AGRICULTURE GLOBAL PRACTICE TECHNICAL ASSISTANCE PAPER RWANDA AGRICULTURAL SECTOR RISK ASSESSMENT Åsa Giertz, Mohinder S. Mudahar, George Gray, Rhoda Rubaiza, Diana Galperin, and Kilara Suit WORLD BANK GROUP REPORT NUMBER 96290-RW OCTOBER 2015 Agriculture Global Practice Technical Assistance Paper RWANDA Agricultural Sector Risk Assessment Åsa Giertz, George Gray, Mohinder S. Mudahar, Rhoda Rubaiza, Diana Galperin, and Kilara Suit © 2015 World Bank Group 1818 H Street NW Washington, DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org Email: feedback@worldbank.org All rights reserved This volume is a product of the staff of the International Bank for Reconstruction and Development/The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. World Bank Group encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright ­ Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone: 978-750-8400, fax: 978-750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, World Bank Group, 1818 H Street NW, Washington, DC 20433, USA, fax: 202-522-2422, e-mail: pubrights@worldbank.org. Cover images courtesy of: Traci Johnson and Åsa Giertz, both of World Bank. Contents Acronyms and Abbreviations vii Acknowledgments ix Executive Summary xi Chapter One: Introduction 1 Chapter Two: Agriculture in the Economy 5 Rwanda in the 21st Century 5 Role of Agriculture in the Economy 6 Climate and Agro-Ecology 9 Agricultural Land and Input Use 10 Chapter Three: Agriculture Sector Risks 13 Food Crops 13 Export Crops 21 Livestock (Dairy and Meat) 27 Regulatory Changes in the Agriculture Sector 32 Chapter Four: Adverse Impacts of Agricultural Risk 35 The Methodology Used to Estimate Production Losses 35 Particularly Vulnerable Groups 41 Chapter Five: Risk Prioritization and Management 45 Risk Prioritization 45 Risk Management 46 Potential Interventions for Agricultural Risk Management in Rwanda 47 References 53 Appendix A: Stakeholder Feedback 59 Appendix B: Climate Change in Rwanda 63 Appendix C: Vulnerability Assessment 69 Appendix D: Detailed Calculations of Provincial Losses 75 Appendix E: Food Crop Supply Chain Analysis 79 BOXES BOX 2.1: Rwanda’s Crop Intensification Program 11 BOX 3.1: Two Actors in the Tea Supply Chain 22 FIGURES Figure ES.1: Agricultural Value Added (annual % growth), 1995–2012 xii Figure ES.2: Frequency and Scope of Losses per Crop, 1995–2012 xiii Agricultural Sector Risk Assessment iii Figure ES.3: Crop Losses in Rwanda by Province (annual, US$, sum of Season A and Season B) xiv Figure 1.1: Agriculture Sector Risk Management Process Flow 3 Figure 2.1: Rwanda’s GDP Growth Compared with SSA and East Africa 6 Figure 2.2: GDP Composition, 2013 6 Figure 2.3: Agricultural Value Added (annual percent growth), 1980–2012 7 Figure 2.4: Gross Production Index Number, Crops and Livestock (2004–06 = 100), 1995–2012 7 Figure 2.5: Sectoral Composition of Rwanda’s Agriculture Sector GDP 7 Figure 2.6: Crops’ Shares of Gross Agricultural Production Value (current, average 2009–11) 8 Figure 2.7: Yield Gaps Compared with SSA, 1980–89 and 2000–09 8 Figure 2.8: National Herd Profile of Livestock (TLUs, ’000 of animals), 2005–11 8 Figure 3.1: Maize Yields (MT/ha), 1995–2012 15 Figure 3.2: Paddy Rice Yields (MT/ha), 1995–2012 15 Figure 3.3: Banana Yields (MT/ha), 1995–2012 16 Figure 3.4: Beans Yields (MT/ha), 1995–2012 16 Figure 3.5: Cassava Yields (MT/ha), 1995–2012 16 Figure 3.6: Irish Potato Yields (MT/ha), 1995–2012 16 Figure 3.7: Sweet Potato Yields (MT/ha), 1995–2012 16 Figure 3.8: Rwanda Food Crop Prices (RF), January 2005–September 2013 20 Figure 3.9: Rwandan Prices of Irish Potatoes (RF/kg), January 2005–September 2013 20 Figure 3.10: Tea Yields in Rwanda (MT/ha), 1995–2011 22 Figure 3.11: Tea Production (MT) and Area (ha), 1997–2013 22 Figure 3.12: Coffee (green beans) Yields (kg/ha), 1995–2012 23 Figure 3.13: Coffee Production (MT) and Area (ha), 2005–13 23 Figure 3.14: Monthly Prices of Rwandan Tea at the Mombasa Auction (U.S. cents/kg), February 1989–October 2013 24 Figure 3.15: Volume (MT) and Value (US$ million) of Rwandan Tea Exports, 2000–13 25 Figure 3.16: Coffee Prices in Rwanda (US$/kg), 2000–03 25 Figure 3.17: Monthly International Coffee Price (U.S. cents/lb), January 2000–July 2013 25 Figure 3.18: Fluctuations in the Volume of Coffee Exports from Rwanda (MT), 2005–13 26 Figure 3.19: Fluctuations in the Value of Coffee Exports from Rwanda (US$ million), 2000–13 26 Figure 3.20: Average Weekly Exchange Rate of RF/USD, January 2008–January 2014 26 Figure 3.21: Systemic Losses to Milk Production (whole, fresh, cow) (MT), 1990–2011 27 Figure 3.22: Systemic Losses to Beef Production, 1990–2011 28 Figure 3.23: Monthly Retail Price Variability of Fresh Milk (RF/liter), January 2005–December 2013 30 Figure 3.24: Meat Prices (US$/MT) and Meat Production (MT), 1996–2011 31 Figure 4.1: Example of Crop Production Loss Calculation 36 Figure 4.2: Losses and Growth in Agricultural Value Added, 1995–2012 37 iv Rwanda Figure 4.3: Agricultural Production Losses and Share of Total Production Value in Specific Years 38 Figure 4.4: Frequency and Scope of Losses per Crop, 1995–2012 38 Figure 4.5: Rwanda’s Five Provinces 39 Figure 4.6: Annual Losses by Province (US$, sum of Season A and Season B) 39 Figure 4.7: Households with Unacceptable Levels of Food Consumption 42 Figure 5.1: Strategic Risk Instruments According to Risk Layers 46 Figure A.1: Prioritization of Possible Interventions: Food Crops 60 Figure A.2: Prioritization of Possible Interventions: Export Crops 60 Figure A.3: Prioritization of Possible Interventions: Livestock 61 Figure B.1: Average Monthly Rainfall and Temperatures in Rwanda, 1960–90 64 Figure B.2: Hot Days, Rwanda, 1961–2000 66 Figure B.3: Hot Days, Rwanda, Projected for 2046–65 66 Figure C.1: Distribution of Food Insecurity in Rwanda, 2012 70 Figure C.2: Food Security and Livestock Units in Rwanda, 2012 70 Figure C.3: Food Security and Number of Crops Cultivated in Rwanda, 2012 70 Figure C.4: Livelihood Zone Mapping in Rwanda 72 Figure E.1: Retail Price Variation in Domestic Markets for Bananas 82 Figure E.2: Price Movements in Domestic Markets for Beans 87 Figure E.3: Monthly Retail Prices of Cassava Flour, 2012 and 2013 89 Figure E.4: Rainfall Anomalies for the March-April-May Period (Season B), by Province 91 Figure E.5: Seasonal Variation in Retail Prices of Maize in Different Markets 93 Figure E.6: Variation in Domestic Market Prices of Irish Potatoes 96 TABLES Table ES.1: Cost of Adverse Events for Crop Production, 1995–2012 xiii Table ES.2: National Risk Prioritization Matrix xv Table ES.3: Proposed Solutions Areas for Agricultural Risk Management in Rwanda xvi Table 2.1: Value of the Top Five Agricultural Export Commodities, 2008–10 (average) 9 Table 2.2: Weather and Crop Seasons in Rwanda 9 Table 2.3: Agricultural Area and Area per Household (ha), by Province, 2010 10 Table 2.4: Agricultural Holdings by Size (%) 10 Table 2.5: Topographic Position of Farms in Rwanda, 2010 11 Table 2.6: Agricultural Households’ Use of Agricultural Inputs 11 Table 3.1: Main Pests and Diseases of Selected Food Crops in Rwanda 19 Table 3.2: Impact of Drought and Dry Spells on Milk Production, Select Years 28 Table 3.3: Total Number of Livestock-Related Disease Outbreaks, 2002–12 (average) 29 Table 3.4: Average Number of Disease Outbreaks Annually in 2002–11 versus 2008 and 2012 30 Agricultural Sector Risk Assessment v Table 3.5: Summary of Regulatory Changes in Rwanda’s Agriculture Sector, 2001–13 33 Table 4.1: Summary of Indicative Production Losses for Rwanda’s Food And Export Crops, 1995–2012 36 Table 4.2: Cost of Major Adverse Events for Crop Production, 1995–2012 37 Table 4.3: Production Volatilities by Province (CVs of yields, %) 39 Table 4.4: Poverty in Different Groups of Households, 2000/01 versus 2010/11 41 Table 4.5: Percentage of Households That Grow Specific Crops and Share of Production Sold on Markets, 2012 43 Table 4.6: Sources of Food and Food versus Nonfood Expenditures, 2012 43 Table 4.7: Gender Division of Crop Cultivation for Different Districts 44 Table 5.1: National Risk Prioritization Matrix 46 Table 5.2: Potential Interventions for Risk Management in Rwandan Agriculture 47 Table B.1: Ideal Growing Conditions for Selected Crops in Rwanda 64 Table C.1: Poverty in Different Groups of Households, 2000/01 and 2005/06 69 Table C.2: Percentage of Households That Grow Specific Crops and Share of Production Sold in Markets 71 Table C.3: Sources of Food and Food versus Nonfood Expenditures, 2012 71 Table C.4: Gender Division of Crops Cultivation for Different Districts 72 Table D.1: Banana Production by Province, 1998–2012 (Season A) and 2000–12 (Season B) 75 Table D.2: Maize Production by Province, 1998–2012 (Season A) and 2000–12 (Season B) 76 Table D.3: Cassava Production by Province, 1998–2012 (Season A) and 2000–12 (Season B) 76 Table D.4: Irish Potato Production by Province, 1998–2012 (Season A) and 2000–12 (Season B) 77 Table E.1: Yields and Total Production of Irish Potatoes by Province in Season A, 2012 94 vi Rwanda Acronyms and Abbreviations Acronym Definition Acronym Definition ARMT Agricultural Risk Management Team IPCC International Panel on Climate Change (of the World Bank) IPM Integrated Pest Management BBTD Banana bunchy top disease ITCZ Inter Tropical Convergence Zone BBW Banana bacterial wilt LSD Lumpy skin disease BCMV Bean common mosaic virus LWH Land Husbandry, Water Harvesting and BXW Banana Xanthomonas wilt Hillside Irrigation Project CAADP Comprehensive Africa Agriculture MCMV Maize chlorotic mottle virus Development Programme MINAGRI Ministry of Agriculture and Animal CBD Coffee berry disease Resources CBPP Contagious bovine pleuropneumonia MINECOFIN Ministry of Finance and Economic Planning CBSV Cassava brown streak virus MLND Maize lethal necrosis disease CIP Crop Intensification Program MT Metric ton CLR Coffee leaf rust MT/ha Metric tons per hectare CMV Cassava mosaic virus NAEB National Agricultural Export Board COMESA Common Market for Eastern and NAES National Agricultural Extension System Southern Africa NISR National Institute of Statistics of CV Coefficient of variation Rwanda DRC Democratic Republic of Congo OIE World Organisation for Animal Health EICV3 Integrated Household Living Conditions PSTA III Strategic Plan for the Transformation of Survey 3 Agriculture in Rwanda FAO Food and Agriculture Organization RAB Rwanda Agricultural Board (of the UN) REMA Rwanda Environment Management FAOSTAT FAO Corporate Statistical Database Authority FMD Foot and mouth disease RF Rwanda franc FOB Free on board SECO Swiss Secretariat of Economic Affairs GAPs Good agricultural practices SSA Sub-Saharan Africa GDP Gross domestic product TLU Tropical livestock unit G-8 Group of Eight USAID U.S. Agency for International GoR Government of Rwanda Development ha hectare USD U.S. dollar IFPRI International Food Policy Research WDI World Development Indicators Institute WFP World Food Programme Agricultural Sector Risk Assessment vii Acknowledgments This report was developed by a team led by Åsa Giertz, Agricultural Specialist from the Agricultural Risk Management Team at the World Bank, and comprising Mohinder S. Mudahar, Diana Galperin, George Gray, Rhoda Rubaiza, and Kilara Suit. The activities were supported by Carlos Arce, Srilatha Shankar, and Yasmine ­Umutoni. Amy Gautam edited the report with help from Traci Johnson. The team is grateful for the leadership and coordination received from Vikas ­Choudhary, Mark Austin, and Valens Mwumvaneza, and also Severin L. Kodderitzsch, Practice Manager (GFADR) and Carolyn Turk, Country Manager, Rwanda Country Office. The team would like to extend its appreciation to the Rwanda Ministry of Agriculture and the stakeholders from major agricultural supply chains who participated at vari- ous moments during the field work and during the workshops to discuss the findings. Their active participation obliged the team to be realistic and practical. Funding for this activity was provided by USAID, without which this activity would not have been possible. Supplemental financing was provided by a Multi-Donor Trust Fund (MDTF) supported by the Ministry of Foreign Affairs of the Government of the Netherlands and State Secretariat for Economic Affairs (SECO) of the Government of Switzerland. Agricultural Sector Risk Assessment ix Executive Summary Background Rwanda has experienced a remarkable recovery since the civil war, with high growth since the mid-1990s; gross domestic product (GDP) has grown 10 percent per year on average. Agriculture is the dominant sector of the economy, contributing a third of the country’s GDP and about half of Rwanda’s export earnings. Because about 80 ­ percent of the population lives in rural areas and is engaged in agriculture to some extent, increasing agricultural productivity is key to improving incomes and decreasing pov- erty. The government of Rwanda (GoR) has therefore made agricultural development a priority and allocated significant resources to improving productivity, expanding the livestock sector, promoting sustainable land management, and developing supply chains and value-added activities. As a result, the sector grew an average 5 percent per year over 2002–12, which is rather high although it fell short of both the government’s own objective of 8–9 percent annual growth for the period (revised to 8.5 percent for the next years in the new Strategic Plan for the Transformation of Agriculture in Rwanda, PSTA III) and of the Comprehensive Africa Agriculture Development Pro- gramme (CAADP) commitment of 6 percent growth in the agriculture sector. At the same time, Rwanda’s agriculture sector faces a series of challenges. Agricultural land plots are very small (80 percent of land holdings are less than 1 hectare [ha], often divided into three to four plots), and over 70 percent of agricultural land is either on hills or on the side of hills. Agriculture is dominated by small-scale, subsistence farming under traditional agricultural practices and rain-fed agriculture. As a result, average crop yields are low compared with potential yields, and crops are exposed to risks such as weather-related shocks and pest and disease outbreaks. Current agricul- tural policies are geared to increasing productivity in the sector by achieving scale in agricultural production. Risks can potentially have significant implications on stakeholders, investments, and development in the agriculture sector. Adverse movements in agricultural commod- ity and input prices, together with production-related shocks (for example, from weather, pests, and diseases), not only affect farmers and firms active in particular Agricultural Sector Risk Assessment xi supply chains, but may also put severe strains on a govern- A consultative stakeholder meeting organized by the Min- ment’s resources. Rapid or significant declines in produc- istry of Agriculture and Animal Resources (MINAGRI) tion and/or trade may reduce government tax revenues, was also held in Kigali to obtain feedback on findings and affect balance of payments, necessitate compensatory (or to discuss areas for risk solution interventions for deeper recovery) expenditures, and/or otherwise adversely affect analysis. a government’s fiscal position. The prevalence of “shock- recovery-shock” cycles vastly reduces the ability of many Risks in Rwanda’s countries to plan for and concentrate on real development issues. The purpose of this report is to assess existing risks Agriculture Sector to Rwanda’s agriculture sector, prioritize them according Compared with many other countries in the region, to their frequency and impacts on the sector, and identify Rwanda is not subject to frequent shocks of large scale, areas of risk management solutions that need deeper spe- such as national droughts or locust events. Still, risks have cialized attention. important consequences for agricultural productivity and growth. Although many countries in Sub-Saharan Africa Methodology (SSA) experience recurring negative agricultural growth because of various shocks, Rwanda has had only one The report takes a quantitative and qualitative approach to year of negative growth in the 20 years since the war in risks and analyzes their impacts on those agricultural com- the early 1990s (figure ES.1): in 2003, agricultural value- modities that jointly make up the top 80 percent of agri- added growth was negative because of a drought that hit cultural production value (cassava, maize, Irish potatoes, the country. On an annual basis, production losses for food sweet potatoes, plantain, beans, rice, and milk and beef) plus and export crops averaged US$65 million between 1995 coffee and tea because of their importance as export crops. and 2012, or about 2.2 percent of Rwanda’s total annual Production risks are quantified in terms of losses and then agricultural production value. Instead, risks, especially mapped by different perils. Marketing and enabling envi- those of pests and diseases, are pervasive in Rwanda and ronment risks are analyzed qualitatively. For the purpose of although they don’t cause large deviations from general this assessment, risk is defined as the possibility that an event yield trends at a national scale, their impacts on produc- will occur and will potentially have a negative impact on the tion likely explain part of Rwanda’s yield gaps. Agricul- achievement of a farm’s or firm’s performance objectives tural risks can thus have an important impact on growth and/or successful functioning of the overall supply chain. objectives and on the government’s efforts to transform In the work previously conducted by the World Bank’s Agri- the sector. cultural Risk Management Team (ARMT) in other coun- tries, time periods of at least 30 years are assessed to secure a proper understanding of the risks to the sector. However, because of the very different systems in Rwanda before and Figure ES.1. Agricultural Value after the mid-1990s, as well as the interruptive civil war in Added (annual % 1994, it is difficult to identify distinct trends over longer time growth), 1995–2012 series. This assessment therefore focuses on risk to the agri- 40 culture sector over the past 20 years. 30 20 Drought To estimate production losses, this report quantifies nega- 10 tive deviations from medium- to long-term yield trends 0 that are greater than what can normally be expected in 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 –10 agricultural production. The value of the loss is then estimated in local producer prices. A broad spectrum of –20 stakeholders was consulted throughout this work, includ- –30 ing the Rwandan government, farmers, traders, proces- –40 sors, cooperatives, agricultural institutions, and academia. Source: WDI 2013. xii Rwanda Figure ES.2. F  requency and Scope of Losses per Crop, 1995–2012 600 Cassava 500 Losses per crop (m US$) 400 Plantains 300 200 Irish potato Sweet potato 100 Beans Green coffee Rice Maize Tea 0 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 –100 Frequency of losses Sources: FAOSTAT; Authors’ calculations. Food crops in Rwanda are mainly subject to production- Table ES.1. C  ost of Adverse Events for related risks whereas export crops are mainly exposed to Crop Production, 1995–2012 market risks. Pests and diseases pose a risk to Rwanda’s Indicative Loss Value food crop producers in particular because food crops have fewer organized supply chains and less access to preven- % Ag. Production tive inputs than export crops. However, coffee producers Value especially suffer from pests and diseases that have impacts (current, on yields and on market access. Weather-related risks are average Causes/Risk less of a concern for food crop producers and the main Year US$ 2009–11) Events impacts on production are from moisture stress caused by 2001 138,241,657 –4.57 Excessive rainfalls in erratic rainfall. Whereas this is true also for export crops, the Northern and the risk is relatively smaller than that of market risks, Western Provinces because international price volatility poses a significant 2004 150,078,184 –4.96 Heavy rains in high risk to the export sector. For food crops, marketing risks altitude areas and a drought in Eastern are limited, with prices responding predictably to seasonal and Southern supply and demand, and domestic prices are not affected Provinces by global prices. 2006 87,062,028 –2.88 Drought/high heat in Eastern and Currently, the crops most exposed to risks are cassava Southern Provinces and plantain, followed by fairly evenly distributed losses 2007 238,236,805 –7.87 Drought in Eastern between Irish potatoes and sweet potatoes (figure ES.2). Province 2008 269,030,202 –8.89 Drought in Eastern Maize has relatively frequent losses but the losses are not Province as large as for the first four crops. The scope of the losses Sources: FAOSTAT; Authors’ calculations. are clearly in line with the importance of the crop in the Note: Plantain, tea, and coffee were calculated from 1980 through 2011 owing total sector, because cassava, plantain, potatoes, and maize to limited data. Cassava, paddy rice, sweet potatoes, maize, dry beans, and Irish dominate agricultural production in terms of value. potatoes were calculated from 1980 through 2012. Because pests and diseases are endemic in nature and out- In years with these events, Rwanda experienced up to breaks are not visible in the national-level yield data, the 9.5 percent losses of total agricultural production value. biggest losses tend to be correlated with difficult weather Nevertheless, omnipresent pests and diseases, including events, that is, drought or excessive rainfall (table ES.1). beanflies, the antestia bug, cassava mosaic virus, coffee Agricultural Sector Risk Assessment xiii Figure ES.3. C  rop Losses in Rwanda slightly higher losses than the other provinces, followed by Province (annual, by the Northern, Southern, and Eastern Provinces and the City of Kigali. Although the Northern Province has US$, sum of Season A the highest aggregate losses in absolute amounts, it is clear and Season B) that the geographic target area for any risk-management Irish potato Maize 30,000,000 Cassava Bananas intervention will depend on the crop. 25,000,000 Livestock is important to Rwandan households, both in terms of income and food security and for the organic 20,000,000 manure produced, which is applied in the fields. Half of all households own a goat, cow, and/or chicken, and of US$ 15,000,000 livestock units, 68 percent are cattle. Thus, this report 10,000,000 looks at the risks to milk and beef production. The pro- 5,000,000 duction of milk and beef has increased dramatically in Rwanda over the past two decades, in part because of 0 government-financed livestock production programs, and North West South East Kigali Sources: Authors’ calculations, based on NISR’s 2014 Seasonal Agricultural in part because of increased incomes that drive consumer Survey. demand for livestock products. From being an importer, Rwanda is now essentially self-sufficient in milk products. leaf rust, and the more recently introduced banana bacte- The key risks for the milk value chain occur first at the pro- rial wilt, are widely spread in Rwanda, causing yield losses bulking/ duction level, then at the marketing level (that is, ­ ranging from a third up to 100 percent in infected plants. collecting and transporting), and finally at the retailing Thus, more systematically mitigating risks of pests and stage. The risks for meat production are mainly related diseases would likely affect the general yield trend and to production. Since the mid-1990s, milk production has narrow the yield gap for crops that currently have yields been affected by droughts and livestock disease outbreaks, much lower than potential yields. such as anthrax, lumpy skin disease (LSD), and foot and mouth disease (FMD). Livestock disease outbreaks in This report looks at indicative crop production losses for 2008 caused a 13 percent loss in milk production in com- Rwanda’s five provinces: Northern, Western, Eastern, parison with the previous year’s production and cost an Southern, and the City of Kigali. Losses were estimated estimated US$10 million in lost income for farmers and for maize, bananas, cassava, and Irish potatoes for 2000– US$163,0001 in the value of destroyed, slaughtered, or 12 using MINAGRI’s disaggregated data. Results indicate dead cattle. Meat production is also affected by drought, that losses are the greatest in absolute terms in the North- albeit with a lagging effect, because production declines ern Province and smallest in the City of Kigali (which also are visible only a year after the drought’s occurrence. Nev- produces much less than the other provinces). Figure ES.3 ertheless, the impacts are limited and cannot be compared provides an overview of the value of annual losses per with those on milk production. province for Irish potatoes, cassava, maize, and bananas. The bulk of the losses of Irish potatoes are in the North- Monthly milk prices are excessively volatile in Rwanda. ern Province but a large amount is also incurred in the In general, this kind of price volatility can occur when Western Province. Most of the cassava losses take place in the Southern Province, followed by the Western and East- ern Provinces. Banana losses are more evenly distributed 1 Based on World Organisation for Animal Health (OIE) data, the “Dairy Value between provinces, but the Eastern Province has slightly Chain in Rwanda” report, and the NISR Statistical Yearbook 2012. The report estimates the value of an exotic bull to be RF 500,000, which is also assumed to higher losses than the others, whereas the Western Prov- be the average value of a milking cow. The total number of destroyed, slaugh- ince has the lowest. Maize production has the lowest losses tered, and dead cattle was multiplied by the estimated value in Rwanda francs in absolute terms, whereas the Western Province sees (RF) and then converted to U.S. dollars (USD). xiv Rwanda Table ES.2. National Risk Prioritization Matrix Impact/Probability of Event Low Moderate High Highly probable •  Potato taste (coffee) •  Price volatility (export crops) •  Pests and diseases (all crops) [1 year in 3] •  Landslide (all crops) •  Livestock disease outbreaks •  Drought and erratic rains •  Floods—local and large scale (all crops and livestock) (all crops) •  Milk contamination (dairy) •  Milk collection center power cuts (dairy) •  Counterparty risk (coffee) •  Price fluctuations (food crops and milk) •  Exchange rate fluctuations (export crops) Probable [1 year in 5] •  Hail (all crops) Occasional [1 year in 10] •  Glut (dairy) •  Frost (tea) •  Losses in transit (tea) •  Aflatoxins in feed (livestock) •  Maize shortage (dairy) daily milk consumption is fairly constant (that is, demand are likely to emerge with more important impacts on the is relatively stable), because even small shifts away from sector. Importantly, land consolidation and monocrop- equilibrium supply levels will lead to high price volatil- ping facilitate the spread of pests and diseases. Similarly, ity. Similarly, high demand price elasticity for milk may Rwandan farmers’ current practice of mixing local vari- magnify volatility at smaller changes in supply, because eties for crops, which mitigates certain risks, is likely to consumers quickly respond to price changes. Prices fluctu- be replaced with single-variety cultivation as output mar- ations are less frequent for meat and occur rather between kets become more sophisticated. There are also signs of years. Finally, especially for milk but also for fodder, the storage-related risks that are currently limited because of supply chains are susceptible to contamination. The milk the limited storage in Rwanda, but these are expected to cold chain already has problems with electricity cuts that expand in scope as storage of commodities increases. put food safety standards at risk, and the fodder chain has sporadic problems with aflatoxin contamination. The livestock sector is predicted to grow along with con- sumption, which will elevate the significance of sanitary Table ES.2 provides an overview of current risks in and food safety risks. An increased number of animals will Rwanda according to the impact and probability of lead to greater impacts associated with disease outbreaks, unforeseen events. When prioritizing investments in risk especially because livestock owners hold more cattle or are management, opting for mechanisms that address risks located in closer proximity to one another. With limited with high impact and high probability would be the first land in Rwanda, more animals are also likely to increase choice. The blue shadings in the table indicate the level of demand for fodder, which would imply greater impact priority among the risks in Rwanda. from aflatoxins in fodder. Further, greater demand for livestock products as a result of income increases makes As Rwanda’s agriculture sector transforms, the risk land- potential impacts from food safety risks greater as supply scape will alter and, unless managed, some of these risks chains grow and products reach more consumers. Agricultural Sector Risk Assessment xv Agricultural Risk failures should be taken into account. Although risks may emerge as the sector develops and markets grow, productiv- Management ity increases are likely to give farmers better financial access It is important to remember that not all investments in risk to inputs and better knowledge about how to mitigate risk. management should be borne by the government, and that However, it is important that appropriate institutions and the private sector has an important role in managing risks. actors are in place to facilitate this transition in the sector. Many risks are already managed to a certain extent by pub- Given the prioritized risks, feedback from stakeholders, and lic and private stakeholders in the sector. Aspects such as ongoing interventions, a shortlist of possible solutions areas private versus public goods, investment gaps, and market is proposed for further assessment (table ES.3). Table ES.3. Proposed Solutions Areas for Agricultural Risk Management in Rwanda 1. Improve water management for crop production Water management in the crop sector, in particular to improve practices in preparation for dry periods and scattered rainfall, but also to better manage rainfall in the valleys to minimize flooding. Solutions areas may include: •  Expansion of on-farm water-harvesting systems •  Viable mechanisms for financing small-scale irrigation •  Expansion and rehabilitation of drainage infrastructures in valleys •  Agricultural practices to improve soil moisture and reduce flooding, including minimum tillage agriculture 2. Improve water and feed access in the livestock sector Weather-risk management in the livestock sector, particularly as it relates to water and feed access. Solutions may include: •  Improving rural water infrastructure •  Developing existing feed supply chains to temporarily substitute for the lack of pastures in provinces where grazing is allowed •  Training of farmers in livestock management in water-scarce situations, and in good hygiene practices with special focus on practices in dry periods 3. Strengthen pest and disease management in crop production Pest and disease management for crops, in particular as it relates to potential future risks caused by land consolidation and increased monocropping. Similarly, potential changes in pest and disease risks caused by climate change integrated in such assessment. Solutions may include: •  Improving agricultural practices and pest management, including further developing integrated pest management •  Strengthening the crop research system on pest and disease management and resilient crops •  Strengthening access to inputs, including developing a network of input dealers •  Developing information system on pests and diseases 4. Develop livestock disease management infrastructure Developing livestock disease management infrastructure to mitigate and manage disease outbreaks to decrease the economic impact on the sector. Solutions may include: •  Developing livestock information systems, including animal registers and disease warning systems •  Developing veterinary services and vaccination programs •  Strengthening animal reference laboratory capacity •  Strengthening regional cooperation in livestock disease management 5. Strengthen sanitary institutions and practices throughout the livestock supply chain Sanitary institutions and practices in the livestock sector, throughout the supply chain and involving both public and private actors. As incomes increase, this sector is likely to grow, so the necessary institutional infrastructure must be in place to mitigate risks and minimize losses. Solutions may include: •  Strengthening animal disease management in relevant institutions •  Introducing farm-level livestock management •  Increasing capacity of food safety institutions •  Improving hygiene practices throughout the supply chain •  Mitigating aflatoxin contamination in the feed supply chain xvi Rwanda Table ES.3. Proposed Solutions Areas for Agricultural Risk Management in Rwanda (Continued ) 6. Support improved price risk management in the export crop sector Assessing possible price management mechanisms for actors in the export crop supply chain. Given the exposure to international prices for actors in the coffee and tea supply chains, scope exists to strengthen price management mechanisms in the sector. By analyzing the physical and financial flows on current transaction arrangements for exports, a set of options on how to reduce exposure to risk can be identified. Potential solutions areas may include: •  Strengthening existing price information systems that allow for transparent price setting throughout the supply chain, and training actors throughout the chain to optimize given available information •  Providing price risk management training to actors in the supply chain, for example in forwarding Price To Be Fixed (PTBF) contracting •  Assessing available policy mechanisms for supporting actors in the sector against price risks •  Assessing possible production and marketing investments for producers and processors that can lessen relevant actors’ exposure to risk 7. Address milk price volatility Analyze milk price volatility to better understand the reasons behind the fluctuations in milk prices. This would include proposing appropriate price risk management mechanisms depending on the identified causes behind existing price volatilities. The GoR is already doing a lot in all of these areas. How- This activity was requested by the Group of Eight (G-8) ever, given the risks identified in this analysis and espe- and principally financed by the U.S. Agency for Interna- cially given the strategic path Rwanda has outlined for tional Development (USAID) and Feed the Future pro- the sector, there is room for strengthening these risk man- grams. Contributions were also received by the Multi agement areas. The proposed solutions assessment could Donor Trust Fund (MDTF) on risk management, financed support Rwanda in preparing the sector for effective risk by the Dutch Ministry of Foreign Affairs and the Swiss management in the coming decades. Secretariat of Economic Affairs (SECO). Agricultural Sector Risk Assessment xvii CHAPTER ONE INTRODUCTION Rwanda is a small, landlocked, agriculture-based country of 26,338 km2. With 12 million inhabitants, Rwanda is one of the 10 most densely populated countries in the world (MINAGRI 2010). Although Rwanda has made remarkable progress over the past two decades and is well under way to achieve its objective of becoming a middle-income country by 2020, 45 percent of the population still lives in poverty, mainly in the rural areas.2 Agriculture is the dominant sector of the economy, contributing a third of the country’s gross domestic product (GDP) and about half of Rwanda’s export earnings. Because about 80 percent of the population lives in rural areas and is engaged in agricul- ture to some extent, increasing agricultural productivity is key to improving incomes and decreasing poverty. The government of Rwanda has therefore made agricultural development a priority and allocated significant resources to improving productivity, expanding the livestock sector, promoting sustainable land management, and develop- ing supply chains and value-added activities. As a result, the sector grew an average 5 percent per year over 2002–12, which is rather high although it fell short of both the government’s own objective of 8–9 percent annual growth for the period (revised to 8.5 percent for the next years in the new, PSTA III) and of the CAADP commitment of 6 percent growth in the agriculture sector. At the same time, Rwanda’s agriculture sector faces a series of challenges. Agricultural land plots are very small (80 percent of land holdings are less than 1 ha, often divided into three to four plots) and over 70 percent of agricultural land is either on hills or on the side of hills. Agriculture is dominated by small-scale, subsistence farming under traditional agricultural practices and rain-fed agriculture. As a result, average crop yields are low compared with potential yields, and exposed to risks such as weather- related shocks and pest and disease outbreaks. 2 See http://www.minagri.gov.rw/fileadmin/user_upload/documents/Publications/Agriculture%20Gender%20Strategy %20Final.pdf. Agricultural Sector Risk Assessment 1 Poor groups are especially vulnerable to the impacts of largest commodities that jointly account for 80 percent risks, because risks tend to reinforce poverty traps through Rwanda’s agricultural production value. The selected of ­ cycles of loss-recuperation-loss that prevent these groups commodities are: from investing in productivity-enhancing measures. In Rwanda, the most vulnerable groups are heavily engaged Food crops: cassava, maize, Irish potatoes, sweet in agriculture and the farming community’s ability to ­ potatoes, plantain, beans, and rice bear risk is low. Farmers tend to diversify to manage risk Export/cash crops: tea and coffee in agriculture, which has also proven advantageous for Livestock: cattle—meat and dairy household food security. However, whereas diversification can reduce agricultural risk for individual households, it It can be noted that neither tea nor coffee technically also tends to prevent agricultural productivity increases falls into this category at production level. Nevertheless, through more efficient use of inputs and technology. they were included because of their contribution to gross national export earnings. Current agricultural policies are geared to increasing productivity in the sector by achieving scale in agricul- The report takes a quantitative and qualitative approach tural production. Consolidation in the agriculture sector to risks. Productions risks are quantified in terms of losses will improve use of inputs and the possibility to mecha- and then mapped by different perils. Marketing and nize part of the sector, but it will also facilitate the spread enabling environment risks are analyzed qualitatively. For of pests and diseases. Because risks often have different the purpose of this assessment, risk is defined as the pos- impacts on different crops and livestock, less diversified sibility that an event will occur and will potentially have a production also makes the actors more vulnerable to risks. negative impact on the achievement of a farm’s or firm’s performance objectives and/or on the successful func- Improved agricultural risk management is one of the tioning of the overall supply chain. In the work previously core enabling actions of the G-8’s New Alliance for Food conducted by ARMT in other countries, time periods of Security and Nutrition. To better understand dynamics at least 30 years were assessed to secure a proper under- of agricultural risks and identify appropriate responses, standing of the risks to the sector. However, because of incorporate an agricultural risk perspective into deci- the very different systems in Rwanda before and after the sion making, and build capacity of local stakeholders in mid-1990s, as well as the interruptive civil war in 1994, risk assessment and management, the Agricultural Risk it is difficult to identify distinct trends over longer time Management Team of the Agriculture and Environ- series. This assessment therefore focuses on risk to the ment Services Department of the World Bank conducted agriculture sector over the past 20 years. an agriculture sector risk assessment. This activity was requested by the G-8 and principally financed by USAID A broad spectrum of stakeholders was consulted through- and Feed the Future programs. Contributions were also out this work, including the Rwandan government, received by the Multi Donor Trust Fund on risk manage- farmers, traders, processors, cooperatives, agricultural ment, financed by the Dutch Ministry of Foreign Affairs institutions, and academia. A consultative stakeholder and the Swiss Secretariat of Economic Affairs (SECO). meeting organized by the Ministry of Agriculture and Animal Resources was also held in Kigali to obtain feed- The purpose of this report is to assess existing risks to back on findings and to discuss areas for risk solution the agriculture sector, prioritize them according to their interventions for deeper analysis. frequency and impacts on the sector, and identify areas of risk management solutions that need deeper special- Figure 1.1 provides an overview of the full process applied ized attention. Three levels of risks are assessed: produc- by the ARMT in the past. The Agricultural Sector Risk tion risks, market risks, and enabling environment risks Assessment in this report constitutes the first phase. Based to selected supply chains. To give a sectorwide over- on the results of this assessment, a solutions assessment view of the impacts of risks, the assessment looks at the will be conducted, under which a few potential risk 2 Rwanda Figure 1.1.  