Report No: AUS13486 Republic of Zimbabwe Economic growth notes ZIMBABWE: SPATIAL INTEGRATION IN ZIMBABWEAN PRODUCT MARKETS December3, 2015 GMF13 AFRICA U Document of the World Bank Standard Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement: The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights(aworldbank.org. ZIMBABWE: SPATIAL INTEGRATION IN ZIMBABWEAN GRAIN MARKETS December 3, 2015 Macroeconomic and Fiscal Management Africa Region Contents 1 Overview. ................................................. ............ -4 2 Background ................5 3 Data and Methodology... ............. .6 4 Market Integration Patterns ................................... .. ..............7 Maize Markets ........... .................... Other Commodities............................................ ....... 10 5 Determinants of Market Integration.......... ..................................12 6 Concluding Remarks....................... ..............................14 References.................. . ........................................ ...17 Annexes ................................................................. 18 Figures Figure 1: Contribution of Agriculture to GDP,................ ....6 Figure 2: Trends in Maize Production (mt).....................6 Figure 3: Segmented Integration Pattern for Maize Grain.,....... .... ...9 Figure 4: Prevalence of food insecurity......................10 Figure 5: Percent of households stating maize is readily available..........................10 Figure 6: Zimbabwe's grain balance................................. .............14 Figure 7: Thin markets for small grains ............................................14 Tables Table 1Commodity Price Data Sources............................................. 6 Table 2: Summary of Cointegrated and VECM results (Average).. ........................1 2 Acknowledgements This policy note was prepared by Shireen Mabdi (Senior Economist, GMFDR) and Matteo Bonato (Consultant, GMFDR) as part of the Zimbabwe Economic Policy Notes. The team benefited from discussions with Johannes Herderschee (Senior Economist, GMFDR) and the overall guidance of Mark Roland Thomas (Practice Manager, GMFDR). The team is grateful to Fortune Chigumira and Malvern Mupandawana for research assistance. The team would also like to thank Camille Nuamah (Country Manager, AFMZW), Mark Cackler (Practice Manager, GFADR), Catherine Tovey (Program Leader, AFCS 1), Ijeoma Emenanjo (Natural Resource Management Specialist, GFADR) and Gibson Guvheya (Senior Operations Officer, AFMZW) for the guidance and feedback provided. 3 I OVERVIEW The purpose of this policy note is to provide evidence on the level of integration between Zimbabwe's domestic markets for grains and staple foods. In order for prices to deliver the appropriate incentives, markets must be efficient and hence, integrated. Integrated markets have a long-term relationship whereby price changes in one market are transmitted to another, providing the necessary signals that guide production and trading decisions. From a theoretical perspective, integrated markets are consistent with the law of one price, which postulates an arbitrage driven equilibrium whereby the difference in price between two markets for the same good remains only with the transaction costs between them, In practice, integration is a signal of well-functioning markets that are characterized by competition, information and absence of policy impediments to supply and demand response. Efficient and integrated agricultural markets are an important vehicle for growth and poverty reduction. A key stimulant to agricultural growth and poverty reduction is small farmers' incentives, whether through higher sale prices at the farm-gate, reduced costs of production and transport, or the spread between them. In order for prices to deliver the appropriate incentives to smallholders, markets must be integrated and assure the participation of the poor. These factors provide the motivation for this policy note, which covers six grain markets, across the ten provinces of Zimbabwe, These markets represent some of the most significant products in the economy as well as some of the most important commodity markets for the poor, both as producers and consumers. Indeed, Zimbabwe's grain markets are estimated to contribute between 17 and 32 percent of rural household incomes The note determines whether Zimbabwe's provincial markets are integrated and explores the determinants of market integration. It begins by determining whether Zimbabwe's provincial markets are integrated, and provides estimates of the speed of adjustment between markets pairs. The analysis is then extended to investigate the extent to which distances and demand for the products determine market integration. The paper concludes with a discussion of other structural or policy related factors that affect integration and efficiency of domestic markets, The literature on market integration and price transmission is vast3, and has contributed to the understanding of market dynamics and transmission channels especially in times of global price volatility. Price transmission analysis has been used to investigate the relationship between world prices and local prices for given commodities, to analyze the prices of two competing commodities (such as maize and cassava), or the prices of two commodities in the same value chain (e.g. maize grain and maize meal). In this study, we focus on spatial integration between regional prices for the same commodities, Maize grain, maize meal, sorghum, cowpeas, sugar beans and groundnuts. 