issued to a community o f beneficiaries who practice individual farming and have collective use o f forest and grazing lands. As a social organization it i s a system o f village governance and constitutes a form o f social capital in rural areas. Institutionally, the ejido i s the historical product o f the Mexican Revolution and the reparto agrario (land distribution), making its reform a politically complex issue. Source: D e Janvry et al, 1998, World Bank 1999 and 2001a, Hollinger, 2004. and ejidatarios had around 113 of the surface per producer and 113 of the heads of cattle per producer of the private sector. s7 There is a large literature on the nature and effects of the ejido reforms. An interesting reading is the collection of monographic studies edited by Cornelius and Myhre (1998). 90 The impact, however, of the reforms on land productivity and farmers' incomes in the social sector was small. After more than ten years since the reform, profitability in agriculture in ejidos or comunidades continues to be low. As indicated in chapter 4, we have made an exercise with ENHRUMdata to examine the profitability o f farm enterprises according to tenure system, measuring profitability with value added per ha and gross benefit per ha (including as cost the imputedvalue o f family labor, and without including it). On either measure farms inejidos and comunidades are less profitable than private farms as shown inFigure5.7 Figure 5.7 Gross Margin per Hectare with Family Labor (GP2) and without Family Labor (GP1) and Value Added F A ) per Hectare in2002 5,000.00 I 4,000.00 3,000.00 2,000.00 P 1,000.00 ............ ............ ............ 0 -1,000.00 -2,000.00 -3,000.00 ' Ejidos Communal Private Mixed Tvpe of Ownership OGP1 B G M 2 K l V A Source: WB staff calculations based on ENHRUM. There are essentially three reasons for the lack of impact of reforms on employment and income opportunities. First, property right changes do not create per se economic development, although they may open the way to it. The reforms were surrounded o f excessive optimism in this respect. Second, little was done to accompany the reforms with the complementary investments and support systems requiredto improve land productivity and farm incomes. Finally, under existing business circumstances in agriculture, private capital was not interested in the ejido sector with the exception o f particularly attractive peri-urban or irrigated ejidos. New major changes inland legislation do not seem politically feasible at present, and it is questionable that they are truly needed at this stage. For the present legal framework to bring important improvements, the following challenges would have to be addressed: There i s a distinct ageing of the farming population inthe ejido sector. According to data from PROCEDE, land right holders in regularized ejidos have on average 54 years, 60 percent have more than 50 years, and 29 percent more than 65 years. Ageing reduces efficiency in land use since the elder tendto make a more extensive use o f the land and are generally less efficient farmers. This has been observed in field studies (Eduardo, L e Moing and Gonzalez, 2004). 91 The positionof young residentsof the ejidos, who are more educated than their parents but have no access to land, is worrying. Sons and daughters o f ejidatarios cannot easily find employment inthe ejido and hence many are forced to migrate. As residents o f the ejido but not ejidatarios, these young workers are a discriminated group without voice or power in ejido decision-making. The market i s failing to transfer ejido farm land from low productivity users (mostly the elder) to potentially high productivity users (the younger) in a satisfactory manner, and young ejido workers suffer from it. The legal norm preventing the fractioning of the land titled to ejidatarios obstructs land sales and forces ejidatarios to pass it to one heir only, generating efficiency and welfare losses. The rationale behindthe norm is to maintain farm size and ifpossible expand it through the consolidation o f holdings in order to promote a class o f viable farms within the ejidos. Small farmers, however, have a multisectoral economy. The important thing for them i s not viable farms but viable multifunctional family enterprises, o f which part-time farming can be a component. Relatively small farms can play crucial complementary roles in the economy o f rural families. Part- time farming has proven to be efficient (interms of both product per ha and family welfare) in many parts o f Europe. Finan, Sadoulet and de Janvry (2003) have shown the importance o f the welfare effect o f even small plots o f land in Mexico, an effect very much enhanced by complementary assets andgood location. Finally, concern is often voiced about the impact of reforms on the system of local governance, inview o f the double nature o f the ejido mentioned above. This is very much linked to the issue of the erosion of the social protection and equity functions o f social property, and i s addressedin chapter 7. The land-shortage of young landlessfarmers is being addressedthrough a new and promising program. The Secretaria de la Reforma Agraria (SRA) i s starting a program to promote the access to land o f young landless farmers through buying and renting, as well as to complementary investments, using as instrumentsa land fund and an investment fund. This i s a welcome response to the above situation particularly in view o f the importance that young workers may have inthe taking off o f the rural economy. POVERTY FRIENDLINESSOF SPECIFIC AGRICULTURAL PROGRAMS Alianzapara el Campo Alianza i s a demand-driven investment support program aimed at farm modernization, which started operating in 1996. It i s a complex program consisting o f various subprograms with different rules each that have changed much over the years, but it mostly operates as a matching grant subsidy.58The program i s large, with a total budget o f some US$ 1.1 bnin2003, some 30 percent ofwhich were contributedby farmers. Alianza is not particularly focused on the poor, with the possible exception of the sub-program Desarrollo Rural. Alianza was not designed as a poverty reduction instrument. Initially it was oriented to the type o f asset investmentspreferredby large farmers, like tractors, pressurized irrigation equipment, electric fencing, and others, clearly catering for the more 58The main ones are the Programa de Desarrollo Rural (36.5 percent of total program cost in 2003), the Programa de Fomento Agricola (36.7percent), and the Programa de Fomento Ganadero(17.2 percent). 92 commercial sector o f the farming community. Over time, however, it moved to address also the needs o f smaller farmers, particularly through the Desarrollo Rural s~bprogram,~~which in 2003 accounted for 36.5 percent o f the cost o f the entire Alianza. Alianza subsidies are not registered separately in ENIGH surveys (contrary to Procampo 's and, since 2002, Oportunidades ' subsidies), and hence we do not know how much reaches the poor. FAO's evaluation classifies beneficiaries in five types according to education, farm size, number o f animals, value o f productive assets, and percentage o f output sold, as shown in Table 5.4. (FAO, 2003). The first type and perhaps part o the second can be identified as "the poor". According to this, 32 percent as a maximum of the beneficiaries o f Alianza in 2002 were poor (Table 5.5), but the figure i s probably lower since not all Type I1beneficiaries would be poor. The Desarrollo Rural program i s clearly more targeted, but even so at least 47 percent of the beneficiaries are non poor. Figures were not available o f the breakdown of program expenses by type o f beneficiary, but the distribution must be worse than that o f beneficiaries since beneficiaries classified in higher types typically received larger amounts per capitathan those inlower types. Table 5.4: Type ofBeneficiariesofAlianza para el Campo VARIABLE TYPE I TYPE I1 TYPE I11 TYPE IV TYPE V Schooling Primary Secondary HighSchool University Universitytitle (Years) (1-6) (7 - 9) (10-12) (13-16) (17 or more) Surface Less than 3 3 to 10 10to 50 50 to 100 More than 100 (has) Heads of cattle Less than 5 5 to 10 10to 50 50 to 100 More than 100 Value of productive Less than 5,000 5,000 to 25,000 25,000 to 100,000 to More than assets (MxP) 100,000 500,000 500,000 Percentage of Less than 20% 20% to 40% 40% to 60% 60% to 80% More than 80% output sold Source: FA0 (2003). Table 5.5: Percentageof BeneficiariesofAlianza para el Campoby Type of Beneficiary Programs Beneficiaries Beneficiaries Beneficiaries Beneficiaries Beneficiaries Type I Type I1 Type I11 TypeIV Type V Rural Development 23 30 33 13 1 Agricultural 4 20 43 27 6 Development Livestock 3 13 41 36 8 Development AllAlianza 11 21 38 25 5 Source: FA0 (2003). 59 This in turn consists of three subprograms: (i) Programa de Apoyo a 10s Proyectos de Inversion Rural the (PAPIR), an investment fund to finance productive investments, (ii) the Programa de Desarrollo de Capacidades en el Medio Rural (PRODESCA), a trainingbuman capital development program, and (iii) the Programa de Fortalecimiento de Empresas y Organizacion Rural (PROFEMOR), a social capital program oriented to the strengthening of farmers' organizations. 93 Several aspects ofAlianza preventit from being a progressiveprogram,whatever its other merits. First, as we have seen, it is mostly oriented to the non-poor. The main reasons for this are (i) untargeted eligibility criteria, (ii) producers with some capital or access to wide, only finance can put up the counterpart contribution required, (iii)asymmetric information: insufficient dissemination o f the scope and rules o f the program makes larger farmers more likely to be better informed, and (iv) the technical assistanceprovided with the program i s more suitable to the needs o f the more commercial farmers (FAO, 2003). Geographical targeting i s also lacking. Highmarginality municipalities (identified by the CONAPO index) have not been a priority in practice, although the rules o f the Desarrollo Rural subprogram indicate that 70 percent o f the funds should go to them.60Finally, like in other agricultural programs, landless farmers are left out. The program could be made more poverty friendly without major change in its objectives. One way would be through improved selection by filtering out better off farmers likely to carry out the investments without program support. Assistance could be provided to these farmers by helping them obtaining investment credit from established sources through risk sharing or other mechanisms. Another way would be to enhance the Desarrollo Rural program and redesign it using a territorial approach to rural development. Other ways are through better dissemination, more focus on marginal areas, and a better link o f Alianza supports to small farmers with their access to rural finance so that poor farmers without counterpart funds can participate. A decentralization of the program brings both risks and opportunities. Some 25 percent o f the Desarrollo Rural funds were being decentralizedto municipalities in2004, and the present administration intendsto fully decentralize the program by the end o f its mandate in2006. This bringsbothrisks and opportunities and it i s not clear what the effect would be. On one hand, local authorities could be better placed than central ones to identify priorities. On the other hand, local authorities are experienced in local infrastructure and service investments but not in productive development projects, an area in which they have not been traditionally involved. There i s hence risk of atomization and lack o f focus o f the investments.The implementation o f a territorial approach to rural development, explained inAnnex 3.I., would reduce this risk. Procampo Procampois a cash transfer programlaunchedin 1994 to compensate grain farmers from income losses due to increased competition brought by NAFTA. It is thus an income support rather than poverty program, and i s supposed to finish in2008 when the transition period expires and there i s full market integration. The only conditionality i s planting o f eligible crops (mostly grains61) three years before 1994, and continued cropping o f the land. The subsidy i s given to the user of the land, not necessarily the owner, and is seasonal, so that double cropping farmers receive it twice in a year. The payment i s o f US$ 80 to 90 per hectare and per season. There i s no limit to the number o f hectares for which a farmer can receive the subsidy other than the constitutional land ceiling62(Davis, 2003). LikeAlianza, Procampo i s a large program with a budget o f some US$ 1.2 billion in 2003 benefiting an average o f some 2.8 million farmers per 6o Thus, for instance, in 2002 only 15 percent of Papir funds (the core subprogram of the Desarrollo Rural Program)went to these municipalities. 61 There was some widening of crop eligibility over the years. The crops currently eligible are maize, rice, wheat, barley, sorghum, beans, soybeans, cardamom and cotton. 62InMexicothere is a constitutional limit to land holdings of 100hectare of irrigated landor the equivalent inlessproductive lands. 94 year with coverage o f some 14 millionhectare. Payments are made directly to farmers included in the Procampo registry, with bank checks, upon verification that the land i s cultivated. It i s a very popular program among farmers and in rural areas in general. Table 5.6 shows the evolution o f Procampo's budget, subsidy per hectare, surface covered, andnumber o f beneficiaries. Procampo has some pro-poor features, including wide coverage and focus on the social sector. Procampo marks an attempt to move from a conventional price and marketing support system to a more decoupled one. Although not a poverty targeted program, it i s a considerable improvement equity-wise compared to the regressive system of protection that it replaced. The most positive feature o f Procampo for the small farmers i s its large coverage, which, even if the program i s not targeted, allows the vast majority o f poor farmers, most o f whom produce grains, to benefit from the subsidy in a stable and predictable way (according to the 1997 ejido survey, some 85% o f ejidatarios received the subsidy). Procampo also releases somewhat the severe financial constraint faced by farmers, thus facilitating on and off-farm productive investments, which have an income multiplier effect.63 This effect, however, i s not equitably distributed, being more powerful for large farmers. Finally, the program has the interestingbut little noticed effect o f drawing poor farmers closer to the banking system and the use o fbanking services. Table 5.6. Procampo: Budget, Subsidy and Hectares Covered, 1994-2003 1994 1995 1996 1997 1998 199 2000 2001 2002 2003 BudgetWxPmn) 4,848 5,864 6,793 7,533 8,492 9,372 10,379 11,005 11,851 14,191 Subsidy(USDha) Fall-Winter 98.3 63.9 58.4 61.4 61.9 66.2 75.0 83.9 86.4 82.1 Spring-Summer 114.4 96.0 66.6 71.2 76.5 70.6 82.6 86.7 94.8 86.8 Surface (000 ha) 13,625 13,321 14,306 13,885 13,869 13,528 13,571 13,420 13,698 13,900 Beneficiaries (000) 3,295 2,934 2,987 2,850 2,780 2,724 2,681 2,695 2,792 2,800 Social sector n.d. 2,445 2511 2,390 2,343 2,320 2,265 2,267 2,348 2,352 Privateowners n.d. 432 419 405 385 371 365 376 391 392 Mixedownership n.d. 57 57 55 52 51 51 52 53 56 Source: IICA (2004: 279) Based on figures from ASERCA. Income from Procampo is important for the rural poor. In2002 it accounted for 4.7 percent o f family income o f the bottom quintile o f the income distribution o f the disperse rural population, and 5.5 percent o f that o f the extreme rural poor. The share o f Procampo income in poor farming families i s o f course much higher. A study by Davis, Handa and Soto (2001) estimated an impact o fProcampo o f five points inthe national poverty rate in 1996. The program also has some less favorable features fromthe point ofview of poverty friendliness: (1) it bypasses landless farmers, and (2) since land i s unequally distributed and the subsidy is proportional to the size ofholdings, the distribution ofbenefits is biasedtowards larger holdings. Seasonal payments favor also farmers with access to irrigation, who can double crop and are normally better off than those who cannot double crop. As seen in Table 5.7, in 2004 farmers with land-holdings o f less than 2 hectare received only 13 percent o f all subsidies, while one third o f all funds were allocated to land holdings above 18 has. In terms o f consumption expenditure deciles, in 2002, the first three deciles received 20.8 percent o f the subsidy while the top decile received 29.2 percent.64The moderately poor received 53.2 percent o f the subsidy and the non-poor 46.8 percent, but only 26.4 percent o f the extremely poor (who typically are land- 63 This is partly because of the incremental cash and partly because Procampo payments can serve as collateral for loans. By comparing the ejido surveys of 1974 and 1997, Sadoulet, de Janvry and Davis (2001) estimated this effect to be between 1.5 and 2.6 for ejidatario farmers. 64Staff calculations based on ENIGH(2002). Rural defined as localities with less than 15,000 residents. 95 less) receivedthe subsidy (Figure 5.8). This said, Ruiz et a1 (2002) have shown a positive impact of Procampo on nutrition, similar per MxP received to that of Progresa, while a study by Gonzalez- Konig and Wodon (2003) found a negative impact on migration, both temporary and permanent. There havebeenimportantimprovementsin the design ofProcampo in recent times that havemadeit friendlierto the poor.First, farmers with less that one hectarereceivethe full payment correspondingto one hectare, andfarmers with less than 5 hectarereceive apayment per hectare above the Second, there has been an improvement in timing, which gives producers the possibility of receivingthe subsidy before plantingso that they can use itto pay for crop inputs.Finally, a longawaitedscheme was introducedwhereby farmers can receivecash the discounted amount of the remaining subsidy, provided they employ that cash in a productive project. Table 5.7. Procampo: BeneficiaryDistributionby Size of Holdings Producers Hectares Size of Holding Number Percentage Number Percentage O t o l 648,785 23.7 594,305 4.4 1to 2 672,794 24.6 1,201,075 8.8 2 to 5 804,6 13 29.4 2,836,562 20.8 5 to 10 391,776 14.3 2,925,391 21.5 10to 18 113,821 4.2 1,543,079 11.3 18 to 100 97,906 3.6 3,3 84,032 24.8 More than 100 6,375 0.2 1,148,490 8.4 Total 2,736,070 100.0 13,632,934 100.0 Source: IICA (2004: 280) based on ASERCA data. Figure5.8. RuralMexico: Share ofProcampo transfers by 1 Poverty Condition, 2002 26.5 40% 73,5 30% 46.8 20% 10% 0% "; II 53 2 Extreme poverty Moderate Poverty Notes: (1) Rural areas defined as localities with less than 15,000 residents. Source: Calculated from ENIGH2002. 65MxP 1,030 per hectare against MxP905 per hectareinthe 2003 spring-summer cycle. 96 According to its originalplan, Procampo should be discontinued in 2008. Inview of the popularity and many functions o f the subsidy in the rural economy, it i s difficult to imagine that this will be politically and socially feasible. The design, however, may change. One possibility i s to substitute Procampo with a different type of subsidy. Thus, SAGARPA i s considering the possibility o f introducing some kindo f social security system for rural elders as a replacement ofProcampo. AsercaPrice and Marketing Supports Aserca (Apoyosy Sewicios a la ComercializacidnAgropecuaria) was created in 1991 to support the marketing of surpluses in an open market environment. It is a government entity operating under SAGARPA. Aserca operates the Apoyos a la Comercializacibn y Desarrollo de Mercados program, which was introduced to ease marketing problems o f surplus producers o f grains and strengthen their capacity to compete with imported produce after liberalization. Until recently, the program consisted o f a subsidy for surplus farmers (mostly medium and large ones) covering the difference between the local price and the import parity price. The objective was that all beneficiary farmers received the import parity price irrespectively o f transport costs, marketing failures and differences inmarketing costs. The volume of the programhas almost tripledsince the mid 1990s,the volume of produce coveredincreasing from 4.5 milliontons in 1995 to 12.1 millionton in2002, andthe subsidy from MxP 806 millionto MxP 4,005 million, as shown inTable 5.8. Aserca subsidies are regressive. There were some 67,000 beneficiaries in 2002 who received an average support o f around US$ 5,200 each, i.e. approximately four times the moderate poverty income line. This suggests that the subsidy goes to medium and large commercial farmers. Also, it mostly benefits rich agricultural states as shown inFigure5.9. Thus, in 2002, Sinaloa, Sonora, Tamauleas and Guanajuato received among them more than 80 percent o f the support. Table 5.8. ASERCA MarketingSupport Program,1995-2002 1995 1996 1997 1998 1999 2000 2001 2002 Volume (000 tn) 4,528 1,726 8,091 6,531 5,915 8,758 16,052 12,085 Value (million MxP) 806.5 555.4 2,068.0 1,930.6 1,573.6 2,928.5 5,133.5 4,005.0 Subsidy inMxPper tn 178.1 321.8 255.6 295.6 266.0 334.4 319.8 331.4 Source: Presidencia de la Republica. I11Informe de Gobierno. Anexo estadistico 2003. There have been some changes and improvements in recent years. Support has been introduced to compensate for the price fall effect due to international competitors' subsidies. This i s done through a system o ftarget income for each product. Also, a 5-year horizon was adopted to give more price certainty to producers. Specialprograms have been set up to promote shifts from excess supplyto excess demand crops, and to promote the rediscount ('pignoracibn) o f warehouse certificates o f deposit. Poverty aspects aside, the program has some weaknesses. Although Aserca supports play an important role in allowing Mexican commercial producers o f import-competing grains to compete in better terms, and compensate somewhat for price reducing practices from trade partners, their efficiency and effectiveness may be called into question. One issue refers to the efficiency issues linked to the attempt at pan-territorial pricing implicit in the subsidy. Another refers to the actual possibility o f replacing the compensatory subsidy by border protection measures, thus shifting the burden o f the subsidy from the taxpayer to the consumer. More 97 generally, it can be asked to what extent a middle income, fiscally starved country with high poverty rates like Mexico can afford in the long run a heavy subsidy program that does not benefitthe poor andi s not conditional on improvements incompetitiveness. Figure 5.9. Distribution by State of ASERCA Marketing Supports in2002, Percentages 45 40 35 20 15 Amount 0 Producers Source: World Bank staff calculation form ASERCA data. POVERTY FRIENDLINESSOF NON-AGRICULTURAL RURAL DEVELOPMENT POLICIES There are three important trends in rural development policies outside agriculture: (1) decentralization o f social infrastructure, which has resulted in an increased role o f both state and municipal governments, (2) the promotion o f territorial development agendas through federal programs, (3) andthe introduction o f direct cash transfers to the poor.66 Decentralizationof RuralInfrastructure Rural social infrastructure funding has been decentralized. Decentralization is important inmany areas o f social provision for the rural poor, but we will be concerned here with social infrastructure only. Since 1998, the instrumentfor decentralizing social infrastructure i s the Fondo de Aportaciones para la Infraestmctura Social (FAIS) o f Ramo 33 o f the federal budget. Municipalities are the main beneficiaries o f social infrastructure decentralization, since most o f this fund (88 percent in2000 according to the Encuesta Nacional sobre Desarrollo Institucional MunicQal 2000) goes to municipal authorities under a sub-fund called Fondo para la Infrastructura Social MunicQal (FISM), and the rest to state governments. On average, FISM accounts for 27 percent o f municipal resources, but its importance i s much larger for rural and semi-rural municipalities for which it goes from 25 to 75 percent (Cabrero Mendoza, 2002). The largest part o f investment (around 22 percent) goes to urban development and the smallest parts 66Educationand health policies are not considered here since they are included inWorld Bank (2004) 98 to housing improvement and productive infrastructure (around 3.5 percent each). Resources are dividedbetween the municipal capital and other municipal areas invaried proportions according to municipalities but, in general, richer (i.e. less marginal) municipalities tend to devote proportionally more resources to the municipal capital (Gonzalez de Alba y Garza Navarrete, 2002). The modification of the Ley de CoordinacidnFiscal in 1997 and the introduction of Ram0 33 and FISM brought two improvements in the distribution of resources for social infrastructure: (1) the allocation o f funds among municipalities according to objective criteria based on municipal needs -the carencia munic@al-, which prevents arbitrariness and clientelistic distortions, and (2) the larger autonomy in investment decisions that the present policy gives to municipalities. Some rural municipalities have a system in place that allows the participation o f civil society in establishing investment priorities, mainly through municipal development councils operating alongside the cabildos as consultative bodies to the presidentes munic@ales. Some o f these councils are the heirs o f the municipal Consejos de Planeacibn del Desarrollo (COPLADEs) created during the Salinas administration, others are more recent. In some municipalities there i s a Plan de Desarrollo Munic@al approved by the cabildo, where the investmentsare included. Yet, many rural municipalities do not have such councils and municipal plans, and the decision on the use o f FISMfunds i s taken by thepresidente municeal. In general terms this system can be considered poverty friendly, but some improvements can neverthelessbe made. To the extent that (1) the allocation o f funds i s made according to municipal needs measured by poverty indicators, and (2) the local use o f funds i s more efficient and transparent and reflects people's priorities better, the current system can be considered poverty friendly. Some improvements, however, are possible. The first i s to introduce an independent monitoring and evaluation (M&E) system for FISM, which does not exist at present. It could consist o f either independent systems operated by the states or one single federal system. The latter seems a better choice because o f the advantages o f having centralized M&E mechanisms for decentralized operations. Another improvement would be the systematic introduction o f participatory approaches for establishing investment priorities and to ensure transparent accountability o f the use o f funds. Finally, it would be desirable that FISM funds are not used in isolation o f other sources o f development assistance to municipal areas from federal or state programs or other sources. An integrated approach for the municipal territory, consolidated in a strategic municipal program designed in a participatory manner, would be the best way to exploit synergies between differentprograms and organize the municipal demand for development assistance. Promotion of TerritorialDevelopment Several federal programs have tried or are trying to create territorial development systems in rural areas. We can single out among them the Programa de Areas Marginales, implemented by SAGARPA, which was developed during the Zedillo administration, the Microrregiones program, implemented by SEDESOL, and the Microcuencas program, implementedby FIRCO, the latter two startedunder the Fox administration. Also important to the promotion o f local development inrural areas i s the Ley de Desarrollo Rural Sustentable, passed inDecember2001, andthe sustainable ruraldevelopment councils whose creationis mandatedby the Law. The Marginal Areas Program was an attempt to achieve rural development in marginal zones through the promotion of productive activities.It started in 1997 operated by SAGARPA in the framework of the rural development subprogram of Alianzapara el Campo, 99 and was supported by a World Bank loan. Six marginal areas were selected for a first phase, 3 in Oaxaca and 3 in Las Huastecas (in the states o f Veracruz, Hidalgo and San Luis Potosi), and another 11 for a second phase in the states o f Chiapas, Puebla, San Luis Potosi, Guerrero, Michoachn and Veracruz. Matching grants were givento farmers to invest inhomestead orchards and small animals, agriculture and livestock projects, land conservation, and transformation units. Funds were also available for extension and training, institutional support, and communal activities and projects. The main novelty o f the program was institutional and consisted in the creation in each region o f a Consejo de Desarrollo Rural o f mixed public-civil society composition. According to original design, these councils were to be largely autonomous and strongly empowered with decision making capacity in investment and technical assistance matters, thus capable o f playing a crucial role in territorial development as core meso-level economic coordination institutions. The program is now terminated, but provided some important lessons.67There were four main limitations some o f which were slowly overcome as the learning process went on. First, the political will to construct truly representative and autonomous councils and empowering them was not always there. Political traditions, administrative limitations, and a less than full understanding o f the advantages o f community-driven development, particularly by state governments, which had a strong say in the implementation o f the program, were the main reasons for this. Second, the program focused primarily on agricultural development and was not really multisectoral. Also, shared long term visions for the regions future and the strategic programs to carry them out were not constructed. Third, there were problems with the extension system that was put in place under SINDER, which lacked continuity and client orientation, and was not up to the difficult technical and organizational challenges posedby poor farmers inthese regions. Finally, there were administrative and financial management limitations (timing in the liberation o f funds, eligibility o f investments, restriction on certain categories o f expenditures, heavy ex-ante controls, and the like), which did not help implementinga program like this which requires plenty opportunity and flexibility in expenditures. Notwithstanding the above limitations, which should be seen as a reflection o f the learning process, the program helped improving the incomes and assets o f many poor farmers inthe targeted areas. Microrregiones i s another federal program oriented to marginal areas. It is implemented by SEDESOL, started operating in 2001 in the framework of Contigo, and is currently on-going.@Microrregiones i s not only a program but also an attempt at developing a new strategic approach to rural development. Activities mainly concentrate on municipalities o f high and very high marginality measured by the CONAPO marginality index. There are 1,338 municipalities inthese categories, which have been grouped into 263 micro-regions. The strategy has three basic features: (1) investments are not dispersed through the micro-region but concentrate on specific localities, the so-called centros estratkgicos comunitarios, assessed to have the potential to act as development poles within a certain area o f influence; (2) although the program has its own development budget, it mostly tries to promote and coordinate investment and service provisionto the Centros comunitarios by other government bodies;69(3) achievement i s measured on the basis o f the supply o f missing development services or assets -the so-called banderas blancas. Since the end o f 2003 the formation has been promoted o f Agencias de Desarrollo Local, consisting o f professionals o f civil society or rural organizations who 67 The program was discontinued in2003, and the World Bank project supporting it closed. 68 Microrregiones can be seenboth as a government program with operation norms and a budget line and as a strategic perspective of the rural development process and attempt to coordinate efforts of different parts of government around that perspective. 69 There are 14secretarias involved inthe program plus state governments and municipal authorities. 100 encourage and assist endogenous processes o f territorial development. Up to the end o f 2004 there were 81 o f these groups operating in 81 microrregiones. They are supported and monitored bythe Microrregiones managmentunit. While the project is pro-poor in targeting, some challenges remain in terms of program implementation. Focusing on marginal areas, working with a territorial approach, clustering investments to achieve critical masses o f infrastructure and services, and trying to coordinate the local development activities o f different agencies, are all very positive features o f the program that make it poverty friendly. Program implementation, however, has come across some difficulties. Thus, it has proved difficult to marshal1 the type o f strong commitment from other federal agencies, state and local authorities and the local private sector and civil society necessary for successfd territorial development. One reason i s that Microrregiones i s very much associated in people's minds with SEDESOL, and thus seen as its more or less exclusive responsibility. Also, some technocratic features in the program and some top down approaches, mostly associated with the desire to achieve quick impact, do not help attracting local commitment. It was also difficult to build strategic plans for the micro-regions capable o f attracting the imagination o f the local private sector and civil and political society. A shared strategic plan for the region built from below would help integrating program investments into coherent strategic plans, and setting in motion endogenous growth processes (see Annex 3.1. on this). This couldbe done inthe framework ofthe Ley deDesarrollo Rural Su~tentable.~~ Instead, investmentstendto consist o f the juxtaposition o f actions by different agencies based on a check lists o f what micro-regions are supposed to have as development assets and services. Finally, most investmentsgo to social infrastructure and services, and only a minority to employment and income generation projects. This i s not surprising since promoting local productive projects requires much work with local communities in the identification o f opportunities and the accompanying o f the processes. This in turn requires two resources in short supply to the program: a large runningbudget, and a large number o f specially trained, highly motivated staff capable o f building virtuous cycles o f public-private synergy (see Evans, 1996, and Tendler, 1997, on buildingpublic-private synergies). The Plan Nacional de Microcuencas has a micro-territorialdevelopment approach. The program, administered by FIRCO, centers on small basins o f some 4,000 ha and 1,300 residents each on average. It i s a low budget71and low profile but effective program from which interestinglessons can be learnt. The program i s national in scope and currently operates in 1030 microcuencas located mostly in the middle and upper ranges o f the respective municipalities, where the concentration o f rural poor tends to be larger, covering some 1.3 million people. Although operational in all states, it i s particularly active inAguascalientes, Chiapas, Coahuila, Guanajuato, Guerrero, Jalisco, Oaxaca, San Luis Potosi, and Veracruz. Implementation agreements have been signed with 325 municipalities, 13.4 percent o f the 2,427 Mexican municipalities. 70 The external evaluation of Microrregiones says the following in this connection: "La Ley de Desarrollo Rural Sustentable constituye una vision avanzada con relacion a1 desarrollo rural y representa una plataforma -que tiene la ventaja de tener ya el caracter de ordenamiento legal- para la articulacion de las politicas gubernamentales de atencion a1 campo, donde se hace knfasis en las regiones de mayor rezago social y economico; esta Ley no vale solo para SAGARPA pues integra economia, sociedad y medio ambiente e instituye a 10s Consejos como mecanismos de concertacion. Sin embargo la estrategia de Microrregiones, que es de caracter rural, ni siquiera la considera, evidenciando una balcanizacion de la administracion publica altamente lesiva cuando el desarrollo social se quiere economicamente sustentable y en el campo esto es imposible sin la coadyuvancia de SAGARPA (Instituto Maya, 2003: XVII). 71 FIRCOs2003 total budget for this program was MxP 38.4 million. 101 The program operates under several principles which can be summarized as follows: (1) natural resources conservation is an important area of intervention but emphasis is also given to economic, social and human development within an integrated approach; (2) program ownership by the municipalities and local organizations i s strongly sought; a Plan de Produccibn y Consewacibn i s prepared for the micro-basin in a participatory way jointly with local communities, and must be endorsed and formally approved by the municipal cabildo; (3) the program does not have an investment budget; it organizes local development demand through the planning process, and once the objectives and priorities are well established it assists in bringingin investment and service supply from different sources72;(4) emphasis is given to the constitution o f a team o f motivatedtechnical staff to implementthe program73,who are rigorously and formally trained as part o f their job, with the assistance o f academic centers, which also help inthe development ofintervention methodologies. Althoughno evaluation is a~ailable'~, several elements of the programsuggest that it i s successful in its objectives as well as cost effective. Contrary to Microrregiones, it i s not targeted to municipalities with high marginality indexes, but it does work in marginal municipalities, and in general the microcuencas selected tend to be located in the poorest municipal areas. Three success elements related to program concept and institutional factors can be singledout: (1) emphasis on ownership by the local population and municipal authorities; (2) the constitution o f a well trained and motivated team of implementing staff, which have created positive synergies with the local population, and (3) a long-term approach based on developing a good understanding o f local conditions and building good relations with the local population, something difficult to do when there i s pressure to disburse budgetary resources and fulfill administrative targets set from the distance. The program, however, cannot be considered a panacea for all rural development ailments. Microcuencas programs can be ideal local components o f strategic projects for larger territorial spaces, but they are too small to reap the economies o f agglomeration and reach the critical mass o f assets and supply capacity that are crucial to the development process. The Ley deDesarrollo Rural Sustentableis an ambitious and comprehensive piece of legislation, that favors rural development as a multisectoral endeavor and embraces a territorial approach. It sets the framework for government action in agriculture and rural development. One o f its central features in this connection i s the creation o f Consejos de Desarrollo Rural Sustentable at various levels -municipal, district, state and national- with mixed composition o f government officials and civil society representatives. Municipal councils are supposed to program rural development in the municipalities, suggest investment programs and clear the use o f Alianza funds decentralized to the municipalities. SAGARPA has been very active in the formation o f these councils particularly at the municipal level and many have already been created. Shared commitment for the joint implementation of the law is insufficient. One limitation o f the territorial rural development system that the Rural Development Law seeks to promote through the councils i s that, in spite o f being a national law binding all government entities, it i s very much associated with SAGARPA in the minds o f most state and federal government officials operating inrural areas, many o f whom do not know well the law and do not see themselves truly boundby it. Progress has been made inthis respect duringthe last two years, 72 By the end of 2003 the total investment carried out inthis way inmicrocuencas was estimated inMxP 882 million. 73 Some 400 technicianshavebeen recruited by FIRCOto implement the project. 74 An evaluation of six microcuencas is beingprepared by the Universidad Autonoma de Quere'taro. 102 but coordination of activities, budgets and operational norms by the various secretarias involved inruraldevelopment andalsowith stategovernments continues to be a challenge. Direct Cash Transfers The main direct cash transfers program and one that has attracted national and international attention is Oportunidades. Started in 1997 during President Zedillo's administration under the name o f Progresa, it continued during that o f President Fox under the name o f Oportunidades.The program consists o f a direct transfer o f cash to women (household heads or spouses) in poor households with school age children. A supplemental food package i s included in case o f children malnutrition. Transfers are conditional on children attending school and going through regular medical checkups. The program has three components, education, health and nutrition, and two objectives: to alleviate extreme poverty through a direct cash transfer, and to promote the development o f human capital. Iti s the largest single federal program operating in rural areas with coverage o f around 3.6 million rural and 1.4 million urban families and a budget o f MxP 25.5 billion in2004 or some US$2.3 bn. The program has been analyzed in the first part o f the Mexico Programmatic Poverty work (see World Bank, 2004) and hence it i s not examined here. 103 6. ISSUES AND CHALLENGES INTHE IMPLEMENTATIONOF RURAL POLICIES This chapter examies the implementation process of rural policies. There are many issues and challenges linked to policy and program implementation which Mexico along with other Latin American countriesface. Themainfindings andpolicy implicationsof the chapter are summarized below. A new vision of social development emphasizes the importance of institutions, the role of civil society in development, and the need for a more active developmentalrole of the state. This role consists notjust inthe provision ofpublic goods and appropriate regulatory frameworks but also in the active support to civil society organization, andthe search for public-private synergies. Mexico is in transition between program administration cultures: from reliance on principles o f hierarchy and control, to reliance on transparency, creating consensus between those who design and those who implement the programs, and active commitment o f the latter to programs' objectives. The challenges faced by Mexico to improve the implementationsystem of rural development programs can be grouped in five categories: political and administrative circumstances o f a macro type; operational and budgetary norms; organizational cultures; client orientation and beneficiary empowerment; and the incentive system for program operators. Among the macro type political and administrative circumstances we highlight the electoral system o f local authorities, with short mandate and no reelection, little functional to the continuity required by rural development. Also, the annual budgetary system which works against program continuity and the adoption o f a long-term perspective. Finally, the organization o f the state apparatus along sectoral lines which i s little sympathetic to a multi-sectoral matter like rural development. Sub-national levels could play an important role in overcoming "sectoralism". At the federal level, the Secretaria de Hacienda, which has a multi-sectoral view and i s responsible for the quality o f public investment, i s well placed to take an active role inpromoting the integration offederalruralprograms with aterritorial approach. Appropriate operational and budgetary norms are critical to program success. Simplifying these norms and making them friendly to the realities of rural areas would improve implementation. One problem i s the time factor, for timeliness o f support i s often more important than amount o f support. Not only are there no multi- annual budgets but operational and budgetary norms often allow a few months only to spendthe allocated budget, thus introducing distortions. Changes from year to year in norms related to issues like eligibility conditions, subsidy amounts, target areas 104 and type o f benefits are also detrimental to program implementation because they introduce uncertainty among beneficiaries and operators. Another challenge relates to the few recurrent funds usually made available for program operation. There i s a conservative attitude towards allocating recurrent costs for program implementation, which i s understandable in view o f the abuses o f the past but potentially damaging to program success. This i s particularly the case with productive programs where the formula for success could be summarized as "recurrent costs +rural finance". Among the challenges posed by prevailing organizational cultures is there a culture of mistrust, evident for instance inthe reluctance o fmiddlemanagers to take decisions for fear o f breaking the norms, and in the ex ante controls that often check operations. The institutional segmentationculture adds to the problems created for rural development programs by the sectorial organization o f the state. One possible option to overcoming this culture would be through enhanced efforts from SAGARPA to involve more other organizations in the application of the Ley de Desarrollo Rural Sustentable. The Secretaria de Hacienda and Presidencia de la Republica could also promote more integration o f rural development programs inthe framework o f the law, encouraging the application o f a territorial approach to rural development. Strengtheningthe evaluation culture could be achieved by introducing M&E systems simultaneously with program design, better dissemination of program evaluation results, constructing action agendas for the recommendations made by evaluation teams and monitoring the progress made on these agendas, and adopting participatory M&E methods. Overcoming the culture o f short-term achievement could be addressed through the design o f a long-term strategy for rural areas as politica de estado cutting across party lines and administration terms, and the recognition o f the importance o f "intermediate policy results" and their value to political constituencies. Finally, efforts could be made to move the culture related to decentralizedimplementation away from the dichotomy betweennormative functions at the center and operational functions at the periphery. Improvingclient orientation and beneficiary empowerment is another challenge, which could be addressed in various ways: enhancing the dissemination o f programs and program norms to prevent biases in the beneficiary selection process; disseminating among beneficiaries evaluation results and the action agendas emerging from them; direct accountability from program operators to client/beneficiaries; and measures to detect and prevent opportunistic and rent- seeking behavior on the part o fprogram operators. The last category of implementation challenges discussed refers to improving economic and moral incentives for program operators. The economic situation o f bottom level program operators i s generally inconsistent with the importance o f their function. It couldbe revisedthrough linkingremunerations to performance and client satisfaction. Maximum performance from bottom level program operators could be sought through a revaluation o f their function, consulting with them on program matters, giving them systematic training, disseminating best practices, carrying out systematic performance evaluations o f their work, and promoting networking systems, a client orientation ethic, and a sound esprit de corps. 