Report No. 46929-AFR Regional Trade in Food Staples Prospects for Stimulating Agricultural Growth and Moderating Food Security Crises in Eastern and Southern Africa December 17, 2008 Agriculture and Rural Development (AFTAR) Sustainable Development Department Africa Region Document of the World Bank CONTENTS EXECUTIVE SUMMARY ............................................................................................... vi 1.INTRODUCTION .......................................................................................................... 1 2.MARKET SHEDS INEASTERNAND SOUTHERNAFRICA .................................. 4 3 SOUTHEASTAFRICAN FOOD STAPLE ZONES ................................................. . -23 4 MODELING THE IMPACT OFREGIONAL TRADE INMODERATINGMAIZE . PRODUCTIONSHOCKS INSOUTHEASTAFRICA ................................................... 35 5 CONCLUSIONS . ............................................................................................................ 49 REFERENCES.................................................................................................................. 55 ANNEX 1.METHODS FOR DEFININGMARKET SHEDS......................................... 61 ANNEX 2 THE MULTI-MARKET MODEL . ................................................................. 64 ... 111 LIST OF FIGURES 1.Domestic andImport ParityPrices inSouthAfrica 2. Maize Marketing Flows inSouthern and Eastern Africa 3. Southern Mozambique Market Shed 4. Domestic andImport ParityPrices inMaputo 5. South East Africa Market Shed 6. Domestic and Import Prices inSouthern Malawi 7. Zones o f Heavy Border Area Procurement bythe Zambian FoodReserve Agency 8. East Africa Market Shed 9. Domestic and World Prices for White Maize Grain, Wholesale inNairobi 10. Trends incassava andmaize production inZambia, Malawi and Mozambique 11.Maize ProductionVariability inSouthern Africa, 1990-2005 12. Spatial Distribution o f Population andUrban Areas 13. Percentages o f Households Growing Maize and Cassava in Zambia, Malawi and Mozambique 14. Spatial Distribution o fMaize and CassavaProduction inSouth East Africa 15. Area per Household Planted inMaize and Cassava inSouth East Africa 16. Food Staples Zones inSouth East Africa 17. Food Imports Moderate Price Shocks during a Drought LIST OF TABLES 1.Size ofAgricultural Markets inSub-Saharan Africa 2. Quantity Flows inMajor Maize Market Sheds 3. FoodExpenditure Shares inSouthern and Central Mozambique 4. High-Side Outliers Among FoodReserve Agency Purchases 5, Average National Production and Consumption o f Cassava andMaize 6. Correlation Coefficients o f Maize Production among Selected Southern African Countries 7. Alternative Criteria for Defining Food Staple Zones 8. Baseline Population and Income Data 9. Baseline Production and Consumption Data 1O.Model Parameters 11. Simulation Results: Northern Mozambique 12. Simulation Results: Malawi 13. Simulation Results:Zambia Drought 14. Simulation Results: Zambia Bumper Harvest iv ACKNOWLEDGEMENTS This report focuses on growing trade infood staples inthe Southern and Eastern African region o fAfrica as one o f the largest growth opportunities available to African farmers. The report was preparedby a MichiganState University team ledby Steven Haggblade, and comprising Jones Govereh, HunterNielson, and David Tschirley under the guidance o fthe World Bank task team comprising Tijan M.Sallah and Paul Dorosh. Overall guidance was provided byKaren McConnellBrooks (Sector Manager) andMartienvanNieuwkoop (Program Coordinator). The report has benefited greatly from specific comments by Stephen Minkand Sergiy Zorya and from Michael Morris, Petros Aklilu (the peer reviewers), and Sonia Plaza (Senior Economist, Regional IntegrationUnit) at the concept review stage and from Louise Fox, JanNijhoff, William Odwongo, David Rohbach (peer reviewers) inthe final stage o fthe report. Meseret Kebede provided the ACS support. The report was b d e d by the Bank Netherlands Partnership Program (BNNP) trust fund supported by the government o fNetherlands. The main stakeholders for the report are COMESA and SADCLFARNRPAN andthe findings o f the study are to encourage the exchange o f ideas about trade and food security issues and should not be attributed to the World Bank, its Executive Board o f Directors, or its membercountries. V EXECUTIVE SUMMARY Cross-BorderTrade inFoodStaples: 1. Over the next generation, growing trade in food staples appears poised to dwarf other agricultural markets in Africa. Currently, the market value o f Africa's food staples amounts to $50 billion per year, or nearly three-fourths o f the value o f all agricultural production. Given growing urbanization and the highest rates o f poverty in the world, Africa's market demand for food staples will grow dramatically incoming decades. As a result, production o f food staples -- for growing urban markets and food-deficit rural areas -- represents probably the largest growth opportunity available to African farmers. Facilitating expansion o f these markets will, therefore, be critical for efforts at stimulating agricultural production, broad-based income growth andpoverty reduction. 2. Political borders frequently separate surplus food production zones from the deficit markets they would normally serve in large cities and in deficit rural areas. Drawn in Berlin in 1885, the continent's highly arbitrary political boundaries cut across natural market sheds, impeding the free flow o f people and goods. They separate food surplus Northern Mozambique from deficit markets inMalawi and Eastern Zambia; food surplus zones in Uganda and Northern Tanzania from deficit markets in Kenya, and surplus cassava and maize producing areas o f Northern Zambia from the deficit mining towns o f Katanga and Kasai provinces in the DRC. Political borders translate into a welter o f tariff and non-tariff barriers, which together constrain cross-border trade in food staples. In turn, these artificial impediments to trade raise the costs and lower the incentives facing both farmers and traders while simultaneously raising consumer food prices in cross-border deficit zones. 3. Inorder to maintainproducer incentives inAfrica's manysurplus food production zones, farmers in these zones need access to growing markets, both internal and across national borders. Without access to regional export markets, production surges in thinly traded national markets lead easily to price collapses, which in turn risk stalling production growth and private investment inagriculture. 4. Inaddition to helpingmaintain producer incentives in surplus zones, cross-border trade can help moderate consumer price volatility and food supply shortages in deficit areas. Import parity price sets an upper bound, while export parity sets a price floor, provided governments allow grain to flow freely across their borders. 5. This paper examines the impact o f regional trade in food staples, both for maintaining farmer incentives in surplus food production zones and for moderating price spikes in deficit areas. The paper begins by identifying the geographic extent o f major maize market sheds in Eastern and Southern Africa. It then focuses on the South Eastern Africa market shed, the one centered in Malawi, Northern Mozambique and Zambia. This analysis concentrates on the region's two most important food staples, maize and cassava. Given the volatility o f the region's rainfed maize production, this section aims vi to empirically evaluate the impact o f maize production shocks on staple food prices, production incentives, consumption and trade. To do so, the paper develops a spatially disaggregated model o f maize and cassava markets in order to evaluate the impact o f supply shocks, with and without cross-border trade. MaizeMarket Sheds inEasternand SouthernAfrica: 6. Traders and other market observers in Eastern and Southern Africa typically identify three broad market sheds for white maize, the most widely consumed food commodity in the region and the focus o f most food policy debates. Moving from south to north along the eastern half o f the African continent, these three principal market basins include Southern Mozambique, South Eastern Africa and East Africa. Southern Mozambique Market Shed. Despite reliable maize surpluses produced in northern Mozambique, and chronic maize deficits in the south, maize does not flow in appreciable volumes from northern to southern Mozambique. Instead, the surplus maize from northern Mozambique supplies nearby deficit markets inMalawi and sometimes in eastern Zambia. Long distances between northern and southern Mozambique, long wait times at the Zambezi River ferry, and high transport costs make it cheaper for Maputo- based maize mills in the south to source their maize from South Africa rather than internally from northern Mozambique. Transporting maize to Maputo from Northern Mozambique costs roughly $100 per ton, transiting 1,500 kilometers and taking about three days time. In contrast, purchases by rail from South Africa travel only 250 kilometers and cost under $20 per ton to transport. Hence, Mozambique straddles two distinct maize market sheds, with the Zambezi River providing the demarcation line betweenthe two market basins. 8. The main deficit markets in southern Mozambique lie in the far south, in the cities o f Maputo, Matola and Xai Xai and to a lesser extent in the central coast city o f Beira. Domestic surpluses from the central interior areas o f Manica Province supply Beira and provide seasonal supplies o fwhite maize to the major urban centers inthe south. Regular supplies to the southern cities also come from South Africa, where silos house grain in close proximity to the Mozambican border along rail lines that permit inexpensive transport into the major mills o f Maputo. In recent years, official imports from South Africa into southern Mozambique have averaged about 210,000 tons per year. 9. South East Africa Market Shed. Northern Mozambique and Southern Tanzania produce the largest maize surpluses inside the South East Africa market shed. In normal production years, they export maize to Malawi, about 140,000 from Northern Mozambique and another 90,000 tons from Southern Tanzania. In years of urgent demand in Malawi, such as 2002 and early 2003, maize exports into Malawi from these two neighboring countries totaled in the range o f 600,000 tons. In addition, a cohort o f large and small commercial farmers in Central Zambia typically market about 500,000 tons o f maize per year, enough to supply Zambia's urban markets and, in most years, exports to the DRC. These surplus maize production zones serve a range o f deficit markets within the market shed. vii 10. Historically, Malawi has constitutedthe core net deficit market inthe South East region. Duringthe first half o f the 2000s, annual inflows o f white maize into Malawi averaged about 200,000 tons, with slightly over half coming from informal trade. Deficit households, in both rural and urban Malawi, have until very recently relied on maize imports from South Africa, Northern Mozambique and Southern Tanzania. Inaddition to supplying Malawi, the large maize surpluses produced in northern Mozambique serve urban Nampula and other cities in northern Mozambique. Lusaka and the Copperbelt mining towns o f Zambia rely primarily on supplies from large and small commercial farms in Central and North Central Zambia. Typically, these same farms supply Lubumbashi and other towns on the DRC side o f the Copperbelt with about 25,000 tons o f maize annually. In drought years, government and private traders supplement domestic production with imports from South Africa and Southern Tanzania. Lubumbashi, a city o f two million people in the DRC Copperbelt, imports maize from Zambian suppliers in years when commercial supplies and government export policy permits. But when Zambian production falls short, or its government bans exports, as they have inrecent years, Lubumbashi traders and millers import maize and maize meal from South Africa. 11. East Africa Market Shed. Deficit markets inKenya provide the center o f gravity for the East Africa market shed, pullingin surplus maize from Kenya's own central highlands as well as from Uganda and northern Tanzania. Given consistent informal imports from Uganda, inthe range o f 100,000 tons per year, official trade statistics understate Kenya's import dependence as well as Uganda's status as a consistent net exporter. In some years, maize surpluses from southern Ethiopia flow into northern Kenya, although poor roads limit these flows. The Ugandan surpluses likewise periodically find their way into deficit markets inRwanda, DRC and Southern Sudan. Although quantities vary from one year to the next, average data from 2000 to 2004 indicate that informal imports supply roughly half o f Kenya's 400,000 ton annual maize deficit. Most o f this comes from Uganda and Tanzania, which together supply roughly 250,000 tons o f maize to Kenya each year. 12. These realities suggest that strategies for accelerating growth insurplus agricultural areas will require a regional perspective. If farmers and traders in surplus food production areas are to invest in soil fertility, productive assets and input supply and marketing systems, they will need reliable access to deficit markets, which in many instances lie across national borders. Import bans, export bans and other trade controls, often intermittent and unpredictable, undercut incentives for farmers and traders and tend to undermine prospects for accelerated food production growth. From the consumer's perspective, ensuring food security in deficit markets likewise requires a willingness and ability to source supplies from regional markets when these offer prospects for low-cost supply. ... Vlll FoodStapleZones in SouthEastAfrica: 13. The remaining portions o f this paper focus on the South East Africa market shed, inorder to assess empirically the potential impact o f regional trade in moderating food supply shocks stemming from a volatile maize harvest. Given data requirements and availability, this analysis focuses on three core countries within the South East African market shed: Northern Mozambique, Malawi and Zambia. 14. Maize and cassava constitute the two most important food staples in the South East Africa market shed. InZambia and Malawi, maize dominates the national food basket, providing just over 50% o f all calories consumed, while cassava provides roughly 10%. However, inMozambique, the rankings are reversed. Cassava serves as the number one food staple, supplying about one-third o f total calories; while maize, the number two staple, furnishes about one-fourth. Likewise inthe cassava zones of northern Zambia and the northern lakeshore region o fMalawi, cassavapredominates. 15. Because o f its vulnerability to moisture stress, maize production varies significantly from one year to the next. Hence the importance o f a range o f drought-tolerant secondary food staples, such as sorghum and millet in the temperate zones and cassava in the tropical zones. 16. Three broad food staple zones exist within South East Africa. The maize belt, where maize dominates staple food production and consumption, includes the interior segments o f the region, including southern Zambia, southern Malawi and central Mozambique. The cassava belt, in turn, includes sections o f northern Zambia, the lakeshore region o f Malawi and pockets along the north coast o f Mozambique. Dual staple zones, where maize and cassava co-exist in significant quantities in production and consumption baskets, cover a large swath o f the region, cutting across northern Zambia, Malawi and Mozambique as well as Mozambique's southern coast. 17. Because households in northern zones o f Mozambique, Malawi and Zambia consume both cassava and maize, and because they can harvest cassava over several years, households can choose to consume more cassava and sell more maize during drought years, thus releasing maize for sale to deficit maize-belt households. Inbad years, when nearby maize belt households face acute deficits, farmers from neighboring cassava and dual staple zones are able to harvest more o f their perennial cassava crop and inturn free up more maize for export to deficit zones. These mixed and dual-staple zones, thus, serve as potentially important food security shock absorbers, enabling the release o f maize to deficit areas in times o f short supply, thereby moderating regional food shortages. Simulatingthe Impactof RegionalTrade inModeratingMaizeProductionShocks: 18. Focusing on the three core countries o f the South East African market shed, this paper applies a multi-market model designed to quantify the impact o f production shocks on domestic food prices. Inturn, the model assesses the impact o f these changing prices on i x consumer, farmer and trader behavior. Consumers reduce maize consumption and increase consumption o f alternate staples as maize price rises. Traders and millers import and export inresponse to differentials between domestic and border prices. And farmers alter planting decisions in response to changing prices. As exogenous variables, the model includes a range o f potential instruments wielded by government and donors. These include trade quotas, tariffs, public imports, government exports, local procurement, government stockholding and sales, and targeted income transfers to vulnerable groups. 19. When maize production fluctuates, and when adjustments must take place within the confines o f small national markets, this production volatility translates into wide swings in maize price and rapid compression in food consumption by vulnerable groups. Our simulations from Malawi and Zambia suggest that a drought which reduces national maize production by 20% can lead to a 60% increase inmaize prices, while a 30% fall in productioncan trigger a price rise inexcess o f 100%. 20. Two import safety valves can help to moderate consumption and price volatility in the presence o f recurring maize production shocks. The first o f these food system shock absorbers i s regional trade in food staples. During the Malawian food crisis o f 2002/03, informal traders from Northern Mozambique, delivered on the order o f 200,000 to 300,000 tons of maize into Malawi while farmers from southern Tanzania fbrnished fbrther 100,000 tons (Whiteside, 2003). At these levels, total informal imports amounted to between 20% and 25% o f normal consumption in Malawi. Our simulation results suggest that even more modest inflows -- o f 100,000 tons o f imports -- in response to a moderate drought, can cut price spikes by as much as 50%. Given these magnitudes, regional trade flows can clearly help to soften supply deficits. 