Report No. 51927-YE Republic of Yemen COPING STRATEGIES IN RURAL YEMEN AND POLICY IMPLICATIONS June 10, 2010 Sustainable Development Department Middle East and North Africa Region Document of the World Bank CURRENCY EQUIVALENTS (As of June 10, 2010) Currency Unit: Yemeni Rials (YRIs) Exchange Rate: US$1 = 219.25 YRIs ABBREVIATIONS AND ACRONYMS CCT Conditional Cash Transfer DPPR Development Plan for Poverty Reduction GOY Government of Yemen HBS Household Budget Survey HH Households MAI Ministry of Agriculture and Irrigation MOE Ministry of Education MOLA Ministry of Local Administration PWP Public Work Program RLDS Rural/Local Development Strategy SFD Social Fund for Development SWF Social Welfare Fund YAR Yemen Arab Republic YR Yemeni Rial Vice President : Shamshad Akhtar Country Director : A. David Craig Country Manager : Bensong Ateng Sector Director : Laszlo Lovei Sector Manager : Luis Constantino Task Team Leader : Marie-Helene Collion Co-Task Team Leader: Hyoung Gun Wang CONTENTS Acknowledgements ................................................................................................................... iv EXECUTIVE SUMMARY ....................................................................................................... v ................................................................................................................................... 1 1. BACKGROUND, OBJECTIVES AND METHODOLOGY .................................1 Objectives ................................................................................................................................ 3 Methodology ........................................................................................................................... 3 Organization of the paper ........................................................................................................ 4 2. SPATIAL INEQUALITIES ......................................................................................5 "Rich" versus "Poor" Districts ................................................................................................ 5 Summary ............................................................................................................................... 21 3. INTER-HOUSEHOLD INEQUALITIES .............................................................22 Rural livelihood strategies ..................................................................................................... 22 Access to land, inequalities and livelihoods .......................................................................... 31 Agriculture and rural livelihood ............................................................................................ 36 Household demography, migration and education assets ...................................................... 42 Women in the rural economy and society ............................................................................. 47 Synthesis................................................................................................................................ 49 4. IMPLICATIONS FOR GOVERNMENT AND DONORS .................................60 Policies and program priorities to target the rural poor ......................................................... 60 REFERENCES ........................................................................................................................ 89 LIST OF FIGURES Figure 1: Rural-Urban Poverty 1998-2005 ..........................................................................1 Figure 2: Distribution of poverty rate and mass ..................................................................6 Figure 3: Household income structure in rich versus poor rural districts ............................9 Figure 4: Access to land according to district poverty levels ............................................12 Figure 5: Number of people per ha of arable land according to district poverty levels. ...12 Figure 6: Access to services and district poverty incidence ..............................................17 Figure 7: Access to services and district average household income ................................18 Figure 8: Participation in economic activities (% of all rural HH) ....................................22 Figure 9: Household categorization according to main source of income ........................23 Figure 10: Level of income and poverty incidence by main source of income .................24 Figure 11: Income structure of the poorest, middle-class and richest households ............27 Figure 12: Land ownership according to livelihood category ...........................................29 Figure 13: Irrigation according to livelihood category ......................................................30 Figure 14: Livestock assets and livelihood strategy ..........................................................30 Figure 15: Chart of gini coefficient of land value (= 0.68) ...............................................32 Figure 16: Access to land and income ...............................................................................33 Figure 17: Gini coefficient of income ...............................................................................34 Figure 18: Structure of crop and livestock monetary income (gross) ...............................37 Figure 19: Share of households involved in marketing agricultural products (in% of all rural HH) ...........................................................................................................37 i Figure 20: Cash transfers of the Social Welfare Fund according to HH income (in% of total rural transfers) ...........................................................................................81 LIST OF TABLES Table 1: Rural districts income structure according to poverty incidence ..........................9 Table 2: Structure of rural districts agriculture monetary income according to poverty incidence ...........................................................................................................11 Table 3: Household livestock assets according to district poverty levels .........................15 Table 4: Human resource assets ........................................................................................16 Table 5: Income and poverty levels according to HH main source of income..................24 Table 6: Livelihood category according to income quintiles ............................................25 Table 7: Total household income of farming households by income quintiles (YR) ........25 Table 8: Structure of income according to income quintiles .............................................26 Table 9: Household capital assets and main source of income .........................................28 Table 10: Enterprise income of entrepreneur HHs with land and without land (YR) .......29 Table 11: Household assets and level of income ...............................................................31 Table 12: Access to land, income and poverty levels (HH with access to land) ...............32 Table 13: Income quintiles: comparison of average and median ......................................34 Table 14: Household capital assets and access to land .....................................................35 Table 15: Size of cultivated land and main source of income (in% of hh in the group) ...36 Table 16: Crop and livestock income structure according to main source of income .......38 Table 17: Structure of agricultural income by income quintile .........................................39 Table 18: Agricultural monetary investments by livelihood categories ............................42 Table 19: Household demography, migration and education according to livehood categories ..........................................................................................................44 Table 20: Household demography and education assets according to income levels .......45 Table 21: Education and participation of rural women in the labor force according to main source of income of the household ..........................................................48 Table 22: Rural women participation in the labor force according to income levels ........48 Table 23: Rural livelihood strategies and welfare .............................................................51 Table 24: Household assets and rural livelihood strategies ...............................................53 Table 25: Household assets and welfare: a base model .....................................................54 Table 26: Household assets and welfare: extended ...........................................................56 Table 27: Household assets and being trapped in rural poverty ........................................58 Table 28: Yields of cereal crops from 2004 to 2009 (tons) ...............................................70 Table 29: Objectives and Priority Programs to Address the Needs of the Rural Poor ......84 Table A. 1 ..........................................................................................................................92 Table A. 1 Characteristics of agro-climatic zones and suitability for crop cultivation .....92 LIST OF BOXES Box 1: The story of Saoud District in Amran Governorate (Poverty incidence: 76%) ....20 Box 2: Empowerment for Local Development Program ...................................................64 Box 3: Consider upscaling: the Rainfed Agriculture and Livestock Project .....................72 Box 4: Pilot experience of support to coffee value chain in Taiz and Haraz. ...................75 Box 5: Worth up-scaling: The Labor-Intensive Work Program of SFD. ..........................80 ii LIST OF MAPS Map 1: Location of poor, medium poor and rich rural districts (poverty rate) ..................5 Map 2: Spatial distribution of rural poverty (mass)............................................................7 Map 3: Agroclimatic zones ..................................................................................................8 Map 4: Rural market potential ...........................................................................................10 Map 5: Average cultivated land size per district ...............................................................13 Map 6: Location of the districts with very large landholdings (Average cultivated land size between 2 and 17 ha) .................................................................................14 Map 7: Access to irrigation per district (number of households, %) .................................15 Map 8: Paved roads ...........................................................................................................19 Map 9: Traveling time to a city of 20 000 inhabitants ......................................................19 Map A 1: Annual average rainfall .....................................................................................91 ANNEXES Annex 1: Agro climatic data and rainfall map...................................................................91 Annex 2: List of Districts cumulating high poverty incidence and low access index .......95 iii ACKNOWLEDGEMENTS This report is the result of team work, between Marie-Hélène Collion (Lead Agricultural Specialist and Task team leader) and Hyoung Gun Wang (Economist and co-task-team leader). The work was carried out at the request of the Ministry of Planning and International Cooperation, as a follow up to the Poverty Assessment Report, under the guidance of Dr. Al- Abassi, Deputy Minister of Planning and International Cooperation. The team is highly indebted to Elisabeth Sadoulet (Professor of Agricultural Economics, University of California at Berkeley) who, with her research assistant, Léandre Bassole organized and directed the preliminary analysis of the HBS data for the rural areas. She and Alain de Janvry (Professor of Agricultural economics, University of California at Berkeley) made invaluable suggestions to solve problems that the team encountered with the data set and provided key advice such as the way to categorize households according to main source of income. The Small Micro Enterprise Promotion Service (SMEPS) and The Royal Tropical Institute (KIT) carried out an analysis of the five agricultural value chains. The analysis has already proven very valuable as it is already being used by SMEPS and donors to develop programs to support the honey and coffee value chains. The team thanks Pierre Rondot for overseeing this work on behalf of the World Bank. Thirumalai Srinivasan with the help of Gudivada Venkateswara Rao carried out an analysis of the pro-poor targeting of the public resource allocation at the district level, which in the end unfortunately could not be used in this report, but that we hope will be made available. The team is grateful to the staff of Social Fund for Development, in particular to Abdo Al-Qubaty and his team who gracefully shared the District Basic Service Access index that they constructed, and to Mahmood Sallam and Wadie Al-Mekhlafi who helped convened a consultation with key rural development resource persons. The Central Statistical Office staff provided priceless support to match the local land measures reported in the HBS with metric land measures, a complex and cumbersome procedure. Without their dedicated work, the important analysis of land access could not have been done. Also, the Agriculture Research and Extension Authority that carried out extensive GIS work provided agroecological maps that have been used in the report. Frederic Pelat carried out a review of on-going rural development programs and Dr. Mukreed provided support with interviews with resource persons and visits in the field. The peer reviewers were Paul Siegel (Consultant), Ruslan Yemtsov (PREM), Wilfried Engelke (Yemen Country economist) A number of Bank staff provided the team with valuable support in particular Naji Abu-Hatim (ARD) as well as suggestions and comments: Pierre Rondot (ARD), Afrah Alawi Al-Ahmadi (HD), Roberta Gatti (SM, HD), Trina Haque (HD), Harsha Thirumurthy (HD) and Yoshiharu Kobayashi (Water). Our thanks also go to Nabila Al-Mutawakel who successfully organized two workshops to gather feedback from Yemeni counterparts, and to Erika Salamanca and Syviengxay Creger who worked on formatting the document. iv EXECUTIVE SUMMARY COPING STRATEGIES IN RURAL YEMEN Background Yemen is predominantly a rural1 country and poverty is by and large a rural phenomenon: 73% of the population, and 84% of the poor, live in rural areas.2 Between 1998 and 2005, while the percentage of poor in urban areas declined by 11.6 percentage points (from 32.2 percent to 20.7 percent), rural poverty remained at approximately 40%.3 Poverty is estimated to have worsened since 2005. Estimates put the number of Yemeni who fell below the poverty line as a result of the food crisis, followed by the financial crisis and its impact on the real economy, by some 10 percentage points. It is estimated that now, half of the rural population is below the poverty line.4 Rural poverty is also quite unevenly distributed: in some rural districts, such as many districts of the Governorates of Amran, Al-Baida, Shabwa and Abyan, more than 60% of the rural people are poor (more than 70% in some of Amran Districts), while in others, poverty incidence is less than 20% (Al-Maharah). The growing disparities are not only among rural-urban lines or between rural areas, there are growing social disparities as well. As the Yemen Country Social Analysis underlines, the gap between rich and poor is widening, along with the concentration of economic and political power, and the weakening of the traditional systems of social cohesion, governance and accountability, while "modern" rules to guarantee good governance and accountability are not fully operational yet. Objective of the study This study purports to: (i) identify and analyze the livelihood strategies that poor people rely upon to cope with unfavorable environment and difficult circumstances; and (ii) understand the factors that are intricately linked with keeping rural people trapped in poverty or conversely what helps them get out of it. How are rural household livelihood strategies associated with household assets (land, labor, physical and human capital) as well as location assets (geographic locations, agro-climatic conditions and existence of basic infrastructure and services)? And how livelihood strategies in turn influence their levels of income and poverty? 1 Rural is defined as any agglomeration of less than 5000 habitants. 2 World Bank (2007). Poverty Assessment Report. Poverty is measured on the basis of household consumption, not income. Each individual has an expenditure level. An individual is considered as poor if his/her level of expenditure is below a "poverty" line defined according to Yemeni specificities. Household expenditures were collected during a one month survey, with interviews four times a week with household members, recording their weekly as well as infrequent expenditures. 3 The decline is a meager 2.4 percentage points (from 42.5% to 40.1%), which is actually not statistically significant 4 Urban poverty is also estimated to have gone up to 31.1%. See Breisinger, C., M.H. Collion, X. Diao, and P. Rondot (2010). Impacts of the triple global crisis on growth and poverty in Yemen. IFPRI Discussion Paper # 955. The analysis uses a CGE model to trace the effects of the three crisis. v On the basis of the analysis, Government and donors should be able to better support poor rural people livelihood strategies through public policies and investments programs. The analysis should ultimately result in better-tailored pro-rural poor program design and public resource allocation. It is hoped that the report would inform the elaboration of the Fourth Socio-Economic Development Plan for Poverty Reduction, currently under preparation and that the recommended priority actions would be reflected in the Plan. Methodology We base the analysis of the rural livelihood strategies on the 2005 Household Budget Survey (HBS) and various Geographic Information System (GIS) data layers from global and local sources. GIS layers are spatially summarized at the district level, converted to a standard data format, and then merged with the HBS data for descriptive and econometric analyses.5 In light of the results of this analysis, we review government and donor rural development programs as well as public investments at the district level with a view to recommend ways that these programs could be better adapted to the rural situations that we describe. Finally, we analyze various agricultural value chains, in order to identify the constraints and opportunities for their promotion. Since we do not have any time series data (the data from the 2005 HBS cannot be compared to the ones of the 1999 HBS because of different sampling frame), we unfortunately cannot identify any trend or evolution. Hopefully, when the data from the next HBS are available, trends can emerge through a comparison between the two set of data. Spatial inequalities The majority of the rural poor are located where the incidence of poverty is the highest. According to their poverty incidence (i.e the percentage of individuals below the poverty line in a particular district), we classify districts into "rich" (less than 30% of poor people), "poor" (more than 60% of poor) and "medium-poor" districts (between 30 and 60% of poor). "Poor" districts combine high poverty incidence and high number of poor, i.e a large share of the rural poor live in those districts (44%). This is quite different from many countries, where the regions with a high percentage of the population below the poverty line, tend to be the least populated, with the result that the majority of the rural poor live not in these high poverty incidence regions but in regions of low poverty incidence. In Yemen, it is the opposite: the majority of the rural poor are located where the incidence of poverty is also the highest. This helps spatially targeting public investments, as described under the proposed Rural Poverty Alleviation Program. "Rich" districts are better endowed with natural resources and location assets; population density (per ha of arable land) is less. In contrast with high-poverty incidence districts, rich districts are better endowed with geographical assets, in particular they benefit from good agro ecological conditions for crop production. More favorable agro ecological conditions result in more profitable agriculture. In addition to better agro ecological conditions, "rich" districts are not as densely populated: the average 5 We use ArcGIS Spatial Analyst Tools. vi landholding size is 1.4 ha per household (versus 0.6 ha in poor districts). Also more households have access to irrigation (32% versus 21%). The importance of agriculture is a distinctive feature of a rich district economy. With adequate land size, and better agro ecological endowment, compounded with access to irrigation which secures returns to agricultural inputs, investing in agriculture becomes a viable economic activity. Thus crop production plays an important role in a "rich" district economy. Together, crop and livestock production represent 47% of rich districts revenues. In rich districts, entrepreneurial activities are the second most important source of income after agriculture. A thriving agricultural sector stimulates the rural non-farm economy, through opportunities to develop rural enterprises based on backward and forward linkages to the agricultural sector. A more developed entrepreneurial sector can also be linked to more favorable economic location assets. "Rich" districts are more likely to be situated on the border with Oman or Saudi Arabia, or on the coast. Both locations provide opportunities for developing businesses through trade or through an additional economic base, with the fishing sector. We confirm this through calculating an index of rural market potential: this index is 150% higher for "rich" districts than for "poor" districts. Worse agro-ecological conditions, less access to irrigation, more people per ha of arable land in poor districts. Water is scarce everywhere in Yemen. In addition, it is inequitably distributed across regions, with water scarcity most extreme in the Western half of the Highlands, and to the East of the mountain range, resulting in unfavorable agro-ecological conditions for crop cultivation. Poor districts are more likely to be found in these areas. In addition, in poor districts, the arable land supports three times more people than in rich districts. With worse agro-ecological conditions, smaller cultivated land sizes (0.6 ha per household) and less access to irrigation (21% of the households), crop production, whether for sale or for home consumption cannot be the mainstay of the economy. The characteristic of a "poor" district economy is thus more its livestock production (comparatively), as a result of the fact that poor districts are more likely to be located in pasture lands and/or in marginal areas for crop production without irrigation. Agriculture (crop and livestock) is still the number one source of wealth as in rich districts, but it represents only 28% of a ,,poor district economy. Wage labor, associated with migration is a key characteristic of a "poor" district economy: wage labor income represents close to 40% of the "poor" districts economy, (compared to 20% of ,,rich districts economy). Because of lack of local economic opportunities (due to a less vibrant agriculture and less favorable location assets), households which have to rely on wage labor, have to send their adult men away from the villages, in search for work. This is the more the case in "poor" than in "rich" districts: in "poor" districts, half of the households send one of their adult male in search for work outside the village, only a third of the households in "rich" districts. vii Better access to basic infrastructure and services in districts with higher household income 6 . Though access to basic infrastructure and services is inadequate in most districts, in districts featuring higher household incomes, access to services tend to be better on average than in districts with lower household incomes, most likely because families with higher income levels, can afford to send their children to school, pay for health services, or invest in some kind of water supply or sanitation. Inter household inequalities Rural household livelihood strategies are diverse: some households make a living mainly from farming, while others are mainly entrepreneurs, workers (on- or off-farm), government employees, fishing entrepreneurs or remittance-dependent. We classify the households according to their main source of income, i.e. the income source that constitutes at least 60% of total income.7 For many households (36%), the income source is diversified, meaning by our definition that there is not a single source of income that makes up for 60% of the family income. These households often include three generations. The younger adult men are looking for another income-generating activity than that of their father, either because the fathers activity does not generate enough revenues to support more than one nuclear family, or to gain economic independence. The remaining households are mainly farmers (20%), entrepreneurs (9%), government employees (10%), private sector workers (13%), agricultural workers (4%) remittance- dependant households (8%) and fishing entrepreneurs (about 0.5%). Household incomes, poverty levels and livelihood strategies are all related. We classify the rural population according to the level of household income, in five quintiles, from the poorest to the richest. Farmers and entrepreneurs (including fishing entrepreneurs 8) are over represented in the richest category. Farmers are also over represented amongst the poorest quintile. Government employee households tend to be found in the middle- class category (third quintile). Agricultural workers and remittance-dependent households are among the poorest, both in terms of income as well as poverty incidence9. Poverty is also widespread amongst the non-specialized households. While poverty incidence is the highest amongst agricultural workers, the bulk of the rural poor (poverty mass) is found in the non specialized household category.10 Thus activity diversification appears to be a coping strategy for poor households. Land access is extremely unequal, with inequality compounded by unequal access to irrigation. While 90% of the rural households are involved in some agriculture or 6 Poverty level and income are different measures. As mentioned in footnote 2, poverty is measured per individual, on the basis of expenditures. Income is the household income, not individual income 7 The 60% cut-off is somewhat arbitrary: it is taken from the WDR08 Agriculture for Development, and was recommended by Profs. Alain de Janvry and Elizabeth Sadoulet, University of California, at Berkeley. 8 Fishing entrepreneurs operate a fishing boat, whether they own it or rent it. Fishing entrepreneurs, a very small percentage of the rural population (0.5%) are by far the richest on average. 9 Poverty is measured at the individual level, while income is measured at the household level. Hence the income of an household could appear adequate, but because of the household large size, the household members could be below the poverty line. Also we need to keep in mind that poverty is measured on the basis of expenditures, not on the basis of income. 10 This is because while the percentage of poor is higher in the agricultural worker category, there are more people in the diversified household category (4% versus 36%). viii livestock production 11 , only 63 percent of them have access to land. 12 The average cultivated land per household with access to land is quite small (0.98 ha). It is also very unevenly distributed. With a land value gini coefficient of 0.68.13 Yemen has one of the most unequal land distribution in the world.14 Twelve per cent of the households control 80% of the land. Households with large landholdings are more likely to have access to irrigation (almost 40%, while the rural average is 26%). Indeed there is a very high correlation between access to irrigation and income, a higher correlation than between access to land and income. Very large landholdings are concentrated in 9 districts of the governorates of Al-Hodeidah, Dhamar and Lahej. In these 9 districts, the average landholding size is between 2 and 17 ha. With the interaction between water technologies and land access, land access inequality could become worse. A recent survey of land issues suggests that there is a trend towards concentration of land and water access. In spite of the Law on water sector reforms which has been enacted since 2003, farmers with financial resources and large landholdings continue drilling or deepening wells without authorization. 15 As a result, the aquifers are overexploited, the water table lowers, adjacent wells and lower elevation springs dry up, forcing adjacent farmers without the resources to keep up with the scramble for groundwater exploitation, to abandon their fields or sell them, with worsening land access as a result 16. Undermined traditional land customary rights also results in increasingly unequal land access. An important institution for the land-poor is waqf land. 17 Though waqf land represents only 10 to 15% of all agricultural land, it is an important opportunity for land-poor farmers to access land as tenants. However, the administration of waqf land has shown serious problems due to weak and unaccountable management and lack of inventory, thereby facilitating the private appropriation of charitable waqf at the expense of the land-poor farmers.18 Land access combined with access to irrigation, is strongly correlated with higher household income through the lucrative qat and fruit production, or through providing the initial asset for starting a business. Only 6% of all rural households produce and sell horticulture products, 25% sell qat. While qat is produced by households in all income categories, more households in the richest quintile sell qat (31% of the HH in that quintile versus 18% in the poorest quintile). Qat contributes a much higher share of the richest household monetary income than it is the case for the poorer income quintiles (50% of 11 The 2004 Population Census gives a higher percentage, 95%. Central Statistical Office, (2004). Population Census. Ministry of Planning, Yemen. 12 Land access is defined as land owned, plus land rented in or share-cropped, minus land rented out. 13 The gini coefficient of land access, calculated on the basis of land size is even higher, 0.74. However, a gini coefficient calculated on the basis of land value is more accurate as it takes into account the difference associated with land quality. 14 Yemen land distribution is akin to that of the most unequal land distribution in Latin America countries. Republic of Yemen and World Bank (2005), Country Social Analysis. World Bank Report No. 34008-YE. p. 21. 15 World Bank (2009). Land Tenure for Social Economic Inclusion in Yemen: Issues and Opportunities. Mimeo. 16 Ibid. 17 Waqf land is private land which its owner placed perpetually in a trust for religious or charitable purpose. Waqf land cannot be sold, donated or inherited, but it is rented out to provide income for the beneficiaries of the waqf. 18 Ibid. ix the rich household agricultural monetary income comes from qat, versus 36% for the poorest). The production of fruit for exports (banana and mango) or baladi19 wheat for sale is associated with large landholdings. Land and business income are also related: the business income of entrepreneur households with land is almost twice that of entrepreneur households without land. The hypothesis is that revenues from agriculture enable households with sufficient landholdings to gather the start-up capital for investing in business activities. 20 Because of the link with qat and horticulture production, land access has an income inequality increasing effect though the inequality increasing effect is less in the case of qat than in the case of horticulture production. Horticulture and/or qat sale are clearly associated with high household income. The gini coefficient of crop monetary income, without qat, is 0.76, a higher gini coefficient higher than that of land value (0.68). The gini coefficient of qat monetary income, 0.63, is lower than the gini coefficient of land. Therefore, interestingly enough, qat has a less income inequality increasing effect than other crops, probably because anybody can produce qat as long as the agro ecological conditions allow it. The efficiency of qat production and marketing is not related to quantities produced and therefore not related to land size.21 Contrary to land access, animal production has a decreasing income inequality effect, evidenced by the gini coefficient of livestock monetary income of 0.52. Animal production is 14 times more powerful at reducing poverty than qat and horticulture production. 