105 Two measuresmight help advancinginthe transformations suggested above:the creation o f a technical committee to examine the implementation issues o f rural development programs and make recommendations; and empowering the Consejo Mexican0 para el Desarrollo Rural Sustentable to take an active role in the evaluation o f rural development programs and the monitoring o f the action agendas resultingfrom program evaluations. The Consejo would also bethe natural institution to promote the preparation o f a long-term rural development strategy to propose to the country as politica de estado. It would be difficult for the Consejo to carry out these functions without appointing a managing council and having a small technical secretariat. Together with the above measures, two innovations could be considered: the introduction of a system of oidores, consisting of well trained individuals or teams who would informally follow rural development programs at the point o f service delivery; and the promotion o f process certification, carried out by independent consulting firms or NGOs, who would certify that the processes related to program operation andbeneficiary participation are sound. GENERALASPECTS We review in this part some o f the issues and challenges found by governments in the implementation o f public policies and programs, discussed in the classical literature on public management. The purpose i s to understand how and why implementation i s a complex process, and also to realize that many of the issues found in the implementation o f rural development programs in Mexico examined inthe following part are recurrent ones, not exclusively Mexican, on which there i s codified knowledge. Institutions, CivilSociety and theRole of theState An integrated approach to rural development requires building institutions that involve civil society and the state. A new vision and a new mood have gained momentum among development theorists and practitioners after the adjustment processes and disappointing growth o f the 80s and early 90s, when development recipes were based on Washington Consensus premises. The upcoming vision stresses, inter alia, (1) the importance that strong institutions have for development (Burki and Perry, 1998), (2) the recognition o f organized civil society as a major "third sector" and o f its contribution to good governance and the development process, and (3) the need for a more active developmental role o f the state. Perhaps more importantly, the new vision recognizes the synergies usually established between the public and private domains in successful development programs, and the contribution that these synergies can make to economic and social progress. In the area o f rural development, largely established boundaries between urban and rural areas, agricultural and non-agricultural activities, on-farm and off-farm employment, are being crossed over in favor of a more integrated concept o f the rural development process (Tendler, 1998, World Bank, 2002b). Strong institutions are needed as much for equity reasons - the fight against poverty and inequality (World Bank, 2000, and de Ferranti et al, 2004)-- as for efficiency reasons -the proper functioning o f markets (World Bank, 2002) -, and are important both in the government sector and in the private sector and civil society. Strong government institutions are particularly relevant for the implementation o f development programs, because most o f the developmental agency o f the state i s carried out through specific programs, whose characteristics and operation depend crucially on the quality and strength o f government institutions. 106 Civil society organizations play a fundamental role in the promotion of common interests of groups in society. Many also perform public-type functions by providing collective goods to their members as in the management o f common property resources (Ostrom, 1990). Of particular relevance for rural development are meso-level civil society organizations that perform economic coordination functions at the regional or sector level (Helmsing, 2001). They are essential for inter-firmeconomic coordination and public-private interaction, and can be o f many types: 2"d level producers' associations, governance organizations of production chains, business societies, 2"d level rural finance associations, territorial organizations like regional or municipal development councils, 2"d level indigenous peoples' organizations and others. Institutional maturity and organizational thickness have been recognized as conditions facilitating endogenous processes o f local development (Putnam, 1993). The state, on the other hand, retains the overarching responsibilityfor economic development. Other than the classical roles of making war and enforcing internal order, states have acquired in modern times an overreaching role in contemporary developing societies -that o f being responsible for their economic transformation (Evans, 1995). This sets the state apart from other social organizations; for no other organization has the recognized role o f designing and implementingdevelopment policies and articulating the development process. The new vision asserts that this developmental role consists not just in the provision of public goods and appropriate governance and regulatory frameworks for the market to operate, but also in (1) the active promotion of civil society organization (Tendler, 1997), and (2) the search for public-private synergies (Evans, 1996, UnitedNations, 2004). Synergy may consist o f exploiting complementarities in the co-production o f development assets and services, where each part contributes that in which it has advantage, as in rural housing programs where the responsible government agency provides engineering design and supervision and some building materials while the local community provides labor and complementary materials. Synergy may also consist o f state officials and the local population going beyond the public-private divide to create a collective action agenda based on the community o f interests set by the development objective. The state i s here embedded in the development process, working hand in hand with other actors and enhancing their development role, not substituting for it. The Complex Process of Policy Implementation The reality ofpolicy implementationincludes pluralityof interestswhich may create a gap between design and realization. Policy implementation i s far from being a linear, hierarchical process; it i s a rather complicated, potentially conflict-ridden game similar to the assembling o f a machine o f many different parts, each with its own autonomy: financial resources, administrative processes, public and private providers, government regulations, beneficiary attitudes, and so on. It i s a process with a life o f its own, usually carried out by large and inflexible administrations, open to the distortions o f bureaucratic interests. Different games are possible within the "implementation game", and policy and program objectives can be deeply distorted, even reversed, by these games (Bardach, 1977 and 2001). It i s important, therefore, when addressing implementation shortcomings to recognize the plurality of interests that exist in the implementation process, and the ensuing potential distance between objectives and broad design, on the one hand, andhardimplementation realities, onthe other. World Bank view on implementation.A World Bank study on service delivery to the poor (World Bank, 2004b) characterizes the process o f implementation as a set o f relationships among four actors: citizens/clients, politicians/policymakers, organizational providers, and frontline professionals, who "are linked in relationships o f power and accountability. Citizens 107 exercise voice over politicians. Policymakers have compacts with organizational providers. Organizations manage frontline providers. And clients exercise client power through interactions with frontline providers" (page 48-9). There are two routes of accountability of frontline providers to clients: a long one whereby citizens influence policymakers and these in turn influence providers, and a short one whereby clients are directly vested with power over providers. Quality of implementation of development programs is better examined at the intersection between program operators -the frontline providers- and the beneficiary public -the clients. Itis at the "point of service delivery" where programs succeed or not, andit i s there where we can check the extent to which the operation rules and their application respond to program objectives (Williams, 1980 in Aguilar Villanueva, 1993). From this vantage point we can assess what Berman (1978) calls micro-implementation: the way in which local-level operators adapt to program norms and central instructions putting in place their own operational procedures. We can also undertake from there a "backward mapping" o f the implementation process until we arrive to the policy or program decision top (Elmore, 1979-80). This i s the opposite o f "forward mapping", the process normally followed in policy design, which starts from the formulation o f the policy decision or program objective, andthentraces the always more detailed steps that should lead to satisfactory implementation. "Forward mapping" thus assumes that policy-makers or program designers control the organizational, political and technical processes affecting implementation. "Backward mapping", on the contrary, does not presuppose such control, and questions at each stage the capacity, resources and motivations o f the responsibleunit or individuals to carry out the operation o fthe program according to objectives. Implementation critically depends on bottom-level operators -public servants, contracted individuals,service firms or NGOs- that have direct contactwith beneficiaries. Itis these "street-level bureaucrats" (Lipsky, 1976)that can make the difference, because (1) they may or may not generate the synergies on which program success will normally depend; and (2) they are the ones who ultimately determine how the program is delivered. To be good, norms must be general since any attempt at disciplining heterogeneous reality by means of complex rules i s destined to fail, imposing a straight jacket likely to paralyze implementation. These general norms are interpreted and enriched at all stages in the implementation process. Norm creation thus continues to exist until the beneficiary i s reached and the program's relation with hidher is exhausted. "Street-level bureaucrats" are probably more important than anyone else in this process. This is why their technical and context capacity to do well their job, their understanding o f and commitment to program's objectives, and their empowerment to carry out the assigned task with imagination and confidence are essential to program success.75 The approach to the implementationof development programs and the quality of implementationprocesses are affected by how public administrationwas shaped over time as a system o f concrete organizations with their own corporate culture, and the dominant concept o f how the system functions. Elmore (1978) contrasts an organizational model that he calls 75 The importance of a committed and innovative attitude of civil servants is recognized in the Mexico's National Development Plan, 2001-2006. Thus: "A travks de la innovacion buscaremos reemplazar 10s sistemas burocraticos por sistemas emprendedores (...) Transformar la orientacion del Gobierno solo sera posible si somos capaces de sumar las voluntades de todos 10s servidores publicos, por lo que debemos dejar atras 10s esquemas jerarquicos basados en el control, que inhiben la creatividad y la innovacion, para dar pasos a esquemas que faculten y fomenten la participacion y el trabajo en equipo" (Mkxico, 2001: 64) 108 "systems' administration" with another one called "organizational de~elopment"~~.For simplicity we call them Model 1and Model 2. In the "systems' administration" model-Model 1- organizationsare supposed to pursue the rational allocation of means to ends so as to maximize some objective function, all organizational behavior being determined by this. The organizing principle postulated i s that o f hierarchical control, with decision-making concentrated at the managerial top. For each o f the organization's tasks there i s an optimal allocation o f responsibilities among subordinate units. In this model, program implementation is a dynamic process consisting of (1) carefully definingthe objectives, (2) allocating to the subordinate units the responsibilities and performance parameters consistent with those objectives, (3) supervising performance and distributing rewards and sanctions as appropriate, and (4) making internal adjustments to improve performance, particularly inresponse to change inprogram environment or demands. In the "organizational development" model -Model 2- the organization is supposed to work so as to satisfy the individuals' need of autonomy and control over their own work and the participation in decisions that affect them, and to promote their commitment with the organization's objectives. Consequently, organizations should be structured so as to maximize individuals' control, participation and commitment at all levels, which implies minimum hierarchical relations. In this context, effective operation depends on the creation of consensus, strong interpersonal relations among individuals, and the construction o f effective task forces. Program implementation in this model consists o f creating consensus and adaptation between those who define the objectives and those who have to implement the program, and seeking the active commitment o fthe latter to the program's objectives and goals. The historicalprocesshas resultedinMexico, as in other LatinAmerican countries, in a state administrationculture closer to Model 1 than to Model 2. Closeness to Model 1 does not mean that the system operates with the rationality and hierarchical control effectiveness impliedinthe model's description. This is preventedby (1) the unavoidable presence of conflict within and across organizations, and of individuals' and units' own agendas, and (2) the use of routines to avoid change. What it means i s that the perception o f how the system should ideally work i s based on the concept expressed by Model 1. An illustration o f the implementation logic impliedinthe two models is provided inBox 6.1. Box 6.1. Illustrationof the ImplementationLogic under Model 1and Model2 To illustrate these logics we can imagine the hypothetical case o f a central government organization in the process o f launching a new development program. A way to proceed under Model 1would be to: (1) design program details at the central level by a selected group o f technicians; (2) prepare program norms; (3) inform o f program characteristics and operation norms to de-concentrated government units and sub-national authorities; (4) recruit incremental personnel under short-term contracts to implementthe program; (5) train that personal on program norms; (6) centrally develop staff work norms and performance parameters, and (7) start operations based on the operation norms, staffwork norms and performance parameters. Under Model 2 we can imagine the routine for launching the program as follows: (1) once a preliminary concept o f the program has been developed, consultations are started through a set o f workshops with de-concentrated offices, sub-national authorities, organizations representing potential beneficiaries, and other sectors o f govemment and society who may contribute to the program, with the purpose o f checking the soundness o f the program's concept, improving it, and 76Elmore considers also another two models called "bureaucratic process" and "conflict and negotiation". 109 developing operation norms; (2) the concept is also presented to relevant government organizations to inquire ifthey want to associated themselves to it; (3) ifthe organization's staff i s not sufficient, new staff is recruited within a fiamework that would allow satisfactory salaries and job continuity upon good performance, or alternatively a private firm or NGO (or a set o f them) are contacted to conduct implementation in the perspective o f a medium or long-term implementation contract; (4) the program concept and norms are thoroughly discussed with the recruited staff (or firms or NGOs); (5) pilot operations are carried out to check the soundness o f the concept and the applicability of the operation norms, and serve as an experimental ground for the implementing staff (or firms or NGOs); (6) the experience gained duringthese operations is discussed with the participants and other stake holders, and the program design and operation norms are revised accordingly; (7) a continuous training program and staff networking and support systems are developed jointly with the staff (or firms or NGOs); (8) work norms and implementation parameters are developed jointly with the staff (or firms or NGOs); and (9) program implementation i s started. ISSUES AND CHALLENGES INTHE IMPLEMENTATION OF RURAL POLICIES INMEXICO Whereare We? The situation inMexico can be described as one oftransitionbetweena view of state organizationand programimplementationbased on Model 1to another based on Model2. This transition in the implementation model reflects wider changes in public administration taking place in Mexico and other Latin American countries towards a "New Public Management" culture (CLAD, 1999, Barzelay, 2003). The essence of the New Public Management view, first developed in the U.K.and rapidly expanding to many other countries, consists o f (1) transparency in budgetary processes and administration responsibilities; (2) management by results, including performance oriented budgeting; (3) professional status o f public servants and civil service careers; (4) accountability to the client-citizen; (5) decentralization towards sub-national governments and specialized agencies; and (6) program and policy evaluation (Guerrero Amparian, 2000, Arellano, 2001 y 2002, Marini, 2002, Arellano and Gil Garcia, 2003). Advances were made in this direction during President Zedillo's administration, under the Programa de Modernizacibn de la Administracibn Publica (PROMAP), when the reform o f Mexican public administration acquired weight in the government agenda. They have continued under President Fox, most notably in electronic government, governmental innovation, federalizacibn, and under various programs o f SECODAM and SEGOB. Support and stimulus are being received from international organizations, inparticular OECD and the World Bank. The National Development Plan 2001-06 emphasizes this transition towards a New Public Management when describing the type o f government the country needs (Box 6.2). Box 6.2. The "New Public Management"View in the Mexico National Development Plan 2001-2006 "A fin de que esta administracion cumpla con su responsabilidad historica de dar respuesta a las grandes demandas y expectativas de la sociedad, requerimos acciones capaces de transformar radicalmente 10s esquemas tradicionales de gestion... Necesitamos un gobierno participativo...que de forma constante se someta a una rigurosarendicion de cuentas, no solo en lo que se refiere a1us0 honesto y transparente de 10s recursos, sino tambien a la eficacia y calidad con que se utilizan. Requerimos un gobierno con un alto sentido de responsabilidad social... Requerimos un gobiemo estrategico y competitivo, que sea l a vanguardia de la sociedad, que establezca democraticamente las prioridades sociales e invierta de manera eficaz sus recursos financieros, humanos, materiales y legales...Requerimos ungobierno inteligente, capaz de usar 10s 110 mas avanzados sistemas administrativos y tecnologicos... Requerimos un gobierno agil y flexible, capaz de captar las oportunidades, atender 10s problemas y adecuarse a las circunstancias rapida y descentralizado, que en sus relaciones con 10s estados y municipios propicie... un nuevo pacto eficazmente... Requerimos un gobierno abierto y transparente... Requerimos un gobierno federal. Necesitamos un gobierno austero ... que someta sus procesos a una estricta validacion... En suma, requerimos un gobierno de clase mundial, un gobierno innovador y de calidad total" (Mexico, 2001: 64) We examine here some of the issues and challenges in the implementationof rural development programs.77Recognition by Mexican government and society of the need to reform the administration system along the principles set by the New Public Administration School i s a big step forward, but the challenges are many. Before examining these challenges, it i s important to highlight the different degrees o f implementation difficulty according to the type o f programs, particularly on whether they focus on social infrastructure, productive development or direct cash transfers. Productive development programs, such as Alianza para el Campo or Opciones Productivas, are generally much more difficult to implement (and to succeed) than other programs. They require (1) a medium-to long-term perspective, (2) a particular type of synergy and agency on the part o f responsible organizations and street-level bureaucrats, (3) a highquality participation o f beneficiaries, and (4) a large recurrent budget for implementation, all o f which place them on a separate class. Productive programs have thus special needs in terms o f planning horizon, recurrent costs, and public-private synergies, which are extremely important to recognize for their success. Directcash transfer programs, likeProcampo and Oportunidades,are comparatively simple to implement, once the technicalities of beneficiary registration, check issuing and the like are straightened out. These programs are based on centrally-established objective targeting criteria and a direct link between the central level and the beneficiary, without community involvement. Social infrastructureprograms, such as those under Ram0 33 and to a large extent Microrregiones, require local participation for efficient allocation of the investment^.^^ Promoting and managing participation places higher implementation demands on social infrastructure than on cash transfer programs, but the works could be completed and the program fully implementedeven in the absence of participation, because social infrastructure is finally a collection o f well definedinvestments o f a public good type that governments should be capable to provide on their own. This would be impossible in productive projects, which deal essentially with private goods, andwhere private-public convergence is ofthe essence. In the remaining part of this section we examine the main challenges faced in restructuring the implementation system o f rural development programs. We organize these challenges in five categories: (1) macro political and administrative circumstances; (2) 77 The analysis is based on discussions with managers, operators, evaluators and beneficiaries of various programs, the consultation of evaluation documents of programs and other relevant literature, field visits, discussions with leaders of farmers' organizations and NGO staff, and the many insights in an input paper prepared by Jorge Franco: "Los Programas de Desarrollo Rural: Operacion Institucional y Alivio a la Pobreza". The discussion among medium level rural leaders inthe consultation "Dialog0 para el Desarrollo Rural" that took place inTequesquitengo on 16-18May 2004 was also very useful. 78 Participation can also serve to reduce costs when works are directly contracted by the communities, as shown by the decentralized poverty alleviation projects inNortheast Brazil. 111 operational and budgetary norms; (3) organizational cultures; (4) client orientation and beneficiary empowerment; and (5) incentive systems for program operators. Macro TypePolitical andAdministrative Circumstances Three macro circumstancesplace severe constraints on the implementation of rural development (and other) programs in Mexico. The first one is the Mexican electoral process whereby municipal authorities have only three years o ftenure and cannot be reelected. Although this system may have merits, it hinders continuity in what needs to be a long-term endeavor. Federalizacibn has aggravated the situation in this respect as more rural development responsibilities are beingdevolved to municipal governments. The annual budgetary system existing in Mexico makes it difficult to design and operate programs with a long term view, especially productive programs. Only with difficulty i s it possible to ensure resource availability for rural development programs beyond the annual budget cycle. Because o f their long-term nature, productive programs are the ones to suffer most from this. Annual budgeting also results in uncertainty for producers, since it i s not known whether program support will last over time or not. The Ley de Desarrollo Rural Sustentable has made clear the importance o f giving certainty to producers with respect to government supports,79 and SAGARPA i s trying to make it happen, but the existing budgetary system does not make it easy. A multi-annual rolling budget would be friendlier to rural development. The state apparatus is organized along sectoral lines, whereas rural development is a multisectoral phenomenon that is best approached in an integrated manner. This i s the result o f how the state was historically conformed and how it evolved over the years, and i s common to other Latin American countries (Piiieiro et al, 1999) and to countries in general. Multisectoral and territorial development approaches to rural development are inherentlydifficult to handle with the existing organization o f the state. Sub-national levels could play an important role, but state governments tend to reproduce the segmented organization o f the federal level, whereas municipal governments, which are more territorial in approach, find it difficult to deal with rural development, particularly with its employment and income generation side. A more integrated and multisectoral administrative organization would probably be easier at the state than at the federal level. It i s also in the states where an integrated approach i s more needed. Administrative innovations by state governments to deal with rural development would hence be valuable. At the federal level, the Secretaria de Hacienda, which i s a truly multisectoral organization and has responsibility inlooking into the quality o f public investment, i s well placed to play an important role inpromoting the integration o f federal ruralprograms within a territorial approach to rural development (see Annex 3.1. on the territorial approach). Operational and Budgetay Norms Establishing operational norms that are simple and friendly to the rural environment inwhich they have to operate and to the characteristics of the beneficiariesis a major challenge. Inadequate norms may create conflict between what Rein and Rabinovitz (1978) call the "legal imperative" andthe "bureaucratic rational imperative". The former refers to the need to adhere to the norm while the latter reflects the need o f operators to act rationally to carry out their work, even against the norm. A case in point i s the system o f justification o f expenses in development programs, for instance in Alianza. The new Ley de Desarrollo Rural 79See articles 13.111, 70.1, 74.1, I1and 111, 90.1, and 191Iand 11. 112 Sustentable has introduced a most welcome innovation, consisting o f allowing program beneficiaries to procure the goods and services that are part o f development program investments. Inpractice, however, the beneficiary has to submit the justification of expenditure before he or she can access the subsidy. Since very few beneficiaries are in a position to advance the required cash, they some times resort to false invoices obtained from accommodating suppliers --at a price, naturally. Although program operators may be aware o f this, they may not pay attention because o f the pressure to implement the program; the "bureaucratic rational imperative" takes thus the upper hand over the "legal imperative". This may introduce an unnecessary element of irregularity inthe program, generatingtransaction costs and opening a door to rent seeking. Timeliness is key, especially in rural areas, where seasonality is a basic feature of livelihoods. The budgetary process i s one o f the factors that may interfere with good timing in rural development programs in Mexico. This i s not only because o f no multi-annual budgets, but also annual budgets depend on the operational norms for the year being submitted by the responsible organization, approved by Hacienda, and the corresponding funds transferred to state fideicomisos or other accounts inthe field. The process may in some cases not be complete before April or May, and disbursement must be finished by November to close the accounts in December. Six or seven months may be left, therefore, to operate the program. The consequence, when this happens, i s pressure to disburse during those months which may affect quality. Under- disbursement i s another possible consequence highlightedinprogram evaluations.*' Lack of stability in programnorms may introduce uncertainty among potential and actual beneficiaries as well as program operators. Change in norms from year to year i s a problem faced by some rural development programs. Eligibility conditions, amounts, target areas, crops, and type o f benefits may change from one year to the next. This may introduce uncertainty among farmers and other rural dwellers and also among the street-level bureaucracy, who may decide to abstain from action untilthey know what the new norm would be like (ver c6mo viene la norma). Stability inprogramnorms is thus another challenge. Few recurrent funds are made available for program operation. Vigilance from Hacienda that program money i s well spent, as much as possible o f it going to the final beneficiaries, i s to be praised, but it must also be understood that shortage o f operational funds mayjeopardize programs. Recurrent costs are essential for the success o f many programs, which often depend on soft more than hard investments. This i s particularly true for productive 8o Thus, the external evaluation of Microrregiones says: "Uno de 10s problemas mas fuertes que enfrento el 100% de las Coordinaciones del Programa Microrregiones en la operacion de 10s programas en 10s estados fue el relativo a la aprobacion, liberacion y radicacion tardia de recursos lo cual afecto de manera determinante la operacion del programa en casi todas sus etapas" (Instituto Maya, 2003: VIII). Similarly, the external evaluation of Alianza states: ". .. la operacion de Alianza comienza mucho despuks del 1de enero, ya que para su inicio es imprescindible la aprobacion de las Reglas de Operacion que nunca fueron publicadas antes del 15 de marzo. Una vez publicadas las Reglas, el Gobierno Federal acuerda con cada gobierno estatal 10s Anexos Tkcnicos que definen la distribucion de 10s recursos... Posteriormente se depositan 10s recursos federales en 10s fideicomisos estatales, lo que en la mayoria de 10s estados provoca que la operacion de la Alianza nunca comienza antes de mayo, aunque en algun cas0 se retraso hasta agosto... Sin embargo, el mayor problema con este calendario operativo es la obligatoriedad de comprometer totalmente 10s recursos antes del 30 de noviembre." (FAO, 2003: 14). Finally, the external evaluation of Opciones Productivas complains that "Hay unserio problema de retraso en la recepcion de 10s apoyos que afectan su eficacia; el 45% de la poblacion consider6 que 10s apoyos no les llegaron a tiempo. La mayoria de 10s apoyos de Opciones Productivas [an informal credit program for poor farmers] llegaron en agosto, despuks de la siembra y en las demas vertientes hasta diciembre". (RDS, 2004: 14). 113 development, where the formula for success could be summarized as "recurrent costs + rural finance". We may notice that one o f the most interesting rural development programs in operation, Microcuencas, has no investment budget, only a recurrent costs one. If there i s no sufficient fiscal space for a program, it may be better to cut its scope than to leave it without sufficient recurrent costs. Organizational Cultures The focus on bureaucratic norms based on hierarchical control principles (Model 1 above) results in a system burdened by controls at each level. Insuch a system, the reluctance by middle management to take decisions because of the fear of breaking the norms, inparticular but not only those related to expenditures and financial flows, may result in slow operation and many ex ante controls on subordinates. These often check and distort operations.81 Due to a segmented state system, rural programs often have different breakdowns of geographical regions, different definitions, norms and procedures for similar things, different timings and disbursement methods, and create their own separate counterpart organizations. Program evaluations have highlightedthis.82There are, however, many instances o f successfuljoint action by differentprograms at the local level. This tends to happenwhen local organization i s strong and there i s good identification o f local needs, mostly arrived at through local participatory planning, so that local populations know well what they want and from where to get it, and can put pressure on program operators. The Ley de Desarrollo Rural Sustentable i s an important step towards the integration o f rural development actions by different actors, offering a good coordination framework. Not all relevant secretarias, however, see themselves equally committed to the type o f rural development approach and institutionality espousedby the law, since the law tends to be identified with SAGARPA. Enhanced efforts from SAGARPA to involve more other organizations in the implementation o f the Law, and of the Secretaria de Hacienda and the OJicina de la Presidencia de la Republica to promote more integration o f actions within the framework established by the Law would be a possible option to advance in meetingthis challenge. A solid evaluation system is key to the implementation of programs. Evaluations are regularly carried out o f most rural development programs by external evaluators, and there are cases inMexico like that o f Progresa/Oportunidades where the evaluation methodology has been praised internationally. Yet, there are still challenges in the evaluation o f Mexican rural development programs. One i s the introduction o f evaluation mechanisms for on-going and impact evaluation at the time of project design, so that evaluation i s embeddedinthe working o f the program, and feedback i s regularly obtained. This would avoid introducing evaluation systems after the program i s designed, as a kind o f expost appendix to it. Another challenge i s to give more relevance to and make more use o f evaluation results. Broad dissemination and open 81Thus, it was reported inone extreme case that seven signatures and five days of waiting were required in a certain office for a gasoline authorization of MxP 100for a work visit to a community (Franco, 2004). 82 Thus, upon examining program evaluations, Jorge Franco (2004: 18) concludes: "Esta deficiencia de articulacion se expresa en la tendencia de cada institucion a crear su propia contraparte de organizacion social con la cual interactuar. Oportunidades, Alianza para el Campo o Microrregiones no dialogan con autoridades o asambleas comunitarias o ejidales, con organizaciones rurales solidas y estructuradas en torno a sus propios fines. Se dialoga con organizaciones ad hoc, sea el Comitk de Mujeres de Oportunidades, el Consejo de Desarrollo Rural Sustentable o el Consejo Microrregional, todos ellos formas organizativas creadas desde la accion publica, de acuerdo a sus reglas y con vida uti1 estrictamente vinculada a la existencia del programa" 114 discussion o f results would be needed for this. Also, mechanisms could be developed whereby recommendations made in evaluation studies are included in an action agenda after discussion and agreement with the relevant parties, and progress on this agenda i s monitored. This would give more relevanceto the evaluation process, makingmanagers more responsiveto it, andwould also force evaluators to provide well thought, feasible recommendations. The systematic adoption o f participatory evaluation methods i s yet another challenge. These methods are important to develop ownership o f programs, and to test how well beneficiaries are informed, if programs respond to client expectations, and how results are affected by local context variables. Another key issue is how to overcome the culture of short-term achievement often present in rural development activity. This culture is characterized by (1) insufficient strategic focus, and (2) attention to pursuing quick results. This has a number o f consequences some o f which are highlightedin program evaluations. Thus, synergies among programs and cooperation among different levels o f government are made more difficult if there i s not a rural development strategy that i s well known and broadly shared by rural actors. Also, inthe absence o f a long-term perspective it i s more difficult to have clarity as to the wider objectives o f specific programs, and rural development efforts may become disperse. A broad strategic long-term vision for rural developmentthat cuts across party lines would facilitate avoiding excessive focus on "quick results". It is a political reality that governments, national or sub-national, like to show results, and this i s indeed part o f the democratic process. Results, however, are o f differenttypes. What political leaders often want to show i s final results, which leads to concentrating on programs where final results can be achieved in a sufficiently short period o f time. Political constituencies, however, are more ready to value intermediate results within a well understood strategy than i s often acknowledged. This i s because political constituencies are increasingly more sophisticated in the understanding o f policy processes, and more influenced by "issue networks" (Heclo, 1978). These networks, made o f academics, specialized government agencies, area professionals and practitioners, civil society organizations, interested individuals, and pressure groups, value strategic approaches and appreciate the merits o f intermediate results. Mexico i s not an exception to this. In the case o f rural development, it would be useful to have a strategic politica de estado cutting across party lines and administration terms in view of the long-term nature o f rural development processes, and also because o f the urgency created by the large incidence o f poverty and dualism in rural areas, with the frustration and social tensions that this generates. Decentralization and increased participation by local partners in normative and design functions would increase program ownership on the ground. A common view o f decentralization stresses a normative role at the centre and an implementation role in the periphery. An alternative view is one where national and sub-national levels are both seen as having normative and operational functions although o f different type. As indicated above, there i s a normative continuum going from basic norms to implementation details. There i s a fundamental role for central authorities in the design o f basic program norms, but those norms would be enriched by discussion and agreement with the sub-national units that will apply them. Similarly, sub-national authorities have a fundamental role in designing norms for the specific application o f programs in their territorial area o f competence, but it i s advisable that application norms are discussed with central agencies, to ensure that they are inline with program objectives. Client Orientation and theEmpowerment of Beneficiaries Improving the dissemination of programs is an important issue. Good dissemination i s needed in all programs in general but it i s essential in demand-driven ones; otherwise a bias i s 115 introduced in the selection o f beneficiaries by excluding the uninformed. Since the likelihood i s that the uninformed are poorer than the informed, the bias i s likely to be anti-poor. Thus, attention to dissemination campaigns and the allocation o f sufficient resources to this end i s conducive to equity and client orientation. Dissemination campaigns are often insufficient inruraldevelopment programs in Mexico, as highlightedin program evaluation^.^^ Most o f the dissemination o f rural development programs i s by word o f mouth. In the case o f Alianza much o f it i s done by the Prestadores de Sewicios Profesionales. They approach individuals or communities to induce them to submit eligible projects to Alianza, which they would prepare. The process, thus, risks turning from demand- to supply-driven, distorting beneficiary and investment selection. Dissemination campaigns need to use means appropriate to rural areas and the characteristics o f the beneficiaries, such as the use o f local radios, languages, and organizationnetworks. Direct accountability o f program operators to clients/beneficiaries empowers beneficiaries and enhances the quality o f program delivery. A classical example i s letting farmers choose who provides technical assistance financed with public money. This i s done for instance inAlianza. Direct accountability i s easier when bottom level implementation i s carried out by private providers. It i s more difficult when it i s civil servants who are at the "point o f service delivery", and still more difficult with respect to middle-levelprogram managers. Direct accountability would be enhanced by the introduction inthese cases o f a system o f incentives for "street level bureaucrats" and middle managers linked to "client satisfaction" and o f suitable ways to measure that satisfaction. Preventing opportunistic and rent-seeking behavior o f program operators i s another challenge to foster client orientation. The tendency o f organizations to provide services to some beneficiaries and not to others has been observed in the literature (Bardach, 2001). This has also been observed in Mexico where, for instance, clientelistic ties may be established between Prestadores de Sewicios Profesionales and their favorite producers, generating distortions in beneficiary selection. Distortions may also be introduced if low level managers deliberately withhold dissemination of their programs in the fear that more demand would be attracted than they can meet. Criticism has been voiced o f the reproduction o f relations o f subordination betweenstreet-level bureaucrats or independent service providers empowered by programs, and beneficiaries. The complain i s occasionally heard, for instance, with respect to the Prestadores de Sewicios Profesionales in Alianza and the local teacher or local nurse in Op~rtunidades.~~ Political biases may be another source o f distortions. Thus, the evaluation o f Opciones Productivas asserts that "the operation o f the program is frequently distorted by the political-clientelistic biases in the country's entities" (RDS, 2004: 13). The use of participatory evaluation approaches i s an opportunity for beneficiary empowerment and for finding out and eventually neutralizing opportunistic behavior by program operators. Incentivesfor Program Operators 83 Thus, for instance, the external evaluation of Opciones Productivas indicates that "En la mayoria de las delegaciones no se publica la convocatoria en un medio de comunicacion masivo, mas bien suelen pegarla en las presidencias municipales y dejan que de manera "natural" fluya la demanda...El conocimiento de 10s beneficiarios de sus derechos y obligaciones, asi como del proceso operativo del Programa, es sumamente limitado" (RDS:2004 12and 14) InOportunidades, teachers must certify the attendance of children to school and nurses the attendance of women to training sessions and of children to medical controls. Since the subsidy depends on certification, teachers andnurses are placed ina position of power, which they are accused to abuse insome cases. 