Report No. 39881-RW Rwanda Promoting Pro-Poor Agricultural Growth in Rwanda: Challenges and Opportunities June 1, 2007 Sustainable Development, AFTS3 Country Department C2 Africa Region Document of the World Bank Contents 1. INTRODUCTION ................................................................................................................. 1 Background............................................................................................................................................... 1 Objectives ................................................................................................................................................. 1 2.Organization of the report., ....................................................................................................................... 1 ................................................... Macroeconomic overview......................................................................................................................... AGRICULTURE INTHE ECONOMY OFRWANDA 2 2 Extent and distribution ofpoverty ............................................................................................................ 2 Rwanda's development strategy ............................................................................................................... 5 Role of agriculture inthe economy ........................................................................................................... 5 Naturalresourcebase ............................................................................................................................... -8 Land ...................................................................................................................................................... 9 Water .................................................................................................................................................. 12 Soils .................................................................................................................................................... 14 Farming systems and cropping patterns.................................................................................................. 14 Agncultural production trends ............................................................................................................... -18 Livestock ............................................................................................................................................ Crops .................................................................................................................................................. 18 -21 Agricultural technology .......................................................................................................................... 22 Agricultural research ......................................................................................................................... 23 Agricultural extension ........................................................................................................................ 23 Improvedproduction inputs............................................................................................................... -23 Agricultural marketing systems ............................................................................................................. Ruralfinancial services...................................................................................................................... 32 -32 Food staples........................................................................................................................................ 32 Export crops ....................................................................................................................................... 33 Policies affecting agriculture .................................................................................................................. 34 Agricultural policies ........................................................................................................................... 34 Tradepolicies 36 3. ..................................................................................................................................... AGRICULTURAL GROWTHDRIVERS ....................................................................... 37 Growth drivers:Demandprospects......................................................................................................... Domestic markets ............................................................................................................................... 37 Regional markets ................................................................................................................................ 37 39 International markets ......................................................................................................................... 40 Summary ............................................................................................................................................ -42 Growth Drivers: Supply prospects.......................................................................................................... 42 Potentialfuture sources of growth .................................................................................................... -43 51 4.Competitiveness ofRwandan commodities ............................................................................................ PROSPECTSFOR GROWTHAND POVERTY REDUCTION .................................. 55 Base Year: Initialconditions ................................................................................................................... Land holdings andpoverty distribution across rural groups ............................................................. 55 56 Income distribution and consumption patterns across rural groups.................................................. 56 Simulationresults.................................................................................................................................... 61 Base Run Scenario.............................................................................................................................. 61 Growth simulations............................................................................................................................ -67 Summary of simulations 78 5. ...................................................................................................................... ................................................................................. Principal constraints to agricultural growth inRwanda.......................................................................... SUMMARY AND CONCLUSIONS 81 81 Getting agriculture going inRwanda: An agenda for action .................................................................. 81 111 ... 82 Deliveringon Growth........................................................................................................................... Immediatepriorities: Agricultural growthpillars.............................................................................. 106 Strengthening the institutions andpolicies that support agriculture ............................................... 106 110 6 . Cross cuttingpolicy issues ............................................................................................................... REFERENCES .................................................................................................................. 111 ANNEX 1 AGRICULTURAL STATISTICS INRWANDA 112 ANNEX 2 ..DESCRIPTIONOF THE REMMMODEL ....................................................... ............................................... 113 iv Tables Table 1. Selectedeconomic performance indicators. Rwanda. 1996-2008 ..................................... Table 2.Agricultural sector growth performance. Rwanda. 2001-05 ............................................. 3 7 Table 4.Distribution o f landholdings among rural households. Rwanda. 2006........................... Table 3.Land endowment indicators. Rwanda and selected countries. 2001-03 ............................ 9 Table 5. Crop production by household groups. Rwanda. 2006.................................................... 11 Table 6.Area, yield. and production o fprincipal crops. Rwanda. 2001-05 .................................. 11 Table 8.Food crop production trends. Rwanda. 2001-05.............................................................. Table 7.Average food crop yields. Rwanda and selected countries, 2003-05 (MTha) ................16 17 Table 9.Evolution o f livestockproduction. Rwanda. 1994-2005 ................................................. 19 Table 10.Use o f improved inputs by province and income quintile. Rwanda.............................. 22 Table 11.Estimated soil nutrient losses inA h c a n countries. 2002-04 (kgha) ............................ 24 Table 12.Fertilizer use. Rwanda. 2005 Season A (numbers o f households) ................................. 24 Table 13. Summary o f experimental data on fertilizer response, selected crops........................... 26 Table 14.Fertilizer imports, Rwanda. 2001-05 ............................................................................. 26 Table 15.Fertilizer use intensity, selected Africa countries .......................................................... 27 Table 16.Production o f improved seed (kg) and demand coverage (%), 200-2005 ..................... 28 Table 17.Pesticide use, Rwanda, 2005 Season A (numbers o f households) ................................. 30 Table 18.PSTA programs and sub-programs ................................................................................ 31 Table 19.Composition o f rural and urban diets, Rwanda. 2000-01 .............................................. 36 38 Table 20 . Structure Table 21. Estimatedyield gap for principal food crops. Rwanda................................................. ofexpenditures, rural vs.urbanhouseholds. 2000-01.................................... 38 -45 Table 23.Producer prices and export prices o f coffee. 2003-05 ................................................... Table 22. Estimated yield gap for selected fruits and vegetables. Rwanda................................... 46 Table 24.Highest prices, Mombasa auction, Jan-Oct 2005 (US centskg) ................................... 48 Table 25.Trends inexports o f Rwanda's crude pyrethrum extract. 1003-2005............................ 49 Table 26.Import and export parity prices. food staples. Rwanda. 2005 ....................................... 50 Table 27.Rwandan exports o f fruits. vegetables. and flowers. 2000-03 ....................................... 52 54 Table 28.Annual commodity expenditures per household, Rwanda, 2003 ($) ............................. Table 29.Annual commodity expenditures by income quintile, 2003 (000 $) ............................. 57 Table 30.Income growth and poverty reduction. REMMsimulationresults................................ 57 Table 31. Projected imports o f agricultural commodities, 2015.................................................... 62 Table 33. Average annual growth inagricultural GDP. African countries. 1996-2005 ................66 Table 32. Projected agricultural export growth and agricultural trade balance ............................. 65 69 Table 35. Poverty-growth elasticities inthe model scenarios........................................................ Table 34. Sources o f income growth andpoverty reduction inthe model .................................... 71 79 reduction........................................................................................................................................ Table 36. Priority action areas and entry points to unlock agricultural growth andpoverty Table 37.Priority action areas and potential entry points to strengthen institutionsthat support 98 agriculture .................................................................................................................................... 108 V Figures Figure 1.Unemployment rates inRwanda. by source o f employment. 2001.................................. 4 Figure 2.Populationdensity and poverty rates. Rwanda. 2005....................................................... 4 Figure 3. Sectoral composition o f GDP. Rwanda. 2005 .................................................................. Figure 4.Distribution o f landholdings by household size. Rwanda. 2001 .................................... 8 Figure 5. Agricultural value added. Rwanda vs. selected African countries. 2003 ....................... 10 Figure 6.Trends inlandproductivity. Rwanda ............................................................................. 12 13 Figure 7.Variability inagricultural GDP. Rwanda. 1966-2004.................................................... Figure 8.Area shares o fprincipal crops. Rwanda. 2005 ............................................................... 14 15 Figure 10.Area trends for principal food crops. Rwanda. 2001-05 .............................................. Figure 9.Productiontrends for principal food crops. Rwanda. 2001-05....................................... 20 Figure 11.Yield trends for principal food crops. Rwanda. 2001-05 ............................................. 20 Figure 12.Proportion o f fertilizer appliedto different crops. Rwanda. 2001-03 average.............20 Figure 14. Trends incoffee prices. Rwanda and world markets. 1975-2005 ................................ Figure 13. Labor productivity inagriculture. Rwanda vs. selected African countries. 2005 ........25 31 41 Figure 16. Quality shares o f Rwandan coffee exports. 1986-2004................................................ Figure 15. Trends intea prices. Rwanda vs.regional and global prices. 1989-2005 .................... 41 42 Figure 18.Average budget shares. by income quintile. Rwanda. 2001......................................... Figure 17.Relationship between landholding size and income. Rwanda. 2001............................ 56 Figure 19.Marginal budget shares. by income quintile. Rwanda. 2001........................................ 58 59 Figure 21. GDP growth by household type. REMMBase Run Scenario. 2005-15....................... Figure 20.Annual GDP growth. REMMBase Run Scenario. 2005-2015 .................................... 63 Figure 23.Annual GDP growth. REMMBase Runvs.Growth Scenario. 2005-2015 .................65 Figure 22. Declines inpoverty rate by household type. REMMBase RunScenario. 2005-15 .....64 Figure24 GDP growth byhousehold type. REMMBase Runvs., Growth Scenario. 2005-15 . ...68 72 Figure 26.Projected price trends. selected crops. REMMGrowth Scenario. 2005-15 .................74 Figure25. Summary o f Poverty Impactsfrom different Simulations 2006-15 ............................. Figure 27.Projected price trends. livestock products. REMMGrowth Scenario. 2005-15 ..........76 Figure 28.GDP growth by household type. REMMBase Runvs., MDG Scenario. 2005-15 ......76 77 Boxes Box 2. Promotingefficient and sustainable fertilizer use: Lessons from Africa ........................... Box 1.Agricultural growth and poverty reduction.......................................................................... 6 Box 4. Lessons from the Loess PlateauProject inChina .............................................................. Box 3. Developing rural input supply systems: Lessons from Malawi. Kenya. Uganda ..............29 84 Box 6. Strengthening producer organizations................................................................................ Box 5. Stimulatingclient-oriented. demand driven research: Lessons from Senegal ..................-88 86 90 Box 8. Impacts o froad infrastructure on markets and productivity .............................................. Box 7.Promotingexport growth and diversification: Lessons from Bangladesh......................... 93 Box 9. Buildinginstitutional capacity for microfinance: Lessons from Uganda........................... 95 97 vi ACKNOWLEDGEMENTS This report was prepared by a World Bank task team that included Michael Morris (Task Team Leader) Liz Drake, Marie-Claudine Fundi, Marie Jeanne Uwanyanvaya, and Hawanty Page. Valuable contributions were also made by the following World Bank colleagues: Kene Ezemenari, Henry Bagazonzya, Toni Kayonga, Sidi Jammeh, Loraine Ronchi, Verdon Staines, and Joseph Baah-Dwomoh. Madhur Gautam, Dina Umali-Deininger, Kene Ezemenari, and Rodney Dyer (DFID) served as Peer Reviewers. Joseph Baah-Dwomoh, Francios L e Gall, Pedro Alba, Mohammed Toure, and Victoria Kwakwa supported the study and ensured that resources were available for its implementation. The Task Team was fortunate to be able to count on the strong support o f many colleagues inthe Government o f Rwanda, including Hon. Anastaze Murekezi (Minister of Agriculture and Animal Resources), Hon. Daphrose Gahakwa (Minister o f State for Agriculture), Dr. Agnes Kalibata (Secretary General, Ministry o f Agriculture), Mr. Ernest Ruzindaza (Director o f Planning, MINAGRI), Mr. Ernest Rwamucyo (Director General o f Planning and Development, MINECOFIN), Mr. Faustin Minani (Agricultural and Rural Development, Strategic Planning Unit, MINECOFIN), Mr. Naphtal Nyandwi (Monitoring and Evaluation Research, MINECOFIN), Mr Paul Lambers (Advisor to MINECOFIN, Belgian Technical Cooperation), and Mrs. Gerardine Mukeshimana (Project Coordinator, Rural Sector Support Project). Colleagues from many development partners in the Rwanda Rural Cluster contributed information and analysis. These included the African Development Bank (ADB), Belgian Development Cooperation, the United Kingdom Department for International Development (DFID), the EuropeanUnion (EU), the Food and Agriculture Organization o f the United Nations (FAO), the InternationalFund for Agricultural Development (FAD), the Japanese International Cooperation Agency (JICA), the Royal Netherlands Embassy, the United States Agency for International Development (USAID), and the World Food Program (WFP). A special debt o f gratitude is owed to the authors o f three backgroundpapers: Georges Aertssen, Augustine Mutijima, Sylvain Mbarubugeye, Frans Goossens, Serge Rwamasirabo, Xinshen Diao, and Binxen Yu. The important contribution made by colleagues from the International Food Policy ResearchInstitute (IFPRI) i s also recognized. Funding for these background studies was provided by the World Bank, The Government o f Belgium through a World Bank-administered Trust Fund and through the Belgian Technical Cooperation Program, the Government o f Japan through a Government o f Rwanda-administered trust fund, and the UK Department for International Development. vii ... Vlll EXECUTIVE SUMMARY 1. This report summarizes the findings o f a study undertaken by the World Bank at the request o f the Government o f Rwanda. The study hadthree main objectives: (9 Validate the argument that agriculture has potential to become a leading engine o f pro- poor growth in Rwanda and identify potential sources o f rapid and sustainable growth within the agriculturalsector; (ii)Identifykeyactionsthatwillbeneededtounlockthesesourcesofagriculturalgrowth, and describe actions in other sectors that will be needed to support the successful implementationo f the government's agriculturalpolicy agenda; and (iii)Confirmthe congruence between the priority actions neededto stimulate increased agricultural growth and the policy reforms, institutional changes, and supporting investments envisioned under the Plan StratCgique de TransformationAgricole (PSTA). AGRICULTUREINTHE ECONOMYOFRWANDA 2. Nationally, agriculture i s the most important sector in terms o f contribution to GDP, employment, and foreign exchange earnings. Agriculture's contribution to growth i s even larger when strong multiplier effects are taken into account. Agriculture also contributes significantly to national food self-sufficiency, as over 90 percent o f all food consumed in the country i s domesticallyproduced. 3, The agricultural sector currently accounts for about 42 percent o f GDP inreal terms. This figure i s likely an underestimate, because it i s difficult to measure the large amount o f food that i s produced and consumed at home. The sectoral share o f agnculture in the national economy has fluctuated in recent years around a modest upward trend. Between 2001 and 2005, agricultural growth averaged 4.2 percent per year, below the target range of 5 to 8 percent set out in the Poverty Reduction Strategy Paper (PRSP) and Vision 2020 document. Considerable year-to-year variability around this trend was caused mainly by climatic fluctuations. 4. In2005, approximately 90 percent of the economically active population was employed in agriculture. Despite government efforts to encourage migration of labor out of agriculture to relieve pressure on the country's severely constrained land resources, agnculture remains by far the main source o f employment. EXTENT DISTRIBUTIONOFPOVERTY AND 5. Rwanda i s one o f the poorest countries in the world, with an average annual per capita income o f only US$245. Results from the most recent Rwanda Integrated Household Living Conditions Survey (EICV, 2001) show that more than one-half o f all Rwandans (52 percent) live in extreme poverty as measured by the international standard of US$1 per day in income, and more than three-quarters o f all Rwandans (84 percent) live inmoderate poverty o f less than US$2 per day inincome. Poverty inRwanda i s concentrated in the countryside: the rural poverty rate i s 67 percent. Poverty i s also strongly associated with working in agriculture, especially for wage laborers and female- or widow-headed households. i x RWANDA'S DEVELOPMENTSTRATEGY 6. The Government's strategy for reducing poverty and stimulating rapid and sustainable economic growth i s laid out in the PRSP and further articulated in Vision 2020. In these two documents, the Government proposes an ambitious development program to transform Rwanda into a middle-income country by 2020. The bold targets o f Vision 2020 include increasing per capita GDP from US$230 in 2000 to US$900 in 2020, reducing the proportion o f the poor from 60 percent o f the population to 25 percent, increasing life expectancy from 49 to 65 years, and increasing the literacyrate from 48 to 90 percent. 7. The apculture sector i s identified in the PRSP and Vision 2020 as a leading engine o f future economic growth. According to Vision 2020, agricultural transformation i s expected to boost growth inboth the formal and informal sectors, with the effect o f reducing the proportion o f the population dependant on agriculture from the present 87 percent to about 50 percent in 2020. Because agricultural productivity i s currently very low, there i s considerable potential to achieve rapid income gains by increasing productivity at the farm level. Vision 2020 calls for greatly increased public investment in agriculture during a "primary growth phase" lasting from 2002 to 2006, to be followed by greatly increased private investment during a "consolidation phase" lasting from 2006 to 2010. 8. A key pillar o fthe agricultural development strategy i s the PSTA, which was finalized in 2004 and launched inJune 2006. The PSTA aims to increase the incomes o f the rural population by improving agricultural productivity and facilitating transformation from a subsistence economy to one that i s geared to productionfor both domestic and export markets. CROPPINGSYSTEMSAND PRODUCTIONTRENDS 9. Agriculture inRwanda i s dominated by small-scale, subsistence-oriented family farming units.Approximately 1.4 million rural households depend on agriculture as their main livelihood source. These households produce a range o f food crops (cereals, roots and tubers, bananas, and vegetables), with approximately 66 percent o f production destined for home consumption. The remaining 34 percent o f production finds its way to local markets. Crops are produced mainly under rainfed conditions using mostly family labor and few or no purchased inputs (improved seed, fertilizer, and crop protection chemicals). Approximately 60 percent o f households also keep animals for milk, eggs, and meat. These animals, are mainly local breeds, are raised using traditional low-input extensive grazing methods, although in the case o f cattle the dwindling availability o f pasture land i s causing a shift to confined feeding with cut fodder supplemented by grain and/or roots and tubers. A minority o f rural households also produces export crops, the most important o f which are coffee (cultivated by approximately one-third of all rural households) and tea (cultivated by less than 1percent o f all rural households). 10. Food crops dominate the area planted to annual crops, reflecting the subsistence orientation o f Rwandan agriculture. In 2005, roots and tubers accounted for the largest share of total cropped area (25 percent), followed by bananas (22 percent), cereals (21 percent), pulses including beans and peas (21 percent), h i t s and vegetables (5 percent), and oilseeds (3 percent). Traditional export crops accounted for only 3 percent o f total cropped area, including coffee (2 percent), tea (1 percent), and pyrethrum (4percent). 11. Livestock production contributes significantly to the agricultural economy, but this contribution i s to some extent hidden, because most livestock products are consumed at home and do not enter the market. Livestock herds were decimated duringthe genocide, but they have since X recovered, and numbers o f cattle, sheep, and goats are now approximately equal to pre-genocide levels. However, productivity remains low wit a limited adoption of technological change. 12. Food security i s a concern for policy makers. Fueled by high population growth and modest income gains, demand for food has outstripped food production increases, so national food self-sufficiency has declined. Food imports, destined mainly for urban markets, increased 11.6 percent between 2003 and 2005. Although its food self-sufficiency rate i s high (more than 90 percent), Rwanda remains a structurally food-deficit country, importing approximately 130,000 tons o f food per year, mainly edible oil, wheat, sugar, rice, beans, maize, cooking bananas, and dairy products. AGRICULTURALGROWTHDRIVERS 13. Rwanda's strategy to stimulate increased agricultural growth, as articulated inthe PSTA, focuses on raising agricultural productivity and increasing production through a series o f interventions directed at the supply side: intensifying sustainable production systems, strengthening research and extension, improving input distribution, building capacity among farmers' organizations, promoting commodity chain development, improving export competitiveness, and strengthening the institutions that support agriculture. These measures are undeniably important. Yet agricultural growth cannot be sustained indefinitely with a supply-push strategy. Over the longer term, it will not be possible to sustain increased production o f food and cash crops unless the increased production finds a ready market. So where will future demand come from for commodities producedinRwanda? Demand 14. Three main sources of demand exist for Rwandan agricultural products: (1) domestic markets, (2) regional markets, and (3) international markets. 15. Domestic markets have received relatively little attention in most discussions o f future sources o f agricultural growth in Rwanda. Yet, with the nation's population growing at nearly 3 percent per year, domestic demand for food will grow at a similar rate in the short to medium term. Inaddition, given present low levels o f food consumption, especially among lower income groups, future income gains will translate into increased per capita food consumption, which could easily add an additional 1-2 percent to demand. Combining these two effects, domestic demand for food can support 5 percent growth in the domestic food crop and livestock sectors duringthe short to medium term. 16. This is not to say that external demand is unimportant. Domestic demand for food will eventually level off when the population stabilizes and income growth slows at higher per capita income levels. From that point on, agricultural growth will be sustainable only if opportunities can be exploited to export into regional and global markets. Developing agricultural exports will be needed as well to diversify the rural economy, making it less vulnerable to short-term fluctuations inthe fortunes o f individual crops. 17. For Rwanda's traditional export commodities, especially coffee and tea, demand in international markets remains strong, although the nature o f that demand will continue to change as consumers shift into higher quality specialty grades. In the short to medium term, traditional export commodities-especially coffee and tea-will continue to make an important contribution to agricultural growth. Over the longer term, development o f new non-traditional export sectors will be neededto sustain export growth. xi Supply 18. What will drive future growth inRwandan agriculture? On the supply side, three sources o f future growth are possible: (i) increased scale o f production, (ii) increased productivity, and (iii) addition. value 19. Increased scale of production: The most obvious way to increase the scale of production would be to expand the land frontier. With the highest population density in Africa and the smallest average farm sizes, Rwanda clearly faces a major challenge with regardto land. There is a widespread perception that the land frontier i s by now largely exhausted, but this perception may be unwarranted. Total cultivated area could continue to increase at a rate o f 1 to 1.5 percent per year for the foreseeable future through intensification. Experience from other densely populated countries, for example in East and South Asia, also suggests that cropping intensity can increase well beyond the level at which Rwanda currently finds itself, although significant investments will be needed in land, soil, and water management technologies. Although there i s little scope for bringing new land under production, intensification o f the currently cultivated area should be possible if underutilized marshlands can be developed and if upland areas can be cultivatedmore intensively(i.e., multiple-cropped). 20. Increased productivity: Crop yields inRwanda are generally speaking low compared to yields in neighboring countries, especially when Rwanda's generally more favorable agro- climatic endowment i s considered. This suggests that possibilitiesexist to increase productivity in Rwandan agriculture. A background study commissioned for this report assessed the prospects for achieving productivity gains in Rwanda, taking into account what i s technically feasible and economically rational for farmers. For food crops, comparison o f farm-level yields (actual yields) with experimental yields (potential yields) shows large yield gaps ranging from a low o f about 25 percent inbanana to 75 percent or more for wheat, maize, and beans. For horticultural crops, the yield gaps are similarly large. While it i s unreasonable to suppose that farm-level yields will attain the same level as experimental yields, the gaps observed in Rwanda are large compared to those observed inmany other developing counties. The yield gap analysis thus suggests that the productivity growth targets being pursued by the Government o f Rwanda are reasonable. Inmost cases, the goal i s to achieve yield growth inthe range o f 3 to 5 percent per year over the short to medium term. Given the very low productivity levels infarmers' fields, and considering the scope for achieving significant yield gains through relatively modest changes in crop and resource management practices, productivity gains o f this magnitude are considered achievable. 21. Increased value of production: Currently, most food staples destined for the domestic market are consumed with little or no processing. The nature o f demand will change, however, as consumers beginto demand more highlyprocessedfoods and/or foods with special qualities, such as organically produced foods or nutritionally fortified foods. In contrast, the traditional export crops o f coffee, tea, pyrethrum, and hides and skins are sold in global markets where demand for quality i s evolving very rapidly and where niche markets for specialized highvalue products offer increasingly attractive returns for those who can compete successfully. For producers and exporters o f Rwanda's traditional export commodities, the challenge will be to adopt quality enhancement measures and develop linkages to high-value niche markets. The premium prices that can be obtained in these markets will help to promote sustained production and contribute appreciably to future agricultural growth. COMPETITIVENESSOFRWANDAN COMMODITIES 22. The fact that Rwanda i s landlocked, with relatively poor links to external markets, has both positive and negative implications for Rwanda's agricultural sector. On the positive side, the xii country's geographical isolation, coupled with its predominantly mountainous terrain, confers a natural level o f protection for Rwandanproducers inthe domestic market, because hightransport costs make the prices o f imported commodities very high within the country. On the negative side, the same factors that confer natural protection to Rwandan producers in domestic markets pose a major challenge to producers looking to regional or global markets. The high cost o f transporting Rwandan commodities to regional or global markets must be absorbed in order for those commodities to be competitive outside the country. These effects are particularly important in the case of unprocessed staples that have a high volume-to-value ratio, including most roots and tubers, bananas, and cereals. For these low-value commodities, transport costs represent a large share of the final price, and this discourages transportation over large distances. 23. Because high external transport costs add appreciably to the landed cost o f food imports, Rwandan food crop producers are generally able to compete effectively with imports indomestic markets, even when domestic production costs are relatively highby international standards. Yet rice, wheat, maize, beans, sugar, and vegetable oils are imported regularly from regional and international markets to compensate for structural deficits in production. Bananas, potatoes, cassava, beef, and fish, are imported in smaller quantities from neighboring countries, mainly Uganda, Democratic Republic o f Congo, and Tanzania, with the volumes varying considerably from year to year depending on localproductionconditions. 24. High transport costs also affect the competitiveness o f traditional export crops, notably coffee and tea, but two factors have enabled Rwandan coffee and tea producers to remain competitive in international markets. First, coffee and tea have high volume-to-value ratios, so transport costs make up a relatively smaller share o f the final selling price than i s the case with food staples. Second, by pursuing a high-quality strategy for coffee and tea, Rwandan exporters have been able to differentiate Rwandan products in global markets. This has allowed them to builddemand for Rwandanbrandedproducts, which inturnhas allowed them to obtain premium prices, especially for specialty coffees. 25, Withthe goal o f stimulating increased growth, increasing foreign exchange earnings, and diversifying commercial agriculture away from its current heavy reliance on coffee and tea, policy makers are promoting non-traditional export crops-for example, horticultural crops including fruits, vegetables, and cut flowers; essential oils such as petunia and geranium; macadamia nuts; vanilla; and silk. Non-traditional export crops show some promise, but they are unlikely to be a major source o f long-term pro-poor growth for the country, for two reasons. First, the scale o f production i s very limited. All o f these enterprises are capital intensive, and while many do also generate employment, wage growth will be constrained by the need to compete with similar enterprises in countries that have better access to end markets. Second, most non- traditional exports are targeted at niche markets, which by definition are small and ephemeral. PROSPECTSFORGROWTHAND POVERTYREDUCTION 26. Options for increasing agncultural growth in Rwanda were assessed using the Rwanda economy-wide multi-market (REMM) simulation model. The REMM also made possible evaluation o f the linkages and the trade-offs between growth and poverty reduction at both macro and micro levels. Base RunScenario 27. The REMMwas developed to allow simulation o f the likely impacts o f alternative policy scenarios on growth, incomes and poverty, food security, and the trade balance. Since these impacts must be evaluated relative to a baseline, first it was necessary to simulate a Base Run xiii Scenario representing the "business as usual" option, under which agricultural and non- agricultural growth are assumed to continue along current trends. The results o f the Base Run Scenario show that business as usual i s not an option if Rwanda i s to meet its national development targets, including the Millennium Development Goals (MDGs) and the targets agreed to under the New Economic Partnership for Africa's Development (NEPAD). The results o f the Base Run Scenario highlight that a continuation o f current policies will bring about a modest reduction inthe national poverty rate, but the absolute number o f people living below the poverty line will increase because of population growth. Food self-sufficiency at the national level will be eroded in the face o f demographic pressure and rising food imports. The trade balance will improve only slightly as increased export earnings stay just ahead o f the rising cost o f food imports. Growth Scenarios 28. To assess how agriculture can contribute to growth and poverty reduction in Rwanda, a series o f simulations was camed out usingthe REMM. Like any simulation model, the REMM has limitations, and projections generated using the REMM should be considered merely indicative. Still, a number o f results emerging from the REMM modeling exercise are compelling, and they have important policy implications summarized below. 29. Agriculture has the potential to be a leading engine of growth for Rwanda's economy over the short to medium term. Based on the assumptions used in the REMM, the agricultural sector i s projected to grow at an average annual rate o f 6.17 percent through 2015. Robust agricultural growth will fuel average annual growth intotal GDP o f 6.24 percent. 