2(1. The second major shock absorber highlighted in this paper i s consumer substitution among food staples. Results from northern Zambia illustrate the importance o f this consumer substitution. Even during a serious drought, reducing national maize production by 30%, poor households in the cassava-consuming north see calorie intake from cassava and maize fall by only 2%. Overall, this cassava substitution for maize compensates for about 40% of the national maize shortfall inthese circumstances. While discussion here has concentrated on cassava as a secondary food staple, these results should be considered illustrative o f the broader opportunities for consumer substitution o f a whole array o f drought-tolerant alternative staples including sorghum, millet, and sweet potatoes. In urban areas, shifting consumption patterns and growing consumer preferences for rice and wheat products open hrther opportunities for substitution among food staples. 22. Livestock feed and livestock products, likewise, offer important substitution possibilities, particularly given growing demand for poultry, meat and other livestock products. In years such as 2008, with sharply rising world cereal prices, the introduction o f alternative local carbohydrate sources (such as cassava and sorghum) in feeds offers a potentially important conduit for relieving pressure on maize demands. X 23. Ingeneral, policy restrictions - such as export bans, permit systems and tariffs - tend to drive cross-border trade from formal into informal channels. Because handlingcosts rise with small lots, and with transfers from truck to bicycle and back onto trucks to avoid border controls, these policy limitations do, in fact, diminish trade flows by driving up transaction costs. Tracking informal maize trade from Southern Tanzania into Malawi, one study notes, "Costs fell inmid 2001, when the export ban was lifted and the traders no longer had to take the maize across the border by bicycle and canoe. This illustrates how lifting an export ban can tip the balance between profitable and unprofitable trade." (Whiteside, 2003, p.33). 24. Production surges, when confined within small national markets, lead quickly to price collapse. In Zambia, for example, a 30% boost in maize production can lead easily to a 50% price fall, if borders remain closed. However, when policy makers allow maize exports, to neighboring DRC, the export parity price forms a price floor, limiting the price fall to a half, to about 26%. Open borders, by increasing farmer incentives in surplus areas and dampening price spikes in deficit zones, can contribute to both agricultural growth inthe surplus zones andto food security throughout the region. 25. Recent increases in world cereal prices have substantially raised the cost o f importing food from international markets. Therefore, the coming crop year will prove particularly difficult for deficit areas unless they can raise domestic production or source supplies at lower cost from surplus regional suppliers in Africa. Attentiveness to regional production forecasts, regular tracking o f regional markets and border prices, ongoing policy monitoring and open dialogue among neighbors can all contribute to improved preparedness and responsiveness to unfolding food requirements. Over the coming year, assuring food security for vulnerable households and deficit zones within Africa will depend, more than ever, on open borders andfluid cross-border flows o f staple foods. PolicyImplications: 26. Open borders for regional maize trade offer potentially important benefits to farmers in surplus production zones and consumers in deficit areas. Because regional maize production varies less than production in each individual country, open borders offer a means o f reducing supply and price volatility in individual countries. The common policy alternative, o f import bans and export bans, leads to highly volatile maize prices, price spikes which risk punishing consumers in some years and farmers in others. Over the medium run, reduced price volatility encourages on-farm investment by farmers in surplus zones and mitigates forced consumption compression by vulnerable households. The key policy instruments for encouraging cross-border trade include cessation o f Regional trade associations -- such as COMESA, SADC and E A C -- provide fora for quantitative controls, tariff reduction, and harmonization o f customs procedures. negotiating and enforcing regional trade agreements. 27. Improved regional infrastructure can complement trade reforms by lowering transport and transaction costs. Adopting a regional perspective, public investment in transport corridors can help to link surplus farming zones with cross-border deficit markets, thus x i benefitting farmers in surplus zones and consumers in the deficit markets. Transport corridors between Northern Mozambique and Malawi offers a clear example o f the potential food security benefits o f regional infrastructure investment program. To be effective, these investments will need to be accompanied by harmonized customs regulations andpolicy reforms facilitating cross-border flows o f food staples. 28. Food policy discussions need to expand beyond maize. Secondary food staples such as cassava, sweet potatoes, sorghum and millet merit increasing attention in domestic and regional food policy debates. Currently, policy attention and government investments focus largely on maize. Yet recurring food shortages and price volatility in the region stem primarily from the dominance o f maize together with its moisture sensitivity, consequently erratic production. Drought tolerant staples such as cassava, sweet potatoes, sorghum and millet offer important food security shock absorbers during drought years. The roots and tubers, inparticular, also offer higher calorie productivity per unit o f labor and landthan does maize. Yet research and extension budgets for these secondary food crops languish in many countries. Livestock and the feed industry also offer buffering capacity in the presence o f maize production and price shocks. Partial substitution o f cassava for maize in feed rations (as inEurope) can help to cushion maize supply shortages during drought years. 29. Ultimately, increased reliance on cross-border trade requires trust. Governments must have confidence that competitive markets will deliver food staples at low cost and in adequate quantity. And traders must trust that governments will not change policies unexpectedly - by announcing public imports o f a certain amount and then failing to bring in the full amount or by unexpectedly releasing public stocks at below-market prices and thereby imposing commercial losses on firms who import food inresponse to a drought. Open lines o f communication between governments and traders, transparency and predictability in policy making, and investments in reliable, accessible market information - on production, stocks, prices and quantities traded -provide the necessary underpinnings o f growing regional trade infood staples. xii 1.INTRODUCTION 1.1.Regionaltrade andagriculturalgrowth: 30. Over the next generation, growing trade in food staples appears poised to dwarf that in all other African agricultural markets. Currently, the market value o f Africa's food staples amounts to about $50 billion per year, or nearly three- fourths o f the value o f all agricultural production (Table 1). Given growing urbanization and the highest rates o f poverty in the world, Africa's market demand for food staples will grow dramatically in coming decades. As a result, production o f food staples -- for growing urban markets and food-deficit rural areas -- represents probably the largest growth opportunity available to African farmers. Facilitating expansion o f these markets will, therefore, be critical for efforts at stimulating agricultural production, broad-based income growth and poverty reduction. 31. A patchwork o f surplus food production zones across Africa supply the growing food markets in Africa's large cities and in deficit rural areas. In many cases, surplus food-producing regions emerge in areas o f favorable soils and rainfall. In other instances, regular food surpluses emerge in flexible ecosystems that combine the production o f multiple staples, particularly cereals in combination with perennial foodcrops such as bananas, cassava or root crops. Because farmers can harvest perennial foodcrops such as banana and cassava any time o f year and over multiple years, they are able to release cereals for sale as a cash crop in domestic or regional markets. 32. Key food surplus zones in Eastern and Southern Africa lie in South Africa, northern Mozambique, southern Tanzania and Uganda, and to a lesser extent in northern Zambia and northern Tanzania. InSouth Africa, mechanization, modem input use and increasing irrigation enable cereal export northward inmost harvest seasons. Innorthern Mozambique, cassava provides local food security, enabling regular exports o f maize to Malawi and in some season to Eastern Zambia. In Tanzania, a favorable climate and a blendo f food staples including rice, cassava, banana and maize enable regular cereal exports both north into Kenya and south into Malawi, into DRC and sometimes into Zambia. InUganda, favorable rains and multiple staples such as banana and cassava ensure food security, thereby enabling regular maize exports to Kenya. 33. For historical reasons, these surplus food production zones frequently lie across national borders from the markets they serve. Africa's political boundaries, drawn in Berlin in 1885, cut across natural market sheds, impeding the free flow of people and goods. As a result, political borders frequently separate surplus food production zones from the deficit markets they would normally serve. They separate food surplus northern Mozambique and southern Tanzania from deficit markets in Malawi and eastern Zambia. They cut off surplus zones in Uganda and northern Tanzania from deficit markets inKenya. And they separate surplus 1 cassava and maize producing areas o f northern Zambia from the deficit mining towns o f Katanga and Kasai provinces inthe DRC. Table 1. Size o f Agricultural Markets inSub-SaharanAfrica, circa 2000 Value ($US billions) Percent Exports out o fAfrica traditional 8.6 13% nontraditional 6.1 9% other 1.9 3% Intra-Africa trade domestic food staples 49.7 73% other 1.9 3% Total 68.2 100% Source: Diao andHazel1(2004). 34. Political borders translate into a welter o f tariffs, export restrictions and other man-made impediments to cross-border trade in food staples. In turn, these impediments to trade raise costs and lower incentives to both farmers and traders while at the same time artificially raising consumer food prices in cross-border deficit zones. Without access to regional export markets, production surges in thinly traded national markets lead easily to price collapses, which in turn risk stalling production growth and private investment in agriculture. Therefore, in order to maintain producer incentives, farmers in Africa's many surplus food production zones require regular access to growing food markets, both internal and external. 1.2. Regional Trade and Food Security: 35. Inaddition to helping maintain producer incentives in surplus zones, cross-border trade can help to moderate price volatility and food supply shortages in deficit areas. Cross-border trade flows can potentially contribute to reduced price volatility in food staple markets (see, for example, T i m e r , Falcon and Pearson, 1983; Koester, 1986; Dorosh, 2001). The import parityprice sets an upper bound, while export parity sets a floor below which prices will not fall, provided governments allow food to flow freely across their borders (Figure 1). 36. Maize remains the most important food staple in Eastern and Southern Africa. Yet dependence on rainfed maize production leads to highly volatile output from one year to the next, throughout much o f the region. Thus, a key food security challenge becomes one o f maintaining calorie consumption o f vulnerable consumers inthe face o fwide swings inmaize production, availability andprices. 2 Regional trade in food staples offers one tool for moderating national maize production and price shocks. Evaluating the potential magnitude o f this buffering capacity constitutes the focus o f the empirical analysis inthis paper. Figure 1. Trends in Domestic and Border Prices for White Maize, Randfontein, South Africa (prices inU S dollars per ton) Source: Traub (2008). 1.3. Objectives: 37. This paper examines the impact o f regional trade in food staples, both for moderating price spikes in deficit areas and for maintaining farmer incentives in surplus food production zones. The paper begins by identifying the geographic extent o f major maize market sheds in Eastern and Southern Africa. In order to lay the ground for subsequent empirical analysis o f these markets, the paper evaluates the approximate magnitude o f quantities produced, consumed and traded within key market sheds. 38. The paper then focuses on South Eastern Africa, on the market shed centered in Malawi, Northern Mozambique, Southern Tanzania and Zambia. Analysis focuses on the region's two most important food staples, maize and cassava. Giventhe volatility o f the region's rainfed maize production, this section aims to empirically evaluate the impact o f maize production shocks on staple food prices, production incentives, consumption and trade. To do so, the paper develops a spatially disaggregated model o f maize and cassava markets in order to evaluate the impact o f supply shocks, with and without cross-border trade. 3 2. MAIZE MARKET SHEDS INEASTERNAND SOUTHERNAFRICA 2.1. An Overviewof RegionalMarket Sheds: 39. White maize, the most widely consumed and traded food commodity in Eastern and Southern Africa, remains the central focus o f food policy debates in the region. Therefore, even a rudimentaryknowledge o f existing maize trade patterns and regional complementarities will prove useful for defining the geographic scope o f market basins to be used in a formal analysis o f trade flows and their impact on prices and consumption and production incentives. This chapter summarizes what maize traders and other market watchers have observed over the past decade. Although the volumes and possibly even directions o f trade flows may differ across seasons and years, as production and policy shifts occur, traders and other market observers typically identify three broad market sheds moving from south to north along the eastern halfo fthe African continent (Figure 2). The following discussion examines each inturn. 2.1.I.Southern Mozambique Market Shed 40. Despite reliable maize surpluses produced in northern Mozambique, and chronic maize deficits in the south, maize does not flow in appreciable volumes from northern to southern Mozambique. Instead, the surplus maize from northern Mozambique supplies nearby deficit markets inMalawi and sometimes ineastern Zambia. Long distances between northern and southern Mozambique, long wait times at the Zambezi River ferry, and high transport costs make it cheaper for Maputo-based maize mills in the south to source their maize from South Africa rather than internally from northern Mozambique. Hence, Mozambique straddles two distinct maize market sheds, with the Zambezi River providing the demarcationline between the two market basins. 41, The maindeficit markets insouthern Mozambique lie inthe far south, inthe cities o f Maputo, Matola and Xai Xai and to a lesser extent in the central coast city o f Beira. Domestic surpluses from the central interior areas o f Manica Province supply Beira and provide seasonal supplies o f white maize to the major urban centers inthe south. Regular supplies to the southern cities also come from South Africa, where silos house grain within about 250 kilometers o f the Mozambican border along rail lines that permit transport into the major mills o f Maputo at roughly $20 per ton. Incontrast, maize sourced from Northern Mozambique must transit 1,500 kilometers to reach Maputo. Given the long wait at the Zambezi River ferry crossing and the poor state o f Mozambican roads, the trip typically requires three days and costs about $100 per ton for transport alone (SIMA, 2008; Cruz et al., 2007). For reasons o f speed, convenience and cost, Maputo mills prefer to source grain from South Africa. In recent years, official imports from South Africa into southern Mozambique have averaged roughly 210,000 tons per year (Table 2). 4 Figure 2. Maize Market Sheds inEastern and Southern Africa Source: Govereh et al. (2008). 5 Table 2. QuantityFlows inMajor Maize Market Sheds (Averages 2000 to 2004, `000 tons)**** FormalTrade InformalTrade*** Total Production Consumption Imports Exports Imports Exports NetImports EastAfiica Kenya 2,442 2,7 16 216 11 197 5 398 Uganda 1,209 757 31 31 0 147 -147 NorthernTanzania 916 1,001 73 27 5 85 -35 Total 4,566 4,473 321 69 202 237 216 SouthEastAfrica SouthernTanzania 1,83 1 1,001 0 54 2 38 -90 NorthernMozambique* 813 699 0 4 0 136 -140 Malawi 1,873 1,518 95 18 104 0 181 Zambia 868 1,261 59 58 8 13 -4 Total 5,385 4,480 154 134 114 187 -53 SouthernMozambique Center** 271 151 0 0 0 0 0 south 148 226 212 0 0 0 212 Total 419 377 212 0 0 0 212 *** IncludesTete Province *** Excludes Tete Province ForMalawi,NorthemMozambiqueand SouthernTanzania, includescroppingyears 2000/01to 2002/03, or marketing **** This table omits years 2001/02 to 2003/04. All other informaltrade flows are averagedfrom2004 to 2007. seeds, feeds, losses and netstock changes. InMalawi, for example, seeds, feeds andlosses averaged495 thousandtons per year duringthis period. Source: FAOSTAT,RATIN, FEWSNET, Whiteside(2003). 