22 While obviously households with more land also own more animals, animal ownership is less unequal than land access. Even households with no land, or tiny plots, own a few sheep and goats, and some even have cows. A much higher percentage of rural households sell livestock and livestock products (48%) than qat (25%) or horticulture (6%). Live animal sales play a more important role for the monetary income of the poor: 34% of monetary income for the poorest HH,23 versus 12% for the richest. Indeed, amongst the richest farming households, the sources of monetary income are quite diverse: livestock, qat, horticulture and, depending on the agro ecological zone, baladi wheat. The average gross income from the sale of livestock is about half of the income of from the sale of qat, and a third of the income from horticulture. 24 However, livestock as a source of income for the poor could be threatened by private appropriation of communal land. People without access to land or whose plots are small are largely dependent on communal and public land for livestock grazing. Customary law entrusts shaykhs with the management of communal land. In recent years, land speculation has increased. There is a growing confusion over the communal 19 Baladi=local or indigenous. Baladi wheat sells for very high price in the Gulf countries or is bought for special occasions by rich Yemeni families. To wit: it is not sold in the regular market with other cereals, such as the imported wheat, but in specialty stores that sell spices. 20 While we do not know in which direction the causality operates, it is likely that rural household with land have been able to invest part of their agricultural revenues in businesses (though one could argue the reverse proposition: successful entrepreneurs invest in land). 21 The average income from qat sale for qat farmers is 262 000 YR21, while the average income from horticulture for horticulture producers is 386 000 YR. 22 See econometric analysis in Chapter III. 23 This is approximately equivalent to the percentage of the revenues coming from qat for the poorest households. 24 The average gross income from the sale of livestock for those who sell animals is 126 500 YR. x land entrusted to shaykhs and the land they own privately resulting in private appropriation of communal land.25 Community rights to communal land are often not documented which makes poor people whose livelihood depends on communal land particularly vulnerable.26 Very small landholdings combined with: (i) erratic patterns of rainfall and limited access to irrigation; and, (ii) the cultivation of qat as a high value crop, result in extremely low levels of food self-sufficiency. Only 31% of the rural households produce some of the food they consume, meagerly covering 10% of their needs. More HH in the richest quintile produce some of their food, but the coverage remains surprisingly low: 13%. Given the high value of qat relative to food crops, which in addition benefits from an extremely efficient value chain, it makes economic sense for farmers to use the small amount of land they have to grow qat for sale and buy their food. Migration plays a key role for those with limited assets. For those who have to rely on their labor for a livelihood, employment opportunities are very limited in rural areas, and they have to resort to migration to the cities or to other countries. Overall, almost one quarter of the adult male population is migrating for work, often for very long periods, more than half of the year, while the rest of the family stays in the village. Salaried-labor households and remittance-dependant households display the highest rate of migration. In these categories, 30 to 38% of the adult males migrate, while only 16% migrate in the farming or enterprise category, and 21% in the diversified-income households. Migration, household income, and household size are all correlated. Larger households with migrants are less vulnerable. Household income is correlated with migration: high-income households have more adult migrants. However, one should be careful not to deduct that migration necessarily results in higher levels of per capita income for the household: a third factor comes into play here, which is household size. High-income households are larger (10.2 members on average for the richest quintile), almost double the size of the low-income households (5.6 members for the lowest quintile). In particular, they have more adult men and lower dependency ratio. 27 With more adult men working, the total household income is necessarily higher, with or without migration. What is worth highlighting however is that these larger households can afford to send one of their adult men in search for job opportunities more so than smaller households with only one active male adult. Larger size households, with more working adult men are likely to be less vulnerable to shocks (whether internal or external) as they are able to diversify their income sources, and more able to use the economic opportunities derived from migration. Education levels are generally low in rural areas except for government employees. The number of years of schooling of male adults is strongly associated with higher household income and lower poverty incidence. Male adults achieve 5.9 years of schooling on the average. The lowest level of education is to be found amongst farmer 25 Some shaykhs have sold land alleged to be communal outside the tribe, a contravention to customary norm. World Bank (2009). Land Tenure for Social Economic Inclusion in Yemen: Issues and Opportunities. Mimeo, p 42. 26 Ibid. p 46. 27 We define dependency ratio as the number of active adult men divided by total household members. xi and agricultural worker households. The number of years of schooling of adult men in government-employee households is 10.7, substantially higher than the rural average. Education is associated with higher incomes, by and large through the opportunity for government employment that higher levels of education open. Nevertheless, educational attainment is also important for the welfare improvement of other households. Years of schooling of male adults are strongly associated with higher household income and lower poverty incidence even for a subset of rural households not in the government sector. 28 Migrant workers are not necessarily better educated. One could expect migrant workers to have a higher level of education than those remaining in the village. It is the case for the heads of the household: they are on average better educated than the heads of households not migrating, but they are also younger. However, surprisingly, it is not the case for the other adult men migrating: their level of education is lower than those staying in the village and their age is not different from the ones not migrating. Women educational attainment is particularly dismal. Rural women educational levels are extremely low, much lower than that of men (1.3 years on average versus 5.9 years). Rural male education levels are improving vis-à-vis those of urban residents, but not those of rural women. The gap is not closing either at younger ages. Rural girl education levels remain quite substantially lower than that of boys: approximately 3 years of schooling for 15 year old-girls versus 6.5 years of schooling for a 15 year-old boys. Women educational attainment in the richest quintile is higher (1.6 years of schooling on average) than that of the poorest quintile (0.9 years of schooling), which most likely reflects the fact that richer households can afford to send their girls to school29. Most rural women (85%) say they work on domestic tasks only. These tasks are likely to include raising animals30. Rural women are very much left out of the labor force. Only 0.6% of them earn a salary either in agriculture or off-farm. Fourteen percent participate in the activities of the family farm or family business, in addition to their domestic tasks. Very few are heads of households, only 3% on average, mostly in the lowest income categories (10% in the lowest quintile). Cash transfer programs are not well targeted. We look at the cash transfers from three social programs: the government Social Welfare Fund, the assistance from the tribal authority affairs, and the assistance from charitable organizations. None of these programs are well targeted, though the charitable organizations are better at targeting the poorest households, the tribal authority affairs being the worst. Only 48% percent of the Social Welfare Fund transfers go to families in the two lowest income quintiles. The negative social impact of the narrowing per capita resource base due to rapid population growth is likely to be exacerbated by rural inequalities. With a population that still grows at a rate over 3% a year, per capita land and water resources are 28 Here again we cannot infer the direction of the causality: richer households may be able to afford sending their children to school more than poorer households, or else, households whose members have a higher level of education can obtain better employment opportunities, or be more successful whatever their economic activities are. 29 This assumes that women of rich households come from rich households. 30 We include raising animals in women domestic tasks though the HSB questionnaire does not give any explanation of "women domestic tasks. xii dwindling. With landholdings already very small, the young generation is going to inherit even smaller pieces of land. In addition, with underground water availability becoming increasingly scarce, famers may have to return to rainfed farming and make better use of water harvesting systems. The situation is going to be worse for poor rural people who do not have other alternatives than farming, whatever small piece of land they have. Land and water access inequalities are likely to have an exacerbating effect on the impact caused by the narrowing per capita natural resource base. Yemen urgently needs a successful exit strategy that can take massive amounts of rural people out of agriculture. To sum up: coping strategies of the asset-poor households We reckon that a large proportion of the rural households, more than 75% of the rural households, are in a situation of limited assets which makes them dependent on their labor for a living. Even if not of all them are poor, most of the non poor are living at the margin of subsistence, which makes them extremely vulnerable to shocks: disease of an active family member, loss of employment, natural disasters, droughts and price fluctuations, as evidenced by the impact of the 2008/9 food and financial crisis. Poverty and vulnerability are not associated with a particular livelihood, but with lack of assets. There are more poor people amongst the households whose main source of income is salaried labor as well as amongst the diversified-income households. Households with limited assets cope with a combination of the following sources of income, though as we saw, one source of income may dominate: For those with access to land, some subsistence production (cereals and legumes), and limited crop sales (mainly, depending upon the location of the household, qat or coffee); Animal production for monetary income, even for those without access to land to grow crops; Either salaried labor, associated with the migration of the adult men; or Permanent out migration of part of the family, while the family members remaining in the village rely on their remittances. How can these households improve their livelihood and what government and donors can do to support them? This will be the subject of the following section. A rural poverty alleviation program Building upon the results of the analysis, the rural poverty alleviation program proposes six strategic areas. Basic infrastructure and services. Our analysis has shown that beyond the rural-urban poverty gaps, spatial differences within rural areas are considerable, with rural poverty being concentrated in some districts/governorates. Basic infrastructure and services, as measured by the District Access index vary considerably from one district to the next. xiii Reducing poverty in rural Yemen requires among others, better access to quality social services. Protection of poor-people assets. In a context of rapid population growth and resulting intense pressure on natural resources, compounded by the fact that rules of law are being blurred - the formal system of governance is not being well established yet while the customary one is weakened- the poor and not well-connected are put at a disadvantage. They are likely to lose out to more influential members of the society in case of conflicts or ambiguous rights. Whatever few assets poor people have in terms of land and water, need to be protected. Other assets that need protection is livestock: animals constitute poor people main savings instrument and insurance against risks. Hence the importance of a well functioning animal health protection system. Finally, one aspect of poor peoples assets which is often overlooked is their arable land. Poorer farmers are more dependent on rain for crop production than better-off farmers. Rain-fed production in Yemen is by and large dependant on an elaborated water and soil harvesting system of terraces and cisterns, which represents thousands of years of labor. Poor farmers may not have the resources to maintain it and need help. Enhanced and secured agricultural productivity. Poor people agricultural activities can be made more productive, and productivity needs to be secured, given the risks associated with rainfall uncertainty, heightened by climate change. Animal production can be enhanced through better animal health and nutrition. Food security can be expanded somewhat, with enhanced drought-resilience and higher crop productivity, that can be achieved through enhanced soil and water management, and through low cost technologies such as better seed management, using more drought-resilient varieties. Some households can make a better living with their coffee; more households can be involved with honey production. Successful exit out of agriculture and rural areas. Yemeni rural labor, as we saw, is largely unskilled labor. An essential aspect to improve the livelihood of rural people whose only asset is their labor will be to help them access more and better economic opportunities either in the rural non farm economy or through migration to urban centers or abroad. However, finding and maintaining employment requires broad-based occupational skills or specific job-related skills. Hence, investing in the skills of young rural people is essential to prepare them (as well as the next generation) to migrate successfully out of overpopulated rural areas. A better future for poor people's children. Many poor families cannot afford to send all their children to school, girls are the most affected. Economic and other social factors keep girls away from school. Concerted efforts are required to solve the issue of girls school attendance, if the next generation of adult women is going to be empowered through education, which will have a positive impact on the well-being of the entire family. xiv Social safety nets. Many poor people will continue to depend on social safety nets, either in the form of labor intensive public works program, or in the form of cash transfers. Indeed, a number of poor households, especially the ones in extreme poverty (aged, disabled, and/or destitute female-headed households) will continue to depend on cash transfers to survive. The objectives of the proposed program for the rural poor are presented thereafter. The issues, objectives, activities and responsible entities are summarized in the Table below. xv Increase investments in districts with the lowest access and highest poverty incidence Improve Access to basic services Improve local governance and capacity of district councils to manage resources efficiently and target the poor Implement the water sector reform together with pro-poor water programs and improved Protect poor people limited groundwater local governance assets Enact rural land regulatory framework to ensure access to waaf and communal land Secure crop production in rain fed areas through water harvesting and soil conservation Enhance and secure the productivity Invest in animal health and nutrition programs of poor people assets Improve cereal and legume production in rainfed areas Create more local value added in lucrative value chains: coffee and honey Ensure a better future for poor Enhance school attendance especially that of girls people's children Enhance the quality of education, with a specific focus on vocational training Improve access to more and better employment opportunities Invest in labor intensive rural public work programs Enhance the social protection Improve the targeting of the Social Welfare Fund system xvi Objectives and Priority Programs to Address the Need of Rural Poor Issues Objectives and Actions Responsible organization Spatial inequalities Increase public investments in the rural districts with highest - Ministry of Local poverty incidence and lowest Access index Administration - Spatial differences within rural areas are considerable, with rural poverty being concentrated in some -through: (i) budget transfers (Decentralization Law); (ii) - Social Fund for districts/governorates, both in terms of poverty rate and investments though SFD and PWP with spatial targeting. Development mass. - 68 districts qualify for increased per capita investments. - Public Works Program - Access to basic infrastructure and services, as measured Improve the capacity of district councils to manage public by the District Access index vary considerably from one funds efficiently and address the needs of the poor and district to the next. vulnerable groups. - Households in richer districts have better access to basic - Worth up scaling: the "Empowerment for Local Development" services. Program of the SFD. Access to irrigation is inequitable and could become worse Implement the Water Sector Reforms together with pro-poor - Ministry of Agriculture and with the effect of irrigation technology water access programs and enhance groundwater local Irrigation governance - Farmers with financial resources and larger - Ministry of Water and landholdings, dig deeper wells. The rapid drawing of - Improve the capacity of local institutions to implement the Environment groundwater has dried springs and wells, forcing groundwater law, through local collective action (community - National Water Resource adjacent or downstream smaller farmers who cannot water management and water user associations) as part of the Authority afford deep well-drilling to abandon their fields. implementation of the National Water Strategy and Investment Plan. - There is a lack of effective local institutions for groundwater governance. - Implement the water sector reform as a package, together with pro-poor water access programs and pro-poor entry criteria in - Water sector reforms have been enacted since 2003 but order to enhance equity. implementation is hindered by political economy constraints. xvii Land access is skewed and could become worse with Protect poor people access to waqf and communal land through - Ministry of religious growing competition for land enacting rural land regulatory framework Endowment and Spiritual Guidance - Poor people depend on communal and waqf land for - Develop an inventory of waqf land and develop regulations to access to land. For example, communal land and land in improve the management of waqf land - Ministry of Interior upper catchment areas are important resources for - Develop an inventory of State land and mobilize this land to - Ministry of Social Affairs grazing animals for people with limited access to land. improve land access to land-poor farmers. There is uncertainty as to the legal status of such land, and cases of private appropriation are occurring. - Document customary rights in order to protect communal properties from private appropriation and clarify rights of the - Waqf land administration is affected by weak supervision communities on this type of land. and lack of inventory, thus facilitating private appropriation of waqf land. - Enforcement of customary rights on these type of lands, in particular communal lands is weakened, with the result that land-poor farmers could lose access. Under low and erratic patterns of rainfall, farmers that Secure crop production and increase resilience to climate - Ministry of Agriculture and do not have access to irrigation are highly dependent on change through water harvesting and soil conservation Irrigation soil and water harvesting systems programs in rain fed areas with low and erratic patterns of - Social Fund for rainfall, using a watershed management approach, including: (i) - Poor rural families are highly dependent on rainfall Development rehabilitation of terraces; (ii) reforestation of the watershed upper availability for their crop or livestock production, either catchment areas; (iii) construction or rehabilitation of cisterns for because they live in areas with poor agro ecological human and animal consumption (could be linked to the conditions, or unfeasibility of irrigation or because they introduction of a drip irrigation system for supplemental do not have the resources to invest in irrigation. irrigation of high value crops); (iv) wadi bank protection works; - Rainfall is limited and highly erratic in many areas of and, (v) the improvement of small spate irrigation schemes. Yemen, and is likely to come with flash floods. Rainfall Worth upscaling: the Rainfed Agriculture and Livestock project ; variability is likely to increase with climate change. the Groundwater and Soil Conservation Project. - Yemeni people survived through centuries thanks to their careful harvesting and conservation of water (and soil) through terraces and management of run-off water (wadi, spate irrigation). However this system is in jeopardy because of migration of adult men (better opportunities for labor and low land productivity) and the costs it represents to maintain these infrastructures. xviii Livestock is poor people's best asset, an asset that they can Protect poor people's assets through animal health and - Ministry of Agriculture and easily lose and that is not as productive as it could be. livestock nutrition programs: Irrigation/General Directorate of Animal - Animals are the best assets of poor people, providing - upgrade the public animal health services, progressively Resources cash income and savings. privatize parts of the services, and extend the present coverage limited to peri-urban areas, through a system of veterinary - Animal health and nutrition are poor. auxiliaries or community animal health workers (ref: the - The availability of veterinary services is low in rural Rainfed Agriculture and Livestock Project, consider up- areas. scaling). - Forage and fodder availability is limited (rainfall and - improve animal nutrition through: (i) better management of the water availability constraints). resource constraints (water, forage and fodder); (ii) fodder productivity improvements and affordable diet supplementation (such as bone meal); (iii) improved management of the upper catchment of the watersheds, introducing animal edible species (napier grass, legume trees) as well as management of other communal rangeland. Cereal and legume productivity in rainfed areas is very Enhance the productivity and climate resilience of rain fed - Ministry of Agriculture and low, well below technical potential and below farmers' cereal and legume production through establishing a farmer- Irrigation/Generals seed yields in comparable environments based seed production system based on local landraces: Multiplication and Conservation Center - Farmers in highy risky environments cannot afford to - Indigenous variety improvement and maintenance. invest in improved technologies. - Social Fund for - On-farm seed production and conservation. Development - One of the reason for low productivity is the deterioration - Establishment of seed producer groups. of the quality of the seeds being used. - Worth up-scaling: (i) the Rainfed Agriculture and Livestock - The availability of veterinary services is low in rural Project; (ii) the Agro-Biodiversity and Climate Adaptation areas. Project. - Better seed management of local varieties is a low cost/low risk technology that can be easily adopted and that could help farmers withstand the shocks of climate change. xix More rural families could be involved in honey production Create more local value added in lucrative value chains: coffee - Ministry of Agriculture Coffee value chain is not effective and honey through: Though relatively few rural households are involved in coffee and honey production (about 2.5 % of the rural HH in - Addressing production constraints each of the value chains), there is potential for improving - Increasing capacity of producer organization; producers income in these two sub sectors and increasing rural employment in the value chains. - Promoting transparent market transactions and timely information - improving productivity (equipment for producers, technical advice); - Introducing and enforce branding and traceability; - Introducing product differentiation; Too many children, especially girls are still not enrolled or Enhance school attendance especially that of girls : School - Ministry of Education drop out of school. planning and design addressing parents' concern regarding - Social Fund for girls' attendance: - Poor families cannot afford to send their children to Development school as they need them to help with domestic tasks, - building smaller schools closer to girls homes; especially girls for household chores, or boys working - involving communities in all aspects of project planning, design, and on the farm. management, especially regarding school location; - The gap between boys and girls education level is not - obtaining the communitys commitment to enrolling girls as a prerequisite to school construction; closing though girls enrollment has improved. There are many social issues regarding girls school attendance. - including water, sanitary facilities and boundary walls; - Education is the only way out of poverty for poor - providing separate classrooms for girls in grades 7-9; peoples children. - building separate secondary schools for girls; - Higher level of education of women will contribute to - increasing the number of qualified female teachers and providing them lower fertility rates, and decrease demographic pressure with special incentives to work in remote rural areas. on limited natural resources. - Conditional cash transfer to offset opportunity costs of sending girls to school and reduce girls' drop out xx Opportunities for employment are very limited in rural Invest in quality education, with a specific focus on vocational - Ministry of Technical areas. Rural people lack skills training, Education and Vocational Training - Rural households lacking assets depend on salaried labor - targeting quality and relevance, through closer linkages with for their livelihood employers (to develop programs, etc..) - Social Fund for Development - Opportunities for employment in rural areas are very - aligning the sector with signals from the market and taking limited. People have to migrate abroad or to urban areas. into account not only the needs of the Yemeni labor market, but also the labor markets needs in Gulf countries and Saudi - Men migrating are typically employed in low paid Arabia. unskilled jobs (construction and services). Invest in labor intensive rural public works programs with a - Yemeni enterprises complain that they cannot find food for work or cash for work approach. Particular reference is qualified workers. the Labor Intensive Work Program of the Social Fund for Development. Extreme poor and vulnerable households will continue to Develop better targeted social welfare programs adapted to -Social Welfare Fund depend on cash transfers in order to survive. The problem rural specificities. Review the list of beneficiaries of the cash is that the Social welfare Fund is not well targeted transfer program and improve the management of the Fund. xxi 31 %73 : 32 .33 %84 32.2 ( 11.6 2005 - 1998 .34 40 ) 20.7 .2005 . 10 .35 : ) 70 ( 60 .) ( 20 . "" . )( : )( . 31 . 5000 32 . . )2007( 33 . )%40.1 %42.5 ( 2.4 34 P. .X M.H. Collion C. Breisinger %31.1 35 .)(2009 1 ( . - ( ) ) . . . 2005 . GIS )HBS( .36 HBS . ( . ) 1999 2005 . . . 37 ) ( ) 30 ( " " ( " " ) 60 ( "" "" .) 60 30 .) 44( . ArcGIS 36 . . 37 ( . .)2007 2 : . " . . " "" . . "" 0.6 ( 1.4 .) .) 21 32 ( . . . 47 . . . . . %150 . . . . ( ) 0.6( 3 ) 21 . )( . / ) ( ."" 28 . : 20 ( 40 .) : "" ."" .38 . : ( . ) 37 .39 60 . 60 ) ( 20( . 13( ) 10( ) 9( ) . 7 . 38 2008 : %60 39 . 4 ) 8( ) 4( ) .) 0.5( . . . ) 40 ( . .) ( .41 . .42 ) ( . . .44 63 43 90 0.98 0.68 . 45 46 80 12 . 40( . .) 26 . 9 . 9 . 17 - 2 . )%0.5( . 40 . . 41 . . %4( 42 .)%36 : .%95 2004 43 .2004 . 44 0.74 45 . )2007( . 46 .21 5 . "" 2003 .47 .48 . .49 15-10 . .50 6 . . 25 .) 18 31( .) 36 50( . 51 ) ( : . 52 . . . . . : )2009( 47 Mimeo . 48 . 49 . )2009( 50 . . = 51 . 52 : ( ) 6 .)0.68( 0.76 0.63 . . .53 .0.52 .54 14 . . ) 48( .) 6( ) 25( 55 12 34 : : . . .56 . . . .57 .58 )( : )( %31 . 262,000 53 86,000 . 54 . 55 126,500 56 . 57 .42 . : )2009( .46 . 58 7 . %10 : . %13 . . . : . 16 38 30 . 21 . : . . . : ) 10.2( .) 5.6( .59 . . ) ( . . . . 5.9 . . 10.7 . . . 59 8 .60 . . . . : 1.3( . .) 5.6 . 3 : . .15 6.5 15 ) 1.6( ) 0.9( .61 . 85 0.6 . .62 14 . 3 . .) %10( . : . 48 . . 3 . . : 60 . . 61 " 62 ." 9 . . . . . : 75 . : . . 2009/2008 . . : ( ( ) ) . . . ./ . " " 10 . . - . - . . . : . . . . . . . . . . . . . . . . . . . . . . 11 ) / ( . . . 12 : 13 - - - - / - ) ( ) ( : - )( - 68 - - - . . " " : - - - ( . . ) - . . - 2003 - 14 - - - . - . . - . - . - )( )( )( ( )( )( ) - . : - )( .) ( ) ( . 15 / : - - - ) ( : - - )( ) ( )( ) ( ) ( - ) ( . : . . )( : . )( / . 16 : (value : )Chain 2.5( ) - - ) ( - - - : - - - - - - - - 9-7 - . - - . 17 ( - - ) - . - . - - . . . .5 : .5 : 2.5( - ) 2.5( - ) ) ( - - - 18 .6 : - - - - - - - 9-7 - - . - - . :7 - - ) ( ). ( - - . . .8 . . . 19 .9 : - - - . - - . . - . ( ) - . . . . 20 1. BACKGROUND, OBJECTIVES AND METHODOLOGY 1.1. Prosperity has by-passed rural Yemen. Yemen is predominantly a rural1 country and poverty is by and large a rural phenomenon: 73% of The population and 84% of the poor live in rural areas. 2 The risk of poverty for a rural person is double than that of an urban person. The severity of rural poverty is also three times worse than urban poverty. The Yemen Poverty Assessment (2007) shows evidence that the rural economy is stagnating and the rural-urban gap widening. Between 1998 and 2005, while the percentage of poor in urban areas has declined by 11.6 percentage points (from 32.2 percent to 20.7 percent); rural poverty remained at approximately 40%.3 In three of the seven rural regions where nearly 40% of the countrys poor live (Central North, Central South and the Eastern)4 poverty worsened by 15 to 20% points (World Bank, 2007). The decline in urban poverty is mainly due to oil revenue-based government spending, which mostly expands the service sector in urban areas. Urban development seems to have had limited and uneven spill-over effects on rural areas, due to poorly functioning rural-urban linkages. In a number of governorates (such as Hajjah, Lahej and Al-Mahweet) rural poverty increased while urban poverty decreased. Figure 1: Rural-Urban Poverty 1998-2005 1.2. The situation may have worsened since 2005. In 2008, the sharp rise in fuel prices accelerated economy-wide growth, but the oil-driven growth has not been pro-poor5. Poor people 1 Rural is defined as any agglomeration of less than 5000 habitants. 2 World Bank (2007), Poverty Assessment Report. Poverty is measured on the basis of household consumption rather than income. Household expenditures were collected during a one month survey, with interviews four times a week with household members, recording their weekly as well as infrequent expenditures. 3 The decline is a meager 2.4 percentage points (from 42.5% to 40.1%), which is actually not statistically significant 4 Central North: Sanaa, Sadah, Mareb, Al-Jaf, Amran, Rimah; Central South Al-Baida, Lahj, Abyan, Al-Daleh 5 The oil sector generates very few jobs. In 2003, for example, the oil sector employed only 21 000 Yemenis, while at the same time 190 000 new workers entering the job market (World Bank, Development Policy review, 2006). 1 suffered not only from the rise in food prices, but also from sluggish economic activities due to reduced non-food consumption. While food prices have gone down from their high 2008 peak, the financial crisis took over, resulting in a further deceleration of the economy of Yemen as well as those of the Gulf countries. Rural people are now suffering from a decrease in job availability for migrant workers and less remittances from urban residents. Estimates put the number of Yemeni who fell below the poverty level as a result of the two crises by some 10 percentage points, bringing the rural poverty level to 49.8% and urban poverty to 31.1%.6 1.3. The disparities among rural areas differ considerably. The Poverty Assessment Report also highlights large intra-governorate differences of poverty. Rural poverty incidence among governorates varies between 5.4 percent and 71 percent. Poverty is highest in the rural part of Amran governorate, where 71 percent of rural population is poor. Amran is followed by rural Al- Baida and rural Shabwah (60 and 57 percent of poverty). The incidence of poverty is the lowest in rural Al-Maharh and Saadah (6 and 16 percent of poverty). 