30. Within agriculture, the main drivers of growth will be food staples. Based on the assumptions used in the REMM, some agricultural sub-sectors will grow more rapidly than others. The contribution o f each sub-sector to GDP growth will depend not only on the rate o f growth achieved inthat sub-sector, but on the absolute size o f the sub-sector. Taking into account the large absolute size o f the food staples sub-sectors, most of the growth inagriculture will come from growth in production o f food staples (cereals, root crops, oilseeds, and livestock). Of the projected growth in agricultural GDP, over two-thirds will come from projected growth in production o f food staples. 31. Export cropswill make a significant contribution to growth. Export crops will make a significant contribution to growth, but the importance o f this contribution will be limited because the export crop sub-sector i s small relative to the food crops and livestock sectors. Of the projected growth in agricultural GDP, slightly more than one-quarter will come from production o f export crops. 32. Agriculture-led growth will lead to significant income gains. Based on the assumptions used in the REMM,robust agricultural growth averaging 6.17 percent per year will fuel growth in total GDP o f 6.24 percent per year, allowing total GDP per capita to grow at an average annual rate o f 3.44 percent, almost triple the rate projected under the Base Run Scenario. 33. Staple-led growth i s more pro-poor. Growth in the production o f food staples (including livestock) has a greater impact in terms of reducing poverty than growth in the production o f export commodities. The largest impacts in terms o f poverty reduction will come from growth in production o f crops that are grown by the poor and also consumed by them: beans, pulses and oilseeds, sweet potatoes, and soybeans. Growth in the production o f export commodities i s more variable. Pyrethrum-led growth shows highpoverty-growth elasticity, and xiv coffee-led growth shows moderately high poverty-growth elasticity, whereas tea-led growth shows low poverty-growth elasticity. 34. At the national level, food self-sufficiency will decrease. Based on the assumptions used in the REMM, strong income growth combined with population growth will fuel consumption increases that will outstrip production gains inmost food crops. The current deficits inmaize, wheat, beans, sugar, and vegetable oils will increase, while those for rice andmilkwill disappear. For all other commodities, including roots and tubers and livestock products other than milk,productionwill increase roughly inline with growth indemand, and current levels o f self- sufficiency will be maintained. 35. Agriculture will help to reduce the trade deficit, but it will not be able to eliminate it completely. Based on the assumptions used in the REMM, Rwanda's total agricultural trade surplus will increase to approximately US$ 66 million by 2015, possibly to $100 million if the strategy o f targeting the specialty coffee market proves successful. The overall balance o f trade will improve, butnot sufficiently to eliminate the trade deficit. 36. Differences in production and consumption growth in individual commodities will lead to imbalancesin supply and demand. For example inthe case o f rice, potatoes, and some livestock products (poultry and eggs) projected increases in production are considerably higher than the projected increases in demand. Domestic prices for these commodities therefore are likely to fall, benefiting consumers, butresultinginpotentialrevenue losses for some producers. 37. Agricultural growth will contribute to the attainment of the first MDG of halving poverty by 2025, but agricultural growth alone will not to be sufficient. With significant focused investments, agricultural growth o f at least 6 percent per year i s achievable, which i s the target growth rate set under the Comprehensive African Agricultural Development Program (CAADP) o f NEPAD. While the goal o f halving poverty by 2015 could be achieved through many possible combinations o f agricultural and non-agricultural growth, overall GDP growth would have to exceed 8 percent per year from 2005 to 2015, which i s unlikely to happen unless agricultural growth can reach 9 percent or more. CONCLUSIONS: PRINCIPAL CONSTRAINTS TO AGRICULTURAL GROWTH 38. The principal constraint to agricultural growth in Rwanda i s the low profitability of agriculture, not only at the farm level but further along the value chain at the levels o f processing, storage, and marketing. The low profitability o f agriculture discourages investment in the sector and leaves the vast majority o f producers trapped in a low-input, low-output existence characterizedby highlevels o f persistentpoverty and chronic food insecurity. 39. The low profitability of agriculture in Rwanda results from many underlying causes, o f which six stand out: 1. Low farm-level productivity inboth the crops and the livestock sub-sectors caused by producers' inadequate knowledge o f improved management practices and compounded by their limited use of land and water management practices that permit sustainable intensification(e.g., irrigation, terracing). 2. Limited availability and high cost of agricultural inputs (especially seed, fertilizer, crop chemicals, and machinery) resulting from poorly developed input distribution systems and exacerbated by hightransportation costs and a weak private sector. xv 3. Low and variable prices received for outputs attributable to poorly developed commodity marketing systems and exacerbated by hightransportation costs, particularly for regional and international exports. 4. Lack of access to rural financial services needed to support productive investment in primary production activities as well as post-harvest value-adding activities. 5. Unfavorablebusiness climate that discourages private investment by subjecting small- and medium-scale enterprises to costly and time consumingregulatoryprocedures. 6. Weak human capitalbase, both interms o f slulls (attributable to the lack o f education opportunities for the rural population, particularly opportunities for vocational training) and also in term of physical capacity (resulting from poor nutrition, inadequate rural health services, and prevalence o f debilitating diseases, especially HIV/AIDS). RECOMMENDATIONS: Ah' AGENDA FOR IMMEDIATEACTION 40. The Government o f Rwanda has set itself a target o f sustained growth in agriculture of 6 percent over the short to medium term. Based on the analysis carried out for this study, recognizing the constraints identified above, and considering the programs and sub-programs identifiedinthe PSTA, what should be the Government's immediatepriorities? 41. The basic conclusion o fthis report i s that the main thrusts o f the PSTA are fundamentally sound and that the areas targeted for reform under the PSTA are consistent with those identified here. At the same time, the agenda spelled out in the PSTA i s extremely comprehensive. Given the financial and human capital constraints facing MINAGRIand the other organizations charged with implementing the government's agricultural development agenda, it will not be possible to implement all o f the PSTA programs and sub-programs immediately. Some prioritization is needed to ensure that the available financial and human resources can have the greatest possible growth and poverty reduction impact inthe short to medium term. 42. Rapid and sustainable growth inRwanda's agricultural sector can be achieved only ifthe productivity, profitability, and competitiveness o f agriculture can be improved. Based on the information and analysis presented in this report, and taking into account the most binding constraints identified above, we argue that seven areas should be targeted for immediate attention (for a detailed list o f actions, see Tables 35 and 36). 1.Stimulatingproductivitygrowthinfood staples(crops andlivestock) 43. Unlocking pro-poor agricultural growth in Rwanda will depend on increasing productivity growth in food staples. The REMM modeling results indicate that most o f the growth in agriculture will come from growth in production o f food staples, both crops and livestock. This growth will be pro-poor in that it will disproportionately benefit low income groups, including households that do not grow cash crops and households headed by women. Sustainable intensification o f staple food crop production will not be possible unless farmers increase their use o f purchased inputs, especially improved seed, fertilizer, crop chemicals, and animal health products. These inputs will not be available unless inputmarkets are working well. Strengthening input supply systems therefore forms an integral component o f the overall agriculturalgrowth agenda inRwanda. 2. Scalingup sustainabledevelopmentof landandwater resources 44. Unlocking agricultural growth in Rwanda will depend on improving the natural resource base on which agriculture depends. Evidence presented in this report on the high rate o f land xvi degradation, soil fertility losses, limited area under irrigation, and low use o f improved water management practices illustrates the urgent need to scale up investment in the sustainable development and management o f land andwater resources. 3. Overhauling nationalagriculturalresearch and extensionsystems 45. Unlocking agricultural growth in Rwanda will depend on overhauling the national agricultural research and extension systems. Technology-driven increases in agricultural productivity are critically neededto get Rwandan agriculture going. While new technology can be imported from outside the country, the distinctive features o f many o f Rwanda's production systems mean that much new technology will have to be adapted to local conditions. For this to happen, research and extension systems will have to be reformed and revitalized, with a clear delineation o fpublic and private roles infundingand service delivery. 4. Strengthening producer organizations 46. Unlocking agricultural growth in Rwanda will depend on strengthening the capacity o f producer organizations to compete effectively in domestic, regional, and international markets. Strong producer organizations will be needed if agnculture i s to move away from traditional subsistence farming towards more market-oriented commercial agriculture. Achieving a successful transformation will depend on farmers' ability to identify actual and potential market opportunities, access cutting-edge technology, procure essential production inputs, produce high- quality products that meet the requirements o f an increasingly quality-conscious market, and negotiate effectively with input suppliers as well as buyers. 5. Promoting agricultural export growth and diversification 47. Sustaining increased levels o f agricultural growth inRwanda will not be possible unless the export commodity sub-sectors can be revitalized and strengthened to meet rapidly evolving market requirements. The REMMmodeling results show that export crops can make a significant contribution to growth. Immediate attention should be directed to strengthening and deepening the ongoing reforms in the coffee and tea sectors, as these traditional export crops have the greatest potentialto contribute to growth, and inthe case o f coffee, poverty reduction, inthe short run. Efforts should continue as well to revitalize the pyrethrum industry and to exploit the untapped potential in the hides and shns sector. Inaddition, opportunities should be explored to develop non-traditional export crops, particularly horticultural crops, through launching o f pilot export promotionprojects that target the development o f value chains for particular commodities. 6. Improving the performance of agriculturalmarkets 48. Unlocking agricultural growth in Rwanda will depend on improving the performance o f agricultural markets. Evidence presented in this report has documented l o w rates o f market participation among rural households, the lack of reliable market outlets for many staples, the highreal marketingcosts associated inmany caseswith deficienttransportation infrastructure and storage facilities, the high seasonal variability o f commodity prices, and the lack o f access to timely and accurate market information. Rapid and sustained agricultural growth leading to structural change and, eventually, diversification o f the rural economy will not happen without well-functioning markets populatedby a dynamic and innovative private sector. 7. Improving access to ruralfinancial services 49. Unlockmg agricultural growth in Rwanda will depend critically on improving access to rural financial services. Current low levels o f investment by financial institutions in the agricultural sector can be explained inpart by their perception that lending to rural clients i s risky xvii and unprofitable. Rural entrepreneurs, not only producers o f primary commodities but also those engaged in post-harvest activities such as processing, storage, and transporting, need resources for productive investment, butthey have difficulty accessing credit through formal channels. DELIVERING GROWTH Strengtheningthe institutionsthat support agriculture 50. Buildingeffectively on the seven growth pillars will require strengthening the institutions and policies that support agriculture. Unlockmg agricultural growth inRwanda will depend on the existence o f strong and responsive institutions with the capacity to provide critical goods and services, while at the same time performing essential coordinating and regulatory functions. Institutional refoms in the agricultural sector have led to the creation of some new agencies (RADA, RARDA, RHODA) and the restructuring of others (ISAR). The capacity of many o f these institutions needs to be further strengthened. 51. The biggest institutional delivery challenge revolves around the decentralization reforms enacted in 2006. Beginning in 2007 public agricultural development activities will be supported mainly through earmarked transfers fiom the central government to districts, as well as investments from the government's development budget. Establishing and implementing a clear institutional framework for decentralization, with clear accountability mechanisms and building capacity for decentralized delivery by stakeholders inthe sector, i s vital. Crosscuttingpolicy issues 52. Inrecentyears, Rwandahashadastable economic climate, afairly valuedexchangerate, a modest rate o f inflation, and reasonably open borders for trade. Thus there are no major macro policy constraints distorting the agricultural sector at present. Still, in support o f longer term growth, policies that impact indirectly on agriculture will require attention. These include: 53. Trade policies: Ongoing efforts to promote regional trade integration under COMESA take on increasing importance for supporting long term growth in agriculture. Competition on domestic food staples production i s likely to increase, especially ifTanzania joins COMESA. 54. Business and ICT policies: Agriculture in Rwanda will not be able to grow without a significant increase in investment, much o f which will have to come from the private sector. An improved business climate will be needed to attract this investment. Developing information communications technology to meet business and market informationneeds i s also a priority. 55. Familyplanning, gender, and education policies: Increasing demographic pressure on limited resources highlights the urgent need for better implementation o f family planning policies. Since food crops are produced mainly by women, the gender dimension o f agricultural policies and programs needs to be considered. Promotion o f equitable access for women to productive assets (e.g. land, credit and labor saving technologies) must be a priority. Education, particularly for women will be important withrespect to empowerment o freproductiverights. xviii 1. INTRODUCTION Background 1. Agriculture, the backbone o f Rwanda's economy, will be central to stimulating economic growth and reducing poverty. Yet inrecent years Rwanda's agricultural sector has fallen short o f expectations. Agricultural productivity has remained flat, and average farm sizes have declined in the face o f steady population growth, leaving farm incomes lower than they were 20 years ago. Meanwhile, marketable surpluses have declined, indicating that more and more households are adopting a "retreat into subsistence" livelihoods strategy (McKay and Loveridge 2005). With less production being sold, the income available to stimulate demand in the non-farm rural sector has fallen. The sluggish performance o f agriculture has eroded national food self-sufficiency, contributed to higher food prices, and increasedfood insecurity for many households. 2. In2004, the Government of Rwanda formulated a National Agricultural Policy (NAP) and developed a Strategic Plan for Agricultural Transformation (PSTA). The expressed goals o f the NAP are to contribute to national economic growth, improve food security and the nutritional status o f the population, and increase rural incomes. The objective o f the PSTA i s to make a "sustainable contribution to poverty alleviation and economic growth through improved productivity o f production factors, creation o f added value, diversification o f income-generating opportunities, and preservation o f natural resources." The PSTA encompasses four broad program areas: (i)Intensification and Development o f Sustainable Production Systems, (ii) Support to the Professionalization o f Producers, (iii)Promotion o f Specialty Crops and Development o f Agri-business, and (iv) InstitutionalDevelopment. Objectives 3. This report summarizes the findings of a study undertaken by the World Bank at the request o f the Government o f Rwanda. The study had three main objectives: (i) Validate the argument that agriculture has potential to become a leading engine o f pro- poor growth in Rwanda and identify potential sources o f rapid and sustainable growth within the agriculturalsector; (ii) Identify key actions that will be needed to unlock these sources o f agricultural growth, and describe actions in other sectors that will be needed to support the successful implementationo f the government's agricultural policy agenda; and (iii)Confirm the congruence between the priority actions needed to stimulate increased agricultural growth and the policy reforms, institutional changes, and supporting investments envisioned under the PSTA. Organization of the report 4. This report consists o f five sections, o f which this Introduction is the first. Section 2 describes the importance o f agriculture inthe economy o f Rwanda and analyzes recent trends in the performance o f the agricultural sector. Section 3 discusses possible future drivers o f growth, considering both the demand and the supply side. Section 4 presents the results o f a modeling exercise that explores the likely future impacts o f alternative growth strategies on incomes, poverty, foreign exchange earnings, and food security. Section 5 concludes by discussing key actions within the agricultural sector that will be needed to stimulate sustainable pro-poor growth. 1 2. AGRICULTURE INTHE ECONOMY OFRWANDA Macroeconomic overview 5. Rwanda has made a remarkable transition from genocide to peace and development. Since 1994, the Government has been able to maintain overall macro stability and implement extensive reforms that have contributed to the strong growth performance observed over the past decade. Throughout most o f this period, economic growth has been driven by a recovery in the agricultural sector, complemented by a construction boom during the post-conflict reconstruction phase. This growth has continued up to the present. During the period 2002-05, overall GDP growth averaged close to 5 percent per year, with low and stable inflation (Table I).Meanwhile, non-income indicators o f well-being also have improved significantly. Primary school net enrollment now stands at 93 percent, and the vaccine coverage rate for all antigens ranges from 80 to 95 percent inmost provinces. 6. On the social and political front, while the country i s at now at peace and secure, Rwandans continue to deal with the legacy o f genocide. Progress in regenerating the torn social fabric and rebuilding a national sense o f unity i s being made through the national unity and reconciliationprocess, ledby the Government. Extent and distribution ofpoverty 7. Rwanda i s one o f the poorest countries in the world, with an annual income of only US$245 per capita. More than one-half of all Rwandans (52 percent) live in extreme poverty as measured by the international standard o f US$1 per day inincome, and more than 3 out o f 4 (84 percent) live in moderate poverty o f less than US$2 per day in income (World Bank, 2005). At least 60 percent o fthe population lives below the national poverty line.' 8. Poverty in Rwanda i s concentrated inrural areas. Results from the most recent Rwanda Integrated Household Living Conditions Survey conducted in 1999 (Enqukte IntCgrale sur les Conditions de Vie des MCnages au Rwanda - EICV) show that poverty i s strongly associated with living in a rural area and with working in agriculture, whether on one's own farm or as a hired laborer working on a farm owned by another person (Figure I).The incidence o f poverty i s higher inparticular ex-provinces, notably those withhigher population densities (Figure 2). 9. Disaggregating the national EICV data, 14.3 percent o f urban residents are classified as poor. This figure increases to 65.7 percent o frural residents, with 97.5 percent o f those identified as falling below the overall poverty line livinginrural areas. While 10.7 percent o f the population lives in urban areas, 2.5 percent o f these are considered as poor. The depth o f poverty (the average proportion by which the poor fall below the poverty line) i s also greater in rural areas (42.4 percent) than inurban areas (28.0 percent). ' The Enqugte Intkgrale sur les Conditions de Vie des Mknages au Rwanda (EICV) uses two consumption poverty measures, based on estimated minimum requirements (taken as 2500 Kcal per adult equivalent per day): (i) A food poverty line of FRW 45,000 per equivalent adult per year, in January 2001 prices. This was estimated based the cost o f purchasing a basket o f foods which satisfied reasonable calorie and protein requirements and which reflected the average consumption pattern of the poorest 60% o f people. (ii)An overall poverty line o fFRW 64,000 per equivalent adult per year, again inJanuary 2001prices. Thiswas obtainedbyadding anallowance for non-foodrequirements, estimated based on the average proportion o f the budget devoted to non-food items by people whose food expenditure was around the food poverty line. 2 ' " 9 " ' " dv,v,4 n Q ent rates inRwanda, by source of e~ployment,2001 80 5 70 h Ev) 60 c0 Q) 50 v) c 40 a * 0 .- 30 n -5Pe20 10 0 Public Para-publlc Prlvate Private Prlvate Own Non-farm Notworking sector sector sector sector sector account (formal) (informal - (informal- agriculture agriculture) non-farm) Source: MINECOFlN 2002a. ulation densi~yandpovertyrates,Rwanda, 2005 Source: MINECOFIN,2002b. 4 Rwanda's developmentstrategy 10. The Government's strategy for reducing poverty and stimulating rapid and sustainable economic growth i s laid out in the PRSP and further articulated in its Vision 2020 document. In these two documents, the Government proposes an ambitious development program to transform Rwanda into a middle-income country by 2020. The bold targets of Vision 2020 include increasing per capita GDP from US$230 in 2000 to US$900, reducing the proportion o f the poor from 60 percent o f the population to 25 percent, increasing life expectancy from 49 to 65 years, and increasingthe literacy rate from 48 to 90 percent. 11. The agriculture sector is identified in the PRSP and Vision 2020 as a leading engine of future economic growth. This i s consistent with a wealth o f global evidence showing that agriculture can play a key role during the early stages o f economic development (see Box 1). According to Vision 2020, agricultural transformation i s expected to boost growth in both the formal and informal sectors, with the effect o f reducing the proportion o f the population dependant on agriculture from the present 87 percent to about 50 percent in 2020. In addition, because agricultural productivity i s currently very low, there i s considerable potential to achieve rapid income gains by increasing productivity in the short run. The Government planned to increase public investment in agriculture during the "primary growth phase" from 2002 to 2006 and subsequently to have the private sector play a greater role in investment during the "consolidation phase" from 2006 to 2010. 12. A key pillar o fthe agriculturaldevelopment strategy is the PSTA, which was finalized in 2004 and launched in June 2006. The strategy aims to increase the incomes o f the rural population by improving agricultural productivity and facilitating transformation from a subsistence economy to one that i s geared to production for both domestic and export markets. Role of agriculture in the economy 13. Agriculture features prominently in the economy o f Rwanda. The agricultural sector i s the most important sector in terms o f contribution to GDP, employment, and foreign exchange earnings. Agriculture also contributes significantly to national food self-sufficiency, accounting for well over 90 percent o f all food consumed inthe country. 14. According to official statistics, the agricultural sector currently accounts for about 42 percent o f GDP (Figure 3). This figure i s likely an underestimate, due to the difficulty o f accurately measuring the large amount o f food that i s produced by subsistence formers and consumed at home. The sectoral share o f agriculture in the national economy has fluctuated in recent years around a modest upward trend. Between 2001 and 2005, agricultural growth i s estimated to have averaged 4.2 percent per year. While not insignificant, this rate was below the target range o f 5 to 8 percent set out inthe PRSP and Vision 2020 (Table 2).Considerable year- to-year variability aroundthis trend was caused mainly by climatic shocks. 15. In2005, approximately 90 percent of the economically active population was employed in agriculture. Despite government efforts to encourage migration o f labor out of agriculture to relieve pressure on the country's severely constrained land resources, agriculture remains by far the main source o f employment. 5 Box 1.Agricultural growth and poverty reduction Broad-based agricultural growth can contribute significantly to poverty reduction, especially, during the early stages o f the structural transformation process. A wealth o f empirical evidence shows that agricultural growth can reduce poverty in rural and urban areas, both directly through impacts on farm incomes and employment as well as indirectly through growth linkages and impacts on food prices. For example: Usingdata for 1985 to 1996 for China, Fan et a1(2005) found that higher growth in agriculture reduces both rural and urban poverty, though the pro-poor effect i s largest for rural areas. On the other hand, urban growth contributes only to urban poverty reduction; the effect o f urban growth on rural poverty i s neither positive nor statistically significant. Based on 33 household surveys inIndia from 1951-1990, Ravallion and Datt (1996) found strong evidence that the urban-rural composition o f growth matters to poverty reduction. While urban growth reduced urban poverty, its effect on the national rate o f poverty was negligible. On the other hand, rural growth reduced poverty in rural and urban areas and hence had a significantly positive effect on national poverty reduction. By disaggregating different types of households in Indonesia, Thorbecke and Jung (1996) found that the agricultural sector contributes the most to overall poverty reduction, followed by the services and informal sectors. Based on data from a broad sample o f developing countries in the early 1970s and mid 1980s, Bourguignon and Morrison (1998) found that variables which measure agricultural productivity are important in explaining income inequality. Increasing agricultural productivity was found to be the most effective path for many countries to reduce poverty and inequality. The growth-linkage effects o f agriculture have proven most powerful when agricultural growth i s driven by broad-based productivity increases in rural economies that are dominated by small farms, as i s the rural economy o f Rwanda. Small- to medium-sized farm households typically have more favorable expenditure patterns for promoting growth o f the local non-farm economy, including rural towns, since they spend higher shares o f income on rural non-traded goods and services which are also generally more labor- intensive (Mellor, 1976; King and Byerlee; 1978; Hazel1andRoell, 1983). 6 9 In 3 0 0 e4 0 In 0 N d 0 0 N m 0 0 N N 0 0 c1 r( 0 0 N Figure 3. Sectoral compositionof GDP,Rwanda,2005 Source: MINECOFINMacrodata, 2006. 16. The performance of the agricultural sector has a major impact on Rwanda's trade statistics, because agricultural exports represent the main source of foreign exchange earnings. Agricultural export earnings, derived mainly from coffee and tea, plummeted in the period immediately following the genocide as harvesting ceased and established plantations fell into neglect. More recently, modest production gains achieved in the coffee and tea sectors have allowed foreign exchange earnings to recover, but the value per capita of commodity exports remains muchlower than the average for sub-SaharanAfrica as a whole. 17. Food security remains a concern for policy makers. Fueled by high population growth averaging 2.9 percent per year and modest income gains, demand for food has outstripped food production gains, contributing to a long-term decline in national food self-sufficiency. Food imports, destined mainly for urbanmarkets, increasedby approximately 12percent between 2003 and 2005. Rwanda remains a structurally food-deficit country that imports annually at least 130,000 tons of food, mainly items for which it has an insufficient production (edible oil, wheat, sugar, rice, beans, maize, cooking bananaand dairy products). 18. Food insecurity at the national level i s reflected in inadequate and hghly variable nutrition for many households. Long-tern trends in food production conceal considerable short- runtemporaland spatial variability, includingperiodswhensignificantportionsofthe population experience varying degrees o f hunger (MINECOFIN2002). National survey data show that most Rwandans are permanently in deficit for calories, protein and lipids. Nutritional indicators for children under five years of age also show that over the period 2000-2005, while the height for age has been increasing, the weight for height and weight for age have declined, also indicating nutritional deficits (Health Sector Survey, EDSR, 2005). Natural resource base 19. Compared to many countries in Sub-Saharan Africa, Rwanda's natural resource endowments are favorable for agriculture. The temperate climate, relatively plentiful and bi- modally distributed rainfall, and predominant fertile volcanic soils confer a high agricultural potential. On the other hand, the generally mountainous terrain poses a challenge, because inthe absence o f proper management practices, a large proportion o f the land surface i s highly susceptible to erosion. 8 20. Rwanda's population o f almost 9 million people i s distributed across a land area o f only 26,340 km2.This gives Rwanda the highest average population density in sub-Saharan Africa- approximately 355 inhabitants per kmz(in some places more than 1,000 inhabitants per kmz). Land i s arguably the single most critical natural resource for Rwandan agriculture. Since agricultural livelihoods depend critically on access to land, land i s a key factor o f productionthat heavily influences the welfare o frural households. 21. According to FAO, o f the 2.47 million ha o f Rwanda's total land area (defined as the country's total land surface excluding area under inland water bodies), approximately 1.94 million ha are currently being used for agricultural purposes. Of the latter area, 1.47 million ha are planted to annual or permanent crops, and nearly 0.47 million ha are classified as permanent pasture. The fact that 78.5 percent o f the country's total land surface i s being used for agriculture-a proportion that i s extraordinarily high by global standards-points to the severe scarcity of this critical production factor. 22. The scarcity o f land in Rwanda becomes even more evident when the nation's land endowment i s related to the population size and especially to the number of agncultural workers who depend on land as their primary livelihood source. The average amount o f agricultural land available per rural resident in Rwanda i s about 0.3 ha, and the average amount of arable land (defined as agricultural landnet o fpermanent pasture) i s about 0.2 ha (FAOSTAT). These figures are lower than those for most other countries in Sub-Saharan Africa, and they are comparable to those for the most densely populated countries in Asia (Table 3). When comparing land endowments, however, it i s important to keep in mind that a large share o f agricultural land in most Asian countries i s irrigated, so it i s far more productive. Table 3. Land endowment indicators, Rwanda and selected countries, 2001-03 Rwanda Burundi Uganda Kenya SSA Brazil China India Population density (people I lan') 355 274 136 58 30 21 138 358 Agricultural land/ ruralpopulation 0.3 0.4 0.5 1.3 2.3 8.4 0.7 0.2 Arable landa I ruralpopulation 0.2 0.2 0.2 0.3 0.4 2.1 0.2 0.2 aArable landandpermanent crops (FA0 definition 61) 2000 data Source: Calculated using FAOSTAT data, 2000-2003. 23. Land availability inRwandahas been declining over time inthe face o f steady population growth combined with a stagnant land frontier. Between 1994 and 2004, the rural population o f Rwanda grew at an average annual rate o f 4.4 percent, while the amount o f arable land remained virtually unchanged. A s a result, the amount o f arable land available per rural inhabitant declined from 0.21 ha to 0.20 ha (FAOSTAT). 9 24. These average figures on land availability mask considerable variability in the access to land by individual households. Since land i s often the most binding constraint to production in Rwanda, it i s important to understand the distribution of land and its relationshipto poverty. 25. Insights into the distribution of land in Rwanda can be gained by classifying the rural population into three roughly equal-sized groups, definedby landholding size: Rural Group 1(Land holdings of less than 0.3 ha): Approximately 40 percent of rural households hold less than 0.3 ha. Cumulatively, these households hold less than 6 percent of all agricultural land inthe country. The average landholding per household i s 0.11ha, equivalent to 0.02 ha per capita when household size is taken into account. This group includes 11.5 percent of all householdsholdingno land, i.e., landless households. 0 Rural Group 2 (Land holdingsbetween0.3 ha and 1.0 ha): Approximately 32 percent of rural householdsholdbetween 0.3 ha and 1.0 ha. Cumulatively, these householdshold 25 percent of all agricultural land inthe country. The average landholding per household i s 0.58 ha, equivalent to 0.12 ha per capita when household size i s taken into account. 0 Rural Group 3 (Land holdings of more than 1.0 ha): Approximately 26% of rural households hold more than 1.0 ha. Cumulatively, these households hold more than 70 percent of all agricultural land inthe country. The average landholding per household i s 1.94 ha, equivalent to 0.35 haper capita when household size i s taken into account. 26. The distribution of landholdings by Rural Group i s shown in Figure 4. Summary statistics on land availability and agricultural production activities relating to each of the three Rural Groups appear in Table 4 and Table 5. Figure 4. Distribution oflandholdingsby householdsize, Rwanda, 2001 2.50 ~~ Rural Group Rural Group Rural Group 1 2 3 2.00 h m I 5 t 0 .E 1.50 a 0 a 0 .-mE ,4 eE 1.00 4 U m c , 0 4 , 0.50 c - .--- * _ - * 0.00 - I - 0 10 20 30 40 50 60 70 80 90 100 Cumulativepercentageof rural households Source: IFPRIcalculationsporn EICV data, 2006. 10 27. Poverty inRwanda i s clearly correlated with landholding size. As shown in Table 4, the poverty head count rate within Rural Group 1 (74 percent) i s significantly higher than the rural average of 66 percent. It i s also 7 percent higher than the poverty rate within Rural Group 2 (67 percent) and almost 20 percent higher than the poverty rate within Rural Group 3 (54 percent). Table 4. Distribution of land holdings among ruralhouseholds,Rwanda, 2006 Rural Rural Rural All Group1 Group2 Group3 Rural 0.3 ha 0.3-1.0 ha >1.0 ha Groups Share of rural households(YO) 39.4 31.7 25.8 96.9 Average householdsize (number of persons) 4.5 4.9 5.6 5.3 Totallandholdings(`000 ha) 64 274 750 1,087 Shareof total land holdings(YO) 5.9 25.2 68.9 100 Land holdingsper household(ha) 0.11 0.58 1.94 0.75 Land holdingsper capita (ha) 0.02 0.12 0.35 0.15 Povertyrate (YOhouseholds poverty line) 74 67 54 66 Source: IFPRIcalculationsfrom EICV data, 2006. Table 5. Crop productionby householdgroups, Rwanda, 2006 Rural 1 Rural 2 Rural3 Rural total Totalproduction (1000 mt) Grains 50 81 160 291 Roots andtubers 483 887 1,716 3,086 Bananas 236 712 1,377 2,324 Pulses and oilseeds 38 80 165 283 Sharein rural total( Grains 17.1 27.9 55.0 100 Roots and tubers 15.7 28.8 55.6 100 Bananas 13.6 28.2 58.3 100 Pulses and oilseeds 13.6 28.2 58.3 100 Averageper capitap Grains 19 35 75 41 Roots andtubers 183 384 804 455 Bananas 89 308 645 308 Pulses and oilseeds 15 34 77 40 Source: IFPRIcalculationsfrom EICV data, 2006. 11 28. Poverty in Rwanda i s also correlated with gender. Within each rural group, the poverty rate i s significantly higher among the female-headed households than among male-headed households. For example, the poverty rate for female-headed households with landholding in Rural Group 1i s 78 percent, compared to only 72 percent for male-headed households within the same Rural Group. 