Agriculture Sector Risk Management Process Flow Phase 1 Phase 2 Phase 3 Phase 4 Client demand Risk Solution Development of risk Implementation and assessment assessment management plan risk monitoring RM plan development Desk review Desk review Implementation by stakeholders Stakeholder In-country Incorporation into Monitoring risks consultations assessment mission existing govt. programs and Stakeholder development plans Finalize analysis Refining RM strategy workshop Source: World Bank ARMT. management instruments are further assessed. Under this the selected food crops, export crops, and livestock are second phase, ongoing activities in the selected areas are analyzed in chapter 3. Analysis of the adverse impacts assessed and gaps mapped to determine activities needed of agricultural risks at aggregate and provincial levels, to minimize the impacts of risks on the sector. along with a stakeholder risk assessment and a discussion of particularly vulnerable groups, is found in chapter 4. The report is structured as follows: chapter 2 provides an Chapter 5 prioritizes identified risks, discusses potential overview of Rwanda’s economy and the role and struc- solutions areas, summarizes feedback from consulted ture of the agriculture sector. Agriculture sector risks stakeholders, and recommends solutions areas for further (production, market, and enabling environment risks) for assessment. Agricultural Sector Risk Assessment 3 CHAPTER TWO AGRICULTURE IN THE ECONOMY Rwanda in the 21st Century Emerging from the 1994 civil war, Rwanda’s economy has seen a rapid expansion over the past two decades. Overall growth has averaged 8 percent, exceeding average SSA growth rates, and Rwanda’s total GDP now amounts to US$7.1 billion (2012 figure). Inflation is relatively low and the government has maintained general macroeconomic stability. In the World Bank/IFC Doing Business 2010 report, Rwanda was the world’s top reformer and now ranks 32 out of 189 countries worldwide in the “ease of doing business” (World Bank 2009). The government’s vision is for Rwanda to become a middle-income country with an annual per capita GDP of US$900 by 2020. This will require annual growth of at least 7 percent. Nevertheless, Rwanda is still relatively poor, ranking 36 out of 48 SSA countries in 2012 in terms of per capita GDP.3 Real per capita GDP was US$390 in 2012, com- pared with the SSA average per capita GDP of US$1,522.4 As shown in figure 2.1, large annual fluctuations in GDP growth rates have occurred over time. With low per capita GDP, poverty persists in Rwanda. About 63 percent of the popu- lation lives on less than US$1.25 per day and 82 percent on less than US$2 per day. Inequality is high: the Gini coefficient is 50.8 percent and 43 percent of the income share is held by 10 percent of the population.5,6 The composition of GDP is 2013 was 33 percent agriculture, 14 percent industry, and 53 percent services (figure 2.2). Over time, the share of agriculture has declined and the share of services has increased. Rwanda is a relatively open economy and trade constitutes almost half of Rwanda’s GDP. The value of imports is close to twice the size of exports and account for 33 percent of GDP, compared with 13  percent for exports. 3 Based on 2005 U.S. dollars, WDI (accessed November 6, 2013). 4 Real GDP, 2005 prices, not including South Africa and the Seychelles. 5 2011 figure; http://data.worldbank.org/indicator/SI.POV.GINI (accessed December 13, 2013). 6 WDI (accessed November 6, 2013). Agricultural Sector Risk Assessment 5 Figure 2.1. R  wanda’s GDP Growth Compared with SSA and East Africa GDP growth (annual %) average of SSA 40 GDP growth (annual %) rwanda 30 GDP growth (annual %) average of East Africa 20 10 0 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 –10 –20 –30 –40 –50 –60 Source: WDI 2013. Note: East Africa follows the definition set out by Food and Agriculture Organization (FAO) and includes the following countries: Burundi, Djibouti, Ethiopia, Kenya, Rwanda, Somalia, Sudan, and Uganda. See http://www.fao.org/africa/sfe/en/. Figure 2.2. GDP Composition, household level. Agriculture is also a major source of 2013 export e­arnings. Despite this, poverty is high in rural areas, where 49 ­percent live below the poverty line com- pared with 22 percent in urban areas. Industry 14% Agriculture Sector Composition and 33% Agricultural Value Added Rwanda’s agriculture sector has experienced two growth trends over the past 30 years, with high volatility pre-1994 and almost uninterrupted growth from 1995 onward. Services Agricultural value added per worker has increased since 53% 1999 and, consequently, agricultural employment and the rural population have declined. In 2007, the pro- ductivity of an agricultural worker was about US$263; by 2012, it had reached US$294. However, the overall Source: WDI 2013. growth rate of agricultural value added has not been lin- ear (figure 2.3). For instance, it went from 2.6 percent in 2007 to 7.7 percent in 2009, followed by another decline Role of Agriculture in to 4.68 percent in 2011.7 The last time that agricultural the Economy value added growth rates were negative was in 2003, when Rwanda experienced a drought that affected more Agriculture plays an important role in the overall econ- than 1 million people.8 omy. As discussed above, agriculture contributes 33 per- cent of GDP and 80 percent of population is engaged in the sector. At the aggregate level, domestic food pro- duction almost equals domestic demand and farmers’ 7 Ibid. own production is an important source of food at the 8 Ibid. 6 Rwanda Figure 2.3. Agricultural Value Figure 2.5. S  ectoral Composition Added (annual percent of Rwanda’s growth), 1980–2012 Agriculture Sector GDP 40 Forestry Fisheries 30 Livestock 7% 1% Drought 20 Civil war 4% Export crops 10 2% 0 19 0 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10 12 8 –10 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 –20 –30 Food crops –40 86% Source: WDI November 2013. Figure 2.4. Gross Production Index Number, Crops and Source: MINECOFIN 2013a; Authors’ calculations. Livestock (2004–06 = 100), 1995–2012 200 Crops Livestock impacts of risks at provincial and commodity levels. This 150 disaggregated analysis might reveal that losses are signif- 100 icant for one or two provinces and/or for certain com- modities, or that agricultural risks lead to significant losses 50 for certain groups engaged in the sector. The agriculture 0 sector in Rwanda consists of four subsectors: crops, live- 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 stock, fisheries, and forestry. The crops subsector is fur- Source: FAOSTAT 2014. ther divided into two groups: food crops and export crops. Figure 2.5 shows the composition of agricultural GDP. Looking at production indexes for crops and livestock over Within agriculture, food crops account for 86 percent of the past 20 years confirms this smooth, positive growth agricultural GDP. The role of other subsectors is relatively trend in agriculture. Overall, there has been little volatil- small and the relative shares of export crops and livestock ity in the sector, which is good from a macroeconomic have been declining.9 perspective and indicates limited systemic risks to the sector. This sets Rwanda apart from many other coun- Plantain, cassava, Irish potatoes, sweet potatoes, and maize tries in SSA, where growth trends are often highly vola- dominate agricultural production (figure 2.6). In particu- tile. The few slight drops in Rwanda’s crop production lar, cassava, potato, and maize production has increased were recorded in 2001, 2004, and 2007 compared with rapidly since the turn of the century. Rice and beans are the respective previous years; livestock production drops potentially important crops. Production increase is largely occurred in 2000 and in 2011–12 compared with 2010 a result of targeted agricultural policies and expansion (figure 2.4). All of these years had either deficit or exces- of production areas. The share of export crops is much sive rainfalls, but with the exception of 2004 (when legu- smaller. However, their contribution to overall exports is minous production showed a marked decline), it is not important and agricultural exports are a major source of clear that these actually affected aggregate production. foreign exchange earnings in Rwanda. However, aggregated data mask volatility; hence, to better 9 The relative share of fisheries has remained constant, whereas the relative understand the real losses it is necessary to analyze the share of forestry has been increasing. Agricultural Sector Risk Assessment 7 Figure 2.6. C  rops’ Shares of Gross Figure 2.8. N  ational Herd Profile Agricultural Production of Livestock (TLUs,* ’000 Value (current, average of animals), 2005–11 2009–11) 2,500 Pigs Sheep Goats Cattle 2,000 Other, 26.7% Cassava, 23.01% 1,500 1,000 Maize, 5.02% Tea, 0.1% Coffee, 0.9% Potato,13.27% 500 Sweet potato, 4.47% Rice, paddy,1.77% 0 Beans, dry, 5.34% Plantain, 19.41% 05 06 07 08 09 10 11 20 20 20 20 20 20 20 Source: FAOSTAT 2013. Source: Rwanda Agriculture Board (RAB).1 Note: Cattle, sheep, and goats are in TLUs. No estimates for pigs were avail- able for 2011 so 2010 figures were used. *  TLU = Tropical Livestock Units. 1 TLU = 1.0 Camels; 0.7 Cattle; Figure 2.7. Yield Gaps Compared with 0.1 Sheep/Goats. See http://www.fao.org/ag/againfo/programmes/en/ lead/toolbox/Mixed1/TLU.htm SSA, 1980–89 and 2000–09 1  Data obtained in 2014 from RAB files during field mission. 200% 1980–89 2000–09 180% The livestock sector is growing, although less than the over- 160% all agriculture sector (figure 2.8). According to the National 140% Institute of Statistics (NISR), half of all households own a 120% goat, cow, and/or chicken. This is partly attributable to gov- 100% ernment initiatives such as the “One Cow per Poor Family” 80% (Girinka) program (which has distributed 134,548 cows and 60% 40% another 40,352 heifers passed by beneficiary households 20% onto other poor households [MINAGRI 2013a]) and UBU- 0% DEHE (a rural development program that has distributed Plantain Cassava Irish potato Sweet potato Maize Beans Rice (paddy) different types of livestock). As a result, milk production has increased nine times since 1999, and meat production Sources: FAOSTAT 2013; Authors’ calculations. three times in the same period. From being a major milk importer, Rwanda is now more or less self-­ sufficient in milk production. Livestock production accounts for just over 9 Longer-term time series of crop yields show a more varied percent (2010–12 average) of total production value, and trend.10 Yield decreased significantly in the late 1980s and the demand for livestock products is likely to increase fur- early 1990s because of instability in the country. Although ther as incomes rise in Rwanda. Milk and meat consump- yields for most crops have now returned to prewar levels, tion increased from 20 l/person/year and 5.7 kg/person/ the yield growth varies between crops. Several crops have year in 2006 to 50 l/person/year and 6.8 kg/person/year yields below the SSA average and the gap between actual in 2012, respectively (FAO Corporate Statistical Database yield and production potential is large (figure 2.7; postwar [FAOSTAT] 2014; 2004–2006 prices). crop yield trends are given in figures 3.4–3.8). Food and agricultural export make up almost half of 10 Average yield of crops from 2009 through 2011 was calculated to make the Rwanda’s export. The top five agricultural export com- comparison (FAOSTAT 2013). modities in terms of value contribute 92 percent of 8 Rwanda Table 2.1. Value of the Top Five Agricultural Export Commodities, 2008–10 (average) Share of Share of Agricultural Total Value Agricultural Share of Exports (%) (USD ’000s) Production (%)* GDP (%)** Tea 45.56 63,605 3.48 1.78 Coffee 35.47 49,514 2.71 1.38 Beer of barley 4.72 6,585 0.36 0.18 Beverages, 3.72 5,192 0.28 0.15 nonalcoholic Cattle 2.53 3,531 0.19 0.10 Source: FAOSTAT 2014. * Constant 2004–06 prices; ** constant 2005 prices. agricultural exports (table 2.1). Remaining exports are Table 2.2.  Weather and Crop Seasons mainly ores and metals. Food and agriculture constitute in Rwanda a much smaller share of imports (18.5 percent), whereas Crop Seasons Period manufactured goods, food, and fuel constitute the bulk of imports (WDI 2013). Season A September–January Season B February–June Season C July–September Climate and Rainy Seasons Agro-Ecology Season A Mid-September–December Season B Mid-February–Mid-May Situated on mountainous terrain in the East African Rift Dry Seasons Valley, Rwanda has a tropical temperate climate (Rwanda December–February Environment Management Authority [REMA] 2009, 97). June–August Temperatures in the country vary with altitude, but aver- Source: MINAGRI 2011a. age annual temperatures range between 16°C and 20°C (REMA 2009, 97). Rainfall in the country is shaped by the effect of the Inter Tropical Convergence Zone (ITCZ), prolonged droughts, whereas the northern and western where the weather systems of the Northern and Southern regions get more rainfall (REMA 2009, 97). Over the Hemispheres meet. The progression of the ITCZ results past 20 years, both floods and rainfall deficits or droughts in two types of seasons: dry and rainy. have been fairly frequent but are often incurred locally or regionally. Therefore, as seen earlier, droughts and floods Four seasons divide a calendar year in Rwanda (table more often than not do not have significant impacts on 2.2): two rainy seasons from September to November and the agriculture sector as a whole. Similarly, hailstorms from March to May, and two dry seasons from December are relatively common in Rwanda, but their impacts are to February and June to August. The crop calendar has highly localized. essentially three seasons, though the most important are September to January (Season A) and February to June Certain changes in both rainfall and temperature patterns (Season B). are already apparent, indicating that Rwanda is affected by global climate change. Temperature and rainfall data over Because of the topography of the country and the exis- the past 30 years show that the rainy season is becoming tence of large bodies of water, the eastern and south- shorter with higher intensity, leading to both more droughts eastern parts of the country experience more frequent and floods simultaneously (REMA 2009, 97). However, Agricultural Sector Risk Assessment 9 different parts of the country are affected differently. The Table 2.3. A  gricultural Area and Northern and Western Provinces are seeing heavier rains Area per Household (ha), and floods whereas the Eastern Province is seeing more by Province, 2010 rainfall deficits (REMA 2009, 98). Area, Ag. Average Area per Despite these observed patterns, climate models are in Province Holdings (ha) Household (ha) disagreement over rainfall changes projected in Rwanda Northern 211,576 0.65 over the next 30 years. For East Africa as a whole, high Southern 237,047 0.71 rainfall extremes (events typically occurring once in every Eastern 439,204 1.10 10 years) are expected to increase in frequency (van de Western 269,964 0.62 Steeg et al. 2009, 27). City of Kigali 32,959 0.65 Source: NISR 2010. Climate models are in more agreement regarding tem- perature increases. Three different climate change Table 2.4. Agricultural Holdings models11 forecast a 1°C to 2.5°C increase in maximum ­ temperatures. It is projected that higher and more vari- by size (%) able temperature will lead to more frequent and severe Area held Share Cumulative droughts and floods in Africa. Because most of Rwan- (ha) (%) (%) da’s agriculture is rain fed and thus exposed to weather Less than 0.20 26.3 26.3 events, it is vulnerable to the climate changes projected. 0.20–0.49 30.5 56.8 For crops that also require cooler temperatures to grow, 0.50–0.99 23.2 80.0 such as beans and potatoes, temperature increases pose a 1.00–1.99 14.0 94.0 particular threat.12 In addition, higher temperatures are 2.00–2.99 3.6 97.6 expected to increase the prevalence of pests and diseases. 3.00–3.99 1.2 98.8 However, no studies have been conducted on how global 4.00–4.99 0.6 99.4 climate change will affect key crops in Rwanda. Appen- Greater than 4.99 0.6 100.0 dix B gives a more detailed overview of projected global Source: NISR 2010. climate change in Rwanda. ­ Agricultural Land and Farm size in Rwanda is extremely small and farms are Input Use fragmented. The average area per agricultural household About 50.6 percent of Rwanda’s land area is agricultural, is 0.76 ha (NISR 2010). As shown in table 2.4, 80 percent of which about 73 percent is actually used to grow crops of agricultural land holdings are less than 1 ha and the (food crops, cash crops, and forages); the remaining 27 land is highly fragmented; on average, each household percent is either kept fallow or used for pastures and affor- has four land plots. Under these circumstances it is very estation. The Eastern Province has the most agricultural difficult for farmers to take advantage of economies of land (439,000 ha) and the Northern Province has the least scale by adopting modern agricultural equipment. The (212,000 ha) (not including the City of Kigali) (table 2.3). government’s Land Consolidation Program and Crop The share of land covered by agricultural holdings also Intensification Program (CIP) are designed to address varies by province: 46 percent in the Eastern Province and this problem by organizing farmers into cooperatives 65 ­ percent in the Northern Province. (box 2.1). Because of Rwanda’s hilly topography, 70 percent of the land is either on hillsides or on the top of hills. Only 30 per- 11 These include CNRM-CM3, ECHAM, and MIROC 3.2. See Tenge, Apho- nse, and Thomas 2012, 264. cent of farms in Rwanda are located on flatland or at the 12 Ibid. bottom of hills, which contributes to a series of challenges 10 Rwanda Box 2.1. R  WANDA’S CROP Table 2.5. Topographic Position of INTENSIFICATION PROGRAM Farms in Rwanda, 2010 The Crop Intensification Program is a flagship project Topography Share (%) for the Ministry of Agriculture and Animal Resources. Top of the hill 24.3 Launched in 2007 and with a current budget of RF 9,092 Side of the hill 45.8 million, it focuses on maize, wheat, rice, Irish potatoes, Bottom of the hill 12.0 beans, and cassava. Its overall objective is to increase agricultural productivity in high-potential food crops and Plain 15.8 ensure food security and self-sufficiency. Its main activities Marsh 2.2 include land consolidation, proximity extension services, Total 100.0 service providers for extension, demonstration plots, seed Source: NISR 2010. distribution, and improved seeds and fertilizer use. Distribution of improved inputs—in a move to raise productivity levels through improved inputs, the govern- Table 2.6. Agricultural Households’ ment decided to initially supply inputs and encourage Use of Agricultural Inputs farmers to use them. Inputs Season A (%) Season B (%) Improved seeds—improved seed from neighbor- Improved seeds 13.3 7.1 ing countries such as Kenya and Tanzania, along with Pesticides/ 15.7 14.2 improved planting materials (cuttings) of cassava and pota- fungicides toes, are distributed. Manure 39.5 29.8 Distribution of fertilizers—vouchers are distributed to Compost 38.3 23.4 farmers through service providers for subsidized fertilizer. Fertilizer 17.7 13.7 Consolidation of land use—because of high demo- Source: NISR 2010. graphic pressure in Rwanda that has led to highly frag- mented agricultural landholdings, this involves successfully also adds to agricultural risks in the event of unfavorable rearranging land parcels to consolidate the use of farm weather conditions. holdings. Under the policy, farmers are required to grow a specific food crop together with the aim of improving pro- ductivity and environmental sustainability. Consolidated Traditional agricultural practices persist around the land area was measured at 503,000 ha in 2011. country. According to a NISR survey, 98 percent of agri- cultural land is rain fed and only 0.6 percent is under Source: MINAGRI 2011a. irrigation. Irrigation is one solution to address drought. However, in Rwanda it will be very difficult and expen- sive to bring more area under irrigation. Furthermore, and risks (table 2.5). Hilly land is subject to drought, soil only about 0.2 percent of the land uses animal traction erosion, and landslides, whereas marshland is subject to or mechanical equipment, whereas 99.8 percent uses floods during heavy rains. The hilly topography makes traditional hand hoe manual cultivation. Similarly, the it very difficult to use modern farm equipment or irriga- number of farmers who use modern agricultural inputs tion. This not only reduces agricultural productivity but is small (table 2.6). Agricultural Sector Risk Assessment 11 CHAPTER THREE AGRICULTURE SECTOR RISKS The identification, analysis, and prioritization of agricultural risks cover three separate categories of agricultural commodities grown in Rwanda: (1) food crops, (2) export crops, and (3) livestock. As discussed in chapter 1, the assessment looks at the largest commodities that jointly account for 80 percent of Rwanda’s agricultural production value as well as the two largest agricultural export commodities. The selected com- modities are: Food crops: cassava, maize, Irish potatoes, sweet potatoes, plantains, beans, and rice Export/cash crops: tea and coffee Livestock: cattle—meat and dairy This chapter looks at production risks, market risks, and enabling environment risks for the commodities in these three categories. Food Crops Food crops for domestic consumption dominate primary agriculture and are therefore the main focus of Rwanda’s Crop Intensification Program. Cassava, Irish potatoes, sweet potatoes, maize, and bananas are the largest commodities in terms of produc- tion value and account for 56 percent of total production value (2009–11 average). Bean production accounts for just over 5 percent of total production but is important because of its wide application across Rwanda, as 92 percent of rural households cultivate various varieties of beans (NISR 2012a). Beans are also an important source of protein. Largely a result of targeted agricultural policies and expansion of production areas, maize, cassava, potato, and plantain production increased rapidly in the early to mid- 2000s. As explained earlier (box 2.1), the CIP supports maize, cassava, rice, Irish potato, bean, and wheat producers with improved inputs; finances research for improved seeds for these crops; and encourages production on consolidated land through cooperative Agricultural Sector Risk Assessment 13 structures. The support for rice, maize, and wheat pro- In Rwanda, maize and rice are arguably the least mois- duction is largely in anticipation of expected dietary shifts ture stress-tolerant crops. Maize requires constant mois- toward more rice and wheat intake, following expected ture for optimal growth and yield is reduced if the maize rises in incomes over the next decade. crop is allowed to wilt consistently for more than 48 hours. Growth is particularly sensitive: (1) when the crop Food crops are mainly subject to production-related risks, is 50 cm high and dry conditions can restrict the develop- but the risks prevalent in the sector are likely to alter as ment of the reproductive organs (15 percent); (2) during the sector transforms. Pests and diseases pose a risk to tasseling, silking, and the completion of pollen germi- Rwanda’s food crop producers, particularly because food nation, when dry conditions can reduce the number of crop producers have less organized supply chains and less grains that will develop in each cob (50 percent); and access to preventive inputs than do export crop produc- (3) during early grain development, when dry conditions ers. Weather-related risks are less of a concern for food can result in shriveled or aborted grains (30 percent). crop producers; the main impacts on production are During the latter two growth stages, the maize plant is from moisture stress caused by erratic rainfall. Market- more developed with a greater leaf area, transpiration ing risks are limited, with prices responding predictably from which may require as much as one liter of water to seasonal supply and demand, and domestic prices per day. If soils are deep and well structured, crops at are not affected by global prices. However, as Rwanda’s these growth stages may be able to extract more water agriculture sector transforms, the risk landscape is likely from greater soil volumes by virtue of their greater depth to alter. Importantly, land consolidation and monocrop- of rooting, but if soils are shallow or of low water-hold- ping facilitate the spread of pests and diseases. Similarly, ing capacity, then the demands of evapotranspiration Rwandan farmers’ current practice of mixing local vari- will exceed the supply capacity of the soil and wilting eties for crops, which mitigates certain risks, is likely to will occur. be replaced with single-variety cultivation as output mar- kets become more sophisticated. There are also signs of Exactly how much is lost on an annual basis through- storage-related risks that are currently limited because out the country is not clear, but a variation of 20 per- of the limited storing in Rwanda, but these are expected cent in seasonal rainfall could reduce yields by as much to expand in scope as storage of commodities increases. as 50 ­percent if the dry spell occurred during the criti- Conversely, as productivity increases, farmers are likely cal tasseling and silking stage of growth. Anecdotal to have better access to inputs and the knowledge to miti- evidence from Rwanda supports this: in 2008, erratic gate these risks. Marketing and enabling environment rainfall caused yield losses for 37 percent and 26 percent risks may emerge as the sector transforms, but this will of smallholders in the Eastern and Southern Provinces, largely depend on the role the government plays in the respectively, compared with 19 percent and 14 percent future. in the Northern and Western Provinces, respectively.13 However, at a national level, systemic losses are not vis- Production Risks ible. Maize yields shifted slightly downward in the late Unpredictable Weather Patterns 1990s, and a decade later increased significantly from Unpredictable weather patterns can pose a risk to 2007 to 2010, arguably in response to the CIP, introduced producers but there are no clear patterns of systemic in 2007. Because of the large shift in the long-term trend, weather risks on food crop production in Rwanda. The it makes more sense to divide the trend into two periods: frequency of substantial rainfall deficit in a given sea- 1995–2006 and 2007–12. ­ Figure 3.1 depicts these two son is low (less than 10 percent), but the probability of trends and shows that in fact, systemic losses to maize at erratic rainfall and short-term moisture stress is high. A the national level are limited. certain degree of yield loss from moisture stress is almost inevitable, contributing to the risks faced by individual farmers. 13 Comprehensive Food Security and Vulnerability data. 14 Rwanda Figure 3.1. Maize Yields (MT/ha), Figure 3.2. Paddy Rice Yields (MT/ha), 1995–2012 1995–2012 3.000 2007–2012 trend Yield (tons/ha) 1995–2006 trend 7 Trend 2.500 Drought Yield (tons/ha) 6 2.000 Rainfall 5 deficit/drought MT/ha 1.500 4 1.000 3 2 0.500 1 0.000 0 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Source: FAOSTAT 2013. Source: FAOSTAT 2013. Note: MT/ha = metric tons per hectare During the preparation of this report, rice farmers For other food crops, moisture stress is not a major a expressed concerns about water availability; however, yield concern in terms of systemic risks. In general, weather- data do not indicate nationwide systemic rice losses from related shocks on production result in short-lasting dips droughts in Rwanda (figure 3.2). Research (Akram, Sat- below the general yield trend, and although there are tar, Rehman, and Bibi 2013) demonstrates that withhold- declines in production for several crops that correlate ing irrigation water from a rice crop for a 14-day period with drought or flood years, these events have either reduced paddy yield by 10–40 percent, depending upon been regional or did not result in multi-year declines in the time at which moisture stress was imposed. Drought production. For example, banana yields saw a decline in stress at panicle initiation had the greatest impact on yield, 2000–01, which correlates with a severe drought in the whereas stress at anthesis and grain filling led to reduced Eastern Province in 2000 related to La Niña (figure 3.3). impacts. In Rwanda, rice is produced under marshland This would indeed affect production because bananas are conditions, which is not the same as irrigated conditions sensitive to drought. However, looking at the subnational (although some irrigation systems do exist) but depends level, the decline in yield clearly occurred throughout the more upon controlled drainage to ensure adequate lev- country; therefore, the drought in the Eastern Province els of moisture are available at key growing periods. Such does not fully explain the decline in banana yields in systems are vulnerable to water shortage, especially at the 2000–01. beginning of the season if delayed rains have precluded the accumulation of adequate moisture for initial germi- Although beans are sensitive to temperature, weather does nation and growth. not have any significant impact on Rwanda’s production of bush and climbing beans (figure 3.4). This is largely because Despite this, no exact figures exist for annual rice losses Rwandan bean producers minimize weather-related risks by caused by moisture stress in Rwanda. At a national level, growing a mix of local and improved varieties and applying most of the drought/rainfall deficits go unnoticed, even different varieties across the country. Applying a mix of vari- for the big drought in 2003. One explanation may be eties, rather than solely improved varieties, results in lower because rice is grown in marshlands across the country yields to a certain extent, particularly because improved and regional droughts are masked by good yields in other varieties can yield twice as much as traditional varieties. parts of the country. Another potential explanation is that However, the reduced risk of adverse effects from unpre- marshlands retain water better than does agricultural land dictable weather events justifies this practice.14 on hillsides, and therefore rice manages better during dry periods. Finally, the timing of the dry periods may affect the impacts on rice production. 14 Rwanda has relatively competitive bean yields internationally. Agricultural Sector Risk Assessment 15 Figure 3.3. Banana Yields (MT/ha), Figure 3.5. C  assava Yields (MT/ha), 1995–2012 1995–2012 Drought years 16 Yield (tons/ha) 10 in the East 14 Trend 9 8 12 7 10 6 8 5 4 6 3 4 Yield (tons/ha) 2 Trend 2 1 0 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 1995 1997 1998 1999 2001 2005 2007 2008 2009 2011 1996 2000 2002 2003 2004 2006 2010 Source: FAOSTAT 2013. Source: FAOSTAT 2013. Figure 3.4. Beans Yields (MT/ha), Figure 3.6. Irish Potato Yields 1995–2012 (MT/ha), 1995–2012 16 Yield (tons/ha) Drought Droughts in the year eastern and 14 Trend Southern provinces 12 Yield (tons/ha) Rwanda 1.2 Trend 10 1 8 0.8 6 0.6 4 0.4 2 0 0.2 1995 1997 1998 1999 2001 2005 2007 2008 2009 2011 1996 2000 2002 2003 2004 2006 2010 2012 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Source: FAOSTAT 2013. Source: FAOSTAT 2013. Figure 3.7.  Sweet Potato Yields Cassava is the most moisture stress-tolerant crop of those (MT/ha), 1995–2012 studied in this assessment, because of its ability to respond 10 Yield (tons/ha) 9 quickly to decreased moisture in the environment (see Trend 8 appendix E). Production is not affected by moisture stress 7 more than 1 in 10 years under Rwanda’s current rainfall 6 5 patterns. At a national level, cassava yield trends do not 4 indicate systemic impacts from weather-related shocks 3 (figure 3.5). 2 1 0 Similarly, Irish and sweet potato yields show few large 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 dips (figures 3.6 and 3.7). The exception is for Irish potato Source: FAOSTAT 2013. yield in 2007, which coincides with a drought in the East- ern Province. Sweet potatoes, conversely, show no par- ticular dips, but rather shifts in yield every half decade Other weather-related risks are less of a concern for that affect the longer-term trend line. Nevertheless, there food crop producers in Rwanda on a systemic basis. are no significant yield drops that can be explained by Floods, winds, and hailstorms are reportedly a prob- shock events. lem and can have devastating impacts for individual 16 Rwanda farmers, wiping out up to 100 percent of the harvest. (or banana xanthomonas wilt [BXW]).18,19 Depending However, these events tend to be highly localized and on the disease and when the plant is infected, an indi- do not affect production at the national or even provin- vidual grower may easily experience 100 percent yield cial level. loss. For cassava, the most damaging pest is the green spider Pests and Diseases mite (Mononychellus tanajoa), which is widespread. In 2007, For most food crops, the bulk of production losses are it was found to infest approximately 40 percent of all cas- seemingly from pests and diseases. Although no data sava plants, causing 45 percent damage on average where exist on aggregate annual crop losses from pest and dis- infestation occurred (Night et al. 2011). The cassava eases, information on the impact from individual pests mosaic virus and cassava brown streak virus (CBSV) can and diseases indicates that losses are significant. How- reduce yields by as much as 95 percent. Currently CMV ever, the existing literature and the reporting of pests is more prevalent; a 2007 assessment found the disease and diseases in Rwanda indicate that pests and diseases at 94 percent of plots visited, with 32 percent of plants are endemic and that outbreaks are relatively localized. infected and the impact on the yield of infected plants Thus it is not possible to see the impact of pests and estimated at 60 percent (Night et al. 2011). However, the diseases as shocks affecting the yield data at the national disease situation in Rwanda has historically been quite or even provincial level. Some diseases, such as cassava fluid, with new virus diseases arising every 10–15 years mosaic virus (CMV), come in cycles and do not affect (FAO 2010), and it is possible that a new form of CBSV all farmers simultaneously but seem to be ever present is spreading rapidly (Bigirimana, Barumbanze, Ndayihan- in Rwanda and likely contribute to existing yield gaps. zamaso, Shirima, and Legg 2011). Both CMV and CBSV Others, such as banana bacterial wilt (BBW) and maize are spread by the white fly, Bemisia tabaci, and by the distri- chlorotic mottle virus (MCMV), were introduced rather bution of infected plant material. recently and are on the rise but have yet to make a sig- nificant mark on national yield levels. Some, such as the Insect pests of rice are limited to the rice fly (Diopsis maize stalk borer, may spread more rapidly and affect thoracica), the larvae of which eat out the center of larger areas as the structure of the sector changes into young tillers, causing blind shoots. Yield losses of 5–20 larger single-crop land areas with more homogenous percent are commonly recorded (Akinsola and Agyen- varieties. Finally, climate change models project a more Sampong 1984), depending on the severity and timing favorable environment within which certain pests and of infestation. The impact of early infestation, if con- diseases will flourish. trolled by insecticides, can be mitigated by compensa- tory growth. Unmanaged, pests and diseases cause high losses for pro- ducers in Rwanda. The main bean pests in Rwanda, the The occurrence of pests and diseases seems to be on the beanfly15 and the bean Bruchid,16 have been estimated to rise and to spread more rapidly now than in the past. reduce bean yields nationally by as much as 25 percent For example, the potential frequency of occurrence of and 30 percent, respectively (Trutmann and Graf 1993; BBW is increasing and the disease is spreading rapidly: Jones 1999).17 Similarly, banana production is highly BBW was first found in Rwanda in 2005; by 2012, it affected by diseases, particularly black sigatoka, banana bunchy top disease (BBTD), and banana bacterial wilt 18 Two main banana pests in Rwanda (nematodes and the banana weevil) limit yield when stands of bananas are not rotated or when cultural practices are inadequate. Nevertheless, under most conditions, the risk to banana production posed by pests is minimal. 