2.1.2.South East Market Shed 42. Northern Mozambique and southern Tanzania produce the largest maize surpluses insidethe South East Africa market shed (Table 2). Inaddition, a cohort o f large andsmall commercial farmers inCentralZambia produces marketed volumes that are typically sufficient to supply Zambia's urban markets and in most years exports to the DRC. These surplus maize production zones serve a range o f deficit markets within the market shed. 43. Historically, Malawi has constituted the core net deficit market in the region.' Duringthe first half o f the ~OOO'S, annual inflows o f white maize into Malawi averaged about 200,000 tons, with slightly over half coming from informal2trade (Table 2). Deficit households, in both rural and urban Malawi, have relied on Land-scarce Malawi, historically dependent on maize imports; has produced large maize surpluses during the past two agricultural seasons, as a result o f large-scale distribution o f subsidized maize input packages. As a result, official estimates suggestthat production has surged to over 3 million metric tons, fromprior norms o f2 million tons, although it i s unclear ifthese subsidiesare sustainable over the long run. T h i s paper considers formal trade to be the quantities officially registered incustoms declarations and reported inofficial trade statistics. Typically, these move by truck and rail. Incontrast, informal trade remains unrecorded by customs officials. Normally, informal trade flows cross the border insmall lots, on bicycles, boats or headloads, often outside of normal customs hours or facilities. 6 maize imports from South Africa, northern Mozambique and southern Tanzania. Inaddition to supplying Malawi, the large maize surpluses produced inNorthern Mozambique provision urbanNampula and other cities innorthern Mozambique. Lusaka and the Copperbelt mining towns inZambia rely on primarily on roughly 500,000 tons o f maize supplied by large and small commercial farms in Central and North Central Zambia (Haggblade and Tschirley, 2007). Typically, these same farms supply Lubumbashi and other towns on the DRC side o f the Copperbelt with about 25,000 tons (Govereh, Chapoto and Jayne, 2008). In drought years, government and private traders supplement domestic production with imports from South Africa and Southern Tanzania. Lubumbashi, a city of two million people inthe DRC Copperbelt, imports roughly 70,000 tons o f maize annually from Zambia, almost all o f it in the form o f maize meal (Mwale, 2008). Nearby Zambian suppliers export to Lubumbashi when commercial supplies and government export policy permits. When Zambian production falls short or its government imposes export bans, as they have in recent years, Lubumbashi traders andmillers import maize andmaize meal from South Africa. 2.I.3. East Africa Market Shed 44. Regularly deficit markets in Kenya provide the center o f gravity for the East Africa market shed, pulling in surplus maize from Kenya's own central highlands as well as from Uganda and northern Tanzania. Given consistent informal imports from Uganda, in the range o f 100,000 tons per year, official trade statistics understate Kenya's import dependence as well as Uganda's status as a consistent net exporter (Table 2). Intermittently, in years o f exceptional maize harvest and low price in Ethiopia, such as 1997 and 2002, small quantities o f surplus maize flow from southern Ethiopia flow into northern Kenya, although poor roads and long distances limit these flows. The Ugandansurpluses likewise periodically find their way into deficit markets inRwanda, DRC and increasingly into Southern Sudan. 45. Although quantities vary from one year to the next, Table 2 provides a quantitative summary of these maize production, consumption and trade flows using average data from 2000 to 2004. As the Kenya data indicate, informal imports supply roughly half o f the country's 400,000 ton annual maize deficit. Most o f this comes from Uganda and Tanzania, which together supply roughly 250,000 tons o fmaize to Kenya each year. 2.I.4. Summing Up 46. Natural market flows cut across national boundaries. Surplus maize production from northern Mozambique serves deficit markets inMalawi and eastern Zambia. Meanwhile maize deficit cities in southern Mozambique source their maize supplies from across the border in South Africa. Tanzania similarly straddles two major maize market sheds. The maize surplus highlands o f southern Tanzania serve deficit markets in Malawi, DRC and intermittently innortheastern Zambia. Yet the northern highlands o f Tanzania supply deficit markets in Kenya. Major 7 mills in Dar es Salaam source a large portion o f their requirements by ocean freight, primarily from Durban. 47. These realities suggest that strategies for accelerating growth in surplus agricultural areas will require a regional perspective. If farmers and traders in surplus food production areas are to invest in soil fertility, productive assets and input supply and marketing systems, they will need reliable access to deficit markets, which inmany instances lie across national borders. Import bans, export bans and other trade controls, often intermittent and unpredictable, undercut incentives for farmers and traders and tend to undermineprospects for accelerated food production growth. .From the consumer's perspective, ensuring food security in deficit markets likewise requires a willingness and ability to source supplies from regional markets when these offer prospects for low-cost supply. The following discussion provides greater detail on the flows within these market sheds, on the resulting price patterns and on the policy stances that affect opportunities for cross-border maize trade. 2.2. SouthernMozambiqueMarket Shed: 2.2.1,Marketing 48. In the southern half o f Mozambique, below the Zambezi River, farmers in the central interior region o f Manica Province produce the most reliable maize surpluses. From Manica, along the Zimbabwe border in the central interior o f Mozambique, a network o f small. and large traders transport maize surpluses to markets and mills in the coastal cities, particularly in the far south (Figure 3). They also serve deficit households in rural areas. In fact, about 70% o f rural households incentral and southern Mozambique are net buyers o f maize, resulting intotal rural market demand for maize roughly equal to that inurban areas. As a result, production remains concentrated, not only geographically but also among a small group o f commercial smallholder farmers. Less than 5% o f maize producers account for over 50% o f production and over 70% o f sales (Tschirley and Abdula, 2007). Unit marketing costs are high, quality is generally poor, and traders find it difficult to source reliable supplies for large buyers, especially in the South. As a result, the largest millers inthe country, located in Maputo, rely almost exclusively on maize grain imported from South Africa. 49. Urbanpopulation, at about 35% o f national total, i s growing rapidly. At current rates, the urban population share will reach 50% by 2020. As a result, under likely scenarios, urban demand for maize will double over the next decade. Growing urban demand represents a huge growth opportunity for Mozambican farmers. Yet the growth in demand could easily be satisfied by imports from South Africa, if productivity in production and marketing in Mozambique does not improve. 8 50. Maize shares in total food expenditure in urban Maputo province are 2.4%, compared to 7.4% for rice and 15.5% for wheat. The maize share rises outside o f Maputo, to 14.5% inother southern provinces and 40% inthe Center. 51. InMaputo, about two-thirds o f consumers primarily purchase refined maize meal, with the remainder purchasing grain for service milling. Outside o f urban Maputo, those shares are reversed. The low share o f consumers in Maputo relying on maize grain likely stems from the low price3 and widespread availability o f rice and the resulting very low consumer expenditure share on maize (Table 3), especially among higher income consumers. The buying habits of low-income consumers also contribute to preferences for refined meal. They tend to buy very small quantities at a time, making service milling at hammer mills less desirable. Figure 3. Southern Mozambique Market Shed Untilthis year. 9 2.2.2. Prices 52. Maize meal prices are extremely high in Mozambique. The leading brand cost about US$680 per ton in early 2007, while the cheapest was about US$440. Maize grain at retail cost about US$270/mt during the same period in Maputo. These prices compare to a range o f US$250-US$330 for comparable meals in Zambia, and grain prices o f US$190. This wide differential between grain and meal prices in Mozambique may be related to the structure o f the industry. The largest miller in the country holds roughly an 80% share o f the Maputo market, and the closest competitor holds nearly all the remaining 20%. Both also sell in major cities and rural areas throughout the country. A 20% duty on imported maize meal reduces prospects for competition from South African millers. At least four new millers have entered the market over the past six years, but they operate much smaller milling capacity. In southern Mozambique, they hold a very small market share and exert no appreciable effect on prices charged by the leading millers. 53. The small-scale milling sector provides little competition for industrial milling in the south. After booming inthe early 1990s, the sector began to decline with the end o f large food aid arrivals after 1993, and by 2003 it was difficult to find hammer mills operating in the city. The decline o f this sector in the south has resultedprimarilyfrom a lack o f demand for its services, and much less to policy. Medium-scale millers inthe Center and South rely primarily on local production, butholdvery small market shares. 54. Maize meal price in Maputo broadly track SAFEX grain prices, on which the large millers depend for grain supplies. Over the seven year period from 2000 to 2006, SAFEX white maize prices averaged $136 per ton, while wholesale white maize inthe informal Maputo market sold for $204 per ton. Second grade maize meal, in contrast, retailed for $420 per ton, roughly double the price o f wholesale maize. Converting retail maize meal prices back to implicit wholesale levels4, Figure 4 tracks domestic grain-equivalent wholesale prices in relation to import and export parityin South Africa. The large mills that produce maize meal for the Maputo market tend to price their meal inreference to SFEX, sometimes with a lag. On average, over the seven year period depicted below, import parity and implicit large mill retail pricing have equaled each other, at just over $200 per ton implicit wholesale price for maize grain. During Mozambique's domestic maize shortages o f late 2005 and early 2006, the SAFEX price appears to have moderated maize meal prices coming out o f the large mills. Meal prices out o f the large mills followed SAFEX price downwards, with a several month lag, and remained at SAFEX levels during the worst o f the lean season. In contrast, the thinly traded whole grain on the informal wholesale market saw prices spike well above import parity for several months. Given low budget shares for whole grain and service milling, and given the disincentives for small-lot maize imports from South Africa, traders evidently felt they could not import profitably in the short 4Inthis conversionwe have usedthe long-termaverage retail to wholesale ratio of420/204 =2.11. 10 window available. Indeed, wholesale maize prices on the informal market fell back below import parity within a few months. 2.2.3.Policy Issues 55. Unlike many other countries in the region, Mozambique has consistently retained an open border policy on maize trade. Since the end o f the civil war in 1992, Mozambique has freely allowed maize imports and exports. This enables the large deficits in the southern cities to be met by large millers who import maize grain from South Africa and mill it for sale locally. The open border policy likewise enables farmers in the surplus zones o f northern Mozambique to export to Malawi and Eastern Zambia as market conditions permit. Table 3. Food Expenditure Shares inSouthern and Central Mozambique Commoditv Share of FoodExDenditure(%) Maize Rice Wheat Cassava Urbanareas Maputo Province* 2.4 7.8 15.5 1.3 Gaza and InhambaneProvinces** 14.5 9.8 6.0 5.2 Manica and Tete Provinces*** 39.9 4.4 2.9 0.5 RuralAreas MaputoProvince* 9.1 11.4 7.4 4.7 Gaza and Inhambane Provinces** 12.2 9.5 3.2 8.4 ManicaandTete Provinces*** 48.0 2.5 1.4 0.5 ***Southernmost province. Southcoast and interior south *** Upperinterior Source:IAF2002, as reportedbyTshirley andAbdula (2007) 56. Although Mozambique does not impose quotas or bans on cross-border maize trade, they do impose a 17% VAT on imported maize, although not on rice or wheat. Maize meal i s exempt but maize grain i s not, meaning that grain imported for sale as grain must pay the VAT, while grain imported for meal receives a reimbursement. Thus, in principle, the application o f the VAT favors rice and wheat relative to maize, favors the availability o f maize meal over maize grain at retail, and favors large industrial millers over smaller traders and hammer millers. 57. Inpractice, imports o f maize for sale as grain have not occurred despite several prolonged periods where such imports would have beenprofitable (Figure 4). We attribute the absence o f imports by small traders to complexities in import 11 procedures and to the high degree o f formality and large scale o f the South African maize marketing system. The lack of imports by larger scale formal traders i s due to a combination of factors: consumers in Maputo have, until the past year, had access to a low cost option inrice, they spend very little on maize, and most o f them are therefore willing to pay the highpremium for refinedmeals on the small quantities they buy. Figure4. Domestic andImport ParityPrices inMaputo 400 I 300 n 3 @FI a 200 *2 a, PI 100 0 1111 I1 I 1 I 1 1 1I 1 I 1 1 1I 1 I1 I 1I 1 I / I 1 / I I / I 1 I 1 1 1I I I / I I I 1 I / 1 1I 1 I 1 I 1 / I I / I 1 I / I 1 1 1/ I / I I 1 I I I I I1 I Maputo infbrmalwholesale gram -Maputo largemill Importparity, SouthM c a - - - Exportparity, SouthAfiica Source: SIMA. 12 2.3. The South-EastMarketShed: 2.3.1.Marketing 58. Maize surpluses from northern Mozambique serve deficit urban markets in the cities o f Nampula and Nacala as well as intermittently significant deficits in Malawi. The highly productive highlands o f southern Tanzania supplement these flows by directing surpluses into Malawi, DRC and sometimes into Zambia. In normal years, flows from northern Mozambique into Malawi range between 70,000 and 140,000 tons while exports from southern Tanzania reach about 90,000 tons, the majority o f that flowing into northern Malawi, with the remainder directed into DRC and sometimes into northern Zambia (see Table 2, Whiteside, 2003 and FEWSNET). In some years, smaller flows travel from Tete Province inMozambique into eastern Zambia (Figure 5). 59. Duringdrought years in Malawi, these inflows increase substantially, as they did most notably during the Malawian food crisis o f 2002 and early 2003 (Whiteside, 2003; Tschirley, 2006). During this period, informal traders from Northern Mozambique, delivered on the order o f 200,000 to 300,000 tons o f maize to Malawi, while formal imports from Northern Mozambique raised that total to about 500,000 tons o f maizeq5 Farmers from Southern Tanzania fbrnished roughly a fbrther 140,000 t o m 6 At these levels, total informal imports amounted to between 20% and25% o fnormal maize consumption inMalawi. 60. Both northern Zambia and southern Tanzania supply maize to DRC. Large and small-scale commercial farms in central Zambia likewise supply Lusaka and in the Copperbelt. The southern Tanzanian highlands direct some surplus maize into Dares Salaam, though most ofthe southern highlands more easily serve southerly andwesterly markets inMalawi, Zambia and DRC (Ashimogo, 2008). 61. In years when Malawi and Zambia face large maize deficits, South Africa supplements these more localized trade flows by supplying white maize to urban centers in central Zambia and Malawi. Following the liberalization o f maize markets in South Africa in 1996, the South African Futures Exchange (SAFEX) added an Agricultural Markets Division trading maize, soya, sunflower and wheat futures. In 1999, SAFEX added options on maize fbtures contracts. Given South Africa's status as a reliable exporter in regional maize markets, the SAFEX spot and fbtures prices have become the standard regional benchmarks for pricing white maize duringdeficit years. This estimate includes 305,000 tons of officially recordedtrade, mostly from Tete Province of Mozambique, plus 223,000 tons o funrecordedtrade, coming mostly from Zambezia Province o f Mozambique into Malawi (Whiteside, 2003, p.56). Tanzanian maize exports into Malawi during the 2002/03 marketing year (the 2001/02 production year) included 112,000 tons of institutional deliveries by the World FoodProgramme andby Tanzania's National Food Reserve Agency plus 22,000 tons o f recorded commercial imports and 7,000 tons o funrecorded, informal trade (Whiteside, 2003, p.32). 13 Figure 5. South East Africa Market Shed 62. Local and regional food aid procurement (LRP)has likewise helped to encourage regional cross border flows o f food staples. Between 2001 and 2005, the World Food Programme (WFP) tripled the value o f its purchases in Africa. As a result, the share o fLRPintotal WFP food aid to Africa rose to about 22%. Ofthis total, nearly 60% has come from South Africa, although Zambia and Tanzania have also provided maize supplies to WFP in some years (Tschirley and del Castillo, 2007). During the Malawian shortfall o f 2001/02, for example, WFP shipped 57,000 tons o f Tanzanian white maize into Malawi (Whiteside, 2003, p.32). These institutional flows complement the growing flows o f commercial cross- border grain flows within the South East African market shed. 2.3.2. Prices 63. Until very recently, Malawi has formed the core deficit market in the South East African market shed. During the seven year period from 2000 through 2006, wholesale maize prices in Lunzu Market, near Blantye in Southern Malawi, 14 averaged $166 per ton, while those in the interior markets o f Northern Mozambique and inthe Southern Tanzanian highlands averaged $120 (Ashimogo, 2008; Phiri, 2008). Because o f its proximity, Northern Mozambique has a slight cost advantage in supplying the Malawian market. In 2001, 2002 and 2005, imports from Northern Mozambique were highly profitable, and import parity prices from Northern Mozambique helped to cap prices in southern Malawi (Figure 6). Followingthe Malawian government offloading their excess stocks in early 2003, and the subsequent price fall, exports from Northern Mozambique became unprofitable, as they did again in 2006, following widespread input distribution and a bumper harvest inMalawi. It remains to be seen how continued input subsidies in Malawi over the past several seasons will affect planting decisions innorthern Mozambique and southern Tanzania. Alternative cash crops may become more attractive to Mozambican farmers if Malawian maize harvests continue to be strong or if subsidized government sales depress market prices, as in2003 (Whitehead, 2003). 