29. As mightbe expected ina land-scarce economy, landproductivity inRwanda (defined as agricultural GDP / cultivated land) i s highcompared to landproductivity inneighboring countries (Figure 5). The relatively high level o f land productivity reflects the generally favorable agro- climatic potential, as well as the intensive nature o f the predominant agncultural production systems. Land productivity remains modest by global standards, however, reflecting the limited use o f external inputs, lack o f water and soil management practices, and low levels o f mechanization. This suggests that there i s considerable potential to further increase land productivity. Figure 5. Agricultural value added, Rwanda vs. selectedAfrican countries, 2003 400 Source. WorldBank WorldDevelopment Indicators and FAOSTAT data, 2005. 30. Land productivity has increased slowly during the last three decades, albeit with considerable year-to-year variability around the longer term trend. Land productivity declined sharply in 1994 and took about three years to recover (Figure 6). A sharp drop in land productivity was also observed in2003, when much o fthe country was affected by drought. Water 31. Water i s another critical resource for Rwandan agnculture. Although total annual rainfall i s sufficient to support two cropping seasons in most parts o f the country, rainfall distribution patterns show considerable temporal and spatial variability that can and often do severely disrupt agncultural production activities. The heavy dependence o f Rwandan agriculture on rainfall affects both crop yields (which are both low and highly variable), as well as in the area planted (which fluctuates from year to year in response to changes in rainfall). The combined effect o f variability in yields and variability in area planted i s reflected in pronounced year-to-year variability in agricultural GDP (Figure 7). The weather-related variability in these two key parameters has been exacerbated insome recent years by outbreaks o f civil conflict. 12 32. One way to increase agricultural productivity and reduce weather-related production variability i s through improved water management, including irrigation. Yet Rwanda's irrigation potential remains underexploited. Currently less than 1 percent of the nation's cultivated land i s irrigated, or roughly 6,000 ha out o f the nearly 1.5 million ha planted to annual and permanent crops. Approximately 165,000 ha o f low-lying marshlands have been identified as potentially irrigable (Aertssen et al., 2006), so clearly there i s scope for expanding the area under irrigation, and the Government has made marshlands development a priority. Under the World Bank-funded Rural Sector Support Project (RSSP) and other donor-supported projects, irrigated area has been expanding at a rate o f 2,500-5,000 ha per year, a rate that the Government would like to see increased. Implementation of the ambitious Marshlands Development Master Plan i s however being slowed by a number of constraints, including the high capital investment costs. Recent RSSP data show that the average cost of developing large scale irrigation in marshlands in Rwanda now runs from US$3,500 - US$ 7,000 per ha. Also there i s a lack o f local capacity to design and construct technically efficient and environmentallysound water-control systems. Figure6. Trends inlandproductivity,Rwanda 1,000 600 - 900 t - Trend In agriculturalvalue added / ha 500 H, c m 400 P P %a -3 500 300 a xfz 400 E E 200 .-3 300 2 .-0 zoo I00 100 0 0 1970 1975 I980 1985 I990 1995 2000 Source: Calculated using FAOSTAT data, 2006. 13 Figure 7. Variability in agricultural GDP, Rwanda, 1966-2004 50 variability caused mainlyby climaticfactors 40 Ivariabilityexacerbatedbycivilconflict 30 .-g.10 E - g -20 -E - a E -30 - a -40 I 1966 1970 1974 1978 1982 1986 1990 1994 I998 2002 Source: Calculated using FAOSTATdata, 2006. 33. Development o f marshlands for irrigation represents an obvious avenue for raising agricultural productivity and reducing production variability, but it will not be sufficient given that marshlands represent less than 5% o f the nation's arable land. Therefore it will be necessary as well to improve water management practices in the much more expansive rainfed upland production areas. Currently water management practices are generally poor. Use o f physical water harvesting structures such as terraces, contour bunds, tied ridges, and check dams i s limited, as i s use o f moisture-conserving crop management practices such as contour plowing, mulching, and planting o f cover crops. S& 34. Rwanda's soils contain many o f the metal compounds found in laterite soils, but they are generally lighter, more fertile, more workable, and less troublesome to farmers than true laterite soils. Insome areas, especially inthe northwest o f the country, soils are o fvolcanic origin and are highly fertile. When well managed, Rwanda's soils have good humus content and highfertility, and they compare favorably with soils found inmany African regions. 35. Currently, few o f Rwanda's soils are well managed. Deforestation, overgrazing, and intensive cultivation-often on extremely steep slopes-have led to serious soil erosion throughout the country. Continuous cropping without adequate use o f soil fertility management practices (mulching, green manuring, nutrient recycling, fertilizing) or fallowing (as a last resort) has led to severe soil nutrient miningand a steady decline in soil fertility levels inmany areas. It i s estimated that between 39 and 51 percent o f the agricultural land in the country i s already moderately or severely degraded (Government o f Rwanda and FAOSTAT data). Farming systems and croppingpatterns 36. Agriculture in Rwanda i s dominated by small-scale, subsistence-oriented family farming units.Approximately 1.4 million rural households depend on agriculture as their main livelihood source. These households produce a range of food crops (cereals, roots and tubers, bananas, and vegetables), with approximately 66 percent o f production destined for home consumption. The e remaining 34 percent o f production finds its way to local markets. Crops are produced mainly 14 under rainfed conditions using mostly family labor and few or no purchased inputs (improved seed, fertilizer, and crop protection chemicals). Approximately 60 percent o f households also keep animals for milk, eggs, and meat. These animals, are mainly local breeds, are raised using traditional low-input extensive grazing methods, although in the case o f cattle the dwindling availability o fpasture landi s causinga shift to confined feeding with cut fodder supplementedby grain and/or roots andtubers. A minority o fruralhouseholdsalso produces export crops, the most important o f which are coffee (cultivated by approximately a one third percent o f all rural households) andtea (cultivated by less than 1percent o f all rural households). 37. Data on area, yield, and production o f major food and cash crops are shown in Table 6 and Figure 8. Food crops dominate the area planted, reflecting the subsistence orientation Rwandan agnculture. In 2005, roots and tubers accounted for the largest share o f total cropped area (26 percent), followed by bananas (22 percent), cereals (21 percent), pulses including beans and peas (21 percent), h i t s and vegetables (5 percent), and oilseeds (3 percent). Traditional export crops accounted for only 3 percent o f total cropped area, including coffee (2 percent), tea (1percent), andpyrethrum(4percent). Figure 8. Area shares of principal crops, Rwanda, 2005 Pulses Oilseeds 3% Fruits& Vegetables 5% EXDOI~ Crops 3% Cereals Roots&Tubers 21% Source: Calculatedfrom MINAGRIandMINECOFINdata, 2006. 38. Food crop yields in Rwanda are low by global standards. When yields in Rwanda are compared to yields inother countries, it is clear that Rwandan agriculture i s facing a productivity crisis. Using FAOSTAT data to make comparisons more consistent, average yields achieved in Rwanda for most major food crops lag behind the sub-Saharan Africa averages (Table 7). The picture becomes even bleaker when yields achieved in Rwanda are compared to the averages in neighboring countries such as Burundi and Uganda, which arguably provide better standards for comparison becausethey feature more similar production environments. Inthe caseoftea, householdsmayproduce tea onownedor rentedland, orthey maygenerateincome by selling their labor to the teaplantations. 15 U e 4 I F d3 5 e L I F 2 ce U a e I F d3 U a : I 5; d3 U : 4 - U F ce 4 E 00 2 3 0 m hl d 9 9 9 3 2 3 2 3 z a 0 00 0 "! m P d t': r: t 3 3 3 3 hl m d 2 2 =t! d m t':m 3 z 0 z 3 cu 2 2 3 FJ m m m m z 3 2 2 9 c? hl d \o 3 3 2 s s 0 m 0 03 t 3 z 2 x 2 0 P 00 t': t z 3 m P 3 3 3 3 2 P 0 m 9 c? 3 3 s 2 s 3 0 3 3 2 9 d 3 P 'c! 3 0 0 0 'c! m 3 00 9 3 m 0 09 3 x 0 2 4 B 3 5 e, 5 0 m Em B 0 m"23 Agricultural production trends 39. The instability associated with the recent period o f social conflict complicates analysis o f crop productiontrends, for at least three reasons. First, inthe wake o f the genocide, the collecting and reporting o f all types o f statistics including crop production statistics was severely disrupted for about a decade, so the data needed for rigorous quantitative analysis are often unavailable (see Annex 1). Second, because the genocide so severely disrupted most types o f economic activity throughout the country, including agriculture, many traditional analytical procedures based on the use o f time series data to detect trends through time are frustrated by huge discontinuities inthe data. Third, while it i s sometimes possible to avoid the two problems mentioned above by restricting one's focus to the post-genocide period when data collection activities resumed, many spheres o f economic activity including agriculture have experienced extraordinarily rapid growth in the post-conflict period, and it is unrealistic to assume that this "catching up" effect will continue indefinitely. Crovs 40. Food crop productiontrends inRwanda are shown in Table 8. Indices o fproduction, area planted, and yields o f major food crop groups are depicted inFigure 9, Figure 10 and Figure 11. Duringthe past five years (2001-2005), growth rates have varied considerably between different groups o f food staples, as have the sources o f growth. 41. Rapid growth inthe production o f fruits and vegetables averaging 49.7 percent per year has been driven mainly by improved yields, which have increased by about 32 percent per year. Area expansion has also contributed to growth in the fruits and vegetables sub-sector; the area planted to fruits and vegetables has expanded at a rate o f 13.4 percent per year, albeit from a small base. Fruitsand vegetables now occupy 5 percent o f Rwanda's total cultivable area. 42. Production o f cereals has increased at a rate o f 8.2% per year, fueled by expansion in cultivated area as well as by increases in yields. The overall growth rate and the relative contribution o f these two sources o f growth have varied between crops. Production o f rice and wheat has increased very rapidly, fueled mainly by expansion incultivated area. The area planted to rice and wheat has more than doubled since 2001, although from very small bases. Incontrast, production o f sorghum and maize has increased at a more modest pace, driven mainly by area growth inthe case o f sorghum and mainly by yield growth inthe case o f maize. Cereals currently occupy just over 2lpercent o f Rwanda's total cultivable area. 43. Significant growth has been recorded also inthe banana sub-sector. The average annual growth o f 5.9 percent recorded since 2001 has come entirely from growth in yields, as area planted to banana has remained unchanged. The area planted to banana remains significant, however, as it represents about 22.3 percent o fRwanda's total cultivable area. 44. Production o f oilseeds has increased at 4.5 percent per year since 2001. The area planted to oilseeds has increased at 7.3 percent per year, but increases inarea plantedhave been offset by declining yields. Performance has differed significantly between different crops, however: in groundnuts, both the area cultivated and yields have recorded strong positive growth, but in soybeans, expansion in cultivatedarea was more than offset by steep declines inyields which fell at an average annual rate o f 7.6 percent. 18 c? 3 3 'm9 0 ' 0 3 9 Figure 9. Productiontrends for principalfood crops, Rwanda, 2001-05 / 500 -8 450 400 r Fruit. and Vegetables f I1 350 300 0 E 250 1 200 150 8 E 100 Legumes Rootsand Tubers 50 I 2000 2001 2002 2003 2004 2005 Source:MINAGRI data and Aertssen et. al., 2006. Figure 10. Area trends for principal food crops, Rwanda, 2001-05 _ _ _ ~ ~ Frultsand Vegetables / "1 aananar Legumes ROO& andTubers 2000 2001 2002 2003 2004 2005 Source:MINAGRI data and Aertssen et. al., 2006. Figure 11. Yield trends for principal food crops, Rwanda, 2001-05 300 f 250 ii I 200 : 150 8 100 0 60 I 2wo 2w1 2wz 2003 2004 2005 Source: MINAGRI data and Aertssen et. al., 2006. 20 45. In contrast to the previously mentioned staple food groups, all of which experienced production growth during the 2001-05 period, production o f roots and tubers remained more or less stagnant, exhibiting an overall decline in production averaging 1.1 percent per year. Within the roots and tubers group, production o f Irishpotatoes increased as a result o f expanding area combined with rising yields, as did production o f other roots and tubers (a category that includes mainly colocasse and yam). Meanwhile, production o f cassava fell modestly, and production o f sweet potato declined sharply by 8.5 percent per year on average. Some have expressed concern at the steep decline in sweet potato production, noting the importance o f sweet potato inthe diets o f the poorest households, but an alternative explanation i s that producers are shiftingland out o f sweet potato and into more profitable crops. This explanation i s also consistent with the fact that sweet potatoes were traditionally planted in many o f the marshlands that have recently been developed for irrigation and are now beingplantedto rice and other high-value crops. 46. The only group o f food staples that contracted significantly during the past five years was pulses(mainly beans butincluding also peas). Productiono fpulses declined at an average annual rate o f 5 percent, with contraction in the area planted to beans and declines in yields o f beans offsetting gains inarea plantedand yields o fpeas. The performance of the pulses group has raised concerns among some policy makers, because beans are not only a major crop by the majority o f rural households (they occupy just over 19 percent o f all cultivable land in Rwanda), but also because they an important source of protein and hence a key food security crop. 47. To summarize, since 2001 the production performance o f major food staples has been mixed. Production increases have been achieved in cereals (sorghum, maize, rice, wheat) and in oilseeds (groundnuts and soybeans). Production gains have been more modest inbananas and in roots and tubers (Irishpotato, sweet potato, cassava). Production o f legumes (beans and peas) has declined slightly. Decomposing production into its two components-area and yield-it is interesting to note that both have increased. Although it i s often said that the land frontier in Rwanda has been exhausted, cropped area continues to increase, albeit very slowly at an average annual rate o f only 1.4 percent. The continuing growth in cropped area reflects both physical expansion o f the land area devoted to cropping activities (e.g., through cropping o f previously uncultivated lands including marshlands and through conversion o f permanent pasture to crop land), as well as intensification (e.g., increased multiple-cropping on existing crop land). Area growth has been concentrated mainly inh i t s and vegetables and incereals. 48. These trends raise concerns regarding food security prospects. Production has gradually declined for four major food staples grown by many poor households for home consumption: sweet potatoes, beans, banana, and cassava. Productiono f sorghum, another important food staple consumed primarily by the poor, has been generally flat. Taken together, these five crops account for more than 70% o f consumption expenditures made by rural households and contribute as much as fourth-fifths o f the caloric intake o f many rural households. Livestock 49. Livestock production activities make a significant contribution to the agricultural economy, but this contribution i s to some extent hidden, because most livestock products are consumed at home and never enter the market. Livestock herds were generally decimated during the genocide, but they have since recovered, and numbers o f cattle, sheep, and goats are now approximately equal to pre-genocide levels (Table 9). Rwanda's livestock producers have been slow to embrace technical change, however, and productivity levels for all species remain low. 50. Cattle, which are kept inall regions o f the country, represent not only a source o f income but also serve as a store o f wealth and furthermore serve an important social function. Cattle 21 production systems inRwandaremain very traditional. Indigenous breeds predominate, especially Ankolt, and these are generally raised using traditional extensive grazing methods. Inreactionto the increasing scarcity o f land, the government i s promoting the intensification o f production, especially dairy production, through the introduction o f improved breeds and conversion to contained feeding productionmethods. Table 9. Evolutionof livestock production, Rwanda, 1994-2005 2000 2001 2002 2003 2004 2005 Increase (YO) Livestock(`000 head) Cattle 755 814,124 960,450 991,697 1,006,572 1,079,206 43% Goats 757 916,753 919,785 1,270,903 1,263,962 1,263,962 67% Sheep 233 266,539 300,600 371,766 686,837 686,837 195% Pigs 177 197,081 207,783 211,918 326,652 342,027 93% Poultry 2,043 1,277,706 1,055,644 2,432,449 2,482,124 2,943,703 44% Rabbit 339 495,290 488,629 498,401 520,057 565,696 67% Livestockproducts(kg) Milk 57,853 63,484 97,981 112,463 121,417 91.3% Meat 22,807 25,608 35,748 39,126 43,589 48,681 70.2% Fish 6,996 7,308 7,612 8,144 8,126 11.2% Eggs 1,471 920 1,015 2,432 3,402 2,452 269.8% Honey 760 819 908 1,029 35.4% TOTAL 98,080 143,175 163,073 177,563 81.O% Source: MINAGRT,RARDA and MINECOFIN, 2006. 51. Goats, sheep, pigs, and rabbits are also produced on a significant scale in Rwanda. For these species, production i s basically artisanal, with small numbers o f animals being raised at the household level using traditional extensive methods. Commercial production using improved breeds and improved feeding strategies i s still limited. 52. Poultry, mainly chickens, are raised by the majority o f rural households as a source o f eggs and meat for home consumption and occasionally also as a minor source of supplemental income. Large-scale commercial poultry operations that use improved breeds and have recently started to appear inurban areas, especially inand around Kigali. The National Hatchery located at Rubilizi near Kigali has emerged as an important source o f improved breeds sought by commercialproducers. Agricultural technology 53. Farm-level productivity will have to improve inRwanda ifthe agricultural growth targets spelled out in the PSTA are to be achieved. For productivity to improve, Rwanda's farmers will need to have better access to and make better use o f improved technology. Weak agricultural research and extension systems, low use o f improved genetics and purchased inputs,and the lack o f availability o f rural financial services all contribute to the low productivity o f Rwandan agriculture and constrain agricultural growth. 22 Agricultural research 54. Agricultural research in Rwanda is carried out mainly in the public sector. The premier agricultural research institute i s the Institut des Sciences Agronorniques du Rwanda (ISAR) located in Butare. Following the loss o f many senior scientific staff during the genocide, I S A R has gradually rebuilt its capacity, but further strengthening i s needed. A strategic plan launched in 2002 set the institute on a new course under which more basic "upstream" research was scaled back in order for increased emphasis to be placed on more applied "downstream" research featuring the adaptation o f importedtechnologies through the use o f farmer-participatory on-farm trials. Lack o f trained scientists remains a problem, however, and I S A R continues to operate with the help o f foreign experts while a cadre o f Rwandan scientists undergoes graduate training within the country and outside. Agricultural extension 55. Agricultural extension has long been identified as a priority in Rwanda. In the past, extension was always considered the responsibility o f the national government, and responsibility for delivery o f extension services was assigned to MINAGRI. The extension model used by MINAGRIwas very top-down and linear. Technical recommendations were formulated centrally based on research results coming out o f I S A R and other research institutes, and these recommendations were conveyed by extension agents to farmers, who were expected to adopt them exactly as formulated. The obvious ineffectiveness o f the extension service led to the complete scrapping o f the national extension service in 1998 and the devolution o f responsibilities for technology transfer activities to the local level. In the absence o f any meaningful follow-up, public extension services basically disappeared, to be replaced by the current patchwork o f project-funded extension initiatives whose coverage i s far from complete and whose technical quality is highly variable. Except for coffee and tea producers, many o f whom report regular contact with extension agents working through OCIR-CafC and OCIR-ThC, most Rwandan farmers today have very limitedcontact with extension agents. Imuroved uroductioninputs 56. Agricultural productivity has stagnated in Rwanda in part because the use o f improved production inputs i s extremely low. The 1999 EICV survey revealed that less than 7% o f rural households reported using improved inputs of any kind (Table I O ) . Use o f improved inputs i s strongly correlated with income, with households in the highest income quintile reporting almost seven times as much use as households inthe lowest quintile. Increasing agriculturalproductivity will require increasing the use o f inputs, especially improved seeds, chemical fertilizer, and pesticides in the crop sub-sectors, and improved breeds and veterinary services in the livestock sub-sectors. Marked differences in the use o f inputs between individual provinces are mainly attributable to differences inlocal farming systems and cropping patterns, but they may also have been resulted from inconsistent coverage o f agricultural development projects and targeted input promotion schemes. Fertilizer 57. Many o f Rwanda's soils are naturally fertile, but their fertility has been severely eroded by poor land and soil management practices. These have contributed to soil nutrient losses that are among the highest on the highest on the continent. The International Fertilizer Development Center estimates that soil nutrients are being lost inRwanda at a rate o f 77 kgha per year (Table II).Adoption of improved soil fertility management practices will be needed to reverse the current highrates o fnutrient losses and sustain higher levels o fproductivity andproduction. 23 Table 10. Use of improvedinputsbyprovinceandincomequintile,Rwanda Proportionof householdsreportingimprovedinput use(YO) 1st 2nd 3rd 4th 5th quintile quintile quintile quintile quintile Total Butare 0.8 5.0 7.8 11.8 19.4 7.0 Byumba 2.9 1.7 5.7 5.1 20.2 6.9 Cyangugu 2.7 4.3 7.7 19.7 13.6 9.0 Gikongoro 10.2 15.1 26.2 33.4 40.4 20.5 Gisenyi 0.0 0.7 4.3 15.0 24.8 9.8 Gitarama 1.1 5.9 12.2 4.0 9.2 6.6 Kibungo 0.0 0.0 0.0 3.3 4.0 1.8 Kibuye 1.8 0.0 2.4 1.o 3.2 1.5 Kigali Ngali 0.0 1.8 2.7 8.9 18.4 4.3 Kigali Ville 0.0 0.0 0.0 8.3 7.4 5.6 Ruhengeri 1.3 6.8 8.3 5.6 15.5 6.3 Umutara 0.0 0.0 0.0 4.0 2.8 1.7 Total 2.1 4.0 7.0 9.2 14.3 6.9 Source: IFPRI calculationsusingEICV data, 2006. Table 11.Estimatedsoil nutrientlosses inAfricancountries, 2002-04 (kglha) Moderate/ Low Medium High (lessthan 30 kg/ha/ year) (from30 to 60 kg/ha/ year) (morethan 60 kg/ha/ year) Egypt 9 Libya 33 Tanzania 61 Mauritius 15 Swaziland 37 Mauritania 63 South Africa 23 Senegal 41 Congo Republic 64 Zambia 25 Tunisia 42 Guinea 64 Morocco 27 BurkinaFaso 43 Lesotho 65 Algeria 28 Benin 44 Madagascar 65 Cameroon 44 Liberia 66 Sierra Leone 46 Uganda 66 Sudan 47 D.R. Congo 68 Togo 47 Kenya 68 Cote d'Ivoire 48 Gabon 69 Ethiopia 49 Angola 70 Mali 49 Gambia 71 Mozambique 51 Malawi 72 Zimbabwe 53 Guinea Bissau 73 Niger 56 Namibia 73 Chad 57 Burundi 77 Nigeria 57 Rwanda 77 Eritrea 58 Equatorial Guinea 83 Ghana 58 Somalia 88 Source: Henao and Baanante, 2006. 24 58. One strategy to improve soil fertility levels is to use organic fertilizer. Approximately two thirds (62 percent) o f rural households in Rwanda use organic fertilizer, especially manure, and available supplies are fully exploited. However the amount o f organic fertilizer used falls far short o f the amount neededto replace the nutrients taken off inthe form o f crops, so there is need to make up the shortfall by applying inorganic fertilizer (sometimes referred to as chemical fertilizer or mineral fertilizer). 59. Inorganic fertilizer is applied mainly to cash crops that respond rapidly to added soil nutrients, especially tea and Irish potato. Fertilizer i s also applied to rice and to other crops (a category that includes mainly horticultural crops) (Figure 12). The number o f farmers using fertilizer i s very limited, and the number of users varies considerably from year-to-year. In2001, 5 percent of agncultural households were found to be applying fertilizer (EICV 2001). By 2003, the figure had fallen to only 1.4 percent (QUIBB4 2003). Preliminary analysis of the 2005 Agricultural Survey data (Season A) indicates that fertilizer use had increased again, with approximately 14 percent of agncultural households reporting use (Table 12). Compound formulations (NPK) are the most popular (used by 7 percent o f households), followed by single- nutrient formulations such as urea and DAP (used by slightly less than 3 percent of households). A very small number of farmers (representing 0.3 percent of all agncultural households) apply lime to their plots to adjust soil pH. Most fertilizer is purchased from private inputs suppliers. Prices vary considerably, even withinindividual provinces and districts. Figure 12. Proportion of fertilizer appliedto different crops, Rwanda, 2001-03 average Other food crops (21%) Coffel Source: MINAGRIand WorldBank Report "Utilisation etAccessibilite`des Engrais Inorganiques au Rwanda,''corrected and averaged,2004. 4 Core Welfare Indicators Survey (MINECOFIN, 2003) Ministere des Finances et de la Planification Economique, Dtpartement de la Statistique (Mars 2003). ``Enqugte sur les Indicateurs de Base du Bien Etre, QUIBB:Rapport d'halyse des R6sultats." 25 Table 12. Fertilizer use, Rwanda, 2005 Season A (numbers of households) Purchased Received free Total YOof all HH NPK 117,102 2,792 118,639 7.4 DAP 33,371 1,256 33,371 2.1 Urea 37,196 5,613 41,413 2.6 Lime 4,275 0 4,275 0.3 Other 16,764 2,708 19,472 1.2 TOTAL 208,708 12,369 217,170 13.6 Total number o f HH= 1,597,264 Source: Preliminay datafrom 2005 Agricultural Survey SeasonA, MINAGRI, 2005. 60. A comprehensive review o f the experimental evidence on crop responses to fertilizer carried out by Kelly et al. (2001) concluded that the potential for profitable fertilizer use i s high (Table 13).These authors concluded that the current low usage results from constraints both on the demand side (e.g., lack o f farmer awareness, lack o f farmer purchasing power, lack o f access to output markets) and on the supply side (e.g., limited availability o f fertilizer, highcost). Table 13. Summary of experimental data on fertilizer response, selected crops Increase Increase as percent National average inyield of national yield yield (1998-01) (mt I ha) (%) (mt I ha) Maize 0.7 2.6 - 89 -330 0.79 Rice 1.0 1.7 - 57 -97 1.75 Sorghum 0.6 1.9 - 62 - 197 0.97 Potatoes 5.0 - 9.3 61 - 113 8.22 Sweet potatoes 4.2 - 7.4 74 130 - 5.72 Beans 0.4 - 1.2 93 -278 6.48 Soybeans 0.4 - 0.8 77 - 154 0.52 Source: Calculatedfrom Kelly and Murekezi, 2000. 61. The amount o f fertilizer that i s officially imported i s reported, but the official import data do not account for the significant quantities o f fertilizer that i s known to cross the border through informal channels. In 2005, the most recent year for which data are available, official fertilizer imports totaled nearly 8,000 tons, nearly double the amount that was imported just two years earlier (Table 14). Even allowing for the possibility o f significant additional informal imports, this amount falls far short o fnational fertilizer requirements. A recent Cabinet Paper on fertilizer estimates that national fertilizer needs for 2006/07 will total nearly 60,000 MT (MINAGRI 2006). With support from the Clinton Foundation, a new program was launched in 2006 to encourage 26 increased fertilizer use by improving availability and affordability at the farm level. Under the new program, MINAGRI i s working with the Clinton Foundation in an effort to achieve economies o f scale in fertilizer procurement and distribution. The program i s still very new, so it i s too early to judge its performance. Early indications are that the amount o f fertilizer imported into the country in2007 will increase to around 14,000-16,000 tons, but it is not clear whether the hoped-for economies o f scale can be achieved at these still relatively modest volumes. In addition, to date most o f the fertilizer distributed has been handled by public agencies, with relatively modest participation by the private sector. A challenge will be to attract increased participation by private distributors, whose involvement will be needed to permit the eventual withdrawal o f the public agencies. Experience from many other countries inAfrica and elsewhere suggests that ensuring the effective participation o f private input distributors i s one o f the key guiding principles that must be followed if fertilizer promotion activities are to be sustainable over the longer term (see Box 2). Table 14. Fertilizer imports, Rwanda, 2001-05 Fertilizer Imports 1991 2001 2002 2003 2004 2005 Volume (T) 6,952 4,172 3,575 7,855 Value 1,488 1,374 3,045 Kg/ha 5.5 2.84 2.43 5.34 3.30 2.83 6.21 Note: IFDC data are calculated assuming 1.26 million hectares of arable land, whereas MINECOFIN data are calculated on the basis of 1.47 million hectares of arable land. Figures in italics show the estimates based on 1.26 million hectares. Source: IFDC and MINECOFIN and BNR Oficial import statistics, 2005. 62. The recent modest gains in fertilizer use are encouraging, but much more progress i s needed. Table 15 highlights fertilizer use trends in 30 African countries for which data are available. Duringthe period 1996-2002 (post-structural adjustment, when fertilizer subsidies had been reduced in most countries), fertilizer use in Rwanda increased by more than 30 percent. However this increase was from an initial low base, and Rwanda remains in the group o f countries with an overall low intensity o f fertilizer use (<25 kgha). 27 Table 15. Fertilizer use intensity, selected Africa countries Mean fertilizer use intensity in 1996-2002 (kg/ha) Fertilizer use intensity (1996-2002) and percentageincrease(mean 1996-2002 / mean 1990-95) Small increase(e30%) Large increase (>30%) Angola (0.7, -69%) Benin (17.6, +76%) Burkina Faso (5.9, -28%) Botswana (11.8, +294%) Burundi(2.3, -6%) Ethiopia (14.4, +71%) DRC (0.5, -47%) Cameroon (5.9, +77%) Gambia (5.2, +15%) Chad (4.3, +93%) Guinea (2.0, -4%) Cote d'Ivoire (11.8, +53%) Low intensity (<25 kg/ha) Madagascar (2.9, -8%) Ghana (3.6, +68% M a l i (9.0, +7%) Lesotho (23.2, +35%) Mauritania (4.0, -64%) Mozambique (3.2, +142%) Niger (0.9, +5%) Rwanda (1.8, +89%) Nigeria (5.6, -73%) Senegal (13.2, +67%) Tanzania (4.8, -47%) Togo (7.0, +30%) Zambia (8.4, -34%) Uganda (0.6, +237%) Malawi (30.8. +9%) Kenya (31.8, +33%) High intensity , I (>25 &/ha) Swaziland (30.5, -40%) Zimbabwe (48.3, +9%) Note: Fertilizer use intensity i s defined as kilograms o f fertilizer applied per hectare cultivated to annual and permanent crops. Increase in fertilizer use intensity is defined as the percentage increase in mean fertilizer use intensity between 1990-95 and 1996-2002. Numbers in parentheses are mean fertilizer use intensity for 1996-2002 and the percentage increase infertilizer use intensity, as defined above. Source: Crawford et al., 2006, adaptedfrom Ariga, Jayne, and Nyoro, 2006. 28 Box 2. Promoting efficient and sustainable fertilizer use: Lessons from Africa Crop yields in Rwanda are low partly because o f soil fertility problems, which can be addressed by increasing use o f fertilizer. Fertilizer use i s unlikely to increase, however, unless the underlying structural problems affecting fertilizer demand and supply can be addressed. A recent World Bank review o f fertilizer promotion schemes in Africa identified 10 guiding principles that can be used to design and implement public interventions to support efficient and sustainable growth infertilizer use. 1.Promote fertilizer as part of a wider sector strategy: Interventionsdesigned to promote increased fertilizer use should be implemented as part o f a wider sector strategy that recognizes the importance of complementary inputs, reliable output markets, and appropriate sequencing o f activities. 2. Favor market-based solutions: Interventions designed to promote increased fertilizer use should be designed to support market development and not undermine incentives for private investment. 3. Promote competition: While the private sector should be in the driver's seat, markets should be competitive to ensure the lowest cost and best quality service. 4. Pay attention to demand: Farmers' effective demand, shaped by the financial profitability o f fertilizer use, should be the ultimate driving force o f input supply systems and the foundation o f a sustainable fertilizer promotion strategy. 5. Insist on economic efficiency: Interventions to promote the use o f fertilizer should be carried out only where it i s economically efficient to use fertilizer. 6. Empower farmers: Interventions designed to promote increased use o f fertilizer should empower farmers to make their own decisions on the most appropriate way to improve soil fertility intheir particular context. 7. Devise an exit strategy: Public sector interventions designed to promote increased use o f fertilizer should be designed with a clear exit strategy, except for a few long-run public good functions, such as, market regulation, infrastructure development, and R&D for natural resourcemanagement. 8. Pursue regional integration: Countries should seek regional integration and harmonization o f fertilizer policies to reap economies o f size, which are especially important inaregionsuchas Africa withmany smallcountries. 9. Ensure sustainability: Interventions to promote increased use o f fertilizer should be economically, institutionally, and environmentally sustainable. 10. Promote pro-poor growth: Public interventions designed to promote increased use o f fertilizer should aim to promote pro-poor growth. In exceptional circumstances, poverty reduction andor food security objectives may be given precedence over efficiency and sustainability goals, if it can be determined that fertilizer interventions are the most cost- effective way o f addressing these problems. 29 Seed 63. To derive maximumbenefits from increased use o f fertilizer, farmers must grow modem varieties that have the capacity to respond to increased fertility levels. The delivery vehicle for modem varieties i s improved seed. Currently few farmers in Rwanda plant improved seed. In 2005, only 12 percent o f households reported using improved seeds, covering only 2 percent of cultivated land. According to preliminary analysis o f the Season A results from the 2005 Agricultural Survey, 90 percent o f seed for food crops i s farmer-saved seed produced on the farm. 64. Responsibility for improved seed multiplication activities was formerly held by a parastatal organization, the National Seed Service, which has now been integrated into RADA. At about the same time, a new seed law was enacted with the goal o f encouraging increased private investment in the seed industry. The effort to attract increased private investment seems to have elicited limited response, so the National Seed service remains the leading producer o f improved seed in the country. Recently there have been some initiatives to distribute improved seeds o f maize, sorghum, rice, wheat, and beans, as well as improved virus-resistant planting materials for potato, sweet potato, cassava, and banana. The amount o f seed produced remains small, however, and it covers only a small fraction o f potentialneeds (Table 16). Table 16. Productionof improvedseed (kg) and demandcoverage (YO),200-2005 2001 2002 2003 2004 2005 Prod Cov Prod Cov Prod Cov Prod Cov Prod Cov (kg) ("/.I(kg) ("/.I(kg) ("/.I(kg) ("/.I(kg) ("/.I Sorghum 495 7.2 64 0.9 58 0.9 367 5.4 206 3 Maize 1,292 11.3 363 3.2 1,228 10.7 1,324 11.6 1,127 9.9 Wheat 111 1.0 54 0.5 25 0.2 0 0 50 0.5 Beans 432 0.5 856 0.9 707 0.8 503 0.6 521 0.6 Soybean 379 4.0 284 3.0 345 3.7 172 1.8 80 0.9 Potato 1,036 0.1 1,020 0.1 1,258 0.1 617 0.1 1,172 0.1 Note : Prod= Quantity o f seed produced(kg); Cov = Proportion o f estimated total seed requirement (%). Source: AFSR Study "Demande Solvable et Rentabilitd de la Production SemenciGre,'' 2006. Pesticide 65. Consistent with the low use o f seed and fertilizer, use o f pesticides i s also low. Preliminary analysis o f data from the 2005 Agricultural Survey (Season A) indicates that pesticides are used in approximately 12 percent o f households (Table 17).Pesticides are applied mainly to coffee and tea, bananas, potatoes, and horticultural crops (fruits and vegetables). The low use o f pesticides can be traced back to demand-side as well as supply-side factors. Weak demand for pesticides results mainly from farmers' poor knowledge o f pest and disease control methods, which in turn i s compounded by the lack o f research being done on chemical pest control practices. The limited availability o f pesticides in the market stems from the difficulty o f procuringpesticides, most o f which are imported. 