116 We reflected above on the importance for program success o f the commitment to program objectives o f street-level bureaucrats and other bottom-level operators. Yet, the way these bottom-level operators are treated by administration bureaucracies in many countries i s not always consistent with the relevance o f their function. The challenge i s to introduce the material and moral incentive systems required to push these operators to give in their jobs the best o f themselves andto do it in a way conducive to achieving program objectives. Program operators dealing with field activities and the interface with clients are generally poorly compensated. The decrease in size o f the rural administration part of the Mexican state that took place inthe 1990s, andthe increase inthe number and importance o f rural programs, has led to a situation where the better paid, tenured government staff concentrate on administrative or managerial office functions. Field activities and the interface with program beneficiaries are left to individuals recruited ad hoc for the task or to contracted technicians or professional bureaus. There are cases when these street-level bureaucrats have nojob tenure, no career prospects, are poorly paid, sometimes with delay, and rotate rapidly. Inoccasions, they are compensated for the precariousness o f their position with mechanisms implying a conflict o f interests, like when rural development coordinators are allowed to present projects to Alianza as if they were Prestadores de Sewicios (Franco, 2004). Program implementation would be enhanced by ensuring that bottom level operators are fairly treated economically and have reasonable career prospects as government officials or continuity and improvement o f contracts if they are contractedtechnicians or professional bureaus. Resource constraints are not a sufficient reason, because (1) as mentioned before, recurrent costs are too important to the success o f programs to make them a saving target upon budget squeezes; it i s usually better to cut the scope o f programs than to allow insufficient recurrent costs, and (2) many problems have to do with payment delays, length o f contracts, and inter contract blanks, which are not due to fiscal constraints but to inappropriate budgetary and financial regulations or simply inadequate management. Moral incentivesare as important as economic ones. JudithTendler (1997) has shown with the example of the state of Ceara in Northeast Brazil the importance in development programs o f virtuous circles, where motivated operators receive the recognition o f beneficiaries and are empowered by local communities in their work, increasing their motivation. Other than with fair economic treatment, motivationof bottom-level operators can be promoted by measures such as (1) recognizing in different ways the importance o f their function, (2) discussing with them program objectives and implementation procedures, and incorporating their ideas to improve the program, (3) introducing systematic training that allows them to broaden their views in addition to being informed about the norms85, (4) organizing network systems, (5) disseminating best practices, (5) carrying out systematic and fair evaluations o f their work, (6) promoting a client orientation ethic, and (7) valuing their esprit de COTS as rural development practitioners. HOW COULD CHALLENGESBEMET? The challenges above are sizeable but probable not larger or more difficult to meet than in other Latin American countries. They are also mostly long-term, and thus to be met over a long period o f time. Government organizations responsible for the design and 85 Thus, for instance, SAGARPA is designing with the assistance of IICA and the Colegio de Postgraduados de Chapingo an interesting broad-based training program on the promotion of rural development with a territorial approach, mostly oriented to rural development practitioners, especially those supporting the work of the Consejos de Desarrollo Rural Sustentable. 117 implementation o f rural development programs suffer the tension generated by the institutional transformations taking place in the country. They also have a difficult bridging function: linking the Mexico o f formal institutions to a complex rural milieu with varied cultures. They have to confront a rapidly changing world and an inherentlymultifaceted sector with instruments which are not always the most appropriate. The need for a modern public administration in Mexico and the path to follow to construct it have been laid down inthe National Development Plan 2001-06, as illustrated in Box 6.2. The policy options highlightedinthis section belong inthat framework. For convenience we summarize them inTable 6.1. Possible Steps toAdvance in theImprovement of Program Implementation We suggest two possible steps to advance in the improvement of program implementationsystems. The first is to create a technical committee to examine the institutional implementation processes o f rural development programs and make recommendationsto improve implementation. The committee could examine administrative and budgetary processes as well as incentive systems and institutional cultures. It could be integrated by a mix o f independent specialists, civil servants and legislators. A significant participation o f Secretaria de Hacienda would be relevant since the committee would respond to the global need o f the nation to ensure the quality o f public investment. The second suggestion i s to empower the Consejo Mexican0 para el Desarrollo Rural Sustentable created by the Sustainable Rural Development Law to take an active role in the evaluation o f rural development programs, including the adoption o f participatory evaluation systems. In order to take more operationally relevant roles, the Consejo would need to have a small technical secretariat and it could also elect a small managing council to be able to function in a more practical way, in view o f its large size. The Consejo could have an important role in monitoring progress in the agenda o f actions to implement evaluation recommendations. It would also be the natural institution to promote the preparation o f a long term rural development strategy to propose to the country aspolitica de estado. TwoPossibleInnovations To conclude, we would like to suggest the possible adoption o f two novelties to improve the monitoring o f rural development programs: a system o f oidores and a system o f "process certification". We discuss thembelow. A system of oidores could improvethe monitoringof rural development programs. We propose considering the introduction o f a system o f oidores (listeners) consisting o f well trained independent individuals or teams who would informally follow rural development programs at the "point o f service delivery" through ad hoc visits, which could follow a randomized system, and inform top management o fthe "mood" o f the program there. They would not be Bardach's type "fixers", and would have no power over program operators; they would just see, listen (especially to beneficiaries and street-level bureaucrats) and report to top management. Oidores would work independently from formal evaluators not substituting for them in any way. The rationale derives from the distance that inevitably exists between top management and street-level operations, particularly when programs are large. Administrative mechanisms makes it inevitably difficult for top managers to have a frank and open two-way dialogue with street-level bureaucrats and receive fresh information and imaginative suggestions from them. In all administrative systems information flowing up i s modified according to the perceptions o f the middle-levelbureaucracy to ensure that the information reaching the top poses 118 no danger (real or imagined) to the intermediate level. This makes it difficult for managers to develop a feeling o fwhat i s the perception from bellow o f how programs work.86 Low profile oidores sustaining recurrent informal dialogues with frontline professionals and clients would be a way for top managers to have fast and frank information from the bottom on operational problems. Monitoring systems give valuable information on progress in physical and financial parameters, but do not convey much regarding the conditions at the bottom that may or may not facilitate program operation, how beneficiaries and direct operators see them, and a first hand impression o f the effects o f the program. Traditional external evaluations are too distant apart and are focused on impact, which makes them little suitable to give fast feedback of program operation and possible improvement measures. The system o f oidores could fill this vacuum. It i s important that oidores have experience and analytical capabilities to reflect on general issues rather than try to meddle with specific situations trying to correct them. To be effective, the system should be kept very small, informal, flexible, and separate from formal evaluation. A system of "process certification" to be carried out by independent consulting firms or NGOs, would certify that the processes related to program operation and beneficiary participation are sound, in the sense o f proceeding according to program objectives and guidelines and to accepted practice. Certification would be most useful when the implementation o f the program or o f some components are delegated to third parties (consulting firms, NGOs, 2"d or 3rd level rural organizations), serving to reassure program managers that implementation processes carried out by these parties are sound. "Process certification" i s different from evaluation and auditing because it does not attempt to assess program outcomes and it does not deal with program financial management and accounting, or only in a broad way and to the extent that this affects implementation processes. The closest parallel i s with environmental certification, andthe inspiration i s taken from there. x6 The work of the oidores is similar to that traditionally carried out by Bank project supervision missions, from where the idea is borrowed. Like Bank missions, oidores would have the capacity to visit projects inthe field, buildup an image of how they are working, and report at highlevel. 119 Table6.1. Summa] of Optionsto ImproveProgramImplementation ProblemArea Options Multi-annualbudgeting Administrative innovations at the state level to deal with the multisectoral Macro type political and and territorialnature ofrural development administrative circumstances An active role of the Secretaria de Hacienda to promote the coordination o f federal ruralprograms Simplifying operational norms and making them more friendly to the realities o f ruralbeneficiaries Better timing in the delivery o f supports and services and closing the gap betweenthe actual expenditureperiod andthe fiscal year. Operational andbudgetary norms More continuity inprogram norms Attention to needs o f recurrent funds, especially inproductive programs Changing mistrust cultures by empowering middle managers and rationalizing the system o f ex ante controls Reducing the institutional segmentation culture through (1) Enhanced efforts from SAGARPA to involve other organizations in the implementation o fthe Ley de Desarrollo Rural Sustentable;and(2) efforts by Secretaria de Hacienda and Presidencia de la Republica to promote more integration o f rural development programs in the framework o f the Law andto encouragea territorial approach to rural development Improving the evaluation culture by (1) introducing M&E systems simultaneously with program design; (2) disseminating better program evaluation results; (3) constructing action agendas for the recommendations made in evaluations, and monitoring progress in these Organizational cultures agendas; and(4) systematically adopting participatory M&Emethods Modifying the short-term achievement culture through (1) designing a long-term strategy for rural areas aspolitica de estado cutting across party lines and administration terms; and (2) recognizing the importance o f "intermediate results" andtheir value to political constituencies Changing the decentralized implementation culture away from the dichotomy between normative functions at the center and operational - functions at the periphery Improving the dissemination o f programs in order to empower Client orientation andbeneficiaq beneficiaries andprevent selection biases empowerment Promoting direct accountability o f program operators to clients Preventing opportunistic behavior from program operators through several - means including participatory evaluations. Providingappropriate economic incentives to bottom level operators Revaluing the function o f bottom level operators and providing moral incentives to promote their capacity and commitment through consultation Incentives for program operators with them, systematic training, networking, fair evaluations, dissemination o f best practices, client orientation ethics, and valuing esprit de corps. The promotion of a system of "process certification"would requirethe formationof a market for these services, and this in turn entails that a number o f independent consulting firms and NGOs specialize on this. A training cum promotion program couldperhaps be started to that effect. The system would facilitate the involvement o f farmers' organizations in the implementation o f rural development programs because it would give government agencies an independent means of knowing if the implementation process is sound. The delegation to farmers' organizations o f implementation functions has been impaired by a history o f cases o f politicization and abuse in the exercise o f the functions delegated. This i s not however the rule and many rural organizations have the capacity to implementprograms and are better placed to do it than other agencies. "Process certification" would be an instrument to facilitate their involvement. 120 7. HETEROGENEITYAND VULNERABILITY OF THE RURALPOOR Poverg is a complex and multifaceted concept which can be approached in many ways each highlighting different dimensions.As a result, poverty programs need not only be targeted to "the poor" but also to "the type ofpoor". In this chapter we examine the heterogeneity ofpoverty and vulnerability in rural areas, and highlight some policy options to decrease the vulnerability of ruralpoor households. Themainfindings andpolicy implicationsare asfollows: Different poverty situations are characterized by the type and amount of assets owned by the poor or their lack thereof, as well as by the return on those assets. Economic assets, both tangible and intangible, are important but political and cultural assets are important too. Different combinations o f access to economic, cultural and political assets result inmultiplepoverty situations. Indigenous groups in Mexico are generally deprivedo fmost o f these assets. Family characteristics, in particular family size and the family life cycle, are related to poverty, especially in their interaction with access to assets. Asset position and demographic family conditions act together to form specific poverty situations. Illness i s the main idiosyncratic shock whereas natural conditions like pests and diseases and droughts are at the origin of the main covariate shock, hitting farmers in particular. Rural households in Mexico typically manage vulnerability combining risk reduction, mitigation, and coping instruments.For the poor, increased labor market participation, i s the most important response to shock. Income diversification, migration and subsistence farming are part o f risk management strategies. There i s evidence that rural households in Mexico are very much affected bybothidiosyncratic and covariate shocks, butthat they are comparatively successful in smoothing consumption. These practices may come at a high cost in terms of future growth prospects, however. Moreover, many mechanisms become ineffective inthe face ofsystemic shocks (e.g. dueto labor surplus or risk-pooling). Existing formal agricultural insurance systems, while not appropriate for the poorest, can be usefulfor farmers intransition, especially ifthey want to diversify into high value crops. Similarly, the introduction o f parametric insurance systems would increase the insurance options for the rural poor, especially if it goes together with the development ofrural finance. Given the difficulties with coping with covariate shocks, programs like Fonden are useful mechanisms to reduce the vulnerability of the rural poor, especially vis-a-vis covariate shocks. 121 Rural financial systems are multipurpose instruments that serve for risk management as much as for capital accumulation, technology adoption and personal welfare. They probably are the single most important formal system to assist the rural poor to manage risks, particularly idiosyncratic risks, because they facilitate savings, personal loans, agricultural insurance, and productive loans that encourage income diversification andmigration strategies. Strong support to the subsistence economy is a major policy option inview of its importance as a safety net for poor producers. Support to the subsistence economy i s probably best carried out at the local level inthe framework o f municipal or micro- regional plans like those promoted by the Microcuencas program. DIVERSITYOF RURALPOVERTY SITUATIONS Poverty is a complex and multifacetedconcept which can be approached in many ways each highlighting different dimensions (Box 7.1). Because of this and because of the geographical and cultural differences o f Mexico rural areas, there are many ways to be a Mexican rural poor. Differences among the poor are not easily recognizable from the distance, or they are but as positions on an income continuum cut off by appropriate poverty lines. They are, however, all too evident for the concerned families. That i s why poverty aggregates, however useful for many purposes, always conceal an element o f deceit. And that i s also why poverty programs need not only be targeted to "the poor" but also to "the type o f poor". Some poverty instruments, such as income support through direct cash transfers or the provision o f basic government services (education, health, social infrastructure), have a broad spectrum and can serve many types o f rural poor. Other instruments, especially those aimed at employment and income creation or at improving risk and environmental management by the poor, are very specific to the particular circumstances and mustbe targeted accordingly. Box 7.1. DifferentViews and DimensionsofPoverty Poverty can be seen as failure to keep up with the standard prevalent in a given society or, as the European Union defines it, as the situation o f "persons, families or groups o f persons whose resources (material, cultural, social) are so limited as to exclude them from the minimum acceptable way o f life inthe member state inwhich they live". Poverty definitions can emphasize basic needs, lack o f participation and self-esteem, vulnerability, lack o f capabilities and opportunities, and constraints to proper human development. Poverty measures can consist o f individual or household indicators, include only private welfare indicators or value publicly provided goods too, consider only monetary components or also non-monetary ones, assess poverty at a point in time or over the life-cycle, include only actual poverty or also potential poverty, use flow or stock measures (e.g. income vs. wealth), use input or output concepts (e.g. income vs. welfare), use absolute or relative indicators (e.g. poverty lines or position in a distribution), and adopt objective or subjective approaches (e.g. assessed income vs. community poverty ranking or self-perceived poverty). See Maxwell (1999). Sen's concept o f poverty is that currently enjoying most popularity among a broad range o f scholars and practitioners o f different disciplines and persuasions. It is a philosophical concept which sees poverty as people's inability to choose among different types o f possible lives because o f the lack o f capabilities or opportunities available to them. They are thus deprived o f what they could potentially do or be, i.e. have reduced "functionings", as Sen chooses to put it. Poverty i s a deprivation o f fieedom, because there is an economic dimension o f freedom related to an increase inthe choice set open to individuals that can only come with development. Hence the idea o f development as freedom (Sen, 1984, 1985 and 1999). 122 Rural Poverty andAccess toAssets We start by examining access of the rural poor to economic assets, which following Siege1 and Alwang (1999) can be grouped in two broad categories: tangible and intangible. Tangible assets are the best known ones and those usually measured in surveys, but intangible assets can be equally important to family or individual welfare especially in the event o f economic shocks. Intangible assets refer to entitlements or capabilities to access certain sources o f employment or income, and hence to future flows o f income. While tangible assets can only have positive or zero value, intangible assets can be negative because other parties can have claims on our future flows o f income. We list economic assets inTable 7.1. Tangible assets are of three types: production assets, consumption assets, and financial assets. The first include access to own and collective lands, favorable agro-climatic conditions, irrigation rights, plantations, animals, and tools and equipment; the second includes house and household goods, stored produce and standing crops; and the third includes cash and financial savings. Differences in access to these assets determine different types o f material poverty and vulnerability. Thus, there i s evident dissimilarity betweenlandless rural laborers that have only human capital as a production asset and farmers who also have land and animals. The sources o f income will normally be different and the types o f risk and risk strategies too; landless laborers will depend on wage income and face the risk o f unemployment, while small farmers will normally depend more on farming income (unless they have very little land) and will face productive and market risks. Vulnerability will also depend on whether the household has cash, liquidsavings, stored food or animals to sell inthe event of emergencies. 123 Table 7.1. Economic Assets of the Rural Poor TangibleAssets Intangible Assets, with effect Positive Negative Ownland Formal or informal entitlement to Economic vulnerability to natural private or public transfers, including and market risks (depends on Rightto use collective rents physical environment, crop lands technology and mix, diversification Personal goodwill: capacity to o f income sources, type o fmarket Favorable ago-climatic readily obtain credit on need and market access, and personal conditions (dependson reputation, contacts, and characteristics) availability o f other assets) Irrigation rights Standing debts and commitments to Command over the labor market: make transfers, including rent payments Large and small animals capacity to readily obtain employment at the going rate Standing crops and ready (dependson personal characteristics, to sell crops including gender, experience, education, commitments to other Plantations activities like agriculture or caring after own children, and on location Stored food or other and contacts) produce Command over off-farm activities: Tools and Equipment capacity to readily enter off-farm markets as a producer (depends on Ownhouse and household skills, equipment, access to inputs or goods operational capital and market contacts) Cash inhand and saving deposits Command over migration: capacity to undertake successful transitory migration (dependson personal characteristics, past migration experience, contacts, and availability o f cash to migrate) Human Canital Source: Author's construction. Poverty and vulnerability will be different accordingto the availability of intangible assets. Thus, for instance a household entitled to Procampo, whose family head has a good standing in the community and good market contacts allowing him or her to get local part time employment when required, and has well established migratory links so that he or she can migrate duringthe low agricultural season and earn a complementary income, is in amuchbetter position than a household without those assets. Access to tangible and intangible assets combine in practice invarying ways producing many different poverty and vulnerability situations. Poverty is affected by the amount of assets (e.g. land, family members available for work), as well as the return on their asset (e.g. prices, wage rate) Thus, agricultural prices are vital to the owners o f farming assets and so i s the wage rate to rural laborers. Higher returns on the assets o f the poor decreasespoverty, and changes inthe distribution o f returns across different assets affect their value to the poor, altering specific poverty situations. 124 In additionto economic assets there are other assets not always easy to identify and generallydifficult to measure that add new dimensions to the situations of poverty.Political assets are an example o f this. By political assets we understand access to things such as not being discriminated on political, religious, ethnic, gender or other grounds, and being able to exercise recognized citizen rights such as voting, parity access to legal, administrative andjustice systems, and the right to one's own privacy and identity and to associate with others to pursue specific interests. Political assets are hence an entitlement to citizenship -to be able to fully and equally participate in all things pertaining to the public sphere. Another example i s cultural assets. We can include here things like (1) the confidence and satisfaction derivedfrom having and sharing with others a recognizedidentity and awareness o f ones own origins, and (2) the ability to manage the (often subtle) norms and idiosyncrasies o f the forms o f life o f one's own society, including language, traditions and folklore. Cultural assets can become a form o f capital, as when musical or plastic arts traditions are exploited as an attraction in tourist business operations, or when the own style o f a community's handicrafts acquires value as a commercial trade mark. As capital, cultural assets can be important to the poor, but they are more than that; they are also a source o f self-esteem and a kind o f cement binding together many forms of social capital. Social capital is at the crossroads between a productive and a cultural asset. It can generate income or other forms o f value and hence it i s a productive asset, but it i s also "social", thus implying solidarity or connectedness among individuals over and above mutualbenefit (Uphoff, 2003). From this perspective, indigenous groups in Mexico suffer from multiple deprivations. Indigenous groups in Mexico tend to have less political and cultural assets than other groups, and this i s generally accompanied by economic deprivation Thus, results from a recent study by Ramirez and Garcia (2004) show that indigenous workers have much lower earnings than their non-indigenous counterparts after controlling for personal characteristics, education and sector o f employment. A 59 percent part o f the difference in earnings i s explained by those characteristics (higher education of the non-indigenous, better employment sector, and so on), but there i s another 41 percent which cannot be explained by them, the only explanation beingfactors such as quality o f education, culture or labor market discrimination. The same i s the case with the probability o f being poor even within the same educational group or sector o f activity. Thus, an indigenous worker in agriculture has a 72 percent probability o f being extremely poor against a 34 percent probability o f the non-indigenous, and an indigenous person with 6 to 11 years of schooling is twice as likely to be poor and four times as likely to be extremely poor than a non-indigenous person with comparable education. When examining participation o f different groups in non-farm income opportunities a similar pattern emerges (Janvry and Sadoulet, 2001). After controlling for other variables, education among them, indigenous groups have more difficulty in accessing non-farm employment. "Young indigenous adults suffer from a double disadvantage for income generation: they lag in educational progress, and they derive lower benefits from education in accessing more remunerative nonagricultural employment" (p. 473). The distribution of assets is extremely unequal, as shown in Table 7.2 where Gini coefficients are calculated for the distribution of differenttype o f rural assets and o f rural incomes from different sources. Gini coefficients for asset are extremely high with the exception o f the average years o f schooling. The number o f migrants i s used as a proxy for migration "capital" and i s divided in internal, USA, and total migration assets. All three Ginis are very high. Coefficients overstate, however, the extent o f asset inequality, because we should not expect all rural households to have access to all assets. Thus, there i s no reason why non-farming households should have land or animals. If we could put some value on human and migration 125 capital and estimate the total value o f the assets owned by each family, the distribution of this value would probably be less unequal than that o f its separate components. This i s the case with respect to income figures; the Gini coefficient o f total income i s smaller than that o f most o f its components. It i s interesting, but not surprising, that incomes from wage employment and government transfers are better distributed than those derived from self-employment (in or outside agriculture) and private transfers. Physical& Human Capital and Migration Assets Net Income from Different Activities Typeof Asset Gini Typeof Activity Gini Household Average Schooling 0.25 Agricultural Wage Employment 0.5 1 Household Head Schooling 0.61 Government Transferences 0.53 Total Migrants 0.79 Natural Resources 0.55 Internal Migrants 0.84 N o nAgricultural Wage Employment 0.56 U S Migrants 0.90 Intemal Remittances 0.64 Landholdings (hectare) 0.85 Intemational Remittances 0.65 Steer 0.95 N o nAgricultural Production 0.68 Horses 0.89 Cattle 0.77 Pigs 0.95 Staples 0.77 Large Animals 0.90 Cash-Crops 0.83 Tractors 0.95 Other Agricultural Production 0.83 Small Animals 0.86 GiniNet Total Income 0.57 Poverty and theRural Family There are different types of families in rural Mexico, influencing the way poverty and vulnerability affects household members. There are also variations in the concept o f family as an institution, even if essential functions like bringingup and socializing the children and looking after the sick and the old remain. The two basic types o f families are the nuclear monogamist family consisting o f parents and children but possibly also o f survivingmembers o f older generations, and the extended family with various related nuclear families forming a single household. In some indigenous groups, like the mazatecos, huicholes, tzeltales, totonacas, coras and others, there are also polygamist families (Nahmad and Carrasco, 2004). A frequent practice inruralMexico is the informal adoption ofabandoned children or children ofmigrant relatives or o f broken families. Rural families tend to be larger than urban ones, and indigenous families are particularly large. Thus, the average size o f rural indigenous families i s 6.4, while the size o f rural non-indigenous families i s 5.8, and that o f urban non indigenous families 5.1. The average family size for Mexico as a whole i s 5.3.87 Family size tends to be an asset in survival strategies and risk management. In principle, large families have four advantages over smaller ones: (1) they have more opportunities o f self-insurance through diversified income sources by having family members employed in different works, (2) they have wider networking opportunities because o f different member occupations, including more migration contacts, (3) they enjoy economies o f scale in 87Data from the 2000 population census compiled by Ramirez and Garcia (2004: Table 9). 126 consumption, and (4) they can mobilize more family labor when required, which may be important for farming families at harvest time or to contribute to communal construction works or to look after animals or collect firewood. Yet, young large families with high dependency ratios are at a disadvantage. Size cannot be seen independently o f the family cycle and the dependency ratio, whose importance was observed in chapter 2, and has long been recognized in the economic literature on the peasantry.88Young couples with many young children have a particularly hardtime because they have many mouths to feed and only their own labor available to earn income, and labor has to be shared between production and very demanding reproduction work. As children grow up and become increasingly able to contribute to income generation and household labor the situation changes. Thus, as reported in chapter 2, in 2002, Mexico's rural families with dependents below 11 years had a probability of being extremely poor 22 percent higher than families without dependents, other things beingequal. The probabilityfalls to 14percent for families with children between 13 and 14 and to 6 percent for children between 15 and 18, becoming negative for families with children o f 18 to 25 years o f age still living inthe household. The asset position of the family i s o f course important inthis respect, acting together with the demographic conditions to form specific poverty situations. Thus, for instance, having access to land i s comparatively advantageous for larger families because they can use family labor in agriculture or animal husbandry, and they can produce their own food (or part o f it), hence being more food secure. Landless families o f similar size dependent on wage labor do not have productive use for family (particularly children's) labor, and have to buy all their food with their wages. They are hence extremely vulnerable to the vagaries o f the labor market. Family size i s probably not an advantage inthis case. Poverty and theRural Community Rural communities are groups of families living nearby who have close interactions and in some ways depend on each other. They may have a collective title on land, as in the ejidos and comunidades campesinas, or they may not, although the more cohesive communities typically hold resources in common. Mexican rural communities tend to share five characteristics: (1) a set o f relations among member families that were formed over time; (2) a definedterritory constituting the community's natural environment, collectively owned or not; (3) productive activities common to many members, and widely shared technical knowledge and practices, which usually go together with other forms o f identity and shared culture; (4) an organizational structure to ensure governance and natural resources management; and (5) a system o f social stratification, which i s more important and visible in the larger communities, with divergent group interests coexisting within the communal system (Nahmad and Carrasco, 2004). Rural communities have different sizes and degrees o f cohesion and their formal status may also be different. Households in small aldeas o f thirty to fifty families tend to be more income poor and less supplied with services than those in larger communities o f 2,000 to 5,000 inhabitants. There i s a hierarchical system o f rural communities, which goes from the rancheria and paraje to the agencia de policia, agencia municeal and cabecera municeal, with marked differences in their access to communications, social services and economic opportunities. Also, the quality o f natural resources i s often better in the higher ranking communities, because richer areas were usually populated earlier andhave also grown more. 88 Chayanov (1966) is the classic analysts of the peasant family cycle and how resources and needs change along it, based on the experience of the Russian peasantry at the turnof the 19thcentury. 127 It is difficult to say how large can a rural settlement be while still remaining a community inthe sense described above, as this depends much on the degree of cohesion. There are fairly large rural settlements in Mexico o f say 10,000 to 15,000 residents which still keep a "sense o f community", mostly due to the effective working o f communal governance structures and a proud sharing o f traditions. There are also rural settlements that never were or were but have stopped being proper communities and function more as pure family agglomerations. Indigenous settlements, even if large, tend to be more cohesive and keep communal traditions more than mestizo villages. The special electoral system o f local authorities known as usos y costumbres that operates in indigenous communities willing to adhere to it contributes to this cohesi~eness.~'Valuable common resources, such as forests and pasture lands, which need to be commonly managed, do also contribute to keeping communities more united and better structured. Communities can coincide with municipalities: In Oaxaca, for instance, where municipalities are characteristicallysmall, there are many single-communitymunicipalities. Inother cases, ruralmunicipalities consist ofvarious communities, one ofwhich is the cabecera munic@al and the others agencias or simple rancherias if very small. Indigenous communities usually have an elaborate system o f ad honorem authorities or cargos with functions o f responsibility at the service o f the collective. Male members are expected to occupy a progressive succession o f these cargos along their lives, and they must do so if they want to be respected or even to keep their full communal rights. Reallocation of farm land was a major function o f communal authorities in the past, but has decreased in importance with the mounting scarcity o f cropping land, the advance o f privatization, and the clear delimitation o f rightsunder PROCEDE. Decision-making with respect to common property resources continues to be a crucial responsibility o f communal authorities. Four economic functions of communities are relevantto poverty and vulnerability. The first is their role in the generation of social capital. Community ties and governance systems are themselves an instance of social capital, but they serve also as facilitating frameworks for more private sources o f social capital such as privately organized collective enterprises or labor exchanges or the functioning o f special interest groups. These forms o f social capital enhance income generation and decrease vulnerability, especially to idiosyncratic shocks. The second relevant function i s the management o f the community's common pool resources. The importance o f these resources varies among communities but i s large in many o f them," allowing members to share in the income and employment generated. Third, communities are distinct providers o fpublic goods. This i s the case with many public works, often carried out with voluntary communal labor underfaenas or tequios, and also with things such as the resolution o f conflicts and the provision o f local policing and petty justice services. Finally, communal authorities and governance systems serve as mediators with the larger society fulfilling a bridging function to attract economic support from government or NGO programs. Demographic pressures, the advance of the market economy, and increasing migration have diminished the social protectionfunction of communities." The historical decline inthe redistribution o f communal lands to young generations, mentioned above, i s a good example. The same forces that erode cohesion in the communities and reduce their social s9 Under usos y costumbres indigenous representatives elected in communal assemblies by a show of hands are recognized as valid municipal authorities for political and administrative purposes without need of a formal ballot. The rationale for this system is not to duplicate governance systems at the locallevel. 90Thus, for instance, 70 percent of Mexixan temperate and tropical forests belongto ejidos and comunidades. 91See Platteau (2002)for the analysis of a similar process ina different context. 128 protection capabilities are the ones that generate increasing inequality. Market development and migration open up new opportunities from which members benefit in different measure. This generates poverty, not necessarily in the sense o f some sectors o f the community beingworst off inabsolute terms thaninthe past, butinthat ofthemfalling behindother groups. And, after all, in a very relevant sense, in many political and cultural settings poverty i s not but the way in which inequality i s historically constructed. Illustration of Poverty Situations We present in Table 7.3 three cases of poor rural families indicating their economic characteristics, their social and cultural capital, and their survival and risk management strategies. The cases are hypothetical but are consistent with the figures presented on rural labor structure, poverty correlates and income sources, and also with ethnographic data. The first family is a young one with children of small age and no land. The characteristics o f this family are summarized in Table 7.3.a. We have assumed that this family lives in a small remote location but could also live in the outskirts o f a rural town. Other than its crude labor force, the family has practically no assets -a desperate case o f lack o f "capabilities". The husband is a casual laborer and the wife hires herself to do agricultural labor to the extent compatible with her family obligations. There are no migrant sons or daughters that can send transfers, because the children are too young. And there i s no Procampo subsidy because the family has no land. They receive instead Oportunidades,and this i s o f critical importance to their well-being. The husbandwould like to migrate temporally to make some income when the local labor market i s inactive, but has no good connections and very little money to travel and survive while he finds work. If the income o f this family were measured it would likely be below the extreme poverty line. The second family, whose characteristics are shown in Table 7.3.b., is a middle-age one living in a small rural community. It has some rain-fed land on poor soils allowing the family to produce some food for its own consumption and to keep some animals. This i s not enough, however, and the husband has to complement family income by working as a casual laborer. H e had migrated temporally in the past, and that allowed him to accumulate enough to builda house andhelpraising their children, but he does not migrate any more; with the food the family produces, payments from Procampo, some money from Oportunidades for the last child attending school, another money sent by a migrant son, the earnings from his casual work, and some income contributedby the wife who helps ina petty commerce operation runby a neighbor, the family has enough to go by. If the income o f this family were measured it would likely be betweenthe extreme andmoderatepoverty lines. 129 Table 7.3.a. Illustration of a Young-to-Middle-Age Poor LandlessFamily in a Remote Small Location Economic Characteristics Social and Cultural Capital Survival Strategies and RiskManagement Main asset: small homesteadwith Parents without education or Rents or borrows aplot of land a mudhouse o f 1or 2 rooms primary incomplete. Children for seasonal cropping when without electricity or running attend primary school, not possible water, located outside the village. secondary Tries to develop continuing work Small kitchen garden with a few Wife and small childrenmono- relations with a local land owner chickens, perhaps a pig. lingual (inindigenous areas) Wife works occasionally as Uses wood for he1 N o migrant sons or daughters domestic help or as farm hand or makes some handicrafts (the Receives Oportunidades Few or no visits to medical unit latter mostly inindigenous areas) (more now with Oportunidades). Does not receive Procampo Use o f traditional medicine Inemergencies, they sell small (especially inindigenous areas) animals and request loan from Receives some grain help from employer, better off relative or relatives with land or from Little participation inreligious local money lender employer and communal activities Temporary migration if and when Main occupation and source o f Fewrelations o freciprocity with possible income is casual labor in relatives and neighbors agriculture N o participationinpolitical Has few migratory connections activities o f community or municipality Marked subordination to employer Possible alcoholism and domestic violence Source: Author's construction based on Nahmad and Carrasco (2004). The last family, illustrated in Table 7.3.c., is also a middle-age one living in a small- to-medium rural location. The family owns three or four has o frain-fed land and another two or three hectare o f irrigated land. It also has some large animals which are used for work, to provide milk for family consumption, and as a form of investment. Rain-fed lands are used to produce food for the family while a cash crop i s grown in irrigated areas using improvedtechnology. The family receives Procampo but not Oportunidades and has other income from various sources: work done by the husband supervisingoperations in the farm o f a distant relative who moved to the state capital, occasional transfers from two migrant sons, one in the USA who has children staying with their grandparents, and income from the wife who participates in a rather successfd women's horticultural group existing in the community. As in the former case, the husband has migrated in the past but does not want to migrate any more. If the income o f this family were measured it would probably be inthe border between poverty and non-poverty. Vulnerability for this family wouldmeanthe risk of falling into poverty. 130 Table 7.3.b. Illustration of a Poor LandedFamily in a SmallRuralLocation Economic Characteristics Social and Cultural Capital Survival Strategies and RiskManagement Main assets: 2 to 5 has o f rain-fed Parents without education or Family diversifies income and risk land inhilly soils. House with primary incomplete. Children with casual work and other earth floor o f 2 or 3 rooms with attended primary school, not occupations tinroof, electricity andno secondary runningwater. Kitchen garden Parents support migration o f elder and a few fruit trees for family Wife and small childrenmono- sons and daughters helpingthemto consumptions. Has a few small lingual (inindigenous areas) keep linkedto the community. animals or one large animal grazing in communal lands Family has migrant son or Parents try to convince the younger daughter son or daughter to stay inthe Produces maize and beans plus community to help inthe farm and some horticultural products with Few or no visits to medical units look after them in old age. minimuminputs.Landrotation (more now with Oportunidades). and fallows. Soil degradation. Use o f traditional medicine Husband sticks to his land without Stores produce for self- (especially inindigenous areas) fractioning it or passing it to his consumption. There is no grain offspring surplus. Participation inminor religious and communal activities Wife works occasionally inpetty Does not use credit or receive commerce or making some technical assistance Relations o freciprocity with handicrafts (the latter mostly in relatives and neighbors indigenous areas). She may Uses wood for fuel. Collects non participate in a women's production timber products from communal N o participationinpolitical or savings group areas activities o f community or municipality Inemergencies, husbandparticipates Receives occasional transfers more inlabor market and wife in from a migrant son or daughter trade or handicrafts. They may sell animals and request a loan from Receives Oportunidadesand better offrelatives or a local money Procampo lender, or ask migrant children for incremental support Works part time on his farm and also as casual laborer Husband has beentemporary migrant but does not migrate any more. Source: Author's construction based on Nahmad and Carrasco (2004). The families discussed do not only differ in their economic condition but also in their social and cultural capital and in the way they manage risk. Culturally, the first family lives in a vacuum o f social relations and communal integration that makes its poverty situation more disheartening because not culturally dignified. Alcoholism and family violence are common features in this situation inrural Mexico, and indeedin many other rural areas of Latin America, for example in the Andes (see for instance World Bank 2002b). The second family i s better integrated although it plays a modest role inthe management o f local collective affairs. The third family i s fully embeddedinthe norms and culture o f its community, and manages well the part o f that culture that consists o fknowing how to use the links with the outside world. 