1.4. The growing disparities are not only along rural-urban lines: there is a widening of social inequalities as well. As the Yemen Country Social Analysis7 aptly points out, the gap between rich and poor is widening, along with the concentration of the economic and political power. This is certainly due to the dramatic changes that Yemeni society has experienced over the last three decades, resulting in a weakening of traditional systems of governance via tribal customs of social accountability and conflict mediation and prevention. Customary mechanisms for protecting property rights are weakened, while not being replaced by yet functional judicial system and formal mechanisms of redressing wrongs. Hence, the poor and politically unconnected are put at a distinct disadvantage, with limited voice and capacity to act within the changing social structures. Productive land in particular is being concentrated in the hands of a small number of powerful families, while the poor have diminishing access to land and water. 1.5. Growing poverty, inequality and patronage also threaten social cohesion. 8 Current systems of social solidarity at the communal levels are stressed as a result of deepening poverty. The situation characterized by formal systems of governance superimposed over customary ones, confers undue power to a small circle of influential people, while traditional systems of accountability, particularly important for the rural poor, are undermined. As power, wealth and influence become increasingly situated in urban areas, rural poor have less opportunity for accessing networks of influence. A growing class-based system of social cleavage, coupled with patronage as the main means of redistribution, risks promoting the fragmentation of Yemeni society. 1.6. The Government is facing daunting challenges to address the needs of the poor. Population growth (over 3% a year according to the 2004 census) is putting tremendous pressure on natural resources and translates into increased demands for services. In addition, since 2008 poverty is on the rise, as a result of the double shock of the food and the financial crisis, with the latter impacting the real economy and translating into less revenue from remittances (especially from the Gulf countries) and less employment opportunities for migrant rural workers. At the same time, the government overall institutional and financial capacity to respond to these challenges is limited. Resources, essentially from oil revenues, are being reduced, due in parts to the decline in 6 Breisinger, C., M.H. Collion, X. Diao, and P. Rondot (2009). The analysis uses a CGE model to trace the effects of the three crisis. 7 Republic of Yemen (2005). p. xv. 8 Ibid. p. xv. 2 oil prices since their peak in 2008, but also to the decline in production, as oil reserves are diminishing. Reduced budget resources place severe constraints on the governments ability to address the needs of poorer sections of society and to improve public service delivery in rural areas, and particularly in poor rural areas. OBJECTIVES 1.7. Given this situation: worsening of rural poverty, the widening of the disparities rural/urban as well as spatial disparities within rural areas, and the loosening of social cohesion, there is an urgent need to better understand the factors that are intricately linked with keeping rural people trapped in poverty or conversely the factors that help them get out of it. The analysis should ultimately result in better informed program design and better-tailored priorities for resource allocation given the constraints on government budget. This paper purports therefore to identify what are the livelihood strategies that poor people devise to cope with unfavorable environment and difficult circumstances and how the Government and donors could better support poor people strategies through public policies and investments programs. In particular, it is hoped that the priorities that result from the analysis would inform the elaboration of the Fourth Socio- Economic Development Plan for Poverty Reduction, currently under preparation. METHODOLOGY 1.8. Rural households combine their assets to devise livelihood strategies, subject to the limitations of their environment. Household income levels and welfare conditions are the outcome. In this way, narrowly defined household assets (lands, labor, physical and human capital) or broadly defined location assets (first and second nature geography, social and institutional assets) affect the extent to which people make a good living or on the contrary are trapped in poverty. Put differently, better geographic locations or favorable ago-climatic conditions attract people, which then leads to infrastructure development. Access to infrastructure, services and markets, then increases the productivity and the accumulation of household private assets. First and second nature of geography interacts, and both influence the productivity of various private and location assets rural households possess. 1.9. Given this asset-based conceptual framework, we examined how Yemeni rural household livelihood strategies, income, and poverty are interacted with and determined by various household and location assets. The analysis is based on the 2005 Household Budget Survey (HBS), and various GIS data layers from global and local sources. Those GIS layers are spatially summarized at the district level, converted to a standard data format, and then merged with the HBS data for descriptive statistics and econometric analyses, using ArcGIS Spatial Analyst Tools. The analysis enabled us to understand the mechanisms, through which rural economy functions, and to identify key drivers of household welfare and coping mechanisms. We then completed the work with an analysis of some sources for rural economic growth, mainly five agricultural value chains: fisheries, honey, coffee, baladi wheat and qat (qat being used as a reference for a well functioning value chain) to identify what are the constraints and entry points for the development of these sub sectors. Finally, in the light of poor people coping strategies, we reviewed the on- going development programs, to identify what should be the priorities, what programs should be reinforced and expanded and what further actions should be considered. 3 ORGANIZATION OF THE PAPER 1.10. The paper is organized in three parts, as follows: On the basis of GIS mapping and correlation with HBS data, the first section analyzes the factors that are correlated with spatial inequalities, in particular agroecological conditions, population densities with regard to arable land, human assets (education, HH demography), migration and access to basic infrastructure and services. 1.11. The second section is devoted to an analysis of inter household inequalities: (i) what are the main income sources; (ii) how sources of income are correlated with poverty, household income and welfare; how sources of income interact with household assets such as access to land, livestock ownership, human assets (education and demography); the role of migration and remittances; how qat and animal sales relate to different income levels. The second section ends with an econometric analysis which combines coping strategies, household assets, and resulting welfare into a consistent analytical framework. This exercise crosschecks the findings of previous descriptive analyses and enables us to bring together more robust results. 1.12. The third and final section draws recommendations for improving and better targeting rural development programs in support of the poor building upon existing programs. While the 8 pillars of the on-going Development Plan for Poverty Reduction (DPPR) remain entirely relevant, the priorities are very broad, and the instruments to achieve the intended results unspecified. We review the on-going programs that are the most relevant given poor peoples strategies. We highlight the programs that should be scaled up and identify priority investments that are likely to have the most impact on poverty reduction. 4 2. SPATIAL INEQUALITIES "RICH" VERSUS "POOR" DISTRICTS 2.1. Spatial distribution of rural poverty. The Poverty Assessment Report highlights huge disparities among districts in terms of poverty incidence. 9 We first attempt to understand the characteristics of "poor" versus "rich" districts. We categorize districts by the severity of rural poverty incidence, defined as the percentage of rural population whose household income is under the household idiosyncratic poverty line.10 If a districts rural poverty incidence is higher than 60% (lower than 30%), it is classified as a poor (rich) district. Map 1 shows the location of the three types of districts. Map 1: Location of poor, medium poor and rich rural districts (poverty rate) 9 Poverty incidence or poverty level is the percentage of poor out of the population, on the basis of the Poverty Assessment report characterization. Hence the poverty incidence of a district is the percentage of poor people out of the total population in that district. 10 The rural poverty definition and data are from the Yemen Poverty Assessment Report (2007). 5 2.2. With this classification, 23% of Yemeni rural households live in poor districts and 41% in rich districts. The average rural poverty incidence of poor districts is 74.2%, compared to 14.9% of rich districts. Almost half (44.2%) of Yemeni rural poor live in poor districts. Both poverty rate and mass11 are high in poor rural districts (Figure 2). In many countries, the mass of rural poor do not live in areas with the highest poverty rate, because these areas are usually the least densely populated. In that sense, Yemen differs: a large share of its poor people is located in districts which have the highest poverty incidence is the highest (Figure 2 and Map 2). Figure 2: Distribution of poverty rate and mass 11 The poverty mass is the percentage of poor out of total poor. Hence the district poverty mass indicates where most poor people live. 6 Map 2: Spatial distribution of rural poverty (mass)12 2.3. Agro ecological assets and poverty incidence. Rainfall quantities, evapo-transpiration, soil temperature and soil moisture determine the length of the growing season and whether crops can be grown with rain only or need supplemental irrigation or full irrigation. Combining these criteria we identify 5 agro-ecological zones (Map 3 and Annex 1 for the data). The black zone with highest rainfall (around 600 to 700 mm rainfall) and longest growing season, allowing for two crops per year has the best agronomic potential. On the opposite side, the green zone requires supplemental irrigation whether summer or winter growing seasons. In between are the red and blue zones, which require some supplemental irrigation for one of the growing season, while for the other season, crops can be produced with rain only, but is nevertheless risky. 2.4. District poverty incidence and agro climatic conditions are correlated. 13 Poor rural districts are located in worse agro ecological conditions: they receive less rain: 86%, relative to that of rich districts. The proportion of cropland cover is also much less: 64% relative to that of rich districts. Poor districts have a higher probability than the rural average (155%) to be located in an agro climatic zone with rangeland only or where crop production requires supplemental irrigation (green zone). Conversely the probability for a poor district to be located in a zone with good agro climatic conditions is 70% that of the rural average. By contrast, the probability that a 12 The bars represent poverty mass, i.e. the number of poor people in the district. 13 Though the analysis shows correlation, it would have been much stronger, had the poverty data been available at a lower level of disaggregation, say sub-districts. The problem with district level poverty data is that districts can spread across two agro ecological zones. 7 rich district is located in areas with good agro climatic conditions where two cropping seasons are feasible is 3.5 times that of a poor district (black zone). Poor districts tend to be located mostly in the governorates of Amran, Shabwa, Al-Baida and the inland parts of Abyan where cropping under rainfed conditions is risky: supplemental irrigation is required and rangeland predominates (Green area of Map 3: Agroclimatic zones). Map 3: Agroclimatic zones Source: Agriculture Research and Extension Authority. The aggregation in five agroclimatic zones is the authors. 2.5. District poverty incidence and main source of income. Table 1 and Figure 3 compare household income structure according to district poverty levels. Given the difference in natural resource assets described above, rich rural districts enjoy more favorable agricultural conditions. As a result, households of rich rural districts have on average a much larger share of their income coming from agriculture: 47% versus 28% in poor districts (Table 1). Income from enterprises represents also a larger share of total income in rich districts: 18% versus 12% in poor districts, a difference that can be attributed to a more thriving agricultural base. Indeed more opportunities in agriculture lead to more opportunities for businesses: agricultural growth stimulates the demand for goods and services from other sectors through backward and forward linkages related to agricultural activities and higher agricultural income levels. Income from agricultural production provides the favorable conditions for rural micro enterprises to begin operations or to expand investment into new activities. 8 2.6. In poor districts, wage income (from agriculture and non agriculture) is an essential source of income, representing on average 39% of HH income, versus 20% in rich districts. Reflecting the lack of agriculture and agriculture-related economic opportunities in poor districts, 0.50 adult males per household migrate in search for work, versus 0.37 in rich districts. The migrants are also away for longer periods: 0.25 adult males per household are away for more than 6 months of the year in poor districts; 0.17 in rich districts. In addition, households tend to rely more on cash transfers and remittances (21% of their income versus 16% in rich districts). Table 1: Rural districts income structure according to poverty incidence Rural poverty Rural poverty Rural poverty incidence incidence incidence All rural > 60% >30% < 60% < 30% Total income (Riyals) 418 000 507 000 865 000 632 000 Agricultural net income 28.0% 31.0% 46.7% 39.3% Agricultural wages 5.3% 3.7% 2.1% 3.1% Private sector wages 14.32% 9.99% 7.13% 9.05% Government wages 19.19% 22.28% 10.48% 15.25% Net income from enterprises 12.3% 13.5% 18.1% 15.9% Cash transfers 1.4% 0.8% 0.5% 0.7% Remittances 7.5% 7.2% 5.9% 6.5% Pensions and rents 12.0% 11.5% 9.1% 10.3% Food self-sufficiency ratio (%) 7.7 9.8 11.1 9.9 A/ Rural poverty incidence:% of poor out of rural population in the district Figure 3: Household income structure in rich versus poor rural districts Rich districts Poor districts 0% 20% 40% 60% 80% 100% Agricultural net income Agricultural wage income Non-agricultural private Non agricultural public Enterprises Remittances transfers, other income 9 2.7. District economic geography assets. To substantiate further the difference in income structure, we calculate indices of rural and urban market potential as well as distances to paved roads and major cities.14 The rural market potential of rich districts is 150% that of poor districts which corroborates the larger share of business income in rich districts (Map 4). In addition to having more agricultural opportunities, a rich district is three times as likely to be located on the border with Oman or Saudi Arabia, which results in additional business opportunities for these districts; and, twice as likely to be located on the coast, a location which allows for lucrative fishing business. Map 4: Rural market potential Source: Authors calculation. The higher the number, the better the market potential. (Formula in footnote) 14 Rural (urban) market potential is a gravity measure of neighboring areas rural (urban) economic size, discounted by Euclidean distance (km) between two districts. Specifically, distance to paved roads and travel time to the nearest town of more than 20,000 population are computed from GIS road and populated area shape file data using spatial analyst tools of ArcGIS software. rural MP i avg. rural j hh income j rural pop j exp 0.1 distanceij urban MP i avg. urban hh income j j urban pop j exp 0.1 distanceij . 10 2.8. Structure of district agriculture income. Rich districts have income from fisheries (a negligible source of income in medium-poor districts, and nonexistent in poor districts). Livestock plays a more important role in poor districts (22.6% of monetary income) than in the other districts (16.6% and 6.8% of monetary income respectively for medium poor and rich districts) consistent with the fact that poor districts tend to be located in more marginal areas with regard to crop production, where livestock raising maybe the most viable agricultural activity (Table 2). Table 2: Structure of rural districts agriculture monetary income according to poverty incidence Poor districts Medium-poor districts Rich districts Rural average Agri net monetary income (Riyals) 56,000 84,150 130,000 96,200 Crops 20.3% 31.1% 30.0% 29.1% Qat 58.9% 51.7% 47.0% 50.1% Livestock products 23.8% 18.8% 10.4% 14.9% Fish 0.0% 1.7% 16.1% 9.4% 2.9. Qat is a major source of monetary income in all districts. In absolute terms, the value of qat income is the highest in rich districts. However, its share of monetary income is more important in poor districts (60% versus 47% in rich districts, and 50% overall), reflecting the fact that households in rich districts enjoy a wider range of economic opportunities in addition to qat, for their monetary income (Table 2). There are slightly less households selling qat in poor districts (21%) than in rich districts (25%). 2.10. District poverty incidence and food self sufficiency. The food self-sufficiency ratio15 is extremely low overall: rural households produce less than 10% of the food they consume (Table 1). However, in rich districts, this percentage is higher, 11% compared to 8% in poor districts, a result which is consistent with the fact that rich districts are better endowed with natural resources, and therefore produce more agricultural goods. In rich districts, households also have access to more land than in poor districts (see below), which again may explain why they have a higher food self-sufficiency ratio. 2.11. Inequalities in district capital assets: land. The previous section analyzed the differences in natural resource endowment between poor and rich districts. We now look at their average household assets. With regard to land access, fewer households in poor districts have access to land: 58% (65% in rich districts) with smaller average cultivated land size per household16: 0.56 ha (1.38 ha in rich districts) (Figure 4). Another way to highlight land asset differential is to calculate population density per arable land: in poor districts one ha of land supports close to 25 people; 8 to 9 people in rich districts (Figure 5). In addition, in poor districts, 15 The food self sufficiency ratio is the ratio of the value of food consumed from own production divided by the value of total food consumption. However, the food consumption ratio is calculated on the basis of the HBS data. Households were interviewed four times over a four week period during the month of December. Whether households would consume more or less from their own production during other months of the year is a moot point. 16 Land size cultivated is land owned plus land rented in, minus land rented out, plus land share in, minus land shared out 11 there are fewer households with land which benefit from access to some form of irrigation: 21% vs. 32% in rich districts17 (26% overall). Figure 4: Access to land according to district poverty levels Figure 5: Number of people per ha of arable land according to district poverty levels. 17 Percentage of households having access to irrigation out of all households with land. 12 2.12. Map 5 presents the average cultivated land size per district. The inequalities of land access between districts are large: in many districts, farm sizes are less than a quarter of a hectare, while in others, farm sizes are between two thirds of a hectare and 17 ha. The very large landholdings are located in a few districts. Nine districts of the Govemorates of Al-Hodeidah, Amran and Lahej18have an average size of the landholdings between 2 and 17 ha. (Map 6) Map 5: Average cultivated land size per district 18 These districts are the following: Al-Hodeidah (Al Munirah: 4.6 ha; Bayt Al-Fajiah: 3.1 ha; Al-Mansuriyah: 3.0 ha; Al-Mighlaf 2.4 ha); Dhamar (Jaran 17.1 ha; Mayfaat Aness: 3.1 ha; Dawran Aness: 2.1 ha); Lahej (Al-Milah 4.0 ha); Taiz (As Silwa; 2.3 ha). 13 Map 6: Location of the districts with very large landholdings (Average cultivated land size between 2 and 17 ha) 2.13. Map 7 presents the distribution of access to irrigation (in percentage of households with access to irrigation). Here again access to irrigation is very unequal: in some districts, more than two thirds of the farms are irrigated, while in others it is less than 20%. 14 Map 7: Access to irrigation per district (number of households, %) 2.14. Other assets: animal ownership. Households in poor districts have also fewer cows and beehives, but more small ruminants (Table 3). Land and other agricultural asset differential may explain why as many as 39% of poor district households are wage earners for their main source of income19 (on and off-farm), while in rich districts, the percentage of such households is 20%. Table 3: Household livestock assets according to district poverty levels Medium Poor Rich poor All rural districts districts districts Number of cows 0.45 0.67 0.71 0.64 Number of small ruminants 6.50 5.84 5.60 5.89 Number of beehives 0.36 0.57 0.60 0.53 19 Main source of income is the source of income representing at least 60% of the total income. 15 2.15. Human resource assets. Looking at human capital assets, there is no significant difference in neither demographic nor education assets. Rural households in poor districts have on the average, 8.1 people, slightly more than 7.6 people in rich districts (Table 4). The number of years of schooling of male and female adults is also quite similar (on average, for all rural areas 5.9 and 1.3 years respectively). Table 4: Human resource assets Poor districts Medium-poor Rich districts Average HH size 8.1 7.5 7.6 7.7 Years of schooling of HH 3.4 3.5 3.4 3.4 head Years of schooling of female 1.4 1.4 1.2 1.3 adults Years of schooling of male 5.9 5.9 6.0 5.9 adults 2.16. Access to services and basic infrastructure. Poverty is not just the result of lack of access to economic opportunities, which results in monetary deprivation the most common measure of poverty. It is also the result of lack of access to services and basic infrastructure, another dimension of deprivation. We seek to analyze whether economic poverty is compounded (or not) with lack of access to basic infrastructure and services. To this end, we use a District Access index constructed on the basis of information from the 2004 Population Census. The index aggregates individual literacy (for people 10 years and older); enrollment rate (for children age 6 to 15); household access to safe drinking water, sanitation, electricity; and use of gas as cooking fuel (rather than wood and charcoal). The District Access index varies from 0 to 100, with 100 indicating that all households in the district have most of their needs satisfied. 20 2.17. We correlate the index with the two measures of economic deprivation we have been working with: the district poverty incidence and the district average household income. The diagrams below purport to illustrate the relationship. Access to services and basic infrastructure vary tremendously from one district to the next: some districts reach almost the 100 index mark, while others are close to zero. The fitted line does show some relationship, albeit not a strong one, between poverty incidence in a particular district and access to services and basic infrastructure in that district: in other words, there are almost as many rich districts with low access as there are poor districts (Figure 6). 20 The index is akin to the UN Unmet Basic Needs Index (as a matter of fact a reverse index). It has been calculated by the Social Fund for Development, using the results from the 2004 Population Census. 16 Figure 6: Access to services and district poverty incidence 80 access to services 60 40 20 0 0 .2 .4 .6 .8 1 poverty incidence poor districts medium-poor districts rich districts Fitted values Source: Access Index calculated by the Social Fund, using census data, and HBS data 2.18. A similar diagram, focusing on average household income of poor/medium-poor/rich districts, shows the same overall spread of access, but this time with a much clearer trend: districts with higher average household income have better access, as indicated by the fitted line (Figure 7). This result is comparable with that of other countries: it has been shown that households with higher income levels tend to purchase the services that are not met through public services, for example for sanitation, cooking gas instead of wood or charcoal, etc.. 17 Figure 7: Access to services and district average household income 80 access to services 60 40 20 0 0 200000 400000 600000 800000 1000000 average income poor districts medium-poor districts rich districts Fitted values Source: Same as Figure 6 2.19. Poverty and lack of mobility. Map 8 shows the paved roads in Yemen. The distance to a paved road is high in all districts (12.5 kms on average) without much difference between poor and rich districts21 . However, distances in Yemen do not mean much, given the extremely rough and steep terrain in certain areas, compared to relatively flat and easy terrain in others. Hence we also look at travel times to cities (Map 9). Travel time to a city of 20 000 inhabitants is on average very high for all districts: almost seven hours. Travel time for people in poorer districts is lower than the average (six hours).22 Hence poverty is not correlated with lesser access in terms of traveling time. 21 The average distance to a paved in poor districts is 10.2kms, 13kms in rich districts. 22 416 minutes on average. Travel time for poor districts is 352 minutes, 402 minutes for rich districts and 470 minutes for medium-poor districts. Travel time to larger cities is slightly higher but not much : seven and a half hours travel time to a city of 50 000 inhabitants on average. 18 Map 8: Paved roads Map 9: Traveling time to a city of 20 000 inhabitants 19 2.20. The story of Saoud district in Amran Governorate, where 76 percent of people are below the poverty line illustrates and summarizes well the situation of poor districts (Box 1). Box 1: The story of Saoud District in Amran Governorate (Poverty incidence: 76%) Saoud district is a very poor rural district of 23 000 people. It is located in the drier Northern part of the Highlands Eastern Mountains. The average rainfall of 350 mm, is characterized by erratic repartition, both inter and intra annual, and according to Local Councils representatives, with a pronounced tendency towards lesser amounts. With a combination of population increase and less rainfall, the land can no longer sustain its people. "Sorghum cultivation is no longer feeding people". Another factor that contributes to the decline of sorghum production is the fact that terraces are no longer maintained as they used to, while flash floods are becoming more destructive. As terraces are not maintained, the erosion is severe downstream when flash floods occur. Cooperation among people used to ensure terrace maintenance. "Before, when there was damage to terraces, people used to get together immediately to repair, because they knew their life depended on it. Nowadays, people go off on their own in search for work". The district extends from highlands to lowlands. Rainfed agriculture, based on sorghum and legumes is more productive in the highlands, because of higher rainfall. Irrigated agriculture would be possible in the lowlands, where the water table is close to the ground. However, most families do not have enough resources to invest in a well and a motor pump. In any case, high value irrigated crops such as vegetables, would not be possible because of the very poor access conditions. Even the district center, in the highlands, can only be accessed with a dirt road, which was being paved at the time of the visit. It takes several hours with a 4x4 vehicle to reach most of the district villages. The best agricultural income source is beekeeping and livestock raising. But even livestock raising is becoming problematic as rangelands are being depleted. In the seventies, men left to Saoudi Arabia and worked as construction workers. Migrant workers returned during the first Gulf War, and since then, either stayed in their villages or looked for work in Yemeni cities. Because of very low level of skills, most people are not even literate, finding jobs other than low paid manual labor, is almost impossible. Source: Interview with Local Council representatives, Mohamed Naser Mohamed Humaid and Mohamed Kaid Ali Serag, both school teachers. 20 SUMMARY 2.21. This section examined spatial differences related to poverty level: the differences between "rich" and "poor" districts. Districts are categorized as "poor", "medium-poor" and "rich" according to the percentage of the rural population below the poverty line, or incidence. 2.22. In contrast with poor districts, rich districts are better endowed with geographical assets, in particular natural resources for crop production. A more vibrant agriculture stimulates the rural non-farm economy and the opportunities to develop rural enterprises. Households in rich districts have more economic assets, in particular land, they cultivate larger land areas on average, and more of them have access to irrigation. The number of rural people per ha of arable land in poor districts is 3 times that of rich districts. 2.23. In poor districts, with worse natural resource conditions, smaller cultivated land sizes and less access to irrigation, opportunities to derive cash income from agriculture (except from qat), or even to produce for home consumption are limited. Raising small ruminants is economically more important in poor than in rich districts, as a result of the fact that poor districts are more likely to be located in pasture lands and/or marginal areas for crop production without irrigation. 2.24. Rich districts are also better endowed with favorable economic location assets, in particular being on the border with Oman or Saudi Arabia, or the coast, which again provides opportunities for developing businesses or adds another economic opportunity with the lucrative fishing sector. Rural market potential is 150% higher in such districts compared to poor districts, which results in thriving entrepreneurial activities in rich districts. In poor districts, people have to rely on their labor to generate income, and because of lack of local economic opportunities, adult men have to leave their villages in search for work, more so and for longer periods than in rich districts. 2.25. There are no differences in education level. Families are slightly larger in poorer districts. There are no differences in terms of access to services and basic infrastructure and district poverty level. However, better access to services and basic infrastructure is positively correlated with district average household incomes. 2.26. In the next section, we will examine various households assets (land in particular), how they interact with household main source of income and resulting welfare (poverty and household income levels) and resulting inter household inequalities. 21 3. INTER-HOUSEHOLD INEQUALITIES RURAL LIVELIHOOD STRATEGIES Sources of income and employment 3.1. Agriculture is practiced by a large majority of rural households: 89.6% of the households in rural areas are involved in some crop, fishery or livestock production.23 Including both crops and livestock, 65.6% of the rural households generate some monetary income from the sale of their agricultural products. The percentage of households with at least one member earning agricultural wages is rather low: 11.5%. Finally, one notes the importance of non-farm employment and remittances: 41.3% of rural households have at least one of their members employed in the non-farm sector (18.9% as government employees) and 47% receive remittances. Approximately 20 percent of rural households are involved in business activities (Figure 8). Figure 8: Participation in economic activities (% of all rural HH) Diverse livelihood strategies and welfare outcomes 3.2. Although most households receive incomes from several sources (own farming, employment in agriculture, fishing, business, non-farm employment and remittances), one of the income source may dominate, with the other sources making minimal contributions to household income. In this case, household activities are diversified but not the incomes. To analyze these different rural livelihoods, for each household, we identify the income source which provides at 23 Surprisingly, agriculture plays also an important role for urban HH: a large number (61%) of urban households cultivate land, mostly for home consumption. indicating the importance of land plots for these households. Most urban people who have a plot produce for home consumption. Only 11% of those urban people who produce agricultural products sell some of them. Agriculture also provides employment for 3.2% of the urban people. 