2.3.3.Policy Issues 2.3.3.1.Import and Export Quotas and Bans. 64. Mozambique and DRC, unlike the other countries in this region, allow the free flow o f maize across their borders. Because Northern Mozambique is typically maize surplus, and because Malawian markets offer better prices than Southern Mozambique (because o f the longer distance and higher transport cost to Maputo), traders in Northern Mozambique routinely look to Malawi to market their surplus grain. Because southern Mozambique requires imports, the Mozambican government has an interest in allowing imports, to serve consumers in the south, and exports, to provide price support for surplus farmers in the North. The DRC, a net deficit country, similarly benefits from open borders. 65. In contrast, Zambia, Tanzania and Malawi carefully control formal cross-border maize trade. They require export and import permits, and periodically they impose bans7. Typically, though not always, they ban imports during good harvest years and exports during poor harvest years.8 Traders and farmers complain about the unpredictability o f these quotas, and considerable evidence suggests that unpredictability concerning government import and export quotas -- coupled with uncertainty about government's own import, sales and pricing decisions -- reduces private trader willingness to engage in formal cross-border maize trade (Mwanaumo et al., 2005; Nijhoff et al., 2003; Ashimogo, 2008). 'See Phiri (2008) on Malawi, Ashimogo (2008) on Tanzania and Govereh, Chapoto and Jayne (2008) on Zambia. A contrary instance occurred following Zambia's good maize harvest o f2006, when government initially banned exports inorder to buildup national stocks for the government FoodReserve Agency (FRA) (Dorosh, Dradri and Haggblade, 2007). 15 Figure6. Domestic andImport Prices inSouthern Malawi 500 sF18400 e 5 a 300 0 0 % s 200 s a 8 loo 0 Import ParityNorthernMozambique -ExportParity, - - - 8 NorthernMozambique SouthernMalawi Source: Annexes 1'3.5. 2.3.3.2. Government Price Subsidies. 66. Farm price subsidies. The Zambian, Malawian and Tanzanian governments all procure grain for a public food reserve agency. Inall three countries, the role o f public food agencies has diminished from highly interventionist levels in the late 1980's and early 1990's. In Tanzania, the least interventionist of the three, government purchased an average of 2.6% o f total maize production during the 1 9 9 0 ' ~with the share purchased by their national Strategic Grain Reserve (SGR) ~ falling to 1.6% in the years since 2000 (Ashimogo, 2008). In Zambia, the dissolution o f the National Maize Marketing Board (NAMBOARD), in the early 1 9 9 0 ' ~launched a decade o f negligible government procurement. Not until the ~ mid-2000's did the newly reconstituted Food Reserve Agency (FRA) begin purchasing significant quantities o f maize, amounting to about 10% o f smallholder maize production by 2004 (Govereh, J a p e and Chapoto, 2008). The Malawian government's Agricultural Marketing Development Corporation (ADMARC) has remained the most heavily interventionist in the region (Smale and Jayne, 2003). 16 67. In Zambia, as in Kenya, the government has typically paid above-market prices for the maize they purchase (Govereh, Chapoto and Jayne, 2008). This has led, in some years, to unusually large border area purchases by Zambia's Food Reserve Agency (Table 4 and Figure 7). Many market observers believe that high procurement prices in Zambia have attracted informal maize imports from southern Tanzania and from Tete Province in Mozambique (Whiteside, 2003; Times o f Zambia, 2006; Zambia Daily Mail, 2007). 68. Consumer price subsidies. To benefit urban consumers, and to dispose o f unwanted maize stocks, both Malawi and Zambia's food agencies have, at various times, sold public maize at below-market price in urban areas. Given the unpredictability of these subsidized sales, many traders decline to enter the import market duringdrought years, for fear o f being undercut by subsidized government grain (Mwanaumo et al., 2005; Whitehead, 2003; Tschirley et al., 2006). 69. On balance, most available evidence suggests that government quantity and price controls tend to discourage private cross-border trade, by introducing significant price risk and raising the cost o f doing business. Most market traders complain most o f all about the unpredictability o f government trade controls. A study o f Zambian maize markets indicates that several international grain trading companies exited the Zambian market after initial forays because o f the heavy risk imposed by unpredictable government interventions in the maize market (Nijhoff, 2003). 2.4. The East Africa Market Shed: 2.4.1. Marketflows 70. Surplus maize-producing areas inEast Africa include Uganda, northern Tanzania, andthe western highlandso fKenya. Intermittently, inyears of exceptional maize harvest and low price in Ethiopia, such as 1997 and 2002, small quantities o f surplus maize flow from southern Ethiopia flow into northern Kenya, although poor roads and long distances limit these flows (RATES, 2003; Awuor, 2007). 71. Major deficit areas emerge in the large urban areas o f Nairobi and Mombassa as well as in coastal Dar es Salaam. Smaller but growing volumes o f staple foods flow into deficit markets in Rwanda and southern Sudan. Recently, improvements in political stability in both northern Uganda and southern Sudan have led to substantial increases in staple food flows heading north from Uganda into southern Sudan. Though maize movements vary seasonally and involve movement inboth directions at given times o f year, the most prevalent flows are depicted inFigure 8. 17 Figure 7. Zones o f Heavy Border Area Procurement by the Zambian Food Reserve Agency, 2006 Source: FSR Table 4. High-Side Outliers Among FoodReserve Agency Purchases in2006 Province Marketed Volumes* FRAPurchases District 1999/2000 2002/03 2006 Eastern Chadiza 1,391 1,586 28,257 Nyimba 1,133 486 10,088 Northern Mbala 5,962 6,406 20,386 Nakonde 1,545 4,521 11,704 Total 10.031 12.999 70.435 * Farm household estimates of total volumes o f maize sold, including all sales to households, tradeers and to the FRA. Source: FSRP (2007). 18 72. Total maize exports from Uganda average about 180,000 tons per year (Table 2). About 140,000 o f these go to Kenya, while traders ship the remainder to Rwanda and other East African neighbors. Northern Tanzania supplies a further 110,000 tons o f maize to Kenya, about three-fourths o f this through informal channels. With annual maize exports o f about 250,000 tons flowing into Kenya, these two East African neighbors supply the majority o f Kenya's 400,000 ton annual maize deficit. 2.4.2.Prices 73. Kenya remains a deficit, high-cost maize market. Over the period between 2000 and 2006, the monthly Nairobi wholesale price for white maize has averaged $194 per ton, compared to $127 in Uganda at the Mbale market near the Busia border post, and $165 in Durban (Alisuma, 2008; Nyoro, 2008; Traub, 2008). Kenya's policy o f maintaining high maize prices, through high procurement prices and tariff protection, has favored farm groups as well as informal traders from Uganda. Even from 2005 onwards, when Kenya lowered its maize import tariff from 25% to 2.75% for E A C countries, they retained the 25% levy on sea shipments coming invia Mombasa. As a result, shipments from South Africa and the rest o fthe world continue to face a 25% entry fee (Nyoro, 2008). These duties on official imports have spurred the rise o f substantial informal maize imports from Uganda and Tanzania. At roughly 200,000 tons per year, these informal imports roughly match the magnitude o f official imports (Table 2). 74. Import parity calculations suggest that informal imports from Uganda have helped to cap Nairobi price increases intermittently over the past six years (Figure 9). Imports from Durbanremainmore competitive inMombasa, where they avoid the Mombasa to Nairobi transit and handlingcosts o f roughly $50 per ton. Given the high cost o f maize in Kenya, farmers there remain generally uncompetitive in export markets. 2.4.3.Policy Issues 75. Chronic maize deficits inKenya, since the early 1 9 9 0 ' ~have provided the central ~ gravitational pull drawing inmaize from surrounding surplus areas inUganda and Tanzania (Magnay, 2004). For this reason, Kenya's trade policies become the most important governor o fregional maize flows. 76. Kenyan maize policy in recent decades has involved supporting high prices for farmers through high government procurement prices. Kenya remains the most highly interventionist government in the region, with National Cereal Produce Board (NCPB) purchases accounting for 11% o f national maize production during the 199Os, though falling to 7% during the 2000's (Nyoro, 2008). In contrast, Tanzania's Strategic Grain Reserve purchased only 1.6% o f national production since 2000, and Uganda does not operate a government maize procurement agency (Adamuga, 2008). 19 Figure8. East Africa Market Shed / SOMALIA 77. Because Ugandan farmers can produce maize at lower cost than Kenyan farmers, Kenya has historically protected its farmers by imposing high import tariffs on maize (Nyoro, Kirimi and Jape, 2004). This import tariff has fluctuated widely and unexpectedly over time, from a higho f 50% in 1999 to a low o f zero in2000. But during the second half o f the 1990's and the first half o f the 2000's, the import tariff averaged 25% (Ariga and Jape, 2007). Recent simulation results suggest that the Kenyan tariffs raised maize prices in Kenya between 4% and 17%, depending on the market and year (Jape, Myers and Nyoro, 2005). 78. Not surprisingly, these high import tariffs discouraged formal trade and instead diverted trade flows into informal channels. A series o f cross-border monitoring studies has documented the widespread practice o f trucking grain to the Uganda- Kenya border, then offloading and carrying bags on bicycles across the border to circumvent customs agents. The bicycle traders then bulkbags at assembly points across the border where Kenyan traders with trucks reload the bags and deliver them to deficit markets throughout Kenya (Ackello-Ogutu and Echesseh, 1997; RATES, 2003b). These added transaction costs clearly dampen incentives to trade. In spite of these impediments, informal flows from Uganda ranged between 40,000 and 80,000 tons per year between 2001 and 2004 (Magnay, 2004). 20 Figure9. Domestic and World Prices for White Maize Grain, Wholesale inNairobi 400 300 n @0 fA 5 200 -2 0 a 100 0 -Nairobi -- -- -- ' Importparity,Durban Exportparity, k b a n Importparity, Mbale Uganda 8Exportparity,Mbale Uganda Source: Aguma (2008), Nyoro (2008) and Traub (2008). 79. Since January o f 2005, with the institution o f the new East African Community (EAC) trade agreement, Kenya has lowered its import tariff from 25% to 2.75% on maize imports from the other member countries o f Uganda, Tazania, Rwanda and Burundi. This, together with COMESA efforts to harmonize food safety, phytosanitary standards and customs procedures, i s expected to facilitate cross border trade incoming years. 2.5. Conclusions: 80. Surplus food production zones often lie across national borders from the deficit markets they serve. Maize farmers inUganda and northern Tanzania serve deficit markets in Kenya. Farmers in northern Zambia serve deficit markets in the mining centers o f Katanga and Kasai Province in the Democratic Republic of Congo (DRC). Small producers innorthern Mozambique supplymaize to Malawi and intermittently to eastern Zambia. And surplus maize from the southern Tanzanian highlands flows regularly into Malawi and intermittently into northern andeastern Zambia. 21 81. As a result, natural markets sheds cut across national boundaries. Tanzania and Mozambique offer the clearest example o f this. While surplus producers in northern Tanzania serve deficit markets inKenya, those inthe southern highlands serve coastal markets and deficit areas to the south, primarily in Malawi. Mozambique i s likewise partitioned into two major market sheds. Despite regular maize surpluses in the north and in Tete Province in northern central Mozambique, the major deficit markets in the south rely primarily on imports from South Africa and to a lesser extent on seasonal surpluses from central Mozambique. 82. To maintain and sustain producer incentives, farmers in food surplus zones will needaccess to growing markets, bothinternal and across national borders. Inthin national markets, without export outlets, production surges lead easily to price collapses. In turn, these disincentives dampen long-term agricultural income growth. Therefore, failure to facilitate the expansion o f intra-regional trade in food staples risks stalling production growth andprivate investment in agriculture in these critically important high potential food production zones. Indeed, the policy-abetted Malawian price collapse o f 2003 left many Mozambican maize farmers without their normal export outlets, inducing them to consider alternative cash crops such as tobacco instead. This suggests that national maize production and marketing policy will be most productive when formulated in a regional perspective that recognizes and facilitates the development o f these external food markets. 83. Successfil expansion o f regional trade in food staples holds the potential to accelerate agricultural income growth in favorable areas while simultaneously diminishing price volatility and hunger in deficit zones. Thus, regional trade in food staples constitutes one key plank in an effective agricultural development and food security strategy for the region. The following discussion focuses in on one particular market shed to assess empirically the potential impact o f regional trade in moderating food supply shocks stemming from fluctuating maize production. 22 3. SOUTH EAST AFRICAN FOOD STAPLE ZONES 3.1. Key Food Staples: 84. This analysis focuses on cassava and maize, the two most important food staples insub-Saharan Africa, as well as inthe three core countrieswithinthe SouthEast Africa maize market shed. InZambia and Malawi, maize dominates the national food basket, providing just over half o f all calories consumed, while cassava provides roughly 10%. However, in Mozambique the rankings are reversed. Cassava serves as the number one food staple there, supplying about one-third o f total calories, while maize, the number two staple, furnishes about one-fourth (Table 5). Likewise in the cassava zones o f northern Zambia and the northern lakeshore region o f Malawi, cassava predominates. 85. Because o f its vulnerability to moisture stress, maize production varies significantly from one year to the next (Figure 10). Hence the importance o f a range o f drought-tolerant secondary food staples, such as sorghum and millet in the temperate zones and cassava in the tropical zones. Estimates o f cross price elasticities o f demand in South Africa, for example, indicate strong substitution effects between maize and other cereals (Alderman and del Nino, 2002). Empirical work in Mozambique similarly shows high levels o f cassava consumption as well as substitution between maize and cassava, in urban and rural areas, particularly inthe north and especially during drought years (Rosling, 1986; Tschirley and Abdula, 2007). In Malawi and Zambia, consumers in the northern parts o f both countries can, likewise, substitute cassava for maize in seasons and years when maize i s in short supply. Data from Zambia illustrate how cassava's well-deserved reputation for drought-resistance translates into much lower production volatility for cassava than maize (Figure 10). As a result, increased regional cassava production since the early 1990's provides a growing buffer against drought-induced volatility inrainfed maize production. 86. Because households in northern zones o f Mozambique, Malawi and Zambia consume both cassava and maize (see Figure 13), and because they can harvest cassava over several years, households can choose to consume more cassava and sell more maize during drought years, thus releasing maize for sale to deficit maize-belt households. In bad years, when nearby maize belt households face acute deficits, farmers from neighboring cassava and dual staple zones are able to harvest more o f their perennial cassava crop for home consumption and in turn free up more maize for export to deficit zones. These mixed and dual-staple zones, thus, serve as potentially important food security shock absorbers, enabling the release o f maize to deficit areas intimes o f short supply and thereby helping to moderate regional food shortages. Official estimates o f cassavaproduction are subject to a wide margino f error. Because cassava farmers maintain a portfolio o f cassavaplots o f differing maturity, total area cropped does not necessarily provide a reliable barometer o f total production. Andbecause farmers typically harvest their mature plots throughout the year, often weekly and insmall lots, yield estimates based on recall data are notoriously suspect. 23 Figure 10. Trends incassava and maize productioninZambia, Malawi and Mozambique 1 I 'g 2500 2000 9 0 8 1500 i. ,g E 1000 2 500 1 -Cassava -Maize I a. Production trends inZambia ;n' 8B 2500 2ooo b 4E 1500 3e 1000 a 500 -b,, , , , , , ,, I I 0 1 -Cassava -Maize 1 b.Production trends inMalawi -Cassava --Maize c. Production trends inMozambique Source: FAOSTAT. 24 Table 5. Average National Production and Consumption o f Cassava and Maize, 2001-2004 Cassava. fresh weight Maize Production (kgkapita) Malawi 138 158 Mozambique 314 64 Zambia 86 94 Consumption (glpersodday) Malawi 225 358 Mozambique 644 171 Zambia 231 351 Consumption (% kcal) Malawi 9% 53% Mozambique 34% 24% Zambia 13% 56% Source: FAOSTAT. 87. Production in the region was also far more covariant during the first period than the second (Table 6). From 1990 to 1999, correlation coefficients on production between South Africa, Zimbabwe, and Zambia were large, positive, and highly statistically significant. Duringthe second period they were much lower and none was significant. Correlations between those three countries and Mozambique and Malawi were small and insignificant during both periods, with one exception: a large, significant, and negative correlation between Mozambique and Zimbabwe during the second period. Mozambique's lack of correlation with other countries" is driven by the predominance o f the North innational production, and by the low correlation o f weather patterns inthis area with those inthe rest o f the region. For example, during the droughts of 1992 and 1995, production in northern Mozambique was largely unaffected. Since northern Mozambique regularly produces exportable maize surpluses, its lack o f correlation with production inthe region makes it a potentially important source o f supply for both commercial and humanitarian responses to drought. 