15 Bean stem maggot, Ophiomyia spp. 19 A fourth banana plant disease, the Panama disease, exists in Rwanda but its 16 That is, a number of Bruchid species. risk is limited. It only affects modern banana varieties and is of no risk to the 17 180–225 kg/ha; Bruchid species infest bean pods in the field and can then East African Highland clone sets, which are resistant to the disease and consti- become important pests of stored beans. tute the bulk of production in Rwanda. Agricultural Sector Risk Assessment 17 had spread to 23 of the country’s 30 districts. Similarly, Eastern Province, ­ infestation of the striga weed can cause levels of maize pests and diseases are currently low; high levels of maize crop loss, but as the weed’s incidence until 2013, only leaf blight and maize streak virus were is predictable, it is less of a risk and more of a constraint recorded as significant diseases of the growing crop to production. For Irish potato growers, blight (Phytophtera (MINAGRI 2008). However, in June 2013, MCMV was infestans) poses a significant risk but is in part exacerbated identified in the Western and Northern Provinces. This by poor agricultural practices. The cool, wet conditions virus is a component of maize lethal necrosis disease under which most potatoes are grown in Rwanda con- (MLND), a disease complex that has spread rapidly in tribute to the spread of this disease, which can result in Kenya since 2012 and can cause up to 100 percent loss up to 100 percent loss of yield and can render inedible of yield. This disease poses a significant threat to future any tubers that might survive. Even mild infections can maize production. result in significant loss of yield. Considerable emphasis is placed on regular application of fungicides to control Stored crop can also be vulnerable to pests. Unless the disease and in some areas, growers delay planting so addressed, pests will continue to be a problem as grain that the crop matures under drier conditions, although is stored in larger volumes and for longer periods in the this increases the risk of yield loss caused by insufficient future. Bruchid species infest bean pods already in the field moisture. But the disease also flourishes in part because and can then become important pests of stored beans of poor crop hygiene, including: reduced rotation periods causing, losses of up to 30 percent (Jones 1999). The (the period between potato crops in the same soil should pest can also be sustained within stores under poor stor- ideally be at least four years); the ubiquity of volunteer or age conditions. Although a minimal level of infestation backyard potato plants grown by noncommercial grow- is inevitable, good storage practices will constrain such ers that can act as a reservoir for disease; and the use of infestations. This includes making use of resistant variet- infected seed (as a result of the limited supplies of clean ies, anaerobic storage, and fumigants and coating seeds planting material). with edible oil (which will kill Bruchid eggs). Insect damage from common pests of stored maize and rice (weevils such Because rice is grown in large areas across valley bot- as Sitophilus zeamais and Sitophilus oryzae) (Dunkel, Sriharan, toms, the crop is vulnerable to the rapid spread of pests Niziyimana, and ­ Serugendo 1990) is not unusual, but and diseases. Rice blast (Magnaporthe oryzae) and bacte- because grain is stored only for a short period, levels of rial disease complexes (leaf and panicle blight caused loss have generally been low; hence, this is not a significant by Xanthomonas spp. and sheath rot associated with Pseu- risk for growers or millers. However, unless addressed, this domonas infection) are the major diseases causing yield problem may increase in the future as postharvest infra- loss in rice, and can affect all known varieties. Control structure expands. is currently based mainly upon crop and varietal rota- tion, but discussions with specialists reveal that for these To a certain extent, the occurrence and losses from cer- diseases pathogen evolution is so fast that within 3 to 4 tain pests and diseases are predictable and attributable growing seasons most grown varieties become suscep- to suboptimal agro-environmental conditions or agricul- tible to the extent of causing total crop failure. Lower tural practices. For example, the cassava pest is ubiqui- levels of yield loss are more common, but can regularly tous and current control options, including breeding for be as much as 20 percent. Other diseases such as rice resistance and biological control, have yet to demon- yellow mosaic virus and smuts also occur but with little strate substantial success. Chemical control of the pest, impact on yield. although effective, is impracticable under current con- ditions, as the patchwork nature of smallholders’ plots allows rapid reinfestation from neighboring land.20 In the The changing agricultural landscape is giving rise to new risks related to pests and diseases. As noted earlier, maize Unless blanket treatment of a large area could be carried out, but this would 20 losses caused by insect pests in fields are rarely significant. be prohibitively expensive. Maize stalk borer (Busseola fusca) is the only pest reported 18 Rwanda Table 3.1. Main Pests and Diseases of Selected Food Crops in Rwanda Crop Pest Disease Banana •  Banana weevil (Cosmopolites sordidus) •  Panama diseases (only affect modern varieties) •  Nematodes •  Black sigatoka (Mycosphaerella) •  Banana bunchy top disease •  Banana bacterial wilt Beans •  Beanfly (bean stem maggot, Ophiomya spp.) •  Angular leaf spot •  Bean Bruchid •  Anthracnose •  Common bacterial blight •  Halo blight •  Ascochyta blight •  Rust •  Bean common mosaic virus •  Root rot Cassava •  Green spider mite (Mononychellus tanajoa) •  Cassava mosaic virus •  Cassava mealy bug (Phenacoccus manihoti) •  Cassava brown streak virus •  White fly (Bemisia tabaci) Maize •  Maize stalk borer (Busseola fusca) •  Leaf blight •  The greater weevil (Sitophilus zeamais, for in-store •  Maize streak virus grain) •  Maize chlorotic mottle virus (component of maize •  Striga weed lethal necrosis disease) Potatoes •  Blight (Phytophthera infestans) •  Sucking pests •  Potato viruses •  Bacterial wilts (caused by Pseudomonas solanacearum and by Erwinia complexes) Rice •  Rice blast (Magnaporthe oryzae) •  Leaf and panicle blight (caused by Xanthomonas spp.) •  Sheath rot (associated with Pseudomonas infection) •  Rice yellow mosaic virus •  Smuts to have caused significant losses.21 However, prior to Market Risks 2007, maize areas in Rwanda were considerably smaller In general, market risks are limited for Rwanda’s food and more dispersed than they are now and the increased crop producers. Because most markets are local, prices consolidation and importance of the maize crop that has fluctuate seasonally and in direct response to supply occurred in the last five years will undoubtedly increase and demand (figure 3.8). As such, price fluctuations do the probability of losses caused by pests and diseases. not constitute a risk but are rather caused by constraints. Again, increasing periods of storage and larger stored vol- Domestic markets for commodities such as beans seem umes will contribute to increases in related risks, as will to be well integrated, with limited disparities in terms of global climate change, because projected temperature fluctuations. Nevertheless, prices in Rwanda are to a cer- increases will provide a more favorable environment for tain degree influenced by the availability of postharvest pests and diseases. infrastructure. The lack thereof, such as for potatoes, can cause volatilities whereas there is evidence that a devel- Table 3.1 gives an overview of the main pests and diseases oped processing industry, such as for cassava and beer in Rwanda. A more detailed description of the impacts bananas, tends to provide more stable producer prices for from production risks on each crop is given in appendix E. the studied commodities. 21 Ibid. Agricultural Sector Risk Assessment 19 Figure 3.8. R  wanda Food Crop Prices (RF), January 2005–September 2013 Dry maize (grain) Shelled local rice Dry beans 700 Irish potato Sweet potato Cassava (root) 600 500 400 RF 300 200 100 0 Jan-05 May-05 Sep-05 Jan-06 May-06 Sep-06 Jan-07 May-07 Sep-07 Jan-08 May-08 Sep-08 Jan-09 May-09 Sep-09 Jan-10 May-10 Sep-10 Jan-11 May-11 Sep-11 Jan-12 May-12 Sep-12 Jan-13 May-13 Sep-13 Source: Authors’ calculations, based on NISR’s 2014 Seasonal Agricultural Survey. Figure 3.9.  Rwandan Prices of Irish Potatoes (RF/kg), JANUARY 2005–September 2013 300 250 200 150 100 50 0 Jan-05 May-05 Sep-05 Jan-06 May-06 Sep-06 Jan-07 May-07 Sep-07 Jan-08 May-08 Sep-08 Jan-09 May-09 Sep-09 Jan-10 May-10 Sep-10 Jan-11 May-11 Sep-11 Jan-12 May-12 Sep-12 Jan-13 May-13 Source: Authors’ calculations, based on NISR’s 2014 Seasonal Agricultural Survey. Sep-13 Prices in neighboring countries affect domestic prices Potato farmers regard domestic price volatility as an because of trade but global price fluctuations have little inherent production risk. Potato price fluctuations are influence over Rwandan prices. This is largely because largely the result of the limited storage and process- high transportation costs effectively insulate Rwanda ing facilities for potatoes in Rwanda; a potato shortage from global price fluctuations, especially for perishable occurs immediately before harvest and a glut immedi- commodities such as bananas. Neighboring markets ately after (figure 3.9). To avoid the impact of each glut, have more impact, but do not show significant volatility growers tend to harvest as early as possible, generally between seasons. For certain products, such as cassava, for- before the tubers are fully mature, which tends to reduce eign markets also help smooth price fluctuations in times shelf life considerably. Price volatility is offset to some of overproduction. For products such as maize, imports extent by three factors: (1) the fact that potatoes can be stabilize seasonal fluctuations. Rwanda’s membership in grown in two seasons in Rwanda; (2) the staggering of the East African Community (EAC) and its adherence to planting across different provinces; and (3) the import of open trade policies support this. Nevertheless, potato, rice, early- or late-harvested potatoes from Uganda. Although and banana producers (of other than beer banana) face both Rwanda and Uganda export to the Democratic certain marketing risks. Republic of Congo (DRC) and Burundi, Rwandan 20 Rwanda prices are determined almost entirely by production costs. The cooperative’s response was that prices were set within Rwanda and neighboring parts of Uganda. Inter- before sowing and would not be increased and that farm- national price volatility does not contribute to the risks ers should seek to improve the fertility of their land for the involved in the production of potatoes and there is no next crop. The risk for farmers thus lies with obtaining a evidence of any global market impact (for example, of high enough yield to cover costs and income needs given potatoes from Egypt or China). set prices. Growing alternative crops is not an option for rice farmers, as by law, lands developed for irrigated rice Banana growers face inherently different market risks production can be used only for that purpose. The stabil- because of the perishability and fragility of bananas. ity of domestic prices, coupled with the significant costs of Dessert bananas are mainly grown for home consump- transport to Rwanda from seaports, create a stable domes- tion in Rwanda and supply chains are therefore not well tic rice market, even though imports from T ­ anzania, developed. As a result, prices for dessert bananas can be Thailand, and Pakistan may make up 50 percent of the variable and considerable risk exists in commercial pro- market volume (figure 3.9). duction for the dessert banana market. Cooking banana prices are more stable because the fruit is harvested when Export Crops it is more resistant to damage and can therefore be trans- Export crops play an important role in Rwanda’s econ- ported to a wider market. Nevertheless, prices still fluctu- omy through their contribution to export earnings even ate, and in some cases, unpredictably. Growers of beer though their share in agriculture GDP is very small (about bananas report that prices offered by processors are more 2 percent in 2013). Tea and coffee exports account for stable. Stable prices are also quoted as a reason for grow- 81 percent of agricultural exports22 and about 20 percent ing beer bananas in preference to the other two types, of Rwanda’s goods exports (WDI 2013). The value of tea even though the beer banana yields are generally lower and coffee exports almost tripled over the 14-year period than those of cooking or dessert bananas. Rwanda also from 2000 to 2013. With government plans to expand imports bananas and prices fluctuate in parity with mar- areas under tea and coffee, these will remain an important kets in Uganda and, to a lesser extent, Kenya, the DRC, source of export earnings. and Burundi. Because of reasons discussed earlier, global prices have little impact on Rwandan banana prices. As export crop subsectors’ structures differ from those of food crops, export crop producers face different risks. As Because of the structure of the rice sector, rice produc- the term “export crops” indicates, a large share of produc- ers are faced with certain income risks. Rice prices in tion is exported and because Rwanda is landlocked, export Rwanda are determined by government policy, which ports are located in neighboring countries (­ Mombasa in sets a minimum price paid to rice mills by licensed trad- Kenya and Dar es Salaam in Tanzania). The subsectors ers. Smallholders, as members of cooperatives, receive are thus exposed to exogenous risks, including interna- inputs and produce rice that is purchased by mills at a tional price volatilities, exchange rate fluctuations, and price determined before the crop is sown. Traders are other countries’ trade policies. not allowed to buy directly from smallholders, so large mills are the only source of rice for traders. As a result of this system, neither growers nor mills face any risk from Production Risks domestic price volatility; prices and potential margins are Tea All tea produced in Rwanda is rain fed and as such is known before any investment in inputs is made. Never- subject to weather-related risks (see box 3.1). Figure 3.10 theless, such prices are not always favorable to growers; shows tea yields over time. The annual fluctuation in the for example, in December 2013, farmers in Muhanga dis- area under tea and the total production of Made tea (con- trict complained that the price they received (RF 250/kg) version rate from green leaves is 4.5) over time is shown in was inadequate to cover the costs of production at the yield they had achieved (3.5 MT/ha). They suggested that RF 300/kg would have been appropriate to cover their 22 2008–10 figures (FAOSTAT 2014). Agricultural Sector Risk Assessment 21 Box 3.1. TWO ACTORS IN THE TEA SUPPLY Figure 3.10. Tea Yields in Rwanda CHAIN (MT/ha), 1995–2011 Yield (tons/ha) Sorwathe Tea Factory was one of the first and is the 2.000 Trend 1.800 largest private tea factory in Rwanda. Construction of the 1.600 factory started in 1975 and tea production began in 1978. 1.400 1.200 At present, the factory employs about 2,500 employees and 1.000 partners with about 4,500 tea farmers who are members 0.800 of the Assopthe Tea Cooperative. Sorwathe Tea Factory 0.600 0.400 accounts for about 15 percent of Rwanda’s tea production. 0.200 The factory produces several types of tea products, includ- 0.000 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 ing black, green, and white tea. Its total annual production is about 3 million kilograms (3,000 MT) of final product Source: FAOSTAT 2014. (Made tea). The total area under tea surrounding this fac- Note: Yield figures are derived from production and area figures. tory is about 1,275 ha, of which 1,000 ha are under the cooperative and 275 ha are under the tea factory. A large share of the tea is produced on marshlands that are sub- ject to floods during the two rainy seasons. The cooperative must maintain the drainage system on a regular basis and Figure 3.11. Tea Production (MT) and rehabilitate the drainage system, when needed, to keep it Area (ha), 1997–2013 operational. The Assopthe Tea Cooperative accounts for 30,000 Area (ha) about 75 percent of green leaf tea production in the pro- duction zone of the factory (the rest is produced by the Made tea (ton) 25,000 factory itself). The factory has plans to expand its capacity as well as the variety of tea products offered in the future. 20,000 Pfunda Tea Cooperative is another key actor in Rwan- dan tea production. This tea plantation began in 1972, and 15,000 tea farmers were later organized into Pfunda Tea Coopera- tive, which presently has 1,988 members. The cooperative 10,000 sells its green tea leaf production to the Pfunda Tea Factory. Green leaf tea production increased from 4,554.5 MT in 2005 to 7,457.2 MT in 2013, an almost 64 percent increase 5,000 in eight years. About 776 ha are under tea production. A large share of the tea is produced on marshlands that are 0 subject to floods during the two rainy seasons. As with the 97 99 01 03 05 07 09 11 13 19 19 20 20 20 20 20 20 20 Assopthe Tea Cooperative, Pfunda Tea Cooperative has to Sources: NISR 2013; NAEB 2014. maintain the drainage infrastructure on a regular basis and Note: NAEB = National Agricultural Export Board. rehabilitate it when necessary to maintain green tea leaf yields. In 2013, tea on about 12 ha was destroyed by floods. Source: World Bank interviews with cooperatives. Tea planted on marshlands is subject to floods caused by heavy rains. This is particularly the case in the Northern and Western Provinces. One tea cooperative recently lost 12 ha (out of 776 ha) of tea because of floods, which now figure 3.11. In 1999, 2002, 2004, 2008, 2011, and 2012, need to be replaced. Another reported about 20 ha (out of tea yields were the result of unpredictable weather events, 575 ha) lost in 2012, damaging some 260,000 trees. Drain- that is, erratic rains, drought, and floods in marshlands. age needs to be regularly maintained to avoid flooding. Although pests and diseases have limited impact on tea production, diseases carried by insects are correlated with Hailstorms damage tea leaves and prevent plucking for up to weather incidents as dry weather makes tea bushes more three months in affected areas. One cooperative with a tea susceptible to these diseases. plantation in a marshland reported that hailstorms used to 22 Rwanda Figure 3.12. C  offee (green beans) Figure 3.13. C  offee Production (MT) Yields (kg/ha), 1995–2012 and Area (ha), 2005–13 1200 45,000 Ton of green beans Ha 40,000 1000 Drought year 35,000 800 30,000 600 25,000 20,000 400 15,000 200 10,000 5,000 0 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2005 2006 2007 2008 2009 2010 2011 2012 2013 Sources: FAO for 1995–2004; NAEB for 2005–12. Source: NAEB 2014. Note: NAEB data are inconsistent with FAO data for some years and thus NAEB data were used when available. affect about 2–3 ha, but that in 2013, 113 ha were damaged coffee production in Rwanda and the total area planted; because of hailstorms, affecting about a fifth of the coopera- yield changes clearly affected total production, with pro- tive’s productive land. Nevertheless, it is difficult to assess the duction drops in 2007, 2009, and 2011. Although coffee impacts that these types of risks have at the national level. production in 2012 and 2013 was somewhat similar to that in 2005, it would have dropped significantly if the area Coffee planted had not doubled between 2005 and 2012. Drought, pests, and diseases are all major production risks to coffee in Rwanda. Coffee yields have been somewhat tur- Pests and/or Diseases Risk bulent since the mid-2000s, with important drops in coffee Unlike tea, coffee is seriously affected by insects and production in 2007, 2009, and 2011, but the overall trend is diseases. The most common disease is CLR, caused by downward sloping (figure 3.12). This decline in coffee yields Hemileia vastatrix. Coffee yields in Rwanda are generally was the result of: (1) bad weather, including erratic rains, low and coffee plants are not in good health; plants are floods, and drought; and (2) outbreaks of coffee pests and therefore more susceptible to insect and disease attacks diseases, including intestia, coffee berry disease (CBD), and than they would be otherwise. Research demonstrates that coffee leaf rust (CLR). Coffee is very sensitive to drought CLR exists across Rwanda, but in the Eastern Province, conditions during the flowering and bean formation period almost 100 percent of the plants surveyed were affected. from October to March, and there is a clear relationship Losses range from 30–90 percent, depending on environ- between coffee yield and drought, as seen in 2007 (EARS mental conditions and varieties. Higher altitudes are less 2008). The 2010 drought also reportedly had an impact on affected (previous research found a negative correlation coffee yields in the affected provinces (Eastern and South- between altitude and CLR of –0.71). Further, the vari- ern) but this is not reflected at the national level. eties most commonly grown in Rwanda are especially Farmers reported other weather-related risks that affect susceptible to the disease. Other pests such as coffee leaf coffee, including hailstorms (which affect the quality and miner, stem borer, and antestia bugs are also a problem. weight of the coffee cherry) and mudslides caused by In the Northern Province, as much as 35 percent of all heavy rains. However, these risks are highly localized and coffee plants are estimated to be infested with coffee leaf do not have systemic impacts on Rwanda’s aggregate or ­ ercent miner, and antestia can reportedly destroy over 35 p even provincial coffee production. of coffee yields (Bigirimana, Barumbanze, Ndayihanza- maso, Shirima, and Legg 2012). CBD is currently a minor Although the drop in yields is masked in the aggregate to disease in Rwanda but as with CLR, the coffee variet- some degree by an increase in area planted, weather-related ies grown in Rwanda are susceptible to CBD and the risks still affect overall production levels. Figure 3.13 shows agro-climatic conditions are advantageous for the disease. ­ Agricultural Sector Risk Assessment 23 CBD epidemics are therefore deemed to be a potential risk based on a range of factors, including international auc- to Rwanda’s coffee growers (Bigirimana, Barumbanze, tion prices and the exchange rate. Figure 3.14 shows the Ndayihanzamaso, Shirima, and Legg 2012). REMA has fluctuations in tea prices at the Mombasa auction (where developed a national Integrated Pest Management (IPM) all Rwandan tea is sold) over the past 15 years. These fluc- framework in the context of the Lake Victoria Basin in tuations are reflected in the farm gate prices for leaf tea, Rwanda. This IPM framework should be relevant for which increased by 31 percent in 2013 and declined by ­ controlling coffee pests and diseases as well as pests and almost 18 percent in 2014. Effective as of 2012, NAEB diseases for other export and food crops. switched from a cost-based price model to the interna- tional price-based model to fix the floor price. However, Potato Taste Risk: So-called “potato taste” in coffee is most of the fall in prices is absorbed by farmers, and the a big problem in Rwanda. No consensus exists as to what farmers interviewed complained that prices are currently causes potato taste, but most experts believe it is caused insufficient to cover input costs. As a result, farmers may by the antestia bug. This insect enters coffee cherries on reduce input use, reduce investment in the rehabilitation plants that are not very healthy. According to some ad of drainage, or delay replacement of old tea plants (for hoc estimates, almost 60 percent of coffee in Rwanda is example, 30 percent of tea plants need to be replaced at affected by potato taste. One coffee exporter in Kigali one cooperative visited). All of this will affect green leaf reported that 9 out of 16 containers (56 percent) were tea yield, Made tea production, and farmers’ profitability rejected by Starbucks because of potato taste. In the over time. Rwanda Cup of Excellence competition, 60–65 percent of samples were found to have potato taste. The direct Fluctuations in tea production and prices also affect Rwan- impact of potato taste is a drop in coffee price by 15 per- da’s export earnings. The value of Rwanda’s tea exports cent or more, which results in almost US$5 million in tripled between 2000 and 2013, but growth has not been annual losses of export earnings. consistent because both the quantity exported and inter- Figure 3.15 shows national tea prices have also fluctuated. ­ Market Risks the quantity and value of tea exports from 2000 to 2013. Price Risks for Tea It is evident that declining prices played an important role Tea production prices in Rwanda are fixed by the National in the value of tea exported, especially in 2001, 2005, Agricultural Export Board (NAEB) every four months 2007, and 2013. Figure 3.14. Monthly Prices of Rwandan Tea at the Mombasa Auction (U.S. cents/kg), February 1989–October 2013 400 350 300 250 200 150 100 50 0 Feb-89 Oct-89 Jun-90 Feb-91 Oct-91 Jun-92 Feb-93 Oct-93 Jun-94 Feb-95 Oct-95 Jun-96 Feb-97 Oct-97 Jun-98 Feb-99 Oct-99 Jun-00 Feb-01 Oct-01 Jun-02 Feb-03 Oct-03 Jun-04 Feb-05 Oct-05 Jun-06 Feb-07 Oct-07 Jun-08 Feb-09 Oct-09 Jun-10 Feb-11 Oct-11 Jun-12 Feb-13 Oct-13 Source: NAEB 2014. 24 Rwanda Figure 3.15. Volume (MT) and Figure 3.17. Monthly International Value (US$ million) of Coffee Price (U.S. cents/ Rwandan Tea Exports, lb), January 2000–July 2000–13 2013 Quantity of made tea exports (ton) 350 70 Value of made tea exports (USD million) 25,000 60 300 20,000 50 250 40 15,000 US cents/lb 200 30 10,000 20 150 5,000 10 100 0 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 50 0 Sources: MINECOFIN 2013a; NAEB 2014. Jan-00 Oct-00 Jul-01 Apr-02 Jan-03 Oct-03 Jul-04 Apr-05 Jan-06 Oct-06 Jul-07 Apr-08 Jan-09 Oct-09 Jul-10 Apr-11 Jan-12 Oct-12 Jul-13 Note: Percentage share is calculated by the authors. Sources: International Coffee Organization: New York ex-dock cash price for other mild Arabica coffee. Figure 3.16. C  offee Prices in Rwanda (US$/kg), 2000–03 but farmers and washing stations are more exposed to Change in price over the Price for coffee (US$/Kg) these price changes. The farmers with whom the World previous year (%) Bank team met confirmed that prices are unpredictable; 6 100 80 one farmer reported that in 2013, prices varied from RF 4 60 130 to 350/kg for coffee of the same quality. Accord- 40 2 ing to this farmer, coffee price fluctuations are his main % change US$/kg 20 0 0 concern as they make it difficult for him to plan his pro- –20 –2 duction activities. –40 –4 –60 –80 Annual coffee price volatility has a major effect on –6 –100 Rwanda’s national export earnings (figure 3.18). The ­ 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 value of coffee exports declined over the previous year Source: MINECOFIN 2013a. in 2001, 2002, 2007, 2009, 2012, and 2013, primarily because of two factors: a decline in coffee production and a decline in the international price of coffee. In Price Risks for Coffee 2011, the quantity exported declined but export earn- Because coffee is sold in the international market, ings increased because of an increase in coffee prices in Rwanda’s coffee prices follow international prices ­ the international market. In 2012 and 2013, the value of ­(figures 3.16 and 3.17). Coffee prices have no consis- coffee exports declined because of a decline in interna- tent predictable pattern, but they do depend on the tional coffee prices, even though the quantity exported international markets. Coffee prices experienced a sig- increased (figure 3.19). nificant decline during 2001 and 2002 as well as dur- ing 2012 and 2013; they increased linearly between Exchange Rate Fluctuations Risk 2003 and 2011; and in 2006, 2008, and 2009, coffee Because both tea and coffee are exported in USD but prices declined slightly or remained stable. Processors the farmer is paid in RF, farmers are subject to exchange and exporters tend to hedge against price volatility risk rate fluctuations. The final impact on farmers, however, through forward contracts or other such mechanisms, depends on which currency appreciates or depreciates. Agricultural Sector Risk Assessment 25 Figure 3.18. F  luctuations in the Figure 3.19. F  luctuations in the Volume of Coffee Value of Coffee Exports from Rwanda Exports from Rwanda (MT), 2005–13 (US$ million), 2000–13 30,000 Quantity ton % change 50 80 US$ million % change 140 40 70 120 25,000 100 30 60 80 % change US$ million 20,000 20 50 60 % change 10 40 40 Ton 15,000 0 20 30 –10 0 10,000 –20 20 –20 –30 10 –40 5,000 –40 0 –60 00 01 02 03 04 05 06 07 08 09 10 11 12 13 0 –50 20 20 20 20 20 20 20 20 20 20 20 20 20 20 05 06 07 08 09 10 11 12 13 20 20 20 20 20 20 20 20 20 Sources: MINECOFIN 2013a for 2000–04; NAEB 2014 for 2005–13. Source: NAEB 2014 from 2005 onward. Figure 3.20.  Average Weekly Exchange Rate of RF/USD, January 2008–January 2014 675 650 625 600 575 550 525 8 8 9 9 0 0 1 1 2 2 3 3 4 00 00 00 00 01 01 01 01 01 01 01 01 01 /2 /2 /2 /2 /2 /2 /2 /2 /2 /2 /2 /2 /2 13 13 13 13 13 13 13 13 13 13 13 13 13 1/ 7/ 1/ 7/ 1/ 7/ 1/ 7/ 1/ 7/ 1/ 7/ 1/ Source: Oanda, www.oanda.com/currency/historical-rates/. Similarly, fertilizer prices in USD are converted into farm Enabling Environment Risks gate prices in RF by using the prevailing exchange rate. Contract Enforcement (Counter Party Risk): Cof- Although exchange rate fluctuations can favor farmers, fee processors and exporters have contracts with farm- they can also work against them, as farmers have no pro- ers and/or washing stations to deliver cherries and/or tection against them. However, exchange rates have not parchment at a certain time and price. Depending upon been overly volatile in past years. Instead, the value of the market price and other prevailing conditions, the the RF has steadily depreciated, which favors Rwanda’s terms and conditions of the contract are not always ful- producers and processors, depending on who captures the filled. Although Rwanda has made substantial progress gains. Conversely, any imported inputs, like fertilizers, will in improving contract enforcement, there is still scope to be more expensive. Regardless, exchange rate fluctuations make more improvements, as noted by coffee stakeholders cannot be considered a significant risk to Rwanda’s agri- in Rwanda. cultural export (figure 3.20). 26 Rwanda Logistics Risk: Rwanda is a landlocked country, known Figure 3.21. S  ystemic Losses to Milk to be made up of a thousand hills. Landslides are com- Production (whole, mon, particularly in the Northern and Western Provinces. fresh, cow) (MT), Landslides damage roads and bridges and pose domestic 1990–2011 transportation risks. Reportedly lengthy border crossings 190,000 Disease outbreaks and insecurities in other countries add further logistical –5.9% 170,000 risks. Nevertheless, although many transportation- and Drought logistics-related problems exist, these are mainly predict- 150,000 –11% able constraints rather than risks. 130,000 Civil war –5.9% 110,000 Livestock 90,000 (Dairy and Meat) 70,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Livestock are important to households in terms of income and food security and for the organic manure produced, Source: FAOSTAT 2014. which is applied in the fields. According to NISR, half of all households own a goat, cow, and/or chicken. Of live- stock units, 68 percent are cattle. The key risks for the milk Drought: A drought affects livestock production in value chain occur first at the production level, then at the a number of ways. Primarily, animals’ water intake is marketing level (that is, bulking/collecting and transport- reduced, which affects their production. Therefore, sup- ing), and finally at the retailing stage. For meat produc- ply of water is critical for livestock well-being and pro- tion, the main risks are related to production and prices. duction. Second, the water available for production is Enabling environment risks are limited for both milk and reduced, which affects activities such as cleaning of ani- meat.23 mal sheds and milk hygiene and handling. This tends to increase the incidence of disease among cattle and affects Production Risks the quality of cattle products. Last, the availability of Milk production has increased exponentially since the feed is affected. This is especially true in Rwanda, where mid-1990s, but shocks to production have been incurred access to commercial feeds is limited, and thus farm- on the way. Overall, three years had systemic shocks to ers rely on rain-fed pastures and open water s ­ources.24 milk production at the national level: in 1994, a 6 percent Milk production can decrease by as much 60 percent loss in milk production was experienced because of the during a drought (Olsson 2012). As table 3.2 shows, the war; in 2002/03, a major drought (affecting 1 million peo- 2002/03 and 2007/08 droughts had the highest impact ple) led to an 11 percent loss in milk production; and in on milk production and milk yield over the last decade. 2008, multiple disease outbreaks of anthrax, lumpy skin In both cases, despite an increase in the number of milk- disease and foot and mouth disease caused a 13 percent ing animals, milk production fell because of lower water loss in milk production compared with the previous year’s availability, which resulted in less milk produced per cow ­ production (figure 3.21). (that is, lower milk yield). In contrast, national milk pro- duction and milk yield significantly increased in 2010, despite a drought in the Eastern Province. Arguably, good rains and improved breeds increased production in 23 The national production data series covered about 12 years; therefore, the team relied on FAOSTAT data, which are available from 1961. The disease data were based on OIE data, which were available from 1997, but with a gap between 1998 and 2002. With regard to price data, the team relied on FAO- STAT data, which were only available from 1995. Therefore, the analysis con- 24 http://amor.cms.hu-berlin.de/~h1981d0z/pdf/2006-02-kenia/kabete-lect.pdf centrates on the period between 1990–2011, depending on data availability. (accessed February 2014); and TechnoServe Rwanda 2008. Agricultural Sector Risk Assessment 27 Table 3.2. Impact of Drought and Dry Both the amount of meat produced and prices increased, Spells on Milk Production, whereas the proportion of slaughtered animals did not go above the average in the 2000s except in 2008. However, Select Years the government developed a strategic plan to increase meat 2002/03 2005 2007/08 2010 production to 165,000 MT, of which beef will contribute Change in milk –11 –1.2 –13.0 26.7 about 60,000 MT. This will require improved beef and production (%) other meat breeds, which will increase the water require- Change in heads 8.9 –0.4 9.9 0.9 ments for animal production as well as food requirements. of milking Currently, the livestock sector relies on rain-fed fodder animals (%) and pastures to feed the national herd, a situation that Change in milk –18.3 –0.8 –20.8 25.5 yield (%) will not be feasible in the future if targeted production is to be achieved. Thus, as meat production increases, it is Source: FAOSTAT, 2014. anticipated that drought risk will become more important to the meat industry. other parts of the country, so the provincial drought did not affect national milk production. Past decades’ droughts have led to the displacement of live- stock in the affected areas, which has negatively affected Drought has less impact on meat production than on the sector. Droughts often force pastoralists to move their milk in Rwanda, although this is likely to change in the herds in search of feed and water, sometimes to neighbor- future. The dry spells and droughts of 2002/03, 2007/08, ing countries or into national park areas. The cattle do not and 2010 seem to have had a lagging effect, in that the cope well during these long moves and yield less milk as a decrease in production is visible a year after the event, consequence, or even die in extreme cases. These moves rather than during the drought year (figure 3.22). Nev- also result in herds being mixed, and livestock being in ertheless, the impacts are limited. For example, the 2005 contact with wild animals, both of which increase the drought did not register on meat production at all. The spread of diseases. FMD is a particular problem, as wild 2009 drop in production was likely to have been exacer- animals are carriers of the disease. The movement of ani- bated by the global financial crisis, which had an impact mals between countries increases the risk of transferring on both production and prices. diseases across borders. Figure 3.22. S  ystemic Losses to Beef Production, 1990–2011 40,000 Production (tons) Annual change (%) 25% Post 20% 35,000 drought –2.8% Postdrought 15% 30,000 Civil war years; –8.7%, –8.7, –13 financial crisis 10% 25,000 –4.2%, –4.3% 5% 0% 20,000 –5% 15,000 –10% 10,000 –15% 1991 1995 1997 1998 1999 2001 2005 2007 2008 2009 2011 1990 1992 1993 1994 1996 2000 2002 2003 2004 2006 2010 Source: FAOSTAT 2014. 28 Rwanda Table 3.3. Total Number of Livestock-Related Disease Outbreaks, 2002–12 (average) Disease New Outbreaks Susceptible Cases Deaths Destroyed Slaughtered FMD 48 266,429 758 93 262 68 CBPP 12 351,219 1,706 97 27 - LSD 123 730,195 2,434 81 91 - Anthrax 160 929,906 2,097 362 122 106 Total 343 2,277,749 6,995 633 502 174 Source: OIE 2014. Diseases: Livestock diseases can have a significant Both 2008 and 2012 were devastating years for Rwan- impact on the sector. Among the most common disease da’s livestock sector caused by the high number of dis- outbreaks are FMD, contagious bovine pleuropneumo- ease outbreaks (table 3.4). These two years accounted nia (CBPP), anthrax, black quarter, and LSD (table 3.3). for half of all new outbreaks, the number of susceptible The increase in incidence is attributed to the movement animals, and the cases seen in the 2000s. Furthermore, a of cattle across the borders with Uganda, Tanzania, and third of deaths and animals destroyed occurred in these the DRC. In the event of an outbreak, RAB quaran- two years. FMD, anthrax, and LSD epidemics struck tines the affected area(s),25 such that all livestock and live- in both years; however, besides LSD, the dominant dis- stock products cannot be sold or transported out of the ease outbreak in 2008 was FMD, whereas in 2012 it was affected area until the ban is lifted, causing a disruption anthrax. in trade as well as the possibility of discounted prices. In addition, depending on the nature of the outbreak, Although the impact of transboundary diseases is thought the government might slaughter and destroy26 animals to have been quite high in 2012, data were not avail- and animal products within the affected area. The losses able and it was therefore not possible to determine the incurred depend on the size of the affected area, the impact on milk production. Milk production decreased by number of farms and animals within the area, and the 13 percent in 2008 from 2007, primarily because of the outbreak’s duration.27 disease outbreaks mentioned above. This translates into an estimated US$10 million28 loss in milk-related income Underlying the aggregate numbers in table 3.3 is signifi- to farmers and US$163,00029 in the value of destroyed, cant variability, with years when no outbreaks occurred slaughtered, and dead cattle. These figures do not include and years when there were several. Additionally, there is the direct costs of disease control measures. variability in the number of susceptible animals and the number of cases, deaths, animals destroyed, and ani- Aflatoxins: Aflatoxins are toxins produced by mycotic mals slaughtered in any given outbreak. Therefore, each organisms that grow in poorly stored animal feed. In outbreak is unique, creating uncertainty for the govern- countries with developed animal feed industries, afla- ment as it plans and prepares for livestock epidemics, toxins have been known to cause poisoning that could a situation complicated by the government’s limited lead to death depending on the level of contamination. resources. 28 Based on FAOSTAT milk production and prices. Estimated as the difference in milk production between 2007 and 2008 multiplied by the average milk price 25 Interviews with MINAGRI staff. in 2008. 