131 Table 7.3.c. Illustration of a MatureBorder-Poor Family in a Small to MediumRuralLocation Economic Characteristics Social and Cultural Capital Survival Strategies and RiskManagement Main assets: 5 to 10 has some rain- Parentswithout educationor primary Family diversifies income and fed and some irrigated (with water incomplete. Childrenattended managerisk with crop diversification rights). House o fbrick and adobe primary and secondary school. and other occupations and with cement floor, o f 3 to 5 rooms, participating in govemment or NGO with tinroof, electricity andrunning Wife monolingual (inindigenous programs water. Kitchen garden and a few fruit areas) trees for family consumptions. Has Parents supports migrationof elder several small animals or three or four Family has migrantsons or daughters sons and daughters helping them to large animal grazing incommunal keep linkedto the community. lands. Family combines formal medical services and traditional medicine Parentstry to convince the younger Producesmaize andbeans plus some (especially inindigenous areas) son or daughter to stay inthe horticulturalproductswith minimum community to help inthe farm and inputsfor self-consumption. Active participation inreligious and look after them inold age. Produces also some commercial crop communal activities for sale Husband sticks to his landwithout Frequent relations o f reciprocity with fractioning it or passing it to his Eventualuse o f credit and relatives andneighbors offspring technical assistance Active participation inpolitical Connection with extemal agents and Uses wood and gas for fuel. activities o f community or development programs i s sought as municipality self insurance andfor petty Receives occasional transfers from accumulation migrant children Husbandparticipatesinproducers' associations andincommittees Wife works occasionally inpetty Receives Procampo sponsoredbyNGO or govemment commerce or making some programs handicrafts (the latter mostly in Husbandworks part time on his farm indigenous areas). Shemay and may also have some other participate ina women's production occupation butnot as casual laborer or savings group Inemergencies, the family sells animals, requestsa loan from outside connections or a localmoney lender, or asks migrant children for incremental support Husbandhas been temporary migrant butdoes not migrate any more. Source: Author's construction based on Nahmad and Carrasco (2004). In view of its asset situation and very small risk pool, the first family is extremely vulnerable and has very few instruments to manage risk. It i s the combination o f having young children, no land and no connections that makes this family particularly vulnerable. Access to land -if the family could rent it on a continuing basis and had access to credit to farm it- would be a good way for this family to improve its income position and decrease vulnerability. Diversification o f income sources and access to public and private transfers make the second family less vulnerable. Not only does it have its own source o f food, but has also good command over the labor market which allows it to intensify wage employment if necessary. The risk position o f the third family i s quite robust; its sources o f income are much diversified and it has access to large social capital and a large risk pool. Only under exceptional circumstances, it would seem, would this family fall into an extreme poverty condition. Yet, there are many families in the coffee growing areas o f Mexico who were in situations similar to this one only a 132 few years ago, who have been struggling during the last years to keep on the other side o f destitution. INCOMEAND CONSUMPTIONVULNERABILITY OF THE RURAL POOR There are differentways in which families can be vulnerable to income shocks and also differentways inwhich they can manage risk. Inthis section we examine the vulnerability o f the Mexican rural poor to idiosyncratic and covariate shocks, the reduction, mitigation and coping strategies that they use, and the effect o f some formal mechanisms, like crop insurance, and some government programs. The conceptual framework for vulnerability used in our analysis i s summarized in Box 7.2. Box 7.2. ConceptualFrameworkfor ExaminingVulnerability There are three aspects o f vulnerability: (1) the risk or risk event, (2) the options for managing risk or risk response, and (3) the outcome o f the risk event and its welfare impact (Alwing et al, 2001). Riskmanagement has ex ante and expost facets. Measures are taken ex ante to reduce risks, for instance when agricultural risks are lowered by using drought resistant low- yielding seeds. Ex ante measures can also be taken to mitigate risk, by diversifying for instance assets or income sources. After the risk event or "shock", steps are taken to cope with its effects, like selling animals or procuring a loan from a relative. A common practice in rural areas i s for households to "pool risk", i.e. to form systems whereby those who have experienced a shock can count on the assistance o f those who have not. This can be done formally, like in the Mexican fondos de aseguramiento, or informally through traditional mutual assistance systems. The extent to which a household can count on the help o f others inthe event o f shocks is its "risk pool". Consumption smoothing mechanisms, while essential to the survival o f a household faced with income shocks, can come at a high cost, especially interms o f future income opportunities. Risk events usually affect directly income and as a consequence consumption. Households try to defendtheir consumption levels through copingmeasures. The result is "consumption smoothing", i.e. a smaller variation o f consumption than o f income. But smoothing i s not without cost, and the cost can be dear to the poor, like the distress sale o f assets or forced migration. Income itself can be smoothed ex ante using risk mitigation measures like those mentioned above. Risk management strategies by individuals would normally combine risk reduction, mitigation and coping instruments according to their accessibility and cost in the particular circumstances (Siege1 and Alwang, 1999). Risk strategies do not only serve to protect consumption levels from eventual shocks, but help also households to take higher levels o f risk, like adopting more profitable but risky farming technologies or moving from low input field crops into high input niche market crops (Holzman and Jorgensen, 2000). Shocks can affect either single individuals or very small groups, i.e. be "idiosyncratic", like most health conditions, or can simultaneously affect a large number o f individuals, i.e. be "covariate" or "systemic", like most shocks related to weather situations. Covariate shocks are more difficult to deal with at the local level because risk pooling mechanisms normally fail, and because these risks are as a rule more difficult to insure against. How vulnerable are theRural Poor in Mexico? Rural households are fairly successful in protecting consumption from frequent income shocks, both idiosyncratic and covariate.Using panel data from ENCEL, the survey applied to evaluate Progresa/Oportunidades, for October 1998, June 1999, and November 1999, covering 506 poor villages and 24,000 households, Emmanuel Skoufias has shown in a recent paper (Skoufias, 2004) that covariate risks significantly affect household incomes and consumption, although households do carry out income smoothing practices that partially protect 133 their incomes from such risks. Systemic shocks, however, are shown to be of secondary importance with respect to idiosyncratic ones. Also, panel data shows that systemic shocks such as weather related risks and natural disasters can have very different impacts on households, and that shocks leading to income changes do not necessarily lead to consumption changes. Successkl practices o f consumption smoothing make consumption more protectedthan income. Illness comes out as an importantsource of risk.Householdswhose headis temporally illexperienced income growth rates 20 to 22 percent below those of other households. Interestingly, consumption decreased by 2 percent only, showing the strength of smoothing practices. Longer illnesses had a milder effect on income, probably because the length o f the illness made households resort to income smoothing practices. Among production-related shocks, the one to come out more strongly in these panel data i s the incidence o f pests and diseases. Households who experienced a problem of pests and diseases had an average income growth rate 16to 17 percent below other households. Again, the impact on consumption was much smaller; a drop o f only 3 percent. Table 7.4. EconomicShocks inRuralAreas in 1989-94 Sector All Farm Non- Farm Bottom Third Fourth Farm & two quartile quartile Non- quartiles Farm Percent of respondentsthat 59 63 56 60 51 66 66 faced an economic crises Type of Shock (%) Low Yields 23 48 5 22 11 27 27 Low Prices 6 11 3 4 4 4 11 Low Sales 21 1 34 33 20 21 22 Weak Demandfor Services 12 13 18 13 18 8 7 Illness of Entrepreneur 4 6 3 2 3 7 2 Other 9 11 9 5 12 6 9 Subtotal 75 80 72 79 68 73 78 High Expenditure Due to Illness of Entrepreneur or 18 18 18 17 20 22 15 Household Member Other 7 2 10 4 12 5 7 Subtotal 25 20 28 21 32 27 22 TOTAL 100 100 100 100 100 100 100 Source: World B a n k (1995). The incidence of shocks is important. A survey of 1,944 rural entrepreneurial households carried out in mid-1994 in Guanajuato, Puebla, Tamauleas and Veracruz, reported economic shocks occurred between 1989 and 1994 (World Bank, 1995). The results are shown in Table 7.4. Fifty nine percent of respondents reported a shock in the period. Incidence was larger among farmers, reflecting the risk o f agricultural activities. For farmers, the most frequent shocks were related to agricultural production resultingin low yield, whereas for non farmers low sales was the most common shock. In both cases weak demand comes second as a source o f risk. Illness o f the entrepreneur (normally the householdhead) or other household member i s an import source o f high unplanned expenditures. Poorer households in the two bottom quartiles experienced less shocks than richer ones, probably because the risk-aversion o f the poor make 134 themtake lower risks to avoid facing the catastrophic consequences of income shock. Low yields due to productionrelatedshocks were significantly smaller inthe bottomtwo quartiles, which can be attributedto the planting o f lower risk crops andthe use o f low risktechnology bythis group. Weather related shocks are extremely frequent for small Mexican farmers. The Encuesta Nacional de Hogares Rurales de Mkxico (ENHRUM) survey of close to 1,800 rural households conducted in 2002 by the Colegio de Mkxico and INEGI gives information on weather-related shocks inthat year for 666 crop farmers. The results are shown inTable 7.5. The frequency o f weather-related shocks i s worrisome: in 2002, they affected 44 percent o f farmers throughout the country. Contrary to Skoufia's results from the ENCEL survey reported above, droughts are inthis survey the shock most commonly experienced, more than pests and diseases. The high incidence o f shocks i s corroborated by information from the survey undertaken by the World Bank and FIRA to examine rural financial markets inMexico (see footnote 3 to chapter 5), shown inTable 7.6. Almost half o f all farmers inthe survey experienced adverse events in 1999- 2002. Importantly, in contrast to micro-entrepreneurs, more than half o f the farmers experienced systemic shocks. Table 7.5. Weather-Related Shocks Registered by Crop Farmers Problems N u m b e r o f Droducers Percentage Withproblems 295 44.3 Excessive Rain or Hurricane 71 10.7 Droughts 162 24.3 Pests and Diseases 40 6.0 Frosts 22 3.3 Without Problems 371 55.7 Total 666 100 Source: Computed from ENHRUM. Table 7.6. Mexico: Incidence of Adverse Shocks on Farmers and Micro-entrenreneurs inRuralTowns in 1999-2002 Farmers Micro-entrep. Incidence of None 53.0 73.3 Adverse Events Only systemic 19.8 6.7 Only Idiosyncratic 8.0 11.8 Both 19.2 8.2 Number of None 53.0 73.3 Adverse Events One 20.9 14.8 Two 9.8 6.5 Three or more 16.3 5.5 Source: World B a n k (2003: Table 9B.1). Rural households are affected by shocks differently according to their characteristics. The distribution of shocks i s not independent o f the distribution o f household characteristics such as size, education, field o f employment, occupation or location. Using panel data from the national employment survey (ENET) for 2000:2 and 2002:4, Maloney et a1 (2003) have investigated the effects o f differenthousehold characteristics on the distribution of shocks in rural and urban areas. The results are reported in detail in World Bank (2004) and will be only summarized here. An important finding i s that the effect o f household characteristics on shocks are rather similar for rural and urban areas, i.e. characteristics such as education, family 135 composition, being self employed or being formally employed affect the distribution o f shocks in the two areas in a similar way. Informal workers and the self-employed show much higher variability in income than formal workers. Thus, income falls in the bottom part o f the distribution o f shocks and income increases in the upper part between 2000:2 and 2002:4 were much bigger for informal and self-employed workers than for their formal counterparts. Another finding is that higher levels of education do not contribute to reduce negative income shocks. Thus, at the median, less educated households (with primary education or less) experienced higher income increases than those with better education (high school, technical or university levels). The World Bank rural finance study (World Bank, 2003) examined also the personal and context characteristics correlated to the incidence of shocks. An interesting finding is that the probability of incidence of negative shocks for individual farmers is positively correlated with farms located in ejidos and those depending on informal traders (coyotes) to sell their produce, and negatively correlated with irrigation and the use of certified seeds. The probability o f negative shocks increases for farmers with Procampo, for farmers who receive remittances, and for those owning cattle (which may be a consequence rather than a cause o f vulnerability). High income variance does not result in a similarly high consumption variance. Maloney et a1 (2003) use also panel data from the ENCEL survey to compare income and consumption vulnerability between October 1998 and November 2000. Like Skoufias, they observe significant consumption smoothing particularly for large positive and negative shocks. We illustrate this in Figure 7.1 taken from World Bank (2004), which shows how large negative shocks induce a less than proportional fall in consumption and the opposite for large positive shocks. Figure 7.1. The Pattern of Annual ChangeinIncomes in a Panel of Mexican HouseholdsinPoor RuralAreas, 1998-2000 Distributionof LogChange Among Percentiles. ENCELSurvey 3.0 2.0 j 1.0 0.0 80 90 100 -1.0 -2.0 -3.0 Income - - - - - - - Consumption Source: World Bank (2004: 84). Risk Reduction,Mitigation and CopingStrategiesof theRural Poor Households resort to a variety of risk mitigating and coping strategies as a way of self-insurancein order to cope with large idiosyncraticriskscombined and inthe absence of formal insurance markets and credit constraints. These include the accumulation of assets, 136 income diversification, sending women and children to work, withdrawing children from school, input and crop choices, precautionary savings, migration, marriage, income transfers among friends and relatives, and other informal risk sharing arrangements such as share cropping or input sharing (Skoufias, 2004). The use of these mechanisms is shown in Tables 7.7 and 7.8. According to the survey o f rural entrepreneurs in Guanajuato, Puebla, Tamauleas and Veracruz mentioned above, one third o f households affected by shocks resorted to sending more family members to work (Table 7.7.). Some 26 percent of households could secure loans and donations, while the percentage o f affected households that resorted to selling assets was small, only 8 percent o f the total. Defense o f consumption levels i s evident from the data: 57.7 percent o f households that experienced shocks didnot reducedtheir consumption. Table 7.7. CopingwithEconomicCrises Sector All Farm Non- Farm& Bottom Third Fourth Farm Non- two quartile quartile Farm quartiles Percent o f respondents that 59 63 56 60 51 66 66 faced an Economic Shock Percentage that received credit inthe form of: Loans with positive interest rate, either 7 7 7 9 4 8 11 formal or informal Loans fiom 5 5 5 9 5 6 3 frienddrelatives Delayed loan 1 1 1 3 1 1 2 repayments Subtotal 13 13 13 21 10 15 16 Percentage that received donations from Frienddrelatives 12 10 16 6 13 14 11 Government or NGOs 1 1 1 2 1 2 0 Subtotal 13 11 17 8 14 16 11 Percentage that balanced income through Sales o f Assets 8 10 7 4 6 4 15 Increasedlabor market 34 38 31 42 40 37 26 participation Reduced consumption 25 22 28 19 22 24 29 Source: World Bank (1995). Income level is more importantfor the type of responsethan being a farmer or not. Differences inthe way farming andnon-farming households reactedto shocks are small, although farming households tended to sell more assets (probably animals), while non-farming households tended to receive more donations from friends and relatives. In all cases the support received from government and NGOs was minimal. The position in the income distribution affected the response. Not surprisingly, households in the top two quartiles resorted to credit more than those in the two bottom quartiles, who turned instead more to an increased participation in the labor market. Consumption smoothing was more marked among households in the bottom quartiles, which i s consistent with the view that the poor, being close to subsistence, can hardly afford to reduce their consumption. 137 Increased participationin the labor market is an importantcoping mechanism. The coping mechanisms used by the individual farmers and rural micro-entrepreneurs interviewed in the rural finance survey that experienced shocks are shown inTable 7.8. The first thing to observe i s that for farmers two thirds o f the systemic shocks were due to climatic conditions, contrary to the case o f micro-entrepreneurs for whom economic factors were the main cause o f systemic shocks. Increasing the labor supply i s the main mechanism to cope with shocks for both farmers and micro-entrepreneurs. Other than this, there are a series o f strategies (depleting financial and non-financial assets, borrowing from friends and relatives) which are also used but to a much smaller amount. It i s interestingthat borrowing i s hardly used as a coping mechanism either by farmers or micro-entrepreneurs. The strategies to cope with idiosyncratic and systemic shocks are somewhat different, but not much; they are both dominated by increases in the labor supply, although somewhat less inthe case o f idiosyncratic shocks. Table 7.8. Actions Takenby Farmersand Micro-entrepreneursto Copewith Systemic and IdiosyncraticShocks Systemic Shocks Idiosyncratic Shocks Total Climatic Economic Others Total HH Health Business Other Income Problem Farmers 100 65.5 33.3 1.2 100 62.0 15.9 19.0 3.1 Borrowingwith interest 1.9 1.o 3.5 6.7 4.6 3.8 4.7 6.5 8.0 Borrowingwithout 1.o 0.9 1.2 __ 2.0 0.8 3.9 3.2 8.0 interest Other financial services 1.1 1.5 0.3 __ 1.2 1.2 0.8 1.9 __ Depletedfinancial assets 5.0 4.6 5.7 6.7 8.2 7.7 10.9 7.7 8.0 Depletednon-financial 6.6 7.3 5.5 __ 5.9 5.0 10.9 5.8 __ assets Transfers from 2.4 2.8 1.5 6.7 1.4 1.2 0.8 0.7 12.0 institutions Transfers from friends & 5.6 5.8 5.2 6.7 7.3 6.3 11.6 6.5 8.0 relatives Increased labor supply 63.3 61.9 63.4 53.3 57.6 64.4 51.2 45.8 28.0 No strategy 13.1 14.2 10.7 20.0 11.9 9.7 5.4 21.9 28.0 Micro-entrepreneurs 100 38.2 58.8 3.0 100 36.1 22.4 35.5 6.1 Borrowingwith interest 0.9 _ _ 1.5 __ 0.7 0.3 1.4 0.6 1.7 Borrowingwithout 1.9 0.8 2.4 4.8 1.8 1.7 1.4 2.1 1.7 interest Other financial services 4.0 4.5 2.9 19.1 3.0 2.9 1.9 4.1 1.7 Depletedfinancial assets 6.0 6.0 5.4 19.1 6.2 6.1 7.9 5.3 5.2 Depletednon-financial 2.6 2.3 2.9 __ 6.1 3.8 13.6 3.8 5.2 assets Transfers from 2.0 4.1 0.7 __ 0.6 0.6 0.9 0.6 __ institutions Transfers from friends & 5.9 10.2 3.2 4.8 10.5 9.6 19.2 4.4 19.0 relatives Increased labor supply 63.6 54.9 70.2 42.9 57.5 64.9 43.5 61.1 44.8 No strategy 13.2 17.3 10.7 9.5 13.6 10.1 10.3 18.0 20.7 Source: WorldBank (2003: Table 9B.8). Systemic shocks, such as the Tequila Crisis, are more difficult to cope with because mechanisms that work for idiosyncraticshocks fail at the systemic level. Thus, the response to the systemic shock epitomized by the 1995 Tequila Crisis, examined by McKenzie (2003), i s different from that o f idiosyncratic shocks indicated in the tables above. Most specifically, rural workers during the Tequila shock could not as a whole resort to increased labor market participation, because o f the depressed condition o f rural labor markets. Nor could they profit from domestic loans and donations from friends and relatives because o f the generalized impact o f the crisis on the household economies; only transfers from friends and relatives in the USA, 138 i.e. sheltered from the impact o f the crisis, increased, mitigating the shock to some extent. The situation was also differentwith respect to consumption smoothing. There was some consumption smoothing during the Tequila crisis, but less than that observed upon the idiosyncratic shocks reflected in the ENCEL and ENET panel data and those in Table 7.4. Thus, for families engaged in agriculture, the average fall in income between 1994 and 1996 was 17 percent, that in monetary consumption 13 percent, and that in non-monetary consumption also 17 percent. The main smoothing that took place was through the reallocation o f consumption expenditures in favor o f the most essential ones, especially food spending which fell only 5 percent (McKenzie, 2003: Table 5). Risk reduction, mitigation and coping mechanisms cannot always be separated, because some decisions by the rural poor affect simultaneously all of them. Also, decisions which have risk implications may not just be related to risk management but also to economic strategies linked to income and acc~mulation.~~ We analyze below some strategies typically followed by the rural poor that have a variety o f implications for the management o f income, investment and risk. They are: the diversification of income sources, the reliance on the subsistence economy as a safety net, and migration. We also examine some specific features o f riskmanagement inold age. Diversijication of IncomeSources Diversifyingincome through non-farmoccupations has been an importantway for Mexicanrural households to concurrently increase income and mitigate risk. We illustrate this with the diversification of rural incomes that took place from 1994 to 1996 and 1998 as a consequence o f the Tequila Crisis, already mentioned in chapter 2. Table 7.9. shows rural households reacting to the crisis between 1994 and 1996 by increasing their involvement innon- farm occupations, particularly low return ones which are easier to access. This was probably related to the difficulty o f increasing their participation in the labor market due to its depressed state, and to that o f increasing farm incomes which requires investment in inputs and labor. The situation was somewhat reversedbetween 1996 and 1998 when the worst impact o f the crisis was over. The shares o f independent farming and agricultural wage incomes increased, and that o f non-farm occupations decreased. 92 As Roumasset once put it, the risks with risk hypotheses is not that they explain too little but that they explain too much(Roumasset, 1976) 139 Table 7.9. Share of Different Sources of IncomeinTotal Income of Rural Households(percentage) 1994 1996 1998 TYPE OF INCOME INDEPENDENT FARMING All households 25.0 22.3 23.8 Extreme poor 27.7 26.7 23.1 AGRICULTURAL WAGE LABOR All households 14.6 12.8 15.0 Extreme poor 23.4 20.5 22.3 NON-FARM OCCUPATIONS All households 32.7 38.4 34.9 Extreme poor 18.7 28.2 26.7 of which low return activities All households 16.5 18.3 17.9 Extreme Door 11.6 17.7 16.1 Source: WB staff calculations basedon ENIGH 1994, 1996 and 1998. Taking the four years between 1992 and 1996,we have estimated from ENIGHdata that the number of individuals in rural areas belonging to what we have called "diversified income household^"^^ passed from 4.4 million (19 percent of the total) in 1992 to 6.5 million (26 percent o f the total) in 1996. While income diversification i s a structural trend inthe rural areas o f Mexico, there i s little doubt that it was accelerated by the Tequila Crisis. What conditions explain the diversification of income sources? We examined this in chapter 3 with respect to participation in non-farm employment, distinguishing between low returnandhighreturnoccupations. We saw there that gender, education, region and the larger or smaller size o f the rural settlement are all correlated to household involvement inthese two types o f non-farm activities. Using data from the ENHRUM survey, Taylor, Yunez-Naude and Ceron (2004) have studied the process whereby rural households decide to participate in different occupations and the incomes derived from them. We show in Table 7.10 the average income shares o f rural household from different sources resulting from this survey. The most important source i s non-agricultural wages (41%) followed by farm production (18.2%). Transfers account for 17.1 percent o f average rural household incomes, o f which two thirds from remittances from abroad. These figures confirm our previous finding o f large diversification o f sources o f income in rural areas. Taylor, Yunez-Naude and Ceron further examine how different household characteristics (size, education, experience), asset endowments (land, livestock, tractor, migration capital for domestic and international migrations), and region explain participation in different income sources. They then examine how these variables explain the income obtained from different sources, after controlling for household participation inthem. 93Households with less than 50 percent of their income coming from a single source. 140 Table 7.10: Share of IncomeSources of MexicanRuralHouseholds Accordingto the ENHRUM2002 Survey Activities Shares FarmProductionActivities 18.2 Livestock 3.7 Staples 2.4 Commercial Crops and Plantations 10.0 Other agricultural 2.1 LocalNon-Farm Activities 8.3 Commerce 6.1 Services 2.1 Handicrafts 0.1 RenewableResourceExtraction 2.3 PublicTransfers 4.4 Private Transfers 12.7 Domestic 1.7 FromU S A 11.0 Salaries andWages 54.1 Agricultural 13.0 Non-Agricultural 41.1 Total (Average Household Income inMxP = 53,456) 100.0 Sample Size: 1,782 households. Source: Taylor, Yunez-Naude and Ceron (2004). Results are interesting. Not surprisingly, household size is directly related to diversification.Larger households tend to participate more than smaller ones in most activities: traditional and modern agriculture, large and small livestock, the exploitation o f natural resources, non-agricultural occupations, and labor markets. They also receive more government transfers. Large households not only participate more in these activities but also obtain more income from them (after controlling for participation). Thus, for instance, an increase in one household member i s associated on average with 23 percent more income from agricultural wage employment and 34 percent more income from non-agricultural wage employment. Migration, instead, i s not related to household size in the regression equations, probably because o f the endogenous factors involved between these variables: larger families may have a higher propensity to migrate, but migration reduces the size o f the household. Human capital (family education, experience o f the householdhead) encourages participation inmodern agriculture and non agricultural activities, and discourages traditional agricultural, small livestock and participation in the agricultural labor market. Income from migration, both international and domestic, i s very much enhanced by the existence o f migrationnetworks or "migration capital"94; families with members resident in the USA have on average a 65 percent higher probability o f receiving remittances, and families receiving remittances have on average an income 1.4 times higher than those who do not. The corresponding figures for internalmigration are 32 percent for receiving transfers and 0.67 times for the increase in income. Land increases the probability o f participating in traditional and modern agriculture and in livestock, but the contribution in the margin o f landto income from these activities i s small. TheSubsistenceEconomy as a Safety Net The subsistence economy is commonly understood as the productionof food crops carried out by farmers in one or several small plots of land for self-consumption, using 94We will see later that not only migration income, i.e. migrant transfers, depends on migration networks, butmigration itself also depends on these networks. 141 family labor. Other components could also be added, like the raising of small animals, a vegetables and fruit orchard close to the homestead, and inmost cases access to communal lands for grazing animals and collecting fire wood, wild plants, construction materials and other products. Lands are commonly on fragile or little fertile rain-fed soils, often steep, which forces farmers to use rotations, with fallow periods that have been decreasing under demographic pressure. Maize and beans are normally produced in association, and often a horticultural crop like calabacita i s also included. Technology i s simple and low risk consisting o f well-tried criollo seeds usually resistant but low yielding, manual and animal labor (the latter if the family has animals or can rent the service), organic manure from own animals and domestic waste, and in some cases a small dose o f chemical fertilizer ifthe family could afford buyingit. Subsistence farming is not exclusive to poor farmers, and is part of a wider family productionsystem. Two things are important to understand about subsistence production. The first i s that it i s rarely practiced alone, but as part o f a wider income and occupation plan o f the family, i.e. o f its production system. Subsistence production may be combined with other agricultural activities, for instance with the production o f coffee or other cash crops, the commercial exploitation o f communal forest resources, participation in the local labor market or non-agricultural occupations or temporary migrations. The second important aspect i s that it i s not exclusive o f very poor farmers. Most farmers practice subsistence production to a smaller or larger extent. It coexists side by side with the commercial operations o f mediumor comparatively rich small farmers intricately woven into them. It i s only when the family stops participating directly inmanualwork that subsistence production i s abandonedby commercial farmers. The importanceof the subsistence economy has been decliningbut it still probably i s the number one safety net in rural areas and it will probablybe for many years. This i s why among our recommendations we include that of strong policy support to the subsistence economy. Its role in food security i s evident but there are also other more hidden roles which need to be highlighted. Thus, cultivating the soil and producing (some of) one's own food i s strongly associated with belongingto a community. Iti s part o f tradition and o f what local society expects o f an ejidatario or a comunero, and i s a sign o f being diligent and having some measure o f independence. Migration Out-rural migration is an important income, employment, and risk management option for rural householdsin MCxico. It has accelerated since the mid-90s as can be seen in Figure 7.2, where migration indexes calculated by Taylor et a1 (2004) on the basis of migration histories contained in the ENHRUM survey are presented. Acceleration since 1995 onwards, particularly o f migration to the USA, i s probably related to the Tequila Crisis and can be seen as one more mechanism used by rural households to cope with this economic shock. Migration to the USA i s growing more rapidly than internal migration, but this i s still much larger; approximately 70 percent o f ruralmigrants in2002 hadmigratedwithin Mexico, and only a small proportion o f them (less than 10percent) were employed in agriculture (Taylor et al, 2004).95 95 Large migration to the USA after the Tequila Crisis is also observed by Davis and Winters (2001) among ejidatarios. 142 Figure 7.2 Trends inRural Out-Migration, Internal and to the USA, 1980-2002 I Source: Taylor, Yirnez-Naude and Ceron (2004), calculated from ENHRUM. I t would be wrong to assume that most rural households have migrant members.We have survey information on this from three sources: the 1994 and 1997 ejido surveys, the 1997 ENCASEH survey, and the 2002 ENHRUM survey. Comparison o f the 1994 and 1997 ejido surveys indicates that 15.9 percent o f the 5,260 individuals in the sample migrated at least once in that period, with 5.2 percent going to the USA, and 10.7 percent to other locations in Mexico, o f which 1.4 percent to do agricultural work and 9.3 percent non-agricultural work. Of all migrants, 71 percent were male and 29 female, and o f migrants to the USA, 80 percent were male and 20 percent female (Davis, Stecklov and Winters, 2002: Table 1). In the ENCASEH 1997 survey, 10.6 percent o f the 2,574 rural households in the sample had members who had migrated temporarily during the last year (Gonzalez-Konig and Wondon, 2003: Table 1). Finally, in the 2002 ENHRUM survey, 16.2 percent o f the 1,782 families in the sample had one or more members living inthe USA and 25.8 percent of the families hadmembers living in other parts of Mexico at the time o fthe survey (January-February 2002). The figures above are not as large as might be expected given the known importance o f out-rural migration in Mexico, but there are two reasons why these figures could be somewhat deceptive. First, there may be, and often are, more than one migrantper household. Thus, when in the ENHRUM survey we measure all migrants against all rural families the result i s that at the beginning of 2002 there were on average 0.35 migrants in the USA and 0.71 migrants within Mexico per rural family (Taylor et al, 2004: Table 3). Second, the figures refer to members o f the nuclear family only. The situation changes if we include siblings and past migration experience. Thus, for instance, it has been worked out from the 1997 ejido survey that 44 percent of ejido families had some connection with the USA, either because the parents had migrated themselves inthe past or hadchildren or siblingslivinginthe USA (Davis andWinters, 2001:Table 1). There are different types of migration of the rural poor accordingto their duration, purpose, and role in the life cycle and accumulation strategy of the migrant. Ethnographic data from field studies (Edouard, L e Moing and Gonzalez, 2004) suggest various types o f migrationwith differentfunctions. Seasonalmigrations of short duration (one to six months) normally have an income diversification function and take place during the low part of the agricultural cycle. Short migrations are often part o f a regular pattern where continuing arrangements exist with an employer. This i s particularly the case in rural-to-rural migrations when the migrant recurrently 143 returns to the same farm or agricultural area to helpduringthe harvest. Recurrent short migrations involve different types o f workers but are typical o f middle age individuals, mostly males, and mostly small farmers, often from communities o f origin in the south to agricultural areas in the central and northern parts o f the country. There are three common routes. Along the Pacific Coast, migrants work seasonally in the harvest o f fruits and sugar cane, and year-round in vegetables. In North-Central Mexico migrant workers help produce key crops such as cotton, apples, and various vegetables (primary between August and January). Along the Gulf Coast, farm operators employ migrants to produce sugar cane, cotton, oranges and coffee, except during July-September. Short migrations can also serve as a coping mechanism upon an income shock, inwhich case they would not be recurrent, and we would expect them to involve individuals of different age and gender. The use o f this coping mechanism i s likely to depend on the depth o f the migration network o f the incumbent, for this determines the possibility o f finding work and shelter at the place o f destination. Individuals who do not have good migration contacts are likely to use other responses to income shocks. Resource accumulation i s generally the purpose of temporary migrations of much longer duration (one to four years). They are mostly rural-urban and many are to the USA. Their purpose is accumulation more than income diversification. Young rural workers are typical candidates for this type o f medium-termmigrations, for they need to accumulate in order to establecerse (get established), which mostly means buying or building a house and getting married. Migration often works in this case as a sort a passage ritual -a passport to the adult world. Young women do also participate inthese long temporary migrations, more probably than in short-term ones. Contact networks are of great importance in these migrations for both USA and domestic destinations and for both men and women but perhaps more for the latter. When migrations are to the USA, male networks are important for the decision o f women to migrate but female networks are more important to decide the destination (Davis and Winters, 2001). Middle age married individuals also participate in medium-termmigrations. In this case it i s mostly the husband who migrates, and the purpose could be either capital accumulation in order to buy property or start some line o f business, or to complement income probably to attend the needs o f a growing family when local opportunities are very scarce and there are good migration contacts. The situation inthese cases of the women left behindis ambiguous; on the one hand they take on new responsibilities as managers o f the household and o f family property in the absence o f the husband, hence becoming more autonomous, but on the other hand the continued absence o f the husbandgenerates many uncertainties and incrementalwork loads. LandAssets and Risk Management in OldAge Current inheritance patterns which tend to discriminate against younger generations are part of an old age risk management strategy. Attention has been called in recent years to the inheritance patterns in rural areas in Latin America with respect to the main rural asset, land, and their impact on agricultural productivity, youth exodus, and demographic imbalances (Abramovay et al, 1998, Dirven, 2002 and 2003). Mexico i s certainly not an exception to this, as indicated in Chapter 5. The resistance o f Mexican small land holders o f advanced age to pass on their lands to the young generation i s part o f a broader survival cum risk management strategy identified in field studies (Edouard, L e Moing and Gonzalez, 2004). Being males the main owners o f land in rural Mexico, the strategy refers mostly to them rather than to women, and i s briefly describe below. Keeping land ensures control of assets as well as bargaining power over other household members' income. The main perceived risks faced by old land-owning men, whose possibilities o f migration, participation in the labor market and venturing into non-agricultural 144 activities are diminishedby age, are (1) to be left alone without anyone to look after them, and (2) to be unable to earn enough income to survive. A response to this situation commonly observed inejidos and comunidades,which probably extends also to small private land owners, consists of three aspects. First, old owners cling to their land resisting any pressures to pass it on in life to their heirs. This allows them to keep control of an asset which (i) provides a flow o f income, (ii) embodies the main family wealth and hence helps keepingthe family together around the parents, (iii) bebequeathedtothewifeoranyofthesonsordaughtersathisdiscretion,thusgivinga can good bargaining power to the title holder, and (iv) gives power and status in the community.96 Old owners are often afraid that ifthey pass on the landto a son or daughter, these may sell it and leave, so that he would be left with nothing. This i s not an overly unreasonable assumption in view o f the breaking o f family ties observed in some communities. Second, parents try also to retain inthe household at least one o f the sons or daughters to look after them and helptilling the land, often with the promise that they will inherit it.97Finally, parents try to keep the family bound together even if physically separated by migration, and try to ensure that transfers keep coming in from distant children. Maintaining uncertainty with respect to the inheritance i s a way o f achieving this. Another way i s helping migrant sons and daughters to maintain ties with the community, by for instance keeping a house there, and also providing services such as looking after the grand children and after the children's property in the community while they are gone. This behavior from old land owners has important consequences on the use o f land, the farming practices, and the access to landby young workers. GovernmentProgramsAffecting Vulnerability We review here various ways in which the Mexican governmenthelps reducing the vulnerability of rural poor people and support their risk management strategies. Many government programs have some direct or indirect effect on the vulnerability o f the rural poor, but we will be only concerned here with programs that are either directly oriented to reduce vulnerability, such as agricultural insurance, or have (or could have) a major impact on it. We only deal with economic vulnerability, not covering issues and programs related to health and education. Agricultural Insurance Systems Most agricultural insurance in Mexico is oriented to middle and large commercial farmers. Crop and livestock insurance is not relevant for very poor farmers but can be important for sectors o f small farmers who are moderately poor but have the conditions to escape poverty through farming. There are three main ways in which government intervenes in agricultural insurance. The first i s through the 2"d level activities o f Agroasemex, the government's agricultural insurance company. The second i s through the support provided to the fondos de aseguramiento, a private cooperative-type risk pooling system funded and run by the farmers themselves. The third i s through Fonden, a natural disasters government fund. We examine here briefly these systems and their impact on the rural poor leaving some recommendations to the latter part o f this chapter. Agroasemex, which replacedthe former governmentinsurancecompany in 1990 has an overarching role as reinsurance company and insurance development agency. Unlike Anagsa, Agroasemex did not try to insure all loans from Banrural and concentrated mostly on 96Thus, for instance, only title holders can vote or hold office inthe ejidos. g7 The situation in the ejidos is complicated by the fact that, as explained in chapter 5, according to law, properties cannot be divided by inheritance or otherwise. Inpractice, however, they often are. 145 highpotential andmoderaterisk areas. Italso left room for private companies to participateinthe rural insurance market. Private companies at present cover some 57 percent of the 1.5 million hectare insured area and more than 90 percent o f livestock insurance. Agroasemex main clients were small commercial and middle producers with productive potential. In 2001 Agroasemex withdrew from the retailinsurance business andbecame a 2"d level institution with two functions: reinsurance company, and insurance development agency. As a development agency, Agroasemex promotes new instrumentsand links between the finance and insurance markets, and carries out training. It also has two important development roles: to administer the Programa de Subsidio a la Prima del Seguro Agropecuario, a subsidy program to the agricultural insurance premium9*,and to provide technical and organizational support to thefondos de aseguramiento, mainly through promotion and technical assistance services. There are no specific activities of Agroasemex targeted to small farmers, but those among them who have access to insurance benefit from the subsidy on the premium. Also, the support given to the fondos benefits small farmers participating inthem. Thefondos de aseguramiento have been fairly successful, concentratingon weather, biological and crop establishment risks. These fondos are farmers insurance associations, originally started inthe early 90s inthe prosperous irrigated ejidos o f the Mayo and Yaqui valleys inthe state ofSonora. Although hardly hitby the financial crisis of 1995, they have been rather successful and expanded fast among commercially oriented farmers o f small or middle size, mostly in irrigated and good rain-fed areas. They specialize in crop insurance, having gone little into livestock insurance, which i s the province o f the private sector. Fondos currently insure some 622 thousand hectare (42 percent o f the insured area) and receive some USD 40 million in premiums. Premiums are collected from participants, and insurance is paid upon assessed damages. They basically insure against weather, biological, and crop establishment risks. There are different types o f insurance with different coverage: the investmentcost o f the input package, the value o f the plantation (inthe case o f plantation crops), and the value o f the expected crop. On average,fondos have some 350 memberswith an insuredarea o f some 3,000 hectare each, and an average premium o f close to USD 60 per hectare, o f which 30 percent i s paid by Agroasemex subsidy. To deal with covariate risks, fondos reinsure with Agroasemex, which has a program to that effect. As mentioned before, fondos also receive support from Agroasemex for training, technical assistance and computing equipment. The Fondo de Desastres Naturales (Fonden)is a federal government insurancefund against natural disasters,which i s part o fthe SistemaNacional de Proteccibn Civil. It covers all major natural disasters, financing the reconstruction o f public infrastructure and compensatingthe ruralpoor for their losses following large covariate shocks. Small farmers and other ruralpoor are protected in four ways: (a) they receive support to rebuildtheir houses if affected by the disaster, (b) receive compensation for crop and livestock losses for a maximum of 5 hectare and 25 heads o f cattle at a rate o f some USD 33 per hectare and USD23 per head, (c) they may also qualify for temporary income and employment support, and (d) they benefit from the reconstruction of local public infrastructure (Secretaria de Gobernacion, 2003). Fonden i s a useful instrumentto absorb part o f the income impact o f large covariate shocks o f natural origin, but it compensates for part only o f the losses and depends on a number o f procedures and discretionary actions such as the declaration o f emergency that limit its impact. Although useful to the affected individuals, it i s 98 The subsidy amounted to some US$ 31 million in2003 distributed over some 1.5 million hectare and 16 million animals, approximately 60% for agriculture and 40% for livestock. The subsidy was equivalent to 30% of the premium and 3.5% of the insured value (9.2% in agriculture and 1.7% in livestock), and an estimate of 2.8% of production costs inagriculture or MxP 140per hectare (Instituto Superior Tecnologico de Monterrey, 2004). 