22 least 60% of the total income, and assign the household to this group. We consider these as "specialized" households. If none of the income source contributes a minimum of 60% to total income, the household is considered as "diversified" or multi-source income and assigned to that category (Table 5). 3.3. With this definition, about 64% of the households are specialized, 36% are diversified.24 Specialized farmers make up 20% of the rural population; entrepreneurs, 9%; employees in the non agricultural private sector, 13%; civil servants, 10%; agricultural workers, 4%; remittance- dependant households, 7.5%. A tiny group of households, 0.6%, are fishing entrepreneurs.25 (Figure 9). Figure 9: Household categorization according to main source of income 3.4. Table 5 and Figure 10 compare household income by main source of income, and the poverty incidence for that category.26 The average household income is by far the highest for households in the lucrative fishing business: 3.5 times the average HH rural income and 2.2 times that of the second richest HHs, the farmers. This can be explained by the fact that most of the households in the fishing category can be considered as entrepreneurs: they own or rent their boat. Entrepreneurs come as third richest. Households with the lowest incomes are the agricultural workers and the remittance-dependent households. The highest incidence of poverty can be found amongst agricultural workers (56% of the households in this category are poor), and diversified 24 This runs contrary to the statement often heard that rural households have diversified income sources Rural income sources are diverse across households, but not within households. 25 These are households own or rent a fishing boat. They are not fishing workers. 26 The reason for comparing household incomes and poverty is because these are two different measures: poverty is an individual measure, calculated by dividing household expenditures by the number of people in the household. Household income is a measure of the income of the household, and does not take into account the number of people in the household. 23 households (43%) The mass of rural poverty is with the diversified household category: 38% of the rural poor belong to that category. Table 5: Income and poverty levels according to HH main source of income Main source of income Income Poverty Poverty mass28 (Yemeni Rials) level27 Fishing enterprise 2 194 000 16.8% 0.4% Farming 989 000 36.6% 21.1% Enterprise 962 000 30.3% 10.4% Government employees 429 000 31.9% 8.8% Private sector workers 708 000 52.0% 11.9% Agricultural workers 351 000 55.9% 3.3% Remittances 290 000 34.1% 6.2% Multi-source 450 000 42.7% 37.8% All rural 632 000 40.1% 100% Figure 10: Level of income and poverty incidence by main source of income Income by HH main source of income Poverty by HH main source of income Farmers 989 Farmers 37 20% Enterprises Enterprises 962 30 9% Non-farm employees 595 Non-farm employees 44 23% Agricultural workers 351 Agricultural workers 56 4% Remittances 290 Remittances 34 7.5% Fishing enterprises 2194 Fishing enterprises 17 0.6% Non-specialized HH 450 Non-specialized HH 43 36% 0 500 1,000 1,500 2,000 0 20 40 60 1000 Rials % 27 Poverty incidence= # of poor out of population in the category. 28 Number of poor in the category out of all rural poor. 24 3.5. We check the above findings by looking at the livelihood categories according to income quintiles (Table 6). Farmers and entrepreneurs are overrepresented in the richest category (compared to the rural average). But farmers are also overrepresented amongst the poorest (21.4% in this category). Table 7 shows the diversity of welfare situation of farming households. The household income of farmers in the richest quintile is 15 times that of farmers in the poorest quintile. Table 6: Livelihood category according to income quintiles Main source of HH Poorest 2nd 3rd 4th Richest All income/income income income quintiles quintile quintile Farming 21.4% 20.0% 17.7% 18.0% 23.7% 20.1% Non-farm enterprise 3.4% 5.1% 8.1% 11.3% 18.7% 9.3% Private sector worker 7.37% 13.92% 12.15% 14.96% 16.35% 12.9% Government employee 1.23% 10.28% 21.15% 12.14% 5.05% 10.% Agricultural wage 3.6% 7.5% 5.3% 1.7% 1.7% 4.0% Remittances 13.1% 9.6% 6.5% 5.9% 2.3% 7.5% Fishing 0.0% 0.0% 0.3% 0.3% 1.9% 0.5% Non-specialized 49.9% 33.5% 28.8% 35.7% 30.4% 35.7% All rural 100% 100% 100% 100% 100% 100% Table 7: Total household income of farming households by income quintiles (YR) Poorest quintile 2nd quintile 3rd quintile 4th quintile 5th quintile 103,000 216,000 341,000 550,000 3,267,000 3.6. Remittance-dependent households and diversified households are overrepresented as well in the poorest category (respectively 13% and 50% of the households in this category). The latter points to the fact that diversification is the coping strategy of the poorest households. Non- agricultural workers tend to be in the middle income quintiles, not amongst the poorest. 3.7. Table 8 presents the structure of average rural income: almost 40% comes from agriculture, 16% from enterprises, 15% non- farm private sector wages, 9% from public service employment, 6.5% from remittances and 3% from agricultural wages. 25 Table 8: Structure of income according to income quintiles All rural Poorest Richest 2nd 3rd 4th quintile quintile Total income (ryals) 89 800 217 100 342 000 538 000 1 978 300 632 500 - agri net income29 26.45% 26.49% 25.92% 28.70% 46.43% 39.3% - agri wage income 5.48% 8.62% 6.74% 2.70% 1.79% 3.1% - public service wage 2.27% 9.83% 19.93% 16.15% 5.45% 9.05% - private sector wage 11.06% 15.00% 13.84% 16.00% 15.50% 15.3% - enterprises 5.19% 6.23% 8.65% 13.52% 19.30% 15.9% - cash transfers 5.58% 2.41% 1.45% 0.75% 0.16% 0.7% - remittances 19.83% 13.91% 9.53% 9.78% 3.73% 6.5% - other incomes 24.02% 17.52% 13.89% 12.36% 7.64% 10.3% Households with self consumption 22.9 29.3 27.4 34.5 39.1 30.7 Food self-sufficiency (%) 7.4 8.4 9.1 11.6 12.9 9.9 3.8. Table 8 and Figure 11 also present the structure of income according to income quintiles. The data confirms the findings above. The HHs in the richest quintile derive their income overwhelmingly (i.e. more than the rural average) from agriculture (46.4%) and enterprises (19.3%). For the poorest and second poorest categories, agriculture also represents the largest share of household income (26.5% for both). However, for the two lowest income categories, remittances, cash transfers and other incomes 30 all together, play a significantly more important role, 49.4% and 33.8% respectively, than for the average rural household (17.5%). 3.9. The middle-income class (3rd and 4th quintile) derive more of their income from public service employment: 20% and 16% respectively, than the rural average. Conversely at the two ends of the income spectrum, the poorest and the richest households derive less of their income from employment in public services. 29 Agriculture income includes income from fishing. 30 Other income includes pensions, rent, sale of household items, etc. 26 Figure 11: Income structure of the poorest, middle-class and richest households Food self-sufficiency 3.10. Overall 31% of the households produce some of the food they consume, covering almost 10% of their consumption from their own production. This percentage increases with the level of income: in the poorest category, 23% of the HH produce part of the food they consume (covering 7.4% of their consumption), while it is 39% of the households in the richest category, covering 12.9% of their food (Table 8). The extremely low level of food-sufficiency of Yemeni rural areas is a result of unfavorable agro ecological conditions (only 2% of arable land and unfavorable dry conditions), high agricultural population density, unequal land access and overall predominance of qat at the expense of cereal cultivation. Relations between assets and livelihood strategy 3.11. As described above, income levels and poverty incidence vary according to the main source of income of the household. In Table 9, we examine how main source of income and household capital assets are associated. We look at access to land and water, and livestock ownership. 27 Table 9: Household capital assets and main source of income Household capital assets: Farmer Farm Nonfarm Entrepreneur Remittances Fishing Diversified All worker employee entrepreneur HH rural Size of cultivated land (ha)31 1.52 0.24 0.26 0.49 0.20 2.76 0.46 0.61 Households with land 62.7 (%) 95.0% 29.1.% 45.5% 46.8% 46.4.% 53.6% 68.9.% % Size of cultivated land (ha, for HH with land) 1.60 0.86 0.58 1.06 0.44 5.35 0.67 0.98 Access to irrigation 26.4 (in% of HH with land) 48.0% 13.5% 11.1% 14.1% 9.5% 0.0% 21.4% % Number of cows 1.01 0.26 0.36 0.57 0.38 1.98 0.70 1.01 Number of small ruminants 10.38 2.06 3.24 4.15 2.63 52.51 5.97 10.38 Number of beehives 1.05 0.03 0.17 0.91 0.05 2.95 0.50 1.05 3.12. Fishing entrepreneurs not only have by far the highest level of income, but also have many more assets, in terms of land (5.35 has for those who have land). They use the land for livestock raising32: they have the largest small ruminant herds (52.5 on average) and they own many more cows and beehives than the farming category, the next category for cow and beehive ownership. 3.13. Aside from the fishing entrepreneurs, somewhat outliers in the rural economy (0.6% of the rural households), the category with the most assets are farmers, with not only land (1.6 ha on average, while the rural average is 0.98 ha) but access to irrigation (48%). In the Yemeni context, characterized by high variability of rainfall and recurrent droughts, secured access to water is as important as access to land. It allows farmers to invest in inputs for their production, which would be too risky without it, and thereby increase their agricultural monetary income. 3.14. Land access makes a difference for entrepreneur households. The business income of the enterprise households with land (1.06 ha on average) is almost twice (1.9 times) the business income of enterprise household without land (Table 10) for those with land, it is likely that they were able to use their farm-based income to start up a business. It is usually the case that rural businesses are financed by the owners own savings accumulated from agricultural activities, rather than borrowed from a formal credit source. One could also argue that, conversely, once they start making money out of their business, rural entrepreneurs invest some of their benefits in livestock and perhaps land (see Section on agriculture). The entrepreneur category also invests in the lucrative honey business. 31 For all households in the category including those who do not have access to land. 32 These households do not produce any crop. 28 Table 10: Enterprise income of entrepreneur HHs with land and without land (YR) With land (1.06 ha on Without land Average enterprise average) income Percentage 53% 47% Income (ryals) 1,068,000 572,000 804,000 3.15. The case of the non-specialized households is interesting. Most of them have some land (69%), with land size of 0.67 ha for those who have land, but perhaps not enough to support a three generation family (Figure 12). In addition, an outside job is the only way for younger men to acquire some independence from their father. Hence it is understandable, that the younger adult males of the diversified households with land complement the household income either with non- farm/migration employment, or start a small business on their own. As to the diversified households without land, the family business (or off-farm wages of the father) may again not be enough to employ/support all the younger male adults (otherwise the household would be in the entrepreneur category), and at least one of the younger adult male is seeking other employment opportunities. 3.16. Most agricultural workers (70%) do not have access to land (about 70%). The average land size is 0.86 ha for those that do have access (Figure 12). 3.17. In all categories, households own some livestock, especially small ruminants (Figure 14). Overall 55% of the households own some livestock. Honey beekeeping seems to be a specialized activity with only 2.4% of the rural households engaged in it (see section on agriculture). Figure 12: Land ownership according to livelihood category 29 Figure 13: Irrigation according to livelihood category Access to irrigation, % of HH Farmers 48 Agricultural workers 13.5 Non-farm employeee 11.1 Entrepreneurs 14.1 Remittances 9.5 Diversified HH 21.4 0 10 20 30 40 50 % Figure 14: Livestock assets and livelihood strategy Number of animals 1.01 Farmers 10.4 1.05 .26 Agricultural workers 2.06 .03 .36 Non-farm employeee 3.24 .17 .57 Entrepreneurs 4.15 .91 .38 Remittances 2.63 .05 .7 Diversified HH 5.97 .5 0 2 4 6 8 10 Count Number of cows Number of small ruminants Number of beehives 30 Relationship between household assets and level of income 3.18. As indicated in Table 11 below, farm assets and level of income are strongly related. The percentage of households with cultivated land and the size of the cultivated land increase steadily from the poorest to the richest, as well as the percentage of households with access to irrigation and the number of animals they own. The number of beehives owned is especially high for the fourth and fifth richest quintile. Table 11: Household assets and level of income Poorest 2nd Richest quintile quintile 3rd quintile 4th quintile quintile All Rural HH with cultivated land (%) 52.2% 55.6% 59.9% 70.1% 75.9% 62.7% Size of cultivated land, ha, for HH with land 0,506 0,521 0,578 1,108 1,832 0,977 Irrigation (in% of HH with land) 19.3% 20.5% 25.2% 27.6% 35.7% 26.4% # of cows 0.35 0.50 0.58 0.75 1.01 0.64 # of sheep and goats 3.40 4.77 5.64 7.25 8.42 5.89 # of beehives 0.22 0.16 0.13 0.83 1.33 0.53 ACCESS TO LAND, INEQUALITIES AND LIVELIHOODS Access to land, poverty and HH income 3.19. Our hypothesis is that land assets are a key driver of rural household income differences. In this section we analyze how access to land33 interacts with other asset accumulation, and is associated with income levels. We first categorize rural households by cultivated land size quintiles. Dropping observations with missing land information, we group rural households by land size quintiles, leaving households without access to land.34 3.20. Overall, 37% of rural households do not have access to land. The average cultivated land per household with access to land is quite small (0.98 ha). It is also very unevenly distributed. With a gini coefficient calculated on the value of the land of 0.6835 (Figure 15), Yemen has one of the most unequal land distributions in the world.36 Twelve per cent of the households control 80% of the land. The average land size in the largest land holding quintile, 3.92 ha is 6.6 times that of the average land size in the second largest land quintile (0.59 ha) and 163 times that of the average land size in the smallest land quintile (0.024 ha, Table 12). The two percent most land- 33 Land owned plus land rented in minus land rented out 34 Out of 4863 rural household observations (without sampling weight) of the HBS, we drop 335 rural households with missing land information. Remaining 4528 observations are used in the analysis. 35 The gini coefficient of land access, calculated on the basis of land size is even higher, 0.74. However, a gini coefficient calculated on the basis of land value is more accurate as it takes into account the difference associated with land quality. 36 Yemen land distribution is akin to that of the most unequal land distribution in Latin America countries. Republic of Yemen (2007), p. 21. 31 rich households own 8.3 ha, compared to less than one ha for the overall rural households with access to land. Table 12: Access to land, income and poverty levels (HH with access to land) Land quintiles Rural Cultivated land size quintiles 1 5 2 3 4 Average (smallest) (largest) Size of cultivated land (in ha, for those HH with land) 0.024 0, 109 0.254 0.588 3.921 0.977 Total income (Riyals) 376 000 452 000 570 000 727 000 1635 000 632 000 Rural poverty rate (% of poor in the category) 50.6% 40.5% 39.5% 33.2% 31.4% 40.1% Food self sufficiency (%) 5.7% 11.0% 15.7% 18.3% 25.8% 9.9% Households with own consumption (%) 29.2% 44.8% 50.3% 56.4% 54.9% 30.7% Figure 15: Chart of gini coefficient of land value (= 0.68) 1 Cumulative land value share .8 .6 .4 .2 0 0 .2 .4 .6 .8 1 Cumulative HH share 45 degree line Cumulative land value share 3.21. As can be expected, households with more land have a higher food self-sufficiency ratio. In the largest land quintiles categories, more households produce their own food and a higher percentage of their consumption comes from their own production. However, even in the fourth and fifth largest land quintiles, the percentage of households with own consumption does not rise above 56%, covering not more than 26% of their food needs (Table 12).37 3.22. Household income is strongly correlated with access to land: household incomes steadily increase as we move from the smallest to the largest land size quintiles (Table 12 and Figure 16). 37 The household budget survey may not capture all the consumption from own production, especially as it relates to own animal consumption: if the month of December 2004, when the consumption survey took place had no religious festivity, such as the Eid, then it is likely that there was no meat consumed from own consumption recorded, since rural people tend to slaughter their own animals on the occasion of religious or family (marriage, etc..) events. 32 The average income of the highest land quintile households (1,635,000 Riyals) is more than four times that of the lowest land quintile households (376,000 Riyals). Rural poverty incidence shows a similar pattern: it is the highest for the households in the smaller land size category, (51%)38. Figure 16: Access to land and income 2 0 15 log(income) 10 5 0 5 10 15 log(cultivated land size) log(income) Fitted values 3.23. With the high correlation between income and access to land, it is not surprising that household income, with a gini coefficient of 0.63 is almost as unequal as access to land (Figure 17). However, the fact that income is slightly less unequal than land points to the fact that sources of income that are not related to land, such as non agricultural employment, tend to smooth out the inequalities stemming from land access. 38 However, the landless households, with an average income of 434 000 YR of HH income, are on average better off than the HH in the smallest land size. This is because the landless category is very diverse, including better-off rural households (government employees and entrepreneurs) as well as extremely poor unskilled laborers. 33 Figure 17: Gini coefficient of income Gini Coefficient of income: 0.63 1 .8 Cumulative income share .6 .4 .2 0 0 .2 .4 .6 .8 1 Cumulative HH share 45 degree line Cumulative income share 3.24. Comparing average and median income by income category suggests that a small percentage of households in the richest income category account for the very high average income in that category (Table 13). Table 13: Income quintiles: comparison of average and median Income quintile Poorest 2nd 3rd 4th Richest Total income, average, ryals 89,800 217,100 341,900 538,000 1,978,300 Total income, median, ryals 94,198 215,650 339,498 529,129 1,047,800 Access to land and other capital assets 3.25. Rural households with larger cultivated land area are able to accumulate other capital assets (Table 14). Compared to the rural average, the largest land quintile households have more capital assets, such as access to irrigation (182%) and more cows, small ruminants and beehives (204%, 197% and 357%) than the rural average. Households without land invest in animals, some of them even have cows, highlighting the importance of animal production, not only for land-poor households, but practically for all rural households, whatever their main source of income. 34 Access to land and main source of income 3.26. Land control strongly influences livelihood strategies (Table 14). Logically, most of the household in largest land quintile group are specialized in farming (46.5%). Farmers are also over represented in the second highest land quintile group (34.3%). Conversely, 41.8% of the landless households are wage earners. The landless group has also the highest share of households whose main source of income is remittances, 10.8%. Though enterprise households can be found in all land size categories, they are overrepresented in the landless category (13.3%), as well as in the largest size land (10.7% in this category). The latter suggests, as mentioned above that revenues from agriculture enable farmers to invest in business39 Table 14: Household capital assets and access to land 1rst Average Household capital assets No land 2nd 3rd 4th 5th quintile rural Size of cultivated land (ha, for those with access to 0 0.024 0.11 0.25 0.59 3,92 0.98 land)) Irrigation by springs, wells ,dams (in% of HH with 18.6 26.1 27.3 38.9 0.0% access to land) 21.4% % % % % 21.4% Number of cows 0.20 0.59 0.76 0.96 0.96 1.20 0.64 Number of small ruminants 4.73 4.63 3.92 5.44 7.38 11.62 5.89 Number of beehives 0.23 0.26 0.29 0.40 0.75 1.89 0.53 3.27. As households have access to more land, they rely less on off-farm employment and migration income. Hence, in the highest land quintile, wage earner households represent less than 11% of the households, compared to 41.8% in the lowest quintile. The figure drops to 1% for remittance-dependant household compared to 10.8% in the landless category and 10.2% in the micro-landed household category. 3.28. The diversified households are underrepresented in the landless and the highest land size categories (respectively 30.7% and 28.9%), and over represented in the lowest land quintile, pointing to the fact that households tend to seek to diversify their source of income as a result of insufficient access to land. Households in the largest land category have enough agricultural income, and households in the landless category have secured an off-farm income. 39 Or conversely, it may be the case that entrepreneurs invest part of their benefice into acquiring land. Given the social structure in Yemen, it is more likely that rich landowners are using agriculture revenues to invest in business activities. 35 Table 15: Size of cultivated land and main source of income40 (in% of hh in the group) Size of cultivated 0 0.024 0.11 0.25 0.59 3,92 All rural land (ha) Farming HH 2.8% 21.8% 22.1% 28.0% 34.3% 46.5% 20.1% Agricultural wage HH 7.7% 1.8% 2.0% 1.2% 2.1% 1.7% 4.0% Non-agricultural wage HH 34.1% 17.8% 21.7% 18.3% 14.2% 9.2% 22.9% Enterprises HH 13.3% 5.1% 6.6% 7.4% 4.9% 10.7% 9.3% Remittances HH 10.8% 10.2% 6.9% 6.0% 3.6% 1.0% 7.5% Fishing HH 0.7% 0.0% 0.0% 0.0% 0.0% 2.1% 0.5% Non specialized HH 30.7% 43.3% 40.6% 39.2% 40.9% 28.9% 35.7% All 100% 100% 100% 100% 100% 100% 100% AGRICULTURE AND RURAL LIVELIHOOD Crop and livestock production and sale 3.29. Agriculture is important for all rural households, whatever their main source of income and whatever their level of income: as seen above 90% of the rural households are involved with crop and livestock production, or fishing. Cereals occupy the main share of cultivated area (58%). The area planted in qat represents about 11% of total cultivable area, and it has been steadily expanding at the expense of cereals, at the rate of 7.3% a year since 200541. The expansion is fueled by the high returns to be expected from qat cultivation, between 10 to 20 times higher than with any other alternative crop, and encouraged by subsidized diesel fuel policies. Qat uses about 70% of the groundwater.42 The rest of the area is planted with vegetables and fruit crops, coffee, pulses and fodder. 3.30. Agricultural income includes own consumption and stocks (53.7 %) and monetary income, from crop (36.8%) and livestock (9.5%).43 All most half of the agricultural monetary income comes from qat, followed by fruit and vegetables. The sale of live animals comes as a third (Figure 18). Overall, monetary income from cereal sale is equivalent to that of animal feed, and is quite small. Income for sale of coffee beans is even smaller. 3.31. Though animal product sale represents much less monetary income than qat, it is an activity that concerns almost half of the rural population: 48%. Qat on the contrary is sold by fewer households: about 25%. Even fewer households are involved in fruit and vegetables (6.3%). Cereals, animal feed and coffee are sold by only, 7.4%, 8.0%, and 2.7% of the households respectively (Figure 19). 40 Reminder: main source of income is the source which is responsible for more than 60% of the total HH income. 41 Area planted in cereals has averaged around 760 000 ha since 2004, but except for 2007 when the area went up, it is being replaced by qat, which has added around 4 800 ha per year since 2005. The area planted in qat was 147 000 ha in 2008. FAO (2009). 42 FAO (2009). 43 Monetary income from fish is excluded because fishing-entrepreneur households are few and somewhat outliers. 36 Figure 18: Structure of crop and livestock monetary income (gross) Figure 19: Share of households involved in marketing agricultural products (in% of all rural HH) 37 Crop and livestock sales according to livelihood strategies 3.32. Crop and livestock monetary income contributes to household income to varying degrees depending upon the livelihood strategy of the household (Table 16). Qat represents the number one source of agricultural cash income for farmers, 53.3%. 52.7% of farming households sell qat. Fruit and vegetables come second, representing 20.3 %. Sale of animal products comes as third in importance, representing 11.5% of agricultural monetary income. 3.33. Households of the other livelihood categories also sell qat, but there are less many households involved in that trade. For example, 25.9% of the diversified-income households sell qat, which represents almost half of their agricultural monetary income (47.7%), more than livestock, which accounts for 26.2%. For entrepreneurs, the sale of qat represents 38.9% of their agricultural income 3.34. Livestock is the most important source of monetary agricultural income in terms of the number of households involved (48% including all livestock products). It is a more important source of monetary agricultural income than qat for wage worker households, and as important as qat for remittance-dependent households. Table 16: Crop and livestock income structure according to main source of income Non Livelihood Farm farm Non All category Farmer worker worker Entrepreneur Remittances specialized rural Ag cash income (ryals) 410,000 15,000 26,000 62,000 22,000 94,000 131,000 -qat (in% of cash income) 53,3% 20,4% 27,3% 38,9% 33,9% 47,7% 49,2% -Fruit & vegetables (%) 20,3% 1,2% 11,0% 14,4% 18,6% 17,2% 18,6% -cereal (%) 4,4% 6,3% 6,% 2,3% 8,6% 4,6% 4,3% -animal feed (%) 4,4% 9,4% 6,7% 2,3% 1,8% 2,2% 3,9% -honey (%) 2,2% 0,8% 1,8% 2,6% 0,0% 1,4% 1,9% -live animals (%) 11,5% 55,0% 38,3% 34,9% 33,7% 23,4% 17,7% -other crop or liv. sales (%) 3,9% 6,9% 8,4% 4,7% 3,3% 3,3% 4,3% Own consumption & stocks (rials) 592,000 14,000 26,000 38,000 23,000 56,000 152,000 Total ag. income (ryals) 1,002,000 29,000 52,000 100,000 45,600 150,200 283,000 HH with qat income 52.7% 5.6% 11.0% 12.1% 11.3% 25.9% 24.6% HH with livestock income 66.2% 29.8% 34.2% 42.5% 26.3% 54.1% 48.1% Cultivated land area (ha, for all HH in the category) 1.52 0.24 0.26 0.49 0.20 0.46 0.61 38 Crop and livestock production and sale according to household income 3.35. Comparison of the structure of agricultural monetary income, by income quintiles. Table 17 and Figure 20 present the structure of agricultural income by income quintile. The richest category obtains almost 20 times more income from agriculture than the poorest. Agricultural monetary income for the richest comes by and large from the sale of qat. The share of qat income as part of monetary agricultural income increases steadily from the poorest to the richest quintile, as well as the number of households who participate in that trade (from 17.9% to 31.3% for the richest). For the poorest, revenues from livestock sales are almost as important as the sale of qat. Fruit and vegetables sales are important for both the poorest and the richest categories, less so for the middle-class categories. Sale of honey is more important for the two richest quintiles. Table 17: Structure of agricultural income by income quintile Income quintile Poorest 2nd 3rd 4th 5th All Ag cash income (rials) 29,000 46,000 78,000 134,000 367,000 131,000 -qat 36.20% 42.47% 51.51% 51.27% 50.00% 49.24% -Fruit & vegetables 18.18% 10.83% 11.18% 10.73% 24.07% 18.57% -cereal 2.69% 4.74% 4.08% 6.11% 3.97% 4.32% -animal feed 2.69% 3.21% 3.32% 4.77% 3.97% 3.89% -honey 1.01% 0.85% 0.91% 2.09% 2.38% 1.94% -live animal 33.84% 31.98% 24.92% 20.27% 12.17% 17.71% other sales 5.39% 5.92% 4.08% 4.77% 3.44% 4.32% Own consumption + stocks (rials) 20,000 32,000 40,000 66,000 604,000 152,000 Total ag. income (Y RL) 49,000 78,000 118,000 200,000 971,000 283,000 Household with qat income 17.9% 22.0% 22.6% 29.1% 31.3% 24.6% HH with livestock sale income 35.7% 43.1% 48.4% 54.3% 54.1% 47.1% Size of cultivated land (for all HH in the category) 0.26 0.29 0.35 0.78 1.39 0.61 39 Figure 20: Comparison of the structure of agricultural monetary income, by income quintiles Rural inequalities, qat and livestock 3.36. We look at various sources of agriculture monetary income as potential areas that could increase/decrease inequalities. We calculate an index of the value of gross monetary income for qat, fruit and vegetables, and live animal, whereby the index equals to 100 for the richest category (Figure 21). The lucrative qat, as well as horticulture production, are overwhelmingly associated with the richest HH quintile: the spread of the index between the poorest and the richest household category is respectively 5.7 and 6.0 out of 100 for these commodities (100 being the monetary value of these products for the households in the richest quintile). Because of the link with high value qat and horticulture production, land access has an income inequality increasing effect. The gini coefficient of crop net monetary income, without qat, is 0.76, higher than that of land value. The gini coefficient of qat monetary income is much lower, and lower than land: 0.63. Interestingly enough, qat has a less income inequality increasing effect than the growing of other crops. 3.37. By contrast, the sale of live animal is more evenly distributed in all income categories, with an index of 23 for the poorest (out of 100, for the richest). Livestock raising can be done on communal land, complemented by household residues, and/or animal feed which can be purchased. Qat and horticulture production on the other hand requires not only land, but also irrigation. Even people with no land nor access to irrigation can raise some livestock, small ruminants at least. Because it enables households to enter the lucrative qat and horticulture business, access to land deepen inequalities, while animal production, which does not require access to land, has the potential to decrease initial inequalities due to land access, as evidenced by the gini coefficient of net livestock monetary income of 0.52. 40 Figure 21: Sale value of qat, fruit and vegetables and live animal according to household income quintile (in% of that of the highest income quintile). 100 80 Qat 60 Fruit and vegetables 40 Live animal 20 0 Poorest 2nd 3rd 4th Richest Agricultural monetary investments by livelihood categories 3.38. Table 18 analyzes agricultural investments according to livelihood categories. Overall, they represent about 15.5% of total agricultural production income. The group that spends the most in absolute terms are the farming households (more than twice the average), but as a percentage of the value of their agricultural production, they spend the least (10.5%) The group that spends the most as a percentage of the value of their total agricultural production is the remittances-dependant household group (62.5%), almost equally shared between crop and livestock, presumably buying animals for the latter. The agricultural worker households invest mostly in livestock. In general, the farming and diversified households invest more in their crops (71.2 and 54.3%) while the other categories invest more in livestock. 