88. Overall, regional production varies less than production in each individual country. While the coefficient o f variation (CV) o f total regional production was 25% from 1990 to 2003, CVs inindividual countries ranged from a low o f 28% in South Africa to highs o f 46% in Zimbabwe and 48% in Mozambique.' This suggests that, despite positive and large correlations in production across countries, there will be scope for intra-regional trade to cover some portion o f ~ loThe negative correlation with Zimbabwe i s a special case, drivenby the economic chaos inZimbabwe contrasted with recovery from the civil war inMozambique. 'Mozambique's high variability is due primarily to steady increases inproduction since the drought and the endingo f the civil war in 1992. The CV of production around a linear trend was only 19%, compared to 48% absolute variation over the same period. 25 national and sub-regional shortfalls in all but the worst drought years, such as 1992 (Tschirley et al., 2004). Since 1992, differences inproduction outcomes and the buffering capacity o f the cassava belt zones have induced cross-border flows into the most heavily affected areas, particularly into Malawi, duringdeficit years. Figure 11.MaizeProduction Variability inSouthern Africa, 1990-2005 20,000,000 7 15,000,000 10,000,000 5,000,000 0 -5,000,000 10,000,000 Harvest Year Note: CoversMozambique, SouthAfrica, Swaziland, Zambia, Zimbabwe, andMalawi. Source:FAOSTAT Source: Tschirley and J a p e (2007). 26 Table 6. Correlation Coefficients o f Maize Production among Selected Southern African Countries, 1990-2005 SouthAfrica Zambia Zimbabwe Mozambique Malawi SouthAfrica 1990- 1999 0.66 0.93 0.18 0.12 0.36 0.51 0.04 -0.18 1990- 1999 1996- 2005 Zambia 0.66 0.77 -0.04 0.36 1996 2005 - 0.36 0.27 -0.08 0.06 Zimbabwe 1990- 1999 0.93 0.77 0.30 0.22 0.05 0.27 -0.88 0.21 Mozambique 1990- 1999 1996- 2005 0.18 -0.04 -0.30 1996- 2005 0.65 0.04 -0.08 -0.88 -0.20 Malawi 1990 1999 - 0.12 0.36 0.22 0.65 1996- 2005 -0.18 0.06 0.21 -0.20 Note: coefficients inbold are statistically significant at the 5% level. Source: Tschirley and Jayne (2007). 3.3. Mapping Food StapleZones" : 3.3.1. Motivation 89. Because existing market sheds do not coincide with national boundaries, our analysis requires subnational spatial disaggregation o f food production and consumption patterns. Given regional differences in weather, soils and crop suitability, in tastes, consumption patterns, food substitutability and prices, we first defined homogeneous clusters according to the composition of staple food production and consumption. Using household-level survey data, we classify districts in each of the three countries into one o f five homogeneous food consumption units: maize belt, maize-dominant mixed zones, dual staple zones, cassava-dominant mixed zones and cassava belt. The ensuing discussion describes how we have defined each of these food staple zones according to the relative importance o fmaize and cassava inregional food production. *'For thls purpose, we obtained nationally representative farm household survey data for each o f the three countries for representative recent years. For Zambia, we have used the latest supplemental survey to the Central Statistical Office's Post-Harvest survey, the 2004 supplemental survey covering the crop year 2002/03. For Malawi, we have used the integratedrural household survey o f 2004105 andinMozambique we have use the 2005/06 Ministryo f Agriculture nationalhousehold survey files. 27 Figure 12.Spatial DistributionofPopulation andUrbanAreas a. Population distribution b. Urbanproximity zones 28 3.3.2. Spatial Dispersion of Population 90. As Figure 12 indicates, population is dispersed very unevenly across the region. Roughly 40% o f Zambia's 11 million people live in urban areas, either in the mining towns o f the Copperbelt or along the line o f rail. Malawi, however, squeezes a slightly larger population, o f over 12 million people, into a small space, leading to much heavier population density. Given that less than 20% o f Malawi's population lives in urban townships, Malawi consists primarily o f densely settled rural areas. Mozambique, the largest o f the three countries geographically, also houses more people, roughly 19 million in 2005, with slightly over 60% living in rural areas. The northern districts o f Mozambique nearly rival Malawi inpopulation density, as does the central and southern coastal corridor (Figure 12). 3.3.3. Percent of households growing each crop 91. To provide an initial feel for the spatial distribution o f food production and consumption, we have mapped the percentage o f households growing each staple food. Using progressively darker shades o f yellow (for maize) and blue (for cassava), the maps in Figure 13 classify districts according to the percentage o f households growing each o f these two staple food crops. While maize predominates in the central part o f the region, cassava production becomes significant in northern areas o f all three countries as well as along the Mozambican coast. 3.3.4. Production Quantities 92. The absolute level o f food production depends not only on the percentage o f households growing the crop but also on their absolute numbers. The population distribution, from Figure 12, combined with the production patterns, from Figure 13, translate into spatial distribution o f staple food production as described in Figure 14 below. This figure clearly shows the dense concentration o f maize production in southern Malawi, Eastern Province o f Zambia, along central Zambia's north-south line o f rail, in the central interior o f Mozambique, and scattered throughout the cassavazone o f northern Mozambique. 93. Production data, though the most useful for measuring levels o f importance among food staples, are also the least reliable o f the three measures o f staple food prevalence: percentage o f households growing each crop, hectares per household and kilograms produced. Production data, particularly for cassava, are subject to large measurement error because farm households typically harvest cassava year- round and over a period o f several years. Aggregating up to annual production from daily small baskets produces wide variation inestimated cassava output. 29 Figure 13. Percentages o f Households Growing Maize and Cassava in Zambia, Malawi and Mozambique a. Regional maize production b. Regional cassavaproduction Source: produced from national farm householdsurveys ineach country. 30 Figure 14. Spatial Distribution o f Maize and CassavaProduction Source: produced from national farm household surveys ineach country. 3.3.5. CroppedArea per Household 94. Area data prove more reliable than production, although even here confusion may arise since some surveys and some farmers report only area under mature cassava (those fields with plants over one year old and hence potentially available for harvest) while others report total areaplanted incassava, including freshly planted first-year plots. Cropped area per household in each crop i s displayed in Figure 15. 3.3.6.DelineatingFood Staple Zones 95. Because o f the unreliability o f district-level production data, particularly for cassava, we have focused on measures o f relative staple food prevalence that relied on either area planted or on percentage o f households growing each crop. The percent o f households offers the simplest measure o f crop prevalence. It likewise has the benefit o f enabling researchers to classify new regions quickly, based on fairly inexpensive rapid reconnaissance visits. 31 Figure 15. Area per HouseholdPlantedinMaize and Cassava a. Maize area planted b. Cassavaareaplanted Source: produced from national farm householdsurveys ineach country. 32 96. Area-based classification requires more careful, structured survey data. While area estimates based on farmer recall are subject to considerable imprecision, we consider them a more accurate measure o f cassava stocks available in the ground than the estimates o f harvested quantities derived from farmer recall o f daily harvests that are then aggregated to an annual figure. 97. The two alternative measures - percent o f households and area per household in each crop -- result in broadly similar zonal classifications (Figure 16). Because the hectare-based measure offers a better sense o f relative proportions of each crop, we have adopted the food staple zones as defined inFigure 16a for purposes o f the following analysis. In this case, we have enjoyed access to detailed household survey data for each country. Infuture work, this may not be the case. So future circumstances may dictate reverting to zonal classification based on household percentages. For now, we have adopted the relative hectares under each crop as our classification tool. Table 7. Alternative Criteria for Defining Food Staple Zones Food Staple Zone a. Percent o f Households b. Ratio o f Cassava Hectares GrowingEach Crop to Maize Hectares (C/M) Maize Cassava Cassavabelt <25% >75% > 10 Cassavamixed 25-50% >75% 2 to10 Dual staple > 50% >50% 0.5 to 1.9 Maize mixed >75% 25-50% 0.1 to 0.49 Maize belt >75% <25% <0.1 33 Figure 16. Food Staples Zones a. Defined by relative area planted to maize and cassava b.Definedbypercent ofhouseholds growing maize and cassava Source: produced from national farm household surveys ineach country. 34 4. MODELINGTHE IMPACT OF REGIONAL TRADE INMODERATING MAIZE PRODUCTION SHOCKS INSOUTH EAST AFRICA 4.1. The Multi-Market Model: 4.1.1. Overview 98. This paper applies a multi-market model designed to quantify the impact o f production shocks on domestic food prices. In turn, the model assesses the impact o f these changing prices on consumer, farmer and trader behavior. As an aid to policy makers and traders, the model likewise evaluates prospects for using trade policy, food aid or various government policy interventions to insulate consumers from production-induced shocks in staple food consumption. This work draws on and extends earlier multi-market models developed by Dorosh (2001), Dorosh andHaggblade (2002), Dorosh, Dradri andHaggblade ( 2007). 99. Following a given shock - such as a drought-induced fall in maize production -- the model developed here measures expected changes inthe market prices for key staple foods and the resulting impact o f price changes on farm household income, food consumption by various household groups, staple food imports and exports, and next season's production. To anticipate these multiple outcomes, the framework incorporates price responses by three key groups: consumers, who reduce maize consumption and increase consumption o f alternate staples as maize price rises; traders and millers, who import and export in response to differentials between domestic and border prices; and farmers, who alter planting decisions in response to changing prices. As exogenous variables, the model includes a range o f potential instruments wielded by government and some by donors. These include trade quotas, tariffs, public imports, government exports, local procurement, government stockholding and sales, andtargeted income transfers to vulnerable groups. 4.1 2. Model structure I 100. At its core, the model estimates how much the domestic maize price will change following an exogenous shock - a drought, flood or pest infestationaffecting farm production; a change inworld prices; public food imports; food aid; or an array o f government policy changes. Figure 17 illustrates the basic functioning o f the model. It depicts the impact o f a major supply shock, most frequently a drought, which causes maize production to decrease from SOto SI. Changes in maize output (which falls from Qo to Q1) and maize price (which increases from POto PI), in turn, affect the income o f maize-producing households as well as consumption decisions of all household groups. With even a rudimentary knowledge o f the price elasticity o f demand (depicted in Figure 17 as the slope o f the demand curve, D), the model i s able to estimate approximate orders o f magnitude for the resulting shift in market price, by tracing out movement along the aggregate demand curve (D) for maize. 35 Figure 17.FoodImportsModerate Price Shocks during aDrought Price D s0 I ~~ 1 Closed border Open border Quantity Source: Dorosh, Dradri and Haggblade (2007). 36 101. When the domestic maize price (PO) lies between import parity (Pm) and export parity (Px),no trade takes place and the domestic price (PO)prevails. But when a drought or other supply shock causes the domestic maize price to rise, import parity (Pm) sets an upper limit on the price increase. Inthe absence o f trade, the domestic maize price would spike to P1 duringa drought. But when governments allows imports, private traders import grain (an amount M1= 42-41)at the import parity price (Pm), capping the domestic price increase at import parity. Conversely, in years o f bumper maize harvest, when domestic prices plunge, the export parity price (Px) sets a floor price below which the domestic price will not fall. Only when government policy limits imports or exports does domestic price move outside these import and export parity bands. The import and export flows modeled inthis paper include both formal and informal trade. 102. To capture key consumption responses to a price shock, the model includes Zambia's two principal food staples, maize and cassava, as well as wheat, rice and other foods. Inthe event o f a drought, the maize price rises and consumers reduce their consumption o fmaize. At the sametime, they reorient consumption towards more readily available, typically more drought-tolerant staple foods such as cassava, sweet potatoes, millet and sorghum. 103, Consumption substitution among food staples occurs principally among households inthe mixed and dual staple zones (Figure 16), where people consume both cassava and maize. Because households inthese zones consume both staple foods; and because they account for over 40% o f maize regional consumption, slight changes in consumption patterns there can release significant quantities o f maize for consumption inthe maize belt (Table A2.2). 104. When prices within a given market shed (or basin) increase up to import parity, imports become profitable and the market shed expands to include supply areas from neighboring zones. Similarly, when prices fall below export parity, export becomes possible and the surplus market shed expands its geographic scope to serve the adjacent, deficit basin. The model developed below includes five geographic units that fit together in different ways to form market sheds o f different sizes depending on their relative price movements. The five geographic building blocks include: Northern Mozambique, Malawi, Southern Zambia (including Eastern Province), Northern Zambia and Southern Mozambique. In most years, the first four blocks form the core o f the South East African maize market shed, while Southern Mozambique remains linked more closely to South Africa. 105. Annex 2 describes the model formally. It also provides detailed baseline data and model parameters. 37 4.2. SimulationResults: 106. This section illustrates application o f the model by reviewing results from a series o f simulations across the South East Africa market shed. Discussion begins from the perspective o f producers and consumers in Northern Mozambique, then moves counter-clockwise across the regional market shed to evaluate the probable impact o f drought and various policy responses inMalawi and then inZambia. 107. These simulations examine two key consequences o f regional fluctuations in production. First, each set o f simulations considers the impact with and without cross-border trade. Secondly, the discussion explores the variable impact o f price increases on different constituencies within the region. Given the diverging impact o f price changes on farmers, millers and consumers, governments face conflicting pressures from the various interest groups. These tensions pose thorny domestic political economy problems for governments, who must weigh the consequences o f price fluctuations on short-run winners and losers, while at the same time considering long-run incentives for broad agricultural growth and poverty reduction. Regional political economy problems likewise arise, as short- runtrade bans, aimed at protecting domestic interest groups, may inflict harm on neighboring countries who count on food inflows or access to regional markets. By viewing the consequences o f various shocks from the perspective o f different food staple zones within the region, the simulations attempt to quantify the magnitude o f these tensions and tradeoffs. NorthernMozambique: Simulation 1,Drought in Northern Mozambique 108. The first simulation considers a scenario in which a regional drought reduces maize production by 10%below its expected base level. Ifthis i s the only shock affecting the system, given a normal price elasticity o f demand inthe range o f 0.2 to 0.3, the price o fmaize will rise sharply, by roughly 50% (Table 11,Column 1). 109. Cassava demand rises, due to cross-price effects, as consumers reduce their consumption o f maize. Because farmers can harvest many varieties o f cassava over a several year period, they are able to increase or decrease harvested quantities in any given year to accommodate surpluses or shortages o f other food staples. Given a high supply elasticity, o f 2.0, cassava quantities harvested increase by 8.8% to accommodate growing demand. Indeed, cassava consumption in rural areas increases by 13%, while maize consumption falls at a similar rate. The cassavaprice rises with growing demand, though only 4.3%, far lower than for maize, because o f the highshort-run supply elasticity.'2 l2 Estimated supply elasticities for maize and other annual cereal crops typically range between about 0.2 and 0.3 (see Annex 2). For perennial rootcrops, short-run supply elasticities are undoubtedly far higher. Because many cassava farmers manage multiple cassava plots o f staggeredmaturities, and because they can harvest many cassava varieties over a period o f two to four years, farmers can decide ifand when to harvest their individual cassava plots (see, for example, Dostie et al. 1999). Duringstudies o f cassava 38 110. In this simulation, which assumes no knock-on effect on nonfarm incomes or other agricultural commodity production, real rural incomes actually rise because the maize price increases by more than production falls. However, real incomes inurbanareasfall nearly5% due to the sharp riseinmaizeprice. Simulation 2. DroughtAffects All Rural Production 111. The second simulationmodels the consequences o f a drought that affects not only maize but all other agricultural commodities, save for cassava, which remains unaffected. Unlike the prior scenario, rural nonfarm income also suffers a 10% fall, as a result o f tight linkages to the agricultural economy. 