26 “Destroyed” refers to animals having to be killed and disposed of and so they 29 Based on OIE data, the “Dairy Value Chain in Rwanda” report, and the cannot be used for commercial purposes because of disease, as opposed to NISR Statistical Yearbook 2012. The report estimates the value of an exotic bull to “slaughtered,” which means that some of the animal’s value may be retained be RF 500,000, which is also assumed to be the average value of a milking cow. through sales. The total number of destroyed, slaughtered, and dead cattle were multiplied by 27 http://www.oie.int/doc/ged/D9251.PDF. Accessed February 25, 2014. the estimated value in RF and then converted to USD. Agricultural Sector Risk Assessment 29 Table 3.4.  Average Number of Disease Outbreaks Annually in 2002–11 versus 2008 and 2012 Period New Outbreaks Susceptible Cases Deaths Destroyed Slaughtered Average 2002–11 21 106,837 370 45 39 19 Actual 2008 41 470,860 191 58 120 - Actual 2012 117 845,358 3,472 168 27 - Total 2002–12 343 2,277,749 6,995 633 502 174 2008 % of total 12% 21% 3% 9% 24% 0% 2012 % of total 34% 37% 50% 27% 5% 0% Source: OIE 2014. Figure 3.23. Monthly Retail Price Variability of Fresh Milk (RF/liter), January 2005–December 2013 Fresh milk Difference with previous Standard deviation month’s price 700 20% Price difference with previous month 650 15% Fresh milk price, RF/liter 600 10% 550 5% 500 0% 450 –5% 400 350 –10% 300 –15% 250 –20% Jan-05 Jul-05 Jan-06 Jul-06 Jan-07 Jul-07 Jan-08 Jul-08 Jan-09 Jul-09 Jan-10 Jul-10 Jan-11 Jul-11 Jan-12 Jul-12 Jan-13 Jul-13 MonthͲYear Source: NISR. MINAGRI is promoting intensification of dairy farming, Market Risks which will require commercial feed production and dis- Domestic Milk Price Volatility: Annual milk pro- tribution. To this end, the Ministry is supporting the con- duction increased exponentially from 55,577 liters in struction of animal feed factories, two of which are now 1999 to 503,130 liters in 2012,30 resulting in a decrease under construction. Aflatoxin is a challenge that could in imports. Today, the proportion of imported milk is less destroy the industry unless its regulation and enforce- percent of all milk consumed in Rwanda, shielding than 1 ­ ment are introduced in the nascent stages of the indus- the domestic market from international price volatility. try’s development. As the gap between domestic milk supply and demand Maize Production Shortages: As the animal feed has narrowed, domestic price volatility has increased. industry grows, it will require a consistent and reliable figure 3.23 shows, the price of milk steadily rose as As ­ supply of maize, the main ingredient, comprising about consumer demand increased. However, as milk sup- 60  percent of the feed. Therefore, should a shock affect ply approached milk demand, prices became increas- maize production, the animal feed industry would suffer, ingly more volatile, with several dips below the standard and the cost would be passed onto farmers. There are few signs that this has affected production in the past, however. 30 Livestock data from MINAGRI. 30 Rwanda deviation beginning in January 2010. With the availabil- Figure 3.24. Meat Prices (US$/MT) ity of water in the wet season, a price drop is expected; and Meat Production however, the monthly price difference is widening. (MT), 1996–2011 A 15 percent drop in the milk price (as seen in Sep- Production (tons) tember 2012) would lead to significant losses in house- 40,000 Meat price (USD/ton) 3,000 hold income for dairy farming households. The reason 35,000 2,500 –38% for this volatility is not clear. In general, this kind of 30,000 –33% –25% 2,000 price volatility can occur when daily milk consump- 25,000 USD/ton 1,500 Tons 20,000 tion is fairly constant (that is, demand is relatively sta- 15,000 1,000 ble), as even small shifts away from equilibrium supply 10,000 levels will lead to high price volatility. Similarly, high 5,000 500 demand price elasticity for milk may magnify volatility 0 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 at smaller changes in supply, as consumers will be quick to respond to price changes. Differences between sup- Source: FAOSTAT 2014. ply and demand, upward price pressure from proces- sors, and a quality-related price premium on milk are some of the explanations given by actors in the sector. At the consultative stakeholder meeting, questions were drops (by 2.9 percent). The level of imports and exports raised regarding the quality of the price data. A deeper were on average below 1 percent of production over the analysis is needed to understand if and why the price period analyzed and are therefore thought to have had volatility reflected in the data exists, but this is beyond a minimal effect on prices. The 2009 event was driven the scope of this analysis. by the impact of the global financial crisis on the Rwan- dan economy, which slowed down in 2009. The 2009 real GDP growth rate was 6.2 percent and real GDP per Meat Price Risks: Meat prices have seen both multi- capita was 3.2 percent, compared with an average of annual decreases and increases over the past decades, with 8.2 percent and 5.1 percent, respectively, between 2008 several deviations from longer-term trends (­ figure 3.24). and 2012 (Ministry of Finance and Economic Planning The fact that animals are slaughtered on order or based [MINECOFIN 2013b]). on a contract partly mitigates this. Because meat is sold fresh, butchers and retailers order based on their experi- ence of what their customers will be able to buy. However, Potential for Milk Contamination: Only about 20 price drops definitely have implications for farmers, who percent of the milk in Rwanda is processed; the rest is assume most of this risk. Using a drop of ≥25 ­ percent as either marketed directly in the community or through a threshold, the frequency of price shocks is one in five traders and retailers. For milk not sold directly in the years. Anecdotal evidence suggests that prices were quite community, milk aggregators are the key to bringing high immediately following the war because of low avail- economies of scale to milk processors and traders, as ability of meat on the market. As production increased, Rwanda’s herd sizes are too small to supply individu- prices gradually reduced to a low of US$1,238/MT ally. Monitoring of standards is limited at milk col- in 2003 except for sharp falls in 1996 and 2000 (of lection centers (MCCs) and farmers have differing 38 ­percent and 25 percent, respectively). After 2003, milk handling standards and levels of hygiene. Fur- prices consistently increased until another sharp drop of ther, hygiene standards differ widely between MCCs. 33 percent in 2009. Because the majority of marketed milk is not pasteur- ized, this creates a risk for contamination, particularly The causes of the shocks in 1996 and 2000 are not clear, with ­Salmonella spp., Escherichia coli, and Brucella spp. but prior to both events there were three or more high- Traders and consumers generally boil milk before it is production years before the fall in prices. In 2000, meat sold or consumed, respectively, thus mitigating the risk production dropped slightly in parallel with the price of contamination. However, this practice is not strictly Agricultural Sector Risk Assessment 31 enforced, so there is a risk of milk spoiling and/or Enabling Environment infecting consumers with a virulent strain of bacteria. Drug and Livestock Inputs’ Contamination and Depending on the size of the batch, this can have large Adulteration Risk: Similar to the risks mentioned consequences for consumers and suppliers. Because above, contamination and adulteration are risks that the of interruptions in electricity supply, the cold chain is livestock sector could face in the future. The regulation not necessarily kept cold and the current conditions of the veterinary drug industry is the responsibility of are conducive to the evolution of a heat-resistant and RAB under MINAGRI. However, its capacity to monitor highly virulent bacterial strain. and regulate veterinary pharmacies is limited, and it cur- rently does not have the resources (financial and human) The risk of contaminated milk lies both with consumers or facilities to test the drugs on the market. As this report and producers. Consumers are at risk of being infected by has shown, diseases pose a real threat to the sector, so it is contaminated milk; producers, traders, and retailers risk important that the main method of solving and mitigating losing markets if they deliver contaminated milk. For trad- disease risk does not become a threat itself. ers and retailers, the main concern is shelf life. Because contaminated milk spoils more quickly and has a shorter shelf life, traders and retailers risk returns on milk stocks Regulatory Changes in if they do not last as long as expected. As milk consump- the Agriculture Sector tion is still relatively at a low level, this is currently not a During the work on this report, many stakeholders in the significant problem. Nevertheless, if past years’ increases private sector (particularly processors) pointed out risks to in milk continue, the risk of contaminated milk is likely to the regulatory environment in frequent policy changes have broader impacts in the future. that make investments unpredictable. However, an over- view of government policy over the past decade does not Potential for Meat Contamination: MINAGRI’s reveal erratic agricultural policy changes (table 3.5). Impor- Strategic and Investment Plan to Strengthen the Meat tantly, no subsector has been specifically targeted or favored Industry in Rwanda highlights the constraints related through specific tax and/or trade regulations. The exception to the sanitary conditions in which meat is slaughtered, seems to be to encourage the domestic processing industry transported, and sold. Given the low level of meat con- through regulation of raw material (leather) and tax breaks sumption in Rwanda, sanitary conditions are considered (processed coffee). Also, Rwanda adheres to the East Africa future risks that will have to be addressed as the meat Common Market Protocol, which was introduced in 2010 industry grows, but are of limited risk at present. and should enhance the predictability of trade policy. 32 Rwanda Table 3.5. S  ummary of Regulatory Changes in Rwanda’s Agriculture Sector, 2001–13 Date Reform/Change Note 2001 GoR lifts the ban on milk imports from Uganda Ban imposed in March 1999. 2005 Income Tax Act (profit and income tax rules and rates) Exemptions related to agriculture sector: or Law 16/2005 •  Farm enterprises are exempt from tax with turnover up to RF 12 million/year. •  Agricultural and livestock products except for those processes (locally processed milk is exempt) as well as agricultural inputs and equipment 2005 Ban on importation of poultry products by GoR Outbreak of avian flu. Sept 2005 Ban on export of raw hides and skins The official position of GoR was that of developing the leather industry. Sept 2005 The ban on export of raw hides and skins is temporarily The first decision was implemented without notice— recalled leaving large stocks with no market. This decision was meant to allow those involved to resume for three more months. At the end of this period, companies were expected to have made progress toward setting up tanneries to produce the material locally. 2008 Import tariff decreased on food products 2008 GoR lifts ban on the importation of poultry products 2009 GoR temporarily lifts ban on export of raw hides and Prices of raw/unprocessed hides and skins went down in skins the meantime (2005: RF 1,500/kg; 2008: RF 500/kg) 2010 GoR removes export tax on owners of coffee processing facilities 2010 The East Africa Common Market Protocol comes into effect, allowing free movement of goods, services, capital, and labor among Members 2012 Law passed governing agrochemicals (fertilizers, pesticides), placing them under regulated imports and introducing requirements for imports 2012 New guidelines on milling and trade of rice 2012 Rwandan Cabinet approves a tea pricing mechanism that This mechanism means that farmers who produce high- provides market-based pricing and rewards quality quality green leaf tea will earn more. In turn, the standards (the previous mechanism was cost-based) quality and price of tea made by factories will increase. Sept 2013 18 percent VAT charge introduced on processed staple foods, including rice Agricultural Sector Risk Assessment 33 CHAPTER FOUR ADVERSE IMPACTS OF AGRICULTURAL RISK The existence of agricultural risk has negative consequences for the productivity and production of agricultural commodities as well as the level of profits and investments in agribusiness for various supply chains. This can be measured in the form of losses resulting from the prevalence of various agricultural risks. The purpose of this chapter is to quantify the production losses for individual food and export crops in different provinces of the country, as well as the aggregate losses. This is important to under- stand how frequently risks occur, the volume and monetary value lost in each risk event or for each crop, and the geographic distribution of these losses. Ultimately, this will help in identifying and targeting risk management interventions in a way that has the greatest ability to minimize risk-related losses. The Methodology Used to Estimate Production Losses Losses that occur because of agricultural risks refer primarily to production losses caused by weather-related events such as droughts, floods, erratic rains, landslides, hailstorms, and diseases and/or pest outbreaks. The following method was applied to calculate production losses in a particular year: (1) a historical linear trend line for the yield of each crop was constructed; (2) a second linear trend line was drawn, represent- ing one-third of the standard deviation of the crop yields; (3) loss years were identified as those in which actual yields were lower than the linear trend line; (4) production losses were calculated using the difference between the predicted value (the original trend line) and actual yield; and (5) losses were totaled and divided by the total number of years examined to determine the average annual loss rate for a particular crop; (6) the annual quantity lost was converted into value terms by using the producer price for each crop; and (7) because producer prices are in local currency, the value was con- verted to U.S. dollars using the average exchange rate. Figure 4.1 shows the outcome of steps (1)–(5) for a hypothetical crop. Information about the production loss for a particular crop and in a particular year can also be used to (1) calculate the loss as a share of agricultural GDP for that crop in a particular year; (2) add production losses for different crops to estimate aggregate Agricultural Sector Risk Assessment 35 production losses for all crops; and (3) add the production This analysis covers the selected food and export crops losses of all crops over a number of years to estimate the in Rwanda. Valuation was done by using an average of indicative production losses in a particular period. each crop’s annual producer prices from 2009, 2010, and 2011 from FAOSTAT. Each crop’s production value was Most of the data used in the loss analysis were obtained calculated by taking the 2009, 2010, and 2011 average. from the FAO Corporate Statistical Database (FAO- STAT). Generally, MINAGRI supplies data to the Food and Agriculture Organization (of the UN) (FAO). FAO Indicative Crop Production Losses in turn sanitizes the data and makes them consistent with Using the methodology outlined above, indicative produc- data from other countries using their own methodology. tion losses caused by various production risks for individ- ual crops were estimated for the selected food and export crops (summarized in table 4.1). The results indicate the following: Figure 4.1. Example of Crop 1.  Average annual production losses are US$65 mil- Production Loss lion for the selected crops; Calculation 2.  These production losses are 2.2 percent of the 20 Yield (tons/ha) y = 0.3033x + 5.5697 total value of crop production; 18 R 2 = 0.7238 Trend 3.  Cassava and bananas account for almost 60 per- 16 0.3 trend Linear (yield cent of all the estimated production losses; 14 (tons/ha)) 4.  Total production losses over a period of 18 years 12 10 (1995 to 2012) are estimated at US$1.16 billion. 8 6 These losses are too large and affect the government’s 4 growth objectives. 2 0 The correlation between losses in the sector and growth 80 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10 12 in agricultural GDP is not exact (figure 4.2). However, 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 Table 4.1.  Summary of Indicative Production Losses for Rwanda’s Food and Export Crops, 1995–2012 Annual % Loss of Crop Average Annual Average Annual Production Total Losses Total Losses Crop Losses (MT) Losses (US$) Value* (MT, 1995–2012) (US$, 1995–2012) Cassava 82,326 24,656,594 0.81 1,481,865 443,818,687 Maize 9,658 3,538,799 0.12 173,849 63,698,384 Potatoes 34,507 7,919,246 0.26 621,117 142,546,426 Bananas 89,458 17,957,199 0.59 1,520,785 305,272,377 Beans, dry 7,586 3,733,660 0.12 136,541 67,205,881 Rice, paddy 665 297,948 0.01 11,971 5,354,960 Sweet potatoes 38,027 6,202,261 0.20 684,492 111,640,698 Coffee, green 969 1,347,368 0.04 16,476 22,905,262 Tea 637 97,371 0.00 10,826 1,655,305 Total 263,833 65,749,995 2.17 4,657,925 1,164,097,981 Sources: FAOSTAT 2013; Authors’ calculations, based on NISR’s 2014 Seasonal Agricultural Survey. *Of 2009–11 average agricultural production value. 36 Rwanda Table 4.2. C  ost of Major Adverse Events for Crop Production, 1995–2012 Indicative Loss Value % Ag. Production US$ (in Value (current, Year millions) average 2009–11) Causes/Risk Events 2001 138,241,657 –4.57 Excessive rainfalls in the Northern and Western Provinces 2004 150,078,184 –4.96 Heavy rains in high altitude areas and a drought in Eastern and Southern Provinces 2006 87,062,028 –2.88 Drought/high heat in Eastern and Southern Provinces 2007 238,236,805 –7.87 Drought in Eastern Province 2008 269,030,202 –8.89 Drought in Eastern Province Sources: FAOSTAT; Authors’ calculations. Note: Plantain, tea, and coffee were calculated 1995–2011. Cassava, paddy rice, sweet potatoes, maize, dry beans, and Irish potatoes were calculated 1995–2012. Figure 4.2. Losses and Growth in Agricultural Value Added, 1995–2012 350,000,000 35.00 Losses Ag GDP growth 300,000,000 30.00 25.00 250,000,000 20.00 % change 200,000,000 US$ 15.00 150,000,000 10.00 100,000,000 5.00 50,000,000 0.00 0 –5.00 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Source: Authors’ calculation based on FAOSTAT 2014 and WDI 2014. between 1995 and 2012, the biggest monetary losses distributed losses between Irish potatoes and sweet pota- occurred in 2001, 2004, 2006, 2007, and 2008 (table 4.2), toes. Maize has relatively frequent losses but the losses are amounting to up to 8.9 percent of total agricultural pro- not as large as for the first four crops. The scope of the duction value in 2008. It is clear that in terms of both mon- losses are clearly in line with the importance of the crop in etary value and as a share of agricultural production value, the total sector, as cassava, plantains, potatoes, and maize losses became significantly greater in the 2000s (­ figure 4.3). dominate agricultural production in terms of value. In an environment of scarce resources, this may have implica- There are important differences in losses between individ- tions for risk management policy decisions when deciding ual crops. Cassava and plantain experienced the biggest on which crops to allocate resources for risk mitigation losses in the period 1995–2012, followed by fairly evenly (figure 4.4). Agricultural Sector Risk Assessment 37 Figure 4.3. Agricultural Production Losses and Share of Total Production Value in Specific Years 350,000,000 300,000,000 2008 Loss (US$) 250,000,000 200,000,000 2007 150,000,000 2004 100,000,000 2001 1999 2006 50,000,000 2000 2005 1995 1996 1997 2009 2010 2012 1998 2003 2011 0 1990 1995 2000 2005 2010 2015 –50,000,000 Source: Authors’ calculations based on FAOSTAT 2014. Note: The size of the bubbles signifies the scope of the loss in terms of its share of total production value in that particular year. Figure 4.4.  Frequency and Scope of Losses per Crop, 1995–2012 600 Cassava 500 Losses per crop (m US$) 400 Possible break-off borders for strategic Plantains risk management 300 200 Irish potato Sweet potato 100 Beans Green coffee Maize Rice 0 Tea 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 –100 Frequency of losses Sources: FAOSTAT; Authors’ calculations. Indicative Production Losses by bananas, cassava, and Irish potatoes using the 2000–12 Provinces disaggregated data from MINAGRI. Although risks are generally assessed in terms of the national impact in this report, there is value in disag- Overall, the results indicate that losses are the greatest in gregating losses and considering them at the provincial absolute terms in the Northern Province and smallest in the level. This will help optimal targeting of interventions City of Kigali (which also produces a lot less than the other geographically. Indicative crop production losses were provinces). Figure 4.6 provides an overview of the value of calculated for the five provinces in the country: Northern, annual losses by province for Irish potatoes, cassava, maize, Western, Eastern, Southern, and the City of Kigali (see and bananas. The bulk of the losses associated with Irish figure 4.5). The relative volatility among different prov- potatoes is in the Northern Province but a large amount inces was measured using the coefficient of variation (CV) also occurs in the Western Province. Most of the losses of of yields.31 Losses were estimated for four crops: maize, cassava take place in the Southern Province, followed by the Western and the Eastern Province. Banana losses are more evenly distributed between provinces, but the Eastern Prov- 31 This is calculated as the standard deviation by the series’ arithmetic median. It shows the extent of variability in relation to the population mean; that is, the ince has slightly higher losses than the other provinces and higher the CV, the worse the risk. the Western Province the lowest. Maize production has the 38 Rwanda Figure 4.5. R  wanda’s Five Figure 4.6.  Annual Losses by Provinces Province (US$, sum of Season A and Season B) Irish potato Maize 30,000,000 Cassava Bananas 25,000,000 20,000,000 US$ 15,000,000 10,000,000 5,000,000 0 North West South East Kigali Sources: Authors’ calculations, based on NISR’s 2014 Seasonal Agricul- tural Survey. Source: Central Intelligence Agency 2010. Note: See appendix D for detailed tables with exact values. Table 4.3. Production Volatilities by Province (CVs of yields, %) Maize Maize Bananas (1998–2006) (2007–12) Cassava Irish Potatoes Season A (1998–2012) Eastern 26.8 31.1 33.6 31.9 46.1 City of Kigali 25.3 32.1 50.4 34.8 46.6 Northern 19.0 9.7 45.6 34.2 78.8 Southern 15.0% 21.1 51.4 46.0 43.0 Western 14.4 12.9 40.8 40.8 61.1 National 18.2  6.8 38.2 35.5 27.6 Season B (2000–12) Eastern 17.5 23.9 55.6 47.8 15.2 City of Kigali 10.2 13.6 38.7 47.1 12.4 Northern 13.1 15.6 42.5 44.2 26.5 Southern 10.3 43.6 37.8 46.3 12.8 Western 11.3 22.9 38.9 39.5% 21.6 National 12 21.9 41.9 44.5 16.9 Sources: Authors’ calculations, based on NISR’s 2014 Seasonal Agricultural Survey. lowest losses in absolute terms relative to the other crops, crop. Detailed loss estimates for each respective province with slightly higher losses in the Western Province than in are presented in appendix D. the other provinces, followed by the Northern, Southern, and Eastern Provinces and the City of Kigali. Although Volatility in the provinces is also high for most crops. For the Northern Province has the highest aggregate losses in cassava, volatility is relatively high across seasons, whereas absolute amounts, it is clear that the geographic target area for Irish potatoes, Season A production is a lot more vola- for any risk management intervention will depend on the tile than that of Season B (table 4.3). Agricultural Sector Risk Assessment 39 The results indicate that banana production volatility is Southern Province, where 40 percent of cassava is planted. highest in the Eastern Province in both seasons, and this During Season A, the Southern Province demonstrates area also has the most production. Whereas production the highest CV of yields (46 percent), although the CVs volatility is relatively even across provinces, from a low range from a low of 31.9 percent in the Eastern Province of 14.4 percent in the Western Province to 25.3 percent to a high of 40.8 percent in the Western Province. During in the City of Kigali (Season A), and a low of 10.2 per- Season B, the Eastern Province is most volatile, with a CV cent in the City of Kigali to 13.1 percent in the North- of 47.8 percent, whereas the Southern Province is third, ern Province (Season B), volatility is markedly higher with a CV of 46.3 percent. Cassava is generally unaf- in the Eastern Province, at 26.8 percent (Season A) and fected by weather, but suffers from losses linked to pests 17.5 percent (Season B). This means that the area most and diseases at a national level. subject to production volatility of bananas also produces the most and so will feel the effect of risks most strongly. Irish potatoes are mostly grown commercially in the The FAO considers bananas as highly sensitive to mois- Northern and Western Provinces, the two provinces with ture stress (­Brouwer, Pins, and Heibloem 1989) and the the highest CVs of yields in both seasons. In Season A, Eastern Province is widely accepted as a dry, hot, lowland the Northern and Western Provinces are significantly zone. Other losses for bananas are generally incurred on more volatile, with CVs of 78.8 percent and 61.1 percent, an individual level, such as localized flooding and wind. respectively. Potato production is very dependent on soil Pests and diseases are also a problem, but it is hard to moisture and calculations show that yields are substan- disaggregate these at the provincial level. As Rwanda is a tially reduced when soil moisture is low, although this only small country, market risks (that is, domestic and interna- occurs 1 year in 10. tional price volatility) do not differ much across provinces. Although disaggregate losses were not estimated, other There is large volatility in maize, particularly in the crops of note have provincial risk dynamics: beans and Southern Province, with a CV of 78.4 percent in Sea- rice. Beans are considered by FAO to have medium-high son A and 70.8 percent in Season B. Seventy-five percent sensitivity to moisture stress (Brouwer, Pins, and Heibloem of all households grow maize, three-quarters of which 1989), so the probability of yield loss of arguably the most is grown in Season A. Research indicates that given the important crop in Rwanda in terms of national consump- differing soil types and depths in the country, it is diffi- tion is particularly high in the Eastern and Southern Prov- cult to attribute all volatility to one factor, such as erratic inces. Calculations indicate that erratic rainfall/moisture rainfall. However, the Eastern and Southern Prov- stress contributes to approximately 50 percent reductions inces received less rain compared with the Western and in the potential output of beans in the Eastern Province. Northern Provinces and do show more volatility (CVs of The impact is less in the other three provinces, particu- 57.7 ­percent and 78.4 percent compared with 48 percent larly the Western and Northern Provinces, but is still sig- and 50 ­ percent, respectively, in Season A). Flooding simi- nificant. Provincial climatic differences also play a role in larly varies greatly with location, and does not exhibit a the impact of some pests and diseases, including anthrac- specific trend, although much of the maize in the Eastern nose and ashocyta blight. Both of these diseases thrive in Province is grown in the lowlands of the Akanyeru River cool and wet conditions and are therefore more prevalent basin, where additional moisture even under dry condi- in the Northern and Western Provinces. tions is above average, but there is an increased risk of losses from flooding. Pests and diseases show provincial Rice is grown almost exclusively in the bottom of the tendencies, with MCMV identified in the Western and lower valleys where temperatures are high enough to Northern Provinces. Striga weed has infested maize pro- sustain growth, and marshy conditions provide adequate duction in the Eastern Province, but is generally predict- water in the Western, Southern, and Eastern Provinces. able so more of a constraint than a risk. Volatility in cassava production is high in all provinces Less data were collected on export crops; however, and across seasons. The highest volatility occurs in the a few conclusions can be drawn on provincial risk 40 Rwanda disaggregation. Tea and coffee, the main export crops, are grown across the entire country although more farmers Particularly Vulnerable in the Southern and Eastern Provinces grow coffee, and Groups in the Northern and Western Provinces, tea. It can gener- Over the past decade, Rwanda has made significant prog- ally be said that coffee farmers are more susceptible to ress in reducing poverty, from 57 percent in 2005/06 drought given the agroclimatic realities in the Southern to 45 percent in 2010/11. Extreme poverty decreased and Eastern Provinces. The Northern and Western Prov- from 36 percent to 24 percent in the same period. The inces, in comparison, are hilly and receive more rainfall, increase in agricultural productivity is partly attributable which means that tea production is more at risk from soil to this achievement. Nevertheless, many groups remain erosion and landslides in the hilly areas and from floods in vulnerable, not the least in rural areas, where 49 percent the marshlands. of the population lives below the poverty line compared with 22 percent in urban areas. Poverty is higher for those Livestock holdings are largest in the Eastern Province, mainly engaged in agriculture and overall 40 percent of which is dry and flat and therefore has the most optimal households in Rwanda can be classified as “low-income” conditions for pastoralism, but this is also where most risks agriculturalists. Further, poverty is higher in female- and are found when disaggregating provincially. This may be widow-headed households compared with the national partially explained by Rwanda’s geographic location and average (table 4.4). shared borders with Tanzania and Uganda in the east, which increases the cases of transboundary pest and dis- ease outbreaks, but also because of the relatively higher Food Security proportion of cattle in this province and the area’s pas- Food insecurity is closely linked to the agriculture sector. toralist movement (transhumance) history. During years The World Food Programme’s (WFP’s) survey reports that of dry spells, pastoralists moved to Uganda, Tanzania, 36 percent of rural households had unacceptable food or within Rwanda depending on where there was water consumption in September 2011 and could be considered and pasture. But the GoR has more recently encouraged food insecure, compared with 3 percent in Kigali City. pastoralists to settle, limited their movements, and distrib- Households with less diversified incomes are more food uted land for farms to them. However, FMD and CBPP insecure, and of those households with only one activity outbreaks have only occurred in the Eastern Province, (43 percent of Rwandan households), most are engaged whereas anthrax and LSD outbreaks have been found in in agriculture. Further, in the WFP Food Security and almost all provinces, but mostly in the Eastern and West- Vulnerability Assessment, agriculture (size of land culti- ern Provinces. vated in Season A, crop diversity, ownership of livestock, Table 4.4. Poverty in Different Groups of Households, 2000/01 versus 2010/11 2000/01 2010/11 Type of Population Poverty Population Poverty household Share (%) Incidence (%) Share (%) Incidence (%) All households 100 60.4 100 45 Urban 18 22 Rural 82 49 Female headed 27.6 66.3 23.8 60.2 Widow headed 22.0 67.7 18.7 59.9 Child headed 1.3 60.1 0.7 56.9 Sources: MINAGRI 2010; NISR 2012b; World Development Indicator Database, accessed in 2011; WFP 2012. Agricultural Sector Risk Assessment 41 Figure 4.7. Households with Unacceptable Levels of Food Consumption Source: WFP 2012. cultivating a kitchen garden, whether the household still food stocks. Seasonal food access problems occur in the had food in stock from the last harvest in April) was one lean seasons just before the two main harvests (from March of four variables found to be statistically significant in to May and from September to November) because food explaining household food consumption. It can be noted stocks run out. The households most exposed to seasonal that food insecurity is highest in the Western Province, food insecurity were the poorest and those relying most although this province has less volatility and lower pro- on seasonal work. Thus, food crop losses caused by risk duction losses in absolute terms than do the Northern and events have a direct impact on seasonal food security. Southern Provinces (tables 4.2–4.5). Instead, food insecu- rity in this province is structural and due to geographic Rural households consume a significant share of their pro- and agro-ecological conditions, such as relatively infertile duce within the household. On average and for all crops soils compared with other provinces, the prevalence of produced, households sold 23 percent of their production land located on steep slopes, and long distances to mar- and consumed 71 percent. The rest was reported as either kets. However, the Northern and Southern Provinces given away (2 percent) or spoiled/lost after harvest (3 per- have higher food insecurity than the Eastern Province, cent). The main consumed cereals, roots, and tubers as and 15–28 percent have unacceptable levels of food con- well as beans and cooking bananas are mostly kept for sumption in these two provinces (figure 4.7). home consumption. In contrast, households sold more than half of their production of cash crops (tea, coffee, The food security status in Rwanda is mixed and about pineapples, and sugar cane—all over 85 percent sold) and 20 percent of those who are food insecure report seasonal fruits and vegetables (tomatoes—80 percent sold, passion food insecurity. Over half of those who are food insecure fruit—60 percent, cabbage—58 percent) in addition to are chronically or acutely food insecure. After Seasons A sorghum (54 percent) and rice (63 percent), meaning that and B, 60 percent of households should have acceptable these are more important sources of income (table 4.5). 42 Rwanda Table 4.5. P  ercentage of Households Table 4.6. S  ources of Food and That Grow Specific Crops Food versus Nonfood and Share of Production Expenditures, 2012 Sold on Markets, 2012 Share of Total Change Households Crop Sold in Consumption since Growing Crop (%) Market (%) (%) 2005/06 (%) Beans 90 12 Food purchases 26.6 + 24 Sweet potatoes 45 11 Consumption of own 15.8   – 6 food Maize 42 22 Total food consumption 42.4 + 11 Plantains 28 30 Nonfood expenditure 57.8 + 38 Irish potatoes 15 32 Total 100 + 24 Cassava 40 23 Source: WFP 2012. Source: WFP 2012. Markets provide little over 60 percent of the household products, are underrepresented in agribusiness, and are food basket, whereas own production contributes about employed in low-paid positions in secondary agriculture. 37 percent (table 4.6). The market is the main source for Female-headed households constitute about 30 percent rice (81 percent), groundnuts (67  percent), fish and meat of Rwanda’s households and these households are very (90 percent—except poultry: 50 percent), and milk (55 poor, which has consequences for their access to produc- percent), meaning that prices affect access to these food tive inputs and assets. High poverty levels in these house- products. holds also make them vulnerable to shocks, as they don’t have assets to cushion the impacts. As discussed above, In conclusion, losses caused by agricultural risks have livestock has important impacts on food consumption and ripple impacts on income, poverty, and food insecurity, income, but because of gender structures, larger livestock and especially seasonal vulnerability among farmers. In (such as cattle and goat) are generally a man’s domain, general, impacts on food security will depend on how which restricts women from profiting from these assets. much is grown for consumption versus sales (for example, Thus, livestock risks should have more impact on men’s whether the farmer is a net producer or net consumer of incomes than on women’s. agricultural products) and if the risk is a production risk or a market risk. Because cereals, tubers, roots, beans, and A clear gender divide exists in the types of crops culti- bananas are mainly grown for home consumption, losses vated. Because land is traditionally controlled by men, of these products have a direct impact on household food crops produced by men are allocated more land. The security, especially for households with limited resources types of crops dominated by men versus women are not to buy these food items to compensate for insufficient pro- consistent across the country, but depend on the potential duction. In turn, prices of rice and animal products affect income from each crop in that particular area. The pro- food access because the majority of households purchase duction of crops with higher income potential tends to be these products. controlled by men. Few women are involved in coffee and tea production activities and their value chains are highly Gender and Vulnerability in gender divided, whereby men seem to benefit more from Agriculture labor inputs and control proceeds from sales. The agriculture sector is worked largely by women, but much of their labor input goes uncompensated or is not Although not comprehensive for Rwanda as a whole, and visible in official statistics. Women are primarily restricted allowance should be given to differences between families to subsistence agriculture, receive low prices for their and individuals, based on table 4.7 and the loss analysis Agricultural Sector Risk Assessment 43 Table 4.7. Gender Division of Crop Cultivation for Different Districts Crops Cultivated by Crops Cultivated Crops Cultivated by District Women by Men Both Men and Women Bulera (North) Beans Irish potatoes Maize, wheat Gasabo (Kigali) Beans, sweet potatoes, cassava, Plantains, coffee, exotic vegetables Fruits maize, amaranth (Amaranthus) (tomatoes, eggplants, cabbage, green peppers) Kirehe (East) Maize, beans, flowers Plantains, coffee, pineapples Sorghum Nyabihu (West) Maize, beans, sorghum Irish potatoes, cabbage, carrots Highlands Beans Tea (but supply chain is gender divided), Irish potato, wheat, and maize Middle veld Beans, sorghum, sweet potatoes, Coffee cassava Ruhango (Kigali) Beans, sweet potatoes, vegetables Cassava, coffee, rice Maize Source: MINAGRI 2010. previously discussed, it is possible to draw some conclu- from ongoing agricultural programs to strengthen poten- sions about the different impacts of risks on men and tial risk management interventions. For example, women women producers: reportedly have less access to technologies promoted »» Beans are mainly grown by women; therefore, any under the CIP. Partly, this has to do with their more lim- risks related to beans will be borne by women. ited access to financial capital and assets, as the improved »» Maize is grown either by women or by both men varieties, fertilizers, and chemicals promoted under the and women depending on location. The highest vol- program are expensive. In particular, female-headed atility in maize production occurs in the Southern households seem restricted from optimal participation in Province, whereas the biggest losses are in the West- the activities under the program. However, the technolo- ern Province. No gender-disaggregated production gies being promoted are also very labor intensive, which information is available for the Southern Province, reportedly restricts women from participating on equal but the large losses in the Western Province have a terms. Similarly, the “One Cow per Poor Family” pro- disproportionately high impact on women. gram planned for 30 percent of the beneficiaries to be »» Bananas are mainly grown by men, so any risks women, but given the financial costs involved (owing to related to bananas will be borne by them. the necessity of developing zero-grazing infrastructure), »» Of all provinces, the Northern and Western Prov- women and especially female-headed households are inces experience by far the highest production largely excluded from this program. It is thus important losses and volatility for Irish potatoes, where it is that any risk management activity is designed to keep in grown predominately by men, who consequently mind gender differences in access to programs and to bear most of these losses. incorporate the needs of both male and female actors in the sector to minimize agricultural risks. This type of analysis is important to secure a complete overview of risk impacts because it can aid the design Appendix C gives a more detailed overview of particularly of risk management interventions. Similarly, important vulnerable groups and gender differences in ­ Rwanda’s lessons on gender access and participation can be drawn agriculture sector. 