41 Table 18: Agricultural monetary investments by livelihood categories Livelihood Farmer Agri Non-agri Enterprise Remittances Non Overall categories wage wage specialized Production monetary costs (Ryals) 104,762 10,281 19,364 34,008 28,503 34,954 43,810 As share of total production, % 10.5% 35.1% 37.2% 33.9% 62.5% 23.3% 15.5% Cost category shares, % - Crop 71.2% 22.1% 41.4% 44.0% 34.1% 54.35 59.0% -Livestock specific 11.5% 66.6% 45.5% 42.2% 36.3% 30.0% 24.2% - Wages 11.9% 8.15% 9.6% 10.7% 8.4% 12.7% 11.6% -Other expenses 5.5% 3.2% 3.5% 3.2% 21.1% 3.1% 5.2% HH using credit for ag production , %44 6.1% 0.0% 2.8% 1.6% 0.9% 2.1% 3.3% HH receiving ag extension services ,%45 3.0% 0.9% 2.8% 1.2% 3.2% 5.0% 3.6% Cultivated land (ha,for all HH, average) 1.52 0.24 0.26 0.49 0.20 0.46 0.61 3.39. Very few households use credit to invest in agriculture, even among the farming households: only 6% of them. Even fewer households receive advice from the extension services: less than 3% of the farming households and 5% of the non-specialized households. HOUSEHOLD DEMOGRAPHY, MIGRATION AND EDUCATION ASSETS Household demography, migration and education assets according to main source of income 3.40. Household demography. With 7.7 members per family, the size of the rural household is only slightly more than that of an urban household (7.1 members). The largest households, with also the highest income, are farmers and entrepreneurs (8.0 and 8.6 respectively). The smallest (and also the poorest) are the agricultural worker and remittance-dependent households (Figure 22). 44 Calculated in percentage of household with land 45 Same as above 42 Figure 22: Household size and number of male adults according to livelihood category 3.41. Education. Overall, education levels of male adults are low: 3.4 years for the household head, 5.9 for all male adults. There are no differences in education levels across livelihood categories, except for the government employee households. Not surprisingly, male education is the highest in these households, whether looking at heads of households only, or at all male adults: respectively 9.5 and 10.7 years of schooling (Table 19 and Figure 23).46 Figure 23: Years of schooling according to livelihood categories 46 Compared to 6.8 years of schooling for household heads in urban areas. 43 Table 19: Household demography, migration and education according to livehood categories Livehood Farm Private Government Entrepreneur Remittance Diversified All Farmer Categories worker sector Employee s rural worker Age of 46.5 37.9 41.4 39.0 43.0 41.2 48.2 44.7 household head Years of schooling of 2.3 2.6 3.6 9.5 3.3 2.1 2.8 3.4 household head Years of schooling of 4.7 3.5 5.5 10.7 5.3 6.0 5.9 5.9 male adults Years of schooling of 0.7 0.3 1.3 2.3 1.4 1.9 1.3 1.3 female adults Number of adults migrating 0.30 0.67 0.81 0.41 0.33 0.31 0.46 0.45 for work per household " more than 6 0.06 0.39 0.47 0.22 0.09 0.28 0.19 0.21 months 3.42. The heads of remittance-dependent and diversified households have levels of education similar to the rural average, but the male adults in these households have higher levels of schooling (6.0 and 5.8 years) than the other categories, an indication that in these households, the younger generation is getting out of farming and employed as civil servants or professionals. Agricultural workers have the lowest level of education, for all male adults (Table 19). 3.43. The overall level of education - at least for men - is clearly improving, with adult men gaining 2.5 years of schooling over their fathers (from 3.4 to 5.9 years of schooling on average). However, this gain is unequally distributed, agricultural wage households gain very little (0.9 years of schooling) reflecting the fact that in these households, boys drop out of school to work in the fields at a very early age. The risk is that poverty in this category will be self-perpetuating to the next generation. 3.44. The "diversified" households may be in a demographic and economic transition: the heads of these households are the oldest (48.2 years compared to 44.7 years for the average age of the household heads); the younger adult men are involved in non-farm or business activities. In the short to medium term, at the death of the father, these households are likely to split into nuclear families. 3.45. Migration. The number of adults migrating for work is quite high, overall 0. 45 male per family, and 24% all adult males overall (Table 19 and Figure 24). The working households have the highest number of adult males leaving their homes in search for work (0.76 and 0.81), many of them being away for more than six months a year. This reflects that employment opportunities, for those who have only their labor to sell, are quite limited. Migration rate is the lowest for farming households, followed by entrepreneur and remittance HHs. The hypothesis is that the farming and entrepreneur households have enough assets and income generating opportunities to employ their young adult males effectively, without them to have to leave their homes and villages 44 in search for work. For the remittance-dependant HH, the situation is quite different: the main adult males have permanently left the household, leaving behind mostly older people, women and children. Figure 24: Migration and livelihood categories Household demography, migration and education assets according to income level 3.46. Household demography and income. Household size increases steadily with income (Table 20 and Figure 25). The richest households are the largest: higher levels of income can obviously sustain more people. However, their dependency ratio 47 is also the lowest. More income earners in the household contribute to higher household income. Table 20: Household demography and education assets according to income levels Richest Poorest quintile 2nd quintile 3rd quintile 4th quintile quintile Overall hh head age 44.3 42.1 43.0 46.0 48.1 44.7 hh head yrs of schooling 2.3 3.3 4.3 4.1 3.2 3.4 yrs of schooling of female adults 0.9 1.2 1.4 1.5 1.6 1.3 yrs of schooling of male adults 4.2 5.2 6.4 6.7 6.7 5.9 hh size 5.6 6.7 7.5 8.3 10.2 7.7 number of male adults 1.1 1.5 1.8 2.1 2.8 1.9 Dependency ratio 80.4% 77.6% 76.0% 74.7% 72.5% 75.3% 47 We define the dependency ratio as the ratio of male adults to household size, due to the extremely low level of paid women in rural areas: 0.6% as we shall see later.. 45 Figure 25: Household size and number of male adults by income levels 3.47. Education and income levels. The lowest levels of education are associated with the lowest levels of income. However, the most educated are not the richest quintiles but the middle- class categories, in particular the household heads. These income quintiles, as already discussed, have a higher representation of government employees which explains the higher levels of education (Table 20). 3.48. Migration and income. The higher the income category of the household, the higher the number of adult males migrating: 0.64 adult male per household in the richest quintile, versus 0.28 in the poorest quintiles (Figure 26). As we saw earlier, richer households have more adult males, hence, they are more able to afford sending one of them on migration and thus diversify their income sources. Migration is thus positively associated with household size and income levels: These larger households with more adult males are less dependent upon only one source of income and therefore less vulnerable to shocks, whether external or internal to the household (illness of the working adult or inability to find work in times of economic crisis for example). 46 Figure 26: Migration and household income 3.49. Migration and education levels. We look at the education levels of migrant workers. Surprisingly, the results differ according to the position in the household: heads of families migrating have a significantly higher level of education (4.3 years) than the average rural heads of households (3.4). However, the average number of years of schooling of adult men migrating is 5.3, versus 5.9 for the rural average: migrants who are not heads of families are less educated than the rural average. In addition, heads of household migrating are younger than the rural average (38 years old versus 41), but there is no age difference between other men migrating and the rural average (approximately 32 years). WOMEN IN THE RURAL ECONOMY AND SOCIETY 3.50. Rural women in the labor force. Only 0.6% of the rural women work for a salary. Most of them, 85%, declare that they are involved with domestic tasks only that include tending livestock. 48 Relatively few women (14%) declare that they work in the family agriculture or business activities without salary. 3.3% are heads of family (Table 21). As the Table indicates, the main source of income of the family does not affect much the way rural women participate in the labor force: in working households, women tend to be less involved in family agriculture, supposedly because there are more landless households in this category. A quarter of the women in the remittance-dependent household are heads of family: these households may have lost their male household head, while adult sons have left the village altogether for permanent employment or to set up their own household. 48 Tending livestock as part of "domestic tasks" is our interpretation. The HBS questionnaire does not give the respondent the choice of the response, and just record whether the woman is working for a wage, as unpaid laborer, or is working on domestic tasks. Rural women answered that they are working "on domestic tasks only", as attending livestock is for them probably part of their domestic tasks. 47 Table 21: Education and participation of rural women in the labor force according to main source of income of the household Farmer Ag Non-ag Entrepreneur Remittance Diversified All rural worker worker Unpaid, domestic tasks 83.8% 93.7% 87.2% 83.4% 81.0% 85.3% 85.1% Paid (all sectors) 0.2% 1.3% 1.2% 0.2% 0.0% 0.7% 0.6% Unpaid, agriculture 14.5% 3.9% 11.2% 15.4% 18.5% 13.0% 13.4% Unpaid non- agriculture 1.5% 1.1% 0.4% 1.0% 0.5% 1.0% 0.9% HH head 1.0% 0.4% 0.2% 0.2% 25.3% 3.2% 3.3% Yrs of schooling 0.7 0.3 1.8 1.4 1.9 1.3 1.3 3.51. We now look at the participation of women in the labor force according to the income level of the household (Table 22). The level of income does not change womens participation. The only noticeable difference is the fact that there are more women head of households in the two poorest categories (10.1% and 5.3% respectively). The higher the income, the less likely that a woman be the head of a household. Table 22: Rural women participation in the labor force according to income levels Poorest 2nd 3rd Richest Quintile quintile quintile 4th quintile quintile All rural Unpaid, domestic tasks 86.4% 84.9% 85.9% 85.2% 83.8% 85.1% Paid (all sectors) 0.6% 0.9% 0.6% 0.5% 0.5% 0.6% Unpaid, agriculture 12.5% 13.7% 12.7% 13.3% 14.2% 13.4% Unpaid non-ag 0.5% 0.5% 0.9% 1.0% 1.4% 0.9% HH head 10.1% 5.3% 2.1% 1.5% 0.3% 3.3% Yrs of schooling 0.9 1.2 1.4 1.5 1.6 1.3 3.52. Women education. Adult female education levels are noticeably lower than that of the adult men: on average, 1.3 years of schooling versus 5.9 for male adults. Women in non- agricultural worker households and remittance-dependent households tend to be slightly more educated (1.8 and 1.9 years of schooling respectively). The former includes government employee households, with more educated women than the average (2.3 years of schooling). The least educated are the women in households whose main source of income is agricultural wages (0.3 years o schooling). 3.53. Table 22 shows that women education level increases with household income. Women earning a salary in the off-farm sector, only 0.3% of all rural women, are considerably more educated than the average (9.5 years of schooling). Interestingly enough they are more educated than men in the same category (7.8 years of schooling). The hypothesis is either that women are employed more as white collar workers than men in this category (i.e. there may be a higher proportion of blue collar workers among the men). Or else, barriers for women to work are so high that only very highly educated females can have access to a formal non-farm job. 48 3.54. Most worrisome is the fact that rural women education is not catching up. To illustrate the progress in education over time, Figure 27 reports average education by age for four groups: urban males and females, and rural males and females. The graphs show substantial improvement for all but rural females. In rural areas, there is important progress for men, but education remains very low for women. Figure 27: Years of education, by age 14 12 Urban male Number of years of education 10 8 6 Urban female Rural male 4 Rural female 2 0 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 Age SYNTHESIS Relationship between rural household assets, livelihood strategies and welfare: An Econometric analysis 3.55. We have so far examined various descriptive patterns on spatial and inter-household inequalities in rural households in Yemen, and how those patterns are correlated with livelihood strategies, household assets, and resulting welfare. In this section, we combine those individual patterns into a consistent analytical framework and implement econometric analysis. This exercise crosschecks the findings of previous descriptive analyses and enables us to bring together more robust results. 3.56. We use the Ordinary Least Square estimation (OLS), the probit model of binary choices (probit) and the multinomial logit model in the econometric analysis. The OLS examines how rural household income is associated with different household characteristics. The probit models the relationship between (i) the discrete poverty status (being in poverty or not) and (ii) household 49 characteristics and livelihood strategies. The multinomial logit model is widely used as a model of occupational choice. We use the model to examine how household individual characteristics are associated with its livelihood strategy selection. A caveat of this approach is the econometrically identified relationship does not necessarily imply a causal one, but only statistical associations. Without well-designed datasets such as panel data, we cannot draw causality conclusions. Our aim here is to ensure that these statistical associations are consistent with those from previous descriptive analyses and infer the structural mechanism regarding rural poverty in Yemen. Rural livelihood strategies and welfare 3.57. We first examine how the main income source of the household is associated with its welfare conditions, which are represented by total household income and poverty level. In estimation, we first control for geographical and locational endowments in which a rural household is located. The first variable is whether the sample rural household is located in poor/medium-poor/rich districts capturing spatial externalities. Households located in a rich district should enjoy better economic opportunities with strong demand for local produce, labor and business opportunities. The second variable is the average travel time (minutes) to the nearest town of more than 20,000 populations. It represents market accessibility assuming local markets would be located in towns of at least 20,000 residents. Finally, agroecological zone variables represent geoclimatic environments where a rural household is located. We use the same agroecological zones as before, with characteristics combining rainfall, evapotranspiration and soil temperature, which indicate whether a household is located in: (i) good agroclimatic conditions for two crop cultivations in a year, (ii) good agroclimatic conditions for one crop cultivation in a year, (iii) only one short cropping season, (iv) cropping possible with supplemental irrigation and rangeland. The baseline is a zone of irrigation only and rangeland. 49 3.58. Table 23 shows the results. Column 1 shows how total household income is correlated with the origin of the main income, after controlling aforementioned locational assets. Column 2 is the results of probit model estimation of being trapped in rural poverty 50 . Comparing both estimation results shows interesting patterns. The baseline is diversified sources of income meaning by our definition that there is not a single source of income that makes up for 60% of the family income. Being specialized in farming (farmers), compared to the diversified income sources, is more likely to be correlated with higher household income and lower poverty incidence. The same is for the rural households specialized in private business activities. Non- farm wage workers, whether public or private tend to have higher household incomes than the diversified households, but they show different poverty patterns: members of households of private non farm workers tend to be poorer than the diversified income households, while government employee households tend to be less poor. There is no statistical difference between agricultural wage workers and those with diversified income sources. 3.59. Finally, interestingly, rural households relying on remittances are more associated with lower poverty incidences even though their income is lower, again compared to households with diversified income sources. Then the question is why we obtain those association patterns in terms of main source of income and corresponding welfare conditions. We will explore further in Table 24. 49 The GIS agroecological zone layers are from the Agricultural Research and Extension Authority (AREA) in Yemen. 50 We recall here that "income" is household income, while "poverty" is individual poverty. The result is that a household could have a relatively adequate income but would still have its members classified as "poor" depending on the number of household members.. 50 Table 23: Rural livelihood strategies and welfare (1) (2) Dependent variable: Total hh income Being in poverty Sample: Rural households Rural households Estimation method: OLS Probit Livelihood strategy dummy: 0.229*** -0.103* Farmers (0.048) (0.059) Livelihood strategy dummy: -0.055 0.116 Agricultural workers (0.074) (0.117) Livelihood strategy dummy: 0.423*** 0.192*** Non-farming workers (private) (0.055) (0.073) Livelihood strategy dummy: 0.318*** -0.347*** Non-farming workers (public) (0.042) (0.079) Livelihood strategy dummy: 0.720*** -0.150* Entrepreneurs (0.058) (0.087) Livelihood strategy dummy: -0.232*** -0.221** Relying on remittances (0.061) (0.101) ln(travel time to town of -0.022 0.031 more than 20,000 population) (0.018) (0.026) Being located in medium-poor 0.095** -0.709*** Districts (0.044) (0.056) Being located in rich districts 0.380*** -1.548*** (0.041) (0.060) Good agroclimatic conditions -0.247** -0.288** for two crops (0.112) (0.114) Good agroclimatic conditions -0.042 0.036 for one crop (0.044) (0.060) Only one short cropping season -0.242*** 0.003 (0.048) (0.066) Risky cropping with irrigation -0.043 0.021 (0.051) (0.069) Constant yes yes Observations 4803 4824 R-squared 0.083 1. Baselines of livelihood strategies, district types and agroclimatic zones are diversified income sources, poor districts and no cropping zones respectively. 2. Robust standard errors are in parentheses. 3. * significant at 10%; ** significant at 5%; *** significant at 1%. 51 Household assets and main source of income 3.60. Table 24 reports the multinomial logit estimation results examining how household individual characteristics are associated with its main source of income (which we refer to as the HH livelihood strategy) while controlling for locational endowments. The estimated equations provide a set of probabilities for job selection choices for a decision maker (a household) with (household) characteristics. We examine how the adult ratio (an inverse of dependency ratio), the years of schooling of male adults, landholdings (the size of cultivated lands), and access to irrigation of a rural household are associated with its livelihood strategy. The baseline is again with diversified income sources. 3.61. First, as more household members are in economically active ages from 15 to 64 years old, the household tends to be in the category of diversified income sources. As already discussed, these households tend to be three generation households; and. presumably if the activities of the head of the household are not enough to employ younger adult males and support more than one nuclear family, the younger adult males have to seek other economic activities than only working with their father. 3.62. Second, higher years of schooling of male adults is associated positively only with government employment and is negatively associated with farmers and agricultural workers. There is no correlation between education and entrepreneurs nor with private non-farm employment. It suggests that in the rural areas, the educated labor force finds jobs mostly in the government sector and this tendency is quite strong. Access to land and irrigation is positively associated only with farming as the main income source. It confirms that irrigation is a major input in order for rural households to specialize in farming activities. Household assets and welfare 3.63. We analyze statistical and econometric associations between main source of income, welfare, and household assets. In this section, we examine the direct linkage between household assets and resulting welfare conditions of rural households. Again, we add to regressors the same set of locational assets as control variables. 3.64. Table 25 examines the statistical associations between household assets and two welfare measures of household income and poverty level. A stronger demographic composition (more household members in economically active ages) is associated with better welfare conditions (higher household incomes and lower poverty incidence). The same is for the years of schooling of male adults, who are responsible for virtually all household economic activities. Interestingly, while both landholding and access to irrigation are associated with higher household incomes, access to irrigation is more strongly linked to lowering poverty incidences. It suggests that regarding pathways out of rural poverty, access to irrigation seems to be more critical than landholding. 52 Table 24: Household assets and rural livelihood strategies (1) (2) (3) (4) (5) (6) Non- Non- Dependent variable: Farm farming farming Entre- Relying on Farmers livelihood strategy workers workers workers preneurs remittances (private) (public) Rural Rural Rural Rural Rural Rural Sample: household household households household households households Estimation method: s s Multinomial logit s 2 Share of adult household -0.756*** -0.085 -0.247 3 -1.915*** -1.423*** -0.709* members (0.246) (0.397) (0.244) 4 (0.310) (0.303) (0.371) ln(yrs of schooling of male -0.385*** -0.405*** 0.017 5 1.711*** -0.016 -0.003 adults) (0.053) (0.089) (0.058) (0.206) (0.064) (0.094) ln(cultivated land, square 0.189*** -0.155*** -0.133*** -0.124*** -0.088*** -0.127*** meters) (0.016) (0.032) (0.016) (0.017) (0.018) (0.022) dummy: 1.338*** -0.785* -1.096*** -0.734*** -0.781*** -0.667** irrigation by springs, wells (0.118) (0.442) (0.243) (0.230) (0.229) (0.331) ,dams ln(travel time to town of -0.008 -0.409*** -0.104 -0.141* -0.330*** 0.277** more than 20,000 (0.064) (0.085) (0.069) (0.073) (0.062) (0.132) population) in medium-poor Being located 0.336** 0.137 -0.104 -0.148 0.288 -0.045 districts (0.139) (0.235) (0.140) (0.153) (0.191) (0.226) Being located in rich districts -0.070 -0.181 -0.098 -0.009 0.920*** 0.289 (0.138) (0.243) (0.137) (0.151) (0.178) (0.214) Good agroclimatic conditions -1.185*** -33.639*** 0.385 0.304 -0.573* 0.860*** for two crops (0.296) (0.290) (0.235) (0.278) (0.307) (0.322) Good agroclimatic conditions -0.348** -0.963*** -0.032 -0.062 -0.033 -0.023 for one crop (0.143) (0.310) (0.158) (0.168) (0.182) (0.244) Only one short cropping season -0.099 -0.573** -0.162 -0.268 0.076 0.423* (0.146) (0.268) (0.169) (0.184) (0.181) (0.224) Risky cropping with irrigation -0.247 -0.514* -0.046 0.034 -0.090 0.189 (0.160) (0.289) (0.176) (0.175) (0.207) (0.254) Constant yes yes yes yes yes yes Observations 4480 4480 4480 4480 4480 4480 1. Baselines of livelihood strategies, district types and agroclimatic zones are diversified income sources, poor districts and no cropping zones respectively. 2. Robust standard errors are in parentheses. 3. * significant at 10%; ** significant at 5%; *** significant at 1%. 53 Table 25: Household assets and welfare: a base model (1) (2) Dependent variable: Total hh income Being in poverty Sample: Rural households Rural households Estimation method: OLS Probit Share of adult household 0.139* -0.418*** members (0.074) (0.108) ln(yrs of schooling of male 0.229*** -0.117*** adults) (0.019) (0.024) ln(cultivated land, square meters) 0.057*** 0.006 (0.005) (0.006) dummy: 0.094** -0.296*** irrigation by springs, wells ,dams (0.043) (0.066) ln(travel time to town of -0.057*** 0.026 more than 20,000 population) (0.018) (0.027) Being located in medium-poor 0.101** -0.707*** districts (0.043) (0.058) Being located in rich districts 0.355*** -1.582*** (0.041) (0.062) Good agroclimatic conditions -0.404*** -0.339*** for two crops (0.117) (0.120) Good agroclimatic conditions -0.147*** 0.038 for one crop (0.047) (0.065) Only one short cropping season -0.286*** 0.005 (0.049) (0.069) Risky cropping with irrigation -0.174*** 0.015 (0.053) (0.071) Constant yes yes Observations 4461 4480 R-squared 0.125 1. Baselines of district types and agroclimatic zones are oor districts and no cropping zones respectively. 2. Robust standard errors are in parentheses. 3. * significant at 10%; ** significant at 5%; *** significant at 1%. Model interpretations and implications for migration 3.65. This section combines econometric analyses in Tables 24 to 26 and discusses model interpretations and implications focusing on household assets. The main productive inputs of a rural household are household adults. The size (measured by the adult ratio) and the quality (measured by the years of schooling) of adult labor force in a rural household are critical for its income and poverty levels. 54 3.66. Landholding and access to irrigation are two major inputs for farming activities and result in rural households specializing in farming. In this regard, we also find strong statistical association between land/irrigation and improved welfare conditions. However, we should understand that farming and business (enterprise) activities requires significant amount of start-up capitals, which poor rural households cannot afford. The statistical association between land, productive capital and household incomes may be misleading. 3.67. Higher education opens up opportunities for government employment. Educational attainment is also important for the welfare improvement of the households not in the government sector. As shown in columns (1) and (2) in Table 26, years of schooling of male adults are strongly associated with higher household income and lower poverty incidence even for a subset of rural households not specialized in public non agriculture employment. 3.68. There can be alternate options which can be replicated to a larger group of poor rural households without adequate productive assets. The clue comes from Table 27. It shows that a rural livelihood strategy relying on remittances from family members that have permanently moved out is statistically significantly associated with lower poverty incidences (even though lower income levels). Relatively lower income levels of remittance-relying households will suggest lack of productive assets to produce sizeable income levels. However, those households are relatively less vulnerable and prone to fall into severe poverty, as family members who have left permanently ensure the subsistence of whose remaining in the village in times of stressful situations. 3.69. We can draw similar conclusions for rural households with temporary migrant workers. Columns (3) and (4) in Table 26 show the welfare consequences of seasonal migration for working. Having seasonal (temporary) migrant household members are positively associated with higher household income and higher poverty incidence after controlling for other key attributes. Poor households tend to send their household members for seasonal/temporary working in other areas, and as a consequence of more income opportunities are more likely to earn higher income than those without temporary migration. In sum, "exporting" family members is one of the strategy to alleviate the severity of the poverty of the sending households. 55 Table 26: Household assets and welfare: extended (1) (2) (3) (4) Total hh Being in Dependent variable: Total hh income Being in poverty income poverty Rural households, excluding public non-farming Sample: Rural households workers Estimation method: OLS Rural households Probit OLS Probit dummy: households with 0.209*** 0.158*** temporary migrant workers (0.034) (0.048) Share of adult household 0.134 -0.522*** 0.107 -0.445*** Members (0.082) (0.116) (0.074) (0.109) ln(yrs of schooling of male 0.232*** -0.071*** 0.232*** -0.115*** adults) (0.021) (0.026) (0.019) (0.024) ln(cultivated land, square 0.059*** -0.002 0.058*** 0.006 meters) (0.006) (0.007) (0.005) (0.006) dummy: 0.097** -0.290*** 0.117*** -0.282*** irrigation by springs, wells (0.045) (0.068) (0.043) (0.067) ,dams ln(travel time to town of -0.064*** 0.018 -0.068*** 0.017 more than 20,000 (0.021) (0.029) (0.018) (0.027) population) in medium- Being located 0.113** -0.722*** 0.109** -0.702*** poor Districts (0.047) (0.063) (0.042) (0.058) Being located in rich 0.374*** -1.601*** 0.378*** -1.568*** districts (0.046) (0.066) (0.041) (0.063) Good agroclimatic -0.449*** -0.293** -0.434*** -0.371*** conditions for two crops (0.134) (0.128) (0.116) (0.121) Good agroclimatic -0.137*** 0.093 -0.169*** 0.021 conditions for one crop (0.052) (0.069) (0.047) (0.065) Only one short cropping -0.296*** 0.002 -0.290*** 0.002 season (0.054) (0.073) (0.049) (0.069) Risky cropping with -0.170*** 0.027 -0.177*** 0.013 irrigation (0.059) (0.077) (0.052) (0.071) Constant yes yes yes yes Observations 3965 3981 4461 4480 R-squared 0.125 0.134 1. Baselines of district types and agroclimatic zones are oor districts and no cropping zones respectively. 2. Robust standard errors are in parentheses. 3. * significant at 10%; ** significant at 5%; *** significant at 1%. 56 Other options for coping or getting out of poverty 3.70. We examine three alternative options which may reduce the likelihood of being trapped in rural poverty. Those are (i) investing in livestock, (ii) investing in female education and (iii) social welfare programs. A caveat of this approach is that we only evaluate the impacts of "current/historical" interventions by statistically associating actual inter-household variations in both dependent and independent variables in the estimation model. The modeling results do not predict the impacts of "future" interventions in different scales. For example, the impacts of massive increase of the government social welfare program, such as a ten-fold increase in the fund, cannot be projected, as rural households will react differently to this windfall and those patterns are not observed in the historical data. 3.71. The first two columns in Table 27 show the impacts of investing in livestock versus crop and qat cultivation in the probit model framework of being trapped in poverty. Livestock is amongst the few assets of the poor households and one of their few income generating opportunities. Raising livestock as a cash income source, measured by the share of livestock net income in the total household income, is strongly associated with lowering poverty incidence (column 1), and the association is statistically significant (after controlling for baseline factors identified above). Column 2 compares another farming income source, namely net income from crop and qat cultivation. Crop and qat cultivation as a cash income source is also associated with lowering poverty incidence. However, the estimated marginal effect of crop/qat cultivation on reducing the probability of being trapped in poverty is quite lower than that of livestock production. The marginal effect, measuring the change in the probability (of being in poverty) for an infinitesimal change in an independent variable, is -0.1% for crop and qat cultivation and - 1.38% for livestock production. Raising livestock as an income source is important viable option for rural households to escape a poverty trap. 3.72. Column 3 of Table 27 examines the impacts on investing in female education. Female education, measured by the years of schooling of female adults, does not seem to show statistically significant effects on rural poverty reduction. This would be mainly due to very limited participation of female labor force in household economic activities: only less than 1% of rural females earn a salary. However, female education should be considered as an investment for the future, enabling more inclusive and sustainable development in a medium and long term perspective. 3.73. Finally, columns 4 and 5 in 27 examine the statistical association between being trapped in poverty and receiving cash transfer from the government Social Welfare Fund, measured by a dummy whether receiving cash transfer from the Fund, and ln (the amount of the government cash transfer). The results show a strong positive correlation between receiving the government cash transfer and being in trapped poverty, which reflects the fund design of targeting the poor. However, we cannot find empirical evidence that the amount of cash transfer from the Social Welfare Fund is large enough to effectively reverse rural poverty incidence even though it may alleviate the severity of rural poverty. 