112. Inthis situation, rural incomes are lower than inScenario 1because o fthe drop in rural nonfarm income and other agricultural production. Urban incomes likewise fall further because o f the more broadly based price inflation in both maize and other agricultural products. Results in the cassava and maize markets remain similar to Scenario 1, however, because lower incomes are offset by cross-price effects o f demand for other agriculture on maize and cassava demand, and hence price. varietal adoption, farmers frequently tell breedersthat they prefer cassavavarieties that they can harvest over many years, exactly becauseit permits themto adjust their harvested volume o f cassavaup or down in any given year inresponseto fluctuating availability of other, annual food crops (Chitundu, 2008). Although short runoutput supplyelasticity o f cassava i s likely to be very high, we are not aware o f any empirical estimates o f output responsiveness to short-term price change. Inpart, practical difficulties in estimating annual cassava output make such estimates inherently uncertain (see Akoroda, 1997). Moreover, because households market only 10% to 20% o f cassavaproduction inthe region, the majority o f year-to-year changes incassava supplyresponse never reachthe markets. Instead, these adjustments take place within the households as families adjust their cassava harvests up or down to compensate for fluctuations incereal production. Currently, micro-level investigations are under way to estimate the magnitude o fthe supply elasticity o f cassavaproduction and its impact on volume o f maize marketedby cassava-producing households. Untilthat work i s completed, we consider the supply elasticity o f 2.0 adopted here to be a conservative (low-side) estimate o fthe short-run elasticity o f cassava supply. 39 Real Income Rural 3.8% 0.1% 3.15% 5.27% 1*Urban -4.9% -7.2% -3.39% -4.46% Calories fiom maize and cassava. I Simulation 3. Normal Harvest Year: 40,000 Tonsof Food Aid Maize Procurement 113. Given regular surpluses o f maize coming out o f Northern Mozambique, food aid agencies have considered purchasing maize there for distribution to deficit households elsewhere in the region. This simulation, therefore, examines the likely impact o f procuring 40,000 tons o f maize, announced after planting so that short-run supply remains fixed at normal levels. The effect o f this procurement i s similar to a mild drought, in that it pulls supply off the local market. Most commonly such purchases take place in good harvest year and therefore serve to bolster farm price duringbumper years. 114. Under these assumptions, procurement o f 40,000 tons o f maize will lead to a 30% increase in maize prices, given fixed output and increased demand. As before, consumers respond to the rising maize price by shifting consumption in favor of other staples. Because o f its relatively elastic short-run supply, cassava harvesting increases to accommodate increased demand. Rural cassava consumption increases by 8.6%. Although rural maize consumption declines, calories available from maize and cassava increase slightly in rural areas due to the higher cassavaconsumption and the income effect o fhigher maize prices. 40 115. The tradeoff between rural and urban incomes becomes apparent in this scenario. Rural incomes rise by 3.2%, while urbanincomes fall by a similar magnitude. Simulation 4. Cross-Border Trade to Malawi 116. If farmers in northern Mozambique expect higher prices in Malawi during the coming season, as a result o f bad weather or changes in government input supply programs, they respond by increasing production. Because o f the long porous border between Malawi and the surrounding areas o f northern Mozambique, rising prices in Malawi trigger additional exports and higher maize price in Northern Mozambique. In a scenario where Malawi were to require imports of Mozambican maize 100,000 tons above normal, the maize price in northern Mozambique would increase by 44%.13 This, in turn, would induce a 7.6% increase inmaize productioninnorthernMozambique. The combination o f rising maize prices and increased output drives rural incomes up by 5.3% in northern Mozambique. Urban incomes, however, fall inresponse to rising maize prices. Malawi: Simulation 5. Drought in Malawi, Without CrossBorder Trade 117. The previous simulation examined the consequences o f a drought in Malawi, looking at the consequences from the perspective o f farmers and consumers in northern Mozambique. This simulation explores in greater detail the consequences o f a Malawian drought, this time looking at circumstances from the Malawian side o f the border. 118. Consider the impact o f a drought in Malawi that results in a 20% reduction in maize production compared to the base year. This simulation assumes that a sharp fall in maize production will also reduce related rural nonfarm income by 10%. As a result o f this drought-induced reduction in maize output, the maize price rises sharply, by 62%, in the absence o f opportunities for cross-border imports (Table 12). The cassava price rises as well, but by far less than the maize price since farmers can increase the supply o f cassava inthe short runinresponse to price increases. Rural incomes fall due the reduced farm and nonfarm output, while urbanreal incomes fall due to the sharp rise inmaize price. 119. As a result o f reduced maize production and its rising price, maize consumption falls inboth rural and urban areas o f Malawi. Consumers respond by diversifying their consumption into alternative staples. Because farmers can harvest cassava over a period o f several years, they are able to harvest additional volumes in droughts and reduce harvested quantities in years when the maize harvest i s good. l3Using data from 1997/98 and 1998/99,Santos and Tschirley (1999) estimated an econometric model suggesting that cross-border maize exports from Mozambique into Malawi led to a 15% to 21% increase in producer maize prices inNorthern Mozambique. 41 Inthis situation, cassava consumption grows by 14% inrural areas andby 2% in urban areas where sweet cassava has become a common snack food (Kambewa and Nyembe, 2007). In spite of the compensating substitution of cassava for maize, calorie consumption falls, by 6% in urban areas and by 8% inrural areas, because o f the reduction inmaize availability. Simulation5 Simulation6 Drought Drought Malawi Malawi Maize imports Closedborder fromNMozambique Imports 0 100,000 tons Production Maize -20.0% -20.0% Cassava 8.3% 3.3% I Price I I Maize I 62.2% I27.1% Simulation 6. Drought in Malawi, With Cross Border Imports from Northern Mozambique 120. The dual-staple zones o f northern Mozambique also respond to the sharp rise in maize price across the border in Malawi. Simulation 6 examines the resulting impact o f 100,000 tons o f imports from northern Mozambique into Malawi. This volume o f imports -- well within the normal flows and significantly below the 200,000 to 300,000 tons exported to Malawi in 2003 - cuts the maize price rise in Malawi inhalf, from 62% to 27% (Table 12). The cassavaprice increase likewise falls roughly inhalf. 121. Urban consumers become the principal beneficiaries o f this price containment. Their real income shock is moderated considerably, falling about 8% inthe maize import scenario (Simulation 6) compared to a 21% fall without cross-border trade 42 (Simulation 5). Caloric consumption, from cassava and maize, remains roughly similar in rural areas, while the reduced maize supply and price shock enables maize-preferring urban consumers to cut their calorie fall roughly in half, from 8% to 4%. Thus, farmers in Northern Mozambique and urban consumers in Malawi both benefit from this cross border trade. Zambia Drought: Simulation 7. Impact of a Drought, No Trade 122. Simulation 7 examines the impact o f a drought affecting 30% o f the Zambian maize crop. If Zambia were to prevent imports in the face o f a drought - by failing to issue import permits to the private sector, by announcing large volumes o f subsidized public imports and then failing to provide adequate funding (as in 2001), or by some combination o f disincentives (as happened in 2005), then domestic maize price would more than double. l4 Without the moderating impact o f private imports, which when flowing unimpeded cap price increases at import parity levels, Zambia's maize price would increase by over 160 percent. Because poor households bear the brunt o f this weather-induced compression in food availability, their maize consumption would fall by roughly 25% below normal (Table 13). Simulation 8. StapleFood Substitution 123. Even in the unlikely event that government could maintain a completely closed economy in the presence o f widespread informal trade flows, the worst-case scenario in Simulation 7 overstates the compression in food consumption by poor households, because Zambian consumers can fall back on alternative staple foods insituations where maize becomes scarce and the maize price spikes. Simulation 8 suggests that a 160 percent increase inthe maize price would induce Zambians to consume roughly 28% more cassava in the North and 21% more (from a nominal base) in the South, thus offsetting about 40% of the shortfall in maize availability. In the cassava-producing regions of northern Zambia, this substitution o f cassava for maize would largely eliminate the vulnerable households' maize deficit, freeing up maize they would have otherwise consumed for sale in other zones where consumers have developed a more pronounced preference for maize. In calorie terms, the maize-equivalent consumption shortfall among poor households would fall from 15% to 2% in the heavy cassava-consuming areas o f the North, though the impact in the South would be negligiblebecause o fthe low levels o fcassavaconsumption there. 124. Both open borders and consumer substitution among food staples moderate maize production shocks, benefiting primarily low-income consumers. These results from Zambia suggest that, under an open trade regime, private imports together 14For a review ofrecentZambiangovernmenttrade policy, see Dorosh, DradriandHaggblade(2007) and Govereh, Jape andChapoto (2008). 43 with increased cassava consumption could fill roughly two-thirds o f the maize consumption shortfall facing vulnerable households duringdrought years. Simulation 9. Drought, With Open Cross-Border Trade 125. More important to vulnerable households in the South are private imports o f maize. With bothprivate imports and consumer substitution o f cassava for maize, national food security improves markedly, even during a serious drought. Compared to base levels, the private sector imports 155,000 tons o f maize, capping the maize price increase at import parity, or 36 percent above normal lean-season levels. Although this price rise still triggers a reduction in maize consumption, even among households who prefer maize as their staple food, the resulting shortfall in staple food consumption by poor households falls to 3% in the North and 9% inthe South (Table 9). Simulation 10.Public Imports, Small Quantities 126. Iffood aid agencies or the Zambian government were to import small volumes o f maize to sell domestically at market price -- where small i s defined as any amount less than the 155,000 tons the private sector would bring in at import parityprices -- the results would be the same as under free trade. Thus the results o f Simulation 10 are identical to those in Simulation 9. In this situation, public imports would simply displace an equivalent volume o f private imports. For this combination o f side-by-side public and private imports to occur, however, the private sector needs to have confidence that public food managers will operate under transparent, predictable decision rules governing quantities, timing and release prices. The private sector needs to believe that government will not sell imported grain at below-market prices, causing commercial losses for private importers. Government, likewise, needs to have confidence that private importers will not collude to artificially boost import prices above import parity. To develop this mutual trust will require good communications and good will on both sides. 44 T d s cd c Y cd P E z 3 5 z, -3 E d s c, 3 cd z, E 5 .3 3 0 P Simulation 11.Public Imports, Large Quantities 127. If government or food aid agencies bring in maize volumes in excess o f what consumers would purchase at import parity, these large-scale public imports will drive domestic prices down below import parity. In the present example, public imports o f 255,000 tons (the maize supply gap projected under Simulation 7) would bring down prices below import parity, to only 8% above normal, even during a severe drought. However, these subsidized government imports would result in government trading losses o f $64 per ton, amounting to a loss o f $16 million. Simulation 12.Public Imports, Private Sector Impeded 128. Given late and unpredictable decision-making by Zambian authorities, many private firms have become wary o f cross-border maize trade. Simulation 12 considers a scenario, similar to 2001, inwhich government announces that it will import large volumes o f maize, thus scaring off the commercial private trade. Then, due to a shortage o f funds or to management difficulties, government ends up bringing in less maize than they intended. If government were to announce they would import 255,000 tons o fmaize (as inSimulation 1l), scaring away thus private traders, but thenimport only 50,000 tons, the maize price would more than double and staple food consumption (of maize and cassava) by low-income consumers would fall 15% below normal levels in the South (Table 13). Given the predominance o f cassava consumption in the North, poor households there remain largely insulated from the impact o f the maize price increase. Zambia BumperHarvest: Simulation 13.Bumper Harvest: No Trade, No CassavaSubstitution 129. During a good harvest year, in which national maize production increases by 30%, the lean season maize price falls by roughly 50% compared to its normal level. Given normal export parity differentials, this affords significant opportunities for export to DRC, Angola and in some years to Malawi and Zimbabwe. In the absence o f export authorization or long-term domestic stock build-up, national maize consumption will rise by 30%, with low-income consumers seeing calorie increases o f 16% to 19% (Table 14). 46 Table 14. Simulation Results: Zambia Bumper Maize Harvest Simulation 13 Simulation 14 Simulation 15 I Trade policy export ban export ban free trade Adiustments simulated Simulation 14.Bumper Harvest, No Trade, With CassavaSubstitution 130. Reversing the drought-year scenario, a bumper maize harvest leads to increased maize consumption and decreases in consumption o f other food staples, o f which cassava i s the most prominent. The, model projections suggest that national cassava consumption by would fall by 12% to 15% as households in the cassava belt and dual-staple zones consume abundant, low-priced maize. This adjustment reduces calorie gains by low-income northernhouseholds from 16% to 8%. Simulation 15.Bumper Harvest, Maize Exports Permitted 131. Private traders would have incentives to export maize at the prevailing export parity price in DRC. By drawing stocks off the domestic market, these exports would prevent domestic prices from falling below export parity, thereby reducing the maize price fall from 50% o fthe base year price under autarky to 26%. Inthis situation, maize exports cut fall the domestic maize price roughly inhalf. 47 4.3. SummingUp: 132. In South East Africa, maize production fluctuates widely from one year to the next (see Figure 10). When adjustments by consumers and farmers must take place inside o f small national markets, with no trade outlets, this production volatility translates into wide swings in the domestic maize price, in farmer incentives and in food consumption by low-income groups. The preceding simulations suggest that a drought which reduces national maize production by 20% can generate maize price increases o f 60%, while a 30% fall in production can trigger price spikes inexcess o f 100%. 133. As a result, government decision-makers face acute pressures from affected groups during years o f significant surplus or deficit. During drought years, millers and consumers lobby for imports as a means o f moderating price spikes. Yet commercial farmer groups, whose members enjoy the benefits of highmaize prices, lobby for closed borders. Conversely, during years o f bumper harvest, farmer groups call for open borders, while millers and consumers prefer to keep domestic prices low. In either circumstance, governments face pressure from some constituents to close their borders to trade, while other interest groups advocate opening up cross-border trade." 134. In most years, prospects for trade-moderated price stabilization appear good in South East Africa, giventhat coastal monsoonweather patterns inNorthern Mozambique differ from the inlandclimatic forces governing rainfall inMalawi and Zambia. The simulations indicate that even modest imports can cut price spikes by as much as 50%. This suggests that expanded cross-border maize trade can potentially help governments to moderate maize price fluctuations, as well as the political pressurethese fluctuations induce. l5Because o fthese divergent pressures, maize trade barriers inSouthernand Eastern Africa pose thorny political economy problems for governments as well as for private sector operators inthe region. Ongoing maize trade policy discussions currently take place within the framework o f the regional trade organizations, although these discussions become highly sensitive given the politicalpressures surrounding maize. As a result, these political economy questions remain priority issues for hture trade policy work in the region. 