2 ZIMVAC Rural Livelihoods Assessment (2011). Sekhar (2012); Jacks et al. (2011); Minot (2011). 4 The main finding of the study is that Zimbabwe's largest domestic grain market, the market for maize, is segmented between surplus and deficit areas. Although maize markets tend to be integrated within these zones, the linkages between surplus and deficit areas are weak. The other commodities (cowpeas, groundnuts and sugar beans) have more robust level of integration with the exception of sorghum, which has the lowest levels of market integration. However, these grains are narrowly produced and traded. A domestic marketing structure that is strong on aggregation and weak on arbitrage, and high transactions costs for small traders, are highlighted as the main drivers of the observed segmentation, In particular, a requirement for traders to obtain an annual permit at a cost of USD 1,000, doubled to USD 2,000 in case of late registration is an onerous cost that may dampen their role in moving maize across zones. High transport costs between the regions and the low production levels are also discussed as drivers. The policy note concludes by offering some policy recommendations to strengthen market integration. First, liberalizing market entry for small traders by removing the current set of administrative fees would reduce their transaction costs and strengthen their role in facilitating spatial arbitrage. Second, prioritizing rural roads and transport costs to and within deficit areas would reduce transaction costs and deepen integration. Lastly, deepening the focus of key agricultural programs on stimulating production in deficit areas would ease local food shortages and help mitigate the effects of weak market integration. 2 BACKGROUND Grain production, maize in particular, is a dominant feature of Zimbabwe's agricultural landscape. Agriculture plays an important role in Zimbabwe's economy and represents a major source of income for the poorest and most vulnerable households. Its contribution to GDP averaged 11.5 percent since 2009 (figure 1) and it is the source of over 40 percent of export earnings. Grains dominate the agricultural landscape. They are grown by over 90 percent of farming households across the country and are both the main source of calories and income for the rural population. Maize alone covers over 60 percent of the total cropped area and represents over 50 percent of the average calorie intake of the population4. These trends underline the importance of efficient grain markets for welfare, by ensuring access and affordability for consumers and the right price signals for producers. Zimbabwe's grain production levels have declined markedly over the past years, and imports have come to play a growing role in domestic supply. In spite of Zimbabwe having had a long history as a surplus producer in the southern Africa region, both yield and production levels have declined in recent years as the sector evolved from a large-scale commercial farming structure to one dominated by low yield smallholder production (figure 2). This trend, combined with periodic 4ZIMSTAT, 2012. Poverty Income Consumption and Expenditure Survey 5 drought episodes, has opened a deficit in the national grain balance sheet. Zimbabwe's persistent grain deficits are plugged by imports, mostly from South Africa and Zambia. During the past decade, imports as share of maize requirement have averaged approximately 20 percent. Figure 1: Contribution of Agriculture to GDP Figure 2: Trends in Maize Production (m t) 14% 12% 2500000 10%000000 8% 1500000 *6% mnc 500000 2% 0 0% 0 .4 kae W 2009 2010 2011 2012 2013 2014 ma2 cn En m m 1m 1 a) 0 Source: ZIMSTAT Source, ZIMSTAT 3 DATA AND METHODOLOGY The analysis is based on monthly and weekly grain price series collected at main provincial markets. Table 1 presents a summary of the dataset and its sources. Maize prices series cover all ten provinces of Zimbabwe, whereas the selection of provinces for other commodities is guided by the areas for which sufficient data was available. These tend to be the areas where the commodity is produced and most frequently traded5. Table 1: Commodity Price Data Sources Commodity Frequency Years Source No. of provinces selected Maize grain Monthly 2010-2014 World Food Program All provinces Maize meal Monthly 2010-2014 FEWSNET All provinces Cowpeas Weekly 2012-2014 Ministry of Agriculture 8 provinces Sugar beans Weekly 2012-2014 Ministry of Agriculture 9 provinces Groundnuts Weekly 2012-2014 Ministry of Agriculture 7 provinces Sorghum Weekly 2012-2014 Ministry of Agriculture 5 provinces In terms of methodology', the analysis of market integration relies on a vector error correction model (VECM) to assess spatial cointegration between markets. Two variables are s Any missing observations were computed using a filling algorithm, 6 http:/foodprices.vam.wfp.org/Analysis-Monthly-Price-DataADV,aspx Annex I presents a detailed description of the methodology applied in this study. 