57 Table 27: Household assets and being trapped in rural poverty (1) (2) (3) (4) (5) Dependent variable: Being in poverty Sample: Rural households Estimation method: Probit share of net income from -0.038** livestock products (0.019) Share of net income from -0.003** crops and qat (0.001) ln(yrs of schooling of female -0.022 adults) (0.033) dummy for receiving cash 0.426*** transfer social welfare fund from the (0.076) ln(cash transfer from the social 0.046*** welfare fund) (0.008) Share of adult household -0.416*** -0.416*** -0.377*** -0.471*** -0.464*** members (0.108) (0.108) (0.110) (0.108) (0.129) ln(yrs of schooling of male -0.120*** -0.120*** -0.116*** -0.123*** -0.115*** adults) (0.025) (0.024) (0.025) (0.024) (0.030) ln(cultivated land, square meters) 0.006 0.006 0.004 0.006 0.006 (0.006) (0.006) (0.006) (0.006) (0.008) dummy: -0.292*** -0.292*** -0.296*** -0.290*** -0.340*** irrigation by springs, wells ,dams (0.067) (0.066) (0.067) (0.066) (0.083) ln(travel time to town of 0.026 0.027 0.027 0.025 -0.000 more than 20,000 population) (0.027) (0.027) (0.027) (0.027) (0.033) Being located in medium-poor -0.706*** -0.707*** -0.714*** -0.688*** -0.629*** districts (0.058) (0.058) (0.059) (0.058) (0.074) Being located in rich districts -1.586*** -1.586*** -1.589*** -1.567*** -1.526*** (0.063) (0.063) (0.063) (0.063) (0.078) Good agroclimatic conditions -0.332*** -0.331*** -0.322*** -0.309*** -0.450*** for two crops (0.120) (0.120) (0.121) (0.119) (0.152) Good agroclimatic conditions 0.038 0.038 0.048 0.044 0.062 for one crop (0.065) (0.065) (0.066) (0.065) (0.078) Only one short cropping season 0.006 0.006 0.006 0.010 0.004 (0.069) (0.069) (0.069) (0.069) (0.084) Risky cropping with irrigation 0.019 0.019 0.023 0.034 -0.009 (0.072) (0.072) (0.072) (0.072) (0.088) Constant yes yes yes yes yes Observations 4474 4474 4430 4480 3088 1. Baselines of district types and agroclimatic zones are poor districts and no cropping zones respectively. 2. Robust standard errors are in parentheses. 3. * significant at 10%; ** significant at 5%; *** significant at 1%. 58 Conclusion 3.74. Either adequate land and water access or education is key to make a decent living. For those households with sufficient land access compounded with access to irrigation (or operating under good rainfed conditions), agriculture as the main source of income can provide a good livelihood. But these are few: according to poverty data, only about 12% of the households are in that situation. These initial land-assets are linked to other assets, animals in particular, as well as investment in entrepreneurial activities 51 . Initial assets are also necessary to make businesses profitable enough to provide a good livelihood for a whole family. Finally, education opens the way for an adequate living, mainly through government employment. However, new opportunities in that sector are likely to be very limited, since the absorptive capacity of the public sector has reached its ceiling. New employment opportunities are likely to correspond to attritions only, if at all. 3.75. Coping strategies for households with very limited assets, in particular land. We reckon that a large proportion of the rural households, more than 75% of the rural households, are in a situation of limited assets which makes them dependent on their labor for a living. Even if not of all them are poor, most of the non poor are living at the margin of subsistence, which makes them extremely vulnerable to shocks: disease of an active family member, loss of employment, natural disasters, droughts and price fluctuations, as evidenced by the impact of the 2008/9 food and financial crisis. Poverty and vulnerability are not associated with a particular livelihood, but with lack of assets. There are more poor people amongst the households whose main source of income is salaried labor as well as amongst the diversified-income households. 3.76. Households with limited assets cope with a combination of the following sources of income, though as we saw, one source of income may dominate: For those with access to land, some subsistence production (cereals and legumes), and limited crop sales (mainly, depending upon the location of the HH, qat, horticulture or coffee); Animal production for monetary income, even for those without access to land for crop cultivation; Either salaried labor, associated with the migration of the adult men; or Permanent out migration of part of the family, while the family members remaining in the village rely on their remittances. How can these households improve their livelihood and what government and donors can do to support them? This will be the subject of the following section. 51 As we saw in Table 10, enterprise income of entrepreneurs with land is almost double (1.9 times) the enterprise income of entrepreneurs without land. 59 4. IMPLICATIONS FOR GOVERNMENT AND DONORS POLICIES AND PROGRAM PRIORITIES TO TARGET THE RURAL POOR Existing policies and programs 4.1. The Government interventions for poverty reduction in rural areas are framed by two main policy instruments: the Third Socio-Economic Development Plan for Poverty Reduction (DPPR) (2006-10) with its 8 priority pillars, and the Rural/Local Development Strategy formulated in 2003-2004. Though the DPPR provides a prioritized framework for action, it does not specifically address the specific poverty environment of the rural areas. Its 8 pillars, though entirely relevant are still too broad to prioritize actions, especially in a context of fiscal deficit.52 A mid-term review of the DPPR in January 2009 took stock of the implementation progress and highlighted these weaknesses. 4.2. As to the RLDS, it remains valid and still provides a solid conceptual framework for interventions in the rural areas. But it embraces all sectors and does not help setting priorities among programs that are competing for scarce resources. .In addition, though it was initially thought that it would support the implementation of the DPPR in rural areas, it does not specifically address the needs of the rural poor: it is conceived as a strategy to develop rural areas. Proposed objectives for a rural poverty alleviation program 4.3. Our intent here is not to propose another rural/local development strategy, but keeping the RL/DS framework and the DPPR as a framework, to build upon our analysis of the livelihood characteristics of the assets-poor rural population, in order to identify priorities for policies and programs that specifically target the needs of asset-poor rural people. The objective is to propose a well prioritized and well targeted portfolio of programs that specifically address rural poverty and that can be part of the Fourth Socio-Economic Development Plan for Poverty Reduction, presently under preparation and scheduled to begin in 2011. 4.4. Basic infrastructure and services. Yemen is far away from achieving the Millennium Development Goals. In the rural areas, the human development indicators are even lower than the national average. Our analysis has shown that beyond the rural-urban poverty gaps, spatial differences within rural areas are considerable, with rural poverty being concentrated in some districts/governorates. Low income districts with also tend to have worse access to basic infrastructure and services than higher income districts. As shown earlier, basic infrastructure and services, as measured by the District Access index vary considerably from one district to the next (Table 13 and 14). Reducing poverty in rural Yemen requires among others, village-level basic infrastructure and better access to quality social services. 4.5. Beyond the obvious need of providing better access to basic infrastructure and social services, how can the Government and donors help rural poor make a better living? We determined that the coping strategies of poor and vulnerable rural people combine limited crop 52 The eight pillars are as follows: (i) macro-economic policies and targets; (ii) good governance: (iii) development of productive and promising sectors; (iv) water, environment and basic infrastructure; (v) human development; (vi) Government services; (vii) social safety net, social protection and social security; (viii) womens empowerment. 60 production for sale and for subsistence, some animal production for monetary income; and/or either salaried labor, largely based on the migration of the adult men or permanent out migration of part of the family, while the family members remaining in the village rely on their remittances. Farming predominates depending upon the extent to which household have access to land and water. 4.6. Protection of poor-people assets. In a context of rapid population growth and resulting intense pressure on natural resources, compounded by the fact that rules of law are being blurred - the formal system of governance is not being well established yet while the customary one is weakened- the poor and not well-connected are put at a disadvantage. They are likely to lose out to more influential members of the society in case of conflicts or ambiguous rights. Whatever few assets poor people have in terms of land and water, need to be protected. Other assets that need protection is livestock: animals constitute poor people main savings instrument and insurance against risks. Hence the importance of a well functioning animal health protection system. Finally, one aspect of poor peoples assets which is often overlooked is their arable land. Poorer farmers are more dependent on rain for crop production than better-off farmers. Rain-fed production in Yemen is by and large dependant on an elaborated water and soil harvesting system of terraces and cisterns, which represents thousands of years of labor. Poor farmers may not have the resources to maintain it and need help. 4.7. Enhanced and secured productivity. Poor people agricultural activities can be made more productive, and productivity need to be secured, given the risks associated with rainfall uncertainty, heightened by climate change. Animal production can be enhanced through better animal health and nutrition. Food security can be expanded somewhat, with enhanced drought- resilience and higher crop productivity, that can be achieved through enhanced soil and water management, and through low cost technologies such as better seed management, using more drought-resilient varieties. Some households can make a better living with their coffee; more households can be involved with honey production. 4.8. Successful exit out of agriculture and rural areas. Yemeni rural labor, as we saw, is largely unskilled labor. An essential aspect to improve the livelihood of rural people whose only asset is their labor will be to help them access more and better economic opportunities either in the rural non farm economy or through migration to urban centers or abroad. However, finding and maintaining employment requires broad-based occupational skills or specific job-related skills. Hence, investing in the skills of young rural people is essential to prepare them as well as the next generation to migrate successfully out of overpopulated rural areas. 4.9. A better future for poor people's children. Many poor families cannot afford to send all their children to school, girls are the most affected. Economic and other social factors keep girls away from school. The issue of girl education in particular needs concerted efforts if the next generation of adult women is going to be empowered, which will have a positive impact on the well-being of the entire family. 4.10. Social safety nets. Finally, a number of poor households, especially the ones in extreme poverty (aged, disabled, and/or destitute female-headed households) will continue to depend on cash transfers to survive. 4.11. To sum up, the objectives of the proposed policies and programs for the rural poor are the following: 61 Increase investments in districts with the lowest access and highest poverty incidence Improve Access to basic services Improve local governance and capacity of district councils to manage resources efficiently and target the poor Implement the water sector reforms together with pro-poor water programs and improve Protect poor people limited groundwater local governance assets Enact rural land regulatory framework to ensure access to waqf and communal land Secure crop production in rain fed areas through water harvesting and soil conservation Enhance and secure the productivity Invest in animal health and nutrition programs of poor people assets Improve cereal and legume production in rain fed areas Create more local value added in lucrative value chains: coffee and honey Ensure a better future for poor Enhance school attendance especially that of girls people's children Enhance the quality of education, with a specific focus on vocational training Improve access to more and better employment opportunities Invest in labor intensive rural public work programs Enhance the social protection Improve the targeting of the Social Welfare Fund system 62 A rural pro-poor program Increase investments in rural districts with high poverty incidence and lowest Access Index 4.12. Spatial targeting of public spending. One way to target the poor is through spatial targeting in public spending for basic services, i.e. increasing public spending per capita in the districts with the highest poverty incidence, a strategy of "investment in people in lagging areas".115 Given the limited budget availability and the immensity of the needs, we suggest here a finer targeting, crossing two indicators: poverty incidence and District Access Index116, therefore targeting districts with both the highest poverty incidence, and, at the same time, the most underserved in terms of access to basic services. Districts would be selected on the basis of a poverty incidence higher than the rural average (40.1%) and of a District Access Index lower than the rural average (39.8%).117 There are 64 such districts, responding to the two criteria, with 2.3 million poor rural people and a total rural population of 3.6 million. The average rural poverty rate of these districts is 63.4%. Annex II presents the list of these 64 districts. 4.13. Two channels: (i) district budget allocation; and (ii), Social Fund for Development and Public works Program. Resources can be channeled to districts in two ways. The first is through district budget allocation, rather than through sectoral ministries. Even if sectoral ministries receive a clear mandate to focus their investments on the poorer rural districts, these ministries may have different priorities. By contrast, decentralized service delivery presents the potential to better reach vulnerable groups. The legal framework exists: Law No 4 of 2000 concerning the Local Authority provides the framework for fiscal transfers and empowerment of elected representatives at the district and governorate levels. 4.14. The other avenue to target poor districts is through channeling funds through the Social Fund for Development (SFD) and the Public Works Program (PWP). The SFD in particular has a solid track record of channeling resources to the neediest areas and people. According to recent studies and evaluations, 69% of SFD resources are benefiting the poorest household (i.e the bottom three deciles). 118 During the period 2004-2008, 65% of geographically targeted investments went to areas with a poverty index above 50%, comprising a population of 9.2 million (73% of them poor).119 Though it was not the case in its first two phases, the Public Works Program, which implements similar infrastructure projects to those of the SFD, has developed a strategy which allocates available funds to governorates according to population density, poverty levels, deprivation and remoteness. Improve local governance and the capacity of district councils to manage resources efficiently and target the poor 4.15. The lack of capacity of district councils stands in the way of improving local governance to target the poor. Simply increasing public spending per capita in the most underserved districts is unlikely to be sufficient to ensure better access to services for the poor, unless improvements in the efficiency of resource allocation and in the management of public funds are achieved. The lack 115 This is in accordance with the World Development Report 2009. Reshaping Economic Geography 116 See paragraph 2.15 to 2.17 for a discussion of the District Access Index. 117 The maximum District Access Index is 100%, which indicates that all households have most of their basic needs satisfied. 118 World Bank (2007). P 61 119 Pelat, F. (2009). 63 of capacity of local councils to allocate and manage resources efficiently and to ensure that funds are invested in quality and well-monitored projects has long been recognized as a major problem and has been one of the reasons for not fully implementing the Decentralization Law. The Ministry of Local Administration with support from a number of donors has been actively working on improving district councils capacity120, following the recommendations of the National Strategy for Local Government, adopted by Cabinet in October 2008. The Strategy envisions: (i) improving the capacities for planning at district level using participatory approaches; (ii) improving the quality of the representation of communities in District Councils. Box 2: Empowerment for Local Development Program The Social Fund for Development has a special unit dedicated to training and capacity building. This unit has been working with the Public Works Program to develop the capacity of the Local Councils to improve the allocation and management of district resources. One of the distinctive objective of this program is to build the capacities of the local councils to enhance targeting and improve access to services for all. To this end, the SFD facilitators are working with both district councils and village-level communities. They train the council representatives in communication skills and in participatory development planning. They help village-level communities to organize from village to sub-district to district and transfer to them skills to analyze their local conditions, identify needs and solutions and prioritize them. In a sense, the SFD is contributing to revive the traditional systems of social-capital and self-help which has been in decline since the 1970s. The end result from the process and SFD support is a participatory district development plan built on the needs and priorities of village-level communities, and incorporating communities own initiatives. The district development plan presents projects that donors and sectoral government programs can choose from. 4.16. Worth up-scaling: the "Empowerment for Local Development" Program of SFD. In line with the National Strategy for Local Government, and in collaboration with the Ministry of Local Administration, SFD has been piloting a program "Empowerment for Local Development" in 10 districts of seven governorates, and 11 additional districts are planned. The program aims at empowerment at all levels, from village-level communities, to sub-districts and District authorities. It provides mechanisms for bridging the gap between those levels, which includes community organization and promotion of self-help initiatives, as well as community control and auditing. The objective is to improve participatory planning capacity of the District Councils and the representation of village-level communities on the Councils so that the Councils be better able to respond to the needs of these communities (See Box 2). This would enable to head towards a more effective decentralization system, based on more capable and more transparent local governments through improved local governance. Part of the strengthening of local governance is 120 The Ministry of Local Administration (MOLA) has been coordinating the capacity-building programs in support to decentralization of a number of donors and agencies, in particular UNDP, Social Fund for Development, Public Works Program. The UNDP-financed Decentralization and Local Development Support Project, launched in September 2003, was the first strategic external support to decentralization. It addressed policy and institutional issues (including fiscal decentralization), local and central capacity, and financing. 64 improving the capacity of the communities to make their voices heard and to increase their bargaining power, so that local councils resource allocation decisions are made upon a solid documentation of the local conditions and well justified requests for investment. A partnership between local authorities and communities is the best way to promote better targeting, transparency and sustainability. Protect poor people assets through implementing the water sector reform together with pro-poor water programs and improve groundwater local governance 4.17. New irrigation technologies and cheap diesel fuel resulted in more unequal land and water access. As the analysis shows, land access is skewed. There are a number of interacting factors that are making the situation worse. One is the growing competition for land, as a result of population growth and of the economic opportunities based on qat production and other high value crops such as fruit. The competition for land to grow high value crops has been driven by cheap diesel fuel and new water technologies that make possible larger scale cultivation through faster water extraction at deeper levels, using tube wells.121 In spite of a national legislation which has been in place since 1980, that requires the granting of permits before installing a well or deepening an existing one, farmers with enough financial resources and larger landholdings have been drilling wells without respecting the well-spacing norms, or deepening existing wells without authorization.122 Water reserves are currently mined at twice the recharge rate, with poor farmers being the main losers, while better-off farmers have been able to capture the lions share of this common good. 123 Unsustainable drawing of groundwater has dried adjacent wells, and lower elevation springs, forcing poorer farmers, not able to invest into deeper drilling, to abandon their fields or sell them. A recent survey of land issues suggests that there is a trend towards concentration of land and water access in the hands of a few. 124 4.18. Water sector reforms have been enacted since 2003 but implementation is hindered by political economy constraints. The Government has undertaken reforms in the water sector, starting with the 2003 Water Law. The Law establishes a system of water rights and regulation that incorporates all the elements to be quite effective, if regulatory enforcement capacity increases. In 2005, the Government adopted a National Water Sector Strategy and Investment Program, which proposes specific institutional, financial and other measures to support water sector management and coordinate efforts. It is presently being implemented through the Water Sector Support Project, approved in 2009, but implementation of the Law remains hesitant because of the political economy constraints.125 4.19. One way of enhancing implementation is through improving local ground water governance. The importance of local action to improve ground water management, in particular through water users associations (WUA), has proven effective in a number of regions around the world with scarce water resources. In Yemen, however, improving local water governance through WUA, is likely to be a slow and arduous process: in many cases, large landowners either ignore the WUA or join them only to dominate the decision-making processes. In either case, the 121 For an analysis of the effect of new water technologies on poverty, see in particular: Ministry of Water and Environment and Ministry of Agriculture (2007). 122 World Bank (2009). P. 42 123 Ministry of Water and Environment and Ministry of Agriculture (2007). P vii 124 World Bank (2009). P 42 125 Ministry of Water and Environment and Ministry of Agriculture (2007). P viii 65 basic rule of the WUA, which is equitable water management for mutual benefit, is undermined126. There is a need to strengthen the WUAs and provide them with skilled administrative support, building on successful cases of water user associations that have started to regulate ground water extraction. 127 4.20. Water sector reforms need to be implemented as a package in order to enhance equity. The Government also reduced the implicit subsidy on diesel price twice, in 2005 and in 2008128 thereby reducing the incentive for extracting ground water at deeper levels. The analysis of the distributive impact of the reforms suggests that smaller farmers and landless laborers are indirectly affected by the increased fuel costs: poor rural people are faced with higher water costs and lower agricultural employment opportunities.129 The water sector reforms of the National Water Sector Strategy and Investment Program need to be implemented as a package, focusing not only on water saving and water productivity enhancement aspects of the reforms, but also enhancing equity, through pro-poor design and pro-poor entry criteria for subsidized water savings and productivity programs. Protect poor people access to waqf and communal land through enacting rural land regulatory framework 4.21. Poor people depend on communal and waqf land; private appropriation of these threatens their access. There are issues related to the management of various types of property rights, in particular waqf 130 and communal land. Waqf land is private land which has been placed perpetually in a trust for religious or charitable purpose. The land under waqf is rented out to provide income for the beneficiaries of the waqf, thus providing an opportunity for land-poor farmers to access land as tenants. Waqf is an important institution for the land-poor, mitigating land concentration. However, there have been problems with the administration of waqf land, in particular weak management and lack of inventory, thus facilitating the private appropriation of waqf, with the result that tenants are losing access. 4.22. People without access to land, or whose plots are small are largely dependent on communal land for livestock grazing.131 It is estimated that under normal weather conditions, 40% of livestock energy requirements are covered by pastures, more than 50% in the case of sheep and goats. Livestock as a source of income for the poor could be threatened by private appropriation of communal land. Customary law entrusts shaykhs with the management of communal land. In recent years, land speculation has increased, especially in areas close to towns, or in areas where land is becoming more valuable as a result of irrigation development projects. There is a growing confusion over the communal land entrusted to shaykhs and the land they own privately, resulting in private appropriation of communal land. 132 Community rights to communal land are often not documented which makes these communities particularly vulnerable.133 126 Ibid. p 28 127 There have been cases where WUAs succeeded in stopping drilling or in reporting unauthorized drilling to the National Water Resource Authority, or in enforcing space restriction between wells os protections zones. See Burns, B and T. Taher (2009). 128 The price of diesel fuel went from 17 to 35 RY/lit, but it still does not reflect the economic value of diesel fuel. 129 Ministry of Water and Environment and Ministry of Agriculture (2007). P 66 130 Waqf land is estimated to be 10 to 15% of all agricultural land. It cannot be sold, donated or inherited. 131 FAO (2009). 132 Some shaykhs have sold land alleged to be communal, outside the tribe, thus contravening customary norms. See World Bank (2009). P42 133 Ibid. P46. 66 4.23. Systematic land registration and recording of land rights is required to protect users' rights. Poor peoples access to communal and waqf land need to be protected and legal measures to prevent private appropriation of these lands need to be enacted. In particular, systematic land registration and recording of land rights is required to protect users rights. A law on land registration in urban and peri-urban areas is currently before Parliament. There is however, much less interest in promoting land registration and recording of land rights in rural areas. With regard to waqf land, in addition to registration, the Government needs to promote better management practices, which would contribute to improving the efficiency of this instrument to provide land access for the poor. Secure crop production in rain fed areas through water harvesting and soil conservation 4.24. Agricultural income and poverty is inversely correlated with access to irrigation (springs, wells, spate). We saw that in districts with high poverty incidence, there is less irrigation; in all districts, poorer farmers are more dependent on rain for crop production than better-off farmers. Farmers without access to irrigation are operating under very harsh conditions: low and erratic patterns of rainfall, compounded with high potential for soil erosion in mountainous areas. 4.25. For poor farmers, the only solution is to better harvest and conserve rain water and soil. Water harvesting takes several forms: (i) terraces, which retain soil and water; (ii) small check dams and retention structures; (iii) tanks and cisterns. Water harvesting has been the foundation of Yemeni agriculture and has been practiced since immemorial times.134 Man-made terraces in particular are part of the Yemeni landscape: it is estimated that they account for 20-25% of all cultivated land. 135 Terraces make crop cultivation possible in areas of erratic and low rainfall, and generally increase crop productivity through retaining soil moisture. It has been estimated that terraces receive up to 40 % more moisture than that from direct rainfall on the soil.136 Some studies in Amran Governorate revealed that with terraces, crop growth is possible even with as little as 150-300 mm rain per year.137 This explains in large part how agriculture has been viable even in areas where rainfall alone would not be enough to grow crops. In addition to retaining moisture, in mountainous areas, terracing is the main solution to controlling water erosion, one of the major causes of soil loss and soil degradation. Terracing builds up plots of fertile soil, years after years. Without terraces, most of Yemen highlands would be barren. 4.26. The externalities of water harvesting through a system of terraces has long been recognized, though hard to quantify. The benefit of water harvesting, in particular terracing goes beyond the field of the farmers who build the terraces. Though the externalities from watershed management in arid lands and mountainous areas are hard to quantify, it has long been recognized that terraces intercept surface runoff and force it to infiltrate. The result is: (i) improved recharge of aquifers and of existing springs and subsurface wadi flows downstream; (ii) controlled runoff velocity thus protecting the lower lands and wadi downstream from destructive floods. 138 Though the models do not agree whether climate change will lead to a wetter or drier climate, overall there is little doubt that climate variability will increase, characterized by even more erratic patterns of rainfall and more intense rainfall events, thus increasing the risk of floods. Watershed 134 The first terraces in Yemeni mountains were built nearly 3000 years ago. See Wilkinson (1999). 135 Bamatraf et al. (2000). Al-Hebshi (2005). 136 Varisco (1991). 137 Ministry of Water and Environment and Ministry of Agriculture (2007). P vii 138 Eger (1984). 67 management that includes terracing is one of the recommended adaptation policy options to climate change. 139 4.27. The problem is that terraces are being abandoned or insufficiently maintained. The problem is that over the last two decades, a process of terrace abandonment as well as terrace degradation has been taking place. Recently, a survey on terraces was conducted in 45 communities as part of the base line study of the Rainfed Agriculture and Livestock Project (RALP), in an attempt to shed some light on the situation. The survey found that 20% of the terraces were abandoned and two-thirds of the communities reported degradation on used terraces. People cite the shortage of rainfall, making the cultivation of terraces non-economical as the number one reason for abandoning terraces. They also cite the lack of manpower due to the men migration as the second reason for abandoning terraces, and the main reason for poor maintenance.140 The high costs of building or maintaining terraces have also been cited in other studies: hired labor, material (walling stones) and transportation of these materials", highlighting in particular the increasing cost of hired labor due to the migration of the adult men, which corroborates the findings of the RALP base line survey.141 4.28. The severe consequences with regard to control of flash floods and groundwater recharge attract only limited attention from donors and GOY. Terrace abandonment progressively leads to gully erosion and collapse of the retaining walls, resulting in the destruction of the whole terrace system through a domino effect. Severe modifications of the hydrological regime of the whole basin may result, earmarked by more violent and destructive flashfloods and drastic reduction in groundwater recharge especially in major coastal wadis. In spite of the importance of watershed management and the maintenance of terraces for groundwater recharge and control of flash floods, the Water Sector Support Project is not concerned with this issue. Only a few projects promote water harvesting and soil conservation, mainly: the Ground Water and Soil Conservation Project 142 and the Rainfed Agriculture and Livestock Project.143 4.29. More investments in watershed management is required, for poverty reduction and externalities, and for climate change adaptation. Public investment on private land is justified because of the externalities of water harvesting and conservation in the Highlands. However, investments under these projects remain limited, compared to the potential not only to improve the productivity and secure crop production of poor farmers with limited land operating under rainfed conditions, but also because of the larger societal benefits of watershed management. These benefits are likely to take on more importance under climate change. Also with the dwindling of ground water resources, farmers may have to revert to rainfed crop production in the short to medium term. More attention and financial resources are needed, taking a watershed management approach, including the rehabilitation of terraces; the reforestation of the watershed upper catchment areas; the construction or rehabilitation of cisterns for human and animal consumption (sometimes linked to the introduction of a drip irrigation system for supplemental irrigation of 139 FAO (2009). 