48 5. CONCLUSIONS 5.1. ModeratingConsumptionShocks: 5.1,I.Importance of Food System ShockAbsorbers 135. Maize production fluctuates widely in Southern and Eastern Africa given the region's heavy dependence on rainfed cultivation. When adjustments must take place within the confines o f small national markets, this production volatility translates into wide swings in maize price and rapid compression in food consumption by vulnerable groups. Consumers confront the pressure o f unexpected consumption shocks, while farmers face difficulties anticipating food prices inwhat can become self-reinforcing boom and bust production and pricing cycles. 136. Given current production volatility in maize, the development and expansion o f available food system shock absorbers promises significant benefits in helping to buffer these shocks and thereby stimulate both agricultural production growth in surplus zones and improved food security in deficit zones. The data and analysis presented inthis report suggest two import safety valves that can help to moderate consumption and price volatility in the presence o f recurring maize production shocks. 5.1.2.Regional Trade in Food Staples 137. The first o f these food system shock absorbers i s regional trade infood staples. In normal years, maize imports into Malawi from Northern Mozambique and Southern Tanzania combined average about 230,000 tons, about 140,000 from Northern Mozambique and 90,000 from Southern Tanzania. During crisis years in Malawi, total maize imports from these two neighboring countries more than doubles. Our simulation results suggest that even more modest inflows -- o f 100,000 tons o f imports -- inresponse to a moderate drought, can cut price spikes by as much as 50%. Given these magnitudes, regional trade flows can clearly help to soften supply deficits. 138. The liberalization o f maize trade in South Africa in 1996 and the subsequent launching o f a maize trading platform on the SAFEX exchange provide a price and a physical delivery mechanism on which traders and policy makers can build to help integrate regional markets and moderate supply shortfalls. Growing interest in regional food aid procurement has built on these new procurement mechanisms. As the Zimbabwean political situation unfolds, the regular maize deficits o f recent years may revert to the prior norm o f significant surplus production. Clearly, regional markets will continue to evolve. Inthe past decade more tools have become available to help moderate price fluctuations and stimulate production growth in surplus zones while at the same time softening price shocks indeficit zones. 49 139. Over the medium term, as consumer diets continue to change, consumption o f wheat and rice products will continue to increase in Southern and Eastern Africa. While direct human consumption o f sorghum and millet seems likely to fall, their value in livestock feeds may increase. Cross border trade in poultry, fish and meat stand poised to complement growing trade in food staples such as rice, wheat and beans. Given specialized production requirements for many non-maize food staples, consumption diversification seems likely to open up new opportunities for commercial cross-border trade in food staples. 5.1.3. Consumer Substitution among Food Staples 140. The second major shock absorber highlighted in this paper is consumer substitution among food staples. The results from Northern Zambia illustrate the importance o f this consumer substitution. Even during a serious drought, reducing national maize production by 30%, poor households in the cassava- consuming north see calorie intake from cassava and maize fall by only 2%. Overall, this cassava substitution for maize compensates for about 40% o f the national maize shortfall inthese circumstances. 141. While discussion here has concentrated on cassava as a secondary food staple, these results should be considered illustrative o f the broader opportunities for consumer substitution o f a whole array o f drought-tolerant alternative staples including sorghum, millet, and sweet potatoes. In urban areas, shifting consumption patterns and growing consumer preferences for rice and wheat products open hrther opportunities for substitution among food staples. 142. Livestock feed and livestock products, likewise, offer important substitution possibilities, particularly given growing demand for poultry, meat and other livestock products. Feed companies in Zambia are currently experimenting with the incorporation o f cassava into their feed formulations. In years such as 2008, with sharply rising world cereal prices, the introduction o f alternative local carbohydrate sources (such as cassava and sorghum) in feeds offers a second conduit for relieving pressure on maize demands. Although maize dominates many food policy discussions, a portfolio o f drought-tolerant secondary staples offers important benefits insubstitutingfor highlyvariable maize supplies. 143. These substitution possibilities help to moderate the food compression induced by maize supply shocks. Our simulations from Zambia suggest that, together, commercial maize imports plus consumer substitution o f cassava for maize would fill roughly two-thirds of the maize consumption shortfall facing vulnerable households during a typical drought year, assuming open borders. 50 5.2. MaintainingFarmer and Trader Incentives: 144. Farmers, too, pay a price for the increased volatility that results from closed borders. When production shortages one season lead to price spikes and expanded planting the following year, closed borders exacerbate the initial spike and well as the subsequent price fall in the next season. Farmers from northern .Mozambique, who plant additional maize area in expectation o f supplying Malawi become discouraged when deprived o f this natural export market. For this reason, many farmers in Tete Province o f northern Mozambique began switching to alternative cash crops such as tobacco following mis-timed public maize distribution at harvest time which triggered the Malawian price collapse o f 2003 (Whiteside, 2003) 145. Our simulations suggest that production surges, which confined within small national markets, lead quickly to price collapse. In Zambia, for example, a 30% boost in maize production can lead easily to a 50% price fall, when borders remain closed. However, when policy makers allow maize exports to neighboring DRC, the export parityprice forms a price floor, limiting the price fall by half, to about 26%. Conversely, during a drought, open borders limit price spikes. Again using the Zambia example, a 30% production decrease in the maize harvest can lead to more than doubling o f the maize price, when borders remain closed. But when imports are allowed, typical border prices limit the price increase to about one-third this level 146. In general, policy restrictions - such as export bans, permit systems and tariffs - tend to drive cross-border trade from formal into informal channels. Because handling costs rise with small lots, and with truck to bicycle to truck transfers at the border, these policy restrictions do, infact, diminishtrade flows by driving up transaction costs. Tracking informal maize trade from Southern Tanzania into Malawi, one study notes, "Costs fell inmid 2001, when the export ban was lifted and the traders no longer had to take the maize across the border by bicycle and canoe. This illustrates how lifting an export ban can tip the balance between profitable and unprofitable trade." (Whiteside, 2003, p.33). Tariffs and border controls between Uganda and Kenya imposed similar transaction costs by forcing bicycle transfers at border crossings, up until 2005 when signature o f the East African Community (EAC) treaty diminished these controls (RATES, 2003; Ariga and Jape, 2007; Jape, Myers and Nyoro, 2005). High transactions costs and uncertainties about government import and pricing policies pose significant risks to traders and tend to discourage private sector food imports during times o f food shortages. 147 Policy reviews in Zambia and Malawi suggest that unpredictable and intermittently interventionist government policies - government procurement, direct import and subsidized sales - have tended to undercut private traders by raising risk and forcing many to exit the formal import market (Mwanaumo et al., 51 2005; Nijhoff et al. 2002 and 2003; Tschirley et al, 2004; Whiteside, 2003). In contrast, Mozambique has maintained generally open border policies since the end o f the civil war in 1992. Northern Mozambique benefits from access to deficit markets in Malawi, while southern Mozambique benefits from access to nearby imports from South Africa. 148. If farmers and traders in surplus food production areas are to invest in soil fertility, productive assets and input supply and marketing systems, they will need reliable access to deficit markets, which in many instances lie across national borders. Import bans, export bans and other trade controls, often intermittent and unpredictable, undercut incentives for farmers and traders and tend to undermine prospects for accelerated food production growth. Extreme price volatility, upwards and downwards, tends to discourage long-term investment inproductive capacity and favor, instead, short-term speculative behavior. Open borders, by increasing farmer incentives in surplus areas and dampening price spikes in deficit zones, can contribute to both agricultural growth in the surplus zones and to food security indeficit markets throughout the region. 5.3. PolicyImplications: 5.3.1. Open borders 149. Open borders for regional maize trade offer potentially important benefits to farmers in surplus production zones and consumers in deficit areas. Because regional maize production varies less than production in each individual country, open borders offer a means o f reducing supply and price volatility in individual countries. The common policy alternative, o f import bans and export bans, leads to highly volatile maize prices prices, spikes which risk punishing consumers in some years and farmers in others. Over the mediumrun, reduced price volatility encourages on-farm investment by farmers in surplus zone and mitigates forced consumption compression by vulnerable households. 150. The key policy instruments for encouraging cross-border trade include cessation o f quantitative controls, tariff reduction, and harmonization o f customs procedures. Regional trade associations -- such as COMESA, SADC and EAC -- provide fora for negotiating and enforcing regional trade agreements. 151. Recent increases in world cereal prices have substantially raised the cost o f importing food from international markets. Therefore, the coming crop year will prove particularly difficult for deficit areas unless they can raise domestic production or source supplies at lower cost from surplus regional suppliers in Africa. Attentiveness to regional production forecasts, regular tracking o f regional markets and border prices, ongoing policy monitoring and open dialogue among neighbors can all contribute to improved preparedness and responsiveness to unfolding food requirements. Over the coming year, assuring food security for 52 vulnerable households and deficit zones within Africa will depend, more than ever, on open borders and fluid cross-border flows o f staple foods. 5.3.2.Regional transport corridors 152. Improved regional infrastructure can complement trade reforms by lowering transport and transaction costs. Adopting a regional perspective, public investment in transport corridors can help to link surplus farming zones with cross-border deficit markets, thus benefitting farmers in surplus zones and consumers in the deficit markets. Transport corridors between Northern Mozambique and Malawi offers a clear example o f the potential food security benefits o f regional infrastructure investment program. To be effective, these investments will need to be accompanied by harmonized customs regulations and policy reforms facilitating cross-border flows o f food staples. 5.3.3. Secondaryfood staples 153. Food policy discussions need to expand beyond maize. Secondary food staples such as cassava, sweet potatoes, sorghum and millet merit increasing attention in domestic and regional food policy debates. Currently, policy attention and government investments focus largely on maize. Yet recurring food shortages and price volatility in the region stem primarily from the dominance o f maize together with its moisture sensitivity, consequently erratic production. Drought tolerant staples such as cassava, sweet potatoes, sorghum and millet offer important food security shock absorbers during drought years. The roots and tubers, in particular, also offer higher calorie productivity per unit o f labor and land than does maize. Yet research and extension budgets for these secondary foodcrops languish inmany countries. Livestock and the feed industry also offer buffering capacity in the presence o f maize production and price shocks. Partial substitution o f cassava for maize infeed rations (as inEurope) can help to cushion maize supply shortages during drought years. 5.3.4. Trust 154. Ultimately, increased reliance on cross-border trade requires trust. Governments must have confidence that competitive markets will deliver food staples at low cost and in adequate quantity. And traders must trust that governments will not change policies unexpectedly -by announcing public imports o f a certain amount and then failing to bring in'the full amount or by unexpectedly releasing public stocks at below-market prices and thereby imposing commercial losses on firms who import food inresponse to a drought. 155. Experiencein opening cross-border trade infood staples suggests several practical steps that can improve understanding and, over time, build trust between government policy makers and private sector groups. First, where governments mistrust traders and fear collusion, increased competition offers one potential 53 antidote. The intense price competition among several hundredBangladeshi rice importers proved key to their effective response to the 1998 floods inBangladesh, when traders staved o f f supply shortages and capped domestic prices at import parity by importing several million tons o f rice from neighboring India (Dorosh, 2001). Similarly, an ex-post assessment o f the 2004 rice crisis in Madagascar concluded that improved competitiveness o f grain import markets required development o f clear and transparent policies along with a level playing field for all actors (Magnay and Jenn-Treyer, 2006). 156. Second, where traders mistrust governments, active dialogue between the public and private sector serves to improve transparency and trust, as both the Madagascar and Bangladesh experiences emphasize (Dorosh, 2008). InZambia, the recent launching o f a joint maize monitoring and stocks review committee involvingfarmers, traders, millers and government represents an important step in this direction (ZNFU, 2007). More generally, ongoing discussions with traders about trade impediments and possible measures to reduce transaction costs and facilitate commercial flows serve to maintain open lines o f communication on ways o f improving market efficiency andreliability. 157. Finally, governments and traders need to monitor staple food production and markets over time and make this information widely available (Minten and Dorosh, 2006). Production estimates normally fall within the public domain. The likely downward revision in official Malawian maize production estimates in 2008 suggests that improvements in crop forecasting methods might yield considerable benefits, enabling both traders and policy makers to more accurately anticipate future seasonal price movements and potential import needs. Policy makers and the private sector likewise need to track price movements, o f both domestic and regional prices, in order to monitor domestic and import parity prices. Government monitoring o f letters o f credit can likewise prove helpful in maintaining a clear indication o f private sector trading intentions. These market monitoring efforts require regional cooperation and data sharing. In Southern Africa, the South African Commodity Exchange (SAFEX) and Famine and Early Warning System Network (FEWSNET) provide an existing backbone on which to buildactive market information systems throughout the region. 158. Inthe end, transparency, open lines o f communication between governments and traders, predictability in policy making, and investments in reliable, accessible market information - on production, stocks, prices and quantities traded - provide necessary underpinningsfor growing regional trade infood staples. 54 REFERENCES Alderman, Harold and del Ninno, Carlo. 1999. Poverty issues for zero rating VAT in South Africa. Journal of African Economies 8(2) 182-208. Aliguma, Lucy. 2008. "Uganda Maize Trade Country Profile." Background report prepared for World Bank under contract No. 7144132, Strengthening Food Security in Sub-Saharan Africa through Trade Liberalization and Regional Integration. Washington, DC: The World Bank. Ariga, Joshua and Jape, T.S. 2007. "Maize Trade and Marketing Policy Interventions in Kenya." Background paper prepared for the FA0 Conference on "Food Marketing and Trade Policy Options in Eastern and Southern Africa, EST Division, FAO, Rome, March1-2,2007. Ashimogo, Gasper. 2008. "Tanzania Maize Trade Country Profile." Background report prepared for World Bank under contract No. 7144132, Strengthening Food Security in Sub-Saharan Africa through Trade Liberalization and Regional Integration. Washington, DC: The World Bank. Awuor, Thomas. 2007. "Review of Trade and Markets Relevant to Food Security inthe Greater Horn o fAfrica." Washington, DC: USAID. Byerlee, Derek, T.S. Jape, and Robert J. Myers. 2006. Managing food price risks and instability in a liberalizing market environment: overview and policy options. Food Policv, 31(4):275-287, August. Chitundu, Maureen. 2008. Personalcommunication, June 17,2008, del Ninno, Carlo, Paul Dorosh and Kalanidhi Subbarao. 2007. Food Aid, Domestic Policy and Food Security: Contrasting Experiences from South Asia and sub-Saharan Africa. FoodPolicy. Devereux, Stephen. 