30 Machinely and tools 66. Agriculture in Rwanda i s generally non-mechanized. Use o f machinery or even draught animals to carry out cropping operations i s very limited. Most cropping operations are carried out manually using hand tools (e.g., hoes, shovels, rakes). Labor productivity i s consequently very low and broadly comparable withproductivity inother Afkican countries with lower agro-climatic potential (Figure 13). Because the recent social conflict and subsequent recovery were characterized by several waves of migration out o frural areas and then back in, labor productivity trends are difficult to measure. WDI data suggest that labor productivity has remained flat during the past two decades. Table 17. Pesticideuse, Rwanda, 2005 SeasonA (numbers of households) Purchased Received free Total % of all HH Dithane 115,412 1,574 116,986 7.3 Dursban 7,182 0 7,182 0.4 Sumicombi 36.839 4,73 1 4,768 0.3 Oxychlorure 7,024 0 7,024 0.4 Ridomil Y'Ifu 1,604 0 1,604 0.1 RidomilY'Amazi 764 0 764 0.05 Other 54,571 4,711 59,282 3.7 Total 186,594 11,016 197,610 12.4 Source: Prelirninaly dataporn 2005 Agricultural Survey Season A, MINAGRI, 2005. Figure 13. Labor productivityinagriculture,Rwandavs. selectedAfrican countries,2005 g 800 , Source: WorldBank WorldDevelopmentIndicators and FAOSTAT data, 2005. 31 Rural financial services 67. Rwanda's financial sector i s poorly developed and weak. Access to financial services i s particularly limited for rural households. Two government-supported banks, the Banque Rwandaise deDdveloppement (BRD) and the Union des Banques Populaires du Rwanda (UBPR) offer financial services to agricultural customers, but the volume o f rural lending by these banks i s very limited. In2003, total bank loans in the apcultural sector amounted to about RWF 2.25 billion, representing less than 2% o f all bank loans. Less than one-quarter of rural households receive credit from any type of organization (EICV 2001). 68. With use o f formal credit being very limited, rural borrowers have responded by turning to a range o f semi-formal and informal lending sources. These include savings and credit associations, government- and donor-funded project financing instruments, and local money lenders. Nearly two-thirds of rural households rely on personal household savings for financing, and an additional 10to 12percent rely on loans from relatives (EICV 2001). 69. Because collateral i s needed to obtain formal credit, households in the highest income quintile obtain almost twice the amount o f loans as households in the lowest quintile. Poor farmers who are unable to meet the collateral requirements needed to secure credit through the formal banking sector are more likely to gain access credit to through membership o f a producer association or cooperative. Broadening and deepening the outreach o f the micro-finance sector will be important for improving access to finance and other inputs by rural households and farmers Agricultural marketingsystems Food stavles 70. Domestic food markets in Rwanda are generally underdeveloped. Marketing chains for locally produced food are informal and often fragmented, with produce typically changing hands several times as it moves from the farm gate to the final consumer. Farmers sell produce to rural assemblers in the field, at the farm gate, or along the road, or they sell to rural traders located in local assembly markets. Most rural assemblers operate at a very small scale. Lacking mechanized transport, they may arrange for the use of wheelbarrows or bicycles to transport produce from the farm gate to collection points from where it can be sold on to rural traders. Rural traders typically load produce onto small trucks and transport it to urban wholesalers. Urban wholesalers in turn supplyurbanretailers, who break loads downinto small lots for resale from open-air market stalls or small neighborhood shops. 71. Expressed as a proportion o f the final consumer price, marketing margins for domestically produced food crops are highinRwanda compared to other countries inAfrica. The highmarketing margins reflect not so much a lack of competition among intermediaries as they do the presence o f high real costs, especially transport costs. Marketing margins on food crops have come under pressure in recent years, reflecting the effects o f increased competition. The increased competition has come from greater numbers o f traders operating at all stages along the marketing chain, but especially in rural assembly and retailing, where barriers to entry are low (Goessens and Rwamasirabo 2006). The recent spread o f low-cost mobile telephone service to rural areas has exerted further pressure on marketing margins by facilitating flows o f market informationand helpingmatch demand and supply. 72. Despite their informal character, domestic marketing chains for locally produced crops and livestock products appear to be relatively efficient. High and essentially unpredictable inter- annual price fluctuations discourage investment in long-term storage, but intra-annual (seasonal) 32 price fluctuations for most crops range from 25-30 per cent, which, while high, i s reasonable considering that many food staples are subject to significant storage losses. Spatial market integration as indicated by price correlations between major markets is good (Loveridge 1989) which i s not surprising considering that Rwanda i s a small country with relatively goodroad links between major urbanpopulation centers. Exvort crovs 73. Marketing systems for traditional export crops are distinct and specialized. 74. Rwanda's coffee marketing system has undergone a series o f reforms that have succeeded in converting what was once a government controlled system into a fully privatized system. The Ofice des Cultures Industrielles au Rwanda-Cafi (OCIR-Cafe!), which was established as a parastatal marketing board to manage the coffee sub-sector, formerly controlled all coffee marketing activities: it provided production inputs, set producer prices, carried out processing, and managed exports through Rwandex, a mixed public-private corporation under majority government control. In 1995, Rwandex lost its monopoly position as private exporters were allowed to enter the market, and following a decade o f progressive privatization, the company was completely privatized in February 2006 when shares held by the government were sold to a consortium o f private shareholders. Despite the privatization o f the coffee marketing system, OCIR-Cafd continues to provide important support to the sub-sector through a wide range o f activities designed to increase production, improve quality, and increase export revenues. 75. Privatization o f Rwanda's coffee marketing system has been accompanied by a shift away from the production o f semi-washed common grade coffee toward fully-washed specialty coffees. This shift in strategy has induced changes in production practices and organizational structures all along the coffee supply chain. With regard to production practices, the most noteworthy changes have occurred at the processing stage, where there has been a proliferation o f privately owned, professionally managed washing stations capable o f producing fully-washed specialty-grade coffees that command much higher prices in global markets than the semi- washed, bulk-grade coffees that were exported in the past. With regard to organizational structures, an important recent development has been the launching of a number o f projects linking growers, processors, and exporters into vertically integrated operations that allow for more efficient management at every stage o f the supply chain and better control over the quality o f the final product. The shift to quality has brought promising results. Prices fetched by fully- washed Rwandan specialty coffees in international markets are on average more than double prices fetched by semi-washed C-grade coffees. Higher export prices received for fully-washed specialty coffees have been transmitted back along the value chain to benefit not only exporters butalso processors and growers. 76. Unlike the coffee marketing system, Rwanda's tea marketing system continues to be dominated by state-owned companies. Incontrast to coffee, which i s grown by a large number o f small-scale private growers, tea i s cultivated mainly on large plantations, which are owned and managed by about a dozen tea "factories" that process green tea into black tea. The factories supplement the tea cultivated on their own (state-owned) land with relatively small amounts o f tea producedby tea cooperatives and private growers. The marketing chain for tea i s necessarily short, reflecting the extreme perishability o f green tea, which must be delivered to the factory within hours o fbeingpicked so as not to suffer a significantloss inquality. Tea is usually picked early in the early morning, when temperatures are cool, and i s transported in baskets to local collection points. After being weighed, it i s loaded onto trucks for delivery to the factory. 33 Following a brief drying period, it i s chopped, fermented, dried down, cut, and packaged for export. Tea exported from Rwanda i s mostly transported by road to Mombasa, where it i s sold at auction. Rwandan tea i s generally o f very high quality, and it fetches premium prices at the Mombasa auction. 77. The Office des Cultures Industrielles au Rwanda--Tht! (OCIR--The?, which was established as a parastatal marketing board to manage the tea sub-sector, has achieved limited progress in privatizing the tea sub-sector. Privatization has been slowed by two main factors: reluctance on the part o f the government to cede control over an asset that has long been a reliable generator o f revenue and foreign exchange, and reluctance on the part o f private investors to acquire tea factories that following years o f underinvestment and inadequate maintenance no longer are able to operate efficiently. The low level o f operating efficiency not only raises productioncosts but also severely affects the quality o fthe final product. 78. Rwanda's pyrethrum marketing system i s extremely localized. Approximately 25,000 farmers cultivate pyrethrum on around 3,200 ha concentrated within a small radius around the country's single pyrethrum processing plant located outside Ruhangeri. Most pyrethrum growers belong to a production cooperative, and most o f the cooperatives in turn are organized into a federation o f cooperatives, which coordinates production and marketing activities. In the past, pyrethrum growers marketed mainly fresh-cut flowers, but most have adopted simple farm-level solar drying technologies and now sell mainly dried flowers, which are less perishable and much less costly to transport. 79. The pyrethrumprocessingplant, which was formerly owned and operated by a parastatal under the management o f the Office du PyrBthre du Rwanda (OPYRWA), was heavily damaged duringthe war, and it remainedout of commission for about a decade. Pyrethrumproduction fell sharply during this period, and most o f the crop was exported to processing facilities in Kenya and Uganda. In 2004, the plant was taken over by a group o f private investors operating as the Socit!tc! du PyrBthre de Rwanda (SOPYRWA). SOPYRWA has embarked upon an ambitious scheme to renovate, expand, and modernize the plant. A key element o f SOPYRWA's strategy i s to migrate from producing the crude pyrethrum extracts that were produced in the past and for which only a limited market exists to producing more highly refined pyrethrum distillates that can be sold to a muchmore wider range o f global buyers. 80. Rwanda's marketing system for hides and skins i s localized and in general, i s not well organized. Hides and skins o f domestic ruminants (cattle, sheep, and goats) are collected by individual tanneries from nearby livestock producers, who are encouragedto deliver them to local collection centers. Poor post-slaughter treatment attributable to a lack o f specialized equipment, low levels o f human capacity, and deficient coordination along the value chain results in severe degradation of hides and skins, which often receive significant quality discounts in international markets. Hides and skins exports currently amount to approximately 2,000 tons per year, over 90% o f which i s low-value dry leather and less than 10% o f which i s higher-value wet blue leather. The leading destinations for Rwanda's hides and shns exports are Hong Kong, Pakistan, Kenya, and Italy. Policies affecting agriculture Agricultural volicies 81. The basic agricultural policy objectives o f the Government o f Rwanda are enshrined in the Vision 2020 document and the first Poverty Reduction Strategy Paper (PRSP1). These documents set a target o f 5-8% annual growth for the agricultural sector over the short and 34 medium term. Inrecent years, this target has not been met: between 2001 and 2005, agricultural sector growth in Rwanda averaged 4.2%. This rate o f growth falls short o f the 6% level Comprehensive Africa Agriculture Development Program (CAADP) target to which all African Heads o f State committed under the New Economic Partnership for Africa (NEPAD). Also, as we shall see, it falls short o f the level neededto achieve the country's development objectives. 82. In an effort to improve the performance of the agricultural sector and strengthen its contribution to the national policy goals o f stimulating growth and reducing poverty, the Government o f Rwanda recently took steps to better align its agricultural policy and the associated implementation strategy with the objectives described in PRSPl and Vision 2020. In 2004, it formulated a National Agricultural Policy (NAP)and later developed a Strategic Plan for Agricultural Transformation (PSTA) to operationalize the NAP.The PSTA was developed using a participatory methodology which included a series o f thematic analytical studies' and nationwide stakeholder consultations. 83. Rwanda's agricultural policy and strategy as embodied in the NAP and the PSTA are aligned around two basic axes: (i) achieving productivity increases in agriculture and livestock production to allow domestic agtlculture to remain competitive in the region and compared to imported food staples; and (ii) developing export crops. With respect to crops and livestock products destined for domestic consumption, the goal i s to foster the emergence in the short to medium term o f competitive mixed farms that integrate improved crop and livestock production. With respect to crops and livestock markets destined for international markets, the goal is to enhance Rwanda's competitiveness inthe medium to long term by exploiting potential efficiency gains at all stages o f the value chain, i.e., at the farm level and beyond. 84. The objective o f the PSTA is to make a "Sustainable contribution to poverty alleviation and economic growth through improved productivity o f production factors, creation o f added value, diversification o f income-generating opportunities, and the preservation and conservation o f natural resources" (PSTA logframe). The underlying rationale o f the PSTA i s to promote the selection o f priority crops; regionalization o f export and cash crops to reduce production costs and optimize comparative advantages; as well as private sector partnerships for agncultural transformation. 85. Ten strategic axes underpin the PSTA: (i)Diversification and intensification of crop and animal resources development; (ii) Diversification o f sources o f incomes and rural employments; (iii)Market-oriented production and integration o f agricultural economy within the national and regional economic context; (iv) Sustainable natural resources management, particularly o f water and soils; (v) Strengthening o f organizational and technical capacities of producers and other rural professional organizations; (vi) Strengthening o f capacities o f proximity service providers, privatization and private sector promotion; (vii) Creation o f a conducive and effective institutional framework for promotion o f producers' professionalization and agricultural transformation process; (viii) Creation o f a conducive environment, to encourage productive investments to promote entrepreneurship and employment development in agribusiness; (ix) Redefining the role o f MINAGRI and the restructuring o f its actions towards the Sector-Wide Thematic studies undertaken: (i) security and vulnerability; (ii) and animal production Food Crop systems; (iii)Integration of crop and livestock production systems; (iv) Soil and water management; (v) Storage, processing and creation of value added; (vi) Commodity chains and specialization; (vii) Entrepreneurship, quality control and competitiveness; (viii) Domestic and international agricultural markets; (ix) Private sector investments; (x) Agricultural credit; (xi) Producer capacities and Farmer organizations; (xii) Agricultural research and extension. 35 programApproach (SWAP) within the decentralization framework and (x) Promotion o f a gender approach and reduction o f vulnerability o f underprivileged groups. 86. Inline withthis analysis, four broadprogramareas and 17 sub-programs were developed as the framework for the PSTA (Table 18).The program and sub-program structure embodied in the PSTA i s now also reflectedinthe MTEFat the nationallevel and fkom 2007, at district level. Table 18. PSTAprograms and sub-programs mentof Sustainable nSystems SP11: SustainableManagement o fNaturalResources andWater and Soil Preservation SP12: Integrated System o f Breeding SP13:Marshland Development SP14: Development o f Irrigation SP15: Supply andUse o f Fertilizer andMechanization SP16: Food Security and Vulnerability Management Program2: Supportto the Pro alizationofProducers SP21: Promotion o f Farmers' Organizations and Capacity Building o fProducers SP22: Restructuring Services inProximity to Producers andRuralInnovation SP23: Promotion o f Research for Development SP24: Rural Financial Systems and pment ofAgricultural Loans Program3: Promotionof Co SP31: Creation o f an Enviro SP32: Promotion and Development o f Specialist Crops SP33: Transformation and Competitiveness o f Agricultural and Animal Products SP34: Rural Support Infrastructures Program4: InstitutionalDevelop SP41:Management Support SP42: ICT Development and Coordinationinthe Agricultural Sector SP43: Planning, Coordination, Monitoring and Evaluationo f the Agricultural Sector Trade uolicies 87. Rwanda's agricultural sector i s also influenced by trade policies. Trade can and indeed must play a key role ifthe nation's ambitious growth and poverty reduction goals are to be met. Currently annual exports per capita in Rwanda total around US$l8, compared to an average o f US$145 for all o f sub-Saharan Africa (World Bank 2005). Thus there i s considerable scope for increasing exports and contributing to increasedincomes andpoverty reduction. 88. Inthe short- to medium-term, increased exports will come mainly from the agricultural sector, because the economy remains predominantly agrarian, and because the industrial sector i s still very small. The government i s pursuing a two-pronged strategy to increase agricultural 36 exports. First, it i s implementing measures designed to increase the competitiveness in international markets o f traditional export commodities, particularly coffee and tea, but also to a lesser extent pyrethrum and hides and skins. These measures range from improving production and productivity, enhancing quality, and lowering barriers that raise the costs o f trade and limit linkages with external markets. Second, it i s working to identify non-traditional export crops that can profitably be pursued by Rwandanproducers, especially high-value products targeted at niche markets. In this respect, the government has identified a number o f promising products, commissioned feasibility studies designed to identify investment opportunities, provided financial support to private entrepreneurs interested inundertaking pilot products, and built capacity inthe support o f trade and export diversification. 89. Trade policies take on significance for the agricultural sector not only by affecting the competitiveness o f agricultural exports but also by influencing the ability o f Rwandan producers to compete with food imports. Rwanda's status as a landlocked-country with high internal transport costs confers a certain degree o f natural protection, because imported foods are expensive by the time they reach Rwandan consumers. Rwandan farmers have benefited from this natural protection, but one consequence is that they are high-cost producers within the region, Meanwhile, domestic demand for food has outstripped supply, and Rwanda has become a structurally food-deficit country. Inan effort to reduce food imports and raise the level o fnational food self-sufficiency, the government maintains a policy o f import substitution and market protection. Localproduction o f food crops i s encouraged, and food imports are taxed. As a result, the prices o f many domestically produced crops and livestock products are higher than in neighboringUganda and Tanzania. 90. The current high levels o f domestic prices benefit Rwandan farmers, because they are able to sell into local markets even with low productivity levels. The situation could change, however, if the tariff barriers that help to sustain the current high domestic prices were to disappear. This could happen following Rwanda's integration in the COMESA free trade zone, especially if Uganda and Tanzania were to become members. Their future integration would likely put considerable pressure on domestic prices o f many crop and livestockproducts. 3. AGRICULTURALGROWTHDRIVERS Growth drivers:Demandprospects 91. Rwanda's strategy to stimulate increased agricultural growth, as articulated in the PSTA, focuses on raising agricultural productivity and increasing production through a series o f interventions directed at the supply side, such as strengthening research and extension, improving input distribution, building capacity among farmers' organizations, and improving export competitiveness. These measures are undeniably important. Yet agricultural growth cannot be sustained indefinitely with a supply-pushstrategy. Over the longer tern, it will not be possible to sustain increasedproduction o f food and cash crops unless the increased production finds a ready market. So where will future demand come from for commodities producedinRwanda? 92. Three main sources o f demand exist for Rwandan agricultural products: (1) domestic markets, (2) regional markets, and (3) international markets. Domesticmarkets 93. Demand for crops and livestock products produced in Rwanda currently i s driven by domestic food consumption. The most complete information about domestic food consumption 37 patterns comes from the household expenditure data collected under the EICV carried out in 1999. As in most countries, diets o f rural and urban households differ significantly in Rwanda (Table 19).This reflects not only differences in lifestyle-related tastes and preferences, but also the fact that a large proportion of the food consumed by rural households i s produced by the household (Table20). Table 19. Compositionof ruralandurban diets, Rwanda, 2000-01 Urban Rural consumption consumption 2000/01 2000/01 (kg/person/year) (kg/person/year) Irishpotato 124.6 68 Rice 23.0 2.5 Bread 16.6 1.o< Cooking banana 35.9 40.0 Sorghum 6.2 ... Maize 2.9 6.4 Manioc 21.5 30.0 Beans 48.0 55.0 Sweet potato 11.6 100.0 Note: Rural consumption figures are rough estimates based on production estimates and EICV results. Source: MINECOFIN32002a. Table 20. Structureof expenditures,ruralvs. urbanhouseholds,2000-01 Rural Urban Average Kigali Other towns Consumption o f own-producedfood 47 1 7 34 Foodexpenditures 30 45 45 35 Non-food expenditures 23 54 48 31 Total 100 100 100 100 Source: MINECOFIN, 2002b 94. Rural diets vary regionally, reflecting geographical differences inproduction activities. In the highlands o f western Rwanda, the rural diet i s based on Irishpotato, beans, and vegetables. In the lowlands o f central, southern, and eastern Rwanda, the rural diet i s based on banana, sweet potato, cassava, and beans. Maize i s consumed in limited quantities, mainly in on the cob. Sorghum i s used mostly for brewing o f beer. Livestock products are generally not consumed on 38 the farm, but rather are sold to generate income. Since few Rwandan farmers produce oilseed crops, rural diets include very little vegetable oil, leading to high levels o f child malnutrition. Consumption of wheat remains very limited in rural areas. Consumption of rice has however increased as the result of the government's strong pushto develop marshlands for rice cultivation. 95. Urban diets are based (indecreasing order of importance) on Irishpotato, beans, banana, rice, cassava, bread and sweet potato. Urban households also consume important quantities of vegetables. Inrecent years, imported cereals (wheat and rice) have become very popular among urban consumers as convenience food staples, and they have partially displaced traditional staples, including banana, sweet potato, cassava, maize, and sorghum. Beans are an important source of protein for urbanhouseholds. Consumption of maize and sorghum i s negligible inurban Rwanda. 96. How i s the strength and composition of demand for domestically consumed food products likely to evolve in coming years, and what are the implications for Rwanda's agricultural sector? Future demand for food will be shaped by three main factors: (1) population growth, (2) income gains, and (3) changes inconsumer tastes and preferences. Populationgrowth: The population of Rwanda is expected to grow at a rate of approximately 2.9 percent per year through 2020. Ceteris paribus, demand for food will continue to grow at least at the same rate, providing an assured market for producers of food staples including livestock products. Income gains: Assuming that the economy achieves the 6 percent annual growth target set by government, and assuming the population continues to grow at the current rate of 2.9 percent per year, income per capita will increase at a rate ofjust over 3 percent per year. Risingincomes will trigger shifts inthe strengthand composition of demand for food. As incomes rise, consumers can be expected to shift away from some root crops (sweet potato, cassava), coarse cereals (sorghum, maize), bananas, and beans into rice, wheat, and Irish potato. Consumption of h i t s and vegetables will increase as well. Perhaps the biggest change in demand will involve livestock products, includingmeat and especially dairy products, which have an extremely strong income elasticity of demand. Changes in consumer tastes and preferences: Separate from the changes in food demand associatedwithpopulationgrowth and income gains, the composition of food demand i s likely to evolve with changes in consumer tastes and preferences. The most important of these will be associated with urbanization, as increasing numbers o f people move from the countryside to towns and cities in search of employment. This demographic shift will fuel increased demand for convenience foods, especially wheat-basedbreads snack foods, as well as rice. Regionalmarkets 97. Rwanda's trade with neighboring countries (regional trade) is difficult to estimate with precision, because most of this trade is informal and fails to show up in official statistics. Still, what i s clear i s that most regional trade occurs in one direction: considerable quantities of agricultural commodities are imported into Rwanda from Uganda, the Democratic Republic of Congo (DRC), Tanzania, and Burundi,while only limitedquantities of agricultural commodities are exported out of Rwanda into those countries. Trade flows with neighboring countries consist mainly of imports of food products, including banana, maize, hshpotatoes, and fish. Significant amounts of coffee and tea are thought to come inas well from Burundiand DRC for re-export as higher quality Rwandan commodities. 39 98. Regional markets represent a largely untapped opportunity for Rwandan producers, and the government has rightly identified promotion o f regional trade as an important policy priority. Rwanda's favorable agro-climatic conditions relative to those found in some neighboring countries confer natural advantages in the production o f a number o f commodities, especially certain categories o f food staples (banana, potatoes and other root crops, milk and poultry). These commodities could compete effectively in regional markets assuming existing market access problems can be overcome. 99. Two types o f barriers will have to be overcome in order for a regional trade promotion strategy to be effective, however: (1) physical barriers (especially the lack o f transportation infrastructure), and (2) institutional barriers (formal rules and regulations as well as informal practices and procedures that impose heavy transactions costs). The Government i s pursuing a two-pronged strategy in addressing these constraints. Investments are being made in strengthening the transportation infrastructure needed to link Rwandan exporters to regional markets, most noticeable the road links to Tanzania. At the same time, the Government i s aggressively seelung to foster improved regional integration through its participation in the Common Market for Eastern and Southern Africa (COMESA) and now the East African Community (EAC). International markets 100. The third important source o f demand for Rwandan agncultural commodities i s international markets outside the region. Rwanda's international agricultural exports have long been dominated by two commodities: coffee and tea. Global demand for coffee and tea remains strong, although international markets for both commodities are subject to supply shocks that manifest themselves in sharp year-to-year price swings (Figure 14, Figure 15). More importantly, the nature o f demand ininternational markets i s evolving rapidly, as buyers continue to shift away from commercial grades to specialty grades. Rwanda's ability to compete effectively in international coffee and tea markets will thus depend critically on the ability of Rwandan producers to improve quality. With support from government, ambitious plans have been launched to improve the quantity and especially the quality o f Rwandan coffee and tea. Thus far, the effort seems to be paying off. Incoffee, the proportion o f exports consisting o f specialty grade coffees has increased (Figure 16).Average prices received by Rwanda's coffee exporters have increased, and higher export prices have been translated into higher prices for farmers. In the case o f tea, Rwanda has long had a reputation for producing extremely high quality teas, and prices received by Rwandan teas ininternational markets remain among the highest inthe world. 101. The two other commodities that have traditionally been exported by Rwanda are pyrethrum and hides and sluns. Global demand for pyrethrum i s expected to grow, providing good opportunities for Rwandan producers. Increased demand for environmentally safe insecticides provides good prospects for SOPYRWA to reach its target o f increasing the cultivated area from 3,500 ha to 7,000 ha within the next year. Exports o f hides and skins could also increase in future, although the high cost o f reaching international processing centers are likely to undermine the competitiveness o f unfinished leather (dried or wet blue). Ensuring the future viability o f the hides and shns export sector will thus depend on the ability o f Rwandan producers to expand along the value chain by developing a domestic tanning and leather-products industrythat is capable o fexporting finished leather products into global markets. 40 Figure 14. Trends in coffee prices,Rwandaandworld markets, 1975-2005 300 I AverageWorld Price, g250 1 . Colombia Mild Arablca g8 200 =.-8 I / kr, I \ v) 150 h 50 Average Producer Price, Rwanda Arabia I 1975 1980 I985 1990 I995 2000 2005 Source: International CofSeeOrganization and FAOSTAT data, 2006. Figure 15. Trends intea prices, Rwanda vs. regionaland globalprices, 1989-2005 5,000 4,500 4,000 8 T 3,500 2 43m 3,000 ! 2,500 Mombasa auction price 2,000 .-k8 0 1,500 1,000 FA0 composite world price 500 1989 1991 1993 1995 I997 1999 2001 2003 2005 Source: FAOSTAT data, 2006 41 Figure 16. Quality shares of Rwandan coffee exports, 1986-2004 100% - 3 h 90% - g 80% - 8 70% - - - 60% 0 0 50% - 40% - 0 Bg 30% - 20% - n 10% - 0% - L L 1 L L L 1986 I988 11990 I993 11996 2000 2002 2004 Source: Aureille and Rwamasirabo, 2006. Summarv 102. Rwanda's agricultural policy goals emphasize the importance o f promoting export-led growth in the agricultural sector, based on revitalization o f traditional export crops and non- traditional exports including horticultural products ( h i t s and vegetables, flowers), essential oils, vanilla, and silk. Food staples and livestock products generally receive less attention in discussions o f future sources o f agricultural growth. Yet, with the population growing at nearly 3 percent per year, demand for food can be expected to grow at a similar rate in the short to medium term. More importantly, given present low levels o f food consumption, especially among lower income groups, future income gains will translate into increased per capita food consumption, which could easily add an additional 1-2 percent to demand. Combining these two effects, domestic demand for food could very easily support 5 percent growth in the domestic food crop and livestock sub-sectors duringthe short to medium term. 103. This is not to say that the export crop sector is unimportant. Inthe long run, domestic demand for food will slow as the population stabilizes at higher per capita income levels, and continued growth in the agricultural sector will be sustainable only if opportunities in regional and global markets can be exploited. Developing agncultural exports will be needed as well to diversify the rural economy, malung it less vulnerable to short-term fluctuations inthe fortunes o f individual crops. For Rwanda's traditional export commodities, especially coffee and tea, demand remains strong in international markets, although the nature o f that demand will continue to change as consumers shift into higher quality specialty grades. In the short to medium term, traditional export commodities-especially coffee and tea-will continue to make an important contribution to agricultural growth. Over the longer term, development o f new non-traditional export sectors will be neededto sustain export growth. Growth Drivers: Supplyprospects 104. Itwill not be possible to achieve and sustain highrates o f agricultural growth inRwanda unless producers are able to increase the value o f their production and maintain those higher levels over the medium- to long term. 42 Potential future sources of growth 105. What will drive future growth inRwandan agriculture? On the supply side, three possible sources o f future growth are possible: (i) increased scale o f production (achievable through extensification), (ii) increased productivity (achievable through intensification), and (iii) value addition (achievable through improvements inquality). Increasedscale of production 106. What are the prospects for expanding the land frontier in Rwanda? With the highest population density in Africa and the smallest average farm sizes, Rwanda clearly faces a major challenge interms o f the dwindlinglandresource. 107. Over the last five years, the total area cultivated in Rwanda has expanded at about 1.4 percent per year on average. This moderate rate o f increase has been made possible both through processes o f extensification (e.g., expansion into previously uncultivated lands including marshlands and conversion of permanent pasture to crop land), as well as through processes of intensification (e.g., increasedmultiple-cropping on existing crop land). 108. Despite the widespread perception that the land frontier i s by now largely exhausted in Rwanda, it i s not unreasonable to assume that total cultivated area will continue to expand at a rate of 1 to 1.5 percent per year over the short to medium term. Experience from other densely populated countries, for example in East Asia and South Asia, suggests that intensification can continue well beyond the level at which Rwanda currently finds itself, although significant investments will be neededinland, soil, and water management technologies. Although the scope is limited for bringing new land under production, increasing intensification o f the currently cultivated area will be possible as a result o f continuing development o f underutilizedmarshlands and increasedmultiple cropping of uplandareas. 109. This is not to say that the area planted to every crop will expand. On the contrary, increasing pressure will raise the value o f the land resource and induce substitution by farmers out o f less profitable crops and into more profitable crops as a way o f maximizing returns to the scarce and costly resource. A decrease in the production o f low value staples such as sweet potato, cassava, and beans can therefore be expected. 110. While the total area cultivated i s likely to increase, the rate o f increase in cultivated area i s unlikely to keep pace with projected population growth, resulting in a decline in land availability per capita. This suggests that individual households will continue to have problems accessing enough land needed for them to meet their consumptionrequirements. Competition for land i s therefore likely to increase, and significant numbers o f rural people will have to rely increasingly on off-farm sources o f income. Significant numbers are likely to leave agriculture entirely to seek employment inthe services or manufacturing sectors or even to migrate in search o f employment. Productivity gains 111. What are the prospects for achieving significant productivity gains in Rwanda's major commodities (crops and livestock)? Earlier it was shown that crop yields achieved inRwanda are generally speaking low compared to yields achieved inneighboring countries. While this suggests that possibilities exist to increase productivity in Rwandan agriculture, the scope for achieving productivity gains in Rwanda can be assessed more realistically by asking what i s technically feasible and economically rational for farmers. 43 112. A background study commissioned for this report systematically reviewed the technical and economic prospects for achieving productivity gains in the principal commodities produced inRwanda (see Aertssen et al., 2006). The study examined food staples including livestock, as well as export commodities. For crops, the approach used involved comparing average yields currently recorded in farmers' fields with experimental yields realized in research plots using improved seed and management practices. Based on these comparisons, judgments were made as to whether or not the production targets embodied in the government's official development programs for individual commodities can be consideredrealistic. 