6 said to be cointegrated if they tend to move together in the long-run. Although deviations from the equilibrium may be observed in the short-run, these are corrected in the long-run if the price transmission mechanism is functioning and hence, the markets are said to be cointegrated. Unlike many other studies which analyze the price transmission between domestic and world markets of a given commodity (in this case the standard hypothesis is that world prices affect local prices), this analysis investigates spatial price transmission between all possible combinations of domestic markets at the provincial level. The VECM approach is appropriate if two conditions are met. First, each price series is non- stationary and integrated to degree 1, written as I(1). That is, the variable (commodity price) is a random walk but the first differences (commodity return) is stationary or 1(0). Second, the variables are cointegrated, meaning that there is a linear combination of the variables that is stationary. If the time series of prices of a given market is not unit root, by definition, no cointegration can be tested. Such market will be excluded from the VECM estimation. Hence, for each pair of markets the analysis consists of three steps: L Unit root tests: to check whether the price process is stationary. These tests are the Augmented Dickey-Fuller (standard, with drift and with trend), and the Phillips-Perron test (standard or with drift). ii. Cointegration tests: the Engle-Granger and Johansen tests are used to determine whether the two series are cointegrated. This indicates whether there is a long-run relation between these two time series. iii. VECM: the model is estimated only for cointegrated market pairs to estimate the speed of price adjustments between them. 4 MARKET INTEGRATION PATTERNS The cointegration analysis detects the existence of a long-term relationship between markets and estimates its strength. Table 2 presents a summary of the results8. Three main indicators are presented. The first is the existence of a cointegration relationship between a market pair for a given commodity9. This identifies the existence of a long-term relationship. The second is the long- term elasticity, which indicates the strength of price transmission in the long-term. The third is the speed of price adjustment between markets. The full results of the analysis are presented in annex 11. 9 Cointegration is accepted provided that either the Engel Granger test or the Johansen tests identifies a long-tern relationship, 7 Maize markets The cointegration results indicate that maize grain markets are characterized by segmentation between the northern surplus and the southern deficit production regions0. In general, maize grain prices in deficit production areas are not integrated with surplus areas. Although integration is found within the two zones, grain prices in low production areas such as Matabeleland North and South do not have a long-term relationship with prices in the main producing areas of Zimbabwe, These regions are concentrated in the northern and eastern parts of the country. In effect, the results show two separate enclaves, with grain moving within the surplus and deficit producing areas but not between them (figure 3). Overall, 41 percent of the market pairs for maize grain are integrated. The average long-term price transmission elasticity for the integrated markets is very close to full price transmission. The speed of adjustment between these market pairs is also high compared to the other commodities in the sample (table 2). The only link between the surplus and deficit regions is through the Harare - Bulawayo trading corridor. Although surplus and deficit markets are segmented, the prices series for Harare and Bulawayo are integrated. This result reflects the aggregator role of Harare, where much of the maize produced in the northern surplus areas is collected for milling or for shipment to Zimbabwe's second trading and processing hubs in Bulawayo. They also reflect the structure of Zimbabwe's road network, whereby the main provincial hubs in the southern deficit regions are connected to northern surplus production areas through Harare. 10 Surplus/ deficit regions are defined as those with cereal production in excess/ deficit of internal requirements on average over the last three crop seasons. See annex IV for average production gaps by region. 8 Figure 3: Segmented Integration Pattern for Maize Grain Harare itemrAd&hnd North Bulawayo The segmentation extends to regional integration, whereby surplus maize grain markets are integrated with South Africa whereas the deficit regions are not. The cointegration between Zimbabwe and South Africa's maize grain markets is not surprising given that South Africa is the main source of imports for closing Zimbabwe's production deficit. The liberalization of Zimbabwe's maize market in 1990 has facilitated trade and integration with this major regional market, What is less expected is the absence of a long-term relationship between maize prices in Zimbabwe and Zambia. This may reflect the characteristics of the Zambia maize markets which might not be efficient and integrated with regional and world markets." In contrast, the markets for maize meal are relatively well integrated. Maize meal is one of the most efficient markets in the sample with full long-term elasticity and the highest speed of adjustment amongst the selected commodities. This is not surprising. Maize meal is the most widely consumed staple food. Its production and distribution infrastructure is well developed with processing hubs in both surplus and deficit areas and the co-existence of large and micro milling industries. However, these results should be treated carefully when making comparisons, given that domestic maize meal movements are determined by the distribution decisions of a handful of large processors. " Myers and Jayne (2012), Minot (2010, 2011). 