2002. The Malawi Famine o f 2002. Institute o f Development Studies Bulletin30(4):70-78. Dorosh, Paul. 2001. Trade liberalization and national food security: Rice trade between Bangladeshand India. World Development 29(4): 673-689. Dorosh, Paul A. 2008. "Food Price Stabilization and Food Security: The International Experience" Bulletinof Indonesian Economic Studies 44(1): 93-114. Dorosh, Paul A.; Dradri, Simon and Haggblade, Steven. 2007. "Alternative Instruments for Ensuring Food Security and Price Stability in Zambia." FSRP Working Paper 55 No.29. Lusaka, Zambia: Michigan State University Food Security Research Project. http://www.aec.msu.edu/fs2/zambia/wp-29.pdf Dorosh, Paul A. and Subran, Ludovic. 2007. "Food Markets and FoodPrice Inflation in Ethiopia." (mimeo). Washington, DC: The World Bank. Dostie, Benoit; Randriamamonjy, JosCe; and Rabenasolo, Lala. 1999. "La filibre manioc: amortisseur oublik des vulnerable." Ithaca, New York: Cornell Food and NutritionPolicy Program. Fackler, Paul L. and Barry K. Goodwin. 2001. "Spatial Price Analysis", in Bruce L. Gardner and Gordon C. Rausser (eds.) Handbook of Agricultural Economics, Volume IB: Marketing, Distribution and Consumers, Chapter 17. Amsterdam: Elsevier. Govereh, Jones; Jayne, T.S.; and Chapoto, A. 2008. Assessment of Alternative Maize Trade and Market Policy Interventions in Zambia. Working Paper No.33, Food Security ResearchProject. Lusaka: Zambia.. Haggblade, Steven and Zulu, Ballard. 2003. "The Cassava Surge in Zambia and Malawi," Successes in African Agriculture Conference Background Paper No.9. Washington, DC: International FoodPolicy ResearchInstitute. Haggblade, Steven. 2006a. Maize Price Projections for Zambia 's 2006/07 Marketing Season.Food Security ResearchProject Policy Synthesis No. 15. Lusaka: FSRP. Haggblade, Steven and Nielsen, Hunter. 2007. Zonal Mapping o f Food Staple Zones in Zambia, Malawi and Mozambique. Cassava Transformation in Southern Africa (CATISA) Startup Report No.1. East Lansing, Michigan: Michigan State University. Harber, R. 1992. Maize market decontrol programme (MMDP) Project No. 611-0223. Annex I. Economic Analysis. Lusaka: U.S. Agency for International Development. Howard, Julie A. and Mungoma, Caroline. 1996. Zambia's Stop and Go Revolution: The Impact of Policies and Organizations on the Development and Spread of Maize Technology. MSUInternational Development Working Paper No. 61. East Lansing, Michigan: Department of Agricultural Economics, Michigan State University. Jansen, Doris. 1988. Trade, Exchange Rate and Agricultural Pricing Policies in Zambia. Washington, D.C.: World Bank. Jayne, T.S.;p Myers, Robert J; and Nyoro, James. 2008. "The Effects of NCPB Marketing Policies on Maize Market Prices in Kenya." Agricultural Economics 38(2008):3 13-325. 56 Jayne, T.S., Tembo, G., Nijhoff, J., 2005. Experiences o f food market reform and price stabilization ineastern and SouthernAfrica. Food Policy. 31(4): 328-342. Jayne, T.S.; Mukumbu, Mulinge; Chisvo, Munhamo; Tschirley, David; Weber, Michael T.; Zulu, Ballard; Johansson, Robert; Santos, Paula and Soroko, David. 1999. Successes and Challenges of Food Market Reform: Experiences from Kenya, Mozambique, Zambia and Zimbabwe. MSU International Development Working Paper No.71. East Lansing, Michigan: Department of Agricultural Economics, Michigan StateUniversity. Jayne, T. and S. Jones. 1997. Food Marketing and Pricing Policy in Eastern and SouthernAfrica: A Survey. World Development 25(9): 1505-1527. Kalinda, T., F. Maimbo, and M. Mushimba, 2003. Zambia Food Security Issues Paper. Forum for Food Security in Southern Africa. www.odi.orn.uk/food-securitv- forum. Kambewa, Patrick and Nyembe, Misheck. 2008. "Structure and Dynamics o f Malawi's Cassava Markets." Cassava Transformation in SouthernAfrica (CATISA) Startup Report 2b. East Lansing, Michigan: Michigan State University. http://www. aec.msu.edu/anecon/fs2/catisa/Dynamics Malawi Cassava Market.pdf Katepa, M.T. 1984. Price controls and fertilizer subsidies: a partial equilibrium analysis of the Zambian corn sector. M.S.Thesis, Department of Agricultural Economics, PurdueUniversity. Koester, Ulrich. 1986. Regional Cooperation to Improve Food Security in Southern and Eastern African Countries. Research Report No. 53. Washington, D.C.: International FoodPolicy ResearchInstitute. Magnay, John and Jenn-Treyer, Olivier. 2006. The 2004 Crisis and Some Lessons Learnt. Chapter 4 in Bart Minten and Paul Dorosh, editors. Rice Markets in Madagascar in Disaray. Africa Region Working Paper Series No.101. Washington, DC: The World Bank. Malan, Tatenda. 2007. State allows 40,000 metric tones maize export. Daily MailMarch 23,2007. Minot, Nick and Francesco Goletti. 2000. Rice Market Liberalization and Poverty in Vietnam. Research Report No. 114. Washington, D.C: International Food Policy ResearchInstitute. Minten, Bart and Paul Dorosh (eds). 2006. Rice Markets in Madagascar in Disarray: Policy Options for Increased Efficiency and Price Stabilization. Africa Region Working Paper Series. Number 101. Washington: World Bank. 57 Moser, C., C.B. Barrett and B. Minten. 2005. "Missing Markets or Missed Opportunities: Spatio-temporal Arbitrage o fRice inMadagascar," manuscript. Muyanga, Milu; Jayne, T.S.; Argwings-Kodhek, G.; and Ariga, Joshua. 2006. Staple food Consumption Patterns in Urban Kenya: Trends and Policy Implications. Tegemeo Institute of Agricultural Policy and Development, Working Paper No.19, Nairobi: Egerton University. Mwale, Geoffrey S. 2008. "Maize Trade Country Profile for Katanga and Kasai Province, DCR." Background report preparedfor World Bank undercontract No. 7144132, Strengthening Food Security in Sub-Saharan Africa through Trade Liberalization and Regional Integration. Washington, DC: The World Bank. Mwanaumo, Anthony; Jayne, T.S.; Zulu, Ballard; Shawa, Julius; Mbozi, Green; Haggblade, Steven; and Nyembe, Misheck. 2005. Zambia's 2005 Maize Import and Marketing Experience: Lessons and Implications. Food Security Research Project Policy Synthesis No.11. Lusaka. Nakaponda, B. 1992. Food security in Zambia: an econometric analysis o f the maize market. M.S. Thesis, Department of Agricultural Economics, Michigan State University. Nijhoff, J.J.; Jayne, T.S.; Mwinga, Billy; and Shaffer, Jim. 2002. Markets Need Predictable Government Actions to Function Effectively: The Case o f Importing Maize in Times o f Deficit. Policy Synthesis No.6. Lusaka: Food Security Research Project. Nijhoff, J.J.; Tembo, Gelson; Shaffer, Jim; Jayne, T.S.; and Shawa, Julius. 2003. How Will the Proposed Crop Marketing Authority Affect Food Market Performance in Zambia: An Ex Ante Assessment to Guide Government Deliberation. Working Paper No.7. Lusaka: Food Security ResearchProject. Nyoro, James. 2008. "Kenya Maize Trade Country Profile." Background report prepared for World Bank under contract No. 7144132, Strengthening Food Security in Sub-Saharan Africa through Trade Liberalization and Regional Integration. Washington, DC: The World Bank. Phiri, Alexander R.; et al. 2001, "Applying a Sub-sector Analysis Approach to Studying the Marketing of Cassava inMalawi: A Qualitative Assessment of the Subsector." Lilongwe: SARRNET. Phiri, Alexander R. 2008. "Malawi Maize Trade Country Profile." Background report prepared for World Bank under contract No. 7144132, Strengthening Food Security in Sub-Saharan Africa through Trade Liberalization and Regional Integration. Washington, DC: The World Bank. 58 RATES. 2003a. "Maize Market Assessment and Baseline Study for Ethiopia." Nairobi: The RATES Center. RATES (Regional Agricultural Trade Expansion Support Program). 2003b. "Regional Maize Trade PolicyPaper." Nairobi: COMESA. Rosling, Hans. 1986. "Cassava, Cyanide and Epidemis Spastic Paraparesis: A Study in Mozambique on Dietary Cyanide Exposure." Uppsala, Sweden: Faculty o f Medicine. Santos, Ana Paula and Tschirley, David L. 1999. "The Effects o f Maize Trade with Malawi on Price Levels inMozambique: Implications for Trade and Development Policy." MAF/MSU Research Report No.34. Maputo -Direcp?ode Econom6, MAP. http://www.aec.msu.edu/agecon/fs2/mozambique Times. 2007. Floods cut offmaize export plans. Times ofZambia March 19,2007. Timmer, C. Peter, Walter P. Falcon and Scott R. Pearson. 1983. Food Policy Analysis. Baltimore, MD:Johns HopkinsUniversityPress. Traub, Lulama Ndibongo and Jape, T.S. 2004. "The Effect of Market Reform on Maize Marketing Margins in South Africa." MSU International Development Working Paper No.83. East Lansing, MI:Department of Agricultural Economics, Michigan State University. Traub, Lulama Ndibongo. 2008. "South Africa Maize Trade Country Profile." Background report prepared for World Bank under contract No. 7144132, Strengthening Food Security in Sub-Saharan Africa through TradeLiberalization and Regional Integration. Washington, DC: The World Bank. Tschirley, David and del Castillo, Anne Marie. 2007. "Local and Regional Food Aid Procurement: An Assessment o f Experience in Africa and Elements o f Good Donor Practice." MSU International Development Working Paper No.91. East Lansing, Michigan: Department o f Agricultural Economics, Michigan State University. Tschirley, David and Jape, T.S. 2007. "Food Crises and FoodMarkets:What has been Learned in SouthernAfrica over the Past Decade?" Presentedat Vulnerability to and Early Warning for Food Emergencies: Conceptual Issues and Practical Implementation. FA0 Global Information and Early Warning System (GIEWS) on Food andAgriculture.FAO, Rome. December 6-7,2007. Tschirley, D., Nijhoff, J. Arlindo, P., Mwinga, B., Weber, M. and Jape, T. 2004. Anticipating and Responding to Drought Emergencies in Southern Africa: Lessons from the 2002-2003 Experience. Prepared for the NEPAD Conference 59 on Successes in African Agriculture, 22-25 November 2004, in Nairobi, Kenya. Online at: http://www.aec.msu.edu/aaecon/maizemarket/index.htm Tschirley, David, Jan J. Nijhoff, Pedro Arlindo, Billy Mwinga, Michael T. Weber, and T.S. Jayne (2005). "Anticipating and Responding to Drought Emergencies in Southern Africa: Lessons from the 2002-2003 Experience". International Development Working Paper No. 89. Michigan State University. http://www.aec.msu.edu/aaecon/fs2/papers/idpw89.pdf" Tschirley, David, Danilo Abdula, and Michael T. Weber (2006). "Toward Improved Maize Marketing and Trade Policies to Promote Household Food Security in Central and Southern Mozambique". Research Paper No. 60E. MA/MSUFood Security Project. Maputo. http://www.aec.msu.edu/agecon/fs2/mozambique/wps60e.pdf Tschirley, David and Abdula, Danilo. 2007. "Toward ImprovedMaize Marketing and Trade Policies to Promote Household Food Security in Central and Southern Mozambique: 2007 Update." Paper Prepared for the Conference on Trade Policyfor Food Products Conducive to Development in Eastern and Southern Africa. Food and Agriculture Organizationo f the U.N. Rome, Italy. March 2007 USDA. 1996. International food consumption patterns. United States Department of Agriculture (USDA). http://www.ers.usda.gov/data/InternationalFoodDemand Whiteside, Martin. 2003. Enhancing the Role of Informal Maize Imports in Malawian Food Security. Lilongwe: DflD. Wood, Adrian Paul; Kean, Stuart A.; Milimo, John T.; and Warren, Dennis. 1990. The Dynamics of Agricultural Policy and Reform in Zambia. Ames, Iowa: Iowa State UniversityPress. Zambia Daily Mail. 2007. "Abolish Mozambique, Malawi Maize Imports." Lusaka, September 20,2007. Zambia: Social Accounting Matrix, 2001. 2005. Washington, D.C.: International Food Policy Research Institute (IFPRI)(datasets). http://www.ifpri.ora/data/Zambia02.htni. Zinyama, Fridah. 2007. WFP awards NMC Zim maize export contract. The Post Tuesday, February 27,2007, p.11. ZNFU (Zambia National Farmers' Union). 2007. ZNFU fortnightly newsletter. (June 2007). Lusaka: ZNFU. Zulu Ballard., J.J. Nijhoff, T.S. Jayne and A. Negassa. 2000. I s the glass half-empty or halffull?An analysis o f agricultural Production trends inZambia. Working paper no. 3. Food Security ResearchProject. Lusaka 60 ANNEX 1.DATAAND METHODSFORDEFININGMARKETSHEDS 1.1. Methods 1.I.1.Qualitative assessment of marketflows Traders active in staple food markets, particularly those from surplus areas and border zones, know generally where food supplies move geographically. This study, therefore, has relied primarily on the direct observations o f trade flows by market participants and officials who monitor cross-border flows. This study has not attempted any formal econometric price analysis o f market integration, o f the type described by Fackler and Goodwin (2001) and Moser, Barrett and Minten (2005), due to the spatial sparseness o f time series price data and the frequent shifts in country policies that often inhibit market flows and hinder integration o f markets across countries. Instead, this study provides qualitative assessment o f the direction o f flows in normal and drought years as well as approximate quantitative estimates o f the extent and size o f existing market sheds using available information on production, prices and trade flows, both formal and informal. Indoing so, we have pulled together information from an array of earlier food marketing studies conducted by the Regional Agricultural Trade Enhancement Support (RATES) Project, the Famine Early Warning Network (FEWSNET), Michigan State University (MSU) and allied researchers, the International Food Policy Research Institute (IFPRI), the Food and Agriculture Organization (FAO), the World Bank, the World Food Programme (WFP) and others. Market monitoring by the Regional Agricultural Trade Intelligence Network (RATIN) and FEWSNET permit us to track changes in flows over time. To help fill in gaps in our understanding, we have commissioned a series o f specific studies on regional maize trade. These studies are available inAnnexes 1-6. Because both volumes traded and the exact geographic extent o f a market shed may vary from one year to the next, depending on weather, production patterns and changing government policies, we have relied on interviews with traders and other knowledgeable keyinformants inorder to put this information incontext. I.I.2. Quantijkation National food production figures are subject to significant margins o f error. Even with the maize harvest, which takes place at a known time and location, available estimates are often subject to a wide margin of error. With cassava, sweet potatoes and bananas, which are harvested over many months, production estimates are far more tenuous, Quantification o f trade flows poses similar problems. Given highly porous borders in many parts o f the region, estimates o f formal and informal trade flows are often subject to substantial undercounting. 61 Indeed, difficulties in correctly anticipating food availability from local and regional production poses persistent problems to government policy makers, food aid donors and to private traders. The Malawian experience o f 2002 and early 2003 offers a classic example o f this, when government misjudged prospects for private sector imports from Mozambique, which led to overshooting on public food imports, difficulties indisposing o f surplus stock and triggered a consequent price slump early in the following harvest season (Whiteside, 2003; Tschirley et al., 2006). To manage this imprecision, we have triangulated and compared quantity estimates using a variety o f available sources. National and provincial production estimates, coupled with national farm household and national food consumption surveys permit estimation ' o f net surpluses. And as a check against official border counts, net imports may be estimated as the difference between consumption (as derived from household surveys), regional production and net injections by public sector agencies (through food aid distribution or net sales o f government food agencies). Finally, we have cross-checked quantity information with available monthly price series and evidence on transaction costs between pairs o f markets within a defined market shed in order to arrive at final best estimate o f quantities. 1.2. Data 1.2.1,Qualitative information Qualitative information provided by market traders i s available from a variety o f valuable field studies. Key studies o f maize marketing have taken place in Zambia (Govereh, Chapoto and Jayne, 2008), Malawi (Chinva, 2006), Mozambique (Tschirley, Abdula and Weber, 2006; Tschirley and Abdula, 2007), Kenya (Ariga and Jayne, 2007), and Ethiopia (RATES, 2003a; Gabre-Madhin, 1998). In addition, several recent studies have examined regional food markets in Eastern and Southern Africa. See, for example, Ackello-Ogutu and Eshessah (1997), Arlindo and Tschirley (2003), RATES (2003b), Whiteside (2003), Tschirley et al., (2006), Chapoto and Jayne (2007), and Dradri (2007). Ongoing monitoring o f food markets and cross-border flows, by FEWSNET in Southern Africa (see FEWSNET, 2008a-d) and by the RATES Project in Eastern Africa (see RATIN, 2003), permit a detailed reconstruction o fproduction andtrade patterns inmany parts o f the region. Likewise, WFP's regional procurement officers in Southern Africa have gained considerable experience in evaluating production and available surpluses over time (see Tschirley and del Castillo, 2007). In an effort to pull together this information, we have benefited from participating in a two-day workshop in Pretoria in November 2007, hosted by FEWSNET and WFP, designed to pool knowledge o f key stakeholder groups on the direction and magnitude o f food flows innormal years and common emergency situations. 62 1.2.2. Quantitative information National production data are available from Ministries o f Agriculture, and are compiled by the FA0 Statistical Service. Customs officers and border monitors track trade flows, with varying degrees o f completeness. To supplement official border data, FEWSNET andWFP havebeenmonitor informal cross-border trade flows inmaize, rice, beans. Price data come from national statistical offices throughout the region as well as from the South African Commodity Exchange (SAFEX) inJohannesburg. Differences inproduct quality, variations lot sizes and differential product tracking across collection agencies (grain versus flour, for example, or retail versus informal wholesale versus millgate prices) sometimes make price comparisons difficult and require expert consultation with key market participants and observers to resolve. 63 ANNEX 2. MULTI-MARKETMODEL A.2.1. ModelEquations Production Trade Private: free trade MPRN' =C' -Xi -MPUB' quotas MPRN' =MPRN Public MPUB' =MGOV' +MFOODAID' Demand D'= +ASTOCKS' +GOVPURCH' -GOVSALE' ci Equilibrium S' =D' Autarky price PD' = equilibriumpricewith MPRNandMPUB=0 Market price Pi = PIMP' ifPD' >PIMP' PD' ifPEXP' PD'