44 Rwanda CHAPTER FIVE RISK PRIORITIZATION AND MANAGEMENT A variety of risks exist across Rwanda’s agriculture sector. Previous chapters described the major risks affecting Rwanda’s agriculture sector and specific commodity groups (food crops, export crops, and livestock), attempted to quantify losses associated with these major risks, and assessed their impacts on actors in the agriculture sector. Special attention was given to impacts on Rwanda’s most vulnerable groups and how those impacts affect men and women differently. The analysis undertaken during the risk assessment allows for prioritization of risks in relation to the probability of events and their degree of impact, followed by the relevant measures to manage these risks. In Rwanda, many programs and activities funded by government and donors are already in place to make the agriculture sector more resilient. Furthermore, businesses, individual farmers, and consumers may adopt other measures, such as managing higher prices and limited availability of certain commodities by substituting for others. This leads to questions about the effectiveness of existing activities and the sufficiency of their coverage. Options for better man- agement, taking into account the capacity to implement and fund them in Rwanda, should also be considered. This chapter prioritizes existing risks and provides recommendations for how Rwanda can more effectively manage risks, based on a consultative stakeholder exercise. It identifies priority areas for risk management interventions that will be explored in depth in the Phase II Solutions Assessment. Risk Prioritization Identifying and prioritizing risks is an important first step in designing a set of compre- hensive and effective measures to manage risks. With scarce budgetary resources, it is crucial that some sort of prioritization takes place to understand the key risks in terms of frequency of occurrence and degree of impact. Given the data constraints, it is important to note that this list is not exhaustive and the ranking of risks is based on the World Bank team’s evaluation from both data analysis and on-the-ground research. The differing significance of these risks to different stakeholders in society can be Agricultural Sector Risk Assessment 45 found in the previous chapter; vulnerability is discussed in to their importance for a particular crop, but according to the vulnerability assessment. their importance to the sector. As such, the main risks to Rwanda’s agriculture sector are: To get an overview of the frequency and the severity of 1.  Pests and diseases for crops and livestock; the key risks to food crops, export crops, and livestock, and 2.  Weather-related risks for crops and livestock; and to prioritize them, a national risk prioritization matrix was 3.  Price volatility for export crops and dairy producers. developed (table 5.1). Importantly, the matrix ranks the risks to individual crops and livestock relative to risks to other crops and livestock; that is, the risks are not ranked according Risk Management Categories of Risk Management Measures Figure 5.1. S  trategic Risk Risk management measures can be classified into three types: Instruments According »» Risk mitigation (ex ante): Actions designed to Risk Layers to reduce the likelihood of risk or to reduce the Layer 3 severity of losses (for example, soil and water con- Very low frequency, very high losses servation measures, changes in cropping patterns, Layer 2 Risk mitigation adoption of improved practices that improve per- + Risk transfer Low frequency, formance and reduce risks such as conservation Probability medium losses + Risk coping Layer 1 Risk mitigation farming, using short duration and tolerant varie­ + Risk transfer High frequency, ties, irrigation and flood control infrastructure). low losses Risk mitigation »» Risk transfer (ex ante): Actions that transfer the risk to a willing third party. These mechanisms will usually trigger compensation in the case of a Severity risk-generated loss (for example, purchasing insur- Source: World Bank ARMT. ance, reinsurance, financial hedging tools). Table 5.1. National Risk Prioritization Matrix Impact/Probability of Event Low Moderate High Highly probable •  Potato taste (coffee) •  Price volatility (export crops) •  Pests and diseases (all crops) [1 year in 3] •  Landslide (all crops) •  Livestock disease outbreaks •  Drought and erratic rains (all •  Floods—local and large scale crops and livestock) (all crops) •  Milk contamination (dairy) •  Milk collection center power cuts (dairy) •  Counterparty risk (coffee) •  Price fluctuations (food crops and milk) •  Exchange rate fluctuations (export crops) Probable [1 year in 5] •  Hail (all crops) Occasional [1 year in 10] •  Glut (dairy) •  Frost (tea) •  Losses in transit (tea) •  Aflatoxins in feed (livestock) •  Maize shortage (dairy) 46 Rwanda »» Risk coping (ex post): Actions that help the in Rwanda (table 5.2). Although agricultural risk manage- affected population and the government cope with ment measures are discussed individually and/or sequen- the loss. They usually take the form of compensa- tially, many of these would in fact be implemented jointly tion (cash or in-kind), social protection programs, and have positive, complementary impacts while address- and livelihood recovery programs (for example, ing multiple risks and would contribute to improved risk government assistance to farmers, debt restricting, management in the short, medium, and long terms. contingent financing). How instruments are applied for a given risk will likely Potential Interventions depend on the probability of the risk and the severity of for Agricultural Risk its impacts (figure 5.1). Strategic choices of risk manage- ment instrument will likely include a combination of the Management in Rwanda three types of risk management instruments. Existing interventions in risk management and national policies are already in place to begin addressing some of The report highlights some of the indicative interven- these key issues. The potential interventions identified by tions that could be undertaken to manage selected risks the report cover all three groups and, in conjunction with Table 5.2. Potential Interventions for Risk Management in Rwandan Agriculture Risk Mitigation Transfer Coping Pests and •  Integrated pest management •  Insurance (livestock) •  Rapid disease response system diseases •  Pest and disease-tolerant varieties •  Vaccination •  Good agricultural practices (GAPs)/ extension services •  Information systems/Increased border surveillance (livestock) •  Vaccination (livestock) Drought/ •  Soil and water conservation •  Insurance •  Social safety net programs and erratic rain •  Training in improved agronomic practices emergency relief •  Drought-tolerant varieties •  Grain aggregation •  Irrigation •  Storage network •  Savings groups Floods •  Soil and water conservation •  Insurance •  Social safety net programs and •  Drainage emergency relief •  Flood-tolerant varieties •  Grain aggregation •  GAPs/extension services •  Storage network •  Infrastructure •  Savings groups Domestic price •  Improved market information systems •  Hedging •  Social safety net programs and volatility •  Training on milk handling and hygiene emergency relief •  Grain aggregation •  Storage network (crops and cold chain storage and transportation for milk) •  Savings groups International •  Improved market information systems •  Futures contracts •  Social safety net programs and price •  Regional trading system •  Hedging emergency relief volatility •  Shorten farm-to-export time •  Options to buy/ •  Grain aggregation •  Training on milk handling and hygiene sell on international •  Storage network exchanges •  Savings groups Agricultural Sector Risk Assessment 47 the risk prioritization matrix, will form the basis for the and build resilience to specific pests and diseases. A seed solutions areas to be focused on during the next phase of Nkuliyimana 2010) funded by the EU and baseline study (­ the agriculture sector assessment. the Common Market for Eastern and Southern Africa (COMESA) was carried out under a regional agreement Soil and water conservation measures can yield signed between Burundi, Ethiopia, Malawi, Rwanda, significant productivity gains and help mitigate the effects Swaziland, Uganda, Zambia, and Zimbabwe. One of this of climate change. These measures (including sand dams, agreement’s objectives was to improve smallholder farm- afforestation/reforestation, conservation agricultural ers’ access to high-quality seed. This built on the 2007 practices, and terracing) are all effective and efficient National Seed Policy, which oversees four types of seeds: mechanisms for mitigating the risks of drought, floods, foundation, basic, certified, and quality declared. The and/or landslides. They are generally undertaken on government promotes the use of seeds and other agricul- individual farmland or at the community level, whereas tural inputs through the National Agricultural Extension those involving a broader watershed or landscape System (NAES), better information on seeds, and better approach require coordinated measures across a number geographic seed distribution. The National Seed Council of ­communities. in Rwanda oversees everything related to monitoring and implementation of the policy, which supports the formal One of Rwanda’s well-known constraints with regard to seed production sector spearheaded by the private sector. soil and water conservation is topography. Several projects, both government- and donor-funded, focus on or have Risk transfer solutions such as agricultural insurance elements that focus on soil and/or water conservation in and commodity price hedging (using forward contracts some way. The World Bank–GoR fund a large ­ project— and futures) could be useful risk management instruments. the Land Husbandry, Water Harvesting and Hillside Irri- Successful functioning of farmer-level agricultural insur- gation Project (LWH)—at US$43 million. LWH’s goal ance requires a number of preconditions such as: afford- is to increase the productivity and commercialization of ability (ability and willingness to pay premiums); relatively hillside agriculture in target areas, achieved by strength- low frequency of events; robust crop and weather data ening human and organizational capacity for hillside infrastructure; and farmers’ access to financial products intensification and transformation and development of and services. At present, two major companies provide the required physical infrastructure. Specific interventions agricultural insurance (Kilimo Salama and MicroInsure). exist for livestock in the form of the Livestock Infrastruc- These companies have been in operation for about four ture Support Project (LISP), which is setting up livestock years. Overall, the experience has been positive. However, watering facilities for farmers. Currently, it is working to sustain these insurance programs in the long term, ways only in Nyagatare district and is focused on dairy farm- must be found to reduce insurance premiums, increase the ers, but it is anticipated that in the future the program size of the insurance portfolio, and aggressively promote will be spread to other districts, especially in the Eastern the likely benefits of insurance programs to transfer agri- Province, which experiences more rainfall variability, dry cultural risks. Weather data infrastructure in the public spells, and droughts than the rest of the country. sector must be further strengthened and incentives pro- vided to private companies to invest in such infrastructure. Widespread availability of tolerant and short-­ maturing seed varieties will help in ensuring crop In terms of livestock, Kilimo Salama (Syngenta Foun- production during drought and flood occurrences in dation for Sustainable Agriculture) plans to develop a addition to reducing losses from pest and disease out- livestock insurance product. The product is expected to breaks. Short-maturing varieties will aid in avoiding the cover accidental death. Kilimo Salama hopes to establish effects of drought at either end of the growing season a network of veterinarians that would carry out post mor- when the frequency of rains tends to be more unpredict- tems in the case of death to determine the cause. In addi- able. Tolerant varieties, on the other hand, are better tion, clients would be trained on disease management, able to survive periods of moisture stress or excess water animal nutrition, and hygiene, and monitored to ensure 48 Rwanda that they follow the recommended livestock management the program is in its early stages, RAB has received posi- techniques. Such a product once available should help tive feedback from farmers; therefore, it is proposed that transfer the risk of animal mortality from diseases. How- a plan be developed to gradually expand the program, ever, decreased production caused by diseases would not initially focusing on border districts and then the rest of be covered, so the emphasis for such diseases should be the country. In addition, traceability systems, such as a on vaccination, epidemiological surveillance, and rapid national identification system, should be developed to response systems. augment existing disease surveillance systems. Pest and disease information systems/increased Irrigation has the potential to aid in controlling erratic border surveillance for livestock farmers, provid- rain, but the performance of this type of intervention is ing ready access to timely, accurate, and localized infor- mixed. The Rwanda Irrigation Policy and Action Plan mation about impending events that could affect livestock, was released in August 2013 after the Irrigation Mas- are a prerequisite to allow stakeholders to reduce expo- ter Plan in 2010, the overall objective of which was to sure and loss. Given the land issue constraints in Rwanda, provide Rwanda with a planning tool to use its soil and pest management plays an important role. Poor manage- water resources. Development of the tool took into con- ment in the past has contributed to some of the low qual- sideration identification of the most favorable areas to ity of soils now experienced. This means that many of establish irrigation water infrastructure; prioritization of the larger projects had a specific IPM component. Under distribution of irrigation water; identification of means of the second phase of the Lake Victoria Environmental transporting water to selected sites; and establishment of Management Project, REMA has a national IPM frame- irrigated agriculture in small-, medium-, and large-scale work. With the introduction of high-yielding varieties and projects on hillsides, marshlands, and other suitable areas. increased use of fertilizer and pesticides for crop intensifi- The current area under irrigation is just over 25,590 ha, cation, development and adoption of a participatory IPM according to government sources.32 system for all major food and cash crops is required. This is a regional approach to reduce reliance on pesticides in Flood control and drainage infrastructure invest- Kenya, Uganda, Tanzania, Burundi, and Rwanda. The ments such as dams, dykes, draining systems, and other LWH project has a requirement for a pest management flood control infrastructure can effectively mitigate the plan to help farmers reduce crop losses, and it encourages impact of floods. Issues of drainage and flood control are appropriate and timely pest management actions. also included in the Rwanda Irrigation Policy and Action Plan 2013 (described in more detail above because of the With regard to crops, Rwanda’s use of pesticides is cur- focus on the policy and action plan on irrigation). rently very low, mainly used for coffee, potatoes, and Improved access to extension services would allow tomatoes. However, with the increased focus on and pro- producers to be better informed and to access advice, motion of horticultural crops, IPM may become increas- technology, and inputs to alter their agronomic practices ingly important. In terms of livestock, the porous border in view of the prevailing and emerging risk profile of the between the livestock community in Rwanda and the com- sector. After 1994 and a nationally disjointed approach munities in Tanzania, Uganda, and the DRC has been to extension services generally driven by nongovernmen- a source of a number of disease outbreaks in Rwanda. tal organization (NGOs) working in isolation, Rwanda Currently eight border posts are manned by veterinarians, restructured MINAGRI and created RAB and NAEB. but several other informal crossing points exist through Most recently, the decision was made to decentralize agri- which livestock cross in and out of Rwanda. In 2013, the cultural extension activities to be overseen by the Ministry government initiated a Community Animal Health Work- ers (CAHW) program whereby 1,000 volunteers in the Eastern Province were trained by RAB to monitor live- 32 “Rwanda Irrigation Policy and Action Plan,” August 2013, p. 6, and “Stra- stock movements as well as to act as a first point of con- tegic Plan for the Transformation of Agriculture in Rwanda, Phase III,” July tact for farmers regarding disease management. Although 2013, section SP 1.2. Agricultural Sector Risk Assessment 49 of Local Government (MINALOC) to increase efficiency MINAGRI field staff to collect the necessary information and address the specific issues of farm households within and upload it into the databases. each district. In 2011, International Food Policy Research Institute (IFPRI), FAO, and the Inter-American Institute Vaccination is used in disease management both as an for Cooperation on Agriculture undertook a Worldwide ex ante and ex post solution to disease outbreaks. RAB has Extension study that indicated that Rwanda had 1,244 been vaccinating consistently since 2002.34 In spite of these extension staff.33 Existing weak linkages and connections vaccination campaigns, several outbreaks occur every year. between research, extension, and farmers need to be On average, vaccination coverage was 30 percent for FMD, addressed, however. Extension staff at many of the exist- 14 percent for LSD, 23 percent for anthrax, and 26 percent ing institutions in charge of agricultural development also for CBPP. Part of the solution may lie in increasing the cov- lack proper training. erage, if possible to 100 percent, particularly for anthrax and LSD, which have a higher incidence, as the surveillance Social safety net programs are limited mecha- mechanisms are further strengthened. Increasing vaccina- nisms but can help affected populations cope with high-­ tion coverage should be complemented by a Performance frequency and high-impact covariate shocks. Emergency of Veterinary Service evaluation and gap analysis with the food aid and disaster relief from donors and government help of OIE. This study should analyze the rapid disease can partially help but are not sufficient to help with recov- response system and give recommendations. ery from income and asset loss. The best-known pro- gram in Rwanda for social safety nets is the Vision 2020 As the meat industry grows, meat handling and Umurenge program. The poorest people in each district hygiene will require attention because the risk of con- (umurenge) are identified and then offered labor-intensive tamination will increase. Therefore, it is recommended work, credit for small businesses, and cash transfers and that the government: (1) support where possible invest- assets to those who cannot work. According to the Minis- ments in transportation, modern abattoirs in every major try of Finance, poverty fell from 57 percent to 45 percent town, inspection both antemortem as well as postmortem, in 5 years, lifting more than 1 million people out of pov- and food safety laboratories; and (2) increase the capac- erty and decreasing extreme poverty from 36 percent to ity of Rwanda Bureau of Standards (RBS) to monitor 24 percent (NISR 2012b). This program was not the only and certify meat products and processing facilities. These one to reduce poverty, but it contributed to the poverty investments should be complemented by training in meat reduction process. handling and hygiene for traders, transporters, and staff in abattoir and processing facilities as well as the inspectors. An improved market information system can pro- vide accurate, timely, and transparent information about A summary of existing interventions for the three group- production and stocks, trade flows, and prices in different ings by the GoR can be found in appendix D. markets to aid in the management of price volatility in domestic, regional, and international markets. MINAGRI, Decision Filters for Solution through a World Bank–funded “ICT for Development” Prioritization project implemented by the Rwanda Information Tech- Using decision filters to evaluate and prioritize the list nology Authority (RITA) has developed e-SOKO, which of solutions can aid in making rational resource alloca- seeks to empower farmers and make them more informed tion decisions in place of a detailed cost-benefit analy- about market pricing, thus making them more successful. sis. Rwanda, like most countries, is resource constrained Some challenges with the system have arisen, not only and decision makers are compelled to find the quickest, because it requires farmers to be trained on the use and cheapest, and most effective measures among the options maintenance of the system, but also because it relies on presented. A detailed cost-benefit analysis can help in selecting the most appropriate intervention. However, this 33 See http://www.worldwide-extension.org/africa/rwanda/s-rwanda 34 Vaccination data were not available for 2004 and 2005. 50 Rwanda exercise was too costly and time consuming to conduct »» Increased use of agricultural inputs (fertilizer, qual- under this risk assessment.35 Furthermore, many elements ity seeds, and extension services); involved in making these decisions are not easily quantifi- »» Animal resources mobilization; and able and not easy to factor in. »» Development of agricultural postharvest handling storage systems and farmer capacity. A number of complex analytic screening tools can be used to assess the various decision filters; this report does Finally, nine interventions were drawn out from the con- not claim methodological rigor in its assessment of fil- sultation exercise and analysis carried out by the World ters. Filters were applied to provide a rapid assessment to Bank team in the field. Overall, stakeholders advocated obtain some form of prioritization of risk solutions based more training in improved agricultural practices, IPM, on feedback from key stakeholders. The following criteria and handling and hygiene practices for livestock prod- were presented by the World Bank team for this purpose: ucts, as well as better access to livestock vaccination and »» Applicability: Whether the proposed interven- improved planting material. Investments rated highly in tion fits into the current policy/programming or the consultation were in market information systems and business objectives of a government department or cold chain storage and transportation. Further informa- private sector firm. tion on the possible interventions proposed in the consul- »» Feasibility: Whether it is or would be easy for a tation exercise and how they were ranked can be found in government/firm to implement this intervention appendix A. in the short/medium term. »» Affordability: Whether or not an intervention is The priority interventions align with existing government affordable, with regard to current and future budgets. policy and interventions already in place, although at a »» Scalability: Whether or not a pilot/small-scale much smaller scale, with effects at a much more localized project can be rolled out to a wider group of level. Greater emphasis should be placed on whether or ­ beneficiaries. not these could be scaled up to the national level by both »» Sustainability: Whether an intervention will be the government and nongovernmental stakeholders in sustainable in the long run or once government the sector, many of whom are already involved, to more funding runs out or resources are directed elsewhere. meaningfully affect Rwanda’s agriculture sector. Some major projects and interventions are already tak- Proposed Solutions for Further ing place in the sector and it is important that proposed Assessment interventions are in line with government priorities. The Given the prioritized risks, the feedback from stakehold- World Bank team enumerated the existing priorities and ers, and the ongoing interventions, seven possible solutions interventions in the sector. This ensures increased buy-in areas emerge as the most relevant for further assessment. from the government and attempts to aid in efficient use of limited resources. MINAGRI’s latest annual report Water management in the crop sector, particularly to 1.  (Annual Report FY 2011–2012) had several overall strategy improve practices in preparation for dry periods and scat- priorities that are consistent with some of the proposed tered rainfall, but also to better manage rainfall in valleys interventions: to minimize flooding. Solutions areas may include: »» Development of quality irrigation and mechaniza- »» Expansion of on-farm water harvesting systems tion systems (using public and private resources); »» Viable mechanisms for financing small-scale »» Comprehensive approach to land husbandry (soil ­irrigation fertility, soil conservation, water harvesting and »» Expansion and rehabilitation of drainage infra- management, livestock feed); structures in valleys »» Agricultural practices to improve soil moisture and reduce flooding, including minimum tillage 35 Any feasibility study for investments would include a cost-benefit analysis. Such analysis could potentially also be part of a solutions assessment. ­agriculture Agricultural Sector Risk Assessment 51 Weather-risk management in the livestock sector, par- 2.  »» Farm-level livestock management ticularly as it relates to water and feed access. Solutions »» Increased capacity of food safety institutions may include: »» Improved hygiene practices throughout the supply »» Improved rural water infrastructure chain »» Development of existing feed supply chains to tem- »» Mitigation of aflatoxin contamination in the feed porarily substitute for the lack of pastures in prov- supply chain inces where grazing is allowed »» Training of farmers in livestock management in Assessment of possible price management mechanisms 6.  water scarce situations, and in good hygiene prac- for actors in the export crop supply chain. Given the tices with special focus on practices in dry periods exposure to international prices for actors in the coffee and tea supply chains, scope exists to strengthen price Pest and disease management, particularly as it relates 3.  management mechanisms in the sector. By analyzing to potential future risks as a result of land consolida- the physical and financial flows on current transaction tion and increased monocropping. Similarly, potential arrangements for exports, a set of options on how to changes in pest and disease risks as a result of climate reduce exposure to risk can be identified. Potential solu- change should be integrated in such assessment. Solu- tions areas may include: tions may include: »» Strengthening existing price information sys- »» Improving agricultural practices and pest manage- tems that allow for transparent price setting ment, including further developing integrated pest throughout the supply chain, and training actors management throughout the chain to optimize given available »» Strengthening the crop research system on pest information and disease management and resilient crops »» Providing price risk management training to actors »» Strengthening access to inputs, including develop- in the supply chain, for example in forwarding ing network of input dealers Price To Be Fixed contracting »» Developing information system on pests and diseases »» Assessing available policy mechanisms for support- ing actors in the sector against price risks Developing livestock disease management infrastruc- 4.  »» Assessing possible production and marketing ture to mitigate and manage disease outbreaks with the investments for producers and processors that can purpose of decreasing the economic impact on the sec- lessen relevant actors’ exposure to risk tor. Solutions may include: »» Developing livestock information systems, includ- Analysis of milk price volatility to better understand 7.  ing animal registers and disease warning systems the reasons behind the fluctuations in milk prices. This »» Developing veterinary services and vaccination would include proposing appropriate price risk man- programs agement mechanisms depending on the identified »» Strengthening animal reference laboratory capacity causes behind existing price volatilities. »» Strengthening regional cooperation in livestock disease management It is important to recognize that the GoR is already doing a lot in all of these areas. However, given the risks iden- 5. Sanitary institutions and practices in the livestock sector, tified in this analysis and especially given the strategic throughout the supply chain and involving both public path Rwanda has outlined for the sector, there is room and private actors. As incomes increase, this sector is for strengthening these risk management areas. The pro- likely to grow, so the necessary institutional infrastruc- posed solutions assessment could thus support Rwanda in ture must be in place to mitigate risks and minimize preparing the sector for effective risk management in the losses. Solutions may include: next decades. »» Strengthened animal disease management in rel- evant institutions 52 Rwanda References Agricultural Marketing Information System. 2012. http://amis.minagri.gov.rw/. Akinsola, E., and M. Agyen-Sampong. 1984. “The Ecology, Bionomics and Control of Rice Stem Borer in West Africa.” Insect Science and Its Application 5: 69–77. Akram, H. M., A. A. Sattar, H. S. U. Rehman, and A. Bibi. 2013. “Impact of Water Deficit Stress on Various Physiological and Agronomic Traits of three Basmati Rice Cultivars.” Journal of Animal and Plant Science 23 (5): 1415–1423. Asiimwe, P., Gashaka, G. Gashaka, D. Nkezabahizi, J. P. Legg, G. Okao-Okuja, R. Obonyo, and M. Mutumwinka. 2011. “Occurrence and distribution of cassava pests and diseases in Rwanda.” Agriculture, Ecosystems & Environment. doi:10.1016/ j.agee.2011.01.014. Bigirimana, S., P. Barumbanze, P. Ndayihanzamaso, R. Shirima, and J. P. Legg. 2011. “First Report of Cassava Brown Streak Disease and Associated Ugandan Cassava Brown Streak Virus in Burundi.” New Disease Reports 24: 26. Brouwer, C., L. Pins, and M. Heibloem. 1989. “Irrigation and Water Management: Irrigation Scheduling.” Training Manual No. 4. FAO, Rome. ftp://ftp.fao.org /agl/aglw/fwm/Manual4.pdf. Central Intelligence Agency. 2010. The World Factbook. Washington, DC: Central Intel- ligence Agency. Conservation Farming Unit. 2007. Conservation Farming and Conservation Agriculture, A Handbook for HOE Farmers In Agro-ecological Regions I and II A—Flat Culture. Zambia Conservation Farming Unit. Dunkel, F. V., S. Sriharan, E. Niziyimana, and A. Serugendo. 1990. “Evaluation of Neem Kernel Extract (Margosan-O) against Major Stored Insect Pests of Beans and Sorghum in Rwanda.” Proceedings of 5th International Working Conference of Stored-Product Protection. 527–535. Bordeaux, France. Durr, G. 1983. “Potato Production and Utilisation in Rwanda.” International Potato Center (CIP) Lima, Peru. EARS (Earth Environment Monitoring). 2008. Coffee Yield and Production Estimate in Rwanda. October. Kigali. http://www.ears.nl. FAO (Food and Agriculture Organization [of the UN]). 1989. Irrigation Water Man- agement Training Manual. Available at: http://www.fao.org/docrep/T7202E /T7202E00.htm ———. 2010. “Cassava Diseases in Africa: A Major Threat to Food Security.” Strate- gic Programme Framework. Rome. March. FAOSTAT (FAO Corporate Statistical Database). 2013. FAOSTAT Database. FAO, Rome. ———. 2014. FAOSTAT Database. FAO, Rome. FEWS NET (Famine Early Warning System). 2011. “Livelihood Zoning ‘Plus’ Activity in Rwanda: A Special Report by the Famine Early Warning System Agricultural Sector Risk Assessment 53 (FEWS NET).” http://www.fews.net/sites/default/files/documents/reports /RW_livelihood%20descriptions%202011.pdf. ———. 2013a. Seasonal Crop Calendar in Rwanda. FEWS NET, Kigali, December. http:// www.fews.net. ———. 2013b. Price Watch. Kigali. http:// www.fews.net. IFPRI (International Food Policy Research Institute). 2011. Agricultural Extension and Advisory Services Worldwide (facilitated by IFPRI), Extension and Advi- sory Services in Rwanda. http://www.worldwide-extension.org/africa/rwanda /s-rwanda. IPCC (Intergovernmental Panel on Climate Change). 2014. Climate Change 2014: Impacts, Adaptation, and Vulnerability. Working Group II Contribution to the IPCC 5th Assessment Report—Changes to the Underlying Scientific/Technical Assessment. Chapter 22: Africa. Jones, A. L. 1999. “Phaseolus Bean: Post-harvest Operations.” FAO IN PhO Post Har- vest Compendium, Rome. http://www.fao.org/fileadmin/user_upload/inpho /docs/Post_Harvest_Compendium_-_Phaesolus_beans.pdf. Jones, Peter G., and Philip K. Thornton. 2003. “The Potential Impacts of Climate Change on Maize Production in Africa and Latin America in 2055.” Global Envi- ronmental Change 13 (1): 51–59. http://www.sciencedirect.com/science/article /pii/S0959378002000900. Lebot, V. 2009. Tropical Root and Tuber Crops: Cassava, Sweet Potato, Yams and Aroids. No. 17 CABI. http://books.google.com/books/about/Tropical_Root_and_Tuber_ Crops.html?id=rFwyrKRSMUMC. McSweeney R., J. N. Semafara, M. Cole, and R. Washington. 2011. “Rwanda’s Cli- mate: Observations and Projections.” July. Smith School of Enterprise and the Environment. MINAGRI (Ministry of Agriculture and Animal Resources). 2008. “Pest Manage- ment Plan for Rural Sector Support Project.” ———. 2010. “Agriculture Gender Strategy.” Ministry of Agriculture and Animal Resources, November, Kigali, Rwanda. ———. 2011a. About Crop Intensification Program (CIP). http://www.minagri.gov.rw /fileadmin/user_upload/documents/CIP/MORE_INFORMATION_ ABOUT_CROP_INTENSIFICATION_PROGRAM.pdf. ———. 2011b. National Post-Harvest Staple Crops Strategy. ———. 2012a. Agriculture Gender Strategy. Republic of Rwanda, Kigali. November. ———. 2012b. “New Tea Green Leaf Pricing Model for Tea Farmers in Rwanda.” Cabinet Briefing Paper. Republic of Rwanda, Kigali. ———. Annual Report FY 2010–2011. http://www.minagri.gov.rw/fileadmin/user_ upload/documents/Reports/Final_AnnualReport_FY2010_2011updated.pdf. ———. 2013a. Strategic Plan for the Transformation of Agriculture in Rwanda (PSTA) Phase III. http://www.minecofin.gov.rw/fileadmin/templates/documents/sector_ strategic_plan/PSTA_III_Draft.pdf. ———. 2013b. Rwanda Irrigation Policy and Action Plan. August. 6. ———. 2014. Briefing for Hon. Minister as Discussant on Transforming Agriculture Using Public Expenditure Reviews as a Growth Catalyst. Republic of Rwanda, Kigali. 54 Rwanda MINECOFIN (Ministry of Finance and Economic Planning). 2000. Rwanda Vision 2020. Republic of Rwanda, Kigali. ———. 2013a. Macro Framework—Public Data Set. Republic of Rwanda, Kigali. June. ———. 2013b. Economic Development and Poverty Reduction Strategy, 2013–2018. Republic of Rwanda, Kigali. Monares, A. 1984. “Building An Effective Potato Country Program—the Case of Rwanda.” International Potato Center (CIP) Lima, Peru. Mukeshimana, G., and J. Kelly. 2001. “Evaluation of Rwandan Varieties for Disease Resistance.” U.S. Department of Agriculture, Washington, DC. NAEB (National Agricultural Export Development Board). 2014. Statistical Information Related to Coffee and Tea. Republic of Rwanda, Kigali. January. National Bank of Rwanda. 2005. Annual Report. Night, G., P. Asiimwe, G. Gashaka, D. Nkezabahizi, J. P. Legg, G. Okao-Okuja, R. Obonyo, C. Nyirahorana, C. Mukakanyana, F. Mukase, I. Munyabarenzi, and M. Mutumwinka. 2011. “Occurrence and Distribution of Cassava Pests and Diseases in Rwanda.” Agriculture, Ecosystems & Environment. doi:10.1016/j. agee.2011.01.014 NISR (National Institute of Statistics of Rwanda). 2010. National Agricultural Survey 2008 (NAS 2008)—Report of National Data Analysis. Republic of Rwanda, Kigali. February. ———. 2012a. EICV3 Thematic Report—Agriculture. Republic of Rwanda, Kigali. ———. 2012b. Third Integrated Household Living Conditions Survey (EICV III), Republic of Rwanda, Kigali. http://www.statistics.gov.rw. ———. 2013. Statistical Year Book 2012. Republic of Rwanda, Kigali. http://www .statistics.gov.rw. Ngabitsinze, Jean Chrysostome, A. Mukashema, M. Ikirezi, and F. N ­ iyitanga. 2011. “Planning and Costing Adaptation of Perennial Crop Systems to Climate Change: Coffee and Banana in Rwanda: Case Study Report.” October. http://pubs.iied. org/G03174.html. Nkuliyimana, G. 2010. “Rwanda Seed Sector Baseline Study.” September. http://afsta .org/wp-content/uploads/documents/RWANDA%20SEED%20SECTOR%20 BASELINE%20STUDY.pdf. NOAA (National Oceanic and Atmospheric Administration). “La Nina.” http:// oceanservice.noaa.gov/facts/ninonina.html. Norgaard, R. B. 1988. “The Biological Control of Cassava Mealybug in Africa.” Amer- ican Journal of Agricultural Economics 70: 366–371. Nzungize, J., P. Gepts, R. Buruchara, S. Buah, P. Ragama, J. P. Busogoro, and J. P. Baudoin. 2011. “Pathogenic and Molecular Characterisation of Pythium Species Inducing Root Rot Symptoms of Common Bean in Rwanda.” African Journal of Microbiology Research 5 (10): 1169–1181. OIE (World Organisation for Animal Health Information). 2014. Database. http:// www.oie.int/wahis_2/public/wahid.php/Wahidhome/Home Olsson, P. 2012. “Ruminants’ Production within Agroforestry Systems in Rural Rwanda—Production, Benefits and Problems.” A thesis submitted in partial ful- fillment of the requirements of Swedish University of Agricultural Sciences for Agricultural Sector Risk Assessment 55 the degree of Bachelor in Animal Science. Uppsala: Swedish University of Agri- cultural Sciences. http://stud.epsilon.slu.se/5104/1/olsson_p_121205.pdf. Pfunda Tea Cooperative. 