9 Given the major role played by maize in Zimbabwe's food basket, the segmentation of this market has direct implications in terms of food security and demand stimulus for agricultural production. Although surplus and deficit regions are connected through the Harare- Bulawayo corridor, the maize traded through this channel is mostly destined for the milling industries that serve these large urban hubs as opposed to the small markets that serve vulnerable and food insecure rural households2, Rural households in deficit areas rely largely on unprocessed maize grain for their needs during pails of the year, and are more likely to purchase maize form their neighbors than from local retailers or markets. They are also more likely to report difficulties in sourcing maize for their basic consumption needs than in other regions (figure 4 and 5). Therefore, the integrated markets for maize meal do not necessarily counteract the inefficiencies stemming from segmented grain markets and their impact on growth and food security. Figure 4: Prevalence offood insecurity Figure 5: Percent of households stating maize is readily available (Percent of households) (Percent of households) 35 32 4539 30 40 35 315 25 30 20 25 15 20 15 5 5 0 0 Deficit Surplus Surplus Deficit Source: ZlMVAC Rural livelihoods assessment, 2014. Source: ZIMVAC Rural livelihoods assessment, 2014. Other commodities The other staple food markets are relatively well integrated in the areas where they are traded, with the exception of sorghum. Unlike maize, small grain markets are more narrowly transacted making the regional coverage of small grains in our sample less broad. The market coverage for these commodities in the sample tends to be concentrated in the drier regions of the country, with the exception of sugar beans which are more widely traded. In those areas, small grains show robust levels of cointegrationl3. On average, 78, 60 and 52 percent of the market pairs for groundnuts, cowpeas and sugar beans are cointegrated. The exception is in the sorghum market, where only 17 percent of market pairs are cointegrated. Although fairly well integrated in the 12 Kapuya et al (2010). 13 See annex III for mappings of cointegration relationships for the other commodities. 10 observed areas, small grains have slightly lower long-term elasticities and slower speeds of price adjustment than maize, with sorghum registering the slowest rates of adjustment. Sugar beans are the only commodity in the sample, other than maize, which is regularly traded across surplus and deficit areas. Unlike the markets for maize grain, sugar bean markets are integrated between surplus and deficit zones. This suggests that the observed segmentation maybe a phenomenon that affects maize more acutely than it applies to other tradable food crops. Table 2: Summary of cointegrated and VECM results (average)14 e pTime needed to close Lontermice Speed of price 50 percent of the % of integrated pairs transmission adjustment" price gap between elasticity5 markets17 Maize grain 41% 1.01 0.53 1.0 month Maize meal 60% 0.99 0.76 1.8 months Cowpeas 60% 0.92 0.33 2.4 weeks Sugar beans 52% 0.98 0.32 9,1 weeks Groundnuts 78% 0.90 0.44 2.7 weeks Sorghum 17% 1.08 0.24 5.7 weeks 1 See annex IV for the results for each estimated market pair. 1 Average for integrated pairs only. 16 This is the average combined speed of adjustment in absolute terms for the integrated pairs only. 1 These estimates are based on the half-life measure, which estimates the amount of tine (months for maize and weeks for other commodities) needed to correct 50 percent of the deviation in prices between market pairs. 11 5 DETERMINANTS OF MARKET INTEGRATION To explore the determinants of market integration, the analysis turns to estimating the role of distance, production and population as driving factors. Having determined the levels of spatial integration between the main provincial markets for the selected commodities, the analysis was extended to consider the extent to which distance between markets and the respective levels of demand explain the observed patterns of integration'8 and the speeds of adjustment between market pairs. The estimations focused on maize grain given that maize is the commodity in the sample with the widest regional coverage, and to investigate the observed segmentation in this market. Three explanatory variables were considered (i) distance between markets on major roads; (ii) difference in production between market pairs as a share of their joint total output9; and (iii) difference in the provincial population between market pairs as a share of their joint total population. The first variable represents transport as the main transaction cost in the domestic marketing of maize. It is expected that shorter distances between