146 probably a better instrument to compensate for the loss o f infrastructure and facilitate its reconstruction. Other Government SupportPrograms With the exception of work fare programs, there are no government programs specifically focused on mitigating the impact of risks. Cash transfers programs such as Oportunidades and Procampo are important at ensuring some minimum income and hence help smoothing consumption in the presence o f income shocks. They are not, however, insurance mechanisms as such. Using ENCEL panel data Skoufias (2004) has shown that Oportunidades does not provide additional insurance for consumption over and above existing formal and informal insurance systems although it helps smoothing income over time. On the other hand, he also found that cash transfers increase the means available to households to cope with shocks rather than displacing existing informal means of insurance. There i s no similar panel data with respect to the impact o fProcampo but one would expect the situation to be similar. Other rural development programs such as Alianzapara el Campo,Microrregiones and Opciones Productivas do not have a specific social protectionfunction against income shocks, although the credit programs included in Opciones Productivas may facilitate some coping strategies following idiosyncratic shocks. Being mostly a social infrastructure program, Microrregiones is not expected to have a major short time effect on income vulnerability, although it could have medium and long-term effects if the infrastructure built served to attract other investments and promote local development thus generating more income earning opportunities. Something similar could be said with respect to the support offered by Alianza to farm modernization. The Programa de Empleo Temporal (PET) promotes employment of poor people, particularly the extreme poor, in public works in rural areas. This program is mostly the responsibility o f SEDESOL but has also some funding from SAGARPA and SEMARNAT. Although its primary function i s employment provision to the poor duringthe low season, it does have an insurance function since incremental funds are usually made available for additional employment in areas that have been affected by systemic shocks due to natural disasters or other causes. The program i s oriented to the extreme poor and consists o f the provision o f employment in labor intensive public works to build infrastructure or in works related to environmental or sustainable agricultural improvements. PET i s particularly oriented to poor municipalities prioritized in agreement between the federal and state governments. Beneficiaries receive 90 percent o f the minimum wage for unskilled workers in the area and can only work for a maximum o f 88 days. Under the ordinary program, works are carried out during the low agricultural season. PET successfully combines a poverty alleviation function with a shock mitigating function. PET serves two purposes: absorbing seasonal unemployment in rural areas affecting especially the very poor, thus alleviating poverty, and providing especial assistance in situations o f emergency. These two functions are entirely compatible and even synergic, because the experience and capability gained in the implementation of the regular program are very valuable inemergencies situations. Itis hence appropriate for PET to be both an employment and a social protection program. 147 POLICYIMPLICATIONSAND OPTIONS Two Viewsof Poverty and Vulnerability in RuralAreas T o close the chapter we contrast two views on poverty and vulnerability. Much o f the current policy debate can be interpreted as product o f the tension between two contrasting views o f poverty and vulnerability inrural areas. These models represent differentviews on poverty, the poor, poverty measurement, poverty correlates, anti-poverty instruments, poverty targeting, migration, social protection and actions to promote rural development. For reasons o f simplicity we label these views "social-liberal" and "rural-corporate" and compact them under Model A ("social-liberal") and Model B ("rural corporate") inTable 7.11. Model A stresses the importance of individual agents, private assets and the market as engines of rural development, but recognizesthe need for governmentintervention in the supply of public goods and basic services inunderservedrural areas, and insupporting with cash transfers the incomes o f rural families fallen into the worst poverty situations. Education, health, social infrastructure and direct cash transfers targetedto the poor are the principal, albeit not only, items in the policy agenda. These items are market compatible and also tend to increase competitiveness by increasing human capital. Education and health in particular result in individual improvements in knowledge and strength. These are portable assets o f varied application, which increase labor productivity and hence returns to labor. If the market makes local opportunities scarcely attractive, especially to the young, migrationi s the answer. Migration under Model A is seen as a Pareto improving individual decision that people freely make, rather than as a social process explained by general economic circumstances. People are viewed as individuals rather than structured groups. Vulnerability i s something, therefore, more related to individual situations than to that o f groups. Self-insurance through the promotion o f appropriate financial assets i s a fitting answer. The rural poor are seen as an undifferentiated collection o f individuals or families with similar needs who have incomes below certain levels. Since measuring the income o f each individual or family i s not possible, poverty targeting should focus on the observable characteristics that make the individual or family more or less likely to fall under the specifiedlevels. Model B focuses on rural organizations, especially rural communities, and collective action by these organizations. The need o f state intervention i s recognized under this model with similar functions to those inModel A (education, health, social infrastructure, direct poverty alleviation), but also with emphasis on the promotion o f rural economic development through support to productive activities. Migration i s seen as an unavoidable process in many rural areas, but one that imposes hardships on the individuals and families concerned as well as socioeconomic disequilibria in the communities. The model i s optimistic as to the possibility o f reducing migration through the promotion o f local development, and proposes working towards situations where people can have a better choice between migrating and taking up local employment opportunities. Population i s seen not as an agglomeration o f individuals but as a structured set o f organizations -territorial, interest-based and others- embodying valuable social capital for development. Participation o f these organizations in the targeting and operation o f development programs, and unconditioned outside support with freedom for them to decide on investments and development paths, are strongly emphasized. Poverty situations are seen as heterogeneous and hence requiring differentiated actions. Economic development policies coupled with people's participation and reliance on rural organizations are the policy recipes o f this view. The alleviation of extreme poverty situations through direct cash transfers is seen as a positive but transitory and partial measure, secondary in importance to production oriented policies. 148 Table 7.11. Visions on PovertyandVulnerability inRuralAreas: Two ContrastingModels VIEWS ON ModelA Model B "Social-Liberal" "Rural-Corporate" Poverty Emphasis on objective and absolute Emphasis on subjective and relative dimensions dimensions The poor Emphasis on characteristics and Emphasis on differences among the needs that are similar among the poor poor Poverty measurement Emphasis on monetary measures, Emphasis on basic needs, human particularly income-poverty development indicators and stock measures Emphasis on observable variables: Emphasis on both observable variables Poverty correlates assets, education, location, gender, and little observable ones like social family characteristics, occupation, capital and cultural dimensions ethnicity Emphasis on investment in"mobile Emphasis on productive development, Anti-poverty instruments assets" (mostly human capital) and CDD andterritorial approaches direct cash transfer Emphasis on objective, centrally Emphasis on self- and community- Poverty targeting established criteria targeting, importance o f context variables and community perceptions Emphasis on the virtues o fmigration Emphasis on migration as a "necessary Migration as an equilibrating economic evil" and on the suffering and social mechanism and an opportunity to disequilibria generated by it escape poverty Social protection Emphasis on market mechanisms and Emphasis on social capital cash transfers Actions to promote rural Emphasis on the individual, private Emphasis on the community and on development assets and markets collective action Source: Author's construction. A measured policy response to rural development challenges should take into account the views from both models. Each model has its strengths and weaknesses. Model A rightly emphasizes the importance of human capital and the need to protect the freedom of individuals to migrate and to place them in the best position to be successkl migrants. Its understanding, however, o f the migrationprocess i s unsophisticated, too narrowly focused on the individual and on existing conditions, without proper attention to the social dimensions and the circumstances generatingmigrationinthe first place. The insistence o f Model B on active policies to stimulate rural pro-poor economic growth i s sound; income supports through direct transfers cannot substitute for them, regardless o f their possible income multipliers. Direct income supports, however, are important because o f their immediate welfare effects, and cannot be dismissed, which is a risk under Model B. By recognizing the importance o f rural organizations, social capital, and the heterogeneity o f the poor, Model B takes a more "organic" view of rural society than Model A, which i s a strong asset for the formulation and implementation o f rural policies. Insisting on local participation and the need to support communal and other organizations while allowing them a free hand to decide on the use o f resources, i s another strong point o f Model B, but limits and risks need to be recognized. Capture o f benefits by local elites i s a well-known risk.Also, central authorities and other supporting partners are entitledto their own views and priorities with respect to development programs, and hence to discuss and agree priorities and implementation guidelineswith local organizations. They are also entitledto a close watch on the soundness o f local processes. These considerations serve as a background for the analysis o f policy options below. 149 Policy Options Formal agricultural insurance schemes are not particularly useful to the extreme rural poor since for many of them income does not come mainlyfrom independentfarming, and when it does agricultural insurance can be comparatively expensive for small rain-fed producers growing traditional crops. Also, these producers use low risk technologies in their farming operations, hardly use seasonal credit, are not exposed to input or output market risks because they buy very few inputs and sell little if any output, and to a certain extent are covered from large natural disasters that trigger the intervention o f Fonden. For these reasons other type o f insurance systems are more appropriate for this bottom sector o f the ruralpoor. Agricultural insurance, however, can be importantto farmers in transition, located between the extreme and moderate poverty lines or immediately above the latter line, especially if they want to diversify into high value crops, which i s always a risky operation. Agroasemex could make additional efforts to assist creating more insurance groups among this sector. This could best be done in the context o f projects promoting crop diversification or technological change among small farmers in a certain area with a modicum o f potential for such changes, even if at a small scale, preferably as part o f a larger territorial development effort. An interestingproposal has been made to introduceparametric insurance systems in rural areas linked to weather parameters (World Bank, 2001, Skees et al, 2002). The system would be useful to farmers in general, including small farmers. Even the non-farm rural poor couldbenefit since many rural enterprises and labor opportunities on which they depend are relatedto weather. To be most effective, the parametric insurance system should work in synergy with the development of rural finance. Micro finance institutions could hold weather derivatives as part o f their portfolio to protect themselves against defaults originated in covariate weather shocks. They could also make loans available to the rural poor to buy these bonds. Weather derivatives would also help taking part o f the covariate risk off insurance groups, and hence facilitate the formation and operation o f risk pool groups such as thefondos de aseguramiento. The system has the advantages inter alia of being simple, easy to administer and free from the adverse selection and moral hazard problems that affect most insurance schemes. Parametric insurance criteria could also be applied to public insurance programs. Thus, Fonden could consider moving to a parametric system o f triggers (World Bank, 2001). Modifyingexisting cash transfer programs in rural areas to enhance their insurance function is not seen as a favorable option. These programs already fulfill certain social protection functions in addition to their own objectives and it would be dangerous to ask them to performtoo many functions. Iti s better to use other instruments. Rural financial systems are multipurpose instruments that serve for risk management, capitalaccumulation, technology adoption, and personalwelfare. The possibly are the single most important formal system to assist the rural poor to manage risks, particularly idiosyncratic risks. Their main roles are as follows: (1) they promote savings and facilitate keeping them in forms more adequate to needs than those traditionally used by the poor (cash hoardings, animals, produce, land), thus favoring the use of savings as a self insurance mechanism; (2) they facilitate obtaining personal loans in response to shocks; (3) they also facilitate obtaining crop insurance, which i s usually bought together with seasonal crop loans; (4) rural financial intermediaries may directly provide certain types o f insurance, such as life or burial insurance; and (5) personal and productive loans facilitate income diversification and migration strategies, which are important self-insurance instruments. 150 Given the importance of subsistence production as a safety net to poor rural producers, strong support to the subsistence economy is a major policy option. The subsistence economy, like the poor, i s there to stay with us for many years, although (hopefully) with a waning away trend over the very long run.As argued in chapter 1, this is a matter of the inability o f the economic system to provide high productivity jobs to surplus laborers who take refuge in subsistence production, many in the zonas de refugio. The same logic that leads to adopting a social policy o f cash transfers to the extreme poor -an underprivilegedstructural part o f society whom we recognize society must help- applies to supporting the subsistence economy. Only that in supporting subsistence production we are not just doing "social" policy; we are also doing production and environmentalpolicy. There are various ways in which government programs could support subsistence production. First, soil management programs in fragile areas would help redressing one of the main problems in subsistence agriculture: the degradation o f soils because o f erosion and falling fertility due to shorter fallows, use o f steep hill sides for annual cropping, and lack o f conservation practices. Second, programs to enhance kitchen gardens, the production o f small animals, and some tree crops could be expanded. Third, environment friendly yield increasing technical packages could be promoted for traditional crops. Finally, access to communal resources could be improved and better managed through the promotion o f community self- regulation norms that would avoid tragedy o f the commons type situations following migration, market development and income differentiation. Communal resources could also be the basis of accessing environmental service markets in the benefit o f the community. Support to the subsistence economy i s probably better carried out at the local level in the framework o f municipal or micro-regional plans such as those promoted by Microcuencas. The federal government could have an important role, however, in developing technological and methodological options, disseminating best practices, coordinating efforts and providing financial resources. 151 8. SUMMARY OF POLICY OPTIONS TO FIGHT RURAL POVERTY This chapter presents policy options to intensifi thefight against poverty in rural areas, building on the Mexican government's policy reforms underway in areas like local development, program implementation, and education. We bring together in this chapter the options discussed in the previous ones. Policy options for improving rural poverty interventions are organized around four areas: 1. Deepeningthe territorial approach to rural development as a form to achieve local economic development and reduce poverty through territorially based economic coordination rather than purely sectoral approaches. 2. Revitalizing the rural economy in favor ofthe ruralpoor, raising overall productivity inthe farm and non-farm sectors, and working to help the poor increase their labor productivity. 3. Improving the design and effectiveness of rural development policies and programs; and 4. Supportingthe rural youth as a critical factor for a dynamic rural economy. GENERALCONSIDERATIONS This report has shownthat strong policy actionwill be neededto overcome poverty in Mexico's rural areas. Rural Mexico is marked by poverty. Poor families inrural (and urban) areas are still reeling from the effects o f the macroeconomic crisis in the mid-1990s. Extreme poverty exists in all o f rural Mexico, but the southern pacific states, far away from economic centers and with large indigenous populations, have particularly high incidences o f poverty. While the situation has improved substantially since 1996, poverty incidence i s still very high in rural areas. The effects o f the Tequila Crisis and a slow-moving agricultural sector were instrumental factors in eroding rural households' income sources. The Mexican government has been successkl in increasing direct cash transfers to the rural poor, but efforts should also be strengthenedto enhance access o f the ruralpoor to non-farm activities as well as more productive agricultural activities and farming methods. This chapter presents some options for reform to deepen efforts to eradicate rural poverty. The discussion is organized around four areas: (1)deepening the territorial approach to rural development; (2) revitalizing the rural economy in favor o f the rural poor; (3) improving the design and effectiveness o f rural development policies and programs; and (4) supporting the ruralyouth as a critical sector for a dynamic rural economy. 152 These specific policy options, discussedbelow, should be seen in the context of more general policy implications derived from this report.Inparticular, we would like to highlight the following issues: Maintaining macroeconomic stability as an essential element for a poverty reduction strategy Keeping the level o f direct transfers to the rural poor in view o f their current importance for their livelihood, but using incremental resources at the margin to promote income and employment growth. Focussing attention on regions and areas where poverty i s more concentrated. Concentrating investments to build up critical masses to trigger endogenous growth process. Continuing focusing attention on improving rural education, with emphasis on quality, expanding secondary facilities and enrolment, and strengthening technical education and vocational training for rural people related to farm and non-farm activities. Mexico's success at regaining and maintaining macroeconomic stability after the Tequila Crisis has beenimportantfor allowing a sustained fall in poverty since 1996. As we have shown, poverty increased dramatically in 1994-1996, under the effects o f the macroeconomic crisis. Mexico's ability to maintain macroeconomic stability over the past decade i s probably one o f the most important explanations for why poverty has fallen. Prudent macroeconomic policies will need to remain at the core o f any ruralpoverty reduction strategy. Public transfers are now instrumentalto the livelihood of the poor but need to be complementedby enhanced income and employment programs. Public transfers to the poor, especially through Oportunidades and Procampo, have increased substantially since the mid 1990s. It i s likely that the number o f poor and the depth o f poverty would be starker had these efforts not been made. Clearly, these transfers are essential to the poor, and couldnot be removed at this stage without much suffering. We believe, however, that this safety net needs to be complemented by more efficient and effective ways o f promoting income and employment growth, and that at the present time incremental resources for rural areas would be better usedto promote income and employment generationthan to expand direct cash transfers. Locationmatters. There is a need to focus attention on marginalized regions and areas, where poverty i s more concentrated, combining incremental resources with an assessment o f the capacity o f specific programs to promote local development, and using targeting mechanisms adequate to each type o f program. Related to this, investments could be clustered to build up critical masses o f productive and support infrastructure to trigger endogenous growth process, rather than disseminated in large areas and many different activities or in a myriad o f small unconnectedprojects. Education remains the most important correlate of poverty. Education is an important tenet o f the Mexican government's social welfare strategy. Continued emphasis needs to be given to access, but also to quality, expanding secondary facilities and enrolment, and strengthening technical education and vocational training for rural people related to farm and non-farm activities. 153 SPECIFIC OPTIONS TO FIGHT RURALPOVERTY Deepening the TerritorialApproach to Rural Development The territorial approach to rural developmentis a means to achieve local economic development through territorially based economic coordination. The main tenets o f this approach, which i s presented inAnnex 3.1, are the emphasis on (1) mutisectoral development, (2) the links between rural and urban areas, (3) the use o f participatory territorial planning as instrumentfor economic coordination and to organize the demand for development interventions, (4) the structuring o f interventions around a long-term strategic plan for the territory, (5) the economic potential o f territorial assets, and (6) the mobilizing capacity o f a shared territorial identity (see Annex 3.H). The territorial approach changes the focus from sectoral to territorial competitiveness, and offers an excellent framework for poverty reduction interventions in rural areas centered on equitable local economic development. A territorial approachis probably best focused at the levelof territories like micro- regions and districts, which are larger than municipalities. The approach can be applied at different spatial levels, including for instance the microcuenca and the munic@io, but larger territories o f the size o f present SAGARPA districts and SEDESOL micro-regions are the most promising for strong development impact. These larger territories can be articulated with broader rural-urban regions sharing integrated labor pool watersheds. In this connection, a spatial policy would be useful to facilitate the geographical concentration o f investments and services for productive development, favor the growth o f rural towns and intermediate cities, and promote the establishment o f linksbetweenthese urban centers andtheir rural hinterlands. For the territorial approach to work in practice it would be usefulto have a policy and planning agency of mixed public-private-civil society composition in each territory, empowered to carry out participatory planning and articulate local development demands. The Ley de Desarrollo Rural Sustentable provide the instrument to create these agencies in the form o f Consejos de Desarrollo Rural Sustentable. District level consejos could be particularly important to carry out economic coordination intheir territories. Economic diversification and the commercial exploitation of territorial assets are key aspects of the territorial approach. There are three basic modalities: (1) increased value added o f goods or services already produced in the territory, (2) improved use o f territorial advantages through the introduction of new commercial products, and (3) the establishment o f synergies between different sectors o f activity in the territory. The European Union Leader Program, based on a territorial approach similar to that summarized above, offers many examples o f these modalities (Box 8.1). While there are severalprograms and mechanismsin placewith a territorial focus, the diverseinitiatives could be brought together under a unified policy framework. The Ley de Desarrollo Rural Sustentable favors a territorial approach to rural development, and there are rural development programs in operation designed with a territorial approach, such as Microrregiones, the Desarrollo Rural sub-program o f Alianza or Microcuencas. SAGARPA has been active in establishing rural development councils at the municipal level. More actions at the federal and regional level, consistent with the Law, could be carried out to advance along this path. A politica de estado to promote rural development with a territorial approach would be required, supported by the Oficina de la Presidencia de la Republica and the Secretaria de Hacienda. Dissemination o f the advantages o f a territorial approach among state governments would also be essential because territorial policy i s -or should be- their responsibility. 154 Four policy actions are possible at the federal level. First would be to empower the Consejo Mexican0para el Desarrollo Rural Sustentable (CMDRS) to act as national agency for the promotion of the territorial approach to rural development. The Consejo could establish a plan for the progressive introduction over time of the territorial approach and monitor its progress. Second possible action refers to the harmonization o f the operational rules o f rural development programs o f federal agencies, and the national coordination of their activities, which could be done within the framework offered by the CMDRS and with the support of the Secretaria de Hacienda. Third action would be the progressive introduction o f a system to coordinate the budgeting of rural development programs by federal agencies. Finally, a national territorial development fund could be created to finance territorial investment programs submitted by territorial agencies operating under competitive biddingprinciples. Box 8.1. The EuropeanLEADER Program The LEADER program o f the European Union first launched in 1991 is an attempt at fostering a territorial approach to rural development in Europe. Based on the identification o f problems and potentials o f specific rural areas, strategic plans are formulated and priority investments co-financed. There are three basic principles: innovation, partnerships, and multisectoral integration. The central characteristics o f LEADER can be summarized as follows: Each LEADER program intervenes in a territory relatively homogenous, defined by the localactors, ofbetween 10 and 100 thousandresidents. 0 A vertical partnership among three different levels of government (European, national and regional) establishes the general orientations, some general rules for the selection o f subprojects, the financial contribution o f each partner, an the M&E procedures. 0 A horizontal partnership, known as Local Action Group (LAG), is established as a civil association integrating local public and private actors: municipalities, producer associations, chambers o f commerce, savings and loans associations, unions, cooperatives, traders and entrepreneurs, NGOs and other CSOs. With the assistance o f a very small technical group, the LAG prepares a strategic plan for the development o f the territory, formulates an operational program for the 5 or 6 years o f operation o f LEADER financing, and receives, examines, approves and supervises the execution of the subprojects submitted for co-financing. Within the general LEADER norms established by the vertical partnership, each LAG has plenty room to decide investment priorities. 0 Integrated, innovative and multisectoral investments are generally favored. They are mostly oriented at exploiting the comparative advantages and assets o f the territory, the strengthening of local clusters of medium and small enterprises, and the exploitation o f synergies among different local economic sectors, as means o f increasing the competitiveness o fthe territory. Since farm production is supported by other instruments o f the European CAP, LEADER support concentrates on off-farm investments (including agro-processing). LEADER financing typically supports marketing innovations, environmental improvements, adding value to local products, small local industrial and services firms, and agro-tourism. 0 Sub-projects are co-financed with European and national hnds and beneficiary contributions. The latter varies according to the norms established by the vertical partnerships and the specific LAGbut is typically o fthe order o f 50 to 60 percent. 0 LAGSreceive some small technical assistance, particularly at the beginning, are supported in their accounting and other ways by participating municipalities, are organized in networks to share experiences through the national and European 155 "Leader Observatories", often cooperate with each other invarious ways, and usually intervene in discussions and policy decisions related to local economic issues. Source: FAO-World Bank, 2003. At the regional level, five possible actions could be considered. First, establishing shared regionalizaciones in each o f the states, agreed upon by federal agencies and state governments, who could take the leadership in this. Second, advancing in the creation o f territorial economic coordination agencies. The Consejos Distritales de Desarrollo Rural Sustentable regulated by the Ley de Desarrollo Rural Sustentable could be the legal form for these agencies. To ensure local ownership, their formation would best be approached as a bottom up process involving the active participation of all relevant federal and state agencies and territorial actors. Third, as territorial councils are being formed, the Centros de Apoyo a1 Desarrollo Rural Sustentable mentioned in the law (art. 29) could be created to act as technical secretariats o f these councils. Fourth, setting up a system o f ventanillas flnicas inthe territories to provide information and to process demands related to the various rural programs. Finally, formulating strategic plans for the territories usingparticipatory planning methods to identifythe development axes aroundwhich territorial investmentscould cluster. This task couldbe promoted bythe territorial agencies. RevitalizingtheRural Economy in Favor of thePoor Rural development needs to build on a comprehensive approach including both farm and non-farm activities.It is suggested that the focus move from the farm to the family as the unit o f analysis for rural development and the receiver o f rural policy. This implies abandoning the concept of "viable farm'' as a guide for policy interventions in agricultural development, and promoting agricultural competitiveness through multiple interventions embracing small and large farms, full- andpart-time farmers. Agricultural intensification offers a way forward to increasing output and incomes. The exhaustion of the land frontier and the comparatively low levels of land and labor productivity inagriculture suggest that intensification i s needed. We advocate a policy o f "diffuse intensification", embracing all types o f farms, irrigated and rain-fed, small and large. Policies are needed to create the conditions enabling farmers to carry out crop diversification and increasing crop yield. Inparticular, poorly endowed farmers with little physical and human capital will need help to move from low to high value crops. Extensive and well functioning research and extension systems, as well as rural finance systems, are essential for this. These systems are also essential to raise crop yields, the other component o f intensification. The small farm economy is the logical center of a poverty-friendly agricultural growth strategy. Focusing on small farmers i s a recognition o f the fact that markets tend to fail more for them, that instrumentslike research and extension and rural finance need to be specially calibrated for this farming sector, and that there i s a merit want quality in the needs o f small farmers. A positive discrimination spirit from policy designers and program operators would favor overcoming the class and technical biases in favor o f large farmers that usually characterize agricultural programs inMexico as elsewhere. Encouraging participation o f small farmers inrural value chains and their governance organizations and in asociaciones integradoras could also be part o fthese policies. In order to improve the poverty friendliness of rural programs, it would be usefulto intensify the dialogue betweenprogram managers and governmentofficials, on the one hand, and leaders from farmers' and other rural organizations, on the other. Intermediate rural leaders 156 have an especially important role to play in highlighting rural needs and priorities in the geographic or sub-sectoral areas o f their competence, and giving feedback on rural programs. Rural organizations have also an important role to play in the implementation o f rural development programs. Education -access and quality- as well as access to better infrastructure are important means of increasing access to non-farm activities and better wage opportunities for the rural poor. Policy should pay attention to variables strongly correlated with better rural wages and enhanced participation in rural non-farm activities. Education -in particular at secondary level- i s one o f them. But education coverage and incentives to attend school are not enough; quality i s also important, as well as combining education with other productive assets. Road connections, communications and energy are other correlates o f better rural wages and enhanced RNF employment. Economic investmentsto promote the RNF sector are integral to rural development- cum poverty alleviation programs. Investment decisions in the RNF sector could be decentralized through a system o f local participatory planning based on the territorial approach. This would be the best way to address the inherent heterogeneity o f the RNF sector and its different opportunities, while exploiting investment synergies and making investment programs easier to implementand more relevant to the specific context. Technology Poorer farmers need enhanced access to agricultural research and extension services to increase land and labor productivity. Rural technology enhancement has many stakeholders. To determineexactly what to do, it would be useful to organize an intensive debate among interestedparties on how the agricultural knowledge system should evolve inthe medium- and long-run to ensure competitiveness with equity. This could lead to the formulation o f a national strategy to enhance rural technology. Of particular importance i s how to accommodate poor farmers needs in the research and extension agendas and methods, and how to mainstream environmental concerns into research and extension. There is need to discuss the technical assistance and extension models most appropriate for different type of farmers and regions. Together with the use o f more traditional methods such as demonstration plots, that o f more modern ones already tried in localized pilot experiences could be mainstreamed, including methods such as farmer to farmer extension and farmer field schools for pest control, soil management, conservation agriculture, and agro-forestry systems. These models are probably best established at the state level through discussion o f local scientific institutions, farmers' organizations, federal and state authorities, and NGOsworking inthe area. SAGARPA and state authorities couldjointly promote these fora. The agricultural knowledge system could benefit from a modernization of the curricula of agricultural learning centers, particularly those o f the technical agricultural schools, like the CBTAs. The new curricula could pay more attention to new crops, processing industries, and commercial and managerial aspects. The promotion of vocational training and technical-level degrees in non-agricultural areas would also be important for local development. It could be achieved through a larger deployment in rural areas of CETIS and CBTIS type centers. 