113. For food crops, the comparison o f average farmer yields (actual yields) with experimental yields (potential yields) reveals large yield gaps ranging from a low o f about 25 percent inbanana to 75 percent or more for wheat maize and beans (Table21). For horticultural crops, the yield gaps are similarly large (Table22). Yield gaps can never be eliminated completely, but the gaps inRwandaareunusuallylarge comparedto those observedinmany other developingcountries. 114. Based on yield gap analysis, it i s fair to say that the productivity growth targets that underpin the government's commodity development programs are reasonable. Inmost cases, the programs assume yield growth in the range o f 3 to 5 percent per year over the short to medium term (for a more detailed discussion, see Section 4 below, as well as IFPRI 2006). Given the very low productivity levels in farmers' fields, and considering the scope for achieving significant yield gains through relatively modest changes in crop and resource management practices, productivity gains on this order o f magnitude are almost certainly achievable. 44 b 00P WP W b s b d m W W m m d m N m d m 2 zn 0 0 0 0 0 0 0 0 8 mi m :-% `0, 3 N 5E & d m % m m N Q\ m * W N m 0 2 W m W s 3 E% P m m W 00 m 0 0 vr 00 0 c1 :- W W w m W m * sn 3 3 N N d W 10 b m P 3 0 N m d 0 03 c1 2 m P 3 m W 0 P W b 5 2 m 0 0 W b W m W 10 c1 % 2 e4 W W 0 m W m 0 W m I? 3 W m 0 P 00 W m z c1 2 2. b b b N m N % 2- r( W W 8 m m hl 0 P s W m 0 P c1 N N W N * m 0 0 m 2 N P 00 W d 0 b P m m d 0 2 c1 Nn o $3 .I N Table 22. Estimatedyield gap for selected fruits and vegetables, Rwanda Area cultivated Currentyield Potential yield Estimatedyield (ha) (tons/ha) (tons/ha) gap W) Fruits PassionFruit 1,168 10-15 25-40 40 -75 Pineapple 1,432 20 40-85 50 -77 Citrus 2,000 15 20-60 25 -75 Avocado 4,000 6-12 20 40 -70 Mangoes 1,300 6-12 20 40 -70 DessertBanana 9,040 10 35 29 Vegetables Tomatoes 4,000 20 30-180 33 - 88 Cabbages 2,000 20 50-80 60 - 75 Carrot 2,000 12 20-40 40 -70 Onions 2,000 12 30-70 60 - 83 FrenchBeans 1,000 15 15 0 Amaranthus 1,000 20 20 0 Source: Calculations based on MINAGRIand ISAR data, 2005 and cited in DTIS study, 2005. Value addition 115. What are the prospects o f increasing the value o f Rwanda's agricultural commodities? Most o f the food staples destined for the domestic market are consumed with little or no processing, which means there i s limited scope for value addition inthe short run.This situation will change as consumers begin to demand more highlyprocessedfoods and/or foods with special qualities, such as organically produced or nutritionally fortified foods. In contrast, the traditional export crops o f coffee, tea, pyrethrum, and hides and skins are sold in global markets where demand for quality i s evolving very rapidly and where niche markets for specialized high value products offer increasingly attractive returns for those who can compete successfully. For producers and exporters o f the latter commodities, the challenge will be to adopt quality enhancement measures and develop linkages to high value niche markets. The premium prices that can be obtained in these markets will help to promote sustained production and contribute appreciably to future agricultural growth. 116. Coffee: InRwanda as inmost other coffee-producing countries, coffee production i s very sensitive to price changes, and significant changes in coffee prices induce substantial corresponding changes in supply. Partly this i s because most coffee growers in Rwanda also cultivate other crops, income from which can be used to offset variability in income from coffee. If coffee prices decline, growers first stop maintaining their trees but continue to harvest the coffee "cherries." If coffee prices remain low, they then abandon their trees and refrain from harvesting. Finally, if low coffee prices persist for an extended period, many growers will begin to uproot their trees andplant alternative crops. 46 117. Rwanda's coffee sector i s currently recovering from a long period o f neglect. Between 1985 and 1992, production o f green coffee varied between 20,000 and 40,000 tons, but production then fell to the range of 15,000-20,000 tons during the late 1990s because o f low world market prices, which depressed the prices received by producers. During this period, the benchmark New York price for C-grade coffee actually fell below the Rwandan cost o f production, causing in a sharp decline in both production and quality. Farmers responded by decreasing application o f inputs and eventually by abandoning their trees. More recently, a strengthening in global prices has induced many producers to return to their trees, although only about 65% o f the 85 million coffee trees growing inthe most favorable agro-ecological zones are today producing. 118. Faced with the declining profitability o f coffee, the Government o f Rwanda has developed a coffee strategy based on the expectation that growth inglobal demand for coffee will be modest relative to growth in overall supply, but that opportunities will emerge to exploit the relatively small segment o f the market that consists o f high-quality Arabica coffee. This segment o f the market, which i s projected to grow at an average annual rate o f 15 percent, commands prices that are more than double the process o f ordinary C-grade Arabica. Consistent with the goal o f conquering this market niche, a national coffee strategy has been developed based around improving the quality o f Rwandancoffee. 119. The national coffee strategy recognizes that strong efforts will have to be made through participation in international exhibitions, demonstrations, and contests if Rwanda i s to develop a reputation a quality coffee producer. At the same time, effective measures will be needed to ensure quality control. In the highquality market segment, coffee is typically sold on the basis of personal relationships between the buyer and the seller. This means that specific buyers can be targeted in efforts to raise Rwanda's reputation for quality. Rwandan producers have already demonstrated they can convert this potential into reality. Fully-washed coffee from the Maraba washing station achieved high scores at international coffee tasting competitions placing it well above the threshold level for a specialty coffee. Rwandan coffee i s now being featured by major internationalretailers such as the US-based Starbucks chain and Union Coffee Roasters. 120. Efforts to increase investment in high quality coffee production appear to be paying off. The number o f fully functional coffee washing stations increased from five in2002 to 46 in2005. Production o f fully washed coffee increased from 17 tons in 2000 to 1,100 tons in 2005. The increasing share o f fully washed coffee in total exports has been reflected in a sharp increase in the average prices received by Rwanda's coffee exporters, which have been passed along to producers in the form o f higher prices for cherries and to processors inthe form o f higher prices for parchment coffee (Table 23). The increases in export prices stemmed mainly from the premium prices received for high quality fully washed coffee, although a strengthening in world coffee prices contributed as well. 121. The early success o f Rwanda's effort to add value to coffee exports augurs well for the future. Although it remains to be seen whether the Government's ambitious plan will be realized to gain specialty coffee status for 100% o f the export crop by 2010, even if the process takes twice as long as planned, the result will still be a significant increase of the overall value o f the coffee export crop and a strong contribution to overall agricultural growth. 47 Table 23. Producer prices and export prices of coffee, 2003-05 2003 2004 2005 Cherry coffee 55 -60 60 90 - 100 120 - Parchment coffee 250 300 - 400 500 -650 Export prices (U Average Maximum Average Maximum Average Maximum Price Price Price Price Price Price Semi-washed 1.oo 1.08 1.16 1.28 2.18 2.44 Fully- 2.32 2.90 2.42 4.85 3.06 7.80 washed Source: OCIR-CafL, 2005. 122. Tea: Like production o f coffee, production o f tea i s also extremely sensitive to price changes, but the relationshipbetween tea prices and tea production plays out differently because o f the way tea i s produced. Approximately two-thirds o f the tea produced inRwanda i s grown on industrial estates belonging to OCIR-ThC and some private investors. The remaining one-third i s grown by small-scale private growers. Tea growers react to low prices in much the same way as coffee growers: first they scale back investment in management, especially application of chemical fertilizer; next they stop harvesting the plants; and finally as a last resort they remove established plants and substitute other more profitable crops. On industrial states where the main variable cost i s the cost o f labor, tea production levels are furthermore heavily influenced by the availability and cost o fworkers. 123. Following the genocide, many o f the government-operated tea estates faced severe budget constraints, which resulted in neglect o f established plants and a reduction in harvested area. Since 2000, tea production has increased in response to strengthening global prices, but green tea yields in Rwanda remain low, especially on the government operated estates. Based on the cultivated area and potential yields, it i s estimated that the estates are producing about 63 percent o f what could potentially be produced 124. The world tea market shares a number o f similarities with the world coffee market. Like the market for coffee, the market for tea indeveloped countries i s basically saturated, even as the demand for specialty teas continues to grow rapidly within the larger market. And like demand for coffee, demand for tea i s growing rapidly in the developing world, especially in Russia, China, Pakistan, and India. Despite these similarities, there are however some major differences between the two commodities. Most importantly, the international tea market i s far more stable. Similar to coffee prices, tea prices ininternational markets have been trending downward, but the decline in US-denominated international tea prices has been much less pronounced. In addition, international tea prices have shown far less cyclical variability. 125. As for coffee, the Government o f Rwanda has developed a strategy for tea that is based on the expectation that growth in global demand will be modest, but that opportunities will emerge to exploit the relatively small segment o f the market that consists o f high-quality teas. 48 Consistent with this strategy to improve quality, the producer price system used by OCIR-ThC factories when purchasing green tea from private outgrowers now includes a quality premium. A pan-territorial national producer price for green tea i s announced each season by OCIR-ThC, but the price actually paid to individual growers depends on a quality index that expresses the percentage o f "good leaves" in each lot. Ifthe proportion o f good leaves i s less than 66%, the lot i s rejected. If the proportion o f good leaves ranges from 66% to 70%, the grower receives the basic producer price. Ifthe proportion o f good leaves exceeds 70%, the grower receiveda quality premium, currently equal to 5 FRW/ kg on top o f the base price o f 65 FRW/ kg. 126. The introduction o f a quality premium for tea unquestionably has improvedincentives for growers to deliver a quality product to the tea factories, and the result has been reflected in the price premiums earned by Rwandan teas inthe Mombasa auction (Table 24). On the other hand, the fact that the basic tea price i s a pan-territorial price set on the basis o f average costs and retums across the entire industry i s undesirable, because it means that the more efficient tea factories are subsidizing the less efficient factories. As a result, the spatial allocation o f resources managedby OCIR-ThC i s sub-optimal, and overall production costs are higher thannecessary. Table 24. Highest prices,Mombasaauction, Jan-Oct 2005 (US centdkg) Tea quality Rwanda Kenya Uganda Burundi Tanzania Malawi Madagascar Broken Pekoe 1 404 318 173 149 150 120 130 PekoeFannings 1 356 252 155 150 140 125 137 PekoeDust 280 232 141 150 138 132 130 Dust 1 275 281 142 141 130 115 128 Source:Aureille and Rwamasirabo, 2006. 127. Pyrethrum: Increased global demand for environmentally safe insecticides provides good prospects for SOPYRWA to reach its increasedproductiontargets. Inaddition to increasing production volume, additional significant gains can be achieved by increasing the value o f exported products. In the past, the Rwandan pyrethrum industry was unable to negotiate from a position o f strength in regional and international markets, because in the absence o f reliable domestic refining capacity, the only option was to export crude pyrethrum extract or even dried flowers to an overseas processing industrythat was dominated by a small number o f companies that were able to dominate the market. This situation i s expected to change, however, as a group o f private investors i s currently renovating the pyrethrum processing facility near Ruhangeri. Once the processing facility comes on line, it will be possible to export high-value Rwandan pyrethrum extracts and distillates, for which the market i s much larger. Prices fetched by Rwanda's pyrethrum exports in international markets have been increasing, reflecting strengthening global demand for "natural" pesticides (Table 25). 49 m 0 0 N B 0 N m 0 0 N N 0 0 N rl 0 0 N 0 0 0 N o\ Q\ Q\ 3 W z Q\ o\ I- z \o Q\ Q\ 3 of m o\ 3 v z o\ m z Q\ 128. Hides and skins:With more than 90% o fRwanda's hides and skins exports consisting o f unfinished dry leather, opportunities clearly exist to add value inthe sector. Two possible stages o f value addition can be identified, achievement o f which will pose quite different challenges. The first stage would be to increase the proportion of hides and skins exported in "wet blue" form. Achieving this stage will require strengthening the system for collecting unprocessed hides and skins, which will require not only strengthening o f the transport and logistics systems o f the tanneries, but also educating livestock producers to improve handling o f hides and skins at the farm level. The second stage would be to increase the quantity and quality o f finished leather goods exports, including shoes, clothing, and accessories (belts, bags, wallets, etc.). Achieving this stage will require developing a leatherworkindustry, which is likely to require considerable capital investments in facilities, training o f skilled workers, and establishing links to international leather distribution networks. Prospects for achieving the first stage are reasonably good, and it i s conceivable that increased exports o f wet blue leather could contribute to agricultural growth even in the short to medium term. Prospects for achieving the second stage seem much less promising, however, given the imposing economic, technical, and institutional barriers that will needto be overcome inorder to make Rwanda competitive inthe global market for leather goods. Competitiveness of Rwandan commodities 129. The fact that Rwanda i s landlocked, with relatively poor links to external markets, has bothpositive and negative implications for Rwanda's agricultural sector. Onthe positive side, the country's geographical isolation, coupled with its predominantly mountainous terrain, confers a natural level o f protection for Rwandan producers inthe domestic market, because hightransport costs make the prices o f imported commodities very high within the country. On the negative side, the same factors that confer natural protection to Rwandan producers in domestic markets pose a major challenge to producers looking to sell inregional or global markets. The highcost o f transporting Rwandan commodities to regional or global markets must be absorbed in order for those commodities to be competitive outside the country. 130. These effects are particularly important in the case o f unprocessed staples that have a highvolume-to-value ratio, including most roots and tubers, bananas, and cereals. For these low- value commodities, transport costs represent a large share o f the final price, which discourages transportation over large distances. Hightransport costs also discourage trade, because they have the effect o f making imports excessively expensive for consumers and exports unprofitable for producers. This can be seen in Table 26, which shows the large difference between import and export parity prices for selected food staples. When the difference i s so large, domestic markets are likely to clear at some price falling inbetween the import and export parity prices, leaving the country ina state o f autarky, meaningno trade inthat commodity takes place. 51 Table 26. Import and export parity prices, food staples, Rwanda, 2005 Source I Import parity Producer Export parity destination price price price at farm gate at farm gate at farm gate Maize Kenya 406 220 -58 Rice COMESA 427 365 79 Wheat Argentina 372 350 -52 Beans (dried) Uganda 552 445 324 Soybeans Tanzania 175 155 -3 Irishpotato Uganda 164 120 6 Sweet potato Kenya 262 135 -82 Cassava Uganda 159 85 1 Source: Calculatedfrom data in Aureille and Rwamasirabo, 2006. Foodcrops 131. Because highexternal transport costs add appreciably to the landed cost o f food imports, Rwandan food crop producers are generally able to compete effectively with imports in domestic markets, even when domestic production costs are relatively highby international standards. Yet rice, wheat, and maize from international markets are imported regularly in significant quantities to compensate for regular structural deficits in production. Bananas, potatoes, cassava, beef, and fish are imported in smaller quantities from neighboring countries, mainly Uganda, Congo, and Tanzania, with the volumes varying considerably from year to year depending on local production conditions. 132. One development that could significantly change the current situation i s the eventual emergence o f supermarkets as a leading marketing channel for food staples. Large regional supermarket operators are still uncommon in Rwanda, but judging from the speed with which they have grown to capture dominant market shares in South Africa, Kenya, and Uganda, their appearance in Rwanda can be expected sooner rather than later. Supermarkets affect the food buyingbehavior o f consumers, but eventually they will also transform production relations by expanding their reach upward in the supply chain in an effort to secure reliable supplies o f products that meet well-defined standards for quality, safety, and nutrition. The influence o f supermarkets will be felt initially in peri-urban production zones surrounding Kigali and other urban centers, but once they gain a foothold and establish themselves, they can be expected to spread rapidly throughout the country. 133. Supermarkets will offer new challenges and opportunities for Rwandan food crop producers. On the one hand, they could try to source as many products as possible locally, and they will introduce technology, capital, and management expertise that will help raise productivity all along the value chain and strengthen linkages between production, assembly, processing, and distribution. This will create new opportunities for Rwandan producers who are able to adapt to the demands o f an increasingly quality-driven market. But on the other hand, supermarkets will be unforgiving intheir search to achieve efficiencies and maximize profits, and they will aggressively seek out alternative sources o f supply, including outside the country. 52 Traditional exportcrops 134. High transport costs also affect the competitiveness o f traditional export crops, notably coffee and tea, but two factors have enabled Rwandan coffee and tea producers to remain competitive in international markets. First, coffee and tea have high volume-to-value ratios, so transport costs make up a relatively smaller share o f the final selling price than i s the case with food staples. Second, by pursuinga high-quality strategy for coffee and tea, Rwandan exporters have been able to differentiate Rwandan products in global markets. This has allowed them to build strong demand for Rwandan branded products, which in turn has allowed them to obtain premium prices. While bulk grade coffee and tea produced in Rwanda will have an increasingly difficult time competing in international markets with similar grades o f tea and coffee produced in low-cost countries such as Vietnam, China, and India, as long as Rwandan producers can differentiate their products and sell directly to the large international companies that dominate global trade at premium prices, Rwandan coffee and tea should remain competitive. For these crops, the key to maintaining competitiveness in future lies only partly in reducing unit production costs through technological innovations leading to productivity gains; equally important will be raising quality as a way o f securing higher prices. Non-traditionalexportcrops 135. Withthe goal of stimulating increased growth, increasing foreign exchange earnings, and diversifying the commercial part of the agricultural economy away from its current reliance on coffee and tea, policy makers in recent years have sought to promote a range o f non-traditional export crops. Efforts are being made to promote horticultural crops including fruits, vegetables, and cut flowers; essential oils such as petunia and geranium; macadamia nuts; vanilla; and silk. Despite their apparent dissimilarities, these non-traditional export crops share some common features. Most are high-value cash crops targeted at niche markets in Europe or the Middle East. Production i s generally speaking capital intensive and labor intensive. Since by definition these are new crops, successful development requires an integrated supply chain approach focusing not only on production but also processing, transportation, and above all direct marketing through dedicated contracting arrangements with external buyers. 136. Under the Government's Export Promotion Strategy, feasibility studies have been prepared for a number o f horticultural production pilot projects. Passion fruit, apple banana, cut flowers and houseplants, and cape gooseberry have been targeted as four product categories with strong export potential. The USAID-funded ADAR and PEARLprojects have producedtechnical papers and market studies on these and numerous other products and product categories, including French beans, peppers, bird's eye chilis, geranium oil, dairy products, cassava, and honey. In addition, the government commissioned a Horticulture Strategy (OTF, 2005), and efforts are underway to establish a Rwanda Horticulture Development Authority (RHODA). Rwanda's horticulture industryremains at an early stage. In2005, horticulture exports totaled 650 tons, valued at $250,000, an impressive figure considering that the industry employs less than 1000people (OTF, 2005) (Table 27). 53 Table 27. Rwandan exports of fruits, vegetables, and flowers, 2000-03 Exportweight (Kg) 2000 2001 2002 2003 Dessert banana 64,585 36,045 43,248 41,615 Passion h i t 9,263 12,342 9,024 13,885 Avocado 3,192 1008 1,576 1,255 Pineapple ... 510 1,377 1,270 Roses 109,123 66,945 110,000 ... Draceana plants 85,708 151,828 4,791 40,940 Eggplants ... 605 430 150 Frenchbeans ... ... 57 ... Snap peas ... ... 415 566 Cooking banana ... 7,920 7,740 370 Source: MINAGRIdata, 2004. 137. Despite having a favorable climate and good soils for horticulture production, Rwanda faces some key constraints compared to its direct regional competitors. Because o f its landlocked character, Rwanda's potential for exports exists in carefully selected segments and markets. Advantages should be based on a niche appeal o f highvalue added products, not on volume or price. Perhaps more important than product identification i s to add-value and build the vertical and horizontal value chains for non-traditional export products. Since, production i s generally speaking capital intensive and labor intensive and these are new crops, successful development will require an integrated supply chain approach focusing not only on production but also processing, transportation, and above all direct marketing through dedicated contracting arrangements with external buyers. This will include attracting investors for commercial production schemes, establishing linkages and relationships between stakeholders at all stages o f the chain. Downstream in the export market, international partners will have to be identified, products promoted and markets tested. In order to facilitate this process, dynamic, pragmatic and flexible facilitators (partnerships between donor-funded projects, public and private stakeholders) and a long-term engagement are needed. 138. There i s considerable interest inthe production o f high-value organic products, but a lack o f quality organic certification services constrains Rwanda's producers from entering this market. Given the susceptibility to diseases and quality requirements for fresh vegetables and fruits inthe export market, only commercial farms or very well structured farmers' associations will be able to supply the required normalized export quality. The European market imposes traceability requirements that are difficult to fulfill by informal small-scale producers. Other technical constraints include: (i) infrequent cargo flights with limited capacity and high cost o f transpod; (ii) ofcoldstorageandhandlingfacilities atallstagesofthechain; (iii) oftechnical lack lack knowledge at all levels; (iv) absence o f certification and quality control services and laboratories; and, (v) absence o f specific inputsandcheap packaging material. 6 Airfreight rates are about 2.30/kg for the European marketand1.60/kg for the Middle-East and South Africa. This rate is relatively high compared to that paidby Rwanda's competitors, mainly South Africa and Zambia, where output is increasing. Kenyan and Ugandanexporters have direct airline connections. Airfreight rates to Europe are estimated at US$1.57/kg from South Africa and US$1.30-1.50/kg from Zambia. At these rates, Rwandan exporters will be starting out at a competitive disadvantage and the negotiationo f a more beneficial rate will be critical to success (Sunderland, 2005). 54 139. For all o f these reasons, promoting horticultural production for exports will be a challenging and costly enterprise. Given the highbarriers to entry, based primarily on standards, quality, and price and the highperishability and susceptibility to diseases o f horticultural produce, a significant amount o f up-front investment will be needed to respond to these challenges if Rwanda i s to begin to compete on an equal footing with key horticultural suppliers. Adapted credit products will be a basic requirement. It would make sense to exploit domestic and regional markets where the market regulations are less complex. Advantage should also be taken o f the increase in large supermarkets across eastern and southern Africa and in the near future also in Rwanda. There should be a push to forge agreements with these supermarket chains to procure locally within Rwanda. 140. Rising interest in non-traditional export crops has been reflected in the preparation o f numerous feasibility studies for pilot projects, but the feasibility studies have not yet generated much actual investment on the ground. Therefore it i s perhaps premature to draw valid conclusions about the growth prospects for this sector. Several pioneering enterprises have enjoyed early success, for example the Rwanda Flora operation outside Kigali that i s successfully producing roses for the European cut flower market. A handful o f h i t and vegetable producers similarly have enjoyed some modest success in exporting fresh as well as processed vegetables and h i t s into the Europeanmarket. 141. Non-traditional export crops show some promise, but they are unlikely to be a major source o f long-term pro-poor growth for the country, for several reasons. First, the scale o f production i s very limited. All o f these enterprises are capital intensive, and while many do also generate significant employment opportunities for semi-skilled and skilled workers, wage growth will be constrained by the need to compete effectively with similar enterprises in countries that have better access to end markets. Second, many non-traditional exports are targeted at niche markets, which by definition are small and ephemeral. Judging from the horticultural sector, for which there i s perhaps the longest track record, the experience inmany other countries, inAfrica as well as in other regions, i s that the early successful innovators have rarely been able to maintain a dominant presence in the market, as effective competitors have always sprung up to capture market share. 4. PROSPECTSFORGROWTH AND POVERTYREDUCTION 142. Options for increasing agricultural growth in Rwanda were assessed using the Rwanda economy-wide multi-market (REMM) simulation model.' The REMM also made possible evaluation o f the linkages and the trade-offs between growth and poverty reduction at bothmacro and micro levels.' Base Year: Initial conditions 143. To generate further insights into the current features o f Rwanda's agricultural sector and to provide additional context for the simulation results that follow, it is useful to review some o f the economic characteristics represented by the base year data. The REMM model was developed by a team from the Washington-based International Food Policy Research Institute (IFPRI), with the help of financial support provided by the UK Department for International Development and World-Bank administered Belgiantrust funds Technical details about the (REMM) appear inAnnex 1. 55 Land holdings and uoverh,distribution across rural prouus 144. Data were presented earlier in this report showing the acute scarcity o f land in Rwanda. The initial conditions prevailing inthe base year o f the REMMsimulations strongly reinforce the idea that limited access to land i s a key indicator explaining income inequality. Across the three rural groups, landholding size i s highly correlated with household income (R' = 0.66) (Figure 17).Among all rural households holding less than 1ha o f land (a category that includes three- quarters o f all rural households), more than 85 percent have an annual income below the national poverty line o fUS$129. Figure 17. Relationshipbetweenlandholdingsize and income,Rwanda, 2001 250 o I 2 3 4 5 6 7 a 9 IO Household landholdingsire (ha) Source: IFPRI calculationsfrom EICV data, 2006. Income distributionand consumptionuatternsacross rural grouus 145. There are significant differences in average income among the three rural groups, but because inall three rural groups the income level o f the majority o f households i s low, there i s no significant difference among the three rural groups in consumption patterns. Consumption patterns do vary considerably among income quintiles, however, especially between the four lowest quintiles and the highest quintile (Table 28 and Table 29).For this reason, the following discussion focuses on income quintiles. 56 Table 28. Annualcommodityexpendituresper household,Rwanda,2003 ($) Maize Rice Wheat Cassava Potatoes Banana Livestock Total Lowest 15 2 2 26 26 10 5 287 Middle 40 6 3 65 67 49 18 756 Highest 61 19 7 99 73 186 100 1,493 Total 40 9 4 69 58 81 37 861 Lowest 9 31 6 32 51 19 36 428 Middle 18 78 24 50 77 56 161 926 Highest 12 102 67 44 75 73 324 1,402 Total 14 80 36 44 70 52 184 974 Source: IFPRI Calculationsfiom EICV data, 2006. Table29. Annualcommodityexpendituresbyincomequintile,2003 (000 $) Maize Rice Wheat Cassava Potatoes Banana Livestock Total Rural Lowest 3.7 0.5 0.4 6.8 6.8 2.6 1.4 74.0 Middle 10.9 1.6 0.8 17.6 18.3 13.4 4.9 205.5 Highest 21.3 6.6 2.4 34.3 25.5 64.5 34.7 519.0 Total 57.8 12.4 5.4 98.5 83.0 116.7 53.6 1,234.2 Urban Lowest 0.2 0.9 0.2 0.9 1.4 0.5 1.o 11.8 Middle 0.5 2.3 0.7 1.5 2.3 1.7 4.8 27.7 Highest 0.5 4.1 2.7 1.8 3.0 2.9 12.9 56.0 Total 2.2 12.6 5.7 6.9 11.1 8.2 29.0 153.4 Source: IFPRI Calculationsfiom EICV data, 2006. 146. Analysis o f the average budget share data fiom EICV (Figure 18) reveals some clear patterns infood consumption patterns: 0 Root crops account for one-third o f total consumption expenditure for the rural households as a whole, and the shares are much higher among the low-income quintiles than that for the highest income quintile. 0 Grain consumption accounts for about 8-10 percent o f total consumption expenditure for rural households (except for the households in the highest income quintile). Low-income 57 households consume more coarse grains, such as sorghum and maize, while high-income households consume more small grains, such as rice and wheat. Among other food staples, bananas are strongly preferredamong high-income households, as expenditure on bananas rises sharply inthese households when income rises. As income increases, all households spend significantly more on livestock products and beverages. Nevertheless in the low-income quintiles, expenditure on livestock products still represents a very small budget share. Figure 18. Average budget shares, by income quintile, Rwanda, 2001 100% 8091, cE 1 t - 60% Bm 0 af % 40% 2091, 0% I 2 3 4 5 Mean 1 2 3 4 5 Mean Rural Urban Source: IFPRI Calculationsflorn EICV data, 2006. 147. The EICV average budget share data reveal also that urban households have quite different spending patterns from the rural households: Expenditures on agricultural commodities account for slightly less than 50 percent o f total household expenditure among urbanhouseholds, as compared to more than 87 percent among rural households. Among urban households, the share o f total expenditure going to grain purchases i s comparable to the share among rural households. However the type o f grain preferred differs: urban households spend mainly on rice and wheat, while rural households spend mainly on sorghum, maize, and millet. Urban households spend a very small budget share on root crops. As income increases, consumption o f livestock products increases significantly among urban households. 58 e As income increases, consumption o f bananas significantly decreases among the urban households. 148. Comparison o f average budget shares across different income quintiles cannot fully reveal the changes inconsumption patterns that occur with income growth. To get a better picture o f income-related changes in consumption patterns, it i s useful to consider also marginal budget shares (Figure 19),which reflect the marginal propensity to consume from additional income. A number o f interesting insightsemerge: There i s no significant difference between the marginal and average budget shares o f agricultural consumption for rural households, except among households in the highest income quintile. This pattern i s typically found in situations o f extreme poverty when households must devote a highshare o f their income to food purchases. For every additional FRW o f income earned, the average rural household in Rwanda spends 0.84 FRW on food. Even the wealthiest rural households have a propensity to spend a large share o f incremental income on food. Among rural households in the highest quintile, the marginal share o f agricultural consumption falls only slightly to 74 percent, compared to an average budget share o f 85 percent. Figure 19. Marginalbudgetshares, by income quintile, Rwanda, 2001 100% 80% 60% Beverages Home pmmes?d100 OSugar, mflee,and te Fruh6and vegetable 2 E 40% 1 moikeds I 0Bananas OLegumes 2091. 0% -20% Rural Urban Source: IFPRI Calculationsfrom EICV data, 2006. 59 149. When rural households acquire additional income, their spending continues to be dominated by agricultural consumption, but the pattern o f marginal spending varies among commodities and across income groups: For root crops the marginal budget share i s generally lower than the average budget share, indicating that as incomes rise, most rural households consume relatively less o f root crops and relatively more o f other commodities. The effect i s particularly pronounced for sweet potato, with the notable exception o f households inthe lowest income quintile, among which the marginal budget share for sweet potato exceeds the average budget share. These relationships confirm the status o f sweet potato as a food for the very poor. For banana, the effect i s the opposite: in the four lowest income quintiles, the marginal budget share i s higher than the average budget share, indicating that as incomes rise, most rural households choose to consume relatively more o f banana and relatively less o f other commodities. Only in the highest income quintile does this relationship change; households inthe highest income quintile consume relatively less ofbanana at the margin, indicatingthat beyond a certain income level, bananabecomes an "inferior" food. Rural households in the highest income quintile display quite different marginal propensities to consume compared to their lower-income neighbors. For example, they spend a much smaller share o f incremental income on root crops and cereals (except for rice), and they spend a much larger share o f incremental income on livestock products, including meat and especially milk. 150. When urban households acquire more income, the extent to which the additional income i s spent on food varies depending on the households' income level. Among poor urban households (i.e., those included in the first two lowest income quintiles), average budget shares and marginal budget shares are similar, indicating that food purchases continue to dominate consumption expenditure. However, as their income rises, these households spend relatively less on root crops. Spending on cereals remains unchanged overall, but there i s a clear shift away from coarse grains (sorghum and maize) toward small grains (rice and wheat). Among wealthy urban households (i.e., those included in the three highest income quintiles), increases inincome result inrelatively less spending on root crops, coarse grains, and banana, and relatively more spending on small grains (rice and wheat) and livestock products. 