2014. Visit by the World Bank Mission. January. ProMusa. 2014. “Bunchy Top.” http://www.promusa.org/tiki-index.php?page= Bunchy+top. REMA (Rwanda Environment Management Authority). 2009. Rwanda State of Envi- ronment and Outlook: Our Environment for Economic Development. Republic of Rwanda, Kigali. http://www.unep.org/geo/pdfs/rwanda_outlook.pdf. ———. 2011. The Integrated Pest Management (IPM) Framework for Rwanda in the Lake ­Victoria Basin. Republic of Rwanda, Kigali. May. Rwanda Meteorology Agency. “Observation Stations.” http://www.meteorwanda. gov.rw/index.php?id=10. Schwartz, H. F., and G. E. Galvez. 1980. “Beans Production Problems: Disease, Insect, Soil and Climatic Constraints of Phaseolus vulgaris.” CIAT Cali, Colombia. Smith School of Enterprise and the Environment. 2011a. “Baseline Report: National Strategy on Climate Change and Low Carbon Development for Rwanda.” Febru- ary. University of Oxford. ———. 2011b. “Appendix E: Rwanda’s Climate: Observations and Projections.” In Baseline Report: National Strategy on Climate Change and Low Carbon Development for Rwanda. July. University of Oxford. http://www.smithschool.ox.ac.uk/library /reports/Rwanda_Climate-data-final_proofed.pdf. Sorwathe Tea Factory. 2014. Visit by the World Bank Mission. January. Stockholm, Environment Institute. 2009. “Economics of Climate Change in Rwanda.” http://sei-international.org/mediamanager/documents/Publications/Climate /cc_economics_rwanda.pdf. TechnoServe Rwanda. 2008. “The Dairy Value Chain in Rwanda.” October. https://cgspace.cgiar.org/bitstream/handle/10568/2410/Dairy%20Value%20 Chain%20Rwanda%20Report.pdf ?sequence=1. Tenge, Ngoga G., M. Alphonse, and T. S. Thomas. 2012. “Rwanda.” In East African Agriculture and Climate Change: A Comprehensive Analysis, edited by Michael Waithaka, Gerald C. Nelson, Timothy S. Thomas, and Miriam Kyotalimye, 247–277, Chap- ter 9. Washington, DC: International Food Policy Research Institute (IFPRI). http://ebrary.ifpri.org/cdm/ref/collection/p15738coll2/id/127932. Thornton, Philip K., P. G. Jones, G. Alagarswamy, and J. Andresen. 2009. “Spatial Variation of Crop Yield Response to Climate Change in East Africa.” Global Environmental Change 19 (1): 54–65. http://www.sciencedirect.com/science /article/pii/S0959378008000812. Thornton, P. K., P. G. Jones, P. J. Ericksen, and A. J. Challinor. 2011. “Agriculture and Food Systems in Sub-Saharan Africa in a 4°C+ World.” Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 369 (1934): 117–136. Trutmann, P., and W. Graf. 1993. “The Impact of Pathogens and Arthropod Pests on Common Bean Production in Rwanda.” International Journal of Pest Management 39 (3): 328–333. 56 Rwanda USAID-EAT (U.S. Agency for International Development Enabling Agricultural Trade Project). 2012. “Rwanda Cross Border Trade Analysis.” Accessed at: http://eatproject.org/docs/USAID-EAT%20Cross-Border%20Trade%20 Analysis%20Rwanda.pdf. van de Steeg, J. A., M. Herrero, J. Kinyangi, P. K. Thornton, K. P. C. Rao, R. Stern, and P. Cooper. 2009. “The Influence of Current and Future Climate-Induced Risk on the Agricultural Sector in East and Central Africa: Sensitizing the ASARECA Strategic Plan to Climate Change.” Research Paper No. 22, International Live- stock Institute, 27. Vorster, Danton I. 2012. “Rwanda Green Leaf: Assessment of the Tea Green Leaf Pricing.” Prepared for NAEB. May. WDI (World Development Indicators). 2001. http://www.ssc.wisc.edu/~walker/wp /wp-content/uploads/2012/10/wdr2001.pdf ———. 2013. Database. Accessed 2013–2014. WFP (World Food Programme). 2012. Rwanda Comprehensive Food Security and Vulnerability Analysis and Nutrition Survey 2012. Rome. http://documents.wfp.org/stellent/groups /public/documents/ena/wfp255144.pdf World Bank. 2009. Doing Business 2010: Reforming through Difficult Times. Washington, DC: World Bank. ———. 2012. “Social Safety Nets on the Rise in Africa.” April. World Bank, Wash- ington, DC. http://web.worldbank.org/WBSITE/EXTERNAL/COUNTRIES /AFRICAEXT/0,,contentMDK:23179724~menuPK:258659~pagePK:286510 6~piPK:2865128~theSitePK:258644,00.htm. ———. 2013. Doing Business 2014: Understanding Regulations for Small and Medium-Size Enterprises. Washington, DC: World Bank. Agricultural Sector Risk Assessment 57 Appendix A Stakeholder Feedback Decision Filters Rwanda, like most countries, is resource constrained and so decision makers are com- pelled to find the quickest, cheapest, and most effective measures among the options presented. A detailed cost-benefit analysis can help in selecting the most appropriate intervention; however, this exercise itself would be costly and time consuming. Fur- thermore, many elements in making these decisions are not easily quantifiable and not easy to factor in. Using decision filters to evaluate and prioritize the list could aid in making rational resource allocation decisions in place of a detailed cost-benefit analysis. The following criteria were presented by the World Bank team for this purpose. A number of complex analytic screening tools are available to assess the various differ- ent filters and this report does not claim methodological rigor in assessing the filters. These filters were applied to provide a rapid assessment to obtain some form of priori- tization based on feedback from key stakeholders. Applicability: Whether a government department or a private sector firm, there are current policy/programming or business objectives. In an aim to make the interventions enter seamlessly into existing practices/plans, it is important that the applicability of specific interventions is taken into account. Feasibility: Whether it would be/is easy for a government/firm to imple- ment this intervention in the short/medium term is an important factor to take into consideration; often “quick-wins” will come across as more attractive and ­implementable. Affordability: One of the most important considerations for anyone getting involved in implementing an intervention is whether or not that intervention is affordable with regard to current and future budgets. Scalability: Whereas an intervention may have positive results in a small pilot, it is important for actors to take into consideration whether or not they will be able to roll out the intervention to a wider group of affected people. Agricultural Sector Risk Assessment 59 Sustainability: Whereas some interventions, such For Rwanda, three matrixes were drawn up, one each for as coping mechanisms, quickly provide resources food crops, export crops, and livestock (figures A.1–A.3). for affected people, for larger and more expensive The results after the application of these decision filters are interventions it is important to think about how sus- shown next, but it should be remembered that they are indic- tainable the intervention will be in the long run or ative and imperfect. However, they do provide a first step once government funding runs out or resources are toward development of a more comprehensive and strategic directed elsewhere. approach for managing risks in Rwanda’s agriculture sector. Figure A.1. Prioritization of Possible Interventions: Food Crops Emergency relief (appeal) Emergency relief (financial reserve) Predictable market interventions (e.g., SGR) Emergency relief (strategic reserve) Proposed solutions Options to buy/sell on international exchanges Insurance Enhanced grain aggregation and storage network Drainage/irrigation Enhanced access to improved planting material GAP (training and finance) Consultation with stakeholders Savings groups 0 20 40 60 80 100 120 140 Feedback total score = 150 Figure A.2. Prioritization of Possible Interventions: Export Crops Potato effect for coffee (PEC) sources (study) Hedging for P and ER Shorten farm to export time Drainage (floods in marsh lands for tea) Sanctity of contracts Social safety net for the most vulnerable Proposed solutions Less frequent regulatory changes Domestic information Insurance for weather/floods Future contracts for P and ER Regional trading system Buyer’s association Stakeholder consultation Fertilizer market information Market information system Training in improved agronomic practices (PEC) IPM for pests and diseases for coffee 0 20 40 60 80 100 120 140 160 180 120 Feedback total score = 200 60 Rwanda Figure A.3. Prioritization of Possible Interventions: Livestock Strategic grain reserves (SGR) Traceability systems Power back up solutions Stabilization of power supply Potential solutions Rapid disease response system, e.g., increased number of labs, personal Enforcement of milk, drugs and inputs standards, e.g., increase in number of RBS/RAB inspectors Insurance Increased border surveillance, e.g., CAHW program Investment in cold chain storage and transportation facilities Training on milk handling and hygiene Vaccination 0 20 40 60 80 100 120 140 Feedback total score = 150 In summary, the following areas ranked the highest in the Export crops three commodity categories: 1.  Integrated pest management (IPM) for coffee 2.  Training in improved agronomic practices 3.  Market information system Food crops 1.  Savings groups Livestock 2.  Training and finance for good agricultural prac- 1.  Vaccination tices (GAPs) 2.  Training on milk handling and hygiene 3.  Enhanced access to improved planting material 3.  Investment in cold chain storage and transportation Agricultural Sector Risk Assessment 61 Appendix B Climate Change in Rwanda Current Conditions Situated on mountainous terrain in the East African Rift Valley, Rwanda has a tropi- cal temperate climate (REMA 2009, 97). The hilly terrain produces a diverse range of agro-ecological conditions in the country, allowing for a wide variety of crops to be grown but also adding to climate forecasting challenges. Temperatures in the country vary with altitude, but average annual temperatures hover between 16°C and 20°C (REMA 2009, 97.). Rainfall in the country is also shaped by multiple factors. On average, the country receives 1,250 mm of rainfall annually (Stockholm Environment Institute 2009, 4). Primary influence of rainfall patterns falls on the Inter Tropical Convergence Zone where the weather systems of the Northern and Southern Hemispheres meet. The progression of the ITCZ results in two types of seasons: dry and rainy. El Niño and La Niña, opposite phases of the El Niño-Southern Oscillation36 (ENSO), also affect Rwanda, creating heavy rainfall and severe dry conditions, respectively, when they occur every three to five years (Smith School of Enterprise and the Envi- ronment 2011a, 4). Furthermore, the Congo air mass, a moist air appendage of the ITCZ, as well as the Mascarene, Azores, St. Helena, and Arabian high-pressure sys- tems, and subtropical anticyclones also influence rainfall (Smith School of Enterprise and the Environment 2011a, 4). Four seasons divide a calendar year in Rwanda: two rainy seasons, from September to November and from March to May (figure B.1); and two dry seasons, from December to February and June to August. 36 The El Nino–Southern Oscillation makes up the cyclical temperature fluctuations between the ocean and atmo- sphere in the East-Central Equatorial Pacific, which occur every 3–5 years. For more information, see NOAA, http:// oceanservice.noaa.gov/facts/ninonina.html. Agricultural Sector Risk Assessment 63 Figure B.1.  Average Monthly Rainfall and Temperatures in Rwanda, 1960–90 200 mm 20°C 150 mm 19.5°C Temperature Rainfall 100 mm 19°C 50 mm 18.5°C 0 mm 18°C Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Rainy Rainy season season Source: World Bank Climate Change Portal. In addition to altitude levels, regions also see varying cli- Table B.1. Ideal Growing Conditions mate patterns. Because of the topography of the country for Selected Crops in and the existence of large bodies of water, the eastern and Rwanda southeastern portions of the country experience more fre- quent and prolonged droughts, whereas the northern and Temperature Requirements western regions receive more rainfall (REMA 2009, 97). Crop in Rwanda Other Crops in Current Bananas* 15°C–30°C Deep soils, typically require 1,000–2,000 mm Conditions of rainfall Bananas and plantains have traditionally grown well in Beans* 14°C–18°C Typically grow in Rwanda’s tropical temperatures and rainfall conditions higher altitudes in (table B.1). These crops are usually cultivated by small- Rwanda because of cooler temperature scale farmers for food security and income-generation requirements purposes (Tenge, Aphonse, and Thomas 2012, 260). Beans are another critical food source that has thrived Coffee** Below 25°C. Ideal for For Arabica varieties, Arabica is 18°C 800–1,000 mm of in Rwanda’s higher altitudes because they require cooler at night and 22°C rainfall and 1,200 mm temperatures. Coffee, a key export crop for Rwanda, during the day for Robusta is required has very particular temperature and rainfall require- Sources: IFPRI 2012; Tenge, Aphonse, and Thomas 2012, 18. ments, alterations of which significantly affect quality and *Tenge, Aphonse, and Thomas 2012, 260. **Ngabitisinze, Chrysostome, thereby export price. ­ Mukashema, Ikirezi, and Niyitanga 2011, 18. Global Climate Change to exceed a 2°C increase above the mean annual tem- Impacts on Rwanda perature of the late 20th century. The report found that The International Panel on Climate Change (IPCC) con- African ecosystems are already being affected by global cluded that there is evidence in Africa of warming over climate change and future impacts are expected to be sub- land. This warming is expected to continue and by 2100 stantial (IPCC 2014, Chapter 22: Africa, 3). 64 Rwanda Current global climate change models, however, can offer only a crude picture of future climate change impacts on Observed Changes Rwanda, as gathering accurate data on current climate From the available information, changes in both rainfall conditions still proves difficult. Although the number of and temperature patterns are already apparent. An anal- rainfall gauges has improved in recent years, the country ysis conducted of these variables over the past 30 years still has few accurate gauges situated around the country.37 shows that the rainy season is becoming shorter with higher intensity, leading to both more droughts and floods Gathering historical data is even more challenging. simultaneously (REMA 2009, 97). Parts of the country Most of the monitoring network was destroyed during are experiencing these events differently. The Northern the 1990–94 civil war, limiting data gathering in subse- and Western Provinces are seeing heavier rains and floods, quent years. Because of this destruction, the only com- whereas the Eastern Province is seeing more rainfall defi- plete records from 1994–2009 are from one station at the cits (REMA 2009, 98). Kigali airport.38 Information prior to the war, when over 100 meteorological stations were operating,39 was col- Furthermore, increased mean, maximum, and minimum lected, including data from the early 20th century, but it temperatures have been observed from 1971 to 2010 in all is difficult to obtain (REMA 2009, 97) and has yet to be four seasons, with all of these temperatures rising roughly fully digitized.40 one-half of a degree per decade (Smith School of Enter- prise and the Environment 2011b, 11–13). This is greater No specific climate modeling has been completed for than the global trend of 0.19°C and 0.32°C for mean Rwanda; instead, projections from global climate change temperature per decade, and 0.29°C for maximum and models have been downscaled to information from a sin- minimum temperatures (Smith School of Enterprise and gle Rwandan weather station, making them a very crude the Environment 2011b, 11). Humidity, meanwhile, has measure (Smith School of Enterprise and the Environ- decreased over the same time period, at a rate of 1.58 per- ment 2011a, 3). cent every decade (Smith School of Enterprise and the Environment 2011b, 14). Regionally, there is little more to offer, as nearly all of Rwanda’s East and Central African neighbors suffer from Forecasted Changes similar data issues. Furthermore, global climate change Despite these observed patterns, many climate models are models are currently not well tuned to model the mag- in disagreement over rainfall changes in Rwanda over the nitude or even direction of regional rainfall changes in next 30 years. For East Africa as a whole, high rainfall Africa, because they omit or underweight conditions that extremes (events typically occurring once in every 10 years) are important for determining regional rainfall in Africa, are expected to increase in frequency (van de Steeg et al. such as dynamic land cover–atmosphere interactions and 2009, 27). Torrential rainfalls present an added problem climate variability drivers such as ENSO (van de Steeg in Rwanda as they create landslides that can wipe away et al. 2009, 23–24). the numerous farms that reside on the sides and bottoms of the hills that spot the country. 37 Today, to record weather the country has 72 rainfall stations, 72 climatological stations, 39 automatic weather stations, and 13 agro-synoptic across the coun- There is more agreement between the climate models try. See Rwanda Meteorology Agency, “Observation Stations,” http://www. regarding temperature increases. Three different climate meteorwanda.gov.rw/index.php?id=10, accessed June 2014. This is a marked change models41 forecast a 1°C–2.5°C increase in maxi- improvement. A mere three years earlier, the country had only 26 rainfall sta- mum temperatures. Studies of the effect of this on key crops tions, 13 agro-synoptic stations, and 5 automatic weather stations. See Smith School of Enterprise and the Environment 2011a, 4. are lacking, but such temperature increases are thought to 38 See Rwanda Meteorology Agency, “Observation Stations,” http://www. increasing the prev- affect crop productivity particularly by ­ meteorwanda.gov.rw/index.php?id=10. 39 See Smith School of Enterprise and the Environment 2011a, 4. 40 For instance, gathering colonial data requires liaising with the Belgian govern- These include CNRM-CM3, ECHAM, and MIROC 3.2. See Tenge, Aphonse, 41 ment. See Smith School of Enterprise and the Environment 2011a, 4. and Thomas 2012, 264. Agricultural Sector Risk Assessment 65 Figure B.2. Hot Days, Rwanda, Figure B.3. Hot Days, Rwanda, 1961–2000 Projected for 2046–65 20 Ensemble median (50%) Ensemble high (90%) Ensemble low (10%) Ensemble high (90%) Ensemble low (10%) 30 Ensemble median (50%) 15 10 20 Days 5 Days 10 0 0 –5 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec –10 Source: World Bank Climate Change Portal. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Source: World Bank Climate Change Portal. alence of pests and diseases. For crops that also require scenarios and two different climate models, the land area cooler temperatures to grow, such as beans, such tempera- that currently has 210 or more growing days is expected ture increases are likely to produce declining yields (Tenge, to shrink from today’s 92 percent to roughly 78 percent Aphonse, and Thomas 2012, 264).42 by 2030, and then further to about 60 percent by 2050 (van de Steeg et. al. 2009, 50). This indicator is by no Figures B.2 and B.3 show the number of hot days (days that means a direct measurement of agricultural productivity, fall in the 90th percentile of high temperatures) projected as climate variability may in fact reduce yields in areas in Rwanda based on an assembly of 15 climate change with longer growing periods (Thornton et al. 2011, 122). models. At its peak during the dry season, the models pre- Nonetheless, a smaller growing window may put more dict an increase of 10–15 more hot days per year. pressure on farmers to produce their crops. The follow- ing summarizes the available studies on climate change’s Crops and Climate impacts on existing crop varieties. Change One study notes that coffee is vulnerable to high tem- In-depth climate change studies and the subsequent perature increases. Temperatures above 25°C affect the impacts on crops in Rwanda have not been completed, plant’s photosynthesis process and spur the development as is the case in other SSA countries.43 Therefore, the full of diseases such as coffee leaf rust (CLR) and fruit blight. effects of climate change on crops are not yet understood. Low temperatures, below 15°C, spur coffee berry diseases For some crops, such as maize, in the near term, Rwanda (Ngabitsinze et al. 2011, 18). Coffee must also have rain- is expected to have yield increases44 as a result of rising fall above 800–1,000 mm for Arabica varieties and 1,200 temperatures;45 however, other crops will likely require mm for Robusta (Ngabitsinze et al. 2011, 18). Therefore, significant alterations in planting or new varieties to suc- should temperatures in Rwanda continue to increase as ceed. expected, coffee will be at risk of declining yield. The study found that there are already delayed flowering peri- Throughout the country, the length of the growing sea- ods for coffee as a result of longer dry seasons and delayed son is expected to decline. Under two different growth ripening periods as a result of a reduction in heavy rains in March. It also found that pests and diseases are migrating more easily, increasing the need for and length of chemi- 42 See also IPCC 2104, Chapter 22: Africa, 20. 43 For instance, there is a comprehensive report by USAID on climate change cal spraying (Ngabitsinze et al. 2011, 27). Furthermore, impacts on crops in Malawi. increasing temperatures may push perennial crops, such 44 See IPCC 2014, Chapter 22: Africa, 20. as coffee, to higher altitudes and thereby likely reduce the 45 This is for the near term. For the long term, at least one study found that potential land area on which they can be grown (IPCC maize yields are expected to decline by 19 percent in East Africa in 2090. See Thornton et al. 2011, 122. 2014, Chapter 22: Africa, 20). 66 Rwanda Climate change impacts on maize production have also likely occur in areas below 1,000 m in altitude, and in been studied. One study (Jones and Thornton 2003) many areas under 1,500 m. However, higher elevations applied a global climate change model to calculate may see yield increases as long as average temperatures do expected impacts on four generic maize varieties in Latin not go above 20°C–22°C, the temperature threshold for America and Africa in 2055. For Rwanda, the authors bean yields in the region (Thornton et al. 2009). For East found an approximately 7 percent reduction in kg/ha Africa as a whole, the authors found that the region could yield from a baseline of 1990 climate conditions (Jones expect 1–3 percent reductions in production levels (kg) and Thornton 2003). However, because of the diversity by 2050 in low-emission scenarios for maize and beans, of Rwanda’s climate, this broad application of global sce- and 11–15 percent reductions in production levels (kg) narios is only a crude measurement of potential impacts. in high-emission scenarios for the same crops (Thornton et al. 2009). For this reason, the same authors conducted a follow-up observing the effects of climate change in East Africa Finally, cassava and bananas may see increased yields in only.46 The authors found that maize in the region may East Africa by 2030 because of increased temperatures see improved yields in high altitudes, because of tem- (IPCC 2014, Chapter 22: Africa, 20). perature increases, but decreased yields in lower altitudes for the same reason (Thornton et al. 2009). Maize yields Whereas the exact impacts of global climate change on will also depend on water balances, and many places in crops in Rwanda remain unknown, it is clear that the cli- the region are expected to see water stress for the crop mate is altering, with increased potential for extreme events. (Thornton et al. 2009). The authors also evaluated the Because of the high population density of the country, impact on secondary-season beans (beans planted after where plot sizes are very small, adaptation measures will maize harvesting) and found that yield reductions will need to be taken to dampen shocks from these changes. 46 In this study, they found that crop yield responses to climate change are heterogeneous, and vary by crop type, location, temperature, and time. ­ Agricultural Sector Risk Assessment 67 Appendix C Vulnerability Assessment Over the past decade, Rwanda has made significant progress in reducing poverty, from 57 percent in 2005/06 to 45 percent in 2010/11. Extreme poverty decreased from 36  percent to 24 percent in the same period. The increase in agricultural productiv- ity is partly attributable to this achievement. Nevertheless, many groups remain vul- nerable, not the least in rural areas, where 49 percent of the population lives below the poverty line compared with 22 percent in urban areas. Overall, 40 percent of households in Rwanda can be classified as “low-income” agriculturalists; 14 percent of households rely on both agriculture and unskilled daily labor; and 13 percent rely solely on agriculture. Further, poverty is higher in female- and widow-headed house- holds compared with the national average (see table C.1). Table C.1. Poverty in Different Groups of Households, 2000/01 and 2005/06 2000/01 2010/11 Type of Population Poverty Population Poverty household Share (%) Incidence (%) Share (%) Incidence (%) All households 100 60.4 100 45 Urban 18 22 Rural 82 49 Female headed 27.6 66.3 23.8 60.2 Widow headed 22.0 67.7 18.7 59.9 Child headed 1.3 60.1 0.7 56.9 Sources: MINAGRI 2010; NISR 2012b; World Development Indicator Database, accessed in 2011; WFP 2012. Agricultural Sector Risk Assessment 69 Food Security Figure C.1. D  istribution of Food Insecurity in Rwanda, Food consumption is closely linked to the agricultural con- text. Households with less diversified incomes are more 2012 14 food insecure, and of those households with only one 12 activity (43 percent of Rwandan households), most are 10 engaged in agriculture. Further, in the WFP Food Security 8 6 and Vulnerability Assessment (WFP 2012b), agriculture 4 (size of land cultivated in Season A, crop diversity, own- 2 ership of livestock, cultivating a kitchen garden, whether 0 No problem Seasonal Chronic Acute the household still had food in stock from the last harvest reported in April) was one of four variables found to be statistically Source: WFP 2012. significant in explaining household food consumption. Figure C.2. F  ood Security and Similarly, food insecurity remains an issue and WFP’s Livestock Units in survey reports that 36 percent of rural households had Rwanda, 2012 unacceptable levels of food consumption in September 60 2011 and could be considered food insecure, compared 50 with 3 percent in Kigali City. However, the food security status in Rwanda is mixed and about 20 percent of those 40 households that are food insecure report seasonal food 30 insecurity. Over half of those who are food insecure are 20 chronically or acutely food insecure (figure C.1). 10 After Season A and B, 60 percent of households should 0 No livestock < 0.8 0.8 and above have acceptable food stocks. Seasonal food access prob- Source: WFP 2012. lems occur in the lean seasons just before the two main harvests (from March to May and from September to November) because food stocks run out. The households Figure C.3. F  ood Security and most exposed to seasonal food insecurity are the poorest Number of Crops and those relying most on seasonal work. Cultivated in Rwanda, 2012 In Rwanda, 85 percent of all working adults cultivate their 60 own farm and the WFP Food Security and Vulnerability 50 Assessment shows that the more crops a household cul- 40 tivated in Season A, the more likely it was to have better food consumption. Households reporting acceptable food 30 consumption cultivated an average of three crops, whereas 20 those with poor food consumption cultivated two. Having 10 a vegetable garden was also correlated with better food consumption. Finally, livestock ownership was associated 0 One Two Three Four and above with higher levels of food security (figures C.2 and C.3). Source: WFP 2012. On average and for all crops produced, households sold 23 percent of their production and consumed 71 percent. cereals, roots, and tubers as well as beans and cooking The rest was reported as either given away (2 p ­ ercent) or bananas were mostly kept for home consumption. In con- spoiled/lost after harvest (3 percent). The main consumed trast, households sold more than half of their production of 70 Rwanda Table C.2. P  ercentage of Households Gender and Vulnerability That Grow Specific Crops in Agriculture and Share of Production The agriculture sector is largely worked by women, but Sold in Markets much of their labor input goes unrewarded or is not vis- Households Crops Sold in ible in official statistics. Women are primarily restricted Growing Crop Markets to subsistence agriculture, receive low prices for their (%) (%) products, are underrepresented in agribusiness, and are Beans 90 12 employed in low-paid positions in secondary agriculture. Sweet potatoes 45 11 Also, female-headed households constitute about 30 per- Maize 42 22 cent of Rwanda’s households and these households are Plantains 28 30 very poor, which has consequences for their access to Irish potatoes 15 32 productive inputs and assets. High poverty levels in these Cassava 40 23 households also make them vulnerable to shocks as they Source: WFP 2012. do not have assets to cushion the impacts. Livestock have important impacts on food consumption and income, but because of gender structures, larger livestock (such as Table C.3. S  ources of Food and cattle and goat) are generally a man’s domain, restricting Food versus Nonfood women from profiting from these assets. Expenditures, 2012 A clear gender divide exists in the type of crops culti- Share of Total Change vated (table C.4). Because land is traditionally controlled Consumption since by men, crops produced by men are allocated more land. (%) 2005/06 (%) The types of crops dominated by men versus women Food purchases 26.6 + 24 are not consistent across the country, but depend on the Consumption of own 15.8   – 6 potential income from each crop in that particular area. food The production of crops with higher income potential Total food consumption 42.4 + 11 tends to be controlled by men. Nonfood expenditure 57.8 + 38 Total 100 + 24 Few women are involved in coffee and tea production Source: WFP 2012. activities. Women tend to have fewer trees than men because they have smaller land lots, but also because they tend to prioritize food crops over export crops. Further, cash crops (tea, coffee, pineapples, and sugar cane—all over men have better access to agricultural extension services 85 percent sold) and fruits and vegetables (tomatoes—80 than women, which affects their choice of crop. Tea and percent sold; passion fruit—60 percent; cabbage—58 per- coffee value chains are gender divided in terms of the type cent) in addition to sorghum (54 percent) and rice (63 per- of work conducted. For example, in the tea value chain, cent), meaning that these are more important sources of men plant and sell the tea, but women maintain the plants income (table C.2). and pick the leaves. Although women spend more time than men on tea production activities, men are paid bet- Markets provide little over 60 percent of the household ter in both value chains and also control the benefits from food basket, whereas own production contributes about coffee and tea production. 37 percent (table C.3). The market is the main source for rice (81 percent), groundnuts (67  percent), fish and meat Importantly, women reportedly have less access to tech- (90 percent; except poultry—50 percent), and milk (55 nologies promoted under the Crop Intensification Pro- percent), meaning that prices affect access to these food gram. This has partly to do with their more limited access products. to financial capital and assets, because the improved Agricultural Sector Risk Assessment 71 Table C.4. Gender Division of Crops Cultivation for Different Districts Crops Cultivated by Crops Cultivated by Both District Women Crops Cultivated by Men MEN and women Bulera Beans Irish potatoes Maize, wheat Gasabo Beans, sweet potatoes, cassava, Plantains, coffee, exotic vegetables Fruits maize, amaranth (Amaranthus) (tomatoes, eggplants, cabbage, green peppers) Kirehe Maize, beans, flowers Plantains, coffee, pineapples Sorghum Nyabihu Maize, beans, sorghum Irish potatoes, cabbage, carrots Highlands Beans Tea (but supply chain is gender divided), Irish potato, wheat, and maize Middle veld Beans, sorghum, sweet potatoes, Coffee cassava Ruhango Beans, sweet potatoes, vegetables Cassava, coffee, rice Maize Source: MINAGRI 2010. Figure C.4. Livelihood Zone Mapping in Rwanda Source: FEWS NET 2011. varieties, fertilizers, and chemicals promoted under the Livelihood and program are expensive. Female-headed households seem especially precluded from optimal participation in the Vulnerability across activities under the program. However, the technolo- Regions gies being promoted are also very labor intensive, which To better understand the impact of livelihoods on vul- reportedly restricts women from participating on equal nerability in different regions (and especially as they terms. Similarly, the “One Cow per Poor Family” pro- relate to agriculture), Famine Early Warning System gram planned for 30 percent of the beneficiaries to be Network (FEWS NET) conducted a “Livelihood Zon- women, but given the financial costs involved (because of ing” exercise in Rwanda (see figure C.4). FEWS NET the necessity of developing zero-grazing infrastructure), drew the following conclusions from this exercise: women, and especially female-headed households, are »» Most livelihoods in Rwanda are considered rela- largely restricted from benefiting under this program. tively food self-sufficient. 72 Rwanda »» Bugesera Cassava Zone is the only food-deficit »» The three Eastern livelihood zones (Bugesera production zone in the country, which happens Cassava, Eastern Agro-Pastoral, and Eastern only in bad years. This zone is drought-prone ­ Semi-Arid Agro-Pastoral Zones) are drought- area. prone areas. »» Eastern Semi-Arid, Eastern Agro-Pastoral, and »» Poor households living in the Eastern Agro-­ parts of the East Congo-Nile Highland Farming Pastoral, Eastern Semi-Arid Agro-Pastoral, and Zones are at risk of acute food insecurity during Eastern Plateau Agriculture Zones purchase bad production years. significant portions of their annual food needs. ­ Agricultural Sector Risk Assessment 73 Appendix D Detailed Calculations of Provincial Losses Table D.1. Banana Production by Province, 1998–2012 (Season A) and 2000–12 (Season B) Annual Loss of Ag. Average Average Production Value Coefficient of Annual Annual in US$ (2009–11) Variation of Average Yield Losses (MT) Losses (US$) (%) Yields (%) (MT/ha) Season A Eastern 17,947 3,602,524 –0.12 26.8 9.2595 City of Kigali 15,087 3,028,396 –0.10 25.3 7.2099 Northern 9,579 1,922,860 –0.06 19.0 7.5191 Southern 8,701 1,746,671 –0.06 15.0 4.9795 Western 9,053 1,817,292 –0.06 14.4 7.4352 National* 48,007 9,636,548 –0.32 18.2 7.3199 Season B Eastern 8,888 1,784,071 –0.06 17.5 9.7224 City of Kigali 2,309 463,473 –0.02 10.2 7.0310 Northern 7,385 1,482,412 –0.05 13.1 7.1201 Southern 8,783 1,762,963 –0.06 10.3 5.7059 Western 4,945 992,644 –0.03 11.3 7.5112 National* 23,276 4,672,171 –0.15 12 7.6081 *Totals are calculated based on the government’s national-level data and are not the sum of the provincial data. Agricultural Sector Risk Assessment 75 Table D.2. Maize Production by Province, 1998–2012 (Season A) and 2000–12 (Season B) Annual loss of Ag. Coefficient of Average Average Production Value Variation of Average Yield Annual Annual in US$ (2009–11) Yields 1998(A)/ 2007–12 Losses (MT) Losses (US$) (%) 2000(B)–2006 (%) (MT/ha) Season A Eastern 2,547 933,275 –0.03 31.1 2.0533 City of Kigali 330 120,926 0.00 32.1 1.5901 Northern 768 281,406 –0.01 9.7 1.9678 Southern 1,314 481,443 –0.02 21.1 1.6966 Western 2,482 909,381 –0.03 12.9 1.8705 National* 7,060 2,586,749 –0.09 6.8 1.9013 Season B Eastern 883 323,560 –0.01 23.9 1.4417 City of Kigali 86 31,312 0.00 13.6 1.2986 Northern 555 203,351 –0.01 15.6 1.5465 Southern 687 251,664 –0.01 43.6 1.3811 Western 1485 543,925 –0.02 22.9 1.6222 National* 3,406 1,247,758 –0.04 21.9 1.5280 *Totals are calculated based on the government’s national-level data and are not the sum of the provincial data. Table D.3. C  assava Production by Province, 1998–2012 (Season A) and 2000–12 (Season B) Annual loss of Ag. Coefficient of Average Average Production value Variation of Annual Annual in US$ (2009–11) Yields Average Yield Losses (MT) Losses (US$) (%) (%) (MT/ha) Season A  Eastern 7,096 2,125,186 –0.07 31.9 7.7885 City of Kigali 4,203 1,258,855 –0.04 34.8 8.5125 Northern 1,991 596,274 –0.02 34.2 7.1621 Southern 13,041 3,905,920 –0.13 46.0 7.6621 Western 9,317 2,790,422 –0.09 40.8 7.1294 National* 25,776 7,719,773 –0.26 35.5 7.7704 Season B Eastern 13,867. 