157 EnvironmentalAssets The commercial exploitation of environmental assets by local communities could bring dynamism to the rural economy. Cultural and landscape assets and biodiversity offer good opportunities for rural tourism and eco-tourism enterprises. Environmental services like carbon sequestration, water regulation and watershed management, and biodiversity stewardship offer also good economic opportunities, which have started being exploited in Mexico often with NGO support. Areas rich in environmental assets are often income poor, like the Southern Mexican states, and hence support for the sustainable commercial exploitation o f these assets i s usually poverty friendly. Forestry resources are a case in point. Although community forestry is rapidly and successfully developing in Mexico, ejido and community forests are still insufficiently exploited. Intensification o f community forestry programs i s an option. Their scope could be broadened to include the sustainable exploitation o f timber and non-timber resources, watershed protection and management, reforestation, agro-forestry systems, commercial forestry plantations, and timber and wood processingindustries. Rural Finance An accessible and viable rural finance system is essential to help the rural poor build assets over the long term and cope with economic shocks in the short term. Rural financial systems are a multipurpose instrument that serves for risk management as much as for capital accumulation, technology adoption and personal welfare. Three policy actions could be considered. First, expanding BANSEFI and Financiera Rural operations with the resources currently used for ad hoc credit programs so as to mainstream and unify rural finance policy. Second, using the Financiera Rural to assist in the development o f the rural micro-finance system. Third, supporting withjudicious incentives the introduction o f innovations inrural micro- finance operations, such as flexibility in disbursement and repayment to adjust to the agricultural cycle, flexible collateral requirements, use o f smart and debit cards and information technology, use of existing rural infrastructures for loan delivery, and introducing differentfinancial products. There is a large collection of formal and informal institutions and programs providingfinancialservices in many rural areas of the country. Stronger coordination among these institutions would facilitate making information available to users on existing financial products from different sources, thus making the market more transparent. It would also help reaching agreements among suppliers to improve the distribution o f financial services in the territory, and to combine financial services to better serve clients. This could best be done at the state level, where fora o f the relevant institutions could be stimulated to discuss local needs o f financial services and provide a framework for economic coordination. BANSEFI and the Financiera Rural could take an active role in the promotion of these state fora, but strong leadership from state governments would also be useful. Vulnerability Policy can help building better insurance and coping mechanisms against natural shocks. Natural shocks are a major source of vulnerability inrural areas. Policy interventions in technology, changes in crops and crop varieties, and sanitary measures could help reducing the incidence and income impact o f these shocks. Formal agricultural insurance systems are o f little help for the bottom sector of the rural population, but could be useful for farmers in transition, especially if they want to diversify into high value crops. Insurance support to this sector would best be carried out in the context o f projects promoting crop diversification or technological 158 change. The introduction o f parametric insurance systems would increase the insurance options for the ruralpoor, especially ifit goes together with the development o f rural finance. Fonden and PET are useful mechanisms for reducing the vulnerability of the rural poor, especially vis-&vis covariate shocks. Modifying existing cash transfer programs in rural areas to enhance their insurance function i s not considered a good option, however. These programs already fulfill certain social protection functions in addition to their own objectives and itwouldbe dangerous to askthemto performtoo many functions. Rural financial systems are probably the single most important means to assist the rural poor to manage risks, particularly idiosyncratic ones. Financial systems can facilitate savings, personal loans, agricultural insurance, and productive loans that permit income diversification and migration. Reducing vulnerability i s hence another reason for recommending the enhancement of rural financial systems accessible to the poor. TheSubsistenceEconomy Strong support to the subsistence economy is another major policy option inview of its importance as a safety net for poor producers. Supporting the subsistence economy is not only a "social" policy; it also favours productive development and environmental improvement. Support to the subsistence economy can take the form o f soil management programs in fragile areas, promotion/enhancement o f small orchards, small animals and some tree crops, promotion o f environmentally friendly, yield increasing technical packages for traditional crops, and more intensive use o f communal resources subject to better environmental management. Support to the subsistence economy i s probably better carried out at the local level in the framework o f municipal or micro-regional plans like those promoted by the Microcuencas program. Improving theDesign and Effectiveness of Rural DevelopmentPolicies and Programs Federal expending on rural development is high and a true mark of the importance traditionally given by Mexican governments to rural areas. The challenge i s to increase the efficiency o f the use o f these substantial fiscal resources. Program implementation i s an important area where efficiency and effectiveness could be enhanced. Quality o f implementation i s better examined at "the point o f service delivery", where program operators and beneficiariesklients intersect. As indicated in chapter 6, good implementation critically depends on bottom-level operators -the "street-level bureaucrats". They may or may not generate the synergies on which program success normally depends, and they are the ones to ultimately determine program delivery. A new public management culture is developing based on creating consensus and adaptation betweenthose who define the objectives and those who implement the programs, and on seeking the active commitment o f the latter to program's objectives. A movement in this direction was endorsed in Mexico by the National Development Plan 2001-06, but the new paradigm still needs to be fully translated into practice. The "New Public Management" approach that i s gaining momentum in many countries i s characterized by transparency in budgetary processes, management by results, professional status o f public servants, accountability to the clientkitizen, decentralization, and program and policy evaluation. There are different degrees of difficulty in program implementation. Cash transfer programs are the easiest ones to implement once the technicalities o f beneficiary registration, check issuingandthe like are sorted out. Social infrastructure programs require localparticipation for efficient resource allocation, which may complicate them, but they can achieve their 159 immediate goals with limited or no participation because outcomes are essentially technical. Productive programs are the most difficult to implement,because they basically deal with private goods and because public-private synergies are crucial to their success. They also require considerable expenditure inrecurrent costs. The challenges faced by Mexico to improve the implementation system of rural development programs can be grouped in five categories: political and administrative circumstances o f a macro type; operational and budgetary norms; organizational cultures; client orientation and beneficiary empowerment; and the incentive system for program operators. These challenges are summarized below. Macro-Type Political andAdministrative Circumstances Here, challengesinclude the electoral system of local authorities, characterized by a short mandate and no reelection, little functional to the policy continuity required by rural development, which i s a long-term endeavor. The annual budgetary system existing in Mexico does not favor program continuity and the adoption o f a long-term perspective in rural development. Finally, the organization o f the state apparatus along sectoral lines i s little sympathetic to a multi-sectoral matter like rural development. Sub-national levels could play an important role in overcoming "sectoralism". Administrative innovations in this sense from state governments would be most welcome. At the federal level, the Secretaria de Hacienda, which has a multi-sectoral view and i s responsible for the quality o f public investment, i s well placed to take an active role in promoting the coordination o f federal rural programs within a territorial approach. Operational andBudgetavy Norms Simplifying operational and budgetary norms and adapting them better to rural areas is suggested. One problem is the time factor, since timeliness o f support i s often more important than amount o f support. Not only are there no pluri-annual budgets, but operational and budgetary norms sometimes allow a few months only to spend the allocated budget, thus introducing distortions. Changes from year to year in norms related to issues like eligibility conditions, subsidy amounts, target areas or type o f benefits are also a damaging factor when they occur because they introduce uncertainty among program beneficiaries and operators. Another challenge relates to the few recurrent funds often made available for program operation. There i s frequently a conservative attitude toward recurrent costs, which is understandable in view o f the abuses o f the past but i s potentially damaging to programs. This i s particularly the case with productive programs where the formula for success could be summarized as "recurrent costs rural finance". + Organizational Cultures The institutional segmentation culture adds to the problems created to rural development programs by the sectoral organization of the state. To overcome this culture enhanced efforts from SAGARPA to involve more other organizations in the application o f the Ley de Desarrollo Rural Sustentable would be useful. Efforts from the Secretaria de Hacienda and the Oficina de la Presidencia de la Republica to promote more integration o f rural development programs in the framework o f the law, and to encourage the application o f a territorial approach to rural development would also be useful. Strengthening evaluation mechanisms is also suggested by introducing M&E systems simultaneously with program design, better dissemination o f program evaluation results, putting 160 together action agendas following the recommendations made by evaluation teams and monitoringprogress on these agendas, and systematically adopting participatory M&Emethods. An excessive focus on short-term achievement is a hindranceto the continuity and long-term focus of rural development programs. Two policy options could be considered in this connection: the design of a long-term strategy for rural areas as politica de estado cutting across party lines and administration terms, and the recognition o f the importance o f "intermediate policy results" andtheir value to political constituencies. Client Orientation andBeneficiary Empowerment Improving client orientation and beneficiary empowerment is another area where the implementation of rural development programs faces important challenges. There are several options. The first i s to enhance the dissemination o f programs and program norms to prevent biases in the selection process. The dissemination among beneficiaries of evaluation results and o f the action agendas emerging from them i s also suggested. Direct accountability from program operators to ClienUbeneficiaries i s a major way to empower beneficiaries which could be expanded, although it may not always be easy to introduce. Finally, measures could be introduced to detect and prevent opportunistic and rent-seeking behavior on the part o f program operators. Goodprogram dissemination and participatory evaluations are ways to achieve this. Incentives to Program Operators Economic and moral incentives for program operators are key to program success. The economic situation ofbottomlevel program operators is often inconsistentwith the relevance o f their function, because o f low salaries, no job security, and payment delays in some cases. Revising the economic situation of these operators and linking it to performance and client satisfaction would encourage good performance. Maximum performance from bottom level operators would also be encouraged by revaluing their function, consulting with them on program matters, giving them systematic training, disseminating best practices, carrying out systematic evaluations o f their work, and promoting networking systems, a client orientation ethic, and a sound esprit de corps. Other Specific Proposals To advance in the transformations suggested above two policy actions could be considered. The first is the creation of a technical committee to examine the implementation issues o f rural development programs and make recommendations. The committee could be integrated by a mix o f independent specialists, civil servants and legislators, and could have significant participation from the Secretaria de Hacienda. The second action i s to empower the Consejo Mexican0 para el Desarrollo Rural Sustentable to take an active role inthe evaluation o f rural development programs and the monitoring o f the action agendas resulting from program evaluations. The Consejo would also be the natural institution to promote the preparation o f a long-term rural development strategy to propose to the country aspolitica de estado. The creation o f a managing council and o f a small technical secretariat o f the Consejo would be important for itto be able to carry out these functions. Two innovations could help raise program efficiency. One is the introduction of a system o f what we call oidores, consisting o f well trained individuals or teams who would informally follow rural development programs at the "point o f service delivery" through ad hoc visits, and directly inform top managers o f the situation o f the programs and the views on them from the bottom. The other innovation consists o f a system o f process certification.This could be carried out by independent consulting firms or NGOs, who would certify that the processes 161 related to program operation and beneficiary participation are sound, in the sense o f proceeding according to program objectives and guidelines and to accepted practice. The logic i s the same as with environmental certification or auditing, only that applied to implementation processes. The system i s expected to facilitate the delegation o f program implementation functions to rural organizations, since it would give government agencies an independent means o f knowing if the implementation processes are sound. SupportingtheRural Youth as a Critical Sectorfor a Dynamic Rural Economy Young male and female workers constitute an important strategic capital for the development of rural areas. They could be a fundamental actor in the modernization o f rural areas in view o f their higher level o f education and more familiarity with the realities o f the market and globalization than those o f the previous generation. Their potential and energies are frustrated, however, by their conspicuous lack o f access to assets. Land i s held by the older generation, and market and inheritance mechanisms fail to transfer it at a sufficient pace from ageing landholders to young farmers. Lack o f credit to buy land and to exploit it prevents also many young rural workers from acquiring land or renting it in and exploiting it with modern methods. Similarly, young rural workers may have the knowledge and entrepreneurship to start independentRNF operations but do not normally have the necessary capital. Thus, ruralyouth, on which a strong educational effort has been made, i s severely constrained by failures in the financial and assets markets. Inter-generational renewal in farm operations would open an opportunity for farming intensification, since young farmers could carry out the type of changes intechnology, crop mix and farm management requiredby intensification, ifthey have access to finance and technology. Governmentcan interveneinvarious ways in this situationto facilitate the access to assets of the rural youth. One form i s to set up a land fund oriented to young workers to make accessible to them the financial means to acquire land or to rent it in on a medium or long-term basis. As a complement to this, there i s need for an investment fund to allow young farmers to have access to the investments and technology necessary to get started as successful farmers. As mentioned before, the Secretaria de la Reforma Agraria i s initiating a welcome program along these lines inthe social sector to facilitate entrepreneurialdevelopment o fyoung farmers. T o facilitate the inter-generational transfer of land other measures may also be needed. One possibility i s allowing the division o f ejido holdings, which would permit old ejidatarios to keep part o f their land for themselves and transfer the other part to one or more o f their offspring. Another form is aprogram for granting some sort of social security benefits to old farmers who decide to transfer their lands. A better understanding i s also needed o f the reasons preventing at present the natural operation of land renting markets for medium- and long-term leases. Once this i s better understood, measures could be introduced to facilitate this type o f leases. Young rural workers also need access to rural non-farm occupations to increase income opportunities. Since good farm land i s scarce in Mexico and farm holdings are small, even the most optimistic assumption o f land transfers to young farmers would fail to satisfy potential land demand in many places. There i s need, therefore, to facilitate the access of young rural workers to RNF occupations through the support to technical and vocational training, and to the establishment o f RNF enterprises. As mentioned before, the rural economy offers many latent opportunities outside agriculture that could be taken up by young workers. A program for the access to assets o f the ruralyouth couldhence include both farm andnon farm opportunities. 162 The best option for carryingout a programof this type in the social sector would be to rely on the communities themselves to discuss which old ejidatarios or comuneros are willing to pass on their lands to young local landless farmers and under what o f the possible program modalities, and for the young farmers to put together a plan for the exploitation o f these lands, with the technical assistance that may be required. Local governance institutions would need to play a key role in this. Community level plans for land transfers would be the result of these decisions. Program operators could facilitate and technically assist in the preparation o f these plans, which could be largely implementedand monitored by the community themselves. The plans could include not just farm operations but also RNF activities, and should have a middle- and long-term perspective to take into account not only the needs of the current local youth in search o f employment but also those o f the local adolescents that would be enteringthe labor market inthe coming years. A programof this type would benefit from an organizationalcomponentto promote the formation of youth organizations and the networking among them. An important dimension could be the empowering o f rural youth in the decision making processes o f the rural space, promoting their self-confidence, facilitating their leadership intheir communities and other rural organizations, and helpingthem build a generational project that would make the most o f their potential as agents of social and economic change inruralareas. 163 Annexes Annex 2.A: Correlates of Poverty Annex 3.A: Rural Labor Force Characteristics in 1992,1996 and 2002 Annex 3.B: Rural Labor Market Indicators by Gender Annex 3.C: Rural Labor Market Indicators by Region Annex 3.D: Shares of Labor Force Primary Occupation Annex 3.E: Mean Hourly Wages of Rural Laborers Annex 3.F: Wage Correlates using Quantile Regression Annex 3.G: Income Shares inRural Mexico Annex 3.H: Correlates of Rural Non-Farm Employment Annex 3.1: The Territorial Approach to RuralDevelopment Annex 4.A: Public FederalExpenditure inRuralAreas Annex 4.B: Small Farm Efficiency Analysis 164 ANNEX 2.A. THE CORRELATES OF POVERTY This annex investigates the marginal impact o f individual characteristics such as labor market association and human capital on the likelihood that a household fall below the poverty line. The analysis i s carried out usingProbit regressiontechniques, and i s based on three ENIGH surveys, for years 1992, 1996, and 2002. This allows evaluating the evolution o f poverty over time and that of the main variables determiningit. Separate regressions are run for each of these years, using data from localities o f less than 15,000 inhabitants. The analysis reveals: (i) conditional correlation between poverty and the characteristics o f household heads; (ii) information about the volatility o f the impact o f the attributes on the likelihood that a household experience poverty duringthe beginning and mid-1990s and the beginningo f the 2000s; and (iii) information about groups that are particularly vulnerable, and changes in these groups over the decade. We also performed a similar exercise comparing the Sur region with the whole o f Mexico. The status o fthe household-poor or non-poor- i s regressed on relevant individual and household characteristics. The dependent binary variable takes the value o f one when income i s below the food poverty line and zero otherwise. We use six sets o f explanatory variables: (1) attributes o f household head: gender, education, and age; (2) labor market connections: whether the household head works, the type o f relation with the labor market, and sector o f employment; (3) family variables: age o f the household head and dependants; (4) spouse characteristics: education and labor market connection; (5) region; and (6) type o f rural area. Probit coefficients are not easy to interpret, since they do not indicate the standard marginal effects of ordinary linear regression coefficients. We chose, therefore, to present marginal effects, which have a straightforward interpretati~n.~~ Probit regressions should be The interpreted as descriptive, i.e. indicative o f association between explanatory and dependant variables but not o f causation. Results are shown in Tables 2.A.1 and 2.A.2. In the following paragraphs, only the statistically significant coefficients in the regressions are discussed. We summarize below the results for some o fthe main poverty correlates. Rural living Households insemi-urban areas were more likely to escape poverty than those in disperse rural areas. In 2002, households in disperse rural areas were around 12 percentage points more likely to be poor than those in semi-urban areas. Rural dwellers in disperse areas inthe Sur were more likely to experience poverty than those inMexico as awhole in 1992butnot in2002. Labor Status and Sector of Work Formal sector workers inrural areas, i.e. those contributing to the social security system, are much less likely to be poor than their informal sector counterparts. The likelihood o f formal 99The marginaleffects for a householdheadi inthe Probit model are given by: d Pr(yi = 1) m . = dxi =@(XiP>P This represents the marginal changes inthe probability of a household head i being poor due to changes in the regressors. Marginal effects are evaluated at the mean of the data. Since similar conditions apply for marginal effects as for Probit coefficients, the same tests can be applied. 165 sector workers escaping poverty increased during 1992-96 and remained fairly constant since. In 2002, households whose heads were formal sector workers were 24 percentage points less likely to fall into poverty than those whose heads worked inthe informal sector. 166 Table 2.A.1 Probability of Being Poor in Rural Mexico in 1992,1996, and 20021'2 1992 1996 2002 DF/dx SE Blzl dF/dx SE P>lzl dF/dx SE Blzl Household Characteristics ' Dependent below 5 years old+ 0.215 *** 0.017 0 0.234 *** 0.016 0 0.225 *** 0.016 0 Dependent 6 - 11years old+ 0.174 *** 0.017 0 0.25 *** 0.015 0 0.218 *** 0.015 0 Dependent 12 - 14years old+ 0.114 *** 0.019 0 0.119 *** 0.017 0 0.135 *** 0.017 0 Dependent 15 - 18 years old+ 0.026 0.019 0.177 0.053 *** 0.018 0.003 0.057 *** 0.017 0.001 Dependent 18 - 25 years old+ -0.069 *** 0.019 0 -0.046 ** 0.018 0.012 -0.059 *** 0.017 0.001 Dependent 65 and over+ 0.047 ** 0.02 0.018 0.051 *** 0.019 0.007 0.003 0.017 0.839 Head Age + -0.008 *** 0.003 0.009 -0.016 *** 0.003 0 -0.021 *** 0.003 0 Age Square + 0 ** 0 0.031 0 *** 0 0 0 *** 0 0 Female Head+ -0.03 0.028 0.295 -0.089 *** 0.024 0 -0.037 * 0.021 0.087 Education Head Primary Complete+ -0.102 *** 0.02 0 -0.082 *** 0.02 0 -0.097 *** 0.016 0 Lower Secondary Complete+ -0.167 *** 0.027 0 -0.232 *** 0.027 0 -0.18 *** 0.022 0 Upper Secondary Complete+ -0.21 *** 0.042 0 -0.343 *** 0.033 0 -0.239 *** 0.029 0 Spouse Primary Complete+ -0.075 *** 0.021 0.001 -0.069 *** 0.021 0.001 -0.095 *** 0.018 0 Lower Secondary Complete+ -0.116 *** 0.034 0.002 -0.182 *** 0.031 0 -0.171 *** 0.024 0 Upper Secondary Complete+ -0.213 *** 0.051 0.003 -0.364 *** 0.046 0 -0.273 *** 0.032 0 Sector ofActivity andLabor Status Head Unemployed+ 0.083 0.095 0.365 0.205 *** 0.063 0.004 0.074 0.092 0.414 Not inthe labor force+ 0.06 0.038 0.103 0.05 0.032 0.128 0.105 *** 0.03 0 Agviculture Self employed+ 0.006 0.025 0.808 0.031 0.025 0.213 0.219 *** 0.024 0 Salaried worker+ 0.166 *** 0.026 0 0.155 *** 0.023 0 0.216 *** 0.023 0 Employer+ -0.156 *** 0.026 0 -0.185 *** 0.033 0 -0.054 0.041 0.199 Off-farm Sector Self employed+ -0.04 0.03 0.195 -0.044 0.029 0.126 0.046 * 0.027 0.083 Salaried worker+ Employer+ -0.255 *** 0.034 0 -0.356 *** 0.039 0 -0.141 ** 0.051 0.015 Second Employment + 0.01 0.017 0.55 -0.031 * 0.017 0.07 -0.061 *** 0.016 0 Social Security + -0.206 *** 0.02 0 -0.251 *** 0.024 0 -0.235 *** 0.019 0 Spouse Employed in agriculture + -0.017 0.031 0.595 -0.019 0.027 0.487 0.065 *** 0.026 0.01 Employed in off-farm activity+ -0.067 *** 0.023 0.005 -0.123 *** 0.021 0 -0.133 *** 0.017 0 Region Norte + -0.166 *** 0.022 0 -0.155 *** 0.026 0 -0.051 ** 0.024 0.041 Golfo Region+ 0.061 ** 0.026 0.017 0.117 *** 0.023 0 0.222 *** 0.024 0 Pacifica+ -0.145 *** 0.022 0 -0.152 *** 0.025 0 -0.01 0.025 0.696 Sur+ 0.134 *** 0.029 0 0.161 *** 0.024 0 0.19 *** 0.023 0 Centro+ 0.037 0.024 0.113 0.096 *** 0.025 0 0.155 *** 0.025 0 Capital+ -0.113 *** 0.038 0.008 0.014 0.05 0.778 -0.086 *** 0.029 0.005 Locality < 2,500 inhabitants+ 0.085 *** 0.02 0 0.132 *** 0.017 0 0.12 *** 0.015 0 Number o f observations = 4752 6165 6481 Log Likelihood= -2442.4 -3176.2 -3145.7 LRchi2(24)= 1360.4 2180.1 2448.7 Prob>chi2= 0 0 0 Pseudo R2= 0.218 0.256 0.28 Source: Authors estimations based on ENIGH 1992, 1996, and 2002. Rural area defined as localities with less than 15,000 inhabitants. SEDESOL's capacity poverty line, (+) dF/dxis for discrete change o f dummyvariable from 0 to 1,*** sign. at 19'0, ** sign. at 5%,* sign. at 10%. 167 Households with heads inactive in 2002 were more likely to experience poverty than those with active heads. This was not the case inthe early and mid-l990s, as the difference inthe likelihood o f falling into poverty for the two groups was no statistically significant. Surprisingly, the probability o f being poor o f households with unemployed heads"' i s similar to that o f households with employed heads. This holds in both 1992 and 2002, but not in 1996, after the crises, when unemployed heads were more likely to experience poverty than employed ones. Employers are the group with lowest probability o f being poor followed by self- employed and salaried workers. In 2002, households headed by employers in non-agricultural activities were 14 percentage points less likely to be poor than those headed by salaried workers, down from 26 percent in 1992 and 36 percent in 1996. In 1992, households with self-employed heads were 4 percentage points less likely to experience poverty than those headed by salaried workers in the off-farm sector. By 2002, this had changed and household with self-employed heads had the same probability o f being poor than those headed by non-agriculture salaried workers. In 1992 households whose head was self-employed in agriculture were not worse off than those headed by wage workers in the non-farm sector, but were 4 percentage points more likely to be poor than those headed by self-employed inthe non-farm sector. During 1992-2002, households o f heads self-employed in agriculture experienced a 22 percentage point increase in the likelihood o f experiencing poverty. Households headed by salaried workers in agriculture saw their probability o f beingpoor increase by 5 percentage points between 1992 and 2002 compared to households inthe non-farm sector. Households where spouses were engaged in the off-farm sector had lower probability o f being poor than households where spouses were engaged in agriculture or not working. This became stronger throughout the 1992-2002 period. In 1992 and 1996,households whose heads took a secondjob were neither more nor less likely to escape poverty than households whose heads did not have an extra job, but in 2002 households with a secondjob were 6 percentage points more likely to escape poverty. The effect o f a second job became hence poverty reducing after 1996, and a way for some rural families to escape or reduce household poverty. We discuss this inmore detail inchapter 3. Education and Skills Complete levels o f education o f the household head and spouse are very important to escape poverty. All education variables in the three models are statistically significant and negatively correlated with the probability o f being poor, starting with completed primary education. Controlling for other variables, educational attainment i s the strongest correlate o f poverty among the variables included inthe right hand side o f the regression. The negative effect o f education on poverty (i.e. the positive effect for poverty reduction) increases with the level o f completed education o f the householdhead and the spouse. The association between the likelihood o f escaping poverty and having completed primary education in comparison with no or incomplete primary education has been rather constant, at around 10 percent over the last decade. The association o f secondary education i s larger. Between 1992 and 2002, households headed by persons with completed lower secondary education had a probability o f beingpoor between 17 and 24 percent lower than their peers with no completed education. It i s interestingthat the probability o f falling into poverty o f secondary Notice that unemployed is different from inactive. The latter indicating persons who are not inthe labor force like for example retired persons, students, and other people not seeking employment. 168 school graduates decreased between 1992 and 1996 and increased between 1996 and 2002, reaching in 2002 a level not much different from that o f 1992. This i s probably related to labor market adjustments after the Tequila crises (see Chapter 3). Highreturns from education inrural Mexico were observed by Taylor and Yunez-Naude (2000) usingdifferent data: a sample o f rural households inthe states o f Michoacan, Jalisco, Coahuila and Puebla. They show increases o f 10 percent in household income per every year o f household head schooling and 5.5 percent per every year of average schooling o f other householdmembers. Human capital has many components. An important one, apart from formal education, i s experience. In the labor market literature, experience i s often proxied by the age o f the worker. We include age and age squared as regression variables, the latter to capture possible non- linearities in the data. In all three regressions, age i s positively associated with the probability o f escaping poverty, becoming increasingly more poverty reducing during the decade. One more year o f age decreases the likelihood o f being poor by 2 percentage points in 2002, up from less than 1 percentage points in 1992. There does not seem to be a turning point in age when the probability o f beingpoor starts to fall, since the age-square variable i s not significantly different from zero. 169 Table 2.A.2 Probability of Being Poor inthe RuralAreas of the Sur region and inthose of Mexico as a whole, 1992 and 20021>*3 1o w -,,- -""- ?nn? Variables multiplied by Sur Variables multiplied by Sur Region dummy Region dummy DF/dx SE P>lz dF/dx SE P+ dF/dx SE P>lz dF/dx SE P>lz Household Characteristics Dependentbelow 5 years old+ 0.21 *** 0.018 0 0.069 0.056 0.205 0.216 *** 0.017 0 0.025 0.039 0.521 Dependent6 - 11years old+ 0.181 *** 0.018 0 0.028 0.056 0.604 0.2198 *** 0.02 0 -0.033 0.038 0.392 Dependent 12 - 14years old+ 0.118 *** 0.02 0 -0.006 0.055 0.911 0.1341 *** 0.02 0 0.023 0.042 0.582 Dependent 15 - 18years old+ 0.021 0.02 0.29 0.085 0.063 0.162 0.0542 *** 0.02 0.004 -0.011 0.041 0.785 Dependent 18 -25 years old+ -0,075 *** 0.02 0 0.127 * 0.069 0.058 0.0729 *** 0.02 0 0.093 ** 0.046 0.037 Dependent65 and over+ 0.055 *** 0.021 0.01 0.013 0.064 0.835 0.0081 0.02 0.671 0.045 0.04 0.265 Household Head Age + -0.01 *** 0.003 0.002 -0.007 0.005 0.146 -0.022 *** 0.003 0 0.002 0.003 0.471 Age Square + 0 ** 0 0.016 0 0 0.15 0 *** 0 0 0 0 0.441 Female Head+ -0.035 0.029 0.24 0.049 0.094 0.596 -0.025 0.024 0.295 0.047 0.055 0.384 Education Head PrimaryComplete+ -0.091 *** 0.021 0 -0.134 ** 0.057 0.041 -0.128 *** 0.019 0 0.086 * 0.052 0.094 Lower SecondaryComplete+ -0.167 *** 0.029 0 0.047 0.141 0.735 -0.196 *** 0.022 0 0.123 0.08 0.113 Upper SecondaryComplete+ -0.205 *** 0.044 0.001 -0.25 *** 0.031 0 0.07 0.124 0.565 Spouse PrimaryComplete+ -0.086 *** 0.022 0 -0.022 0.081 0.785 -0.122 *** 0.02 0 0.049 0.06 0.403 Lower SecondaryComplete+ -0.138 *** 0.033 0 -0.117 0.148 0.482 -0.186 *** 0.024 0 0.047 0.086 0.583 Upper SecondaryComplete+ -0.229 *** 0.051 0.003 -0.035 0.284 0.905 -0.268 *** 0.035 0 -0.204 0.143 0.274 Sector of Activity and Labor Status Head Unemployed+ -0.01 0.087 0.908 -0.017 0.092 0.855 Not inthe labor force+ 0.044 0.039 0.249 -0.158 0.09 0.16 0.078 ** 0.032 0.014 0.044 0.083 0.589 Agviculture Self employed+ -0.026 0.026 0.318 0.091 0.091 0.3 0.19 *** 0.026 0 0.1387 ** 0.07 0.03 Salariedworker+ 0.144 *** 0.027 0 -0.053 0.085 0.546 0.224 *** 0.025 0 0.0459 0.07 0.48 Employer+ -0.165 *** 0.027 0 -0.004 0.103 0.965 -0.054 0.048 0.274 0.0619 0.11 0.55 Off-Farm Sector Self employed+ -0.065 ** 0.031 0.046 0.105 0.113 0.335 0.07 ** 0.029 0.014 -0.071 0.06 0.29 Salariedworker+ Employer+ -0.277 *** 0.031 0 0.078 0.29 0.782 -0.165 *** 0.053 0.008 0.1408 0.19 0.45 SecondEmployment + 0.039 ** 0.018 0.029 -0.076 0.045 0.115 -0.054 *** 0.018 0.003 -0.013 0.04 0.75 Social Security + -0.235 *** 0.019 0 -0.027 0.111 0.81 -0.241 *** 0.019 0 0.0097 0.1 0.92 Spouse Employedinagriculture + -0.023 0.033 0.498 0.068 0.107 0.511 0.038 0.031 0.218 0.1236 ** 0.06 0.03 Employedinoff-farmactivity+ -0.085 *** 0.024 0.001 0.178 ** 0.084 0.029 -0.135 *** 0.02 0 0.0607 0.05 0.21 Region Locality< 2,500 inhabitants+ 0.039 * 0.022 0.089 0.22 *** 0.075 0.003 0.111 *** 0.017 0 -0.078 ** 0.04 0.05 Number o fobservations = 4743 6478 Log Likelihood= -2483.4 3223.9 LRchi2(53)= 1269.1 2286.8 Prob>chi2= 0 0 PseudoR2= 0.2035 0.262 line, (+) dF/dxis for discrete change o f dummy variable from 0 to 1,*** sign. at 19'0, ** sign. at 59'0, * sign. at 10% level Source: Authors estimations based on ENIGH 1992 and 2002. Note: lRural area defined as locality with less than 15,000 inhabitants; SEDESOL's capacity poverty Gender Gender made a statistically significant difference for poverty in 1996 and, with smaller significance, in 2002, when female-headed households were less likely to be poor than male- headed ones. Instead, in 1992 the probability o f female-headed households being poor was not statistically different to that o f male-headed households. This finding i s different from that for other countries, for example Brazil, where male headed households have been found to have a lower probability o f beingpoor (see Elberset al, 2001), or inEcuador where there seems to be no association betweenpoverty and gender o f the Household head (see World Bank, 2004~). 170 Household Composition The presence o f children or youth inthe householdmakes it more poverty prone, but the probability o f beingpoor falls monotonically with increasing child age. Households with children under 5 are more likely to be poor than childless families, and their higher probability of being poor has been rather constant over the past decade. This indicates that households with young children are vulnerable relative to those without children. One direct policy intervention would be to facilitate access to childcare. The poor often find the shortage o f affordable childcare a large obstacle to their daily chores, and for mothers in particular an obstacle to find employment outside their homes. Households with children between 6 and 11 years have also higher probability o f being poor than those without children, albeit the likelihood i s lower than that for families with smaller children. For these households, the probability o f being poor compared to those without children increased from 17 percent in 1992 to 25 percent in 1996, and fell to 22 percent in 2002. Households with young members between 15 and 18 experienced the same pattern in the likelihood o f falling into poverty as households with children in the 6-11 year bracket, butthe impact was significantly smaller. The picture changes dramatically when the age o fyoung householdmembersincreases to 19-25 years. Households with members aged 19 to 25 were significantly less likely to fall into poverty than households with no children. Having members in the household aged 19 to 25 can hence be considered a protective factor against poverty. The fact that young members enter the labor market and bring home an income contributes positively to the household's poverty situation. In the Sur region, however, households with young members do not experience the same lower probability o f being poor in 2002, although the situation was different in 1992. This may be related to higher migration in 2002 with the drainage that this may entail o f the more skilled and able. The presence o f older members (above 65 years o f age) in the household makes it more poverty prone. In 2002, households with members o f old age experienced a higher likelihood o f poverty than those without them but the magnitude was lower than in 1992: 0.3 percent in 2002 comparedto 5 percent in 1992. Regions Households in the Norte, Pacifica, Capital, and Centro were in general less likely to be poor than those in the Centro-Norte throughout the decade. On the contrary, households in the Golfo, Centro, and Sur were more likely to be poor than those in the Centro Norte. In 2002, households living in the Sur and Golfo had roughly 20 percent more probability o f being poor than households inthe Centro-Norte. Ifwe look at trends, the Centro-Norte improvedits position vis-&vis all other regions duringthe decade, since comparing 2002 with 1992 rural households in all regions hadincreased their probability of beingpoor relative to the Centro-Norte. 171 '" " 0 ' N =22 w w r - N U N m w v ; r j "N mqNN"i "Or- > m -- Annex 3.C. - Table 3.C.1. Rural' Labor Markets Indicators in Mexico, by Region. Not in the labor force Region 1995 1996 1997' 1998 1999 2000 2001 2002 2003 Total 7,279,960 7,168,950 8,175,566 7,134,766 7,373,689 7,918,307 8,288,969 8,320,254 8,599,179 Norte 822,433 655,958 722,777 605,943 637,113 674,210 695,430 718,416 715,469 Capital 477,821 375,912 587,016 452,521 527,124 516,253 565,549 623,060 705,628 Golfo 1,531,541 1,276,139 1,308,778 1,285,842 1,291,970 1,380,364 1,630,660 1,602,782 1,614,280 Pacifico 517,435 657,019 808,074 678,244 738,489 749,667 683,514 727,102 702,618 SLU 1,581,378 1,849,748 2,027,270 1,788,119 1,775,037 2,033,804 2,169,255 2,248,395 2,403,730 Centro-norte 1,178,860 1,411,504 1,808,104 1,414,121 1,457,901 1,462,524 1,498,163 1,393,099 1,458,492 Centro 1,170,492 942,670 913,547 909,976 946,055 1,101,485 1,046,398 1,007,400 998,962 In the Labor Force Region 1995 1996 1997 1998 1999 2000 2001 2002 2003 Total 9,808,413 9,316,996 10,915,253 9,463,444 9,490,050 9,302,190 9,128,347 9,430,459 9,265,957 Norte 898,803 856,325 1,268,799 881,232 856,799 816,414 790,204 788,055 779,539 Capital 604,507 858,955 1,100,605 736,685 674,049 737,411 721,378 665,976 605,635 Golfo 1,768,582 1,538,406 1,470,448 1,677,011 1,708,643 1,730,994 1,507,139 1,556,638 1,540,116 Pacifico 589,533 934,075 1,169,803 955,607 889,055 869,426 937,779 983,581 977,527 SLU 2,944,769 2,535,458 2,879,795 2,619,908 2,718,212 2,596,868 2,545,112 2,605,679 2,549,012 Centro-norte 1,397,501 1,435,311 1,639,633 1,415,729 1,470,445 1,459,124 1,443,330 1,581,671 1,538,886 Centro 1,604,718 1,158,466 1,386,170 1,177,272 1,172,847 1,091,953 1,183,405 1,248,859 1,275,242 Employed Region 1995 1996 1997 1998 1999 2000 2001 2002 2003 Total 9,605,455 9,210,159 10,792,852 9,389,503 9,454,164 9,243,682 9,060,090 9,354,843 9,202,363 Norte 879,776 842,808 1,244,061 868,143 848,743 809,473 776,213 774,597 766,160 Capital 587,435 835,218 1,073,021 721,862 673,813 728,755 718,698 653,356 603,310 Golfo 1,725,884 1,523,797 1,450,252 1,666,032 1,702,035 1,723,682 1,495,798 1,547,849 1,537,097 Pacifico 574,003 920,261 1,153,411 949,795 883,310 863,937 931,328 974,269 973,019 SLU 2,914,843 2,526,662 2,876,724 2,615,837 2,715,399 2,585,943 2,538,637 2,595,379 2,539,804 Centro-norte 1,347,671 1,417,006 1,630,967 1,405,679 1,461,728 1,448,949 1,433,406 1,569,048 1,516,329 Centro 1,575,843 1,144,407 1,364,416 1,162,155 1,169,136 1,082,943 1,166,010 1,240,345 1,266,644 TJnemnlnved ... .. .... - Region 1995 1996 1997 1998 1999 2000 2001 2002 2003 Total 202,958 106,837 122,401 73,941 35,886 58,508 68,257 75,616 63,594 Norte 19,027 13,517 24,738 13,089 8,056 6,941 13,991 13,458 13,379 Capital 17,072 23,737 27,584 14,823 236 8,656 2,680 12,620 2,325 Golfo 42,698 14,609 20,196 10,979 6,608 7,312 11,341 8,789 3,019 Pacifico 15,530 13,814 16,392 5,812 5,745 5,489 6,451 9,312 4,508 SLU 29,926 8,796 3,071 4,071 2,813 10,925 6,475 10,300 9,208 Centro-norte 49,830 18,305 8,666 10,050 8,717 10,175 9,924 12,623 22,557 Centro 28,875 14,059 21,754 15,117 3,711 9,010 17,395 8,514 8,598 Source ENE 2ndquarter, various years. * Rural1997 survey area defined as localities with less than 2,500 inhabitants The has some sampling problems for the rural areas, according con INEGI's staff 177 ANNEX 3.D. - Table 3.D.1. Share of the Workforce by Primary Occupation" 1995 2003 Mean Hourly Mean Hourly Labor Composition Wage" Labor Composition Wage Urban Ruralb Urban Rural Urban Rural Urban Rural Agriculture 9.63 62.82 12.04 8.3 5.4 55.61 13.46 7.44 Cultivation 7.71 53.75 11.4 7.79 4.25 43.38 12.92 6.76 Animal rearing 1.35 5.13 14.47 9.18 0.69 6.02 14.3 9.3 Forest oroduct 0.09 1.02 11.98 13.35 0.04 1.27 12.12 9.44 Fishing 0.47 2.92 13.17 11.37 0.41 4.95 16.74 11.1 Mining/extraction 0.34 0.68 18.63 12.67 0.39 0.13 36.55 16.59 Manufacturing 24.75 11.05 17.14 11.78 26.26 18.51 19.07 12.35 Foodprocessing 3.26 1.23 14.11 13.11 3.75 3.22 16.57 10.85 Beverages 0.69 0.18 15.12 13.93 0.68 0.18 19.69 10.51 Tobacco products 0.05 0.02 21.11 14.19 0.01 0 26.86 Textiles 0.74 0.87 15.75 10.13 0.98 0.94 15.88 7.14 Clothing 2.05 1.44 13.91 9.53 2.24 2.44 14.55 9.2 Leather 0.19 0 17.08 9.08 0.14 0.11 22.37 8.64 Footwear 0.61 0.04 15.05 8.75 0.52 0.01 19.58 7.51 Wooden goods 0.48 0.45 18.16 10.44 0.44 1.08 17.95 9.22 Furniture 0.91 0.62 14.72 11.84 0.95 0.82 18.45 12.78 Paper 0.39 0.08 16.01 10.91 0.35 0.11 18.27 12.83 Printing 0.91 0.06 17.81 18 0.72 0.02 21.31 13.36 Chemical 0.86 0.21 26.03 15.06 0.48 0.03 31.87 19.8 Plastidrubber 0.81 0.16 20.73 12.03 0.77 0.13 18.48 12.93 Ceramdcement 1.08 0.55 17.97 8.7 0.83 1.13 18.03 12.28 Pharmaceuticals 0.22 0 27.21 0.2 0.02 32.99 24.36 Cosmetics 0.26 0 15.75 0.23 0.09 17.8 11.56 Metals 0.38 0.04 23.06 17.4 0.26 0.03 24.1 16.96 Machinery 1.98 0.21 17.75 11.32 2.07 0.55 19.92 15.47 Electronic goods 1.08 0.12 16.66 9.49 1.17 0.08 19.94 13.34 Vehicles 1.39 0.1 17.39 15.09 1.51 0.49 20.31 14.29 Precisioninstruments/others 0.43 0.03 19.31 13.04 0.56 0.09 18.11 16.44 Construction 5.68 4.59 17.28 12.46 6.75 6.81 19.61 14.53 Utilities 0.3 1 0.05 26.8 19.34 0.66 0.12 26.75 18.31 Commerce 21.15 11.66 16.6 10.78 21.54 9.97 17.16 10.53 Wholesaling 3.05 0.72 21.49 12.72 3.33 0.7 21.25 13.01 Formal sales 14.33 8.99 15.36 10.09 18.21 9.27 16.33 10.3 Street sales 3.77 1.95 16.66 12.67 0 0 0 0 Services 44.14 13.79 21.74 13.94 46.41 15.77 23.02 14.13 HoteURestaurant 5.74 1.6 15.37 11.26 6.3 1 2.32 16.99 12.18 Transport 4.78 1.84 18.21 13.82 4.88 1.56 19.99 14.28 Communications 0.46 0.13 25.83 14.48 0.58 0.06 27.58 9.01 Financial services 1.25 0.04 35.14 24.38 0.9 0.03 31.19 21.16 Professional services 2.92 0.29 24.52 9.98 4.05 0.44 25.19 13.48 Education 6.15 1.78 34.73 26.4 5.89 2.13 37.7 26.28 Adentertainment 1.25 0.26 22.58 16.21 1.42 0.35 25.06 16.27 Medical services 3.06 0.47 25.28 16.55 3.41 0.64 30.13 18.07 Sewicinghepair 8.26 2.57 17.08 12.38 7.33 2.21 18.24 13.53 Personal services 4.81 2.98 12.63 10.19 5.71 3.96 14.18 9.65 Renting services 0.23 0.05 26.09 10.81 0.25 0.02 26.33 13.94 Government 5.12 1.72 24.18 12.81 5.66 2.04 27.28 15.98 NE 0.11 0.08 12.62 8.87 0.03 0 14.68 Non-agricultural Total 90.37 37.18 94.6 44.39 Employed 24,267,941 9,605,455 31,430,834 9,202,363 Source: ENE 1995 and- 2003. "People age 12 and over. bLocalities with less than 2,500 inhabitants. "2002 pesos 178 2 2 i 2 ? i M 0; 2 N 2 ". o! $1 K Ici i N i o! i 2 ". ? 2 z? 5 i ". r. z M Ici i r. N Ici ". i c? ". i s o\ !? 2 i $1 N 2 2 2 o! 2 o\ 2 2 ". i $1 NG 2 i v; o! ". ? i 2 x i Ici i o\ N c? Ici 2 i 2 r- ". i ". ". W G 2 M o\ N ". i ". x W 2 5 w W i N "i 2 x 2 i c i 2 2 09 i N ? i 2 "i x Icix 2 2 !? i !? M N sr. 2 2 09 r. z 2 z". W N ". ? : W K z o\ o\ N 2 s 2? 2 2 2 ". N M c? x ? i 2 2 -! i i 0; 2 i N ? r- ? o\ 2 i 2 2 z? U W -! 2 M 0; ? i z".$1 2 N s" tia ANNEX 3.F.WAGE CORRELATESUSINGQUANTILEREGRESSION A. Methodology Economic model The underlyingeconomic model usedinthe analysis will simply follow Mincer's (1974) human capital earnings fbnction extended to control for a number o f other variables that relate to location. Inparticular, we apply a semi-logarithmic framework that has the form: where Inyi i s the log o f earnings or wages for an individual i,xi i s a measure o f a number o f personal characteristics, including human capital variables, ethnicity, etc.; and zi represents location specific variables-for instance, metropolitan living. The last component, ui, i s a random disturbance term that captures unobserved characteristics The functional form i s left unspecified inequation (1). Quantile regression Labor market studies usually make use o f conditional mean regression estimators, such as Ordinary Least Squares. This technique i s subject to criticism because o f the often heroic assumptions underlying the approach. One is the assumption o f homoskedasticity in the distribution o f the error terms. If the sample i s not completely homogenous, this approach, by forcing the parameters to be the same across the entire distribution o f individuals, may be too restrictive andmay hide important information. The method applied inthis paper is quantile regression. The idea is that one can choose any quantile and thus obtain many different parameter estimates on the same variable. In this manner the entire conditional distribution can be explored. By testing whether coefficients for a given variable across different quantiles are significantly different, one implicitly also tests for conditional heteroskedasticity across the wage distribution. This i s particular interesting for developing countries where wage disparities are large and returns to, for example, human capital may vary across the distribution. The method has many other virtues apart from beingrobust to heteroskedasticity. When the error term i s non-normal, for instance, quantile regression estimators may be more efficient than least square estimators. Furthermore, since quantile regression minimizes a weighted sum o f absolute deviations, the estimated coefficient vector i s not sensitive to outlier observations on the dependentvariable.'" lo1 That is, if j-.:Be > 0 ,then y can be increased toward + w, or if ;Be< 0 ,y can be decreased toward -w, without altering the solutionBo. In j- other words, it is not the magnitude of the dependent variable that matters but on which && of the estimated hyperplane the observation is. This is most easily seen by considering the first-order-condition, which can be shown to be given as (see Buchinsky 1998) J-Z(8 +++sgn( n - y i- xz!p,>>xi= 0. i=l 180 The main advantage o f quantile regressions is the semi-parametric nature o f the approach, which relaxes the restrictions on the parameters to be fixed across the entire distribution. Intuitively, quantile regression estimates convey information on wage differentials arising from non-observable characteristics among individuals otherwise observationally equivalent. In other words, by using quantile regressions, we can determine if individuals that rank in different positions in the conditional distribution (i.e., individuals that have higher or lower wages than predicted by observable characteristics) receive different premiums to education, tenure, or to other relevant observable variables. Formally the method, first developed by Koenker and Basset (1978), can be formulated ado2 where Quanh(yi I xi) denotes the Oth conditional quantile o f y given x, and i denotes an index over all individuals, i 1,...,n. = Ingeneral, the Oth sample quantile (0 <8< 1) ofy solves Buchinsky (1998) examines various estimators for the asymptotic covariance matrix and concludes that the design matrix bootstrap performs the best. Inthis paper, the standard errors are obtained by bootstrapping, using200 repetitions. This i s inline with the literature. B.Main results We compare workers located at different points in the wage distribution to analyze this issue. The wage determination model i s based on the ENE survey from 2003 (2"d quarter) using quantile regression (see above for details on the quantile method). Wages are compared across workers grouped by gender, education, experience, labor status, and locationlo3.Results for rural areas with more than 15,000 inhabitants are presented in Table 3.F.1. We analyze for each quantile how the above variables explain the wage at the quantile, and also whether the impact o f individual characteristics on wages i s homogeneous across the wage distribution. Findings indicate that wages are by no means determinedinthe same way for high and low paid workers. This can be seenboth as a strength and weakness of the method. To the extent that a given outlier represents a feature of "the true" distribution of the population, one would prefer the estimator to be sensitive to such an outlier -at least to a certain degree. IO2SeeBuchinsky (1998). 103Wages are modeled using log monthly wages as the dependent variable. The general wage model contains explanatory variables in levels and allows for non-linearities in the data. For example, the logwage equation is found to be non-linear ineducation and experience, indicating that returns to education and experience are not constant but decreasing over the life cycle. The model contains dummy variables that take the value of one if, for example, a worker holds a job inthe formal sector, and zero otherwise. A positive dummyvariable inthis example reveals that there is a wage premium in formal employment related to informal employment. We use standard quantiles, namely the lo*, 25*, 50*, 75*, and90* quantiles. 