market pairs would increase the level of integration between them. The second and third variables are proxies of respective levels of demand for maize between two areas. The estimated model (OLS linear regression) is as follows and the results of the estimations are reported in annex IV: Cointegration, = f3z distancei + P2 production difff9 + j3 population diff.. + e The results indicate that distances have significant and negative effect on the levels of market integration for maize grain. In contrast, differences in production levels between market pairs and the respective population shares were not found to have a significant effect. These results partly confirm the observed pattern of market integration, whereby the integrated markets are clustered within two distant zones, one in the northern fertile regions and the other in the southern arid areas of Zimbabwe20. The distances between these two areas are large. In terms of the speed of adjustment, the explanatory variables were not found to not have an effect. Overall, these results suggest that although distances have the expected effect on market integration, distance, production and populations have relatively low levels of explanatory power for market integration2l. Factors other than these are likely to be important determinants of both the level of integration and the speed of adjustment between Zimbabwe domestic markets, Some of these factors are discussed below. 1 As measured by the test statistic of the cointegration test between market pairs (Johansen's trace statistic). 1 The proxy for the production differentials is computed as follows: Production = %PA+%Pf where %PAand %Padenote the annual percentage of a given commodity production over the total production for the two provinces. 20 Annex VII presents a map and description of the agro-ecological zones of Zimbabwe. 21 The adjusted R-square is 0.1 12 Other factors that influence the levels of integration The formal marketing chain has evolved in favor of aggregation and distribution rather than arbitrage. Zimbabwe's formal grain marketing and processing industries absorb a large share of total production. The industry is composed of large buyers that import maize and that purchase it on a wholesale basis in domestic markets, and large processors that mill grain to distribute it through national networks. The industry also includes some firms that are vertically integrated across both segments. The main function of these firms in the domestic marketing chain is to consolidate grain in volumes large enough to feed the large scale milling industries. In other words, they aggregate grain in the most cost effective way and process it for delivery to urban and semi/urban consumption hubs. Arbitrage, which involves the movement of grain between surplus and deficit areas, is the domain of a smaller, less organized group of agents: small agro dealers and traders. Arbitrage through informal channels is limited by the high transaction costs faced by small traders. The spatial arbitrage between surplus and deficit areas relies largely on the operations of small itinerant traders and agro dealers. Unlike the large formal industries, these agents are fragmented, dispersed and under-capitalized. They typically trade in small volumes at high transaction costs relating to search, financing and transport, amongst others. Their margins are low and they face high entry costs associated with the administrative procedures and fees for small traders. An important example is a requirement to obtain an annual permit at a cost of USD 1,000 for buyers, brokers or traders, which is doubled to USD 2,000 in case of late registration22. This fee alone represents a large cost and an obstacle to market entry for small traders. The late registration penalty also limits flexibility to enter the market for traders who are likely to operate on an opportunistic basis as they become aware of seasonal opportunities for arbitrage or free-up operating capital during the course of the year. Other transaction costs are also likely to be important, particularly those related to transport. Taken together, high entry and transaction costs and low margins will limit the number agents engaged in spatial arbitrage. These factors also dampen the level of competition governing the first transaction in the domestic marketing chain at the farm-gate, to the disadvantage of small farmers that have limited information and access to markets. The national production deficit and thin markets for small grains limit the volume of trade and dampen the potential for integration between markets. Zimbabwe has transitioned from being a surplus to a deficit producer of food over the past two decades. It posted a national deficit for grain in three of the past five years (figure 6). Hence, the shortage of production may be one of the factors limiting the volume of trade and integration between surplus and deficit regions, particularly since a large share of surplus production in the northern regions is absorbed by the large urban population of Harare. Moreover, most food commodities (except maize) such as small 22 These charges are instituted through the Agricultural Marketing Authority's by-laws (2013). See annex VI for an extract of the by-laws with a summary of the charges. 