140 Egel,D. and T. Al-Bass Yeslam (2010). 141 In 2006-7, the Ground Water and Social Conservation Project estimated the costs at 220 000 YR/ha, approximately US $ 1,100 or 6-8 $/m of rehabilitated wall. The World Bank study on climate change cites much higher costs: 25 $/m of rehabilitated wall. See World Bank (2010a). The cost of terrace maintenance depends on the extent of degradation. It can reach one third of the cost of building the terrace. 142 A social impact assessment study is presently being carried out for the Ground Water and Soil Conservation Project. 143 The NGO CARE pioneered the funding of terrace rehabilitation and watershed management in Al-Mahweet Governorate through the Community Management of Local Natural Resources Project, now completed. 68 high value crops); wadi bank protection works and the improvement of small spate irrigation schemes. Invest in animal health and nutrition programs 4.30. Targeting animal health and productivity would be one way to improve poor people income and to reach rural women through their main economic activity. As we saw animal production is a widespread economic activity in rural Yemen, practiced even by people without access to land. Animals are often the only assets of the poorest households and one of their few monetary income-generating opportunities. These households could significantly increase livestocks contribution to their family income. In addition, women are the main livestock caretakers as they are in charge of herding, feeding, watering, milking and making ghee, with the help of their children. As such, livestock is one of the few economic activities where public interventions could directly impact rural womens livelihood. 4.31. Animal health is a serious concern for poor people, who often lose their only assets due to various endemic diseases144. The average annual animal mortality rate is high.145 it is likely that about one third of the herds are lost every year. Less than half of the households have a veterinarian to treat a sick animal and less than half vaccinate their animals, which is attributed either to the high cost, distance or unavailability of the veterinarian in 85% of the cases 146. In addition to loss of animals, another main constraint to livestock productivity and probably the cause of higher animal susceptibility to disease is poor animal nutrition due to lack of fodder and forage availability as a result shortage of rainfall. 4.32. Donors and GOY could do more in this area. Given the special role of livestock in the Yemeni rural economy, the limited attention given to the animal resources sector is somewhat puzzling. The General Directorate for Animal Resources of the Ministry of Agriculture and Irrigation, which lacks staff and resources to operate, has been mainly focusing on public funded disease control programs. The animal production, nutrition and livestock extension program is almost inexistent. Donor investments remain piecemeal. The main efforts are under the World Bank Rainfed Agriculture and Livestock Project, implemented in five governorates, with a component addressing animal health and veterinary services. IFAD area-based projects also supports livestock rearing and animal health among a number of other rural development activities. 4.33. Interventions in the sector should focus on: (i) upgrading the public animal health services, progressively privatizing parts of the services, and extending the present coverage limited to peri-urban areas, through a system of veterinary auxiliaries (See Box 2); (ii) improving animal nutrition through a better management of the resource constraints (water, forage and fodder), fodder productivity improvements and affordable diet supplementation (such as bone meal) to address phosphorus deficiencies. Livestock nutrition improvement programs would thus require good linkages with water harvesting (cisterns, fog or roof water harvesting) and watershed management programs (improved management of the upper catchment of the watersheds, introducing animal edible species (napier grass, legume trees) as well as management of other communal rangeland. 144 Diseases include rinderpest, foot and mouth disease, rift valley fever and sheep pox. 145 The households of the RALP base line survey questionnaire that reported losses, reported a loss of 35% of their herds in 2008/09. See Egel, D. and T. Al-Bass Yeslam (2010), p. 12 146 Ibid. p. 12 69 4.34. Programs to implement known livestock improving technologies must be carefully considered to ensure that: (i) proposals are acceptable to women; (ii) these proposals would reduce, and not increase the already heavy work load of women and children; and (iii) proposals are technically feasible and financially attractive. Improve cereal and legume production in rain fed areas Agricultural productivity, and in particular cereal productivity is very low, well below technical potential, and below actual farmer yields in comparable environments (Table 28). Table 28: Yields of cereal crops from 2004 to 2009 (tons) 2004 2005 2006 2007 2008 2009 1/ Sorghum 0.614219 0.61325 0.887005 0.964184 0.850749 0.853794 Maize 0.84252 0.80792 1.599824 1.666558 1.507847 1.513372 Millet 0.66823 0.66816 0.728396 0.74074 0.653592 0.651211 Wheat 1.23226 1.31336 1.347433 1.544333 1.38458 1.398509 Barley 0.70835 0.61391 0.750169 0.808358 0.715426 0.724763 Total 0.71522 0.71955 0.964767 1.056388 0.938897 0.933557 Source: FAO (2009) 4.35. Productivity improving technologies that could increase crop production without using more water exists.147 However, one cannot expect poor farmers to adopt improved seeds, fertilizer and other chemicals under arid conditions and highly variable rainfall patterns that make the use of improved technologies highly risky. Even when some form of supplemental irrigation is available, poor farmers face additional limitations that prevent them from adopting improved technologies, in particular the farm sizes, the lack of financial resources and difficulties to access markets.148 4.36. Among the reasons that explain low productivity is the deterioration of the quality of the seeds being used, which could be addressed with better seed management. The recommendation here would be to work with local landraces, rather than imported or improved varieties. Though imported/improved varieties would yield more, they would not be as well adapted to the specific local soil and climate conditions of the very diverse Yemeni agro ecological systems. The risk of crop failure would be much higher than with the well adapted local varieties, in case of insufficient rain availability, notwithstanding the fact that improved varieties are also more demanding in terms of inputs to express their potential. Hence better seed management is at least one area of crop technology improvement which can be proposed to poor farmers as it is a low cost/low risk technology (See Box 3). 147 World Bank (2010a). 148 The land size that will make investing in crop production a profitable business very much depends on the type of crop (i.e. qat vs. cereals) as well as whether the land is irrigated or not. 70 4.37. In addition improving crop production with local landraces would have the additional benefit of building climate resilience of rain fed agriculture. Climate change will bring more uncertainly than it is already the case. Poor people are likely to be affected even most severely as they have few other resources to withstand shocks. The agro-ecosystems of Yemeni highlands hold critical agro-biodiversity resources that can provide locally-based solutions to cope with climate change and enhance food security. The wide range of local varieties of cereals and legumes is the result of farmers purposeful selection under harsh and varying agroecological conditions. In this domain the activities of the Rainfed Agriculture and Livestock Project, at the moment implemented in 23 districts are certainly worth upscaling linked with the activities of the GEF financed Agrobiodiversity and Climate Adaptation Project (See Box 3). 71 Box 3: Consider upscaling: the Rainfed Agriculture and Livestock Project Community animal health workers. The project identified the lack and poor quality of animal health services (apart from the occasional vaccination campaigns) as one of the main threats for livestock, the assets of poor household. The project emphasizes the use of veterinary auxiliaries ("paravets"). The paravets are young livestock producers, mainly women, who receive a very basic training on how to deal with simple and common livestock ailment. Though the idea is not new and has been introduced with success in a number of countries, the innovation comes from the way that the trainees are identified. Under another component of the project, the Social Fund for Development supports the formation of community producer groups, i.e. villagers getting together around a common economic interest such as livestock raising. The livestock raising groups, mainly women groups, are given the opportunity to select one of them for being trained as paravet. The potential trainee has to meet a number of basic education criteria set by the General Directorate for Animal Resources. Once trained, the paravet receives a first aid kit and basic drugs that s/he is expected to replenish as s/he gets paid for the drugs and a minimum fee for the service rendered. S/he is expected to be working under the supervision of a full fledge veterinarian whom s/he can refer to in case of difficulties or particular concerns. The fact that the trainee is selected by the communities and returns to the communities after the training gives assurance that the paravet will be accountable to the villagers and trusted by them. An indirect benefit of the program is to create a network of private "first aid" villagers, that would provide limited "first aide" assistance in their village, but also play a role in alerting veterinary authorities to outbreaks etc. of specific diseases, creating thereby the base for an epidemio-surveillance network. Farmer-based informal seed production system. The RAL Project also focuses on the enhancement of rainfed cereals and legumes production through the conservation and improvement of local landraces and the establishment of a farmer-based informal seed production system in 23 districts. Landraces of sorghum, maize, millet, wheat, barley, cowpea, peas, lentil, faba bean and fenugreek are concerned. This includes: (a) Crop improvement: (i) the collection of indigenous landraces in different rainfed locations; (ii) their characterization and evaluation, testing the landraces for resilience to climate change effects; (iii) participatory genetic improvement for the most prominent varieties, and (iv) the ex-situ and on-farm conservation of the genetic resources. (b) Variety maintenance of the most prominent landraces preserving their physical, genetic or physiological deterioration. (c) Enhancing on-farm seed production and conservation of prominent landraces through farmer capacity building and the transfer of appropriate technology packages for seed production, variety improvement and maintenance, and seed storage and quality assurance (d) Promotion of seed producer groups to establish a farmer-based system of production of seeds of local landraces. 72 Create more local value added in lucrative value chains: coffee and honey 4.38. As part of the present study, an analysis of five agricultural value chains: coffee, wheat, fish, honey and qat has been carried out with the objective to identify what could be done to improve the functioning of these lucrative value chains so that more rural people could benefit from them. The analysis of the qat value chain was to serve as a reference for the other chains, the qat value-chain being notoriously highly efficient.149 4.39. Though few households are involved in honey-bee raising and coffee production, there are opportunities for improving farmers income from coffee and honey production, as well as bringing additional households into honey-bee raising, a very lucrative economic sub-sector. Coffee production is a smallholder activity and honey production does not require land access, though cash and social networks are required to move honey-beehives to feeding areas according to the different flower seasons. Finally, these sub-sectors bring employment in the rural areas through the various backward and forward linkages they command. Hence investments in these sub-sectors are likely to increase economic opportunities in rural areas and improve the livelihood not only for the producers themselves, but for all the actors in the value chain. 4.40. Yemeni producers respond to market prices and demand when they are aware of the market requirements and when there is a premium for the quality they produce (cf. qat and honey). Producers need to be fully integrated in the chain and quality must be rewarded with clear and transparent incentives or sanctions: the key is transparent market transactions mechanisms and timely market information. Traceability and certification are essential instruments to facilitate the differentiation of products (market segmentation) and transparent market transactions when products are not perishable, allowing rewards (or sanctions) for the quality that consumers demand. 4.41. Credit flows among the various actors in the chain when minimum trust and confidence exist. Volumes of transactions could be increased through access to formal loans. In all sub- sectors, there are significant opportunities to improve chain performance through the delivery of financial services at the various stages of the value chain. Honey production. 4.42. More rural people could enter the lucrative business of honey production, provided adequate support. Though only a very small percentage of rural households are involved in the production and sale of honey (2.4%), for these households, it is a valuable source of income: the annual gross income from honey is 105 000 YR. 150 Honey production and sale is clearly associated with the richest households. In order to increase value to the rural economy by increasing volumes and/or increasing the number of honey producers, a number of constraints need to be overcome, in particular the start up costs, the technical knowledge (raising honey-bees is a very specific skill) and availability of honey feeding trees. Subject to an economic and environmental analysis, it may be feasible to open the sector to new producers through the plantation of Seder trees, as recommended by the researchers of the AREA research station of Sieyoun151. 149 Small Micro Enterprise Promotion Service (SMEPS) and The Royal Tropical Institute (KIT), (2009). 150 By comparison, the average agricultural monetary income for a household whose main activity is farming is YR 410 000 and the average income from qat is 263 000 YR. 151 AREA researchers recommend the plantation of 500 000 Seder trees in the Hadramaouth valley. 73 4.43. Other constraints to productivity improvements reside in the loss of beehives due to disease, which reflect the lack of knowledge regarding the diseases that affect bees, and how to treat them on the part of beekeepers.152 In addition, beekeepers use old technology: in spite of the availability of modern beehives, most beekeepers rely on the traditional beehives, which are much less productive.153 Any program to improve the productivity of honey production would need to include training on beekeeping disease and treatment as well as: (i) the introduction of modern beehives and equipment; (ii) technologies to produce bee wax sheets locally and breeding of queen bees. 4.44. The honey value chain is the second most efficient value chain after qat, but efficiency gains are possible. Honey producers demonstrated their capacity to respond to increasing market prices: since 1990, production increased 16 times, although market price was multiplied by 2.5. However, producers are not as well integrated in the value chain as it is the case for qat. Honey is not perishable and the rapidity of transactions throughout the chain is not an issue. Chain transactions from producers to consumers can take up to one year. Quality characteristics and requirements are well known by all chain actors. 4.45. There is no immediate market sanction if consumers are displeased with the quality of honey bought. Flow of market information along the chain is not as transparent as it is the case for qat. One main problem is that different qualities of honey are being mixed, generating a growing mistrust in Yemeni honey. For the honey value chain, putting in place traceability and quality certification mechanisms would be essential to guaranty the differentiation of products and market transaction transparency all along the chain. There are two domains of recommendations for improving honey value chain efficiency: Branding and traceability. Support chain actors in implementing a quality certification system to guarantee the integrity and traceability of Yemeni honey. Product differentiation. Encourage clearly differentiated lines of products with differentiated markets: high value Seder honey with recognized medicinal value, for nontraditional niche markets; lower value honeys (lavender, black seeds, mix blends, etc.) but with clearly marked varieties indicating the type of floral mix. Coffee production. 4.46. More value­added could accrue to coffee producer with better functioning coffee value-chain. Households involved in coffee production are also few: 2.7%. Coffee is less lucrative than honey. The average gross income for coffee is 52 000 YR. Contrary to the qat and honey producers, coffee producers are poorly integrated in the value chain. All chain actors including producers are aware of the different coffee quality grades, the indicators for each quality, the price differential and the factors affecting product quality. However, market transactions lack transparency with the result that quality-rewarding signals do not reach the producers. While the market rewards freshness, grading and traceability, producers are storing their coffee over long periods of time, often several years and mix quality. There is no clearly defined grading system applied by the chain actors and information flow between producers, processors and exporters is very weak. As a result, production performances have improved, but market performances are decreasing; more coffee is entering the informal market to the detriment of exports. 152 The RALP baseline survey indicates that nearly 55% of the beekeepers experienced a reduction in the number of beehives, , the average beekeeper loosing 60% of his bees. Less than 50% of those that can recognize when their bees are sick said they would know how to treat the disease. Egel, D. and T. Al-Bass Yeslam (2010), p. 13 153 Ibid. p.14 74 4.47. Here again traceability and certification mechanisms are a priority to improve the chain efficiency to the benefit of all stakeholders, of course as long as market information is available for all. A systematic introduction of labeling, traceability and certification would allow for market segmentation which would be perfect for the circumstances of coffee growing in Yemen with its many coffee growing regions with different characteristics (see Box 4). Box 4: Pilot experience of support to coffee value chain in Taiz and Haraz. The introduction of certification, origin branding and quality grading requires that coffee growers be organized. The Small Micro Enterprise Promotion Service (SMEPS) is working with two producer groups in Taiz and in Haraaz to introduce organic coffee production. GEPA a German Fair Trade label has expressed interest in the Taiz group, a women's association. The project purports to achieve the following: Improve organizational capacity (management, procedure, transparency). Build capacity of local service providers (also members of the Association) to deliver high quality extension services to members (in particular for organic production) and monitor practices. Support the participation of Association representatives together with other actors of the chain (such as exporters) in specialty coffee events (Biofach, and Specialty Coffee Association exhibitions). This will support information flow. Arrange visits to more advanced coffee producer associations in Africa Support developing origin labels. This can be modeled along a number of approaches e.g.: (i) Farmer level which will include investing in small scale processing for the groups where green beans are packaged and labeled and shipped by a logistics farmer to the buyer e.g., GEPA (ii) In close cooperation with a processor partner in Sana'a developing markets for traceable beans, and branding according to region alongside their current existing brand. The processor is the intermediary, contracting with group as well as with the international buyer, and ensures that standards are met. As both associations' management capacity improves, they should be able to add more members to the group. As an example in Talook district there are approximately 1000 farmers. Only 200 farmers are members of the association. The wider district level includes many more farmers. The challenge is to build into the pilot a market up-take mechanism that allows for considerable scaling. So a 3 year pilot targeting a few hundred farmers in each region (Taiz, Haraaz, say Yaffee and Raima or Burra) should have a scale-up target of 5000-10000 coffee growers within 6 years. 75 Helping poor families to enroll their children in school, especially girls.154 4.48. School enrollment is highly correlated with the economic status of the households.155 Poor families do not send their children to school or withdraw them from school, in times of economic hardships. The problem is more acute with girls whose education is not as valued as that of boys and who are needed for household chores: fetching wood and water, and tending animals. These tasks, especially the former two, are female tasks. As an illustration, a 2005 survey in rural areas of Hajjah governorate showed that 36% of the families send their girls to school compared to 96% of their boys. In the same communities, women and girls spend four and a half hours daily fetching water and three and half hours collecting wood.156 Though anecdotal, the NGO "International Community Service" reports that in villages of Hajjah where community cisterns had been rehabilitated, girls enrollment increased ten fold. In addition to the opportunity costs of sending girls to school, there are direct costs for school attendance that poor families may not be able to afford for all their children.157 4.49. Higher levels of education of girls will bring new economic opportunities to poor families and reduce fertility rate in the next generation. Given the extremely low levels of education of rural women and the fact that the gap between rural girls and boys is far from closing, there is a need for a special action plan for girls education. In addition to the fact that higher levels of womens education will open up new economic opportunities for poor households, girls and women education should be seen as one of the component of a program aiming at reducing fertility rate. It has been shown in a number of developing country contexts and Yemen is no exception, education can reduce fertility rate thus slowing down demographic growth, which Yemen badly needs to achieve. The average number of children ever born is 5.5 in both urban and rural areas among uneducated mothers. It declines to 4.9 children in rural areas when mothers complete grade 6, and is reduced further to 4.6 children if basic education is completed. Thus, the decline associated with acquiring only basic education is quite significant and affects primarily those girls from relatively under privileged backgrounds.158 4.50. Increasing girls' enrollment is possible with specific actions to overcome social and economic constraints. The World Bank is presently helping the GOY to formulate a holistic approach to the development of the education sector. 159 The on-going country-wide Basic Education Development Project (2004-2010) gives priority to the underserved districts in each governorate and has introduced specific actions to promote girls enrollment. Through this project, the following measures are being implemented: (i) building small schools closer to girls homes; (ii) involving communities in all aspects of project planning, design, and management, especially regarding school location; (iii) obtaining the communitys commitment to enrolling girls as a prerequisite to school construction; (iv) including water, sanitary facilities and boundary walls; 154 The emphasis on quality is highly relevant. Given the GOY aim to reach 100% enrollment by 2015 (one of the MDGs), the pressure is extremely high to expand education access at the expense of quality, while quality is already a key challenge especially in rural areas. 155 For an analysis of the factors affecting childrens enrollment in school, see Annex C of Government of Yemen and World Bank, (2009). 156 Al-Hebshi, Mohamed, The Role of Terraces Management on Land and Water Conservation in Yemen: Case Study in Kuhlan- Affar/Wide Sharis Districts, Third International Conference on Wadi Hydrology, Sanaa, 2005 157 The cost of sending one student to school is estimated at 2,600 YR/year/student. It includes books, school uniforms, school supplies, food and board, and other school-related items. School attendance fees have been suppressed in 2006. Government of Yemen and World Bank (2009). p 40. 158 Government of Yemen and World Bank (2009). p. 96 159 Ibid. 76 (iv) providing separate classrooms for girls in grades 7-9; (v) building separate secondary schools for girls; (vi) increasing the number of qualified female teachers and providing them with special incentives to work in remote rural areas. Unfortunately, the latter has proven difficult to implement. Special incentives in particular are not effective to date.160 4.51. Through these projects (as well as others161), rural girls enrollment rate has increased, and the drop-out rate has decreased but rural girls still fare worse than rural boys on these two indicators. The issue is now to address the economic constraints to girls school enrollment. Therefore, the idea would be to compensate poor families for the economic loss through what is known as conditional cash transfer (CCT) education programs. In addition to compensating for economic loss, the monetary incentives may help overcome also other social and religious prejudices against girls education. The World Bank, which has helped governments successfully introduce CCT programs in a number of countries, has been piloting such a scheme since 2007 in two governorates: Lahj and Hodeidah. Paying poor families for enrolling and maintaining their girls in school would not only raise girls education level, it would also be transferring cash to the poorest HH. An alternative would be to envision such a program in collaboration with the World Food Program thereby the payment be in kind (wheat flour for example) would at the same time improve the food security of the poorest households Helping rural poor to exit agriculture successfully though quality rural162education, with a specific focus on vocational training 4.52. Promoting successful exit of rural poor out of agriculture is an essential component of a rural poverty reduction strategy. With a population growth rate which remains extremely high at 3% per year, the pressure on natural resources, evidenced by the high number of people per ha of arable will become unbearable in the near future. Compounded with the unsustainable groundwater-irrigated agriculture and increased variability of the rainfall that can be expected from climate change, the situation is likely to become worse. Reducing the number of people living off the land and promoting the exit of rural population out of agriculture becomes essential. 4.53. Poor and densely populated rural areas in Yemen have a long history of exporting their labor force to neighboring countries (mainly the Gulf countries Saudi Arabia and Asia to a certain extent). As in the past, large cities will also continue to attract job-seeking migrants from the rural areas. More recently, migrations have become more seasonal and directed to nearby city centers and neighboring countries. 4.54. Skill development of migrant workers is required for them to take advantage of better and more opportunities. The main opportunities of migrant workers are low paid jobs in construction or services. Rural households, which are currently benefiting from remittances from 160 Ibid. p 122 161 Such as the Fast Track Initiative-Catalytic Fund with contributions from 12 donors that closed in 2007; the on-going projects: the Broadening Regional Initiative for Developing Girls Education Project supported by the Japanese International Cooperation Agency; the Dutch support to the Education component of the "Child Development Project and Improving Education Effectiveness" implemented by UNICEF since 2001 in several governorates with a strong institutional capacity component; 3) the recurrent GTZ support to the MOE through a Basic Education Improvement Program from 2002-2010 and 4) CARE NGOs commitment in the education and literacy sector focusing more particularly on rural women from Al-Mahweet, Abyan and Hajjah (Strategic Plan 2005-2008). 162 The emphasis on quality is highly relevant. Given the GOY aim to reach 100% enrollment by 2015 (One of the MDGs), the pressure is extremely high to expand education access at the expense of quality, while quality is already a key challenge especially in rural areas. 77 emigrants abroad, would benefit more if emigrants were able to apply for skilled jobs. Increasing the level of general education and vocational training would benefit the region and the country in the long term. 4.55. Improving the relevance and quality of the programs of the technical schools is key. Currently, graduates from the technical education and vocational training schools do not fare well on the labor market. Quality and relevance of the programs are the main issues. Restructuring the programs, focusing on quality and relevance should take precedence over expansion. Closer linkages with the employers and aligning the sector with signals from the market would help achieving this objective. The Ministry of Technical Education and Vocational Training has developed a strategy for intervention, which is now being implemented through the Second Vocational Training Project. It includes measures to make the programs of the vocational schools more relevant for the domestic and foreign labor market. Successfully implementing these measures remains a challenge. Enhance social safety net programs: Invest in labor intensive rural public works programs with a "cash for work" approach. 4.56. There are three social safety net programs, the Social Welfare Fund, the Social Fund for Development and the Public Works Program that complement each other. The GOY intervenes to alleviate poverty through three key public action programs that complement each other: (i) the Social Welfare Fund (created in 1996) that provides cash assistance to the chronically poor and vulnerable; (ii) the Social Fund for Development (1997) and the Public Works Program (1998). The latter two programs are designed provide support for productive employment and build assets of the poor to reduce poverty in the long-term. Therefore, even if SFD and the PWP reach their objectives, there will still be households that cannot take advantage of additional economic opportunities for a number of reasons and remain trap into extreme poverty, or need temporary support because of some shocks, like disease of a family member, or natural disasters. For these households, there is a need to have a well functioning public cash transfer or income support program, which is the role of the Social Welfare Fund. 4.57. The role of intensive public works program. There is evidence that poor rural families are suffering from less remittances from urban residents and less job availability for migrants due to the slowdown of the Yemen and Gulf countries economy as a result of the financial crisis. Though it would be unrealistic to assume that enough off-farm employment can be generated in every village, labor intensive public works programs can provide additional local jobs with payment in cash (or food) for the poorest families while at the same time investing in much needed basic infrastructure for the most-underserved communities. Such programs can therefore be part of and complement an approach that focuses on poorest districts for fiscal transfers. In addition, this type of programs, once established can operate as productive safety net program that can be scaled up to respond to shocks of various kinds, such as crop failures or other natural disasters. 