181 For example, female workers are paid much less than males working inthe high end o f the wage distribution relative to their peers inthe low end o f the distribution, and returns to lower levels o f education are far smaller inthe upper income quantiles than inthe lower ones. All coefficients of the explanatory variables included have the expected signs and are statistically significantly different from zero for all quantiles. In the following pages we discuss the results obtained for each o f the explanatory variables: (1) education, (2) experience, (3) labor market association, (4) occupation and employment sector, (5) gender, and (6) disperse rural area and regional location. Table 3.F.1. Wage Determination in Rural Mexico (Quantile Regression), 2003 Quantile Log monthly labor income 0.1 0.25 0.5 0.75 0.9 Coeff SE Coeff SE Coeff SE Coeff SE Coeff SE Worker Characteristics Age 0.054 0.004 0.049 0.002 0.043 0.002 0.038 0.001 0.038 0.002 Age Square -0.001 0 -0.001 0 -0.001 0 0 0 0 0 Marriedwoman w/o children -0.271 0.049 -0.289 0.065 -0.261 0.025 -0.288 0.038 -0.248 0.04 Marriedwoman with children -0.443 0.026 -0.371 0.019 -0.321 0.013 -0.329 0.013 -0.336 0.016 Single woman w/o children -0.317 0.03 -0.303 0.016 -0.33 0.009 -0.351 0.011 -0.33 0.013 Single woman with children -0.255 0.031 -0.252 0.022 -0.28 0.016 -0.327 0.014 -0.35 0.017 Labor Status Employer 0.218 0.061 0.388 0.027 0.507 0.024 0.593 0.031 0.71 0.035 Self-employed -1.516 0.052 -1.163 0.028 -0.623 0.014 -0.278 0.016 -0.118 0.026 Informal Salaried -0.116 0.036 -0.12 0.017 -0.133 0.012 -0.173 0.013 -0.255 0.027 Formal Salaried 0.418 0.027 0.235 0.017 0.139 0.015 0.079 0.017 0.012 0.032 Contract -0.901 0.066 -0.439 0.039 -0.16 0.022 -0.071 0.023 -0.024 0.036 Primary Complete+ 0.258 0.016 0.245 0.016 0.202 0.011 0.175 0.011 0.173 0.013 Lower Secondary Complete+ 0.42 0.023 0.367 0.016 0.296 0.013 0.269 0.011 0.264 0.015 Upper Secondary Complete+ 0.523 0.03 0.495 0.018 0.456 0.02 0.456 0.02 0.51 0.024 University Complete 0.956 0.036 0.98 0.021 0.986 0.022 0.988 0.019 1.03 0.03 Technical Education 0.633 0.03 0.579 0.022 0.537 0.022 0.536 0.024 0.624 0.033 Region ' Norte + 0.376 0.035 0.376 0.019 0.376 0.019 0.253 0.012 0.274 0.026 Capital+ 0.363 0.046 0.325 0.022 0.325 0.022 0.187 0.014 0.152 0.025 Golfo + -0.105 0.034 -0.086 0.023 -0.086 0.023 -0.088 0.015 -0.081 0.013 Pacifica+ 0.372 0.038 0.343 0.022 0.343 0.022 0.198 0.014 0.186 0.012 SUI+ 0.101 0.028 -0.022 0.023 -0.022 0.023 -0.038 0.013 -0.039 0.015 Centro-Norte+ 0.215 0.037 0.202 0.025 0.202 0.025 0.104 0.012 0.093 0.014 Locality < 2,500 inhabitants+ -0.156 0.016 -0.192 0.011 -0.192 0.011 -0.141 0.009 -0.123 0.012 Constant 5.616 0.088 6.218 0.063 6.218 0.063 7.213 0.034 7.491 0.052 Rural area defined as localities with less than 15,000 inhabitants. Source: Calculations based on ENE 2003, 2nd quarter. Note: figures in italics are statistically significant at 10 percent only. Education Human capital has long proven to be important inenhancing economic gr0~th.l'~more A educated workforce i s likely to increase productivity by virtue o f being more flexible and innovative, and to facilitate the adoption and use o f new technologies. Knowledge about wage differentials or gaps due to education serves at least three different purposes. First, wage differentials reveal the magnitude o f the incentives that workers have to acquire education, and hence the incentives for educational demand by individuals. Second, informationo f the impact o f education on wages makes it possible to assess whether it i s worth investing in this area with preference to others. Third, wage differentials disclose how the labor market translates educational inequalities into wage inequalities -a usefbl information to reduce inequality. See, for example, Barro (1991) and Mankiw, Romer, and Weil(l992). 182 Furthermore, the relation betweeneducation and wage levels links educationto labor productivity and thus points to the magnitude o f the contribution o f education to economic growth. It i s o f interest, therefore, to estimate the association between different levels o f education and experience and money wagedo5. Results in Table 3.F. 1 confirm the findings o f many other studies, namely that education plays an important role in the wage determination process. Better-educated individuals earn higherwages thantheir less-educatedpeers. Wages vary according to education levels.lo6 In this analysis, we compare workers who have not completed any level o f education (the reference group) with workers who have completed primary school, lower secondary school, higher secondary school, tertiary education, and some form o f technical ed~cati0n.l'~ In2003, the association with the wage level ofprimary, lower secondary, upper secondary, tertiary, and technical education relative to no or incomplete primary education was positive at all quantiles, controlling for other individual characteristics. Having completed primary education contributes to better wages, and the premium increases rapidly with the level o f education attainment. Compared to the wages o f non-educated workers and those with incomplete primary, median wages o f workers with complete primary, lower secondary, upper secondary and tertiary education were respectively 22, 34, 58, and 168 percent higher1'*. Workers with complete technical education received a 71 percent higher return compared to peers with no complete education. Better-educatedindividuals in rural Mexico earn therefore dramatically higherwages than their less-educated counterparts. Returns across the wage distribution are fairly constant for workers with complete upper secondary and tertiary education, i.e. workers in the low end o f the income distribution are not beingpaid comparatively less than their peers inthe high end. This would seem to indicate that: (1) there is no wide heterogeneity in the quality of education in rural areas across the wage distribution, and (2) the capacity o f workers to convert their educational capital into higher earnings through labor market networks i s similar for poorer and richer workers. Hence, poor people with education seem to benefit from good labor market connections to the same degree as richer people. Workers with complete primary and lower secondary education face decreasing returns across the wage distribution, i.e. those at the low end are paid proportionally more than those at losAnissue to be flagged is the possible endogeneity of education inthe regression. There is vast evidence of a positive correlation between earnings and education but it is difficult to decide whether the higher earnings observed for better educated employees are caused by their higher level of education, or whether employees with more earning capacity have chosen to acquire more education. Also to be flagged is that wage levels are not only related directly to education as stated in the wage equation but also indirectly through the relation between education and labor status -a choice variable. Educationmay affect the choice of labor status (e.g. being an employer) and in turn affect the level of wage derived by an employer. The importance of education is hence likely to beunderestimated inour wage equation. 1% Unmeasured ability and measurement error problems have been dealt with in the literature applying data on twins, see for example Card (1998) and Arias, Hollack, and Sosa (1999). I O 7According to the so-called "sheepskin effect" there are wage premiums for completing the final year of elementary school, high school, or university. It has been argued that credentials such as a school diploma or university degree are more important than years of schooling per se. This is one of the reasons for not having a continuous education variable inthe regression. The percentage return is calculated as (exp (coefficient estimate) - 1)* 100. All figures presented in the following paragraphs are percentage premiums thus calculated from the marginal coefficients inTable 3.5. 183 the high end, indicating that workers with the same level o f education are not compensated equally. The poor (1Oth quantile) receive a wage premiumwhen completing primary education o f 29 percent, while the rich (90thquantile) receive only 19 percent. Inthe case o f lower secondary schooling, workers in the low end (loth quantile) obtain a premiumo f 52 percent, while workers in the top end (90th quantile), obtain only 30 percent. One possible explanation is that social networks that facilitate labor market connections operate better among the poorer than the richer segments o f the rural labor force. Another i s that these levels o f schooling are more relevant for employers hiringworkers at the low than at the highend o fthe wage distribution. Experience There are several reasons for including experience inthe analysis. One i s that experience, together with education, provides flexibility in adapting to changes in technology or other economic changes. Experience and years o f schooling are widely used in analyses o f wage determination (see Mincer 1974, and Levy and Murnane 1992). The proxy used here i s general experience gauged by the age o f the worker. We use two variables, age and age squared, to take into account possible non-linearities. We investigate two questions: (1) is experience important to explain wages? and (2) are returns to experience homogeneous across the population? According to the results presented in Table 3.5 the answer i s yes to the first question and no to the second one. The experience (age) coefficient i s significantly different from zero and positive for all five quantiles, controlling for other individual characteristics. The impact o f experience on wages i s positive and increases until workers reach 49 years o f age. Thereafter, the returns fall in all quantiles. One explanation may be that older workers adapt less easily than younger ones to new technologies or they are simply less productive because o f their age. Returns to experience tend to fall as we move up the wage distribution, butthe variation i s not large. Labor market association Workers in the informal sector obtain a significantly lower pay after controlling for other variables. The negative impact o f informality increases across the wage distribution; a worker in the lothquantile has an 11percent wage discount because o f informality, whereas a worker inthe 75thquantile has a discount o f 16 percent. The informal sector generally provides lower quality jobs than the formal one. Since higher quality jobs may require more skills, the informal sector variable may be capturing skill differences not signaled by other variables included in the regression. The wage gap may also be due to lower productivity inthe informal sector relative to the formal one not captured by education and experience. Labor status The labor status o f workers is included as another determinant o f wages, taking "other workers" as the reference group. Coefficients for all occupational groups included are statistically significant at the 99 or 95 percent level (except contract workers and formal salaried workers at the 90thquantile). Looking at the median o f the distribution, employers obtain the highestreturn: 66 percent. For the 90thquantile, the premiumgap i s even larger: 103 percent. The self-employed, informal salaried and contract workers are systematically worse off than "other workers", particularly the self-employed. It i s interesting that in the case o f self-employed and contract workers the negative gap decreases sharply as we move up the wage distribution. Thus, richer self-employed and contract workers are not as penalized with respect to "other workers" as poorer ones. The opposite i s the case with informal salaried workers. 184 Gender Discrimination takes place when otherwise identical persons are treated differently by virtue of personal characteristics such as ethnicity or gender. Estimating economic discrimination i s difficult. Worker productivity i s seldom observed directly, so other variables must be used to proxy for the relevant productivity characteristics. The crucial problem i s to assess (1) whether relevant omitted characteristics differ according to ethnicity or gender, and (2) whether included characteristics directly capture productivity differences or are instead proxies for ethnicity or gender. The following section reports findings on gender differences. In chapter 7 we report results from Ramirez and Garcia (2004) and de Janvry and Sadoulet (2001) related to ethnic discrimination. The analysis includes four gender variables: married women with and without children, and single women with and without children. The reference group i s male workers. Regression results show large inequalities between men and women. Female wages are significantly differentfrom male wages at all quantiles. Results also suggest that the gender gap i s homogeneous across quantiles for women without children (married and single), but heterogeneous across quantiles for women with children (married and single). Married women with children experience the largest wage gap at the low end o f the distribution; they obtain 36 percent lower wages than their male peers in the lothquantile, with the gap narrowing along the distribution to reach 28 percent at the 90thquantile. The gender gap may be explained to some degree by choice ofjobs by women. Women are more likely to select more flexible jobs. They may choose, for example, part timejobs or jobs with lower working hours than men'''. A second factor may be gender differences in unmeasured skills. Education levels are taken into account inthe regressions, butthe skill level o f some women may be lower than that o f men for some jobs, and they may also be under- capitalized interms o f actualjob experience. Direct discrimination may hence be less strong than it appears from the results inTable 3.5. DisperseAreas and Regional Effects Data inTable 3.5 refers to rural areas definedas settlements o f less that 15,000 residents. We have included inthe regression a dummy variable for settlements o f less than 2,500 residents to see if dwellers o f disperse rural areas experience significant differences in wages compared to those in semi-urban settlements. We have also included dummies for the various geographical regions, taking as referencethe Centro region. Regression results show that workers in disperse rural areas are paid significantly less than workers in semi-urban rural areas, after controlling for other factors. The semi-urban - disperse rural wage gap i s significantly different from zero for all quantiles and varies across the wage distribution. It increases from the lothto the 50thquantile and declines from the 50thto the 90th, controlling for other covariates. The semi-urban premium i s 12 percent for the median worker. One possible explanation for this gap i s that prices, for example that o f urban land, are higherinsemi-urban areas, andhence the higherwage is a compensationfor this -a reflection of the fact that semi-urban workers have a labor supply curve above that o f disperse rural workers. Another possible explanation i s that work opportunities, i.e. labor demand, are higher in semi- urban areas, pushingupwages. On average Mexicanfemale workers work 24 percent fewer hours than their male peers. See annex Table 3.B.1 185 All regionswith the exceptionof the Golfo and Sur enjoy awage premiumwith respect to the Centro, and this is consistentacross the whole distribution (except for the Sur). The Norte, Capital and Pacijico regions have the highest premiums.Workers in the Sur have an advantage over their peers in the Centro (but not over those in other regions) in the bottom part of the distribution, which they soon lose as we moveupto higher quantiles. 186 : t UC { --t < c 0 + i t1 i : t UC { < c -t a + i t -9 . fa ..i2 : i C E T FC m u m w w " Z Z " ? 0 O N 2 U 9 i 2 i u ". N i ci $ ? Ici i 2 ICI i r j ". M o! i 2 i b eB : t UC { --t < c 0 + i t1 i : t UC { < c -t a + i t - 2 ° C N ? - c a Li W --.-ikE tE i 6 k c i a L i W tE i -.-ik a k c d z - - " z " 0 0 0 0 zc6 I ; c c c I S fc1 I a c c C .- .- I c c c1 C c c c c c .-2c 5 c c1 p: .-S .-SCB p: ?: S c1 a V ki c9 I 2 s V i2 c S a c CI $ a a..OI I ic c (I *; CI z4 cC 0 ET ANNEX 3.H. CORRELATES OF RURAL NON-FARM EMPLOYMENT Several studies have tried to explain the participation o f households in RNF activities using different data bases. We present an exercise carried out on the basis o f the ENE 2003, 2"d quarter survey, using a Probit model to determine the probability o f individual involvement in non-farm activities as primary occupation, conditional on a range o f personal, household and geographical characteristics. The specification o f the model draws on findings from previous sections, which suggest that the choice of primary occupation i s affected by geographical location, education, gender, age, and labor status. We use a wide definition o f rural in the regression, and include a dummy to check the impact o f a more restrictive definition. Results are shown in Table 3.H. 1. Rather than reporting parameter estimates, which are difficult to interpret, Table 3.H.1 presents the marginal effects associated with each explanatory variable, which measure the effect o f a percentage change in the explanatory variable on the probability of involvement in non-farm activities, taking all other variables at their means."' The table presents three Probit regressions linking the probability o f having primary employment in non-agricultural wage-labor occupations to a range o f explanatory variables. The dependent variable takes the value o f 1 if the person i s primarily employed in non-agricultural wage labor and zero otherwise. Included in the analyses are household size, age, age squared, gender, labor status, schooling variables, residence in settlements with less than 2,500 people, and regional dummies. The first model comprises all non-farm activities in rural Mexico. The second and third models have the same specification o f regressors but split workers in the non-agricultural labor force into two groups: those with low-retum jobs and those with high- return jobs. As in the previous section, we examine in turn regression results for the different variables, comparing themwith results from other studies. Gender Considering all non-farm employment together, women have considerable higher probability than men to participate in RNF activities, controlling for all other variables"'. This result holds for married and single women, with and without children, with marginal effects that are not very different among these groups o f women. This i s consistent with the results obtained by Araujo (2003: lSt essay) for Mexican rural poor communities using the 1997 ENCASEH data base, but differs from results for other countries, for example for Northeast Brazil, where women are less likely to be represented in the RNF sector (see Ferreira and Lanjouw 2001). Inthe ejido sector, married and old women were found to be less likely to participate in RNF activities, with proximity to urban centers increasingthe probability ofwomen participation inthese activities, particularly that o f younger women (de Janvry and Sadoulet, 2001). The picture becomes more complex when we consider high and low return activities, because women are significantly more likely than men to participate in low return occupations but significantly less likely to participate inhighreturn ones, andthis does not change with marital status or having or not children. For dummy variables, the marginal effect is the change inthe dependent variable associated with a move from a value of zero to avalue of one of the dummy, holding all other variables constant at mean values. The reference variable for gender comparisons is single male. 195 Table 3.H.1. Probability of Being Employed in the Non-agricultural Sector, RuralMexico 2002l Low-Productivity'Non- High-ProductivityNon- DependentVariable Non-Agricultural Employment Agricultural Employment Agricultural Employment Worker Characteristics dF/dx SE PZlzl DF/dx SE PZlzl dF/dx SE PZlzl Age 0.002 ** 0.001 0.03 -0.008 *** 0.001 0 0.012 *** 0.001 0 Age Square 0 *** 0 0 0 *** 0 0 0 *** 0 0 Marriedwoman w/o children+ 0.275 *** 0.006 0 0.498 *** 0.019 0 -0.141 *** 0.014 0 Marriedwoman with children+ 0.314 *** 0.004 0 0.511 *** 0.006 0 -0.154 *** 0.005 0 Single woman w/o children+ 0.281 *** 0.004 0 0.487 *** 0.008 0 -0.161 *** 0.005 0 Single woman with children+ 0.287 *** 0.004 0 0.534 *** 0.008 0 -0.161 *** 0.005 0 Labor Status Employer+ 0.056 *** 0.016 0.001 0.405 *** 0.018 0 -0.151 *** 0.007 0 Self-employed+ -0.146 *** 0.013 0 0.405 *** 0.014 0 -0.326 *** 0.006 0 Informal Salaried+ -0.056 *** 0.012 0 0.319 *** 0.014 0 -0.198 *** 0.006 0 Formal Salaried+ 0.181 *** 0.01 0 0.263 *** 0.015 0 -0.071 *** 0.008 0 Contract+ 0.093 *** 0.013 0 0.328 *** 0.017 0 -0.1 *** 0.009 0 Family Worker+ -0.464 *** 0.013 0 0.216 *** 0.016 0 -0.322 *** 0.003 0 Other Education N o education Primary Complete+ 0.114 *** 0.006 0 0.06 *** 0.007 0 0.082 *** 0.006 0 Lower Secondary Complete+ 0.196 *** 0.006 0 0.09 *** 0.008 0 0.144 *** 0.008 0 Upper Secondary Complete+ 0.228 *** 0.007 0 0.071 *** 0.012 0 0.223 *** 0.013 0 University Complete+ 0.278 *** 0.005 0 -0.214 *** 0.007 0 0.537 *** 0.012 0 Technical Education+ 0.238 *** 0.008 0 -0.093 *** 0.011 0 0.367 *** 0.014 0 Region Norte + 0.019 ** 0.009 0.042 0.031 *** 0.009 0.001 0.002 0.008 0.769 Capital+ 0.175 *** 0.009 0 0.061 *** 0.012 0 0.091 *** 0.012 0 Golfo + 0.015 * 0.008 0.056 0.027 *** 0.008 0 -0.021 *** 0.007 0.003 Pacifica+ -0.027 *** 0.009 0.002 0.019 ** 0.008 0.025 -0.023 *** 0.007 0.002 Sur+ -0.019 ** 0.009 0.03 -0.026 *** 0.008 0.001 -0.009 0.008 0.265 Centro-Norte+ 0.033 *** 0.008 0 0.014 * 0.008 0.074 0.022 *** 0.008 0.002 Locality < 2,500 inhabitants+ -0.237 *** 0.005 0 -0.152 *** 0.005 0 -0.088 *** 0.005 0 obs. P 0.629 0.332 0.297 Pred. P (at x-bar) 0.707 0.302 0.229 Number o f observations = 46501 46501 46501 Log Likelihood= -20843.1 -23166.4 -20634.5 LRchi2(24)= 19671.5 12766.4 15278.6 Prob>chi2= 0 0 0 Pseudo R2= 0.321 0.216 0.27 Source: ENE2003,2nd quarter 'Rural area defined as localities with less than 15,000 inhabitants The worker i s employed ina low-productivity non-agriculturaljob ifher monthly labor income i s below the average non-agricultural labor income. The worker i s employed ina low-productivity agriculturaljob ifher monthly labor income i s below the average agricultural labor income. 196 The probability o f employment in non-agricultural jobs rises with age, after controlling for other characteristics, although the marginal effect i s small. The association i s somewhat larger for high return occupations and i s negative for low return ones. N o evidence was found of the association declining at a certain age. This finding contrasts with that for the BrazilianNortheast obtained by Ferreira and Lanjouw (2001), also with the results obtained for the ejido sector by Winters, Davis and Corral (2002) and de Janvry and Sadoulet (2001), and with the results for the rural poor communities covered in the ENCASEH survey, examined by Araujo (2003: lSt All these studies found that young individuals, particularly young men, tend to essay). participate more than older ones in RNF activities. It i s not clear why this difference in results, but it may be due to the wider coverage ofthe ENEsurvey, andthe wide definition ofruralusedinour regressions. Education The effect o f education is strong, and the results are consistent with findings from other studies, like those o f Yunez-Naude and Taylor (2000), Ferreira and Lanjouw (2001), Winters, Davis, and Corral (2002), de Janvry and Sadoulet (2001), Araujo (2003: lSt and Taylor, Yunez-Naude and Ceron (2004). Table 3.H.1 essay), shows that the probability o f involvement in the non-farm sector i s positively and significantly related to education levels. Relative to the non educated, workers with education are generally more likely to find employment inthe non-agricultural sector. As education levels rise, so does the probability o f being employed in the non-agricultural sector both inlow return and highreturn occupations. The exception is university and technical education, which, not surprisingly, diminish the probability o f engagement in low return RNF activities. It should be acknowledged that the exogeneity o f education in these models can be questioned, so more research i s neededto understandemployment possibilities inRNF sectors. We did not combine in our regressions gender and education variables, but it has been done in other studies with interesting results. Thus, Winters, Davis and Corral (2002) find from income regression equations for the ejido sector that increasing levels o f education translate into higher income for rural women at higher levels o f education only, probably because at lower levels o f education low paid work in domestic help i s the only or the most frequent opportunity available to women. Location Location influences the probability o f participation in the RNF sector. Workers living in localities smaller than 2,500 inhabitants are less likely to be employed in RNF occupations than those living in localities under 15,000 residents. The regression model for low-productive non-agricultural occupations reveals that workers insmall localities are even less likely to be employed inthis sector. Other studies confirm the importance o f location. Araujo (2003: 2"d essay) finds that proximity to urban centers increases the probability o f participation in manufacturing activities, irrespectively o f education, ethnicity, wage levels or initial employment. Participation in services i s less related to proximity to urban centers andmore to the characteristics o fthe area. Region Relative to those living inthe Centro, which i s taken as reference, workers inthe Norte, Capital, Golfo, and Centro-Norte regions are more likely to be employed inRNF activities. Instead, workers inthe Paczjico and Sur are less likely to participate in the RNF economy than their peers in the Centro. D e Janvry and Sadoulet (2001) found also important differences between regions, with dwellers in ejidos located in the Sur having fewer opportunities to work inRNF activities. 197 Other Variables Other variables not included in our regressions because o f lack o f data in the ENE survey may also be important determinants o f labor market participation, and have been included in other studies. One o f them i s access to land. Finan, Sadoulet and de Janvry (2002) find that young educated men from land scarce households inMexico are more likely to participate in off-farm non-agricultural employment. Araujo (2003: lSt essay) also found that access to irrigated landhas a negative impact on RNF employment. Araujo (2003: lSt essay) measures the role o f social networks on labor market behavior in rural Mexico. She finds that neighbors' participation in off-farm non-agricultural employment has a significant impact on the individual choice o f occupation, even after controlling for the availability o f opportunities. The role o f employment choices by neighbors i s more important for groups that are less likely to participate innon- agricultural rural employment such as women, indigenous people, the elder, and land-owners. This finding suggests an important role for networks and referrals in the job-search process o f rural households, specially since it appears that social networks have an equalizing effect, compensating more to those who are less endowed and therefore less likely to participate in off-farm non-agricultural employment. Ethnicity is another important variable. Yunez-Naude and Taylor (2001) found from a sample of 391 household in eight rural locations in four states o f Mexico that indigenous workers are less likely than non- indigenous ones to participate in RNF activities, and more likely to participate in staple and cash crop production, wage employment and national migration. A similar conclusion was reached by de Janvry and Sadoulet (2001) for young indigenous workers in the ejido sector at low levels o f education, although the difference was not significant at high levels o f education. Araujo (2003: lSt also found that indigenous essay) workers tendto participate less inRNF activities. Infrastructureand connectednesshave also been found to be important to promote RNFparticipation by Winters, Davis and Corral (2002) and Araujo (2003: 2"d and 3rdessays). Araujo finds that interventions in roads and secondary education are effective in reducing poverty through non-farm rural employment in rural municipalities. 198 ANNEX 3.1. THE TERRITORIAL APPROACH TO RURAL DEVELOPMENT^l2 The notion o f Rural Territorial Development (RTD) is parallel to that o f local economic development, with the difference that the latter concept is traditionally used more for urban areas and urban-industrial activities while the former i s more rural, although RTD advocates insist in the links between rural an urban activities and the role inrural development o f intermediate and ruraltowns. RTD should be seen as an approach and a method to promote rural development with relevant policy implications rather than as a theory about development. The main tenets o f the RTD approach can be summarized as follows: (1) a view o f rural development that consists o f a combination o f productive transformation and institutional change; (2) a widened concept o f the rural space to include small rural towns and the links with intermediate cities; (3) a multisectoral approach to economic development covering different economic sectors and including farm and non-far activities; (4) a recognition o f the differences among rural territories and the need to tailor productive investments and other interventions to their diverse characteristics and needs; (5) a concept o f the territory that presupposes some territorial identity and the possibility o f building a collective project o f local actors for the development o f the territory; (6) participatory planning as a means o f economic coordination andprioritization o f investmentsinthe territory; (7) conscious involvement o f different local actors (public, private and civil society) in the economic coordination process, and alliances between these actors; (8) emphasis on territorial competitiveness and on maximum economic use o f territorial assets; (9) search of economic synergies and clustering o f activities around development axes to achieve critical economic masses; (10) construction o f an institutional architecture to facilitate economic coordination processes; and (11) a medium- and long-term development horizon `I3.The intellectual background from economic theory and the social and political sciences o f the RTD concept i s summarized inbox below. The central purpose o f RTD is to facilitate endogenous growth processes, centered on the capacity o f local agents to promote territorial development on the basis o f exiting resources. Elements favorable to this are the presence o f valuable -and marketable- territorial assets (natural resources, landscapes, culture and traditions, accumulated local knowledge and know-how in certain areas, etc.), strong institutional development and social capital, and the potential to generate growth process around territorial economic linkages. Return migrations have also been identifiedas a factor favoring local growth processes. A crucial question is that of the extent to which public policies can generate this type of growth dynamics. Each territory that has gone through a successfbl endogenous growth process has done it in a differentway, based on factors that in some sense are always unique. The process may have been triggeredby the presence o f strong leadership from a local agent, or the finding o f a good market niche for a local product, or the transformation o f what may have been an economic obstacle, like a mountainous environment, into an economic opportunity, e.g. for tourist development, or the introduction o f new economic activities by returning migrants, or in many other ways. Policy action cannot replace these triggers, and in this sense it cannot "pick winners", butit can help create favorable conditions so that development arises whenever the triggering factors appear, and can stimulate local actors to search for those triggers. This will normally be done via the supply o f public goods like infrastructure, education, training, research, extension, etc., and also by creating a favorable institutional framework, acting to overcome market failures (e.g. in the financial market), and providing targeted assistancewhen required. 112This annex is based on material inWorld Bank (2004~). 113These tenets are very similar althoughno identicalto the TRD criteria proposed by Schejtman and Berdeguk (2003). 199 IntellectualBackgroundof the Territorial Approachto RuralDevelopment From the perspective o f economic theory three influences can be traced in the RTD notion. The first i s the classical theory of location and territorial specialization of von Thunen, Weber, Christaller, Isard and others, and the recent revival o f economic geography partly under the authority o f Paul Krugman. The second influence i s that o f the theorists o f territorial competitiveness, with their view o f competition as a systemic phenomenon going beyond single markets, their insistence on economies external to the firm but internal to the territory, and their theory of "knowledge environments". Relevant here is the literature on industrial districts initiated byAlfred Marshall, which has flourished around the Italian experience (e.g. Bagnasco, Trigilia, Putnam), the contribution o f Michael Porter on territorial competitiveness and clustering, and the discussions on "knowledge environments" following the Silicon Valley experience. The third strand is the theory of economic coordination originated in the work o f the classical development economists. Rosenstein Rodan's famous paper on the industrialization o f Eastern and Southern- Eastern European countries is the piece most frequently quoted, but the exposition o f Tibor Scitovsky in his " Two Concepts o f External Economies" is theoretically more compact. The theory has revivedwith the flourishing o f information economics and game theory, with notable contributions from Karla Hoff, Joseph Stiglitz and others. The crucial idea is that economic agents do not just take information from the price system but also interact outside that system. For investment decisions, in particular, the use o f non-price information on the investment plans o f different actors will normally lead to superior equilibria than if only price information were used. Economic coordination refers to the exchange of information and resulting agreements by different economic agents to carry out complementary actions that are mutually beneficial. Participatory planning, for instance, i s an instrumentfor doing this. From the perspective o f the social and political sciences, we can identify another three strands o f influence. The first are the discussions on decentralization, the developmental role o f local governments, and the alliances and synergies between different levels o f government. The second influence is that o f the theorists o f the role o f public and private actors inthe development process, with the emphasis on the role o f collective action and the "third actor" (for instance by Elinor Orstrom), and on the potential for public-private synergies in the development process (for instance by Judith Tendler and Peter Evans). The third strand o f influence i s that o f the theorists o f participation and empowerment like Robert Chambers and John Friedmann, and o f the theory and experience o f community-dnven development. The concept of territory is not an easy one. Various issues are relevant, like its size, the criteria for its delimitation, and its ideological dimension, i.e. the way inwhich it i s present inthe imagination o f its actors. Most o f the tenets o f RTD presented above apply to territories o f different sizes, from small micro- basins or municipalareas to entire sub-regions, basins or provinces. The approach can hence be used at different levels in the rural space. Economic coordination and the clustering o f activities to achieve critical economic masses would normally require, however, a certain size, which cannot be too small. This i s why RTD requires strong meso-level economic coordination institutions that would normally go beyond a single municipality or micro-basin. The district level singled out inthe Ley de Desarrollo Rural Susutentablemay be such level. The demarcation ofterritories is a complex issue. There are inprinciple three ways inwhich this can be done: (1) in a top-down technocratic way, using ex-ante technical criteria o f homogeneity, complementarity, water basins, market basins, and the like; (2) in an administrative way, usingexisting administrative boundaries (munic@ios, distritos, provincias); and (3) in a bottom-up way, letting local actors define their own territories, as inthe EuropeanLeader experience. The relevance of the ideological dimension of territories is larger than generally acknowledged. Local populations and institutional actors usually identify themselves with a territory. That identification i s important for people's identity and i s hence an element o f human development in the Sen's sense o f development as freedom to be who you can be. But territorial identity i s also an economic asset, inasmuch as it i s a source o f social capital, a condition that facilitates economic coordination, and itself a marketable asset if territorial 200 products can be favorably differentiated in the market. RTD i s about using identity for the construction o f a development project o f the territory that can catch the imagination o f local actors and produce outcomes in which they can recognize themselves. This i s at the core o f the "development with identity" concept, which can beparticularly relevant for indigenous areas (see WorldBank, 2004d). Keyto RTD are territorial planning and the investment decision process. As indicatedbefore, territorial planning i s an instrument o f economic coordination, and should be carried out with a strategic perspective and in a way that involves all relevant local actors inthe decision making process. It is a means to identify local potentials and constraints, and to locally "pick winners" inthe sense o f identifyingthe strategic axes considered promissory by local actors around which they propose to cluster investments. It i s also a way o f organizing the local demand for development assistance. Participatory territorial planning i s a system o f compromises among the interests o f local actors arrived at in confrontation between spontaneously perceived opportunities and needs, and technical, market and cost criteria. Cost sharing in the investmentsby the beneficiaries, an ex-ante budgetary restriction, and effective decision-making capacity o f the participatory planning body, are essential elements to give economic meaning to the entire exercise and arrive at trade-offs between alternative investment options. This type o f planning i s hence the opposite o f the simple collection and listing o f local spontaneous investmentrequests that i s common inmany community-driven development programs. The final element o f RTDto be mentioned is that o f the new institutional architecture. We can include here the creation and/or strengtheningo f meso-level economic coordination entities, the strengtheningo f local civil society organizations, and the strengthening o f public-private synergies. Among the economic coordination entities we should include territorial institutions, and sector coordination entities like activity- based producer organizations and value chain organizations. Within civil society organizations we have to include membership and service organizations, and in the strengthening o f public-private synergies we should include central government programs and local government actors. 201 . . . iii v; % a U i c c 0 0 0 m 0 z 0 0 M Ici 0 c c Ici c 2r- 2 0 M 0 0- W U Ici 0 U i 0 c c o! c 0 m W w 0 0 0 i 0 r 9 c c 0 W Ici 0 0 0 9 i c c c 2r- 0 0 0 W U U U r 0 1 0 0 0 U ? U U r 0 m m 0 0 0 U r U MW U r 0 0 0 0 \c U w U r 0 M 0 0 0 r- U U N Ici U r 0 0 0 0 W m 8 sk 1 p: 4cl m a8* 8 s s2 0 s 2 ss 2 0 8 0 0 m p: E 8 Eu U c c c c c c c c r c c c c c U U U r U U U r r U U r \c U U r U U U r U c U c c c h c c v c c c h c c h c c c h r r c c c c h c c c c c c h U U 0 U 0 r 0 r U U o( U 0 r 0 r r r U U r 0 0 r \c U \c U 0 r 0 r U U 0 U 9 P ANNEX 4.B. SMALL FARMEFFICIENCY ANALYSIS Using ENHRUM data, we have carried out an exercise to determine the economic efficiency o f small farmers. We do this in various steps. The first step consists in fitting a production fbnction to 666 crop farms using a stochastic frontier analysis method. The econometric basis o f the method i s explained in Appendix 1 o f this Annex. The core o f the method i s to separate intwo the residuals o f the production fbnction regressionequation. The first residual i s the ordinary stochastic one with zero mean and constant variance, which accounts for non included variables and measurement errors. The second i s an efficiency residual which results from differences in efficiency among farmers. The method allows to tests if farmers are efficient, which farmers deviate from the efficiency frontier, and to what extent. The second step consists o f regressing the efficiency residual against a set o f explanatory variables to see if they helpexplaining inefficiency. A Cobb-Douglas production fbnction was cho~en"~.The dependent variable is the gross value o f crop output o f farms (q), and the explanatory variables are: variable capital (kvar), which includes seed, fertilizer and chemicals, fixed capital (kfix), which includes the value o f machinery and draft animal services, labor (labor), which includes all labor usedduringthe entire production process, and land (sland), which enters inthe equation in a standardized form. Land standardization i s necessary because measuring land in crude hectares implies that very different qualities of land (irrigated and not irrigated, with and without good access, steep and flat, etc.) all count equally. It i s appropriate to give different weights to different types o f land. The standardization process carried out i s explained in the Appendix 2 to this Annex. It basically consisted in estimating the implicit or hedonic prices o f relevant land characteristics included in the survey by regressing these characteristics on the value o f land declared by the farmer'l5. Using the coefficients o f the regression equation we estimate the value per hectare o f land for each farm'l6. Comparing this value with the average value per hectare o f all farms declared by farmers, we obtain a weight for the land o fthe particular farm. We useda logarithmic specification o fthe Cobb-Douglas fbnction as follows: lnqi = bo+bl lnkvari+b2lnkfixi +b3lnlabori +b4lnsland+ ei We first run the regression for all 666 farmers in the sample. All explanatory variables turned out to be significant at 95 percent level. The values o f the coefficients, which indicate the elasticity o f crop output to each o f the factors in the production fbnction, are shown in Table 4.B.1117. The first thingto benoticed i s that variable capital i s the main element explaining output, with a big difference to the others factors, followed by land. A one percent increase in variable capital increases output by 0.43 percent. Labor has little weight, signaling the probable presence o f surplus labor in many farms. The sum o f elasticities i s 0.913, i.e. less than one, which would indicate the presence o f diseconomies o f scale. The confidence intervals o f the coefficients are 114A translog specification was also tried but did not perform as well. 115 Regression results are presented in Appendix 2. Irrigation, accessibility, and regional location are the characteristics that influence most landvalues. Location inthe Centro region increased muchthe probability of higher land value, and the opposite was the case for location inthe Sur-Sureste. 116Inthe regressionequation we included thisestimated value, notthenumber of standardized hectares. 117Complete results with relevant tests are given inannex Table 4.B.7. 206 sufficiently large, however, not to reject the null hypotheses that the sum o f the coefficients i s one. Hence, there i s no statistical evidence o f diseconomies o f scale, but there i s little likelihood o f economies o f scale for the entire sample"'. Table 4.B.1 Value ofElasticities inthe ProductionFunction RegressionEquationfor allENHRUM Sample ofCropFarms Parameter Variable Coefficient S.E. P>z bi Variable 0.43 0.035 0 capital bz Fixed 0.159 0.035 0 Capital b3 labor 0.093 0.381 0.02 b4 sland 0.232 0.03 0 bo constant 3.259 0.367 0 Source: calculated from ENHRUM. The values of the elasticities of the productive factors are influenced by imperfections in factor markets. This seem to be particularly the case with variable capital whose high elasticity seems to be the result o f the underutilization o f fertilizer and chemicals due to the lack o f access o f small farmers to seasonal credit (see chapter 5 on this). Lack o f credit to buy inputs prevents farmers from using them optimally, i.e. up to the point when the marginal contribution to production equals the cost to the farmer. The test for inefficiency rejects the nullhypothesis that there is no inefficiency. When we group the inefficiency residuals by categories o f farms we can see which types o f farms are less efficient. We show this in Table 4.B.2, where the average inefficiency o f each category o f farms i s measured as the distance to the efficiency frontier. Thus, an average residual o f 0.94 means that that category o f farmers would need to increase output, with existing inputs, by 94 percent to reach the efficiency frontier. Ilx Notice that economies of scale inthis context do not refer to increased farm size, which is the common meaning in usual parlance, but to the simultaneous increase of ALL factors in the production equation, whichis the technical meaning. 207 Table 4.B.2. Distributionofthe InefficiencyError Termby Category ofFarms No. of Average Variables farms Inefficiency S.D. Sur-Sureste Region 211 0.94 0.3809 Centro Region 233 1.03 0.398 Centro-Occidente Region 122 0.78 0.2391 Noroeste Region 36 0.69 0.4104 Noreste Regions 64 0.73 0.4012 Maize and Beans farmers 456 0.98 0.3845 Coffee Farmers 43 0.65 0.2273 Vegetable Farmers 33 0.69 0.2467 Perennial Crops Farmers 79 0.75 0.3419 Oilseeds and Other Grain Far. 55 0.86 0.4288 Farmerswith natural shocks 295 1.05 0.4481 Farmerswithout natural shocks 371 0.8 0.2862 All Farmers 666 0.91 0.387 Source: calculated from ENHRUM. An important result is that maize and beans farmers, farmers in the Sur-Sureste and Centro regions, and farmers that experienced natural shocks are the least efficient. Producers o f coffee, other perennial crops and vegetables, and producers in the Noreste and Noroeste are the most efficient. At the national level an effort would be required to increase production by 91 percent with existing factors to reach efficiency, i.e. production would almost need to double. One surprising result is the large number o f farmers that suffered from natural disasters"'. We discuss this more inchapter 7 butwe mustnotice here that this i s a frequentlyoverlooked element with important repercussions on efficiency. Nextwe partitioned the sample andrepeated separately the stochastic production fbnction exercise for maize and beans farmers, farmers that experienced shocks, and farmers that did not experience shocks. We also carried out regression exercises for these categories o f farmers to try to explain the factors causing inefficiency12'. 119This was defined as farmers who reported having suffered from rains, hurricanes, droughts, frosts or pests and diseases, and whose output was less than 50 percent that of a good year. l20We only carried out a separate analysis for these categories of farmers because sample sizes were too small for the others. 