151. Inconsidering these expenditure data, it is important to keep inmindthat average and marginal budget shares do not necessarily give a good indication o f the total amount o f a given commodity consumed in the country (which i s what determines the strength o f overall demand and hence production opportunities for farmers). This i s because income i s distributed very unequally, so total consumption tends to be heavily influenced by the expenditure choices made by wealthy households. The average and marginal budget shares for a given commodity may be very highamong poor households, which make up the vast majority o f all households, but ifthose shares are low among wealthy households, total consumption o f that commodity will be low. Conversely, the average and marginal budget shares for a given commodity may be very low among poor households, but if those shares are high among wealthy households, total consumption o f that commodity will be high. 152. In summary, the overall strength of demand for a given commodity is determined by a combination o fthe propensity o f households to consume that commodity (at the margin as well as 60 on average) and the amount o f income commanded by households. Taking into account the EICV budget share data and the EICV absolute spending data, the following insights emerge: 0 Demand for many commodities i s dominated by demand emanating from wealthy households, given the much higher income and expenditure levels o f these households compared to poor households. 0 Given current low consumption levels among poor households and significant differences in the amount o f staples consumed by wealthy households and poor households, demand for most staple foods will increase significantly with income growth, especially income growth among the poor households. 0 If future economic growth benefits wealthy households more than poor households, then demand growth in some staples, such as sorghum, cassava and maize, will be limited, given that the marginal propensity to consume these commodities i s l o w among wealthy households. At the same time, demand for commodities preferredby wealthy households will increase significantly, including wheat, rice, andlivestock products. Simulation results 153. The REMMwas developed to allow simulation o f the likely impacts o f alternative policy scenarios on growth, incomes and poverty, food security, and the trade balance. Since these impacts must be evaluated relative to a baseline, first it was necessary to simulate a Base Run Scenario representing the "business as usual" option, under which agricultural and non- agricultural growth are assumed to continue along current trends. BaseRun Scenario 154. Under the Base Run Scenario, the REMMmodel was used to simulate results over a 10- year period from 2005-15. The parameters used for the Base Run scenario are described inIFPRI (2006). Values for agricultural and non-agricultural growth parameters were set to reflect the rates that were actually registered during the previous decade. Some growth rates were subjectively adjusted downward, however, inthe expectation that the unusually highgrowth rates recorded as the nation recovered from the effects o f the genocide will not be maintained over the longer term. Ineffect, the Base Run Scenario represents the "business as usual" option. 155. Key results generated under the Base Run Scenario appear in Table 30, Table 31, Table 32 and Figure 20, Figure 21, and Figure 22 and are summarized below. 61 Table30. Income growthand povertyreduction,REMM simulationresults Growth rate Growth rate Additional Annual growth rate, 2000-05 (%I in in growth vs. Base BaseRun Scenario 23 Run GDP 3.88 6.24 2.36 Ag GDP 3.60 6.17 2.57 NonAg GDP 4.08 6.28 2.21 GDP pc 1.15 3.44 2.29 Ag GDP pc 0.87 3.37 2.50 NonAg GDP pc 1.34 3.49 2.15 Income for rural HHwith cash crop 3.89 6.33 2.44 Income for rural HHwithout cash crop (10% o ftotal rural HH) 3.73 6.01 2.28 Income for rural male-headedHH 3.87 6.37 2.50 Income for rural female-headed HH 3.90 6.18 2.28 Income for Rural Group 1 3.70 6.21 2.51 Income for Rural Group 2 3.89 6.33 2.45 Income for Rural Group 3 3.91 6.34 2.43 Staple production 3.76 6.21 2.45 Grainproduction 4.12 9.62 5.50 Root production 2.21 3.27 1.06 Pulse and oilseed production 1.44 3.69 2.25 Livestock production 4.28 7.82 3.54 Export crop production 1.21 9.93 8.72 Poverty Poverty measures Poverty rate in Poverty rate by 2005 2015 reduction (%) National 59.2 42.4 -16.8 Rural 64.5 46.6 -17.9 RuralHHwith cash crops 62.2 43.6 -18.7 RuralHHwithout cash crops 88.5 78.3 -10.2 Rural male-headed HH 61.9 43.6 -18.4 Rural female-headed HH 71.4 54.7 -16.7 Rural Group 1 73.1 56.9 -16.2 Rural Group 2 66.0 46.2 -19.8 Rural Group 3 52.5 34.4 -18.0 Source:IFPRIREMMsimulations, 2006. 62 Growth 156. Modest agricultural growth averaging 3.60 percent per year will combine with slightly stronger non-agricultural growth averaging 4.08 percent per year to produce overall GDP growth o f 3.88 percent per year from 2005 to 2015. Continuedrobust population growth averaging about 2.7 percent per year will reduce GDP growth per capita to only 1.15 percent, reflecting the combined effects o f 0.87 percent annual growth in agricultural GDP per capita and 1.34 percent annual growth innon-agricultural GDP per capita. 157. Within the agricultural sector, growth inthe grains sub-sector (4.12 percent per year) and inthe livestock sub-sector (4.28 percent per year) will be considerably higher than growthinthe export crops sub-sector (1.21 percent per year). Figure20. Annual GDP growth, REMMBase RunScenario, 2005-2015 4.50 c a 1.50 Q Ea 1.00 U 0.50 0.00 GDP AgGDP NonAgGDP GDPpc AgGDPpc NonAgGDP PC Source: IFPRIREMMsimulations, 2006. Incomes andpoverty 158. The income gains associated with GDP growth will be distributed unevenly (Figure21). Income growth will be highest among the relatively wealthy households in Rural Group 3 (3.91 percent year) and lowest among the relatively poor households inRural Group 1(3.70 percent per year). Female-headed households will do slightly better than male headed households, experiencing income growth o f 3.90 percent per year compared to only 3.87 percent among male- headed households. As a group, the 90 percent o f rural households that engage in cultivation o f cash crops (defined as coffee, tea, pyrethrum, hides and skins, potatoes, fruits and vegetables, sugar, and beverages) will experience higher income growth (3.89 percent per year) than will the 10percent o f households who do not cultivate cash crops (3.73 percent per year). 63 Figure 21. GDP growthby householdtype, REMMBaseRunScenario,2005-15 3.95 E3.90 - A C 3.85 50b - 3.80- 9m 3.75 Q - -nm 85 3.70- - ' C J 3.65 C 3.60- 3.55- L L L Rural HH Rural HH Rural male- Rural Rural Rural Rural wlth cash without headed HH female- Group 1 Group 2 Group 3 crop cash crop headed HH Source: IFPRIREMMsimulations, 2006. 159. The overall national poverty rate will fall from 59.2 percent in 2005 to 53.9 percent in 2015. The number o f people living below the poverty line will actually increase, however, rising from 4.95 million to 5.88 million because o f continuing robust population growth. The poverty- reducing effects of growth will differ across the population. For example, the rural poverty rate will decline from 64.5 percent in2005 to 58.9 percent in2015, while the urbanpoverty rate will decline from 13.8 percent to 11.7percent duringthe same period. Withinthe rural population, the incidence of poverty in 2015 will remain much higher in Rural Group 1 (68.0 percent) than in Rural Group 3 (45.8 percent). The incidence o f poverty among rural households that do not engage in cash cropping will actually rise, to 86.2 percent, and it will be much higher than the incidence o f poverty among rural households that engage in cash cropping (52.3 percent). Rural female-headed households will not benefit from growth nearly as much as rural male-headed households; the poverty rate among rural female-headed households will remain stubbornly high at 66.4 percent, compared to only 56.0 percent among male-headed households (Figure 22). 64 Figure 22. Declines in poverty rate by householdtype, REMMBase RunScenario, 2005-15 Rural HH Rural HH Rural Rural E h wlth without male- female- cash cash headed headed Rural Rural Rural National Rural crop crop HH HH Group 1 Group2 Group3 Source: IFPRIREMMsimulations, 2006. Food availability 160. Rwanda currently depends on imports to meet domestic consumption requirements for a number o f key food commodities, including vegetable oil, wheat, sugar, maize, rice, and dairy products. Under the Base Run Scenario, projected increases in production o f these commodities fails to keep pace with rising demand, resulting in increasedlevels o f food imports by 2015 and a decline innational food self-sufficiency (Table 31). Table 31. Projectedimports of agricultural commodities, 2015 Baseyear (2003) Growthsimulation(Scenario 23) Projected Annual Production Imports productionin Projected import 2015 imports in2015 growth rate (2005-15) (1000 mt) (1000 mt) (1000 mt) (1000 mt) (%I Maize 81 11 189 42 15.0 Rice 28 13 108 0 --- Wheat 15 20 36 30 3.9 Beans 240 8 367 96 16.7 Vegoil 1 6 2 9 3.0 Sugar 7 11 15 20 5.7 Milk 129 3 323 0 Source: IFPRIREMMsimulations, 2006. 65 Balance of trade 161. Rwanda depends heavily on agriculture to generate export earnings. Inthe base year o f 2005, agncultural exports-mainly tea and coffee-contributed to a net apcultural trade surplus o f about US$ 15.4 million. Under the Base Run Scenario, agricultural exports grow at an average annual rate o f 2.26 percent, while agncultural imports grow at an average annual rate o f 2.34 percent. Because imports are smaller in absolute terms than exports, even though import growth i s higher inpercentage terms, under the Base Run Scenario the agricultural trade surplus actually increases to US$66.3 million per year by 2015 (TubZe32). Table 32. Projectedagricultural export growth and agricultural trade balance BaseRunScenarioannualgrowth rates(YO),2005-2015 Production Exports Coffee 1.09 0.95 Tea 1.09 0.97 Pyrethrum 7.28 7.28 Hides & Skins 4.29 4.29 Annual growth rate(YO) Agricultural exports 2.3 Agricultural imports 2.3 Agriculturaltrade surpluses Agricultural trade surpluses inthe base year (1000 $US) 15,474 Agricultural trade surpluses by 2015 (1000 $US) 19,04 1 GrowthScenarioannualgrowth rates(YO),2005-2015 Production Exports Coffee 8.8 8.9 Tea 9.9 10.1 Pyrethrum 22.4 22.4 Hides & Skins 4.9 4.9 Annual growth rate(YO) Agricultural exports 9.9 Agricultural imports 5.3 Agriculturaltrade surpluses Agricultural trade surpluses inthe base year (1000 $US) 15,474 Agricultural trade surpluses by 2015 (1000 $US) 66,348 Source: IFPRIREMMsimulations, 2006. 66 Summary of Base Run Scenario 162. Business as usual i s not an option if Rwanda i s to meet its national development targets, including the MDG and NEPAD goals. The results o f the Base Run Scenario highlight that a continuation o f current policies will bringabout a modest reduction in the national poverty rate, but the absolute number o fpeople living below the poverty line will actually increase because o f population growth. Food self-sufficiency at the national level will be eroded in the face o f rising food imports, and the trade balance will improve only slightly as increased export earnings stay just ahead of the risingcost o ffood imports. Growth simulations 163. To assess how agriculture can contribute to growth and poverty reduction in Rwanda, a series o f simulations was carried out using the REMM.These simulations, numbering 25 in all, included scenarios based on: (i) projected growth in production o f individual commodities, (ii) projected growth in production o f groups o f commodities, (iii) projected growth in the non- agricultural sectors, and (iv) projected growth inall commodities / groups o f commodities as well as inthe two non-agricultural sectors. 164. The first set o f simulations models the effects o f exogenous increases in the production growth rates o f each commodity considered one at a time, holding the production growth rates o f all other commodities at their baseline levels (i-e., the levels used in the Base Run Scenario). Because the baseline levels o f production o f the various commodities differ, similar increases in production growth rates for different commodities can generate very different levels o f economy- wide growth and impacts on poverty. To assess the extent to which a given commodity can drive economic growth, both the linkage effects on the economy and poverty, as well as the growth potential (determined by supply and demand factors) need to be considered. 165. The exogenous increases in production growth rates used to drive the simulations were generated intwo ways. For about one-third o f the commodities included inthe REMM(12 out o f 30), official national growth targets have been promulgated by the Government o f Rwanda. Typically these official national growth targets appear in the form o f the total production o f a particular crop that i s expected to be achieved ina specified future year, the area that i s expected to be planted to the crop, and/or the average national yield that is expected to be realized. The official national growth targets were used to calculate projected future average annual rates of production growth. For the remaining commodities, projected future average annual rates o f production growth were set equal to or slightly higher than the baseline growth rates (themselves derived from historical trends) to reflect expected future productivity gains attributable to technical change. (Details o f the growth rate projections appear in IFPRI 2006.) Under the various growth scenarios, the increasedproduction growth rates are maintainedthrough 2015. 166. Key insightsemerging from the growth scenario simulations are summarized in Table 30, Table 31, Table 32 and Figure 23 and Figure 24. Growth 167. Ifthe projectedgrowthrates are realized for all commodities, agriculturalGDPwill grow at an average annual rate o f 6.17 percent between 2005 and 2015, roughly double the 3.60 percent average annual rate projected under the Base Run Scenario. Increased agricultural growth will have a pronounced impact on the economy as a whole. Average annual growth in total GDP registered over the 10-year period would rise from 3.88 percent under the Base Run Scenario to 6.24 percent. These projected gains in total GDP growth include the effects o f assumed 4.08 67 percent average annual growth in the manufacturing and services sectors, some o f which i s induced growth linked to growth inagriculture. Figure 23. Annual GDP growth, REMMBase Runvs. Growth Scenario, 2005-2015 3 6.00 v `sPrn5.00 2&8 4.00 3.00 =n al 2.00 C a 4 1.00 0.00 GDP AgGDP NonAgGDP GDPpc AgGDPpc NonAgGDP PC Source: IFPRlREMMsimulations, 2006. 168. The projected agricultural growth rate o f 6.17 percent per year i s the same as the growth rate that has been achieved in Rwanda during the past decade, during a period when the rural economy has been recovering from the effects o f the genocide. This rate i s highcompared to the rates that have been recorded in other developing counties over the longer term (Table 33). While some would question whether agricultural growth in excess o f 6 percent per year can be maintained, it is considered achievable in Rwanda, for at least five reasons. First, the rural economy has not yet recovered completely from the effects o f the genocide. Second, productivity levels for many commodities are currently very low, so there are large yield gaps that can be exploited immediately. Third, the higher yield levels assumed in 2015 are still very modest by international standards. Fourth, ongoing marshlands development activities are providing a strong impetus to agricultural intensification efforts. Fifth, the strategy o f targeting the specialty coffee and tea markets will add significant value to the already important coffee andtea crops even ifthe quantity o fcoffee and tea exports increase only modestly. 68 n 10 9 3 169. Because production and consumption patterns vary among household groups, there i s considerable variability among household groups that i s not evident from the overall (population-wide) average^.^ The sources of growth for different household groups are shown in Error! Not a valid bookmark self-reference.Tuble 34. The following results are particularly noteworthy: Growth in cereals i s very important across all households, contributing between 16.9 and 27.7 percent of income growth. Cereals are extremely important for rural households that do not engage in cash crop production, contributing 26.3 percent to their income growth under Scenario 23. Growth inlivestock i s important generally, contributing 13.6 and 20.1 percent of income growth for the three household groups. Livestock-led growth i s particularly important for Rural Group 3, contributing 20.1 percent of the income growth realized by this group. Livestock-led growth i s relatively more important for male-headed households in general, contributing 18 percent o f the projectedincome growth for these households. 0 Growth inexport crops i s generally speaking much more important for male-headed households than for female-headed households. The importance of export crop-led growth is also positively correlated with land holding size: growth in export crops contributes 21 percent of the projected income growth for Rural Group 3, compared to only 16 percent of the projected income growth for Rural Group 1. Inthe REMM,poverty is measuredbasedonthe income andexpenditures of individualhouseholds, but average income growth rates are measured at more aggregate levels (e.g., national level, household group level). Although they are calculated as an average o f individual household incomes and expenditure measures, the average income growth rates obviously do not represent the income increase for each individual household. Because the production and consumption activities o f individual households differ, some individual households achieve more rapid income gains than others. For example, export crops account for 21 percent o f income growth on average for Rural Group 3, but for the poor households in this group who engage in limited production o f export crops, export crops represent a minor source o f income. Becauseproduction o f export crops i s much more concentrated among a smaller number o f households (especially poor households) than production o f staples, if growth at the national level i s ledby the export sector, fewer households benefit from such growth, and poor households whose income derives mainly from non-cash crop production will benefit less. In considering alternative growth strategies, policy makers therefore need to take into account household-level heterogeneity in production and consumption patterns, since these can lead to big differences between the income effect andthe poverty reduction effect resulting from growth ina specific agricultural sub-sector. 70 c! c? 3 N N N 3 2 3 3 i 2 2 3 2 W gb 0 2 N Incomes andpoverty 170. The simulations make clear that the benefits o f accelerated agricultural growth will be distributed fairly evenly across rural households (Table 30, Figure 24). Under Scenario 23, average annual income growth realized by Rural Groups 1, 2, and 3 are 6.21 percent, 6.33 percent, and 6.34 percent, respectively. Figure 24. GDP growth by householdtype, REMMBase Runvs., Growth Scenario, 2005-15 5 c 5 6.00 5.00 D m 4.00 U ,p 2.00 G53.00 m Ea U 1.00 0.00 Rural HH Rural HH Rural Rural Rural Rural Rural with cash without male- female- Group 1 Group 2 Group 3 crop cashcrop headed headed HH HH Source: IFPRI REMMsimulations, 2006. 171. Within each o f the three rural groups, income growth will vary depending on the demographic characteristics o f the household and the particular mix o f cropping activities in which its members engage (Table34). Gender: Income growth among male-headed rural households will average 6.37 percent per year, compared to only 6.18 percent per year among female-headed rural households. This difference in projected income growth rates, while relatively small (0.19 percent per year), will further exacerbate the existingpoverty gender gap. Cash crops vs. food crops: Income growth among the approximately 90 percent o f rural households that engage in cash crop production will average 6.33 percent per year, compared to only 6.01 percent among the approximately 10 percent o f rural households that engage exclusively in food crop production for home consumption. This indicates that the poverty rates among these two rural household groups will widen. By 2015, the poverty rate among households that engage in cash crop production will have fallen to 74.6 percent, while the poverty rate among households that do not engage incash crop productionwill remain at 88.5 percent. 172. With the income growth projected under Scenario 23, the national poverty rate will fall from 59.2 percent in 2005 to 42.4 percent in 2015, a reduction of 16.8 percentage points. Although poverty rates will fall within all three rural groups, the size of the declines will vary 72 between groups, and by 2015 there will be significant differences in poverty rates among the three groups (Table 30). Within Rural Group 1, the poverty rate will remain stubbornly high at 56.9 percent, whereas in Rural Group 3 the poverty rate will fall to 34.4 percent. As observed under the Base Run Scenario, the benefits o f accelerated income growth will be distributed unequally among different types o f households. Rural female-headed households will benefit less than rural male-headed households; the poverty rate among rural female-headed households will fall from 71.4 percent in 2005 to 54.7 percent in 2015, compared to a decline among rural male- headed households from 61.9 percent in 2005 to 43.6 percent in2015. Similarly, the incidence o f poverty among rural households that do not engage in cash cropping will remain at 74.6 percent, significantly higher than the incidence among rural households that do engage in cash cropping (39.0 percent). 173. The contribution to poverty reduction o f different commodities and commodity groups varies across rural household groups. Because production and consumption patterns vary among households and household groups, there i s considerable variability among household groups that i s not evident from the overall (population-wide) averages. The sources o f poverty reduction for different household groups are shown in Table 34. The impacts of individual commodities or commodity groups in terms o f reducing poverty may or may not be similar to their impacts in terms o f contributing to increased growth. The following results are noteworthy: 0 Growth in food staples i s the dominant source o f poverty reduction. This i s true for all types o f rural households: food crops and livestock account for 35.8 to 46.6 percent o f poverty reduction across all o f the different rural household groups. 0 Growth in pulses and oilseeds and roots and tubers and bananas i s very important for reducing poverty across all households, especially among households that do not engage in cash crop production, accounting for 16.1 and 6 percent o f poverty reduction respectively in these households. This compares to their respective contributions to income growth inthese households o f 7.4 and 4.3 percent, respectively. 0 Growth in cereals i s more important in terms o f poverty reduction for this household group. Growth in cereals production accounts for 26.3 percent o f income growth among households that do not engage in cash crop production, and 19.5 percent o f poverty reduction. 0 Growth in livestock has approximately the same impact on poverty reduction (12.1 - 20.9 percent across the three rural groups) as it does in raising incomes (13.6 - 20.1 percent). 0 Growthinexport crops has approximately the same impact on poverty reduction (20.2 - 22.8 percent across the three rural groups) as it does in raising incomes (19.7 - 23.5 percent). lo lo The contribution of export crops to both income growth and poverty reduction is slightly underestimated, because the REMMfails to account fully for wage income earned by rural households whose memberswork on tea estates. 73 174. The impact on poverty o f different growth scenarios are shown in Figure 25. The greatest amount o fpovertyreduction comes from a balanced combination o f agncultural and non- agricultural growth. Under the Agricultural and Non-Agricultural Growth Scenario, the national poverty rate falls from 59.2 percent in 2005 to 42.4 percent in 2015. Among the various agricultural sector-led growth scenarios, staples-led growth contributes the most to poverty reduction (the national poverty rate falls from 59.2 percent in 2006 to 48.7 percent in 2015). In comparison, exports-led growth has a much more limited impact on poverty, reducing the nationalpoverty rate from 59.2 in2006 to 58.2 percent in2015. I::Figure25. Summary ofPovertyImpacts from differentSimulations2006-15 8 60 [: -Base -Roots Run 81tubers -Pulses -Cereals !!c -Hlgh -Export crops 50 -Livestockvalue crops I! 't: 48 -Staples lncl. livestock -All agriculture -Ag and Non-ag 42 40 2005 2007 2009 2011 2013 2015 Source: IFPRlREMMsimulations, 2006. Food availability 175. Assuming the official growth targets for cereals are met, domestic supplies o f cereals will increase significantly. However the projected production gains will not always be large enough to reduce or eliminate imports. In wheat, production gains will reduce reliance on imports, but imports will still be needed to meet rapidly growing demand. In rice, increased production will allow domestic consumption needs to be met, and imports will cease. Inmaize, production gains will not keep pace with projected strong increases in demand for livestock feed, leading to increased imports. The combined effect o f these changes will be an increase in the overall level o f cereals imports (Table31). 176. The projected increased reliance on imported cereals i s not surprising given the projected strong growth in cereals consumption. Demand for rice and wheat i s highly income-elastic, so rising demand for these two cereals is driven not only by population growth, but also by income gains. Because o f projected strong income growth, per capita demand for rice and wheat are projectedto increase significantly. Per capita demand for rice will rise from 5.2 kg in 2003 to 8.4 kgin2015 (average annual growth o f 5.0 percent). Per capita demand for wheat will rise from 4.5 kg in 2003 to 6.5 kg in 2015 (average annual growth o f 3.9 percent). Demand for maize i s less income-elastic than demand for rice and wheat, but still consumption o f maize will increase very 74 rapidly, driven by steep declines inmaize prices. Per capita demand for maize will double, rising from 11.3 kgin2003 to 24 kgby 2015 (average annual growth o f 7.8 percent). 177. For root crops, the general picture i s one inwhich domestic supplies increase roughly in proportion with growth in domestic demand, with endogenous price effects helping to clear markets. In the case o f potatoes and sweet potatoes, the large production increases implied by official growth targets would soon lead to an excess o f supply over demand, but the projected surpluses are not likely to materialize, because prices for these two crops will fall, causing farmers to shift into other, more profitable crops. Inthe case o f cassava and banana, productioni s projected to grow at rates slightly higher than those implied by official growth targets, driven by increases indomestic market prices. 178. Inthe case of beans,which similar to rice and wheat have a highincome elasticity of demand, production growth i s unlikely to keep pace with consumption growth. Inthe absence o f an official production target for beans, it was assumed that bean production will increase by 3.5 percent per year. With this projected growth rate, bean productionwill increase significantly, but even so, bean imports will rise sharply. Under Scenario 23, bean imports account for only 3.3 percent o f total supply in 2003 (8 thousand tons out o f 248 thousand tons consumed), but this figures rises to 20.9 percent by 2015 (96 thousand tons out o f 463 thousand tons consumed). 179. For most livestock products-milk, beef, eggs, and poultry-afficial targets call for very highrates o fproduction growth. While consumptiono flivestockproducts isprojectedto increase rapidly, not only because current consumption levels are low, but also because livestock products have highincome elasticity o f demand, the ambitious production growth rates impliedby official targets may not be achieved. If they are achieved, however, imports o f livestock products will cease. The only livestock product that is traded in significant quantities is milk. Under Scenario 23, milk production will grow at around 12.5 percent annually between 2005 and 2015, resulting innational self-sufficiency inmilkproductiondespite significantincreases inconsumption. Balance of trade 180. Assuming official growth targets for export commodities are met, and assuming no changes in real international prices for Rwanda's export commodities, the nation's total agricultural trade surplus will increase to about US$ 66 million by 2015, four times higher than the level in2003 (Table32). 181. The overall trade balance could improve even further if efforts to improve the quality o f coffee exports prove successful. Fully washed coffee currently accounts for only three percent o f all coffee exports. If the proportion o f coffee that i s fully washed can be raised to 60 percent by 2015, in line with government targets, this will have the effect o f raising average coffee export prices by three percent annually between 2005 and 2015, since the price o f washed coffee i s about 60 percent higher than the price o f green coffee. Achieving the target for fully washed coffee will result in an additional increase o f US$ 55 million per year in coffee export revenues by2015, further boostingthe total agriculturaltrade surplustojust over US$ 100million. 75 Price trends 182. Mix ofproductive activities: Income growth among rural households will vary depending on the particular mix o f productive activities in which the households engage. Differences in production and consumption growth in individual economies will lead to imbalances in supply and demand for those commodities, with consequent effects on prices that will benefit some households and disadvantage others. In Rwanda, the official production targets for some crops and livestock products are very high, and if the production growth rates implied by the official targets are achieved, supply o f these commodities would soon outstrip demand, leading to strong downward pressure on domestic prices. Falling prices will benefit consumers o f these commodities and hurt producers. For example in the case o f rice, potatoes, and some livestock products (poultry and eggs) projected increases in production are considerably higher than the projected increases in demand. Domestic prices for these commodities therefore are likely to fall, benefiting consumers butresulting inrevenue losses for producers (Figure 26, Figure 27). Figure 26. Projectedprice trends, selected crops, REMMGrowth Scenario, 2005-15 1.4 .-2. v) o 1.3 Cassava k 1.2 v) 88 1.1 e 1.0 ti 0.9 0 a g 0.8 'C u n 'g0.6 f 0.7 n 2005 2007 2009 2011 2013 2015 Source: IFPRIREMMsimulations, 2006. Figure 27. Projectedprice trends, livestock products, REMMGrowth Scenario, 2005-15 z v) .@1.3 - pf m 1.2- Beef C 76 Reaching the Poverty MDG 183. The REMMwas used to evaluate the growth rate that would be needed to meet the first MDG of halving the national poverty rate by 2015. As discussed above, if the government's agricultural growth targets are achieved, Rwanda can achieve the 6 percent per year agricultural growth target established under CAADP. Together with expected growth in the non-agricultural sectors, this will significantly increase incomes and help to reduce poverty. However with this level o f growth, the poverty MDG i s not likely to be met by 2015. The REMM shows that meeting the first MDGwill require overall GDP growth o f 8 percent per year from 2005 to 2015. Although many different combinations o f agricultural and non-agricultural growth could produce overall GDP growth o f 8 percent per year, one possible combination (presented for illustrative purposes) would be 9.0 percent annual growth inagricultural GDP and 7.2 percent annual growth innonagriculturalGDP. 184. Under this hypothetical scenario, the national poverty rate falls to 30.9 percent in 2015, while the rural poverty rate falls to 34.0 percent. However the poverty reduction varies between household groups (Figure 28). Poverty remains much higher in Rural Group 1(42 percent) than inRuralGroup 3 (25 percent). Forhouseholdswithout cashcropproduction, the povertyrate will be cut by only 20 percent by 2015, from its extremely high level o f 89 percent in 2003 to 69 percent by 2015. The povertyrate for rural female-headed household will be as highas 44 percent by 2015, which is only 24 percent lower than in 2003. This suggests that targeted growth and poverty reduction policies will be needed if the majority o f rural households are to share the benefits from growth. For households with very limited landholdings and that are unable to participate high value agricultural production, increasing non-farm employment opportunities may provide the best path out o fpoverty. Figure 28. GDP growth by householdtype, REMMBase Runvs., MDGScenario, 2005-15 3k 90 .-a a0 E .-2 -4 0) 70 60 50 c0 Q) 40 u) c 2 30 3a! 20 lo 8 a - Rural HH Rural HH Rural Rural Rural Rural Rural with cash without male- female- Group 1 Group 2 Group 3 crop cash crop headed headed HH HH Source: IFPRlREMMsimulations, 2006. 77 Summaw of simulations 185. Like any simulation model, the REMMhas limitations, and projections generated using the REMMshould be considered merely indicative. Still, a number of results emerging from the REMMmodeling exercise are compelling, and they have important policy implications. These are summarized below. Growth 186. Agriculture has the potential to be a leading engine of growth for Rwanda's economy over the short to medium term. Based on the assumptions used in the REMM, the agricultural sector i s projected to grow at an average annual rate of 6.17 percent through 2015. Robust agricultural growth will fuel average annual growth intotal GDP of 6.24 percent. Slightly more than one-half of the growth in total GDP projected through 2015 will come from agriculture. 187. Within agriculture, the main drivers of growth will be food staples. Based on the assumptions used in the REMM, some agricultural sub-sectors will grow more rapidly than others. The contribution of each sub-sector to total GDP growth will dependnot only on the rate of growth achieved in that sub-sector, but on the absolute size of the sub-sector. Taking into account the large absolute size of the food staples sub-sectors, most of the growth in agriculture will come from growth inproduction o f food staples (cereals, root crops, oilseeds, and livestock). O f the projectedgrowth inagricultural GDP, over two-thirds will come from projected growth in production of food staples. 188. Export cropswill make a significant contribution to growth. Export crops will make a significant contribution to growth, but the importance of this contribution will be limitedbecause the export crop sub-sector i s small relative to the food crops and livestock sectors. Of the projected growth in agricultural GDP, slightly more than one-quarter will come from projected growth inthe production of export crops. Incomes andpoverty 189. Agriculture-led growth will lead to significant income gains. Based on the assumptions used in the REMM,robust agricultural growth averaging 6.17 percent per year will fuel growth in total GDP of 6.24 percent per year, allowing total GDP per capita to grow at an average annual rate of 3.44 percent, almost triple the rate projectedunder the Base RunScenario. 190. Staple-led growth i s more pro-poor. Growth in the production of food staples (including livestock) has a greater impact in terms o f reducing poverty than growth in the production of export crops. Poverty-growth elasticities" for individual commodities and for 11 The poverty-growth elasticity measures the responsiveness o f the poverty rate to changes in the per capita GDP growth rate. The formula for this elasticity is shown below APO/PO - APO .-GDPpc AGDPpclGDPpc AGDPpc PO where bP0 andAGDPpc are average annual changes (from the base-year) inthe poverty headcount rate and level of per capita GDP; and PO andGDPpcare the base-year poverty headcount rate and per capita GDP. The poverty-growth elasticity measures the percentage change in the poverty 78 commodity groups are shown in Table35. The largest impacts interms of poverty reduction will come from growth inproduction o f crops that are grown by the poor and also consumed by them: beans, pulses and oilseeds, sweet potatoes, and soybeans. Growth in the production of export crops i s more variable, showing considerable differences between export crops. Pyrethrum-led growth shows high poverty-growth elasticity, and coffee-led growth shows moderately high poverty-growth elasticity, whereas tea-led growth shows low poverty-growth elasticity.'* Table 35. Poverty-growth elasticitiesinthe model scenarios Poverty-growth elasticity Scenarios National Rural -1.30 Cereal-led growth (scenario 4) -1.22 -1.23 Maize-led growth (scenario 1) -1.70 -1.70 Rice-led growth (scenario 2) -0.49 -0.50 Wheat-led growth(scenario 3) -0.83 -0.83 Root-led growth (scenario 8) -1.56 -1.58 Cassava-led growth (scenario 5) -1.56 -1.50 Potato-ledgrowth (scenario 6) -1.50 -1.53 Sweet potato-led growth (scenario 7) -2.26 -2.32 Banana-led growth (scenario 9) -1.03 -1.05 Pulses and oilseed-led growth (scenario 12) -2.36 -2.36 Bean-led growth (scenario 10) -2.37 -2.36 Soybean-led growth (scenario 11) -2.16 -2.20 Livestock-led growth(scenario 20) -1.05 -1.07 Poultry and egg-led growth (scenario 17) -1.33 -1.35 Other meat and milk-ledgrowth (scenario 18) -0.98 -1.00 Export crop-led growth (scenario 16) -0.85 -0.87 Coffee-led growth (scenario 13) -1.45 -1.49 Tea-led growth (scenario 14) -0.33 -0.34 Pyrethrum-led growth (scenario 15) -2.97 -3.OS Agricultureled growth (scenario 22) -1.16 -1.18 Source: IFPRIREMMsimulations, 2006. headcount rate caused by a one-percent increase in per capita GDP. This is not equivalent to a percentage point change inthe poverty headcount rate. As mentioned earlier, the contribution of export crops to income growth and poverty reduction is underestimated, because the REMMfails to account fully for wage income earned by rural households whose memberswork on tea estates. 