13 grains, pulses and oil seeds operate in thin markets that are characterized by low production volumes and limited marketing activity. These crops represent between 10% and 14 % percent of household production3, and households that grow these crops are more likely to produce them for own consumption. Only a very small share is traded (figure 7). Small grains and pulses are also less represented in the commercial farming portfolio and in trade flows. It follows that thin markets are less integrated and are more exposed to local demand and supply fluctuations. Figure 6: Zimbabwe's grain balance Figure 7: Thin markets for small grains Metric Tonnes Main marketing channels 400 200100% 0 60% 40V9 74.8 8. -20020 20% -400 0% Maize Small grains -600 -800 Other households in the area Private traders O0MB N Local Markets -1,000 :Local Millers Distant Markets 2010 2011 2012 2013 2014 MOther Source: ZIMVAC. Source: Z[MVAC Rural livelihoods assessment, 2014. 6 CONCLUDING REMARKS Using spatial price transmission analysis, this paper empirically measures the levels and patterns of integration for some of Zimbabwe's most important domestic product markets. Weak domestic market integration is indicative of slow spatial arbitrage. In the absence of robust market linkages, prices are vulnerable to local availability and may fluctuate widely if domestic markets do not respond by increasing supply during times of scarcity or by exporting surplus during times of plenty. Weak integration would also contribute to widening the spread in prices between surplus and deficit areas as prices in low production areas continue to reflect tight supply conditions. The analysis has shown that Zimbabwe's domestic markets for maize are segmented between surplus and deficit areas. Although regional markets tend to be integrated within these zones, the linkages between them, as indicated by systematic price co-movement, are absent. It is a finding that is indicative of a failure in arbitrage and weak trade flows between these two zones. This matters not only for production incentives and agricultural growth, but also for vulnerability and 21 ZIMVAC Rural livelihoods assessment, 2013. 14 food security in some of Zimbabwe's poorest regions. The markets for the other small commodities reviewed show more robust integration patterns across surplus and deficit regions. The policy note highlights four main drivers of the observed segmentation in domestic maize markets. (i) a domestic marketing structure that is strong on aggregation and retail distribution but weak on arbitrage; (ii) high transaction and market entry costs facing small traders who are main agents for arbitrage; (iii) costly domestic transport given the large distances between surplus and deficit zones; and (iv) low national production levels and thin markets for domestic commodities other than maize. The note concludes by discussing potential interventions and priorities that could ease the obstacles to spatial market integration. First, liberalizing market entry for small traders, agro dealers and brokers would reduce their transaction costs and strengthen their role in facilitating spatial arbitrage. The current set of administrative fees required of these agents increases their transaction costs and dampens their role as the primary agents for spatial arbitrage. Waiving all licensing and permit fees, or reducing them to a nominal amount, would increase the number of active traders and their flexibility to enter the market in response to arbitrage opportunities. Other interventions that increase their access to finance and information, and that deepen their linkages to the formal marketing chain would also strengthen this segment of the domestic market. Second, prioritizing rural roads and transport costs to and within deficit areas would further reduce transaction costs and deepen integration. Prioritizing infrastructure investments in these areas would tangibly reduce transaction costs for domestic marketing, resulting in better trade linkages and wider spatial arbitrage. Interventions that reduce the incidental costs of road transport such as reducing toll fees and the costs associated with controls points would also contribute to this result. Third, deepening the focus of key agricultural programs on stimulating production in deficit areas would ease local food shortages and help mitigate the effects of market segmentation. Raising agricultural production and productivity has been a priority issue for Zimbabwe, and the recommendations to attain this goal are addressed in numerous policies, studies and textS24. Amongst these, the findings of this paper point to the importance of increasing productivity in deficit areas, especially for the production of small grains given the comparative advantage of deficit areas in the production and marketing of these commodities, Lastly, the analysis revealed differences in the extent of spatial integration between maize (and sorghum) and the other small commodities. Various factors could explain these results. Firstly, unlike maize and sorghum, the marketing chains for small grains do not have large formal processing industries, Hence, their marketing networks rely more heavily on informal channels and small traders that look to arbitrage for profit. Second, the marketing profit margins of each 4 See World Bank (2012) and World