208 Table 4.B.3 ProductionFunctionElasticities for DifferentCrop Farmersinthe ENHRUMSample Elasticities Variables All Maize With Without Farms and Shocks Shocks Beans Kvar 0.43 0.443 0.399 0.391 Kfix 0.159 0.265 0.273 0.183 Labor 0.093 0.107 0.276 0.207 Sland 0.232 0.132 0.171 0.249 Sum Elasticities 914 0.947 1.119 1.03 Constant 3.259 3.669 2.658 2.226 Source: calculated from ENHRUM.All coefficients are significant at 95% level. There are substantial changes inthe elasticities o fproduction factors for differenttypes o f producers, although variable capital remains always the highest. The comparison i s shown in Table 4.B.3. For maize an beans producers the importance o f fixed capital i s higher than for the entire sample o f farmers, while that o f labor remains low. This indicates that more use o f animal power and/or tractor services for these producers would have a large effect on output. Contrarily, reducing the amount o f labor put in these crops would not have a large impact on output. Land elasticity inmaize and beans production i s surprising low, less than half o f that for all farms and for farms without natural shocks. Increasing output inmaize and beans production depends hence more on improvedtechnology embodied invariable and fixed capital than on increasing the area. This is good news for small peasant farmers who are the main producers o f maize and beans, for itmeans that they couldboost output intheir small farms ifthey had access to better technology. Underpresent conditions, however, shiftingland fi-om maize andbeansto other crops would raise total output. The sum o f elasticities grows when we partition the sample, which suggests that for some categories o f farmers economies o f scales are more o f a possibility than for farmers ingeneral. In the case o f farmers who experienced natural shocks, labor i s much more relevant than for all farmers in general, and there i s no clear explanation for this. Instead, for farmers who did not experience shocks, the importance o f land i s the highest. This group o f farmers represents "normal", i.e. natural shocks fi-ee, farming conditions'21, and their elasticity coefficients are illustrative o f this situation. Hence, under "normal" conditions land i s more relevant than under "abnormal" ones. The presence o f a large proportion o f maize and beans farmers in the sample, nearly 70 percent, decreases the elasticity o f land since, as we have seen, the elasticity o f land i s low for these farmers. In farms, therefore, producing other crops under "normal" conditions land mustbemuch more important. An interesting result for the "farmers without shocks" sub-sample i s that the inefficiency test failed to reject the hypothesis o f no inefficiency. This does not necessarily mean that these producers are all efficient, but points to the strong link between "normality" inproduction conditions and farming efficiency. Our econometric analysis to explain inmore detail the causes o f inefficiency gave modest results (Tables $.B.7 to 4.B.14). Many o f the explanatory variables included, like gender, age, education, existence o f services in the community (measured by a services index), land tenure, farm size, and government transfers, were not statistically significant, although the sign o f the 121 But we can see in Table 4.11 that this "normal" condition is not so normal: 295 out of 666 farmers, 44 percent, experiencednatural shocks. 209 coefficients was generally the expected one. The reason seems to be that these variables influence more the choice o f technology, i.e. the combination o f inputs in the production fbnction, than the efficiency o f production given a combination o f inputs, which i s what we investigate by regressing the efficiency residuals on these variables. More econometric work remains hence to be done, tryingto include these variables directly inthe production fbnction. Table 4.B.4 Frequency ofDummyVariables Variables Dummy Frecuencia Porcentaje Sex0 deljefe del hogar (mujer) 60 9 Jefe de familia habla algunalengua indigena 225 33.8 Existencia de problemas con el cultivo 295 44.3 Existencia de organizacionesagricolas 124 18.6 Tipo de tenencia de latierra (privada) 235 35.3 Tipo de tenencia de latierra (ejidal y comunal) 400 60.1 Tipo de tenencia de latierra (mixta) 31 4.7 Si e l hogar recibe transferenciasde gobierno 482 72.4 Pequeiio, menos de 5 ha cultivadas 411 61.7 Grande. con 5 ha o mas cultivadas 255 38.3 Source: Enhrum, 2003. Table 4.B.5 Crop Specialization Cultivo Frecuencia Porcentaje Porcentaje Maiz y frijol 456 68.5 68.5 Oleaginosasy otros granos 55 8.3 76.7 Cafe 43 6.5 83.2 Hortalizas 33 5 88.1 Perennes 79 11.9 100 Source: Enhrum. 2003. Table 4.B.6. Summary Statistics ofVariables Variable Obs. Media Desviacion Min Max 4 ($1 666 29,639.60 141,276.80 45 2,550,000.00 kvar($) 666 5,024.70 32,033.10 10 564,500.00 kfij(horas-tractor) 666 41.4 81.6 0.1 920 mo Cjornales) 666 114.9 157.3 1 1,559.00 sup (ha naturales) 666 4.9 7.7 0 78.50 supest ($, valor estandarizado) 666 209,030.80 813,229.80 185 12,800,000.00 restfinan ($, tCrminos netos) 666 597.7 3,523.70 -1,875.00 69,265.30 restfinanor @roporci6n delvbp agricola, 4) 666 3.3 33.8 -5.9 642.00 edad (aiios) 666 51.5 15 18 94.00 educ (aiios de escolaridad) 666 3.8 3.3 0 19.00 dispmo @roporci6n) 666 0.7 0.3 0 1.oo indserv @roporci6n de servicios en comunidad) 666 0.2 0.1 0 0.80 Source: Enhrum,2003. 210 Table 4.B.7 Stochastic Production Function. AllFarms Stoc. Frontier normalhalf-normalmodel Numberof obs 666 Wald chi2(4) 686.67 Log likelihood -1102.2665 Prob > chi2 0 lq Coef. Std. Err. z P>z [95% Conf. Interval] lkvar 0.4304 0.03544 12.14 0.000 0.36094 0.49986 lkfij 0.15869 0.03534 4.49 0.000 0.08942 0.22796 lmo 0.09303 0.03809 2.44 0.015 0.01838 0.16769 lsupersta 0.23155 0.03024 7.66 0.000 0.17227 0.29082 - cons 3.259 0.36657 8.89 0.000 2.54054 3.97745 llnsig2v 0.13042 0.16018 0.81 0.416 -0.18353 0.44437 llnsig2u 0.25885 0.3909 0.66 0.508 -0.5073 1.025 sigmav 1.06738 0.08549 0.91232 1.2488 sigmau 1.13818 0.22246 0.77596 1.66946 sigma2 2.43475 0.3565 1.73602 3.13348 lambda 1.06633 0.30044 0.47747 1.65518 Likelihood-ratio test of sigmau=O: chibar2(01)= 3.13 Prob>=chibar2= 0.039 211 Table 4.B.8 EfficiencvDeterminants.All Farms Regression with robust standard errors Numberof obs 666 F(23,599) 15.47 Prob >F 0 R-squared 0.3577 Root MSE 0.31515 RobustStd. inefic Coef. Err. t P>t [95% Conf. Interval] olegra 0.0267 0.0554 0.48 0.630 -0.0821 0.1356 caf -0.3722 0.0422 -8.82 0.000 -0.4551 -0.2893 hort -0.2 166 0.0462 -4.69 0.000 -0.3073 -0.1258 peren -0.1506 0.0375 -4.02 0.000 -0.2242 -0.0771 proble 0.1833 0.0267 6.87 0.000 0.1309 0.2357 restfinanor 0.0032 0.0006 5.66 0.000 0.0021 0.0044 indserv 0.1693 0.117 1.45 0.148 -0.0604 0.3989 organi 0.0703 0.0377 1.87 0.062 -0.0037 0.1443 dispmo -0.014 0.0453 -0.3 1 0.757 -0.103 0.0749 proguber 0.0108 0.0284 0.38 0.703 -0.0449 0.0665 mixta -0.0144 0.0646 -0.22 0.824 -0.1413 0.1125 priv -0.0096 0.0268 -0.36 0.720 -0.0622 0.043 gran -0.029 0.0286 -1.01 0.3 11 -0.0852 0.0272 sex0 0.0645 0.047 1.37 0.170 -0.027 0.1568 lengua 0.0005 0.0298 0.02 0.987 -0.0581 0.0591 educ -0.0009 0.0043 -0.22 0.828 -0.0094 0.0075 edad 0.0009 0.001 0.93 0.355 -0.001 0.0028 r l -0.1082 0.0356 -3.04 0.002 -0.178 -0.0383 r3 -0.2663 0.0343 -7.76 0.000 -0.3337 -0.1989 r4 -0.3338 0.0688 -4.85 0.000 -0.469 -0.1987 r5 -0.3281 0.0571 -5.74 0.000 -0.4403 -0.2159 cons 0.9156 0.0889 10.3 0.000 0.741 1.0902 212 Table 4.B.9 Stochastic ProductionFunction. Maize and BeanProducers Stoc. Frontier normalhalf-normalmodel Number of obs 435 Wald chi2(4) 555.76 Log likelihood -644.4 Prob > chi2 0.000 lq Coef. Std. Err. Z P>z [95% Conf. Interval] lkvar 0.4427 0.0396 11.18 0.000 0.3651 0.5204 lkfij 0.2647 0.0396 6.69 0.000 0.1871 0.3422 lmo 0.1074 0.0478 2.25 0.025 0.0137 0.2011 lsupersta 0.1321 0.0323 4.09 0.000 0.0688 0.1953 - cons 3.6686 0.3893 9.42 0.000 2.9056 4.43 15 llnsig2v -0.6157 0.1957 -3.15 0.002 -0.9993 -0.232 llnsig2u 0.5398 0.2047 2.64 0.008 0.1386 0.941 sigmav 0.735 0.0719 0.6067 0.8905 sigmau 1.3098 0.134 1.0718 1.6008 sigma2 2.2559 0.2793 1.7086 2.8033 lambda 1.782 0.1945 1.4008 2.1631 Table 4.B.10 EfficiencyDeterminants,Maizeand BeansProducers Regressionwith robust standarderrors Numberof ob 435 F (17,393) 10.3 Prob >F 0.000 R-squared 0.268 Root MSE 0.526 [95% Conf. inefmafri Coef. Std. Err. t P>t Interval1 indserv 0.157 0.200 0.790 0.432 -0.236 0.550 organi 0.122 0.071 1.730 0.085 -0.017 0.261 dispmo -0.179 0.098 -1.830 0.068 -0.370 0.013 proguber 0.007 0.062 0.120 0.905 -0.114 0.129 mixta -0.050 0.135 -0.370 0.711 -0.3 14 0.2 15 priv -0.058 0.057 -1.030 0.305 -0.170 0.053 gran -0.041 0.061 -0.670 0.500 -0.160 0.078 sex0 0.142 0.101 1.410 0.160 -0.056 0.340 lengua -0.044 0.058 -0.760 0.449 -0.158 0.070 educ -0.007 0.010 -0.720 0.474 -0.027 0.013 edad 0.000 0.002 0.070 0.947 -0.004 0.004 r l -0.100 0.069 -1.450 0.147 -0.235 0.035 r3 -0.386 0.073 -5.270 0.000 -0.530 -0.242 r4 -0.658 0.103 -6.400 0.000 -0.860 -0.456 r5 -0.252 0.142 -1.770 0.077 -0.532 0.027 -cons 1.151 0.186 6.180 0.000 0.785 1.518 213 Table4.B.11 Stochastic ProductionFunction, Producerswho ExperiencedShocks Stoc. Frontier normayhalf-normalmodel Numberof obs 278 Wald chi2(4) 322.750 Log likelihood 446.91 Prob > chi2 0.000 Std. [95% Conf. lq Coef. Err. Z P>z Interval] lkvar 0.399 0.056 7.150 0.000 0.290 0.509 lkfij 0.273 0.055 5.000 0.000 0.166 0.380 lmo 0.276 0.068 4.060 0.000 0.143 0.409 lsupersta 0.172 0.044 3.870 0.000 0.085 0.259 - cons 2.658 0.544 4.890 0.000 1.592 3.723 llnsig2v -0.175 0.233 0.750 0.453 -0.63 1 0.281 llnsig2u 0.563 0.336 1.670 0.094 -0.096 1.221 sigmav 0.916 0.107 0.729 1.151 sigmau 1.325 0.223 0.953 1.842 sigma2 2.595 0.448 1.716 3.473 lambda 1.446 0.315 0.829 2.063 214 Table 4.B.12 Efficiency Determinants, Producerswho ExperiencedShocks Stoc. Frontier normayhalf-normal model Number of ob 278 F (16,264) 7.620 Prob >F 0.000 R-squared 0.292 Root MSE 0.479 [95% Conf. inefconp Coef. Std. Err. t P>t Interval] olegra 0.188 0.152 1.230 0.219 -0.112 0.487 caf -0.604 0.090 -6.740 0.000 -0.780 -0.428 hort -0.167 0.108 -1.550 0.121 -0.379 0.045 peren -0.039 0.115 -0.340 0.732 -0.265 0.187 restfinanor 0.004 0.001 5.920 0.000 0.003 0.005 indserv 0.208 0.293 0.710 0.478 -0.368 0.784 organi 0.045 0.092 0.490 0.624 -0.136 0.226 dispmo -0.144 0.099 -1.450 0.147 -0.340 0.051 proguber 0.021 0.070 0.300 0.765 -0.116 0.158 mixta 0.046 0.120 0.380 0.702 -0.190 0.282 priv -0.011 0.074 -0.150 0.882 -0.157 0.135 gran 0.018 0.066 0.280 0.783 -0.112 0.149 sex0 0.068 0.109 0.620 0.537 -0.148 0.283 lengua 0.045 0.071 0.630 0.530 -0.095 0.184 educ 0.000 0.011 0.010 0.992 -0.021 0.022 edad 0.002 0.003 0.900 0.371 -0.003 0.007 r l -0.018 0.080 -0.230 0.819 -0.176 0.139 r3 -0.348 0.074 -4.680 0.000 -0.495 -0.202 r4 -0.444 0.116 -3.840 0.000 -0.672 -0.2 16 r5 -0.290 0.151 -1.930 0.055 -0.587 0.006 215 Table4.B.13 Stochastic ProductionFunction, Producerswho did not ExperienceShocks Stoc. Frontier normalhalf-normalmodel Number of ob 346 Wald chi2(4) 484.030 Log likelihood -526.305 Prob > chi2 0.000 lq Coef. Std. Err. Z P>z [95% Conf. Interval] lkvar 0.391 0.043 9.070 0.000 0.306 0.475 lkfij 0.183 0.043 4.250 0.000 0.099 0.268 lmo 0.207 0.057 3.600 0.000 0.094 0.319 lmo 0.207 0.057 3.600 0.000 0.094 0.319 lsupersta 0.249 0.035 7.020 0.000 0.179 0.3 18 - cons 2.226 0.915 2.430 0.015 0.434 4.019 llnsig2v 0.204 0.077 2.660 0.008 0.054 0.355 llnsig2u -8.228 125.030 -0.070 0.948 -253.282 236.826 sigmav 1.108 0.042 1.027 1.194 sigmau 0.016 1.022 0.000 2.670E+51 sigma2 1.227 0.096 1.039 1.414 lambda 0.015 1.028 -2.000 2.029 216 Table 4.B.14 EfficiencyDeterminants, Producerswho did not ExperienceShocks Numberof obs 346 F( 16,325) 11.81 Prob >F 0.00000 R-squared 0.38370 RootMSE 0.00007 inefsinprob Robust `Oef* Std. Err. t P>t [95% Conf. Interval] olegra -0.00001 0.00002 -0.25000 0.80600 -0.00004 0.00003 caf -0.00010 0.00001 -7.08000 0.00000 -0.00013 -0.00007 hort -0.00007 0.00002 -3.83000 0.00000 -0.00010 -0.00003 peren -0.00007 0.00001 -5.29000 0.00000 -0.00010 -0.00004 restfinanor 0.00000 0.00000 0.3 1000 0.75800 -0.00001 0.00001 indserv -0.00002 0.00002 -0.79000 0.43200 -0.00007 0.00003 organi 0.00003 0.00001 2.27000 0.02400 0.00000 0.00005 dispmo 0.00002 0.00002 1.39000 0.16600 -0.00001 0.00006 proguber 0.00000 0.00001 0.42000 0.67300 -0.00002 0.00002 mixta -0.00001 0.00003 -0.42000 0.67600 -0.00007 0.00005 priv -0.00001 0.00001 -0.59000 0.55700 -0.00002 0.00001 Tan -0.00001 0.00001 -1.27000 0.20700 -0.00003 0.00001 sex0 0.00001 0.00001 0.97000 0.33300 -0.00001 0.00004 lengua -0.00001 0.00001 -0.69000 0.49000 -0.00003 0.00001 educ 0.00000 0.00000 -0.26000 0.79800 0.00000 0.00000 edad 0.00000 0.00000 -0.05000 0.95600 0.00000 0.00000 rl -0.00001 0.00001 -1.24000 0.21500 -0.00004 0.00001 r3 -0.00006 0.00001 -4.99000 0.00000 -0.00009 -0.00004 r4 -0.00014 0.00002 -7.27000 0.00000 -0.00017 -0.00010 r5 -0.00011 0.00002 -4.56000 0.00000 -0.00015 -0.00006 -cons 0.01308 0.00003 483.30000 0.00000 0.01303 0.01314 217 APPENDIX 1:THE STOCHASTIC FRONTIER METHOD El enfoque economktrico utilizado esta basado en la construccion de la fi-ontera de produccion y la medida de ineficiencia respecto a ksta, es decir, aisla 10s efectos de la perturbacion aleatoria estandar de 10s correspondientes a la ineficiencia. Posteriormente, a partir de las estimaciones de la ineficienciapara cada productor, se trata de encontrar sus determinantes. L a fi-ontera de produccion estocastica se determina por la estructura de la tecnologia de produccion y por un componente que contiene las desviaciones observadas de la fbncion de produccion, cuyas fbentes pueden ser de dos tipos: 1) efectos especificos del productor que puedenser de cualquier signo y 2) laineficienciaproductiva. Esto es: hYi=BO+ 0 B, InX,i +vi - ui hYi= Bo + 0 B, InXni +ei donde el vector de insumos X pertenece a R', ;e l vector de product0 Y pertenece a R+;B es e l vector de parametros de tecnologia a estimar; i=l,.es e l indice de 10s productores. En ..,I este modelo de error compuesto, vi es e l tkrmino de perturbacion aleatoria simktrico, idkntico e independientemente distribuido (iid) como N(0, 0.") que captura 10s efectos de la perturbacion estocastica, y ui es el componente no negativo atribuible a la ineficiencia tkcnica , distribuido independientemente de vi. Dado que Ui2 0, ei = vi ui es asimktrico. Bajo el supuesto de que vi y ui se distribuyen - independientemente de Xi, la estimacion por M C O proporciona estimadores consistentes de B,, a excepcion del intercept0 Bo, debido a que E(ei) = - E(ui)< 0. Sin embargo, bajo 10s siguientes supuestos de la distribucion de ui y vi el Mktodo de Maxima Verosimilitud es mas eficaz. 1) vi se -iidN(O, 0,") ii) uise -iidN+(O,0,") como unamedia normal no negativa iii) viyuisedistribuyenindependientementeunadeotrayde10sregresoresX,. Dado que ei = vi-ui,,la fbncion de densidad conjunta de uiy ei es f(%e)=2rI0,0, 1 exp{-- U 2 20; - ?}V2 20, y la fbncion de densidad marginal de ei resultante de integraruisobre f(u,e) es Donde 0 = (0."+0,")1'2,. O= (Ou/O,), y O(.) y O(.) es la distribucion normal acumulativay la fbncion de densidad normal, respectivamente. Utilizando *, lafuncion de maximaverosimilitud para unamuestra de Iproductores es: 218 L a maximizacion de esta fbncion proporciona estimadores consistentes de maxima verosimilitud de todos 10s parametros cuando I+m. El siguiente paso consiste en obtener las estimaciones de eficiencia tkcnica para cada productor u hogar bajo la siguiente idea. Dado que e.=v.-u.1, entonces: s i ei>O,ui no es suficientemente grande (dado que E(vi) =O), l o cual implica que el productor es relativamente eficiente, en e l cas0 contrario (cuando eiF 0.00 Residual 1316.17 944.00 1.39 R-squared 0.33 Adj R- Total 1974.63 957.00 2.06 squared 0.32 Std. logvth Coef. Err. t P>t [95% Conf. Interval] hum 0.883 0.109 8.080 0.000 0.669 1.097 mi -0.243 0.086 -2.840 0.005 -0.411 -0.075 mmi -0.547 0.201 -2.720 0.007 -0.941 -0.153 acca 0.266 0.115 2.3 10 0.021 0.040 0.492 t l 0.914 0.271 3.370 0.001 0.383 1.446 t2 0.663 0.262 2.530 0.012 0.149 1.177 t3yt4 0.463 0.262 1.770 0.078 -0.051 0.977 tPP 0.205 0.085 2.420 0.016 0.039 0.372 r l -1.097 0.108 -10.190 0.000 -1.309 -0.886 r3 -0.848 0.115 -7.340 0.000 -1.074 -0.621 r4 -1.142 0.158 -7.240 0.000 -1.451 -0.832 r5 -1.945 0.133 -14.630 0.000 -2.206 -1.684 us02 0.203 0.094 2.150 0.032 0.018 0.388 - cons 9.230 0.287 32.170 0.000 8.667 9.793 222 ListofVariables 4 logaritmo delvalor bruto de l aproduccion agricola de 10s hogares lhar logaritmo del gasto en pesos en semilla, fertilizante y plaguicida lkjj logaritmo de la utilizacion de maquinaria en l a actividad en terminos de horas-tractor lmo logaritmo delnumero dejornales empleados durante las tres etapas delproceso productivo lsupest logaritmo delvalor en pesos del gasto en semilla, fertilizante y plaguicida Etapa I1 sex0 dummyque captura10s hogares donde eljefe es mujer lengua variable dummy que capta 10s hogares donde e ljefe habla una lengua o dialect0 ademas del castellano edad la edad deljefe delhogar educ e l nivel educativo deljefe, afios de escolaridad dispmo proporcion de la disponibilidad de mano de obra familiar restfinan restriccion financiera, capta e l acceso a financiamiento de otras fuentes distintas a las delpresupuesto familiar, en terminos netos restfinanor variables restfinan como porcentaje delvalor bruto de laproduccion agricola (q) proguber dummyque capta el acceso de 10s hogares agricolas atransferencias delgobierno proble dummyque captura si laproduccionagricola delhogar experiment0algim tip0 de siniestro y la cosecha h e menor a150% de lo esperado en afio normal mafri dummyde especializacion de laproduccionagricola en el cultivo de maiz y frijol olegra dummyde especializacion de laproduccionagricola en el cultivo de otros granos y oleaginosas caf dummyde especializacion de laproduccionagricola en el cultivo de cafe hort dummyde especializacion de laproduccionagricola en el cultivo de hortalizas peren dummyde especializacion de laproduccionagricola en cultivos perennes cuyos ciclos sonmayores a unafio gran dummyde hogar con 5 ha cultivadas o mas Peq dummyde hogar con menos de 5 ha cultivadas indserv proporcion que refleja l a dotacion de servicios en la comunidad a la que pertenece el hogar de untotal de nuevetipos organi dummyque capturalaexistencia de organizaciones de tip0 agricola, ejidal o campesina dentro de las comunidades ejicom Tipo de tenencia de l atierra ejidal y comunal priv Tipo de tenencia de la tierra privada mixta Tipo de tenencia de la tierra mixta, es decir, privada y ejidal o comunal r l region 1, Sur-sureste r2 region 2, Centro r3 region 3, Centro-occidente r4 region 4, Noroeste r5 region 5, Noreste 223 Bibliography Abramovay, Ricardo [et al.] ed. (1998). Juventude e AgriculturaFfamiliar : Desafios dos Novos Padrdes Sucesorios. - Brasilia :UNESCO. Adato, Michelle (2004). "Programas de Transferencias Monetarias Condicionadas : Beneficios y Costos Sociales". - In: La pobreza en Mkxico y el mundo 1 Julio Boltvinik y Araceli Damian. Mkxico : Siglo XXI Editores : Gobierno del Estado de Tamaulipas. - Aguilar Villanueva, Luis F. ed. (1993). La Implementacion de las Politicas.- Mkxico : Miguel Angel Porrua. Aguirre Bertran,G. (1967). Regiones de Refbgio. Mkxico : Instituto Nacional Indigenista. - Alwing, Jeffiey, Paul B. Siege1 and Steen L. Jorgensen (2001). Vulnerability : A View fiom Different Disciplines. - Washington, D.C. : The World Bank. - (Social Protection Discussion PapersNo. 0115). Araujo, Maria Caridad (2003). The Role of the Local Context andNon-AgriculturalEmployment on Poverty Reduction : Micro and Macro Evidence fiom Rural Mexico. - Berkeley : University of California. PhD. Thesis inAgriculture and ResourceEconomics. - Arellano, David (2001). Problemas y Potencialidades de 10s Presupuestos Orientados a Resultados : Limites del Gerencialismo en la Reforma Presupuestal. - Mkxico : Centro de Investigaciony Docencia Economica. - (Working Papers). and Ramon Gil Garcia (2003). Public Management Policy and Accountability inLatin America : PerformanceOriented Budget inColombia, Mexico andVenezuela. Mkxico : - Centro de Investigaciony Docencia Economica. - (Working Papers). Arias, Omar, Kevin F. Hollack and Walter Sosa Escudero (1999). IndividualHeterogeneity inthe Returns to Schooling : Instrumental Variable Quantile Regression Using Twins Data. Champaign, 11. :University of Illinois. (Working Paper). - Arrighi, Giovanni (1967). The Political Economy of Rhodesia. - Mouton :.The Hague. Audley, John [et al.] (2003). NAFTA's Promise and Reality : Lessons fiom Mexico for the Hemisphere. Washington D.C. :Carnegie Endowment for International Peace. - Avila, Flavio and R.E. Evenson (2004). Total Factor Productivity Growth in Agriculture : The Role of Technological Capital. -- Paper prepared by the Yale Economic Growth Center and the Fondo Regional de Tecnologia Agropecuaria (FONTAGRO) (Annual IDB Meeting March : Lima, Peru). Bardach, Eugene(1977). The Implementation Game. - Cambridge :MIT Press. Bardach, Eugene(2001). Los Ocho Pasos para e l Analisis de Politicas Publicas :UnManual para laPractica.-Mkxico : Centro de Investigaciony Docencia Economica. Barro, R.J. (1991). "Economic Growth in a Cross Section of Countries". - In:Ouarterlv Journal of Economics, (106)2. Barzelay, Michael (2003). "New Public Administration: Invitation to a Cosmopolitan Dialogue". -In:Gestionv Politica Publica, 12(2) SecondSemestre. Bellon, M.R. [etal.] (2004a). "Geospatial Dinesions of Poverty andFood Security :A Case Study of Mkxico".- Mkxico CIMMYT. http://www .cimmyt.org/gis/povertymexico/povei%-mapping-fmal-draf't.pdf 224 (2004a). Making Poverty Mapping Relevant to Crop Breeding to Benefit Poor Farmers :the Case of Maize inMexico. -- Mexico : International Maize andWheat Improvement Centre : CIMMYT. - (Working Document). Berman, Paul (1978). "The Study of Macro- and Micro-Implementation". - In: Public Policv, 26(2): 157-184. Bigman, David [et al.] (2000). "Community Targeting for Poverty Reduction inBurkinaFaso". . -In: The WorldBank Economic Review (14): 167-194. Blom, Andreas and Carlos Vklez (2001a). The Dynamics of the Skill-Premium in Brazil : Growing Demand or Insufficient Supply.-Washington, D.C. :The World Bank. Nina Pavcnik, and Norbert Schady (2001b). Trade Liberalization and Labor Market AdjustmentinBrazil. Washington, D.C. :The World Bank. - Boltvinik, Julio andAraceli Damian (2004). Lapobrezaen Mkxico y e l mundo. - Mkxico : Siglo XXI Editores :Gobierno delEstadode Tamaulipas. Buchinsky, Moshe (1998). "Recent Advances in Quantile Regression Models : A Practical Guideline for EmpiricalResearch". -In: The Journal of Human Resources33(1): 88-126. Burki, Shahid Javed and Guillermo E. Perry (1998). Beyond the Washington Consensus : Institutions Matter. - Washington, D.C. : The World Bank. -(World Bank Latin American and Caribbean Studies, Viewpoints). Cabrero Mendoza, Enrique (2002). "Los municipios y sus Haciendas Locales : Un Escenario de Carencias y Oportunidades?". - In: Perfil v Perspectivas de 10s Municipios Mexicanos para la Construccion de una Politica de Estado. - Mkxico : Instituto Nacional de Desarrollo Social. Casares, Enrique y Horacio Sobarzo, ed. (2004). "Diez afios del TLCAN en Mkxico : Una PerspectivaAnalitica". - In:ElTrimestre Economico. Serie LecturasNo. 95. - Mkxico : Fondo de Cultura Economica. Casco Flores, Andrks (2001). Gasto Publico para e l Desarrollo Rural en Mkxico : 1994-2000.- Santiago de Chile :FAO, OficinaRegionalparaAmkrica Latina. Andrks Rosenzwig, ed. (2000). La Politica Sectorial Agropecuaria en Mkxico :Balance de unaDkcada. -Mkxico :IICA. Centro Latinoamericano de Administracion para e l Desarrollo (CLAD) (1999). Una Nueva Gestion Publica paraAmkrica Latina. Venezuela : CLAD. -- (Documento del CLAD). - Comitk Tkcnico para la Medicion de la Pobreza (CTMP) (2004). Comunicado 3: Observaciones sobre las Cifras Oficiales para el Aiio 2002". - Mkxico : CTMP, (mimeo). (2005). Recomendaciones Metodologicas sobre la Estimacion Comparativa de 10s Niveles de Pobrezaen Mkxico : 2000-2002. -- Mkxico : CTMP. Consejo Nacional de Poblacion (CONAPO) (2004). Proyecciones de la Poblacion de Mkxico, de las Entidades Federativas, de 10s Municipios y Localidades : 2000-2050. - Mkxico : CONAPO. - (Coleccion ProspectivaDemografica). Consultative Group to Assist the Poor (CGAP) (2004). Microfinance Institutions Moving into Financing ofAgriculture. -Washington, D.C. : The World Bank : CGAP. - (Information Notes on Microfinance and Rural FinanaceNo. 2 January). Cortks, Fernando(2003). LDisminuyo lapobreza? :Mkxico 2000-2002..-Mkxico :ElColegio de Mkxico, Centro de Estudios Sociologicos, (mimeo). 225 Rosa Maria Rubalcava (1991). Autoexplotacion Forzada y Equidad por Empobrecimiento. Mkxico : ElColegio de Mkxico. - Chayanov, Alexandre V. (1996). Theory of the Peasant Economy. -- Homewood, Ill.: Richard Irwin. Davis, Benjamin and Paul Winters (2001a). "Gender, Networks andMexico-USMigration". -In: Journal of Development Studies 3X(2). Susdhanshu Handa and Humberto Soto (2001b). Crisis, Poverty and Long Term Development : Examining the Mexican Case. Roma :FA0 : IADB : PROGRESA. - Guy Stecklov and Paul Winters (2002). "Domestic and International Migration from Rural Mexico : Disaggregating the Effects of Network Structure and Composition". -In: Population Studies 56: 291-309. (2003). Agricultural Reform inMexico : The Experience of the PROCAMPO Program. - Roma :FAO. D e Ferranti, David [et al.] (2002). From Natural Resources to the Knowledge Economy : Trade and Job Quality. .-Washington, D.C. ;The World Bank. (World Bank Latin American - and Caribbean Studies). [et al.] (2004a). Beyond the City : the Rural Contribution to Development. - Washington, D.C. ; The World Bank. - (World Bank Latin American and Caribbean Studies). {et al.] (2004b). Inequality in Latin America : Breaking with History? Washington, D.C. ;The World Bank. (World Bank Latin American and CaribbeanStudies). - D e Janvry, Alain and Elisabeth Sadoulet (1996). Growth, Inequality and Poverty in Latin America : A Causal Analysis, 1970-94. Berkeley, Cal. : University of California, Department ofAgriculture and ResourceEconomics. Gustavo Gordillo and Elisabeth Sadoulet (199Xa). Entre e l Control Politico y la Eficiencia : Evolucion de 10s Derechos de PropiedadAgraria en Mkxico. In:Revista de - la CEPAL (66): 149-166. (199Xb). "El TLCAN y la Agricultura : Evaluacion Inicial". In: TLC v Agricultura : Funciona el Experimento? / Rita Schwentesius [et al.] : Mkxico : Universidad Autonoma de Chapingo, CIESTAM. andElisabeth Sadoulet (2001). "Income Strategies Among Rural HouseholdsinMexico :The Role of Off-farmActivities". - In:World Development (29)3: 467-480. (2002). Targeting and Calibrating Educational Grants : Focus on Poverty or on Non- enrollment Risk?. - Berkeley, Cal. : University of California, (Manuscript Version December). Dirven,Martine (2002). Las Practicasde Herencia de Tierras Agricolas : LunaRazonmaspara el Exodo de la Juventud?.- Santiago de Chile : CEPAL. - (Serie Desarrollo Productivo No. 135). (2003). "Entre el Ideario y la Realidad : Capital Social y Desarrollo Agricola, Algunos Apuntes para la Reflexion". - In: Capital Social v Reduccion de la Pobrezaen Amkrica Latina v e l Caribe : en Busca de un Nuevo Paradigma / Raul Atria [et al.] comp. - Santiago de Chile : CEPAL :Michigan State University. 226 (2004). Alcanzando las Metas del Milenio : Una Mirada Hacia la Pobreza Rural y Agricola. - Santiago de Chile : CEPAL. - (Serie Desarrollo Productivo No. 146). Drake, Deborah and Elisabeth Rhyne, ed. (2002). The Commercialization of Microfinance : Balancing Business andDevelopment. Bloomfield : Kumarian Press. - Edouard, Fabrice, Anne-Marie L e Moing andMarco Antonio Gonzalez Ortiz (2004). Esquemade Acceso de 10s Jovenes a la Tierra en e l Campo Mexicano. -- Field Studies for the ResearchProject Preparedfor the FAO, World Bank Cooperative Program. Elbers, Chris [et al.] (2001). Poverty and Inequality in Brazil: Estimates from Combined PPV- PNAD Data. -Washington, D.C. :The World Bank. Elmore, Richard F. (1979-80). "Backward Mapping : Implementation Research and Policy Decisions". - In:PoliticalScience Ouarterlv 94(4)Winter: 601-616. (1978). "Organizational Models of Social Program Implementation". - In: Public Policv 26(2)Spring: 185-228. Escobal, Javier and Maximo Torero (2000). LComo Enfrentar una Geografia Adversa? : Elrol de 10sActivos Publicos y Privados.-Lima : Grade. -(Documento de Trabajo No. 29). Evans, Peter (1995). EmbeddedAutonomy : States and Industrial Transformation. -- Princeton : Princeton University Press. (1996). "Government Action, Social Capital and Development : Reviewing the Evidence on Synergy". - In:World Development 24(6): 1119-1132. FA0y Banco Mundial (2003). Lanueva ruralidad en Europay su inter& paraAmkrica Latina. - Roma :FAO. http://www.grupochorlavi.org/php/doc/documentos/ruralidadEuropa.pdf Ferreira, Francisco H.G. and Peter Lanjow (2001). "Rural Nonfarm Activities and Poverty inthe Brazilian Northeast". - In:WorldDevelopment (29)3: 509-528. Finan, Frederico, E. Sadoulet and A. de Janvry (2002). Measuring the Poverty Reduction Potential of LandinRural Mexico. Berkeley, Cal. :University of California. - Foster, Andrew D. and Mark Rosenzweig (2003). Agricultural Development, Industrialization and Rural Inequality. -Washington, D.C. : The World Bank, (Draft). Foster, J, J. Greer and E. Thorbecke (1984). "A Class of Decomposable Poverty Measures". -In: Econometrica (52): 761-766. Fox, Jonathan (1995). "Governance and Rural Development in Mexico : State Intervention and Public Accountability". - In:The JournalofDevelopment Studies 32 (1): 1-30. Franco, Jorge (2004). Los Programas de Desarrollo Rural : Operacion Institucional yAlivio a la Pobreza. Mkxico : Banco Mundial, (InputPaper). Georgescu-Roegen, Nicolas (1960). Economic Theory and Agrarian Economics. -- Oxford Economic Papers. G6mez Cruz, Miguel Angel y Rita Schwentesius Rindermann ed. (2001). Estrategias para el Cambio en el CampoMexicano. Mkxico :Plaza and Valdks. - Gonzalez-Konig, Gabriel and Quentin T. Wodon (2003). "Do Cash Transfers to Farmers Reduce Migration? : ProcampoinMexico". -- In: Internal Migration, Urbanization and Povertv : Case Studies for Latin America. -Washington, D.C. The World Bank. " 227 Gonzalez de Alba, Ivan and Thania de la Garza Navarrete (2002). "La Descentralizacion del Gasto Social para Infiaestructura : LLas Autoridades Locales Asignan Mejor 10s Recursos que la Federacion?". - In: Perfil v Perspectivas de 10s Municipios Mexicanos para la Construccion de unaPoliticade Estado. -Mkxico : SEDESOL, Indesol. Guerrero Amparian, JuanPablo (2000). La Reforma de laAdministracion PublicaMexicana en el Nuevo Rkgimen Politico : LPor Donde Empezar? : Ideas para la Creacion de un Gobiemo Transparente, Responsable y Cercano a la Ciudadania. - Mkxico : Centro de Investigacion y DocenciaEconomica. (Working Papers). - Heclo, Hugh (1978). "Issue Networks and the Executive Establishment". - In: The New American Political System / A. King ed. - Washington, D.C. : American Enterprise Institute. Helmsing, Bert A.H.J. (2001). Partnerships, Meso-institutions and Learning: New Local and Regional Economic Development Initiatives in Latin America. -- Institutes of Social Studies, The Hague. Hemandez-Laos, Enrique (1994). "Altemativas de largo plazo para erradicar la pobreza en Mkxico". -- In: Frontera Norte : Special Survey on Poverty / Diana Alarcon ed. -- El Colegio de laFronteraNorte. Hollinger, Frank (2004). Fostering Inter-Generational Transition by Imporving Access to Land and Livelihood Options of Rural Youth in Mexico : Analytical Report. - Washington, D.c.:The World Bank. Holzmann, Robert and Steen Jorgensen (2000). Social Risk Management : A New Conceptual Framework for Social Protection and Beyond. - Washington, D.c.: The World Bank. - (Social Protection Discussion Paper Series No. 0006). Instituto Interamericano de Cooperacion para la Agricultura (IICA) (2003). Evaluacion Institucional Extema del Instituto Nacional de Investigaciones Forestales, Agricola y Pecuarias,INIFAP. -- Mkxico :IICA. (2004). Ley de Seguridad Agropecuaria e Inversion Rural 2002 de EE.UU. y sus Implicaciones en Mkxico y en 10s Mercados Agropecuarios Intemacionales. - Mkxico : IICA. Instituto Maya (2003). Informe Final de la Evaluacion Extema del Programa de Microrregiones. -Mkxico :Instituto Maya. Instituto Superior Tecnologico de Monterrey (ITESM), Campus Querktaro (2004). AGROASEMEX : Evaluacion Extema a1 Ejercicio 2003 del Programa de Subsidio a la Prima del SeguroAgropecuario". Kerrigan, George (2001). Public Spending for the Agricultural Sector and Development of Rural Areas: LAC, Trends and Challenges. -- FA0 Consultant Report Presentedat the Seminar Management of Public Spending for Agricultural and Rural Development in Latin America andthe Caribbean (Santiago de Chile, August 2001). Kjollerstrom, Monica (2004). Competitividad del Sector Agricola y PobrezaRural : El Papel del Gasto Publico en Amkrica Latina. - Santiago de Chile : CEPAL, Unidad de Desarrollo Agricola, -- (Serie Desarrollo Productivo No. 155). Koenker, R. and G. Basset Jr. (1978). "Regression Quantiles". - In: Econometrica 46(91): 410- 414. Kula, Witold (1970). Thkorie Economique duSsyst&meFkodal. - Paris :Muton. 228 Lanjouw, Peter (1999). "Rural Non-agricultural Employment and Poverty in Ecuador". - In: Economic Development and Cultural Change48(1). Lanjouw, Jean 0.and Peter Lanjouw (2001). "The Rural Non-farm Sector: Issues and Evidence from Developing Countries". - In:Agriculture Economics (26): 1-23. Ledgerwood, Joanna (1999). Sustainable Banking with the Poor. Microfinance Hanbook : An Institutional andFinancial Perspective. - Washington, D.C. :The World Bank. Levy, F. and R.J. Murnane (1992). "US Earnings Levels and Earnings Inequality : A Review of Recent Trends and Proposed Explanations". - In: Journal of Economic Literature (30): 1333-1381. Lipsky, Michael (1997). "Towards a Theory of Street-Level Bureaucracy". - In: Theoretical Prespective on Urban Politics / Willis Hawley and Michael Lipsky ed. -- Englewood Cliffs : Prentice Hall. Lopez, Ramon (2002). Agricultural Growth and Poverty Reduction. - Rome : FAO. (Technical - Note for the Roles ofAgriculture Project). Lustig, Nora (1992). Mexico : The Remaking of an Economy. - Washington, D.C. : The Brooking Institution. Maloney, William F. [et al.] (2003). Notes on Income and Consumption Shocks inMexico. Notes on Income and Consumption Shocks in Mexico. - Washington, D.C. : The World Bank, Latin America and CaribbeanDepartment, (mimeo). Mankiw, N.G., D. Romer, and D.N. Weil (1992). "A Contribution to the Empirics of Economic Growth". - In:Ouarterlv Journal ofEconomics (107)2. Marini, Caio (2002). "0 Context0 Contemporikeo da Administraqgo Publica naAmkrica Latina". - In:Revistado ServicoPublico 53(4), October-December:31-51. Maxwell, Simon (1999). The Meaning and Measurement of Poverty : OD1 Poverty Briefings. - London : ODI. Mckinley, T. and D. Alarcon (1995). "The Prevalence of Rural Poverty inMexico". - In:World Development 23(9): 1575-1585. McKenzie, David (2003) "How do Households COPE with Aggregate Shocks? : Evidence from the Mexico Peso Crisis". - In:World Development 3l(7): 1179-1199. Mellor, John W. (1976). The New Economics of Growth : A Strategy form India and the Developing World. - Ithaca : Cornel University. Mellor, John W. (2000). Faster, More Equitable Growth : The Relation Between Growth and Agriculture and Poverty Educations. - Harvard : Harvard Institute for International Development. (CAER I1Discussion PaperNo. 70). - and G. Desai (1985). Agricultural Change and Rural Policy. - Baltimore : John Hopkings University Press. Mincer, Jacob (1974). "Schooling, Experience and Earnings". - In:NBERWorking Paper. Mufioz-Piiia, Carlos, Alain de Janvry and Elisabeth Sadoulet (2003). "Recrafting Rights over Common Property Resources in Mexico". - In: Economic Development and Cultural Change (52)l: 129-158. 229 Nahmad, Salomon and Tania Carrasco (2004). La Construccion de una Tipologia de la Pobreza Rural : 10s Riesgos y las Estrategias de Sobrevivencia en Mkxico. - Mkxico : Banco Mundial, (Input Paper). Nicita, Alessandro (2004). Who Benefitedfrom Trade Liberalization inMexico? : Measuring the Effects on Household Welfare. - Washington, D.C. : The World Bank.-(Policy ResearchWorking Paper Series No. 3265). Ostrom, Elinor (1990). Governing the Commons : The Evolution of Institutions for Collective Action. -- Cambridge : Cambridge UniversityPress. Partida Bus, Virgilio (2003). "Aspectos Demograficos de la Urbanizacion". - In: La Situacion Demoarafica de Mkxico. -- Mkxico : CONAPO. Piiieiro, Martin [et al.] (1999). La Institucionalidad en el Sector Agropecuario en Amkrica Latina : Evaluacion y Propuestas para una Reforma Institucional. - Washington, D.C. : Banco Inter-American0 de Desarrollo. - (Serie de Informes Tkcnicos. Departamento de Desarrollo Sostenible). Otero, Maria and Elisabeth Rhyne, ed. (1994) The New World of Microenterprise Finance : BuildingHealthy Financial Institutions for The Poor. -- Hartford :Kumarian Press. Paes e Barros, Ricardo, Mirella de Carvalho and SamuelFranco (2004). PobrezaRural e Trabalo Agricola no B r a d ao Longo da Dkcada do Noventa. -- Rio de Janeiro : IPEA, (Manuscript .Work on Progress). Parker, Susan W. (2003). Evaluacion delImpact0 de Oportunidades sobre la Inscripcion Escolar : Primaria, Secundariay Media Superior. Mkxico : Centro de Investigacion y Doncencia - Economica. (Documentos de Trabajo, SDTE 263). - Platteau, Jean-Phillipe (2002). The Gradual Erosion ofthe Social Security Function of Customary Land Tenure Arrangements in Lineage-Based Societies. -- Helsinki : United Nations University-(Wider DiscussionPaperNo. 2002/26). Presidencia de la Republica (2001). Plan Nacional de Desarrollo 2001-2006. - Mkxico : Presidencia. Putnam, Robert D. (1993). Making Democracy Work : Civic Traditions in Modern Italy. -- Princeton : PrincetonUniversityPress. Puyana, Alicia and Josk Romero (2004). Evaluacion Integral de 10s Impactos e Instrumentacion del Capitulo Agropecuario delTLCAN. Mkxico :ElColegio de Mkxico. - Ramirez, Alejandro and Vicente Garcia (2004). Poverty : Human Development and Indigenous People in Mexico 1982-2002.. Washington, D.C. : The World Bank, (Draft Paper July - 2004). Ramirez Santos, Tayde and Francisco J. Ramired Diaz (2002). "~QukPas6 con la Nueva Forma de Legislar? : El Caso de la Ley de Desarrollo Rural". -- In: Estrateaiaspara el Cambio en e l Campo Mexican0 / Miguel h g e l Gomez Cruz y Rita Schwentesius ed. - Mkxico : Plazay Valdks. Ramos, Dora (2000). "Diner0 del Diablo (Progresa)". - In:Equidad, Participacion v Coherencia. ElBanco Mundial en Beiiina +5, Sus Politicas de Equidad. Banco Mundial en laMira de las Mujeres. Mkxico. Rao, J. Mohan and Josk Maria Caballero (1990). "Agricultural Performance and Development Strategy : Retrospect and Prospect". - In:World Development 1X(6): 899-913. 230 Ravallion, Martin and Gaurav Datt (1996). "How Important to India's Poor i s the Sectoral Composition of Economic Growth?'.-In: The World Bank Economic Review 1O(1): 1- 25. Reardon, Thomas [et al.] (2000). "Effects ofNon-Farm Employment on Rural Income Inequality in Developing Countries : An Investment Perspective". - In: Journal of Agricultural Economics 51(2): 266-288. Julio Berdeguk and German Escobar (2001). "Rural Nonfarm Employment and Incomes in Latin America : Overview and Policy Implications". - In: World Development (29)3: 395-409. Red para el Desarrollo Sostenible de Mkxico, A.C, (RDS) (2004). Informe Final de Evaluacion Externa : ProgramaOpciones Productivas 2003. Mkxico : SEDESOL. - Rein, Martin and Francine Rabinovitz (1978). "Implementation : A Theoretical Perspective". - In:American Politics and Public Policy / Walter D. Burnham andMartha W, Weinberg, ed. Cambridge :MIT Press. - Rello, Fernando and Y. Trapaga (2001). Libre Mercado y Agricultura : Efectos de la Ronda Uruguay en Costa Rica y Mkxico. - Santiago de Chile : CEPAL. - (Serie Estudios y PerspectivasNo. 7). Rhyne, Elisabeth and Linda S. Roblatt (1994). What Makes then Tick? : Exploringthe Anatomy of Major Microenterprise Finance Organizations. - Somerville : ACCION International. - (Monograph Series No. 9). Robinson, Marguerite S. (2001). The Microfinance Revolution : SustainableFinance for the Poor. -Washington, D.C. : The World Bank. Roumasset, James A. (1976). Rice and Risk : Decision-Making Among Low Income Farmers. -- New York ;Amsterdam :American Elsevier Pub. Co. Roseboom, Johannes (2004). Case Study on the Financing of Agricultural ResearchinMexico. - Washington, D.C. :The World Bank, (Consultant Report July 2004). Ruiz-Arranz, Marta [et all (2002). More Calories or More Diversity? : An Econometric Evaluation of the Impact of the PROGRESA and PROCAMPO Transfer Programs on Food Security inRural Mexico. - ( E S M A O Working PaperNo. 06-2002). Sadoulet, Elisabeth, Alain de Janvry and Benjamin Davis (2001). "Cahs Transfer with Income Multiplier:PROCAMPO inMexico". -- World Development 29(3) March. Scott, John (2003). "Mkxico : Programa de Educacion, Salud y Alimentacion (PROGRESA)". - In: La Pobreza Rural en Amkrica Latina : Lecciones para una Reorientacion de las Politicas.-Santiago de Chile : CEPAL :FA0 :RIMISP. (2004). Comparabilidad ENIGH : Ingresos -Transferencias Publicas. Comitk Tkcnico ParalaMedicion de la Pobreza.Febrero 2004. SchejtmanAlexander y Julio Berdeguk (2003). Desarrollo Territorial Rural. -- Santiago de Chile : Centro Latinoamericano para el Desarrollo Rural Schwentesius, Rita, Manuel Angel Gomez Cmz y Gary W. Williams, ed. (1998). TLC y Agricultura : Funciona el Experimento?. -- Mkxico : Universidad Autonoma de Chapingo, CIESTAM. 231 Secretaria de Desarrollo Social (SEDESOL) (2002). Medicion de la Pobreza : Variantes Metodologicas y Estimacion Preliminar. - Mkxico : SEDESOL, Comitk Tkcnico de Medicion de la Pobreza. - (Series Documentos de InvestigacionNo. 1.). Secretaria de Agricultura, Ganaderia, Desarrollo Rural, Pesca y Alimentacion (SAGARPA) and FA0(2003). Evaluacionde laAlianzaparael Campo 2002. (2004). Programa Especial Concurrente. -- PowerPoint Presentation from the Subsecretaria de Desarrollo Rural. Secretaria de Gobernacion (SEGOB) (2003). "Normas de Operacion del Fondo Nacional de DesastresNaturales, FONDEN'. -In : Diario Oficial de laFederacion Mavo 23. Sen, Amartya K. (1984). Poor Relatively Speaking : Resources Values and Development. - Oxford : BasilBlackwell. (1985). Commodities and Capabilities. -- Amsterdam :Elsevier. (1999). Development as Freedom. -- New York : KnopfPress. Siegel, Paul B. and Jeffrey Alwang (1999). An Asset-Based Approach to Social Risk Management : A Conceptual Framework. - Washington, D.C. : The World Bank. - (Social Protection Discussion Paper Series No. 9926). Skees, Jerry [et al.] (2002). "Can Financial Markets be Tapped to Help Poor People Cope with Weather Risks?' .--Helsinki : UnitedNations University. - (Wider Discussion Paper No. 2002/23). Skoufias, Emmanuel (2004). Household Vulnerability and its Determinants in Rural and Urban Mexico. - Washington, D.C. : The World Bank, (UnpublishedPaper). Snyder, Richard (2001). Politics after Neoliberalism : Reregulaion in Mexico. -- Cambridge : Cambridge UniversityPress. Szkkely-Pardo, Miguel (1994). "Estabilizacion yAjuste con Desigualdad y Pobreza". - In: El Trimestre Economico LXI(1). Soloaga, Isidro and Mario Torres (2003). Agricultural Growth and Poverty Reduction. -- Roles o f Agriculture Project International Conference, Rome, October 2003. Taylor, Edward J. and Antonio Yunez-Naude (2000). "The Returns from Schooling in a Diversified Rural Economy". -- American Journal o f Agriculture Economics 82(May): 287-297. Hazael Ceron (2004). The Effects o f Rural Mexico Household Assets on Rural Non- Rural Activity Selection, on Incomes and Technology Adoption. - Mexico : Banco Mundial : ElColegio de Mkxico : PRECESAM. Tendler, Judith (1997). Good Government in the Tropics. - Baltimore : The John Hopkins UniversityPress. (1998) "Rural Development without the Urban-Rural Divide : Decentralization, Participation and Other Local Things". In: Politicas Agrarias SpecialNumber. - UnitedNations (2004). Strengthening Public Administration for the MillenniumDevelopment Goals : A Partnership Building Approach. New York : Committee o f Experts on Public Administration. -(Report o f the Secretariat, 16 January 2004). Uphoff, Norman (2003). "El capital social y su capacidad de reduccion de la pobreza". -In: Capital Social y Reduccion de la Pobrezaen Amkrica Latina y el Caribe : en Busca de un 232 Nuevo Paradigma / Raul Atria [et al.] comp. -- Santiago de Chile : CEPAL : Michigan State University. Valencia Flores, Jos6 (2004). Dialog0 para el Desarrollo Rural. -- Consultation Report Prepared for the World Bank.-(Tequesquitengo Mayo 16-18,2004). Vences-Rivera, Jos6 (2003). Estimaciones de Pobreza, 2000-2002 : Prueba de Hipotesis. - M6xico : INEGI, (mimeo). Warman, Arturo (2001). El Campo Mexican0 en e l Siglo XX. M6xico : Fondo de Cultura - Economica. Williams, Walter (1980) The Implementation Perspective. -- Berkeley : University of California Press. World Bank (1995). Mexico Rural Financial Markets. - Washington, D.C. : The World Bank. - (Report No. 14599-ME). (1997). World Development Report 1997 : The State in a Changing World. - Washington, D.C. :The World Bank. (1999). Mexico's Ejido Sector Responds : Economic Adjustment and Institutional Reform. Washington, D.C. :The World Bank. (Report No. 18897). - - (2000). World Development Report 2000/01 : Attacking Poverty. - Washington, D.C. : The World Bank. (2001a). Mexico Crop InsuranceReport. .-Washington, D.C. :The World Bank. (2001b). Mexico Land Policy :A DecadeAfter the Ejido Reform. . -Washington, D.C. :The World Bank. (ReportNo. 22187-ME). - (2002a). World Development Report 2002 : Institutions for Markets. . - Washington, D.C. :The World Bank. (2002b). A Rural Development Strategy for the Peruvian Sierra. . - Washington, D.C. : The World Bank. (2002~).Reachingthe RuralPoor : RuralDevelopment Strategy. .-Washington, D.C. : The World Bank. (2003a). Mexico's Financial Markets and the Rural Economy. . - Washington, D.C. : The World Bank. (2003b). Southern States Development Strategies. . - Washington, D.C. : The World Bank. (2004a). Poverty in M6xico : An Assessment of Conditions, Trends and Government Strategy. -Washington, D.C. :The World Bank. (Report No. 28612-ME). - (2004b). Mexico Public Expenditure Review. Washington, D.C. : The World Bank.- - (Report No. 27894-ME). (2004~).World Development Report 2004 : Making Services Work for the Poor. - Washington, D.C. :The World Bank. (2004d). Ecuador Poverty Assessment. - Washington, D.C. : The World Bank. - (Report No. 27061-EC). (2004e). A Strategy for Rural Development in Ecuador. - Washington, D.C. : The World Bank, (Draft). 233 Winters, Paul, Benjamin Davis and Leonard0 Corral (2002). "Assets, Activities and Income Generation in Rural Mexico : Factoring in Social and Public Capital".-In:Agricultural Economics 27: 139-156. Yunez-Naude, Antonio and J. Edward Taylor (2001). "The Determinants o f Nonfarm Activities and Incomes o f Rural Households in Mexico with Emphasis on Education". - In:World Development (29)3: 561-572. Antonio and Fernando Barceinas (2003). Lessons from NAFTA : the Case o f Mexico's Agricultural Sector.-Washington, D.C. : TheWorld Bank. (2004). "El TLCAN y la Agricultura Mexicana". - In: Diez afios del TLCAN en Mkxico : Una Perspectiva Analitica / Enrique R. Casares y Horacio Sobrazo, ed. - Mexico :Fondo de Cultura Economica. - (ElTrimestre Economico, Lecturas). 234