79 Food availability 191. At the national level, food self-sufficiency will decrease. Based on the assumptions used in the REMM, strong income growth combined with population growth will fuel consumption increases that will outstrip production gains inmost food crops. The current deficits in maize, wheat, beans, sugar, and vegetable oils will increase, while those for rice and milk will disappear. For all other commodities, including roots and tubers and livestock products other than milk, production will increase roughly in line with growth indemand, andcurrent levels ofself-sufficiency willbemaintained. Balance of trade 192. Agriculture will help to reducethe trade deficit, but it will not be able to eliminate it completely. Based on the assumptions used in the REMM, Rwanda's total agricultural trade surplus will increase to approximately US$ 66 million by 2015, possibly to $100 million if the strategy of targeting the specialty coffee market proves successful. The overall balance o ftrade will improve, but not sufficiently to eliminate the trade deficit. Price trends 193. Differential production and consumption growth within some agricultural sub- sectors will lead to imbalancesin supply and demand. This will result in price changes that will benefit some households and disadvantage others. For example inthe case o f rice, potatoes, and some livestock products (poultry and eggs), the increases inproductionthat will be achieved if the government's growth targets are met will be considerably higher than the projected increases in demand. Domestic prices for these commodities therefore are likely to fall, which will benefit consumers but is likely to result inrevenue losses for producers Reaching theMDGs 194. Agricultural growth will contribute to the attainment of the first MDG of halving poverty by 2025, but agricultural growth alone will not to be sufficient. With significant focused investments, agricultural growth o f at least 6 percent per year i s achievable, which i s the target growth rate set under the Comprehensive African Agricultural Development Program (CAADP) o f NEPAD. While the goal o f halving poverty by 2015 could be achieved through many possible combinations of agricultural and non-agricultural growth, overall GDP growth would have to exceed 8 percent per year from 2005 to 2015, which i s unlikely to happen unless agriculturalgrowth can reach 9 percent or more. 80 5. SUMMARY AND CONCLUSIONS Principal constraintsto agricultural growth in Rwanda 195. The information and analysis presentedinthisreport indicate that the principal constraint to agricultural growth in Rwanda i s the low profitability of agriculture, both at the farm level but also further alongthe value chain at the levels ofprocessing, storage, andmarketing. The low profitability o f agriculture discourages investment in the sector and leaves the vast majority o f producers trapped in a low input-low output existence characterized by high levels o f persistent poverty and chronic food insecurity. 196. The low profitability o f agriculture in Rwanda results from many underlying causes, of which six stand out: Low farm-level productivity in both the crops and the livestock sub-sectors caused by producers' inadequate knowledge o f improved management practices and compounded by their limited use of land and water management practices that permit sustainable intensification(e.g., irrigation, terracing) Limited availability and high cost of agricultural inputs (especially seed, fertilizer, crop chemicals, and machinery) resulting from poorly developed input distribution systems and exacerbated by hightransportation costs Low and variable prices received for outputs attributable to poorly developed commodity marketing systems and exacerbated by hightransportation costs, particularly for regional and international exports Lack of access to rural financial services needed to support productive investment in primary productionactivities as well as post-harvest value-adding activities Unfavorablebusiness climate that discourages private investment by subjecting firms to costly and time consumingregulatory procedures Weak human capital base, both interms o f skills (attributable to the lack o f education opportunities o f the rural population, particularly opportunities for vocational training) and also in term o f physical capacity (resulting from poor nutrition, inadequate rural health services, and widespread prevalence o f debilitating diseases especially HIV/AIDS) Getting agriculture going in Rwanda: An agendafor action 197. The Government of Rwanda has set itself a target o f sustained growth inagriculture o f 6 percent over the short to medium term. Based on the analysis carried out for this study, and recognizing the constraints identified above, and considering the programs and sub-programs identifiedinthe PSTA, what should be the Government's immediate priorities? 198. The basic conclusiono f this report i s that the main thrusts o f the PSTA are fundamentally sound and that the areas being targeted for reform under the PSTA are consistent with those that have been identified here as requiring attention. At the same time, it must be recognized that the agenda spelled out in the PSTA i s extremely comprehensive. Given the financial and human capital constraints facing MINAGRI and the other organizations charged with implementing the Government's agricultural development agenda, it will not be possible to implement all o f the PSTA Programs and Sub-programs immediately. This suggests that some prioritization i s needed 81 to ensure that the available financial and human resources can have the greatest possible impact inthe short to mediumterminterms ofgrowthandpovertyreduction. Immediate priorities: Agricultural growth uillars 199. Rapid and sustainable growth inRwanda's agricultural sector can be achieved only if the productivity, profitability, and competitiveness o f agriculture can be improved. What i s needed to make this happen? Based on the information and analysis presented inthis report, and taking into account the most binding constraints identified above, the following seven growth pillars should be targeted for immediate attention: 1. Stimulatingproductivitygrowth infood staples (crops and livestock) 2. Scaling up sustainabledevelopment of land and water resources 3. Overhauling nationalagricultural research and extensionsystems 4. Strengthening producer organizations 5. Promoting agriculturalexport growth and diversification 6. Improving the performance of agricultural markets 7. Improving accessto ruralfinancial services GrowthPillar 1. Stimulatingproductivitygrowth infood staples (crops and livestock) 200. Unloclung pro-poor agricultural growth in Rwanda will depend on increasing productivity growth infood staples, both crops and livestock. The REMMmodeling results make clear that most o f the growth in agriculture will come from growth in production o f food staples (cereals, root crops, oilseeds, and livestock). This growth will be strongly pro-poor in that it will disproportionately benefit the lowest income groups, including households that do not grow cash crops and households headed by women. Sustainable intensification o f staple food crop production will not be possible without significant increases in the use by farmers o f purchased inputs, especially improved varieties o f seed; organic and inorganic fertilizers; crop protection practices (e.g. integrated pest management) and more effective and efficient use o f crop chemicals, including pesticides, herbicides, and fungicides; and animal health-related products including vaccines, medications, and nutritional supplements. These inputs will not be available intimely fashion and at affordable prices unless inputmarkets are working well. Strengthening input supply systems, especially for the staple food sector, therefore forms an integralcomponent o f the overall agricultural growth agenda in Rwanda. (See Box 3 for a description o f how input markets were strengthenedinEastern and Southern Africa.) 201. The PSTA recognizes the need to stimulate productivity growth in food staples, and many PSTA sub-programs target activities that will directly or indirectly do so. The most explicit support comes under Program 1: Intensification and Development o f Sustainable Production Systems, which includes Sub-program 12 (Integrated System o f Intensive Agricultural and Livestock Production), and Sub-Program 15 (Supply and Use o f Soil Fertilizers, Inputs and Mechanization). 202. Under Sub-program 12, RARDA has been assigned responsibility for improving the quantity and quality o f livestock products. A key component of the Government's strategy i s the program to provide one cow to every poor family. The livestock promotion strategy includes 82 measures to promote private sector delivery o f veterinary services (e.g. artificial insemination services) and to encourage all farmers to keep some animals, whose number will depend upon the size of the household land. It i s recognized that promoting increasedlivestock productioncan also contribute to increased soil fertility and reduced land degradation, through the production o f manure and zero grazing practices. 203. Under Sub-program 15, the proposal i s to increase fertilizer use by 25 percent per year from the 2005 estimated level of 4 Kgha. Incentives are being provided to encourage private sector engagement in importation and distribution o f fertilizer, as well as in the delivery o f improved extension services to farmers. The goal i s to be importing 15,000 to 20,000 tons o f fertilizer each year by 2008. This year for the first time the Government o f Rwanda entered into an agreement with the Clinton Foundationto set up and pilot a revolving fund to import in bulk 14,000 tons o f chemical fertilizer at cost price from Europe. Farmers' Cooperatives have already paid for the supply o f fertilizers via BRD, and distribution has begun intime for 2006 Season A. The Dutch are also supporting a new project intervention from IFDC aimed at strengthening the role o f the private sector in fertilizer supply and distribution. Sub-program 15 also contains measures to increase the production and use o f seed o f improved varieties. With the National Seed Service now having been integrated into RADA, the public sector will play a key role in ensuring that high-quality, certified seed o f key food crops i s available in a timely and reliable manner. Finally, Sub-program 15 includes measures aimed at obtaining support from development partners to import and locally produce, maintain, and disseminate improved tools and agricultural machinery. 204. The efforts that have been launched under the PSTA to stimulate increased productivity in food staples should be continued and intensified. Key issues that still need to be addressed include: e Strengthening private sector-led fertilizer procurement and distribution systems and clarifiing the role of thepublic sector in thefertilizer sector. e Increasing theproduction and distribution of improved seed andplanting materials. e Improving access to timely and afordable veterinary services. 83 Box 3. Developingruralinput supply systems: Lessons fromMalawi, Kenya, Uganda Duringthe late 1990s, following the introduction o f market liberalization measures, private traders in Malawi, Kenya, and Uganda began to import and distribute agricultural inputs, including seed, fertilizer, and agrochemicals. It soon became apparent that these traders were targeting mainly wealthier commercial farmers, most o f whom could be served through distribution outlets located inand around urban areas. Meanwhile, poorer subsistence-oriented farmers located inrural areas were underserved. The Rockefeller Foundation, worlung inpartnership with local NGOs, launched initiatives in all three countries to develop agricultural input supply pipelines into rural areas. The approach followed was to train, certify, organize, and provide credit to rural input suppliers, with the goal o f improving farmers' knowledge about and access to quality agricultural inputs. The initiative resulted ina marked increase inthe availability and use o f agricultural inputsinrural areas. Several key lessons emerged: 0 Investing in the skills of rural stockists to strengthen input distribution systems can be an effective means of accelerating access by the rural poor to agricultural inputs. Stockists were trained using commercially-delivered training modules to develop their technical knowledge and business management skills. Upon completion, the stockists were certified as "agrodealers." To help achieve economies o f scale insourcing and transporting fertilizer and other inputs, the agrodealers were encouraged to organize themselves at the district level into purchasing groups, with group members providing joint collateral to guarantee the supply o f inputs from the companies. They also were encouraged to form national agrodealers associations, enabling them to better negotiate lower prices and improved credit financing arrangements with the agricultural input supply companies. Joining together into associations has helped the agrodealers to influence government policies relatingto imports, pricing, distribution, and marketing o f agricultural inputs. 0 Offering seed and fertilizer in smaller packs improves their affordability to the rural poor. Agrodealers were encouraged to sell seed and fertilizer in small packages, ranging from 1to 5 kilograms for seed and from 2 to 10 kilograms for fertilizer. This significantly reducedthe cost to farmers, many o f whom had previously been unable to afford the larger quantities o f these inputs available in the market. Regulation o f non-certified stockists by government regulatory agencies was deemed important to prevent adulteration duringthe repackagingprocess. 0 The volume of trade in fertilizers and other agricultural inputs is strongly correlated with the availability of credit. Credit guarantees provided through the initiative were extremely effective in helping input suppliers finance expanded inventories. The credit guarantees eased the capital constraints faced by many rural stochsts and allowed them to significantly increase the range and volume o f fertilizers and other inputs supplied. 84 Growth Pillar 2. Scalingup sustainable development of land andwater resources 205. Unlocking agricultural growth inRwanda will depend on improving the natural resource base on which agriculture depends. Evidence presented in this report on the high rate o f land degradation, soil fertility losses, limited area under irrigation, and low use o f improved water management practices illustrate the urgent need to scale up investment in the sustainable development and management o f land and water resources. 206. The PSTA recognizes the need to scale up development in sustainable development and management o f land and water resource management. Under Program 1: Intensification and Development o f Sustainable Production Systems, three sub-programs address this priority: Sub- Program 11 (Sustainable Management of Natural Resources and Soil Conservation), Sub- Program 13 (Marshland Development), and Sub-Program 14 (Irrigation Development). 207. Under Sub-programs 13 and 14, efforts are being made to implement the national Marshland Development Master Plan. A centerpiece o f the implementation strategy i s the World Bank-supported RSSP, under which marshland development i s being undertaken at a rate now approaching 1,500 ha per year. Additional marshland development work i s being supported or will soon be supported by other development partners, including JICA and African Development Bank. Under Sub-Program 11, efforts are being made to address issues o f land degradation, soil erosion, declines in soil fertility, as well as mitigate damage caused by uncontrolled rain water run-off and better manage water resources to increase agricultural production. By 2005, nearly 10,000 ha o f arable land had been improved through so-called "radical terracing," and in 2006 performance contracts were negotiated at the district level between mayors and the President o f the Republic calling for the development o f an additional 100 ha o f radical terraces per district. Fundingfor this initiative is being provided through the Community Development Fund(CDF). More recently the Government has expressed interest in launching a large-scale land rehabilitationprogram modeledon China's Loess Plateau Project (see Box 4). 208. These and other efforts that have been launched under the PSTA to scale up investments insustainable development of land andwater resources shouldbe continuedand intensified. Key issues that still needto be addressed include: Scaling up sustainable development of land and water resources through investments in marshlands, hillsides and irrigation. Completing the implementation of the land reform agenda. 85 Box 4. Lessonsfromthe Loess PlateauProjectinChina China's irrigated croplands support some o f the most productive farming systems inthe world, but agriculturalproductivity inrainfed areas o f China has lagged. Many o f these rainfed areas are characterized by high population pressure and unsustainable agricultural practices. The Loess Plateau in Northwest China i s one o f the poorest and most environmentally degraded regions inthe country. Early efforts to reverse land degradation on the Loess Plateau involved campaigns to terrace slopes, plant trees and shrubs, and build check dams in gullies to intercept sediment runoff. These interventions were largely unsuccessful, inpart because they were poorly integrated with efforts to raise agriculturalproductivity and farm incomes. Planners and farmers working to improve the productivity o f the Loess Plateau came to the realization that land conservation and rehabilitation measures had to be compatible with sustainable and productive agriculture, as these goals are mutually reinforcing. The World Bank-supported Loess Plateau Watershed RehabilitationProjects (I and 11)were designed with this strategic approach in mind. The two projects, which have now been implemented on 700,000 ha, aim to increase agricultural productivity and incomes and improve ecological conditions in tributary watersheds o f the Yangtze and Yellow Rivers located in Shanxi, Shaanxi, and Gansu Provinces, as well as the autonomous region o f Inner Mongolia. Usinga comprehensive approach centeredon the integratedrehabilitationo f small watersheds, the two projects have transformed formerly abandoned wasteland into productive farmland that i s now being used for sustainable production o f field crops and orchards. Highly erosion- prone sloped lands have been taken out o f crop production and planted with trees, shrubs, and grasses used for fuel, construction materials, and fodder. An important key to the success o f the projects was the use o f a participatory approach that ensured the active involvement o f local officials andpublic institutions. 86 Growth Pillar 3. Overhauling agricultural research and extensionsystems 209. Unlocking agricultural growth in Rwanda will depend on overhauling the national agricultural research and extension systems. Technology-driven increases in agricultural productivity are critically neededto get Rwandanagriculture going. While new technology can be imported from outside the country, the distinctive features o f many o f Rwanda's production systems means that much new technology will have to be adapted to local conditions. For this to happen, existing researchand extension systems will have to be reformed and strengthened, and it will have to be made moreresponsive to clients' needs (see Box 5). 210. The PSTA recognizes the need to overhaul the national agricultural research and extension systems. Under Program 2: Support to the Professionalization o f Producers, two sub- programs target the reform o f research and extension services: Sub-program 22 (Restructuring Services in Proximity to Producers and Rural Innovation), and Sub-program 23 (Promotion o f Research for Development). 211. Under Sub-program 23, efforts are being made to restructure ISAR, RARDA, and RADA, with the goal o freorienting agriculturalresearchactivities and improvingthe delivery o f research products to beneficiaries. With all o f these organizations, the restructuring activities are being driven by the desire to become more client-oriented and demand driven. In the case o f research, I S A R recently created five new outreach centers, and 20 new research program themes have been identified that are believed to be more relevant to farmers' needs. Lack o f scientific capacity remains a problem, however, and although a number o f staff have been trained and continue to be trained, I S A R continues to rely heavily on regionally recruited expatriate staff. In the case o f extension, a well-defined national policy framework i s lacking, and different models are being used under different projects. For example, RSSP has been using farmer-based extension (FBE) methods, and Belgian Technical Cooperation i s supporting the development o f decentralized extension delivery systems in a few districts. The lack o f a clear framework for extension services delivery i s a clear constraint, because it leads to a proliferation o f technical messages and makes it difficult to exploit scale economies in the production and delivery o f extension services. 212. The efforts that have been launched under the PSTA to overhaul agriculturalresearch and extension services should continue and be intensified. Key issues that need to be addressed include: Developing a national client-oriented and demand-driven agricultural research strategy. Developing a national client-oriented and demand driven agricultural extensionpolicy. Securing long-term funding support for public research organizations based on clear public goods rationale and outlining plans to link with the private sector where appropriate. Securing long-term funding support for a pluralistic, decentralized extension sewice delivery system with clear of public andprivate sector roles in the agreed model adopted. 87 Box 5. Stimulating client-oriented, demand driven research: Lessons from Senegal During the 1980s and 1990s, the Senegalese Agricultural Research Institute (ISRA) consistently received poor performance ratings. Review after review concluded that the organization was inefficient, ineffective, and unresponsive to clients' needs. Management reforms, including the introduction o f an incentive scheme designed to reward good performance, remainedpending, mired inred tape. Investments in critical areas such as post- harvest technology and agro-processing were neglected, because they were the responsibility o f a different researchinstitute, the Food Technology ResearchInstitute (ITA). Inan effort to break the impasse, reforms were launched in 2000 with the goal of shifting I S M ' S strategy and operational procedures. Whereas formerly the focus o f the organization had been exclusively on ensuring the supply o f research services, the new strategy called for expanding the focus to include as well supporting the demand for research services. The logic underlying this shift was simple: without strong demand from end-users, it was felt that ISRA was unlikely to make the difficult but necessary reforms needed to ensure the relevance o f its researchto clients. Inline with the new strategy, programs were launched to strengthen the capacity of producer organizations to serve as effective research partners, and performance evaluation mechanisms were introduced to make ISRA more accountable to clients. In addition, a new funding mechanism was devised that allowed core institutional functions to be supported in parallel with competitively chosen researchprojects. Core institutional functions continued to receive earmarked funding, but the National Agricultural Research Fund (NARF) was introduced as an instrument for channeling funds to qualified entities, public and private, to support competitively selected appliedresearchprojects. Today, NARF finances researchproposals submitted through two mechanisms: (1) researcher- developed proposals that are directly related to I S M ' Sor ITA'Sstrategic plans, and (2) client- developed proposals that are received in response to calls-for-proposals issued by NARF on themes identifiedjointly by researchers and end-users. The two types o f proposals undergo the same two-tier screeningprocedure. First, they are screened for scientific quality by a scientific and technical committee comprised o f 15 scientific resource persons (six o f whom come from outside Senegal), and then they are screened for technical soundness and financial viability by a management committee with a majority o f producer organization and private sector representatives. Once a proposal i s approved, NARF signs a contract with the lead research institutionto carry out the work. 88 Growth Pillar 4. Strengthening producer organizations 213, Unlocking agricultural growth in Rwanda will depend on strengthening the capacity o f producer organizations to compete effectively in domestic, regional and international markets. Strong producer organizations will be needed if agriculture i s to move away from traditional subsistence farming towards more market-oriented commercial agriculture. Achieving a successful transformation will depend on farmers' ability to identify actual and potential market opportunities, access cutting-edge technology, procure essential production inputs,produce high- quality products that meet the requirements o f an increasingly quality-conscious market, and negotiate effectively with input suppliers as well as buyers (see Box 6). 214. The PSTA recognizes the need to strengthen producer Organizations. Under Program 2: Professionalization o f Producers, this priority i s directly targeted by Sub-Program 21 (Promoting Farmer Organizations and Strengthening Farmers' Capacities). 215. Under Sub-Program 21, efforts are being made to directly strengthen the technical slulls o f farmers (for example through the establishment o f demonstration plots in all districts and through funding o f learning events in the national agricultural show). More importantly, efforts are being made to strengthen the institutional foundations o f producers' organizations by funding projects designed to launch farmers associations, producers' cooperatives, and other communal bodies and equip them with the knowledge, resources (including financing), technical capabilities, and management skills neededto operate effectively. Many o f these efforts are being led by MINAGRI, sometimes worlung in partnership with MINICOM, but support i s also being providedby numerous development partners working through projects. 216. The efforts that have been launched under the PSTA to strengthenproducer organizations should be continued. However, this area stands out as one where the lack o f a national strategy is a constraint which urgently needs to be addressed in order to ensure that these funds are well spent. Key issues that need to be addressed include: e Working with other public organizations, development partners, NGOs, and private sector groups such as industry organizations and trade associations to develop a national strategy to ensure an efficient, efective, and well coordinated approach to strengthening producer organizations. 0 Usingproject andprogram based approaches to strengthen capacity in key areas. 89 Box 6. Strengtheningproducerorganizations Producer organizations are membership-based organizations or federations o f organizations, with elected leaders positioned to be accountable to their constituents. They take on various legal forms according to national legislation, such as cooperatives, associations, and societies. Their functions can be grouped inthree main categories: 0 Commodity-specific organizations focusing on economic services and defending their members' interests ina particular commodity, such as cotton, coffee, or cocoa; 0 Advocacy organizations that represent producers' interests, such as national producers unions; and 0 Multipurpose organizations that respond to the diverse economic and social needs o f their members, often inthe absence o f local government or effective public services. Producer organizations have expanded rapidly inthe developing world, but existence does not guarantee effectiveness. To be effective, producer organizations must overcome five major challenges, both internal and external to the organization: 1. They mustbe able to resolve conflicts with efficiency and equity. 2. They mustbe able to deal with heterogeneous membership. 3. They must be able to develop managerial capacity. 4. They mustbe able to participate inhigh-level negotiations. 5. They must be able to deal with a sometimes unfavourable external environment. Governments and donors have supported producer organizations, often through specialized NGOs.This support generally takes the form o f buildingsocial capital within the organization to strengthen management capacity. But building social capital is not easy, and there i s no easily replicable blueprint or recipe to ensure success. Any effort to build effective and long- lasting producer organizations must take into account the local circumstances, especially the political environment, and it must be tailored to the legal characteristics o f producer organizations in each country. A guiding principle, however, i s that support should empower producer organizations, rather than positioning them as instruments o f donor or government policy. This i s because donor or government support can be a double-edged sword, potentially empowering the organization, but also potentially creating dependency and thus undermining its autonomy and effectiveness. 90 GrowthPillar 5. Promoting agricultural export growth and diversification 217. Sustaining increased levels o f agricultural growth in Rwanda will not be possible unless the export commodity sub-sectors cab be revitalized and strengthened. The REMM modeling results show that export crops can make a significant contribution to growth, even though this contribution will be smaller than the contribution made by food staples. Immediate attention should be directed to strengthening and deepening the ongoing reforms in the coffee and tea sectors, as these traditional export crops have the greatest potential to contribute to growth inthe short run.Efforts should continue as well to revitalize the pyrethrum industry and to exploit the untapped potential in the hides and skins sector. Inaddition, opportunities should be explored to develop non-traditional export crops, particularly horticultural crops, through launching o f pilot export promotion projects that target the development o f value chains for particular commodities. (For a description o f how this has been achieved inBangladesh, see Box 7). 218. The PSTA recognizes the need to revitalize the export commodity sectors. Under Program 3 (Promotion o f Specialist Crops and Agribusiness), this priority i s directly targeted by three sub-programs: Sub-Program 31 (Creation o f an Environment Conductive to Business and Entrepreneurship Development); Sub-program 32 (Promotion and Development o f Specialist Crops; and Sub-program 33 (Transformation and Competitiveness o f Agricultural and Animal Products). 219. The strategy being pursued under these sub-programs i s to promote commodity chain development for the priority commodities identified in the NAP, both crops (tea, coffee, rice, maize, beans, wheat, oil crops, h i t s , and legumes) and livestock products (milk and meat). Measures to improve competitiveness o f agricultural and livestock products mainly target improved productivity, quality enhancement and meeting grades and standards, improving storage and processing infrastructure reduced cost o f production and ensure stability o f availability for the markets. 220. With support from On the Frontier (OTF) and OCIR Cafk and OCIR ThC, strategies are being implementedto promote exports o f tea, coffee, horticultural crops, and hides and skins. The aim i s to increaseproductivity, improve quality, and add value inthese commodities. The greatest amount o f effort has been directed to the coffee sector; the number o f coffee washing stations has grown dramatically from 10 in 2003 to 79 in 2006, and the proportion o f high-value specialty coffee being exported has increased as a result. In the tea sector, plans to privatize the tea factories are less well advanced, but efforts are underway to upgrade the facilities prior to their sale to private investors. Inthe horticulture sector, cold storage facilities have been constructed at Kanombe Airport, and plans are being developed to add similar facilities at a number o f other strategic locations nationwide. Meanwhile, the National Horticulture Task Force established in 2005 has created the Rwanda Horticulture Development Authority (RHODA), with the aim o f supporting export diversification in horticulture. A proposal by MINAGRI to establish the Rwanda Agricultural Agency for Quality Control (RAAQC) was rejected by Parliament in2006, so quality assurance functions were devolved from MINAGRI to the Rwanda Bureau o f Standards and the Agencies, which are hosted byRARDAand M A . 221. The efforts that have been launched under the PSTA to strengthen export commodities should be continued and intensified. Key issues that needto be addressed include: 0 For allproducts: investments in meeting grades and standardsfor food safety andphyto- sanitary requirements throughout the commodity chain. 91 Continuing coffee sector liberalization through implementation of the coffee strategy and continuing to build international market linkages for the specialty coffee sector. Continuing investment in quality assurance measures and training. Encouraging withdrawal of OCIR Cafb fiom marketing and support theprivatization of cogee washing stations. Accelerating tea sector reforms by speeding the withdrawal of OCIR Thbfrom marketing and privatizing the public tea factories. Investing in tea quality assurance measures. Reviewing the teapricing system, with a view to introducing reforms in order toprovide incentives for tea farmers to deliver high-quality tea. Explore the possibility of developing direct links to specialty marketsfor high quality teaproducts, with a view to avoiding theMombasa market. Supporting the development of the horticulture export sector by introducing an effective public quality certification system that would allow Rwandan products to meet international grades and standards. Finalizing and implementing the Hides and Skins Strategy. Supportingmarket accessfor a refined Pyrethrumproduct. 92 Box 7. Promoting export growth and diversification: Lessons from Bangladesh The government of Bangladesh has identified as key pillars o f its poverty reduction strategy the promotion of agricultural export growth and diversification. It i s pursuingthese goals in part by promoting the development of export-oriented commercial horticulture. However the country has limited experience in commercial horticulture production and marketing, and it lacks the research and especially the extension systems needed to provide farmers with improvedproductiontechnologies and marketing information. To foster the development of the export horticulture sector, an innovative approach has been adopted that uses a semi-autonomous agency to build competitiveness among producers and exporters in the emerging horticulture export industry.The World Bank-funded Agricultural Services Innovationand ReformProject is supporting a quasi-private organization, HORTEX, to pioneer horticultural production and export activities that emphasize contract farming with NGOsor private entrepreneurs usingparticipatoryapproaches to extension. HORTEXis an autonomous, nonprofit, facilitating organization that is not itself commercially involved in production and exports. It i s managed by a Managing Director and a governing body consisting o f 11members (three from government, one from the central bank, five from the private sector, and two from NGOs). HORTEX works with private entrepreneurs, NGOs, and government agencies to forge productive alliances and provide catalytic technical services, combined with limited investment, in export operations that may operate at a loss until adequate export volumes are developed. HORTEX provides assistance in grading, packaging, quality control and export logistics; assists inmarketing; facilitates contracts with overseas interests; and explores potentialjoint venture investments. In1998, its first seasonof successful operation, HORTEXarrangedwith an NGOfor contract production o f French beans, identified interested buyers in Europe, selected appropriate varieties, developed and contracted for supply o f export-quality boxes, trained NGO extension staff, and organized export logistics. HORTEX initially bore the costs o f seed, fertilizer, and cartons as demonstration costs. Based upon analyses o f market opportunities and domestic production possibilities and constraints, HORTEX has identified other potentially profitable products such as chili peppers, okra, bitter gourd, yard long beans, and baby pineapple. Key lessons from the Bangladesh experience includethe following: 0 As farming systems become more complex and market demand becomes increasingly sophisticated, agricultural extension systems must become more decentralized to be able to address location- and market-specific issues. 0 Agencies such as HORTEX that assume a catalytic role should focus on providing critical technical services, rather than making direct investments ininfrastructure. 0 Cooperation between technical agencies with access to international marketing experience and NGOs that have gained the trust o f small-scale farmers can be effective inintroducingnewhigher-value crops inacountrydominatedby small-scale farmers. 93 Growth Pillar 6. Improvingthe performance of agricultural markets 222. Unlocking agricultural growth in Rwanda will depend on improving the performance o f agricultural markets. Evidence presented in this report has documented low rates o f market participation among rural households, the lack o f reliable market outlets for many staples, the highrealmarketingcosts associated inmany caseswith deficient transportationinfrastructure and storage facilities, the high seasonal variability o f commodity prices, and the lack o f access to timely and accurate market information. Rapid and sustained agricultural growth leading to structural change and, eventually, diversification o f the rural economy will not happen without well-functioning markets populated by a dynamic and innovative private sector which in turn implies strengthening the agribusiness sector. 223. The PSTA recognizes the need to improve the performance o f agricultural markets. Under Program 3: Promotion o f Specialist Crops and Agribusiness, this priority i s directly targeted through Sub-program 34 (Rural Infrastructure). 224. Throughout the last few years, considerable investments have been made in the infrastructure needed for the production, transformation, and marketing o f agricultural products. These have focused mainly on rehabilitation o f rural feeder roads designed to link rural production zones to the main trunk road system, as well as targeted investments made at the district level in facilities needed for the collection, processing, storage, and marketing o f crops (rice, maize) and livestock products (milk). Investment in rural roads can lead to increased agriculturalmarketing activity, but the impacts can be variable (see Box 8). 