78 4.58. Worth expanding: the Labor-Intensive Work Program. A good example of such an approach is the Labor-Intensive Work program, piloted with success by the Social Fund for Development under the Global Food Crisis Response Program. The demonstrated that not only did it respond to emergency needs through quick disbursement, it also contributed to a long-term future economic development. It has also encouraged womens participation and designed some work for them, consistent with their physical ability and culture of the community (see Box 5).163 Given the very low participation of rural women in the paid labor force, such programs present considerable added value for the rural households. In addition, some types of the works (e.g., road paving) have an opportunity to be combined with the development of skills for future job opportunities. 163 This emergency program has been financed by the World Bank grant from Global Food Price Increase Response (GFRP) facility. 79 Box 5: Worth up-scaling: The Labor-Intensive Work Program of SFD. The program paid US$ 6.1 million wages to 36 000 people, benefitting 16,000 households within the most seriously affected communities to help mitigate the impact of increased food prices through temporary work opportunities while providing needed basic infrastructure to the affected communities. The districts for the program coverage are selected based on the poverty indicators. Communities within the districts are selected based on a set of criteria: (a) a clear evidence that at least 50% of households are affected by the food price increase; and (b) at least 50 households are ready to participate. Since the wages are set 10-20% below the market rate, this program is targeting the poorest through self-selection mechanism. Following office screening of communities/villages, consultation was conducted with the local authorities to confirm that these selected areas are eligible and to identify any new areas eligible but left out. Further, participatory rural rapid assessment method was used to identify the target communities and households, identify appropriate areas of intervention that meet the programs goal and collect the information to design the sub-projects. The communities decide the type of sub-projects by consensus. At least 50% of the cost of the project must be labor costs and the communities must be able to implement it by themselves. A review of 5 projects in 4 different governorates, conducted in May 2009 as part of supervision of the project revealed that the program implemented by 8 SFD Branch Offices is well managed by competent teams. Beneficiaries confirmed that their wages have been paid weekly on a regular basis and the money is well spent to fulfill immediate needs. The program demonstrates that not only did it respond to emergency needs through quick disbursement, it also contributed to a long-term future economic development. Terrace rehabilitation is a good example. It is a labor- intensive and requires manual laborers using local materials. As the terraces are being rehabilitated, it protects communitys assets from land erosion and flood while the terraced land is used for agricultural activities. The review team also noted that most of the communities visited were suffering from a severe drought in addition to food and financial crisis. The already evident success of the program has led to a GOY request for expansion under EU financing (Euro 17.5 million) through the EU Food Price Crisis Rapid Response Facility under the Global Food Crisis Response Program. Source: Aide memoire from a joint progress review mission of the SFD, June 2009 and Project Appraisal Document for the SFD Phase IV project (February 2010). 4.59. While the program was developed as part of the response to the 2008 food crisis, its evident success has led to a request for its expansion, building upon lessons learnt from the pilot as well as from other programs of this kind in the world. One lesson is that this kind of programs are highly demanding in terms of management resources at the field level, and that significant monitoring and accountability mechanisms and rigorous evaluations are required to ensure 80 effective and equitable resource use.164 SFD also learned that they need to strengthen targeting and selection criteria and ensure better communication with the communities. In order to have a lasting impact in building the capacity of poor households, multi-annual interventions, rather than just one year as under the Food Crisis Response Program are required. The Fourth SFD project recently approved165 intends to reach 300 000 in 45-50 poorest districts. While basic infrastructure will continue to be implemented using this approach, there will be greater emphasis on rehabilitation of agricultural lands and terraces for the benefit of the poorer households, as well as on activities that maximize female participation. To the extent possible, the new LIW will be based on integrated local analysis and planning, to ensure synergies between different interventions within the community. Enhance social safety net programs: improve the targeting of the Social Welfare Fund 4.60. The SWF fares rather poorly in its targeting. 166 The Program expanded quite substantially, from 100 000 beneficiaries at its start to over 1 million households over a ten year period, reaching 14 percent of the extremely poor (urban and rural) and 13 percent of the poor in 2005. The expansion has come at a cost ­the leakage to the non-poor increased. Non-poor beneficiaries have been absorbing 47 percent of all benefit payments.167 We looked at the situation for the rural areas only which shows similar bias: the two poorest income quintiles (household income) get about 48% of the SWF transfers to rural households, while the three richest quintiles are getting 52% (Figure 20). Figure 20: Cash transfers of the Social Welfare Fund according to HH income (in% of total rural transfers) 164 World Development Report, 2008. p 238. Yemen could learn from one of the largest scheme of this kind, the National Rural Employment Scheme, launched by the Government of India, to invest in rural infrastructure and restore degraded natural resources in order to raise farm and nonfarm productivity. 165 The project was approved in February 2010. 166 World Bank (2007) 167 World Bank (2007). p 60 81 4.61. The SWF urban bias may be less pronounced than initially thought. The SWF has also been suspected to be urban-biased. Actually our analysis on the basis of the HBS survey data indicates that the bias may be less pronounced than initially suspected when looking at several indicators: (i) the rural areas are receiving on average 2,764 YR per rural dweller, while urban areas are receiving 1,767 YR per urban dweller. When the amount is calculated per family receiving cash transfers, urban households receive slightly more (21,351 YR) than rural households (20,687 YR), but the difference is not significant. The percentage of poor people in urban areas is 17.3% vs. 82.7% in rural areas. If the SWF were to evenly distribute its resources based on HH's poverty status, the amount of SWF resources going to rural vs. urban areas should be 17.3% vs. 82.7%. The actual percentages are 20.8% vs. 79.2%. There is therefore only a slight bias, not significant. 4.62. The basis to reform the Social welfare Fund is there. As a result of these findings, the Government of Yemen (GoY) has authorized a program of fundamental reforms, enabled by the 2008 Law on Social Welfare, to reorient SWF objectives so that: (i) poverty is the primary focus of assistance and is more clearly defined; (ii) beneficiaries include those in economic difficulty (unemployed, etc.) as well as the poor covered by previous social categories (female headed households, widows, elderly, disabled, etc.); (iii) a case management system will be established to reassess beneficiaries eligibility and recertify them within a defined period of time for either re-enrollment in, or graduation from, SWF benefits; and (iv) the SWFs beneficiary development role to assist the transitory poor (i.e. those around the poverty line) to exit poverty through skills training and linkages with the labor market is legally mandated. 168 4.63. ....And reforms have already started. Reducing inequalities through cash transfer programs can be done for the lowest cost possible when transfers are targeted properly to the poor and ultra poor. Beginning in 2008, the SWF embarked on a reform of its targeting. It completed an extensive national survey to further identify the poor and vulnerable. It then applied improved targeting methods to the data collected in order to expand the program to cover those in need and to improve its current beneficiary/applicants inclusion/exclusion error rate and deliver cash transfers only to the poor and vulnerable. 169 In addition, in response to the food crisis, in 2009 the government doubled the maximum SWF benefits to YR 4,000 (US$ 20) per case per month. Furthermore, the Government has announced a plan to increase the coverage of beneficiaries to 1.5 million households, permitting coverage of nearly all those below the poverty line. 170 4.64. It is expected that beneficiary application for SWF support and assessments of applicants eligibility will be a continuous process. The SWF newly established database is the most comprehensive national record of poor and vulnerable individuals available in Yemen. Such a national database can be used to target and coordinate other funds and benefits across a range of social programs. 168 World Bank (2010b). 169 Proxy Means Tests (PMT) were applied to the beneficiary and applicant survey. PMT is a targeting method based on a score that is attributed to households on the basis of observable characteristics and the estimated relationship of these characteristics with consumption and poverty. 170 SWF 2008 Survey: over 1.6 million households were covered in this survey, 1 million of which are current SWF beneficiary HHs, and 0.6 million new applicants. 82 Synthesis 4.65. Table 29 below summarizes the issues, the objectives and the priority programs together with the responsible organizations. 83 Table 29: Objectives and Priority Programs to Address the Needs of the Rural Poor Issues Objectives and Actions Responsible organization Spatial inequalities Increase public investments in the rural districts with highest - Ministry of Local poverty incidence and lowest Access index, Administration - Spatial differences within rural areas are considerable, with rural poverty being concentrated in some -through: (i) budget transfers (Decentralization Law); (ii) - Social Fund for districts/governorates, both in terms of poverty rate and investments though SFD and PWP with spatial targeting. Development mass. - 68 districts qualify for increased per capita investments. - Public Works Program - Access to basic infrastructure and services, as measured Improve the capacity of district councils to manage public by the District Access index vary considerably from one funds efficiently and address the needs of the poor and district to the next. vulnerable groups. - Households in richer districts have better access to basic - Worth up scaling: the "Empowerment for Local Development" services. Program of the SFD. Access to irrigation is inequitable and could become worse Implement the Water Sector Reforms together with pro-poor - Ministry of Agriculture and with the effect of irrigation technology water access programs and enhance groundwater local Irrigation governance - Farmers with financial resources and larger - Ministry of Water and landholdings, dig deeper wells. The rapid drawing of - Improve the capacity of local institutions to implement the Environment groundwater has dried springs and wells, forcing groundwater law, through local collective action (community - National Water Resource adjacent or downstream smaller farmers who cannot water management and water user associations) as part of the Authority afford deep well-drilling to abandon their fields. implementation of the National Water Strategy and Investment Plan. - There is a lack of effective local institutions for groundwater governance. - Implement the water sector reform as a package, together with pro-poor water access programs and pro-poor entry criteria in - Water sector reforms have been enacted since 2003 but order to enhance equity. implementation is hindered by political economy constraints. 84 Land access is skewed and could become worse with Protect poor people access to waqf and communal land through - Ministry of religious growing competition for land enacting rural land regulatory framework Endowment and Spiritual Guidance - Poor people depend on communal and waqf land for - Develop an inventory of waqf land and develop regulations to access to land. For example, communal land and land in improve the management of waqf land - Ministry of Interior upper catchment areas are important resources for - Develop an inventory of State land and mobilize this land to - Ministry of Social Affairs grazing animals for people with limited access to land. improve land access to land-poor farmers. There is uncertainty as to the legal status of such land, and cases of private appropriation are occurring. - Document customary rights in order to protect communal properties from private appropriation and clarify rights of the - Waqf land administration is affected by weak supervision communities on this type of land. and lack of inventory, thus facilitating private appropriation of waqf land. - Enforcement of customary rights on these type of lands, in particular communal lands is weakened, with the result that land-poor farmers could lose access. Under low and erratic patterns of rainfall, farmers that Secure crop production and increase resilience to climate - Ministry of Agriculture and do not have access to irrigation are highly dependent on change through water harvesting and soil conservation Irrigation soil and water harvesting systems programs in rain fed areas with low and erratic patterns of - Social Fund for rainfall, using a watershed management approach, including: (i) - Poor rural families are highly dependent on rainfall Development rehabilitation of terraces; (ii) reforestation of the watershed upper availability for their crop or livestock production, either catchment areas; (iii) construction or rehabilitation of cisterns for because they live in areas with poor agro ecological human and animal consumption (could be linked to the conditions, or unfeasibility of irrigation or because they introduction of a drip irrigation system for supplemental do not have the resources to invest in irrigation. irrigation of high value crops); (iv) wadi bank protection works; - Rainfall is limited and highly erratic in many areas of and, (v) the improvement of small spate irrigation schemes. Yemen, and is likely to come with flash floods. Rainfall Worth upscaling: the Rainfed Agriculture and Livestock project ; variability is likely to increase with climate change. the Groundwater and Soil Conservation Project. - Yemeni people survived through centuries thanks to their careful harvesting and conservation of water (and soil) through terraces and management of run-off water (wadi, spate irrigation). However this system is in jeopardy because of migration of adult men (better opportunities for labor and low land productivity) and the costs it represents to maintain these infrastructures. 85 Livestock is poor people's best asset, an asset that they can Protect poor people's assets through animal health and - Ministry of Agriculture and easily lose and that is not as productive as it could be. livestock nutrition programs: Irrigation/General Directorate of Animal - Animals are the best assets of poor people, providing - upgrade the public animal health services, progressively Resources cash income and savings. privatize parts of the services, and extend the present coverage limited to peri-urban areas, through a system of veterinary - Animal health and nutrition are poor. auxiliaries or community animal health workers (ref: the - The availability of veterinary services is low in rural Rainfed Agriculture and Livestock Project, consider up- areas. scaling). - Forage and fodder availability is limited (rainfall and - improve animal nutrition through: (i) better management of the water availability constraints). resource constraints (water, forage and fodder); (ii) fodder productivity improvements and affordable diet supplementation (such as bone meal); (iii) improved management of the upper catchment of the watersheds, introducing animal edible species (napier grass, legume trees) as well as management of other communal rangeland. Cereal and legume productivity in rainfed areas is very Enhance the productivity and climate resilience of rain fed - Ministry of Agriculture and low, well below technical potential and below farmers' cereal and legume production through establishing a farmer- Irrigation/Generals seed yields in comparable environments based seed production system based on local landraces: Multiplication and Conservation Center - Farmers in highy risky environments cannot afford to - Indigenous variety improvement and maintenance. invest in improved technologies. - Social Fund for - On-farm seed production and conservation. Development - One of the reason for low productivity is the deterioration - Establishment of seed producer groups. of the quality of the seeds being used. - Worth up-scaling: (i) the Rainfed Agriculture and Livestock - The availability of veterinary services is low in rural Project; (ii) the Agro-Biodiversity and Climate Adaptation areas. Project. - Better seed management of local varieties is a low cost/low risk technology that can be easily adopted and that could help farmers withstand the shocks of climate change. 86 More rural families could be involved in honey production Create more local value added in lucrative value chains: coffee - Ministry of Agriculture Coffee value chain is not effective and honey through: Though relatively few rural households are involved in coffee and honey production (about 2.5 % of the rural HH in - Addressing production constraints each of the value chains), there is potential for improving - Increasing capacity of producer organization; producers income in these two sub sectors and increasing rural employment in the value chains. - Promoting transparent market transactions and timely information - improving productivity (equipment for producers, technical advice); - Introducing and enforce branding and traceability; - Introducing product differentiation; Too many children, especially girls are still not enrolled or Enhance school attendance especially that of girls : School - Ministry of Education drop out of school. planning and design addressing parents' concern regarding - Social Fund for girls' attendance: - Poor families cannot afford to send their children to Development school as they need them to help with domestic tasks, - building smaller schools closer to girls homes; especially girls for household chores, or boys working - involving communities in all aspects of project planning, design, and on the farm. management, especially regarding school location; - The gap between boys and girls education level is not - obtaining the communitys commitment to enrolling girls as a prerequisite to school construction; closing though girls enrollment has improved. There are many social issues regarding girls school attendance. - including water, sanitary facilities and boundary walls; - Education is the only way out of poverty for poor - providing separate classrooms for girls in grades 7-9; peoples children. - building separate secondary schools for girls; - Higher level of education of women will contribute to - increasing the number of qualified female teachers and providing them lower fertility rates, and decrease demographic pressure with special incentives to work in remote rural areas. on limited natural resources. - Conditional cash transfer to offset opportunity costs of sending girls to school and reduce girls' drop out 87 Opportunities for employment are very limited in rural Invest in quality education, with a specific focus on vocational - Ministry of Technical areas. Rural people lack skills training, Education and Vocational Training - Rural households lacking assets depend on salaried labor - targeting quality and relevance, through closer linkages with for their livelihood. employers (to develop programs, etc..) - Social Fund for Development - Opportunities for employment in rural areas are very - aligning the sector with signals from the market and taking limited. People have to migrate abroad or to urban areas. into account not only the needs of the Yemeni labor market, but also the labor markets needs in Gulf countries and Saudi - Men migrating are typically employed in low paid Arabia. unskilled jobs (construction and services); Invest in labor intensive rural public works programs with a - Yemeni enterprises complain that they cannot find food for work or cash for work approach. Particular reference qualified workers. is the Labor Intensive Work Program of the Social Fund for Development. Extreme poor and vulnerable households will continue to Develop better targeted social welfare programs adapted to - Social Welfare Fund depend on cash transfers in order to survive. The problem rural specificities. Review the list of beneficiaries of the cash is that the Social welfare Fund is not well targeted transfer program and improve the management of the Fund. 88 REFERENCES Al-Hebshi, M. (2005). The Role of Terraces Management on Land and Water Conservation in Yemen: Case Study in Kuhlan-Affar/Wide Sharis Districts, Third International Conference on Wadi Hydrology, Sanaa. Bamatraf A., A. Aw-Hassan, and M. Alsabani. (2000). Impact of Land Tenure and Socioeconomic Factors on Mountain Terrace Maintenance in Yemen, CAPRI Working Paper. Breisinger, C., M.H. Collion, X. Diao, and P. Rondot. (2010). Impacts of the triple global crisis on growth and poverty in Yemen. IFPRI Discussion Paper # 955. Burns, B. and T. Taher (2009). Yemen Water User Association Study. Mimeo Central Statistical Office. (2004). Population Census. Ministry of Planning, Yemen: Egel, D. and T. Al-Bass Yeslam (2010). Conditions in Rural Yemen: Findings from the RALP Baseline Survey. Mimeo Eger, H. (1984). "Rainwater Harvesting in the Yemeni Highlands: the effect of rainwater harvesting on soil moisture status and its implications for arable farming, a case study of the `Amrn region" In: Kopp H. and G. Schweizer, (eds.) Entwicklungsprozesse in der Arabischen Republik Jemen, Wiesbaden, Jemen Studien 1: 147-169. European Training Foundation and World Bank (2004). Technical Education and Vocational Training in Yemen and Its Relevance to the Labour Market. FAO (2002) Economic Incentives and Comparative Advantage for Strategic Crops in the Highlands. A Policy Note on Yemen. FAO (2009). FAO/WFP Crop and Food Security Assessment Mission to Yemen. Special report. Mimeo. Ministry of Water and Environment and Ministry of Agriculture. (2007). A Poverty and Social Impact Analysis (PSIA). Yemen Water Sector Reform Program. Pelat, F. (2009). Yemen Rural Development Programs and Projects: A review. Mimeo. Rappold G., Ergenzinger P., Gerke H., and Scholz F. (2003). Hydrological analysis of terraced catchments: a case study of the Taizz region, Yemen, in Indigenous Knowledge and Sustainable Agriculture in Yemen, Les Cahiers du CEFAS #3, Sanaa Republic of Yemen and World Bank. (2005). Country Social Analysis. World Bank Report No. 34008-YE. Republic of Yemen and World Bank. (2004). Rural/Local Development Strategy: Implementing the Poverty Reduction Strategy in Rural Areas. World Bank Report No. 25905-YE. 89 Republic of Yemen and Word Bank. (2007). An Integrated Approach to Social Sectors : Towards a Social Protection Strategy. Republic of Yemen and World Bank. (2009). Yemen Education Country Status Report. Republic of Yemen, (1999). Ministry of Agriculture and Irrigation, General Directorate of Animal Resources, Yemen: Sector Note on Animal Resource. Republic of Yemen, Ministry of Technical Education and Vocational Training. (2004). Technical Education and Vocational Training: Strategic Development Plan. Small Micro Enterprise Promotion Service (SMEPS) and The Royal Tropical Institute (KIT). (2009). An Analysis of Five Agricultural Value Chains in Yemen: Fish, Wheat, Coffee, Qat and Honey. Taylor-Awny, Adam. (2002). Community management of Local Natural Resources: Al-Mahweet Governorate, External Evaluation of CARE Project. Varisco, D.M. (1991). The future of terrace farming in Yemen: a development dilemma, in Agriculture and Human Values #8. Wilkinson. (1999). Settlement, soil erosion and terraced agriculture in highland Yemen: a preliminary statement, Proceedings of the Seminar for Arabian Studies #28, London. World Bank. (2007). Yemen Poverty Assessment Report. World Bank. (2009). Land Tenure for Social Economic Inclusion in Yemen: Issues and Opportunities. Mimeo. World Bank. (2010a). Assessing the impact of climate change and variability on the water and agriculture sectors, and the policy implications. World Bank (2010b). Yemen. Social Welfare Fund Institutional Support Project. Project Appraisal Document. World Development Report. (2008). Agriculture for Development. World Development Report. (2009). Reshaping Economic Geography. 90 Annex 1: Agro climatic data and rainfall map Map A 1: Annual average rainfall Source: WorldClim. 1950-2000 average 91 Table A. 1 Characteristics of agro-climatic zones and suitability for crop cultivation Soil Rainfall PET Growing moisture mm/year mm/day Period- days P/PET Soil temp regions* Regims Suitability of zones for crop cultivation Grouping Typic tropo Suitable for one long cycle species or two short 700 -1200 2.8 Winter, >240 >1 isohyperthermic Ustic cycles crops Black Suitable for one long cycle species or two short 3.7 Summer cycles crops Black Typic 3.1 - 3.5 1st GP = 0.6 tropo 600 -850 winter 170 - 190 - 0.75 isothermic Ustic More risky situation for the first crop Black 4.5 - 4.8 2nd GP = 0.8 More reliable for growing annual crops or two summer -1.0 shorter growing crops Black Typic 3.1 - 3.5 1st GP = 0.6 tropo 700 - 850 winter 170 - 190 - 0.75 isothermic Ustic More risky situation for the first crop Black 4.7 - 4.9 2nd GP = 0.8 More reliable for growing annual crops or two summer -1.0 shorter growing crops Black Typic 2.9 - 3.2 1st GP = 0.6 tropo 600 - 700 winter 170 - 190 - 0.75 isothermic Ustic More risky situation for the first crop Black 3.5 - 3.9 2nd GP = 0.8 More reliable for growing annual crops or two summer -1.0 shorter growing crops Black 3.2 - 4.0 1st GP = 90 - 400 - 700 winter 100 0.5 - 0.6 isohyperthermic - aridic can grow annual crops with short cycles Red 4.5 - 5.0 2nd GP = 50 - summer 60 hyperthermic supplemental irrigation Red 2.7 - 3.3 weak 400 - 700 winter 1st GP = 60 - 80 0.35 aridic one crop of short cycle in winter Red 4.5 - 4.8 one short cycle crop in summer (C4 plants like summer 2nd GP = 50 sorghum, Tef,...) or irrigated crops Maize or melon Red 4.5 - 5.3 cropping possible for short cycle summer and 300 - 600 winter 1st GP = 80 0.5 - 0.6 isohyperthermic - aridic winter crops Red 5.0 - 5.6 2nd GP = 70 hyperthermic cropping possible for short cycle summer and Red 92 Soil Rainfall PET Growing moisture mm/year mm/day Period- days P/PET Soil temp regions* Regims Suitability of zones for crop cultivation Grouping summer winter crops 3.0 - 3.5 Can grow many crops but under irrigation because 200 -450 winter 1st GP = 30 0.35 isohyperthermic - not much rain Blue 5.0 Blue summer 2nd GP = 70 hyperthermic thermic can grow summer short cycle crops 200 - 500 4.0 winter 1st GP = 20 0.4 - 0.6 hyperthermic aridic risky cropping without irrigation Blue 6.0 - 6.5 Blue summer 2nd GP = 70 Summer crop possible 3.0 - 4.0 weak Blue 200 - 400 winter 1st GP = 60 0.4 thermic/isothermic aridic can grow short cycle winter crop 5.0 Blue summer 2nd GP = 40 risky cropping without irrigation 200 - 450 4.0 winter 1st GP = 30 risky cropping Green 5.0 - 5.5 Green summer 2nd GP = 50 can grow short cycle summer crops 200 - 400 6.0 1st GP = 30 0.25 aridic risky cropping without irrigation Green 2nd GP = 40 risky cropping without irrigation Green 3.0 -3.5 Green 100 - 200 winter 40 0.3 - 0.45 risky cropping without irrigation 5.0 - 6.0 Green summer risky cropping without irrigation 3.0 -3.5 Green 100 - 200 winter 30 0.3 hyperthermic aridic risky cropping without irrigation 5.0 - 6.0 Green summer risky cropping without irrigation 3.0 -3.5 Green 100 - 200 winter 20 0.25 - 0.50 risky cropping without irrigation 5.0 - 6.0 Green summer risky cropping without irrigation 100 - 250 4.0 winter 1st GP = 20 0.4 hyperthermic aridic risky cropping or no cropping Green 6.0 - 6.5 Green summer 2nd GP = 50 cropping summer crops with irrigation 175 - 400 3.4 winter 1st GP = 20 - 40 0.4 hyperthermic aridic risky cropping Green 93 Soil Rainfall PET Growing moisture mm/year mm/day Period- days P/PET Soil temp regions* Regims Suitability of zones for crop cultivation Grouping 4.5 - 6.0 Cropping is possible with short cycle crops but need Green summer 2nd GP = 90 supplemental irrigation 2.5 - 3.0 cropping is possible for winter crops but need Green 175 - 325 winter 2nd GP = 90 0.5 thermic aridic supplemental irrigation 4.5 Cropping is possible with short cycle crops but need Green summer 2nd GP = 90 supplemental irrigation 100 - 250 2nd GP = 90 0.3 thermic/hyperthermic risky cropping Green 3 - 3.5 No 50 -125 winter No GP hyperthermic aridic No cropping Colour 6 - 6.5 No summer No cropping Colour No 10 - 200 3 - 4 winter No GP hyperthermic aridic No cropping Colour 4.5 - 5 No summer No cropping Colour 3.5 - 4 extreme No 0 - 100 winter No GP hyperthermic aridic No cropping Colour 7-8 No summer No cropping Colour No 50 - 200 3.5 winter No GP hyperthermic aridic No cropping Colour 6.5 - 7.5 No summer No cropping Colour 94 Annex 2: List of Districts cumulating high poverty incidence and low access index 171 Governorate District District name Rural poverty District Access Index incidence Ibb 1101 Al Qafr 57.9% 32.1 1110 Far Al Udayn 65.9% 28.6 1111 Al Udayn 52.7% 37.1 Abyan 1201 Al Mahfad 95.0% 39.2 1207 Sarar 54.8% 38.7 Al-Baida 1402 Nati' 83.5% 17.6 1411 As Sawadiyah 68.6% 28.1 1412 Radman Al Awad 100.0% 36.8 1420 Al Malagim 63.2% 33.3 Taiz 1503 Shara'b As Salam 51.0% 39.5 1512 Al Misrakh 63.2% 39.5 1523 Sama 64.3% 38.2 Al-jawf 1601 Khabb wa ash Sha'af 59.0% 25.8 1603 Al Matammah 69.6% 25.9 1607 Al Maslub 57.1% 6.9 1611 Rajuzah 55.9% 23.8 1612 Kharab Al Marashi 70.2% 20.1 Hajjah 1705 Hayran 69.9% 19.3 1706 Mustaba 48.0% 7.8 1707 Kushar 80.0% 24.2 171 The District Unmet Basic Needs, calculated by the Social Fund for Development, aggregates individual literacy (for people 10 years and older); enrollment rate (for children age 6 to 15); household access to safe drinking water, sanitation, electricity; and use of gas as cooking fuel (rather than wood and charcoal) on a scale from 0 to 100, with 100 being "all needs satisfied".. 95 Governorate District District name Rural poverty Index of unmet basic incidence needs 1708 Al Jamimah 44.0% 10.9 1710 Aflah Ash Shawm 59.2% 31.7 1711 Khayran Al Muharraq 55.5% 18.2 1713 Qafl Shamer 88.2% 26.0 1714 Aflah Al Yaman 76.3% 31.3 1717 Al Maghrabah 55.0% 16.1 1722 Ku'aydinah 63.0% 23.5 1724 Bani Qa'is 53.5% 12.5 1727 Bani Al Awam 50.7% 22.9 1730 Washhah 63.3% 15.2 1731 Qarah 55.6% 8.3 Al-hodeidha 1807 Az Zaydiyah 67.5% 27.4 1809 Ad Dahi 67.3% 22.3 1810 Bajil 57.6% 17.7 1812 Bura 93.9% 25.0 1819 Hays 66.3% 21.1 1825 Al Garrahi 43.6% 26.9 1826 Al Tuhayat 44.2% 25.2 Shabwah 2108 Merkhah Al Ulya 87.4% 29.2 Sa'adah 2214 Kitaf wa Al Boqe'e 63.3% 35.0 Sana'a region 2308 Al Haymah Ad 53.4% 31.8 Dakhiliyah 2309 Al Haymah Al 82.3% 23.2 Kharijiyah 2311 Sa'fan 65.5% 28.2 96 Governorate District District name Rural poverty Index of unmet basic incidence needs Lahej 2509 Al Musaymir 50.0% 29.3 2513 Al Madaribah Wa Al 52.6% 31.0 Arah Mareb 2604 Harib Al Qaramish 100.0% 23.7 2605 Bidbadah 96.7% 18.8 2608 Rahabah 72.8% 34.5 2611 Al Abdiyah 71.0% 30.5 Al-Mahweet 2705 Milhan 57.2% 16.0 2707 Bani Sa'd 49.0% 19.1 Amran 2901 Harf Sufyan 83.3% 23.3 2902 Huth 56.3% 34.8 2904 Al Qaflah 68.9% 15.6 2905 Shaharah 75.7% 29.8 2907 Suwayr 91.9% 12.6 2908 Habur Zulaymah 94.6% 21.0 2914 As Sawd 88.2% 24.8 2916 Maswar 95.6% 30.6 2920 Bani Suraim 66.7% 37.9 Al-dhaleh 3009 Al Husha 57.9% 31.6 Raymah 3101 Bilad At Ta'am 45.9% 21.1 3103 As Salafiyah 44.8% 28.3 3106 Mazhar 41.1% 21.9 97