4,153,319 –0.14 47.8 9.9042 City of Kigali 1,382 414,013 –0.01 47.1 8.5548 Northern 3,050 913,498 –0.03 44.2 7.9180 Southern 29,224 8,752,597 –0.29 46.3 9.6543 Western 13,164 3,942,641 –0.13 39.5 7.8525 National* 62,625 18,756,143 –0.62% 44.5% 9.1216 *Totals are calculated based on the government’s national-level data and are not the sum of the provincial data. 76 Rwanda Table D.4. Irish Potato Production by Province, 1998–2012 (Season A) and 2000–12 (Season B) Annual Loss of ag. Coefficient of Average Average Production value in Variation of Annual Annual Losses US$ (2009–11) Yields Average Yield Losses (MT) (US$) (%) (%) (MT/ha) Season A  Eastern 2,339 536,730 –0.02 46.1 2.9022 City of Kigali 2,123 487,280 –0.02 46.6 4.2955 Northern 70,908 16,273,446 –0.54 78.8 22.9434 Southern 10,550 2,421,261 –0.08 43.0 4.2931 Western 40,247 9,236,643 –0.31 61.1 14.5475 National* 45,214 10,376,654 –0.34 27.6 9.2085 Season B Eastern 1,089 250,021 –.01 15.2 4.8118 City of Kigali 159 36,408 0.00 12.4 5.6629 Northern 15,041 3,451,808 –0.11 26.5 11.4993 Southern 3,567 818,701 –0.03 12.8 6.6173 Western 11,235 2,578,447 –0.09 21.6 10.5824 National* 23,706 5,440,585 –0.18 16.9 9.6239 *Totals are calculated based on the government’s national-level data and are not the sum of the provincial data. Agricultural Sector Risk Assessment 77 Appendix E Food Crop Supply Chain Analysis INTRODUCTION The environment for staple crop production benefits from relatively consistent rainfall, occurring in two clear seasons, Season A (November to January) and Season B (April to June), although rainfall may also occur between these periods. Rainfall statistics indicate that within the rainy seasons, few periods of severe drought have occurred in the last 30 years, although seasonal variations may be as large as +/−15 percent, occurring approximately every 10–15 years. At the same time, Rwanda’s elevation at 1,500–4,500 meters above sea level promotes a cooler climate that permits the produc- tion of a wider range of crops than elsewhere in Sub-Saharan Africa. Potatoes in par- ticular grow well under the cooler conditions prevailing at higher altitudes; conversely, at lower altitudes crops such as bananas and rice grow well. Nevertheless, overall the temperature range is generally below the harsher high temperatures that can curtail crop growth. The moist conditions that prevail during the rainy seasons promote the spread of fungal diseases, whereas the consistent temperatures favor the development of insect pests. High levels of rainfall contribute to the leaching of soil nutrients and acidifica- tion of the soil. Moreover, the steep slopes that dominate much of the country render soils liable to erosion, especially once cultivated. Rwanda’s meteorological conditions are thus a two-edged sword that favors the growth of crops and their pests and diseases whereas also promoting soil degradation. Almost all staple crops (with the exception of rice) are grown by a large proportion of smallholders, many of whom produce only on a subsistence basis. The average farm size in 2006 was 0.72 ha. This is insufficient to provide sustainable food security and in rural areas, agricultural income (including the value of home consumption) aver- ages RF 120,697 (US$180) and represents only approximately 52 percent of aver- age household income of RF 235,000 (US$350). Of this amount, home consumption amounts to approximately 87 percent (NISR 2012b). Given the emphasis on subsis- tence production, markets have developed based on the commercial surplus that is sporadically generated and that can vary from season to season. In most cases, prices Agricultural Sector Risk Assessment 79 also fluctuate bimodally within the seasonal framework, This suggests that at the individual household level, risk albeit in a relatively predictable manner. Nevertheless, may play a larger part in restricting investment and reduc- such markets are generally poorly developed so that local ing production. This aspect of risk and its impact upon surpluses and shortages can arise as a result of traders’ production is assessed in the following analyses of the six limited capacity to take advantage of opportunities for food crops in turn and the different aspects of the risk spatial arbitrage. (Temporal arbitrage opportunities are inherent in the production of each. limited by the fact that most crops are produced twice in each year, so that price variations are short term in nature, BANANAS as well as by limited storage capacities among traders and Bananas are the most important crop in Rwanda in terms processors.) of volume and are grown throughout the country. Three types of bananas are recognized: Integrated Household The financial investment by most smallholders in crop Living Conditions Survey 3 (EICV3) data indicate that production has been low in the past. Growers have tended in 2010/11, 59.3 percent of rural households grow cook- to apply livestock manure and compost to nourish crops ing bananas, 47.9 percent grow beer bananas, and 38.8 as opposed to inorganic fertilizer. Fungicides and insecti- percent grow dessert bananas.47 The crop is produced cides have been rarely applied, and traditional varieties throughout the year, with marginally higher levels of have predominated over new improved seeds and planting ­ offtake occurring during Season A. material. Much of this is now changing as a result of the government’s Crop Intensification Program, which has The value chain for dessert bananas is poorly developed, promoted the use of inorganic fertilizers and the dissemi- and the extent of damage between production and con- nation of improved seeds and planting materials. This has sumption can be substantial, so most dessert bananas are resulted in a substantial increase in the production of all consumed at home. Cooking bananas are more resilient the crops falling under this program, although the level of and are both consumed at home and widely marketed fertilizer usage remains low (only 34,200 MT of inorganic domestically and in neighboring countries. Beer bananas fertilizer were imported into Rwanda in 2010/11, equiva- are both used at home and sold into the domestic mar- lent to less than 38 kg/ha over the entire seasonal crop ket, which consists mainly of small beer and wine produc- area per annum and less than 20 kg/ha per crop. This ers, although some larger processing plants also exist (for is higher than many SSA countries, but still considerably example, the COVIBAR factory, with a capacity of over less than the economic optimum. At the same time, most 2 million liters per annum). smallholders grow crops in mixture (in some cases mix- ing not only crops but varieties of crops as well). Beans In Rwanda, bananas are grown under three systems: are often grown in mixture with maize, whereas bananas (1) backyard cultivation (1–10 plants); (2) on small plots act as a shade crop for other lower-growing plants. This where bananas are planted as a second-story shade crop strategy allows optimal use of whatever plant nutrients in mixture with either perennial or annual crops; or (3) are available, and may reduce the spread of pests and dis- in monoculture. The bulk of Rwanda’s bananas are pro- eases, but also results in less than optimal yields of all the duced from small mixed plots where the bananas receive crops that are sown. little direct fertilizer but benefit from the fertilizer applied to other crops or generated by them through nitrogen Under conditions of good rainfall and equable tempera- fixation. tures, coupled with low input/output farming systems, it is not surprising that the following analysis of risk for six food crops grown in Rwanda shows little impact of most hazards upon the generally increasing production trends 47 Almost all bananas grown in Rwanda are part of the East African High- land Bananas subgroup. These are largely derived from Musa acuminata and are at a national level. Nevertheless, it is also evident that pro- genetically distinct from plantains. The subgroup contains five clone sets; one duction levels are still substantially below those that could consists exclusively of beer bananas, whereas the other four clone sets include be achieved given adequate investment in crop inputs. both cooking and dessert bananas. 80 Rwanda Production Risks number of different species feed on the roots of bananas, Moisture Stress causing lesions that can promote bacterial infection as The FAO defines bananas as being highly sensitive to well as restricted root growth that can lead to a higher moisture stress (Brouwer and Heibloem 1989). Yield frequency of toppling. The most common nematode reductions from lack of moisture can occur at any stage infesting bananas (Pratylenchus goodeyi) is found predomi- of growth, although it might be expected that the greatest nantly above 1,400 m, where it is present in most soils. impacts would occur during fruit development. Neverthe- Nematode populations increase over time in dense stands less, the period from flowering through to the develop- of bananas, and may eventually cause significant loss of ment of ripe fruit is long (105–155 days depending upon yield, but their impact can be reduced by cultural prac- variety), allowing opportunities for compensatory growth tices, including crop rotation and intercropping. Where so that given the relatively consistent rainfall regimes in these are practiced, yield losses are small. both Season A and Season B, substantial loss of yield is Banana weevils feed upon exposed corms of bananas and unlikely. It is only during extreme seasons (which occur are most prevalent below 1,200 m and in monoculture with less than 10 percent frequency) that significant loss plots. The damage caused by banana weevils is gener- because of moisture stress occurs. This does not imply ally limited, although in extreme cases, individual plants that rain-fed bananas in Rwanda can be expected to pro- can die and yield loss will be 100 percent. Nevertheless, duce the yields observed under irrigation. as with nematodes, the impact of banana weevils can be reduced by good cultural practices, including minimizing Flood and Wind the exposure of corms above the surface of the soil. Bananas are susceptible to flooding and short periods of submergence can lead to the death of plants. Neverthe- The two main banana pests thus constitute constraints to less, the occurrence of flooding is generally localized and production, limiting yield when stands of bananas are not to some extent predictable so that flooding is not consid- rotated or when cultural practices are inadequate. Under ered a significant risk to banana production. most conditions, the risk to banana production posed by either pest is minimal. Localized intense storms and high winds occur frequently in Rwanda. Winds in excess of 70 km/hr can cause fruit- Four diseases affect banana production in Rwanda. The ing banana plants to topple. Plants that are completely first, Panama disease, caused by Fusarium oxysporum, only uprooted can be replanted but will lose any fruit that has affects modern varieties of bananas and is of no risk to been developed. Partially toppled plants may continue to the East African Highland clone sets which are resistant bear fruit, but the resulting bananas are small and yields to the disease and constitute the bulk of production in are significantly reduced. Losses caused by wind are Rwanda. The risk caused by Panama disease is thus lim- unpredictable and can be substantial for individual farm- ited. The second, black sigatoka (Mycosphaerella fijiensis), is ers, although some damage can be prevented by propping capable of infecting all known varieties of bananas and plants as the bunches begin to develop. Reports suggest of causing substantial loss of yield. The disease caused that losses caused by wind although potentially severe at serious loss of yield in Rwanda in 2004 (National Bank an individual level, are restricted to no more than 100–200 of Rwanda, 2005 Annual Report) and has been present hectares each year, and are hence of minimal significance throughout much of Rwanda since that time. Under ideal to national production. Neither is the frequency or extent warm and wet conditions, transmission of black sigatoka wind damage sufficient to affect the investments made by can be rapid. It is spread by wind-borne ascomycetes, by growers in banana production. raindrop splash, and by poor crop hygiene. Cultural con- trol is of limited use and a preventive fungicidal regime of Pests and Disease up to 10 applications annually is required if substantial The two main pests of bananas are nematodes and loss of yield is to be avoided. This is beyond the capacity the banana weevil (Cosmopolites sordidus). Both pests are of many growers, so the risk posed by black sigatoka to widespread and lead to chronic losses. Nematodes of a banana production is considerable. Agricultural Sector Risk Assessment 81 Banana bunchy top disease is caused by a virus spread by destroying diseased plants and rotating with other crops sucking insects and poor crop hygiene. It causes the defor- for at least six months. BBW/BXW ranks with BBTD in mation of emerging leaves leading to eventual death of its severity of impact. the plant. Early infection causes 100 percent loss of yield, whereas late infection results in small and deformed fruit. Overall, the risks posed to banana growers by disease The disease is relatively easily transmitted, causes severe are considerable. An individual grower may easily expe- yield loss and most importantly, is rarely detected until the rience 100 percent loss of yield, the potential frequency characteristic “bunchy top” appears, by which time infec- of occurrence of any of the three diseases (black siga- tion is extensive and the plant will already be acting as a toka, BBTD, and BBW) is increasing, and the remedies source of infection to new plants. It is not surprising that available to growers are few. At a national level, the risk BBTD is considered the most devastating viral disease to the banana subsector posed by these diseases is also affecting bananas (ProMusa 2014). substantial. Although not quantifiable using a historical methodology, the potential impact to the industry is severe Banana bacterial wilt (BBW—also referred to as banana enough to warrant substantial investment in research and Xanthomonas wilt—BXW), was first identified in Rwanda extension to control all three diseases. in 2005. The disease is caused by Xanthomonas campestris pv. musacearum; bacteria multiply within the plant, produc- ing a gummy exudate that blocks vascular tissues (causing Market Risks wilting) and rots fruit. Infected plants die, with 100 per- Domestic Price Volatility cent loss of yield. The disease is spread by insects that The domestic market for bananas fluctuates, with lowest carry the bacterial exudates from plant to plant and by prices obtained at the end of each season as the bulk of poor crop hygiene. It can remain in the soil for up to five the fruit matures and comes to market (figure E.1) months. As of 2012, BBW had spread to 23 of Rwanda’s 30 districts. All varieties of banana are susceptible and Markets also vary according to type. Dessert bananas are no chemical treatment exists. Control is by uprooting and grown on a subsistence basis by most households. The Figure E.1. R  etail Price Variation in Domestic Markets for Bananas 2 year average 2012–2013 2013 2014 Banana: Nominal retail prices in Kimironko (Kigali City) Banana: Nominal retail prices in Kabacuzi 250 250 200 200 150 150 RWF/kg RWF/kg 100 100 50 50 0 0 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Banana: Nominal retail prices in Nyakarambi 250 200 150 RWF/kg 100 50 0 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Source: FEWS NET 2013b. 82 Rwanda value chain is not well developed and the risk of damage sigatoka, BBTD, and BBW) for which no effective treat- to fruit is high. As a result, prices for dessert bananas can ment exists and which can therefore cause dramatic loss be variable and there is considerable risk in engaging in of yield. From this perspective, a low-level system of pro- commercial production for the dessert market. Cooking duction that requires minimal investment appears to be banana prices are more stable because the fruit is har- entirely justified. Only if it is clearly evident that these vested when it is more resistant to damage and can there- three diseases can be effectively controlled can smallhold- fore be transported to a wider market. Nevertheless, prices ers be expected to increase the intensity of their produc- still fluctuate, in some cases unpredictably. By contrast, tion systems. Viewed in this way, the risks inherent in growers of beer bananas report that the offtake and mar- banana production are currently restricting national pro- ket for their product is more stable; this is cited as a reason duction to no more than 40 percent of potential output. for growing beer bananas in preference to the other two types, even though the yields of beer bananas are gener- BEANS ally lower than those of cooking or dessert bananas. Beans are arguably the most important crop in Rwanda in terms of national consumption. They are grown through- Nevertheless, Rwanda currently consumes more bananas out the country by 92 percent of rural households. Almost than it produces, with the deficit imported mainly from all beans grown in Rwanda are of the Phaseolus type,48 Uganda, so that prices remain close to import parity for which can be divided into bush beans, and the more much of the year. From this perspective, the probability of recently introduced climbing beans. A wide number of a significant fall in prices to unprofitable levels is minimal. different local and improved varieties exist for both types. They are adapted to different purposes and conditions, International Price Volatility with different lengths of growing period, degrees of deter- Given the consistent domestic deficit in bananas, it is minacy, and components of yield. Different varieties are inevitable that the Rwandan market will be affected by grown in different parts of the country as mixtures of international prices. Nevertheless, the international prices varieties, in mixture with other crops, or in monoculture. in question are those of markets in Uganda and to a lesser Beans are produced in both Season A and Season B. extent Kenya, the DRC, and Burundi. Global banana prices, which normally peak in March and are at their lowest in October–December, have little impact on the Production Risks international market that surrounds Rwanda, which is Moisture Stress effectively insulated against global market price fluctua- The FAO defines beans as being of medium-high sensi- tions by the high cost of transport of bananas into the tivity to moisture stress (Brouwer, Prins, and Heibloem area. As a result, international price volatility is of a simi- 1989). Sensitivity is particularly high during the periods lar order to the domestic price volatility and is not a major of flowering and pod initiation, when moisture stress can factor affecting the risk involved in banana production. result in the abortion of pods, and during pod filling, when moisture stress can cause reduced bean size. The Summary probability of yield loss through moisture stress is clearly Rwandan banana producers operate mainly with an greater in the drier climate in the eastern and southern extensive mixed cropping system that provides few inputs parts of Rwanda, but can be reduced through the use of to bananas and relies upon cultural practices to achieve varieties that have a shorter growing period, albeit with a adequate levels of pest and disease control. The intensity reduced yield. Bush beans have a shorter growing period of production could be substantially increased and yields than climbing beans and are therefore more suited for could be increased by 120 percent if a package of inputs production in the Eastern and Southern Provinces. more in keeping with international standards could be accessed by growers and applied. Nevertheless, such a level of investment is currently very much at risk from the 48 Other types of bean including Vicia faba and Vigna types are also grown, but impact of at least three potentially severe diseases (black the quantities produced are very small. Agricultural Sector Risk Assessment 83 Significantly, in Season A 2012, climbing beans in the beanfly is widespread throughout Rwanda. Loss of yield Eastern Province yielded 1.45 MT/ha, whereas bush is caused by the beanfly larvae, which emerges from eggs beans yielded only 0.78 MT/ha (approximately half). This laid within young leaves and mines its way through the suggest that a lack of moisture is not a constant constraint plant to the base of the stem where it completes its devel- to the production of beans in the Eastern ­ Province— opment. Damage can be extensive depending upon the rather that growers have adopted a risk mitigation strat- severity and timing of infestation, but has been estimated egy based upon bush bean production because given the to reduce bean yields nationally by 180–225 kg/ha (that variability of rainfall amounts in the Eastern Province, it is, as much as 25 percent of yield) (Trutmann and Graf is more certain that bush beans will produce a crop than 1993). At an individual farm level, the extent of dam- will climbing beans. From this perspective alone, it would age is also affected by the vigor of the crop. Adequate appear that erratic rainfall/moisture stress contributes soil moisture and nutrients are associated with vigorous to the risks involved in bean production that result in an crop growth and limited damage from beanfly, whereas approximately 50 percent reduction in the potential out- stressed or stunted crops tend to exhibit higher levels of put of beans from the Eastern Province. damage. Beanfly can be controlled through integrated pest management, including the use of resistant varieties, The impact of moisture stress in terms of varieties selected and by chemical spray, including neem, but it represents and consequent production is less in the other three prov- a constant threat to growers. It is almost inevitable that inces, especially in the Western and Northern Province of some damage will occur, but there is a lower probability the country, but it is nevertheless clear that moisture stress that such damage will result in severe yield loss. Never- is a significant component of risk affecting bean growers’ theless, beanfly damage represents more than a constraint cultural practices throughout the country. to production and must be considered as contributing toward the risk faced by bean growers. High and Low Temperatures Beans are sensitive to temperature, growing optimally Bruchid species infest bean pods in the field and can then at day temperatures of 20°C to 26°C. Temperatures of become important pests of stored beans, causing losses 30°C or above during flowering can lead to the abscission of up to 30 percent (Jones 1999). The pest can also be of flowers and low pod set, whereas temperatures below sustained within stores under conditions of poor stor- 20°C will delay maturity, thereby increasing exposure to age management. Control can be achieved through the moisture stress. The many different varieties of bean in use of resistant varieties, by coating seeds with edible oil Rwanda exhibit different degrees of sensitivity to tem- (which will kill Bruchid eggs), through anaerobic storage, perature and it is possible that the common practice of and through the use of fumigants. The two Bruchid species planting different varieties in mixture mitigates the impact that infest beans are widespread in Rwanda and a mini- of extreme temperatures by ensuring production from at mal level of infestation is inevitable. Good storage prac- least some proportion of the mixed crop. This risk reduc- tices will constrain such infestations. Poor practices will tion strategy will inevitably reduce yields below those that result in higher levels of damage. As a result, the impact could be achieved using single improved varieties (many of Bruchids is less of a risk inherent in bean production and of which yield at least twice as much as the traditional more of a constraint that obliges growers to invest in the landraces), but the risk of yield loss through extreme tem- basic requirements for good storage. perature is enough to justify accepting a lower level of production to guard against the more severe losses that Beans are susceptible to a wide range of diseases, in Rwanda, would occur if a single modern variety were to be planted. at least seven important diseases of beans exist: angular leaf spot, anthracnose, bacterial blight, aschocyta blight, rust, Pests and Disease bean common mosaic virus (BCMV), and root rot. There are two main pests of beans grown in Rwanda: the beanfly (bean stem maggot, Ophiomyia spp.) and the Angular leaf spot is the most important cause of loss of bean Bruchid (that is, a number of Bruchid species). The yield in Rwanda (Mukeshimana and Kelly 2001). It is 84 Rwanda caused by the fungus Phaeoisariopsis griseola, which infects the phaseolicola. These are the two most important bacterial dis- leaves, causing cell necrosis and consequent loss of yield. eases of beans in East Africa. The incidence of blight in The disease can occur at a range of temperatures, and is Rwanda can vary both locally and from season to season. favored by humid conditions; water is essential to infection. Both diseases are favored by high levels of humidity and Angular leaf spot is spread between crops on infected plant the continuous growing of beans in the same area, and can debris, and within crops by raindrop splash and by air cur- cause substantial loss of yield if crops are infected early. rents that can distribute spores over a wide area. Control Control measures include increasing the length of time can be improved through improved crop hygiene, but the between crops of beans in a plot and the use of modern risk of airborne infection remains. Damage to infected varieties that are effectively resistant to both CBB and HB. crops can be reduced through the use of fungicides. Aschocyta blight is a disease of cool humid climates and New varieties of bean are available that show resistance as such is more prevalent in the Northern and Western to angular leaf spot. Nevertheless, resistance tends to be Provinces. The disease is caused by a number of Ascho- more to specific local isolates of P. griseola, so that some cyta species and although many bush varieties of beans risk of breakdown of resistance remains. As such, angu- are susceptible, most of the modern climbing cultivars are lar leaf spot poses a significant risk to bean producers in resistant to aschocyta blight. As a result, it is expected that that its incidence is unpredictable and its impact can be although the disease has been a significant constraint to substantial. production in the past, with the spread of improved seeds into the Northern and Western Provinces, its significance Anthracnose is the second most important cause of yield will diminish. loss in Rwanda, and can reduce production by 35–95 per- cent according to the extent and timing of infection. The Bean rust is caused by the fungus Uromyces appendicula- disease, caused by the fungus Colletotrichum lindemuthianum, tus. This pathogen infects leaf material and disrupts cell is spread between crops on infected material, and in metabolism, leading to the production of new fungal infected seed. Within crops, the disease spreads by rain- spores at the expense of plant growth, causing severe loss drop splash. Anthracnose development is favored by cool, of yield in the process. The disease is favored by moist, wet conditions and is thus more prevalent In the Northern warm conditions, which promote both the growth and and Western Provinces. Control can be achieved through multiplication of the pathogen. U. appendiculatus can persist crop rotation combined with the use of disease-free seed. between crops as tough teliospores, but multiplies rapidly Some varieties of beans show different degrees of resis- within crops through the formation of uredospores, which tance to anthracnose so selection and planting of resistant are spread on wind currents to infect new plants. Bean varieties can also provide effective control. Chemical con- rust exists as a wide range of races of varying virulence, trol is also possible, but must be provided on a preventive with new races continually arising. Conversely, the many basis to be effective. This is generally too expensive for different varieties of beans in Rwanda possess different most smallholders. degrees of resistance to the different races of rust. Conse- quently, the development of a new and virulent races of Although anthracnose can cause severe loss of yield, it is rust can result in the rapid development of an epidemic an avoidable disease if clean seed of resistant varieties is if it can overcome the resistance of existing varieties. This planted in fresh ground. From that perspective, the disease occurs irregularly every 10–20 years. Chemical control of is less of a risk and more of a constraint to production in rust is possible, but the disease multiplies so rapidly that those situations where control has not been adequate, in fungicides must be applied on a preventive basis to be which case some level of infection is almost inevitable. effective and this is too expensive for most smallholders. Thus, although bean rust can be partially controlled by Bacterial blight of beans includes two diseases, common breeding for disease resistance, there is a continual risk bacterial blight (CBB) is caused by Xanthomonas campestris of the development of an epidemic, and consequent high pv. phaseoli, and halo blight (HB) by Pseudomonas syringae pv. levels of loss caused by bean rust. Agricultural Sector Risk Assessment 85 The bean common mosaic virus is a disease of beans new resistant varieties to combat the development of new in Rwanda that can cause yield losses of 35–98 percent varieties of the pathogens. As such there will always be a (Schwartz and Galvez 1980). It infects the entire plant, risk of resistance breaking down and of epidemic infec- causing root necrosis and leaf chlorosis and die back. It is tions caused by the development of new and virulent vari- spread by sucking insects, especially the black aphid. Con- eties of these diseases. trol can be achieved by good crop hygiene, particularly the use of clean seed, and by varietal resistance. Many of the From this perspective it is clear that a limited number of newly introduced climbing varieties are resistant to BCMV, specific diseases constitute a significant risk to bean grow- but as with some other diseases, resistance is specific to ers. Intensive investment in the production of beans will specific varieties of the virus and can break down if new be inhibited until such time as either new varieties with varieties of BCMV arise. In common with many other dis- strong “horizontal” resistance that is not easily broken eases of beans, the spread of infection can be reduced by down can be bred and disseminated, or until growers can planting beans in mixture with other crops or as mixtures afford the cost of preventive fungicides to combat these of varieties. BCMV is thus unpredictable in its incidence “high-risk” diseases. and infection, and once begun cannot be controlled. From this perspective, it poses a risk to bean producers. Market Risks Root rots caused by a range of fungal agents (mainly It is estimated that up to 30 percent of the bean crop is Fusarium, Rhizoctonia, Pythium, and Sclerotium spp), either marketed. Most production is sold domestically, but the alone or in complexes, can cause loss of yield varying in value chain is not well developed and it is hard to acquire severity with the timing of infection and the condition of commercial volumes of beans for trading purposes. Nev- the plant. Weak plants growing in waterlogged soils can ertheless, the domestic market appears to be well inte- be killed if infected at the seedling stage, whereas older grated, with prices moving in parallel in different markets plants in drier soils may appear unaffected. A number of (figure E.2). local and improved varieties, especially climbing varieties, have been found to be resistant to one or more of the vari- Price data show small seasonal fluctuations, but overall ous fungal agents (Nzungize, Chrysostome, Mukashema, domestic price volatility is limited and the element of Ikirezi, and Nivitange 2011), suggesting that the genetic risk because of poor market prices in any given season diversity of Rwanda’s bean subsector has evolved as an is not great. effective mechanism to cope with various risks, including A very small amount of Rwanda’s bean production is in this case, root rots. Control therefore consists of good exported, mainly to Burundi and the DRC (and some cultural practices (especially crop rotation) and the use beans are occasionally imported from Uganda depending of resistant improved or traditional varieties. If these are upon local price fluctuations), but the market is isolated available, the probability of root rot infection is low and by transport costs from the global trade and overall the the risk posed by the disease is similarly minimal. impact of international price fluctuations is negligible. Overall, it is evident that a wide range of diseases pose a significant threat to bean production in Rwanda. In some Summary cases, diseases exist as constant constraints to production Rwandan bean growers produce yields that are higher (for example, anthracnose, aschocyta, and root rots), caus- than that of most of the rest of Africa, but lower than ing relatively predictable levels of loss under certain con- the commercial optimum and substantially less than the ditions. Other diseases can cause higher levels of loss with potential of new varieties. The element of risk is clearly much less predictability and little effective control other a factor in the reduced investment that leads to the lower than improved cultural practice and the use of resistant level of production. That risk includes the impact of irreg- varieties. Such diseases, including angular leaf spot, rust, ular rainfall and moisture stress, high and/or low temper- BCMV, and blights, require the constant development of atures, and especially the impact of specific diseases. 86 Rwanda Figure E.2. Price Movements in Domestic Markets for Beans Previous year 2012 Current year 2013 Beans: Nominal retail prices in Rukomo Beans: Nominal retail prices in Nyakarambi 400 500 350 400 300 250 RWF/kg RWF/kg 300 200 150 200 100 100 50 0 0 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Beans: Nominal retail prices in Bugarama Beans: Nominal retail prices in Kabaya 500 500 400 400 RWF/kg RWF/kg 300 300 200 200 100 100 0 0 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Source: FEWS NET 2013b. The mixed crop cultivation methods used by most Rwan- crops maize and beans are included). The crop is grown dan growers are appropriate for the level of risk inher- throughout the country by 52.3 percent of rural house- ent in bean production. They provide an optimal strategy holds and is produced throughout the year. in the face of diverse risks of uncertain frequency and potentially substantial impact. If growers have access to In Rwanda, cassava is often interplanted with beans or disease-resistant varieties, a significant element of risk is other low-growing crops. Cassava thus benefits from removed and higher levels of investment are appropriate whatever fertilizer is applied to the other crop, or in the (and are indeed used—the level of fertilizer application case of beans, indirectly from nitrogen fixation. The crop on the more disease-resistant climbing beans is twice that is consumed both as a subsistence crop and marketed in used on local varieties). This clearly demonstrates the sig- raw and processed forms. Raw tubers can be found in nificant impact of risk upon bean production and suggests local markets, but the crop must be processed if it is to that if such varieties could be more widely disseminated, be stored and a substantial proportion of Rwandan cas- yields might be increased by 50 percent overall. sava is sold as dry chips or flour. Processing is generally a cottage industry; chips are often produced by individual Thus the longer-term yield trends for beans are largely smallholders and flour by micro millers (although a large dependent upon the dissemination of disease-resistant cul- cassava processing plant was opened at Kinazi in 2013). tivars which remove a substantial element of the risk fac- Cassava is the fifth most widely consumed commodity in ing growers and thereby permit more intensive cultivation. Rwanda, making up 3.4 percent of national consumption. CASSAVA Global data suggest that under rain-fed conditions, cas- The area and production of cassava in Rwanda are sava yields of over 25 MT/ha/year can be regularly slightly less than those of Irish potatoes. As such, it is achieved through the use of improved modern varie­ the third most important perennial crop (after bananas ties and the application of approximately 100 kg/ha N, and Irish potatoes) and fifth crop overall (when the grain 50 kg/ha P2O5, and 100 kg/ha K2O as inorganic fertilizer. Agricultural Sector Risk Assessment 87 Current national yields of 12.1 MT/ha/year are mark- 1988). As a result, although the cassava mealy bug can be edly below this level. This may be the result of a number found throughout Rwanda, its impact on yield is relatively of factors, including the continued use of less productive low and it can be considered more as a constraint to the varieties, poor cultural practices, high levels of disease, achievement of maximum yields than a risk. The impacts and reduced levels of soil fertility. The following analysis of nematodes are similarly a constraint to yield rather attempts to quantify these effects. than an unpredictable cause of significant yield loss. The impact of the green spider mite on yield can be Production Risks considerable, but the pest is sufficiently ubiquitous that Moisture Stress it might currently be considered more of a constraint to Cassava is widely considered to be a drought-tolerant production than a risk. Current control options, includ- crop, although the soil moisture levels at which cassava ing breeding for resistance and biological control, have demonstrates symptoms of stress are actually higher than yet to demonstrate substantial success and chemical con- those for maize and beans. However, it is this characteris- trol of the pest, although effective, is impracticable under tic of cassava (the capacity to respond rapidly to moisture current conditions. stress by closing stomata and limiting cell metabolism) that allows it to endure periods of low moisture and equally Two main diseases currently affect cassava production rapidly regain production once adequate soil moisture is in Rwanda, caused by the cassava mosaic virus and the restored (Lebot 2009). Nevertheless, cassava is sensitive to cassava brown streak virus. Both viruses are spread by moisture stress during the period of root growth (30–150 the white fly B. tabaci and by the distribution of infected days) and prolonged drought can reduce yields by 30–60 plant material. Both viruses can reduce yields by as much percent. Under the rainfall conditions that prevail in as 95 percent. Currently CMV is more prevalent, but the Rwanda, cassava production will be substantially affected disease situation in Rwanda has historically been quite by reduced rainfall approximately 1 year in 10 at most. It fluid, with new virus diseases arising every 10–15 years is therefore unlikely that moisture stress contributes sig- (FAO 2010) and it is possible that a new form of CBSV is nificantly to the risks inherent in cassava production in spreading rapidly (Bigirimana, Barumbanze, Ndayihanza- either season in Rwanda. maso, Shirima, and Legg 2011). In 2007, an assessment of CMV found the disease at 94 percent of plots visited, with Pests and Disease 32 percent of plants infected and an impact on the yield of Pests of cassava include the green spider mite (Mononychellus infected plants estimated at 60 percent (Night et al. 2011). tanajoa), cassava mealy bug (Phenacoccus manihoti), white fly (Bemisia tabaci), and nematodes. Of these, the green spi- Control of viral diseases in cassava relies upon good crop der mite is widely distributed and in 2007 was found to hygiene to limit the spread of infection and upon the infest approximately 40 percent of all cassava plants, introduction of new disease-resistant planting materials. causing 45 percent damage on average where infestation New cultivars with good resistance to CMV are available, occurred (Night et al. 2011). The predatory mite Typhlo- but given the relatively slow rate at which cassava can be dromalus aripo has been introduced into Rwanda as a bio- multiplied through conventional processes (one plant gen- logical control agent of green spider mite, but appears erally yields about 10 cuttings for planting) and the bulk to be only moderately successful in controlling levels of of the planting material required, the diffusion of new infestation. White fly is present in all areas, and although varieties through the country will take several years. the direct impacts of this pest are minimal, it is significant as a primary means of transmission of viral diseases. The Nevertheless, planting material resistant to CMV has incidence of cassava mealy bug has been reduced by the been introduced to many growers, and yields of cassava introduction 30 years ago of the parasitoid wasp Apoana- have increased substantially since 2007. For those grow- gyrus lopezi as a biological control agent, which effectively ers who continue to use older susceptible varieties, CMV controls more than 90 percent of infestations (Norgaard and CBSV must be considered constraints to production, 88 Rwanda Figure E.3. Monthly Retail Prices of Cassava Flour, 2012 and 2013 2 year average 2012–2013 2013 2014 Cassava flour: Nominal retail prices in Birambo Cassava flour: Nominal retail prices in Kimironko (Kigali City) 500 500 400 400 300 RWF/kg 300 RWF/kg 200 200 100 100 0 0 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Cassava flour: Nominal retail prices in Muhanga Cassava flour: Nominal retail prices in Ruhuha 500 500 400 400 RWF/kg 300 RWF/kg 300 200 200 100 100 0 0 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Source: FEWS NET 2013b. limiting yields and thereby constraining levels of invest- Summary ment in the crop. By contrast, for those growers who have Overall, there are few risks inherent in the production of planted disease-resistant varieties, the reoccurrence of dis- cassava. This is surprising, because although the crop is ease is now a much-reduced risk, allowing