225. The efforts that have been launched under the PSTA to improve the performance of agricultural markets should be continued and intensified. Key issues that need to be addressed include: Working with MINECOFm, MINICOM and MININFRA to strengthen local, regional and international infrastructure to improve access to markets (e.g. road, rail, and air transport links to improve regional and international integration and reduce market access costs). Parallel efforts should be made to reduce administrative barriers that impede thefree flow of gaods,for example by streamlining customsprocedures. Working with MININFRA to improve rural energy and water supply systems (to improve the reliability and reduce the cost of mechanized production and transformation processes). Stepping up investments in the infrastructure needed to reduce marketing costs and increase margins for food crops and exports. Priorities include: (i) food storage, processing and dyingfacilities at district level; (ii) quality market facilities (rural and urban wholesale); (iii) cold chain storagefacilities; and (iv) agro-processingfacilities at district and national levels. Shielding producers from seasonality in commodity prices by promoting private investment in storagefacilities andpiloting warehouse receipts systemsfor strategicfood crops. Developing a private sector-led market information system, using mobile telephone technology. 94 Box 8. Impacts of road infrastructure on markets and productivity Investment in rural r o a d s b o t h construction o f new roads and rehabilitation o f existing roads-can reduce transport costs and generate market activity. It can increase the variety o f goods that households sell inthe market, encourage greater participation intrade and services, generate opportunities for off-farm employment, and boostproductionby reducing the amount of time producers and traders spend travelling. These effects will be mediated, however, by geographic, political and economic factors. In addition, complementary inputs and policies may be required to realize the full potential benefits from roads. And even if aggregate output gains are forthcoming, there will almost certainly be losers too. Whether the gains outweigh the losses i s ultimately an empirical question. Recent work to evaluate the impacts o f investments inrural roads suggests that to be effective as catalysts o f increased marketing activity, rural road projects must be sensitive to local contexts and settings. Policy makers should focus on the complementary role of rural roads. Policy makers often fixate on the supply o f rural roads as a catalyst to development and market activity, instead o f focusing on the complementary role played by roads. Poor road conditions often interact with other constraints to inhibit agricultural productivity and slow economic development; these other constraints typically include poor agro-climatic endowments, low population density, lack of transport services, low education levels, lack o f electricity, high risk, and credit market failures. Policy makers need to consider more than the absence of a road or the dire condition o f a road before deciding that a new road i s critical. Ineach specific case, they should ask whether roads are the right instrument for overcoming the constraints to a given welfare outcome and if so, what other policy initiatives andor investments are needed. Heterogeneities across households will determine who gains and who loses. Within the same community, some households will be better placed to take advantage of a new road, due to their endowments and the nature o f their occupations. Households differ in what they buy and sell and hence how much they stand to gain or lose from price changes induced by better roads. Poor households are more likely to rely on the production o f non-traded goods and services that may actually be displaced by better roads bringingincreasedcompetition. On the other hand, road improvement has a general income effect that can generate increased demand for services from poor providers. The net effect i s an empirical question. Recent impact evaluations provide a mixed picture. InNepal, better road access benefitedthe poor as well as the non-poor, but the gains were proportionally higher for the non-poor. InEthiopia, access to all weather roads in 15 villages reduced the incidence o f poverty by 6.7 percent (Dercon et al. 2007). Given the heterogeneity o f impacts, roads may need to be provided along with other interventions that improve the ability o f certain groups to capture benefits and that protect or compensate groups which may lose. Political economy factors are important in determining impacts. Dependingon how local- level institutions operate, road projects may not end up paying for the goods and services that were originally envisioned, meaning they may have little or no impact. Infrastructure projects present opportunities for graft and the diversion of resources, but it may be possible to reduce these opportunities by careful structuring o f incentives. InIndonesia, the threat o f an audit on road projects significantly increased the actual amounts spent on road construction, thereby bringingthe quality o fthe roads nearer to that originally intended. Insummary, while roads do matter for market development, how much they matter depends on many factors that mustbe taken into account bypolicy makers. 95 Growth Pillar 7. Improving access to rural financial services 226. Unloclung agricultural growth in Rwanda will depend critically on improving access to rural financial services. Current low levels o f investment by financial institutions in the agricultural sector can be explained in part by their perception that lending to rural clients- particularly small-scale producers-is risky and unprofitable. Rural entrepreneurs, not only producers o f primary commodities but also those engaged in post-harvest activities such as processing, storage, and transporting, needresources required for productive investment but have difficulty accessing credit through formal channels, often because they lack legal title to the land that they farm, which for many represents the most obvious form o f collateral. 227. Many investments needed to enhance the productivity o f agriculture are dependent on access to appropriate financial services. At the production level, lack o f financing for agriculture constrains the ability o f farmers to hire labour; purchase inputssuch as seed and fertilizer; pay for machinery services; finance harvesting, storage, marketing activities; pay for extension and information services; bridge the pre-harvest income gap; avoid having to sell immediately following the harvest at low prices; smooth seasonal income flows; and insure against price or yield variability. At the post-harvest level, since agribusinesses have poor access to financial services, this constrains their capacity to finance and supply farmers as well as their capacity to buy andprocess farm produce. 228. The PSTA recognizes the need to improve access to rural financial services. Under Program 2: Professionalization o f Producers, t h s priority i s directly targeted by Sub-program 24 (Rural Financial Systems and Agricultural credit Development). 229. Under Sub-Program 24, the Government i s investing considerable resources in an effort to increase access to agricultural credit and rural finance. The strategy i s to broaden the reach o f rural financial institutions in order to increase the proximity o f affordable financial services to smallholders. (For a description o f how this was achieved inUganda, see Box 9).Measures being implemented include (i) extending agricultural credit lines to commercial banks, specifically BNR and BRD, (ii) supporting refinancing facilities in commercial and development banks; and (iii)establishing a 1.7 billion FRW agricultural credit guarantee fund. Development partners are also contributing to this effort. For example, the World Bank i s supporting a U S $10 million matching grant program through the Rural Investment Facility (RIF), and USAID i s providing partial loan guarantees for rural investments through the DevelopmentCredit Authorities (DCA). 230. The efforts that have been launched under the PSTA to strengthen rural financial services should be continued. Key issues that need to be addressed include: Strengthening rural banking institutions, including micro-finance institutions (MFIs). Oflering training to through farmers associations, producer organizations, and trade associations to build capacity of rural borrowers to accessfinancial services. 0 Piloting innovative rural financial sewices programs, including risk avoidance mechanisms (e.g. insurance). 0 Scaling up successful financial sewices programs, with afocus on improving access to financing, rather than providing credit at subsidized interest rates, which may create distortions andprevent the development of more sustainable market-based solutions. 96 Box9. Buildinginstitutionalcapacity for microfinance: LessonsfromUganda Rwanda, like many developingcountries, has struggled to ensure that rural households have reliable and affordable access to financial services. Rural areas are generally underserved by formal financial institutions owing to the highcost and inherent risk o f providing financial services to mostly small-scale rural clients, who generally lack collateral and must depend on unreliable income from agriculture. While it would be an exaggeration to say that "best practices" have been identified to strengthen the delivery o f rural financial services in Africa, pilot schemes in several countries have identifiedpromising approaches. One such scheme has been implemented in Uganda to strengthen the microfinance sector. Urban andrural entrepreneurs inUganda had very little access to financial services before 1990, but the financial landscape changed significantly after that with the rapid development o f microfinance institutions (MFIs). With strong support from donors, the government played an enabling role by withdrawing from subsidized lending, reforming banlung regulations, and fostering stakeholder collaboration. At the same time, industry-wide training in financial services provision was provided through the Private Enterprise Support Training and Organizational Development (PRESTO) program o f USAID. PRESTO'S Centre for Microfinance, a "one-stop shop" for microfinance information and training, provided training in good practices for micro- lending, offered technical assistance to interested organizations and firms, and administered a business grants program. The result has been a build-upof local capacity among financial institutions, creating a "virtuous circle" in which skilled local and international specialists are being attractedto the Ugandanfinancial sector. MFIs now compete for clients inUganda's urban markets, but rural areas remain largely underserved. The chronic challenge o f reaching rural areas may soon be overcome, since the momentum that has been generated in urban markets i s poised to spill over into rural markets. For example, a finance team from Hewlett-Packard recently began testing a point- of-sale payment application that i s designed to reduce transaction costs in rural areas. The development o f this web-based transaction management application i s facilitated by the presence of a highly trained, skilled human resource base. This application facilitates the provision o f a wide variety o f financial services at low cost, while extending the hours o f service and reducing the distance that rural customers must travel (from 20 kmto less than 5 km). By reducing transaction costs and eliminating certain forms o f risk, the new application i s expected to create additional incentives to borrow and save. The application also creates a customer payment history that can be used for credit decisions, provide risk information, and assist in product development. Loans were also introduced, with collateral at a ratio of 1.5:1 to the loan amount. PRODEMfurther minimizes risk by restricting final loan payments to a maximum o f 60 percent o f the loan amount, and by limiting each office's portfolio ineach economic sector to 30 percent. 97 e e e e e e e e e e e e e e 0 e i -i 0 0 U 0 0 0 0 0 0 0 0 o 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 s L 3z3 .I Y 2 .I 0 .eL & e e e 0 e 0 e . e . e e e 0 2 P kA 0 e e e e e e e e e Delivering on Growth 232. Buildingeffectively onthe seven growthpillarswill require strengthening the institutions and policies that support agriculture. Strengtheningtheinstitutions andpolicies thatsupport agriculture 233. Unlocking agricultural growth in Rwanda will depend on the existence o f strong and responsive institutions-public, semi-autonomous, and private-that have the capacity to provide goods and services, perform essential coordinating and regulatory functions, and ensure the existence o f a policy environment that encourages sustainable pro-poor growth. Evidence presented in this report makes clear that the capacity o f the public institutions charged with overseeing the agricultural sector needs to be strengthened. Many key functions are not currently being carried out effectively either by the public or private sector (e.g. agricultural research and extension, inputs supply and distribution, supply chain coordination, enforcement o f quality grades and standards, monitoring o f food health and safety, export promotion, statistical collection and analysis, policy analysis and planning, etc.) 234. Institutional reforms in the agricultural sector began in earnest in 2005 with Cabinet approval for the creation o f RADA and RARDA, as well as the efforts to restructure ISAR. These efforts have continued in2006 with the move to establish RHODA.Also during200516, the legal and regulatory framework supporting land administration was strengthened. These efforts to formally restructure key institutions were accompanied by efforts to upgrade staff capacity and improve internal processes. These institutional reforms are now beginning to bear fruit in the form o f improved program planning and execution, as reflected by better sector coordination, alignment and reporting against sector wide indicators at joint sector and budget reviews. These initiatives to improve the "soft" aspects o f the institutions involved in agriculture have been accompanied by measures to the "hard" aspects. For example, investments are being made in upgrading ICT capacity inthe sector, 235. InMarch 2006, the Government launched the next phase of its decentralization process by further empowering districts and sectors as the focal points for service delivery. Districts now represent a legal decentralized entity, and they are responsible for overall coordination o f local economic development, and ensuring the coordination o f planning, financing and implementation o f service delivery at sector levels. 236. Under the proposed decentralization framework, beginning in 2007 public agricultural development activities will be supported mainly through earmarked transfers from the central government to districts as well as investments from the government's development budget (Finance Interne). The government's development budget will focus on district level investments for animal disease control vaccines, hillside irrigation and erosion control. Earmarked transfers for agriculture will also be incorporated into the Local Government Budget Support Fund (LGBSF), under the following lines: agricultural extension; animal disease control operations; rural infrastructure; food security: banana and cassava multiplication; water harvesting structures and terracing. 237. Performance targets relating to these areas o f activity will be integrated into the District Performance (Imihigo) Contracts signed between the district mayors and the President. Districts will be responsible for implementation and accountability, while line ministries and other government agencies will play a supporting role with regard to quality assurance, capacity reinforcement, and monitoring and evaluation. It i s envisaged that the relationship between the 106 districts and the various government agencies will be formalized through memoranda o f understanding tied to the Imihigo contracts. The agencies themselves will also sign performance contracts withMINAGRIwhich give performance indicators for a definedperiod. 238. Beneficiaries will be able to provide client feedback on the quality o f service provision from the different service providers. Citizens' report cards will be introduced to enable this evaluation to take place (e.g., through pilots implementedin a few districts for selected services). Since districts will have to report under the Imihigo contracts to the President via MINALOC, their accountability to the agencies and sector will be ensured. The agencies themselves will be accountable to MINAGRI via their performance contracts, while MINAGRI in turn will be accountable to MINECOFIN for monitoring and evaluation o f overall sector outcomes and performance. 239. Ongoing efforts to reform and strengthen agricultural institutions should continue. Key issues that needto be addressed include: Strengthening sector planning, budgeting, coordination, procurement, and M&E (including a Monitoring Information System), especially at decentralized levels. This includes developing and implementing a clear Decentralization Frameworkfor delivery on National Priorities in theAgricultural Sector. Building capacityfor improved service delivery, especially at decentralized levels. Implementing legal and regulatory reform in the sector, With National Institute of Statistics (NIS), developing, using and maintaining robust national agricultural statistics. 107 e, Y 0 d 5:p! k0 a a * a * a * * . a e a a a . e e e e Cross cuttinauolicv issues 240. The current macro policy environment in Rwanda i s generally benign for agriculture. In recent years, Rwanda has had a stable economic climate, a fairly valued exchange rate, a modest rate o f inflation, and reasonably open borders for trade. Hence, unless there are significant changes in this regard, there are no major macro policy constraints distorting the agricultural sector at present. Still, in support o f longer term growth, policies that impact indirectly on agriculture will require attention. These include (i) trade policies; (ii)business and ICT policies; and (iii)family planning, gender, and educationpolicies. 241. Trade policies: Inthe short run, agricultural growth can be driven by domestic demand for food staples, but once domestic demand has been met, any additional growth will be sustainable only if Rwanda can succeed in exporting, particularly to regional markets. Ongoing efforts to promote regional trade integration under COMESA thus take on increasing importance for agriculture. Competition on domestic food staples production i s likely to increase, especially ifTanzaniajoins COMESA. 242. Business and ICT policies: Agriculture inRwanda will not be able to grow without a significant increase in investment, much o f which will have to come from the private sector. An attractive and enabling business climate will be needed to attract this investment. Under the PSTA, MINAGRI i s taking steps designed to create an enabling environment for agribusiness, rural enterprises and export developments. More broadly, the government i s emphasizing export promotion and financial sector development, and it i s establishing a new Export Processing Zone. Developing information communications technology (ICT) to meet business and market informationneeds i s also a priority. 243. Familyplanning,gender and educationpolicies: At current rates o fpopulation growth (2.9 percent per year), the population will double by 2020, increasing demographic pressure on limited resources. The REMMmodeling exercise shows that continuation o f current agricultural policies would result in modest reduction in the national poverty rate by 2015, but the absolute number o f people living below the poverty line will actually increase. This highlights the urgent need for better implementation o f family planning policies. Most local food consumption i s mainly producedby women, therefore equitable access for women to productive assets (e.g. land, credit and labor saving technologies) i s a priority, It can also help to create better linkages to off- farm employment and incomes. Education, particularly for women will be important, not only in terms o f access to employment outside o f the sector, but also with respect to empowerment of reproductive rights. 110 6. REFERENCES Aertssen, G., A. Mutijima, and S. Mbarubugeye. 2006. Rwanda Agricultural Policy Note, Background Study #1: Prospectsfor achieving sustainableagricultural growth inRwanda. Unpublished. Aureille, Y.G., and S. Rwamasirabo. 2006. RwandaAgricultural Policy Note, Background Study #3: Patterns o f Competitiveness and Comparative Advantage inRwanda Agriculture. Unpublished. Dercon, Stefan, J. Hoddinott, and T. Woldehanna. 2007. Shocks and consumption in 15 Ethiopian villages. Journal ofAfiican Economies 14 (4): 559-85. Diao, X, and B, Yu. 2006. RwandaAgricultural PolicyNote, Background Study #4: Agricultural Growthand PovertyReductionOptions inRwanda. InternationalFoodPolicy Research Institute (IFPRI). ForeignInvestment Advisory Service, InternationalFinance CorporationandWorld Bank. 2005. "Sector Study of the Effective Tax Burden, Rwanda." Goessens, F.,and S. Rwamasirabo. 2006. Rwanda Agncultural Policy Note, Background Study #2: Rwanda Agricultural markets Overview. Unpublished. Government of Rwanda. MINISANTE2005. Health Sector Survey, EDSR. Government o f Rwanda. 2004. MINAGRIPlan Stratkgique pour laTransfomation Agricole. Government of Rwanda. MINECOFIN2002a. Integrated HouseholdLiving Conditions Survey EICV (2000-2001). Finalreport. Government o f Rwanda. 2002b. MINECOFINProfile of Poverty inRwanda. (An analysis based on the results ofthe HouseholdLivingCondition Survey 1999-2001) Government o fRwanda. MINECOFIN2003. Core Welfare Indicators Survey (QUIBB). Government of Rwanda. 2001. MINECOFINPoverty Reduction Strategy Paper. Government of Rwanda. MINECOFIN2000. Vision 2020. Final Report. Henao, J., and C. Baanante.2006. Agricultural Productionand Soil NutrientMininginAfrica: Implications for ResourceConservation and Policy Development. Muscle Shoals: InternationalFertilizer Development Center (IFDC). Kelly, V.A., E.Mpyisi,A. Murekezi, andD.Neven, with assistance from E.Shingiro. 2001Fertilizer Consumption inRwanda: PastTrends, FuturePotential, and Determinants. Paperpreparedfor the Policy Workshop onFertilizer Use andMarketing, organized byMINAGRIandUSAID, Rwanda, 22-23 February. Loveridge, S. 1989. The Role of Beans and Sorghum inRwanda's FoodProduction: Current Availability and Projections for the Future.Miscellaneous Paper.East Lansing, MI:Michigan State University. McKay, A., and S. Loveridge. 2005. Income and nutritional outcomes inRwanda's ruralareas, 1990 and 2000. Policy Synthesis No. 74. East Lansing, MI:Michigan State University. Onthe Frontier. 2005, UnpublishedPowerPoint presentation. World Bank. 2006. Rwanda Country Economic Memorandum. World Bank. 2005. Diagnostic Trade Integration Study. 111 ANNEX 1.AGRICULTURAL STATISTICS INRWANDA The quality o f agricultural statistics in Rwanda has varied over the years. During the 1980s and early 1990s, the United States Agency for International Development (USAID) supported MINAGRI's Division o f Statistics (DIS), which included more than 150 staff trained in data collection, processing, and analysis. The DIS conducted annual rural surveys using robust sampling methods and well-designed survey instruments, giving an accurate picture of agriculturalactivities throughout the country down to the household level. Inthe wake of the genocide, formal field data collection activities by the MINAGRIwere abruptly suspended. From 1994-99, no regular annual agricultural surveys were carried out, leaving a major gap in the statistical record. Nonetheless, during this period some efforts were made to monitor what was going on inthe agricultural sector. Working closely with staff from the Famine Early Warning System (FEWS) project, the World Food Program (WFP), and the Program d'Appui a la SkcuritC Alimentaire au Rwanda (PASAR) project, MINAGRI developed and implemented a system for making pre-harvest crop assessments. Multidisciplinary teams visited rural areas and obtained estimates o f land use patterns and production trends. Relying mainly on information provided by key informants-local authorities, MINAGRIfield staff at the province and district levels, and farmers-the teams compared the situation in the current year with that o f the previous year and came up with forecasts o f agricultural production. The pre- harvest crop assessmentswere followed later inthe year bypost-hawest crop assessments, under which information was collected by agricultural officers working at the commune level. The information was aggregated at the provincial level before being sent to MINAGRI's Planning Department for review prior to publication and dissemination. More formal data collection activities resumed in 1999-2000, when a comprehensive agricultural survey covering 1,200 rural households was carried out under the Food Security Project supported by USAID and Michigan State University (MSU). The results o f the 1999-2000 survey were the first accurate agricultural statistics produced using a robust methodology to have appeared since 1994. Unfortunately the 1999-2000 survey proved to be a one-off event, because it was not repeated following the closing o f the Food Security Project in 2002. Pre-harvest crop assessments were revived from 2002-04, but not post-harvest crop assessments. Inany case, the methodology used for these assessmentshave been challenged by data users, many o fwhom have had indicated that they have limited confidence inthe accuracy of the estimates produced through the assessments. Recognizing the need for a much stronger agricultural data base, the government has recently taken steps to restore agricultural statistical collection and reporting capacity. MINAGRIand the National Institute o f Statistics (NIS) under MINECOFIN are now collaborating within the framework o f the new statistical law and decentralization reform. Post harvest rural surveys were undertaken from 2004-06, and since 2005 a bi-annual agricultural survey known as the Enquite Agricole has been implementedinevery cropping season. Inorder to better align agricultural data with other key data sets, beginning in 2006 the sample for the Enqugte Agricole will be a sub- sample o f the Household Living Standards Survey (HLSS, or EICV). Design and management o f the EnquiteAgricole will be the responsibility o fMINECOFIN/NIS. 112 ANNEX 2. DESCRIPTIONOFTHE REMMMODEL 244. The Rwanda Economy-wide Multi-market Model (REMM) is based on neoclassical microeconomic theory. The structure of the model and the initial conditions were designed to reflect the current state of Rwanda's economy in terms of sector and income structure. Key features of the model are summarized below. Structure of the model 245. Because (1) agriculture accounts for more than 40 percent of GDP in Rwanda, (2) approximately 90 percent of the population of Rwanda lives in rural areas, and (3) poverty in Rwanda i s predominantly rural, the REMMhas an agricultural focus. It includes 30 agricultural commodities, which can be classified into eight broad agricultural sub-sectors: (1) grains, (2) bananas, (3) roots and tubers, (4) pulses and oilseeds, (5) export crops, (6) other cash crops, (7) livestock and livestock products, and (8) other food and non-food agricultural commodities. In addition, the REMMincludes two generic non-agriculturalsectors: manufacturingand services. 246. The REMM accounts for regional heterogeneity in production and consumption by disaggregating productionand consumption data for all 30 agricultural commodities into 12 zones representing the 11 former provinces plus the municipality ofKigali. 247. The REMM captures linkages between location, gender, and poverty by disaggregating households into four sub-groups: (1) rural female-headed households, (2) rural male-headed households, (3) urban female-headed households, and (4) urban male-headed households. To allow more refined exploration of poverty-related questions, the REMM further disaggregates rural households into three groups according to landholding size. Landi s one of the most binding constraints for agricultural development inRwanda, and size of landholding i s highly correlated to poverty. For each province, there are thus eight household groups, disaggregated by land holding and household type (Table). Table Al. Householdgroupsdistinguishedinthe REMM Male-headed Female-headed Urban X X Rural RuralGroup1(0.0 -0.5 ha) RuralGroup2 (0.5 - 1.0ha) RuralGroup 3 (More than 1.O ha) Source: IFPRI Calculations@om EICV data, 2006. 248. The REMMincludes separate supply and demand functions for each of the 8 household groups in each of the 12 provinces and for each of the 32 sectors (30 commodities plus manufacturing and services). In total, the REMM thus includes 3,072 production and consumption functions (8 household groups x 12provinces x 32 sectors). 113 SUDPlV 249. Producers are represented in the REMM by the eight household groups (see above definitions), each o f which includes a set o f households who engage in the production o f one or more commodities. Consistent with other multi-market models, supply functions are defined at the aggregate level by household group and are used to capture the group's overall market response. The supply functions are derived under the assumption o f producer profit maximization behavior and are basedon the producer prices o f all commodities (including the prices for the two generic non-agncultural commodities, manufactured goods and services). Risk factors and market imperfections are not taken into account and therefore do not affect producers' decisions. Inthe crops sub-sectors, the supply functions have two components: (i) yield h c t i o n s that are used to capture supply response to the crops' own prices, given the area already allocated to the crop; and (ii) allocationfunctionsthatarefunctionsofallpricesandhenceareresponsivetochanging land profitability across different crops, given the total amount o f available land. 250. The own-price elasticities used in the yield functions are typical values for each crop drawn from other studies, while the cross-price elasticities used in the area functions are calibrated according to the share o f each commodity in total provincial production o f that commodity produced by that particular household group. To account for the fact that land i s in short supply in Rwanda, the price elasticities in the area functions are further restricted by a condition which requires that the total area cultivated be fixed in a given year. Thus any increase in the area planted to a given crop implies an equivalent decrease inthe area planted to one or more other crops. The average elasticities used in the area, yield, and supply functions are reported in IFPRI (2006). The values have been averaged from those defined by province, ruralhrban, gender, and landholding sizes, which vary across these household groups. Demand 251. Demand in the REMMi s disaggregated to the province level and i s estimated based on characteristics o f household groups as well as individual cons~mers.'~Demand i s affected by prices and by per capita income. The data used to calibrate the demand function are taken from the 1999-2001 EICV. Income elasticities o f demand and price elasticities o f demand for specific commodities vary across household groups, reflecting differences in consumption patterns at different income levels. Such differences not only imply that the aggregate effect o f consumers' market responses i s often non-linear and much more complicated than that in the case where demand i s defined at the national level, but also indicate the possible differential effect on poverty reduction with similar income increases at the national level. The average estimated income elasticities o f demand and price elasticities o f demand are reportedinIFPRI (2006). 252. Per capita income within each household group i s endogenously determined in the REMMand depends on the group's production revenue. This feature distinguishes the REMM from many other multi-market models, which tend to be partial equilibrium models. Because per capita income i s endogenously determined as a function o f production revenue, the REMM operates like a general equilibrium model, because production and consumption decisions are linked at the household group level. Since intermediate inputs are not explicitly modeled, however, producer prices must be manually adjusted upwards to reflect post-harvest value addition. These adjustments are made in such a way that the sum o f all agricultural production valued at the adjusted prices i s equal to agricultural GDP (AgGDP).For the two non-agricultural 13 Every household group includes a number o f representative consumers, each o f whom consumes all kinds of goods and services. Herhis demand for each of these goods and services i s derived from maximizing a Stone-Geary utility function. The subsistent level of consumption i s calibrated to the households' home consumption (by household groups). 114 sectors, manufacturingand services, the sectoral GDP i s used to represent production output, with a unit price. Total GDP (at the national level and also at the regional level) thus i s calculated as the sum o f Agricultural GDP (AgGDP) and Non-Agricultural GDP (NonAgGDP), both o f which are endogenous inthe model. Trade 253. Inthe base year of the model (2003), eight commodities were being imported (maize, rice, wheat, beans, vegetable oil, sugar, milk, "industry"), and four commodities were being exported (coffee, tea, pyrethrum, hide~/skins).'~Since trade i s taking place, the REMMmust be able to represent an open economy. To accommodate trade, a multiple market structure i s specified that includes both domestically and internationally produced commodities. For simplicity, these are assumed to be perfectly substitutable inconsumption. Prices 254. Prices o f each commodity vary depending on whether the commodity i s produced domestically for local consumption, imported for local consumption, or produced domestically for export. In the case o f domestically produced and consumed commodities, market-clearing prices are endogenously determined within the model. In the case o f traded commodities, domestic prices are adjusted by transportation and other marketing costs to derive import parity prices (for imports) or export parity prices (for exports). In cases where there i s a structural imbalance between domestic production and consumption o f a given commodity and the domestic market fails to clear, the gap between supply and demand i s made up through trade, with the traded commodity valued at the relevant parity price, which is linked to an exogenously determined international price. The parity prices thus effectively set a limit on domestic price movements. Data used in the model 255. Data used to design the structure o f the REMMand to calibrate the base year values for all variables were drawn from many different sources. Agricultural production data came mainly from MINAGRI (crops) and FA0 (livestock). Import and export data came from MINAGRI, MINECOFIN, FAO, or the relevant commodity associations. Consumption data at the national level were calculated by adjusting production data to account for net exports and are generally consistent withMINAGRIand FA0 data. Price data were taken from Goessens and Rwamasirabo (2006) and from the EICV survey, with price variations between provinces calculated based on data reported inthe latter. Data on the two non-agricultural sectors were taken from the aggregate industrial and services series published inthe World Bank World Development Indicators (2005). Disaggregation o f the non-agricultural data to the level o f the representative household groups was based on EICV data. 256. The base year in the REMM i s 2003, the most recent year for which data were available for most o f the variables o f interest. Since 2003 was known to be a drought year throughout many parts o f the country, agricultural production data for 2003 were compared with agricultural production data for 2000-03. Overall, there was very little difference between the 2003 data and the 2000-03 data (see IFPRI 2006, Table 13), suggesting that while the effects o f the 2003 l4 Due to the negligible economic importance of the so-called "non-traditional export crops" (e.g., flowers, maracuja, essential oils, vanilla), the REMMdoes not include any non-traditional export crops as individual commodities. To the extent that these crops are included inthe statistical category "Fruits & Vegetables," their production contributes in the model to growth, income, and poverty reduction indicators. However foreign exchange earnings derivedfrom exports of these crops are ignored. 115 drought may have been severe inmany locations, production measuredat the aggregate (national) level was affected less than i s normally believed. 257. Future changes in the area planted to individual crops, crop yields, and livestock production were projected basedon data from the EICV, MINAGRI, MINECOFIN, and FAO, as well as from AGRER (2006) and Goessens andRwamasirabo (2006). Performance indicators 258. Four variables generatedby the REMMare of interest as performance indicators. 259. Growth is measured as the average annual percentage increase in the total value of production. Overall GDP growth can be disaggregatedto show the contribution to growth made by individual sectors (agriculture, industry, services), by individual regions, by individual crops and commodity groups, etc. 260. Poverty i s measured as the proportion of people whose income i s below the nationally defined poverty line of RWF 64,000 per adult per year. Income i s measuredbased on household total expenditure data from 1999-2001 EICV, which are used to develop a micro-simulation modelthat capturesdetailed household consumption patterns. 261. Food self sufficiency is measured as the proportion of domestic food consumption requirementsthat canbemet from domestic production. 262. Foreign exchange earnings from agriculture are measured as the total value of agriculturalexports. 116