them to apply relatively unaffected by weather, it is significantly affected more inputs and greater attention to the cassava crop in by pests and diseases. Moreover, national yields are sig- the knowledge that they are likely to achieve substantially nificantly below the levels that could be achieved under greater yields. optimal investment. The reason for this appears to lie in the frequency with which pests and disease impacts occur; Market Risks that is, among those growers using traditional varieties, Price Volatility the impacts of pests and diseases are sufficiently con- In Rwanda, cassava is marketed as a raw tuber, as flour, or sistent for these hazards to be considered constraints to as chips. The multiplicity of markets and the opportunities yield. Nevertheless, among growers using disease-resistant for processing help smooth market prices over the course varieties, whereas green spider mite infestation may still of the year and domestic price volatility is low (figure E.3). be a constraint, disease is not, and hence cassava produc- tion can be undertaken with a much higher degree of Cassava is exported from Rwanda to Burundi and the certainty that potential yields will be achieved. For these DRC as flour and to Burundi as raw tubers. This gen- growers, there remains the uncertainty of yield loss from erally occurs when cassava production in either of these green spider mite and this may still be a factor affecting two countries is reduced but has little impact upon domes- investment decisions. Nevertheless, with the exception of tic prices in Rwanda other than to support the market in this one pest, the production of disease-resistant variet- times of surplus. Global cassava prices do not affect mar- ies of cassava exposes smallholders to few risks. This may kets in Rwanda and there is little price risk as a result of account for the consistent increase in production observed either domestic or international price volatility. since 2007. Agricultural Sector Risk Assessment 89 MAIZE Nevertheless, the issue can be addressed from two per- spectives. From a meteorological perspective, the data Maize is widely grown in Rwanda: 75 percent of all house- show a general trend across the country, according to holds grow maize, making it the fourth most frequently which lower rainfall amounts are received in the Eastern grown crop after dry beans, sweet potatoes, and Amaran- and Southern Provinces (<900 mm per year) compared thus (pigweed). Almost three-quarters (72 percent) of the with the Western and Northern Provinces (1,200 mm per maize grown in Rwanda is grown in Season A. year). This reducing trend might be expected to be associ- ated with a higher degree of variability but in fact, statisti- Maize is the most rapidly expanding crop in Rwanda. cal analysis of rainfall data has shown no significant trend As one of the CIP crops, its production has benefited in the variability of rainfall amounts across the country from subsidized inputs of seed and fertilizer (which are (indeed the Western Province appears to be marginally also provided as loans). In 2012, it was the second largest more variable) and no significant difference in variability grain crop after beans in terms of volume produced (over between the two main rainy seasons (McSweeney, Sema- 500,000 MT). Nevertheless, according to the EICV3 con- fara, Cole, and Washington 2011). At the same time, sumption survey conducted in 2010/11, only 11 percent rainfall records do indicate a degree of interannual varia- of the crop is actually sold; a substantial proportion of the tion, especially when data are disaggregated across differ- crop is consumed as green maize, before the grains ripen; ent parts of the country. Variations equivalent to +/–20 maize grain (or flour) itself accounts for only 2.6 percent percent of the seasonal total are relatively common, with by value of the national diet (NISR 2012b). negative anomalies occurring at least once every 10 years (figure E.4). Production Risks Although meteorological data indicate few distinct trends, Moisture Stress this may be more a reflection of the very limited num- Maize requires constant moisture for optimal growth and ber of weather observations available for analysis than yield is reduced if the maize crop is allowed to wilt consis- of a real lack of difference in variation, because from tently for more than 48 hours. Growth is particularly sensitive the perspective of smallholders, erratic rainfall in 2008 to moisture stress during three periods: (1) when the crop is caused a loss of yield among 37 percent and 26 percent 50 cm high and dry conditions can restrict the development of smallholders in the Eastern and Southern Provinces, of the reproductive organs (15 percent); (2) during tasseling, respectively, as opposed to 19 percent and 14 percent in silking, and the completion of pollen germination, when dry the Northern and Western Provinces.49 These figures, conditions can reduce the number of grains that will develop recorded in a year of above average rainfall nationally, in each cob (50 percent); and (3) during early grain develop- show that uncertain rainfall is perceived to have a signifi- ment, when dry conditions can result in shriveled or aborted cant impact upon yield and might therefore be considered grains (30 percent). During the latter two growth stages, the a significant risk faced by smallholders. Indeed, a varia- maize plant is more developed, with a greater leaf area, tran- tion of 20 percent in seasonal rainfall could reduce yields spiration from which may require as much as one liter of by as much as 50 percent if the dry spell occurred during water per day. If soils are deep and well structured, crops at the critical tasseling and silking stage of growth. these growth stages may be able to extract more water from greater soil volumes by virtue of their greater depth of root- Smallholders can respond to this perceived risk in a num- ing, but if soils are shallow or of low water-holding capacity, ber of ways. By reducing plant density, more soil water then the demands of evapotranspiration will exceed the sup- will be available per plant so that a reduced sowing rate ply capacity of the soil and wilting will occur. is a common adaptation in drier areas. The use of maize varieties that, although they may yield less, complete their Given the multiplicity of soil types and depths across the growth cycle in a shorter time (90 days as opposed to the country, it is effectively impossible to quantify the impact of erratic rainfall upon maize yields in different parts of Rwanda on the basis of an analysis from first principles. 49 Comprehensive Food Security and Vulnerability data. 90 Rwanda Figure E.4. R  ainfall Anomalies for the March-April- May Period (Season B), by Province 500 North East South West 400 300 Rainfall anomaly (mm) 200 100 0 –100 –200 –300 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 Year Source: McSweeney, Semafara, Cole, and Washington 2011. traditional 120–150 days) also reduces exposure to the risk The significance of flooding as a risk varies greatly with of erratic rainfall, especially the late onset or early cessa- location. In valley bottoms, the probability of flooding tion of rains. Good soil management can make a substan- is much higher than on the hillsides. A significant pro- tial difference, although the application of organic matter, portion of the maize currently produced in the Eastern often promoted as a way of increasing the water-holding Province is grown in the lowlands of the Akanyeru River capacity of soil, has little impact, whereas improving root- basin, where the additional moisture under otherwise dry ing depth through the removal of hoe- or plough-induced conditions results in above-average yields, albeit with an soil pans has been shown to have a much greater impact increased risk of losses caused by flooding. In these areas, (Conservation Farming Unit 2007). improved drainage infrastructure is the most appropri- ate measure to mitigate this risk. Nevertheless, for the Overall, however, a limited number of options are open to majority of smallholders growing maize, flooding is not a farmers who wish to grow maize and it must be accepted major risk and is unlikely to affect yield. This is reflected that at least 1 year in 10, individual crop yields may be in national data, which suggest that the impact of flooding substantially reduced. on maize production is negligible. Floods Flooding is an occasional hazard reported by smallhold- Pests and Disease ers in Rwanda. Young maize plants are very sensitive to Prior to 2007, maize areas in Rwanda were considerably flooding and can survive for only two to four days under smaller and more dispersed than they are now, and the water.50 But this sensitivity decreases over time and once increased consolidation and importance of maize that has maize has reached the stage of grain formation, shallow occurred in the last five years will undoubtedly increase depths of flooding will not cause any noticeable damage.51 the probability of losses caused by pests and diseases. Currently, however, pests and diseases levels remain low; until 2013, only leaf blight and maize streak virus were 50 http://www.ag.ndsu.edu/procrop/env/fldwhb07.htm recorded as significant diseases of the growing crop. In 51 Ibid. June 2013, however, maize chlorotic mottle virus was Agricultural Sector Risk Assessment 91 identified in the Western and Northern Provinces. This country is a net importer of maize, seasonal variations in virus (together with sugar cane mosaic virus—SCMV) is price occur against a backdrop of import parity pricing, a component of maize lethal necrosis disease, a disease with prices falling below import parity during periods of complex that has spread rapidly in Kenya since 2012 and immediate surplus and rising to import parity levels over can cause up to 100 percent loss of yield. This disease the remainder of the year (figure E.5). poses a significant threat to future maize production. It can be controlled through the introduction of resistant Historically, maize prices show few unexpected variations varieties and through stringent crop hygiene measures. and appear to be at least as consistent as those in global These include improved scouting to detect early out- maize markets. Domestic price volatility cannot therefore breaks and immediate disposal by uprooting and burning be considered a significant risk for maize producers. all diseased plants. Similarly, losses caused by insect pests in the growing crop International Price Volatility have rarely been significant. Maize stalk borer (Busseola Because Rwanda is situated at a considerable distance from fusca) is the only pest reported to have caused significant any seaport, costs of inland transport reduce the relative losses.52 In stored grain, insect damage from common importance of short-term fluctuations in global markets; pests of stored grain (such as Sitophilus zeamais, the greater thus Rwanda is more dependent upon the prices prevailing grain weevil) is not unusual, but because the grain is stored in neighboring countries. Recently these prices have been only for a short period, levels of loss have generally been at or above average, with the exception of the Zambian low. The larger grain borer (Prostephanus truncatus) is not yet market, which can be accessed via Lake Tanganyika and a threat in Rwanda. which in 2011/12 was oversupplied with grain. Neverthe- less, even though free on board (FOB) grain prices in Zam- One other threat to maize production in the Eastern Prov- bia were only 50 percent of domestic prices, supplies were ince is infestation of the striga weed, which can cause high limited and did not significantly affect the Rwandan mar- levels of crop loss. Nevertheless, because the presence of ket. Overall, a comparison of national and global price this weed is generally predictable, this is less of a risk and indexes suggests that Rwandan maize markets have not more of a constraint to production. been greatly affected by international price volatility and this has not posed a significant risk to growers. Overall, the risk to smallholders growing maize from pests and diseases has been historically low, but is expected to Risk inherent in maize production appears to be primarily increase as plots of maize are consolidated and as grain associated with the availability of adequate moisture and is stored for longer periods. Viral diseases in particular this factor more than any other can be expected to con- are a potential threat and warrant an intensive extension strain the investment decisions of growers. A potential risk program to help smallholders learn how to identify and may exist as a result of increased disease pressure, particu- dispose of diseased plants. larly the threat of MLN disease, but that has not yet been widely experienced and as a result is unlikely to be fac- tored into smallholders’ investment decisions at present. Market Risks Domestic Price Volatility Nevertheless, it is not yet possible to determine how much Maize prices in Rwanda are determined primarily by of the substantial gap between the current levels of maize local supply and demand. The relatively consistent pro- production and those that could potentially be achieved duction that occurs twice a year results in limited and can be ascribed to smallholders’ perception of risk and predictable fluctuations in price, with the lowest prices how much is attributable to the limited availability of occurring immediately after harvest in February/March inputs, especially inorganic fertilizers. Under the CIP pro- and to a lesser extent in August/September. Because the gram, fertilizer inputs have been made available for maize both on credit and at a 50 percent subsidy. The volume of 52 Ibid. fertilizer purchased in 2011 for the production of maize 92 Rwanda Figure E.5.  Seasonal Variation in Retail Prices of Maize in Different Markets 2 year average 2012–2013 2013 2014 Maize: Nominal retail prices in Kimironko (Kigali City) Maize: Nominal retail prices in Kirambo 500 450 400 400 350 300 RWF/kg RWF/kg 300 250 200 200 150 100 100 50 0 0 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Maize: Nominal retail prices in Ndago Maize: Nominal retail prices in Rukomo 450 450 400 400 350 350 300 300 RWF/kg RWF/kg 250 250 200 200 150 150 100 100 50 50 0 0 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Source: FEWS NET 2013b. and wheat was 8,000 MT of urea and 10,000 MT of DAP. all food purchases and 8.3 percent of all food consump- This was theoretically applied to 223,000 ha of maize and tion. Much of the crop is consumed locally; only a small 43,000 ha of wheat, giving an average nitrogen applica- volume is exported, although Rwanda is a net exporter tion rate of only 20 kg/ha. That is an amount that would of Irish potatoes. be removed by 1.2 MT of maize grain per ha, suggesting that the current levels of fertilizer availability are inad- National yields of potatoes in Season A have tended to equate to sustain yields above those c­ urrently achieved. be higher than those in Season B and areas sown in Sea- son A have been consistently higher, so 60 percent of Thus, whereas the impact of agricultural risk, especially the national crop is now produced from Season A and the risk of yield loss caused by erratic rainfall, on the percent from Season B. 40 ­ production of maize might be significant, it is currently masked by the limited availability of inputs. The risk to Potential Potato Yields growers of crop loss caused by disease might be of greater Irish potato yields have been recorded at up to 35 MT/ha significance in the short term if MLN disease becomes (Durr 1983); field trials of the Kinigi and Gahinga varie­ more widespread. ties (both still widely grown in Rwanda) when they were first released in 1982 by the National Potato Program (PNAP) achieved yields of 25.8 MT/ha (Monares 1984). IRISH POTATOES This would suggest that an economic optimal yield would In 2012, Irish potatoes were the second largest vegetable be between 20 and 30 MT/ha. crop after bananas in terms of volume produced (over 2.1 million MT). EICV3 data for 2010/11 indicate that it Internationally, a yield of 25 MT/ha is often considered was grown by 52.9 percent of all households. Irish pota- the commercial optimum under ideal commercial con- toes are the second most important staple in Rwanda. ditions. In practice, countries such as the United States, EICV3 data indicate that they constitute 7.6 percent of Canada, and the United Kingdom achieve average yields Agricultural Sector Risk Assessment 93 of 20 MT/ha, 31 MT/ha, and 45 MT/ha, respectively, indicate that 60 percent of potato production was con- whereas in Africa, Ugandan and Kenyan growers produce centrated in just three districts: Niyabihu (19 percent) approximately 7 MT/ha. Part of the wide disparity is due and Rubavu (23 percent) in the Western Province and to climate, which is certainly more variable in Kenya than Musanze (20 percent) in the Northern Province. Never- it is in the United Kingdom. A major difference, however, theless, the crop is an important staple and although the lies in the higher level of inputs applied than the more volumes recorded from the Southern and Eastern Prov- stable climate justifies. In the United Kingdom, the aver- inces are small, the proportion of households producing age rates of N, P2O5, and K2O application recommended the crop remains substantial (table E.1). for optimal yield are N: 150–210 kg/ha, P2O5: 250 kg/ha, and K20: 360 kg/ha. By contrast, potatoes in Rwanda receive on average only 12 kg/ha of each plant nutrient as Production Risks inorganic fertilizer. It can be argued that additional plant Moisture Stress nutrients are applied as manure and compost, but the The production of potatoes is very dependent upon the volumes necessary to achieve the rates of nutrient appli- availability of adequate soil moisture. The FAO classifies cation required to produce optimal yields are massively the potato crop as highly sensitive to soil moisture (FAO greater than the amounts of organic manure available 1989) throughout the period from stolonization through and applied in practice. to tuber initiation to early ripening. Moisture stress during this period can result in the development of fewer tubers, Yields of potatoes in Rwanda are thus currently constrained whereas moisture stress during tuber growth causes mis- by the amounts of plant nutrients available to growers. This shapen potatoes. Stress at either time will reduce yields. is unfortunate given that the use of other inputs of potato Although the variability of rainfall during both Season A production—especially fungicides for disease control—is and Season B is not large compared with that of rainfall in quite widespread, but has not resulted in the potential yield other SSA countries, it is nevertheless possible that yields benefits that could be achieved through the application of may be substantially reduced by limited moisture availabil- adequate nutrients. It is possible that growers have limited ity at least 1 year in 10 and that lesser yield reductions may their use of inputs to those they consider most appropriate occur more frequently. The risk of crop losses as a result given the perceived risks. The following analysis considers of moisture stress is to some extent offset by the impact of whether or not those decisions are justified. drier conditions upon disease (see below) and by the high returns that potato production can generate when demand exceeds supply. As a result of these factors, growers that Distribution can absorb the risk of drought-induced loss normally The production of Irish potatoes requires low tempera- adopt an intensive approach to potato cultivation. This is tures to restrict pests and diseases and high soil moisture not the case in Rwanda, suggesting either that growers: availability. Consequently, Irish potatoes are grown on a 1.  Perceive the risks of drought to be substantial; commercial basis almost exclusively in the Northern and 2.  Lack the capacity to absorb crop losses when they Western Provinces. In fact, data for Season A in 2012 do occur; or Table E.1. Yields and Total Production of Irish Potatoes by Province in Season A, 2012 Province Southern Western Northern Eastern City of Kigali Yield (MT/ha) 5.26 17.07 18.04 5.17 3.52 Production (MT) 81,419 712,394 502,547 38,172 1,446 Production (%) 6 53 38 3 <1 Households (%) 58.3 46.5 58.1 57.3 20.9 Source: Agricultural Marketing Information System 2012; NISR 2012b. 94 Rwanda 3.  Are unable to source the inputs necessary to adopt that blight need not be so much a risk to potato grow- a more intensive approach to potato production. ers, but a predictable constraint to profitability measured simply in terms of the cost of the chemicals required and their application, as a necessary and inevitable input to Pests and Disease the production process. This does in fact appear to be the All conventional varieties of potatoes are susceptible to general trend among growers, who are increasingly aware blight (Phytophthera infestans) to some degree. This disease of the necessity for the preventive use of fungicides and more than any other reduces potato yields in Rwanda are investing accordingly. Nevertheless, to be fully effec- (and indeed elsewhere in the world). Other important dis- tive, this trend should be accompanied by government eases include a range of potato viruses that are most com- regulations to control the spread of diseases on noncom- monly spread by sucking insects and which can cause leaf mercial plots, including the destruction of volunteers and yellowing or other deformities, and bacterial wilts caused diseased plants. by Pseudomonas solanacearum and by Erwinia complexes (also causing soft rots), which are spread by latently infected Insect pests do infest potatoes, but are of less significance tubers and by volunteer plants. All of the above diseases themselves than as carriers of disease, especially a number are exacerbated by poor crop hygiene, including reduced of different viruses that infect Solanaceae. Nevertheless, the rotation periods (the period between potato crops in the presence of sucking pests is a constant constraint to the same soil should ideally be at least four years), the ubiquity production of new planting material, which must be kept of volunteers or backyard potato plants grown by non- as free of insects as possible if it is not to contain viruses commercial growers that can act as a reservoir for disease, when sold to farmers. and by the use of infected seed (as a result of the limited supplies of clean planting material). Nevertheless, the occurrence of blight is the greatest risk Market Risks faced by potato growers, because the spread of this dis- Domestic Price Volatility ease is favored by the cool, wet conditions under which Neither storage nor processing facilities exist for potatoes most potatoes are grown in Rwanda. Indeed, the overall in Rwanda, hence domestic price volatility is considerable impact of blight may well be mitigated by the fact that the (figure E.6). There is no government intervention in the conditions that favor its spread are also those that lead to market for potatoes, which faces shortages immediately the highest yields of potatoes. The impact of blight can before harvest and gluts immediately afterward. be devastating, resulting in up to 100 percent loss of yield To avoid the impact of each glut, growers tend to harvest and rendering such tubers as might be produced ined- as early as possible, generally before the tubers are fully ible. Even mild infections can result in significant loss of mature, which tends to reduce their shelf life considerably. yield and it is not surprising that considerable emphasis Price volatility is offset to some extent by three factors: (1) is placed upon the regular application of fungicides to the fact that potatoes can be grown in two, if not three, control the disease, whereas in some areas growers delay seasons in Rwanda; (2) the staggering of planting across planting so that the crop matures under drier conditions. different provinces; and (3) the import of early- or late- This reduces the incidence of blight, but increases the risk harvested potatoes from Uganda. Nevertheless, farmers of yield loss through insufficient moisture. regard domestic price volatility as a significant risk inher- The probability that a potato crop in Rwanda will be ent in the production of potatoes. Investment in storage infected by blight is inherently high. The risk that such and/or processing facilities might help to offset this risk. an infection might lead to devastating loss of yield might be somewhat higher than in other countries, but can def- International Price Volatility initely be reduced through timely application of a fun- There is a market for potatoes produced either in Rwanda gicidal regime using a chemical (Dithane) that has been or Uganda in both the DRC and Burundi, and potatoes widely available for 40 years. It might therefore be argued from either source can be found in both destinations. To Agricultural Sector Risk Assessment 95 Figure E.6. Variation in Domestic Market Prices of Irish Potatoes Irish potato: Nominal retail prices in Musanze Irish potato: Nominal retail prices in Mahoko 400 400 350 350 300 300 250 250 RWF/kg RWF/kg 200 200 150 150 100 100 50 50 0 0 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Source: FEWS NET 2013b. that extent, it could be expected that the domestic market Given that additional investment in plant nutrients can would be closely linked with the international market and result in substantial additional yield, it might appear that that prices in both markets would follow similar trends. the primary constraint to potato production is indeed This is broadly true, but neither domestic nor international the lack of available fertilizer. Nevertheless, the volatil- markets for potatoes are well developed. The number of ity of prices and especially the marked declines in price large traders active in these markets is small (reportedly associated with overproduction (even if only in the short less than 10) and clear opportunities for spatial arbitrage term) can offset the beneficial impact of an increase in frequently exist (USAID 2012). It is evident though that yield. Evidence for the counterargument (that the pri- the prices available to producers and traders are deter- mary constraint to investment is perceived risk) comes mined almost entirely by production within Rwanda and from Uganda, where average yields of 7.0 MT/ha are neighboring parts of Uganda, and there is no evidence of lower than those in Rwanda, even though conditions are any impact of global markets (for example, of potatoes similar and the same technologies are available in both from Egypt or China). International price volatility does countries. not contribute to the risks involved in the production of potatoes. If it is indeed perceived risk that constrains growers’ investment in and production of potatoes, then mea- sures such as the increased availability of multi-seasonal Summary finance, crop insurance, and subsidized inputs may all Potato producers appear to be at risk as a result of erratic serve to offset that risk and should lead to increased pro- rainfall, disease (mainly blight), and fluctuations in mar- ductivity. Further investigation is required before this can ket price. The responses of growers to these risks include be determined with certainty. changes in sowing and harvesting dates as well as the use of fungicides and pesticides. Overall, cultural practices achieve yields of no more than 50 percent of what has RICE been shown to be commercially achievable under rain-fed Rice is a CIP crop that benefits from subsidized inputs conditions in Rwanda. Much of the difference between of seed and NPK fertilizer. In 2012, it was the third larg- actual yields and the economic optimal yield (in the main est grain crop (after beans and maize) in terms of volume potato-producing districts) can be ascribed to inadequate produced (over 80,000 MT). It is planned to expand the crop nutrition. What is unclear at present is whether or not area sown to rice substantially, but this will require signifi- inadequate nutrition is a result of the limited availability cant development of drainage and irrigation infrastruc- of fertilizer within Rwanda, or of a reluctance on the part ture, and areas sown are currently static. Domestic rice of growers to invest in the application of additional nutri- consumption exceeds that of maize flour. Local rice con- ents because of the perceived risk that the returns may not stituted 3.7 percent of all food purchases in 2010/11 and justify the additional investment. imported rice, 3.1 percent (NISR 2012b). Local supply meets approximately 50 percent of demand at present. 96 Rwanda The production of rice in Rwanda is subject to two major Pests and Disease constraints: temperature and moisture. As a result, it is Given that rice is grown in large areas across valley bot- grown almost exclusively in the lower valley bottoms, toms, the crop is vulnerable to the rapid spread of pests where temperatures are high enough to sustain growth and diseases. From a research perspective, these appear and the marshy conditions provide adequate water. This to be the main risks now inherent in rice production. Rice restricts the area under production; only 4.5 percent of all blast (Magnaporthe oryzae) and bacterial disease complexes households grow rice in Rwanda. The crop is produced (leaf and panicle blight caused by Xanthomonas spp. and mainly in three provinces (Western, Southern, and East- sheath rot associated with Pseudomonas infection) are the ern); lower temperatures preclude its production in the major diseases causing yield loss in rice and can affect all Northern Province, and it is most common in the Eastern known varieties. Control is currently based mainly upon Province, which is both warmer and contains the valley crop and varietal rotation, but RAB noted that for these areas necessary for optimal production. Approximately 40 diseases, “pathogen evolution is so fast that within 3 to 4 percent of the rice grown in Rwanda is grown in Season A growing seasons most grown varieties become susceptible and the balance in Season B. to the extent of causing total crop failure.” Lower levels of yield loss are more common, but can regularly be as much as 20 percent. Other diseases such as rice yellow Production Risks Moisture Stress mosaic virus and smuts also occur but with little impact Rice in Rwanda is produced under marshland conditions, on yield. which are not the same as irrigated conditions (although The few chemical treatments available to constrain the some irrigation systems do exist), but depend more upon spread of these diseases appear to have little effectiveness controlled drainage to ensure adequate levels of moisture in Rwanda. Some experts noted that this may be because are available at key growing periods. Such systems are vul- the disease is often recognized and pesticides are generally nerable to water shortage, especially at the beginning of applied only after the disease has become well established the season if delayed rains have not allowed the accumu- and affected yields. lation of adequate moisture for initial germination and growth. Thereafter, a prolonged dry spell may reduce Insect pests of rice are limited to the rice fly (Diopsis tho- growth and ultimately yield, as may excessive flooding, racica), the larvae of which eat out the center of young although varieties capable of withstanding both dry and tillers, causing blind shoots. Yield loss depends upon the wet conditions are increasingly available. severity and timing of infestation because the impact of early infestations, once controlled by insecticides, can be Research (Akram et al. 2013) has demonstrated that with- mitigated by compensatory growth. Nevertheless, yield holding irrigation water from a rice crop for a 14-day losses of 5–20 percent are commonly recorded (Akinsola period reduces paddy yield by between 10 and 40 per- and Agyen-Sampong 1984). cent, depending upon the time at which moisture stress was imposed. Drought stress at panicle initiation had the Insect pests of stored rice have been reported, espe- greatest impact on yield, whereas stress at anthesis and cially the rice weevil (Sitophilus oryzae) (Dunkel, Sriharan, grain filling led to reduced impacts. In all cases, however, Niziyimana, and Serugendo 1990), but these do not ­ yield reductions exceeded 10 percent. appear to be a significant risk for growers or millers given the limited time for which the crop is stored. It is not surprising that when interviewed, rice farmers’ key concern appeared to be the availability of water, both from adequate rainfall and from its equitable distribution through Market Risks communal drainage systems. Farmers considered it essential Domestic Price Volatility that available water be effectively distributed both through Rice prices in Rwanda are affected by a government pol- improved drainage and irrigation channels and through icy that determines a minimum price paid to rice mills by proper management of those channels once in place. licensed traders. Extensive restructuring of the rice-­milling Agricultural Sector Risk Assessment 97 subsector has resulted in the closure of small private mills, The high levels of production that have been observed which have been replaced by new and more efficient large can be viewed from another perspective: an individual mills owned by rice cooperatives. Smallholders as mem- marshland smallholder is obliged to invest as much as he/ bers of the cooperatives receive inputs and produce rice she can to achieve an economic return from rice because that is purchased by the mills at a price determined before he/she is effectively tied into a communal drainage/irri- the crop is sown. Traders are not allowed to buy directly gation system that offers no alternative sources of income, from smallholders and the large mills are the only source and no market other than the local cooperative, and which of rice for traders. As a result of this system, neither grow- charges a membership fee, irrigation fee, and manage- ers nor mills face any risk from domestic price volatility, in ment fee, as well as rent for the land (which the coopera- that prices and potential margins are known before any tive leases from the state and allocates to growers). Under investment in inputs is made. such circumstances, a low-risk strategy is futile, because overheads will inevitably be incurred and the most effec- Nevertheless, such prices are not always favorable to grow- tive strategy will be to maximize returns through intensive ers; for example, in December 2013, farmers in Muhanga investment. district complained that the price they received (RF 250/ kg) was inadequate to cover the costs of production at the Provided the local cooperative provides a rice grower with yield that they had achieved (3.5 MT/ha). They suggested the support necessary to absorb the downside impact of that RF 300/kg would have been appropriate to cover risk (for example, through insurance, deferred loan repay- their costs. The cooperative’s response was that prices ments, or subsidized inputs), then high levels of risk can were set before sowing and would not be increased and be incurred and a high input/high output system of pro- that farmers should seek to improve the fertility of their duction can be sustainable. If that support is not available, land for the next crop. then growers facing losses will be unable to participate in rice production on an ongoing basis. International Price Volatility The stability of domestic prices, coupled with the signifi- Current observations suggest that the degree of risk inher- cant costs of transport to Rwanda from seaports, have cre- ent in rice production will increase as areas sown to rice ated a stable domestic rice market, even though imports expand and disease pressures increase. If growers can be from Tanzania, Thailand, and Pakistan may make up assisted to develop the capacity to absorb the increasing 50 percent of the market volume. International price vol- risk, then levels of production can be sustained. If not, atility is not a significant risk to either growers or proces- they may not increase beyond current levels unless risk sors in the Rwandan rice value chain. itself can be reduced—primarily through the develop- ment of disease-resistant varieties and improved disease Summary scouting to assist in the identification and control of dis- Although there is only limited market risk for growers and ease outbreaks. processors of rice, growers in particular are vulnerable to the impacts of erratic rainfall and disease. It is therefore unexpected to see such high levels of investment and con- CONCLUSION sequent high yields achieved across much of the rice-pro- The above risk analysis for selected staple crops in ducing areas by smallholder producers, who have limited Rwanda suggests that even though Rwanda has a fairly capacity to absorb the downside impact of the risk incurred consistent climate and stable markets, growers still face through such investment. Actual yields are consistently considerable risk and adjust their level of investment and more than half of the potential maximum that could be production accordingly. All growers face risks associated economically achieved, suggesting that growers are either with climate. Although the frequency of significantly confident that they can mitigate the impacts of moisture low rainfall amounts in a given season is not high (less stress and disease or that they can absorb those impacts, than 10 percent), the probability of erratic rainfall and although in practice neither of these situations are realistic. short-term moisture stress is much higher, so that some 98 Rwanda element of yield loss from moisture stress is almost inevi- by growers who have tailored their output to match normal table. This factor contributes to the risks faced by grow- domestic demand. From this perspective, poorly developed ers of most crops, especially maize, beans, bananas, and, markets might be considered as constraints to production, to a lesser extent, rice and potatoes. Only cassava is not but there is little evidence that they make a significant con- much affected. tribution to the risk faced by growers. A significant component of risk is related to the unpredict- One significant exception to the response to risk observed able impact of disease, especially in beans and bananas, for most crops is found in the production of rice, for which and to a lesser extent in rice and maize. In Irish potatoes, growers apply significantly higher levels of inputs and disease can be expected to occur with a high frequency achieve yields closer to those of commercial producers. and thus constitutes more of a constraint than a risk. This assumption of risk appears to be due on one hand to the commercial nature of the crop and on the other Market risks appear to be of limited significance to the pro- to the particular circumstances under which the crop is duction of most staple crops, at least insofar as variations in grown, according to which growers are faced with signifi- retail prices indicate little unpredictable volatility in domes- cant overheads regardless of the level of production they tic markets and only limited linkage with international mar- achieve. As a result, they are obliged to assume a higher kets. This may be a result of the limited level of production level of risk to achieve profitability. Agricultural Sector Risk Assessment 99 A G R I C U LT U R E G L O B A L P R A C T I C E T E C H N I C A L A S S I S TA N C E P A P E R W O R L D B A N K G R O U P R E P O R T N U M B E R 96290-RW 1818 H Street, NW Washington, D.C. 20433 USA Telephone: 202-473-1000 Internet: www.worldbank.org/agriculture