Bank (2014) for recent analysis and references to on this body of work. 151 crop are an important factor in determining the incentive to trade in small grains. Higher margins would raise the returns from trading, and support integration across regional borders. An analysis of these margins by crop may yield further insights to these underlying dynamics of market transactions. Lastly, the marketing of maize grain is a delicate issue in Zimbabwe, and is subject to heavy attention by the authorities, Hence, its spatial market integration may be dampened to the extent that it is subject to more burdensome administrative controls. Further analysis of these questions that, combines quantitative value chain analysis with and qualitative assessments of market regulations at the local level would yield further insights to the constraints to stronger spatial integration in Zimbabwean markets. 16 REFERENCES Sekhar, C.S.C,(2012). Agricultural market integration in India: An analysis of select commodities. Food Policy. Volume 37, Issue 3, June 2012, Pages 309-322. Jacks, DS., O'Rourke, KH., Williamson, J.G., (2011). Commodity price volatility and world market integration since 1700. Review of Economics and Statistics-, Vol. 93, No. 3, Pages 800- 813. Myers R.J and Jayne T.S (2012). Multiple-Regime Spatial Price Transmission with an application to Maize Markets in Southern Africa. American Journal ofAgricultural Economics. Vol 94.1. pp 174-188. Minot, N.,(201 1). Transmission of World Food Price Changes to Markets in Sub-Saharan Africa. IFPRI Discussion Paper 01059. Minot, N.,(2010). Paper to be presented at the COMESA policy seminar "Food price variability: Causes, consequences, and policy options" on 25-26 January 2010 in Maputo, Mozambique under the COMESA-MSU-1FPRI African Agricultural Markets Project (AAMP). World Bank (2012). Recovery and Growth of Zimbabwe Agriculture. Zimbabwe Growth Recovery Notes Series. World Bank (2014). Zimbabwe's Food Grain Economy. ZIMSTAT( 2012). Poverty Income Consumption and Expenditure Survey. Government of Zimbabwe. ZIMVAC (2011). Rural Livelihoods Assessment. Food and Nutrition Council. Zimbabwe. ZIMVAC (2013). Rural Livelihoods Assessment. Food and Nutrition Council. Zimbabwe. ZIMVAC (2014). Rural Livelihoods Assessment. Food and Nutrition Council. Zimbabwe. 17 ANNEXES I: Methodology II: Cointegration analysis results III: Cointegration maps IV: Regression results V: Surplus/ deficit regions (2008/09 to 2013/14 average) VI: Agricultural Marketing Authority trader/ buyer registration fees VII: Natural regions of Zimbabwe I: Methodology Unit root tests for the order of integration in the price series (ADF, Phillips Perrmn and KPSS) Test the null of no cointegration There cannot be cointegration for betwcou prices in different markets Accept this market. Stop here. (Engle-Ganger and Johanaen) Rejee Specify and estimate VECM, assess dynamics of speed of adjustment overalltransmission and market integration Unit root tests A time series y, is said to have a unit root if it follows the autoregressive process y1 = yI + e, where e, is a white noise. The process y, is said to be 'non-stationary' as mean is not constant over time and variance is diverging to infinity with t --> o. Augmented Dickey-Fuller test This test, dubbed ADF test, is the extension of the Dickey-Fuller test allowing for more lags to be considered. Consider an AR(3) process y,= O y1+ 62y,2+ e3y,3+ e, Rewriting the model as 19 A y,= 0+ Pt+ TTy-t+ c1A y,-+ c2 A y,2+ e, under the null hypothesis of unit root we have that Ho :fw = Oagainst H, :r<0 The t-test on HGis called Augmented Dickey-Fuller test. Allowing a and 0i to be different from zero is the version of the test which include drift and trend, respectively. The optimal number of lags to be included can be selected relying on some goodness-of-fit criterion, e.g. AIC or BIC. Phillips Perron test Whereas the ADF test involves estimating (assuming no drift/trend/lag) A y,= ny,. + e, the Phillips Perron test (PP) estimates y1= p yt_[+ el In the ADF test e, is 1(0) and can be heteroskedastic. The PP tests correct for any serial correlation and heteroskedasticity in the errors e, non-parametrically by modifying the Dickey Fuller test statistics. One advantage of the PP tests over the ADF tests is that the PP tests are robust to general forms of heteroskedasticity in the error term e,. Kwiatkowsld, Phillips, Schmidt, and Shin test This test, known as KPSS test, differs from the previous two as the null hypothesis is now that the process is stationary, i.e. 1(0). Assuming no trend is present, the starting point of the test is the process Y, e, where e, is stationary and is a random walk, i.e. +v,, v, is IID(0,a2). If a, = 0 then 4 = o hence y, is a stationary process. 20 The simple regression y= +e2, can be used to get an estimate of the stochastic component. Under the null hypothesis, e, is stationary. The test reads Ho:o = 0 againstH:o- >0. The test statistic is given by lS' KPSS=- where S,= Z is a partial sum and ca is a HAC estimator of the variance of F,. Cointegration tests As we are analyzing two prices at the time, the cointegration relation reads: P1 = a + fP2 + e or P - a - #P2 = e where cis a stationary process. Engle and Granger (1987) Consider the following equation; Pit P2 + Ut If u, is non-stationary, then p, - fp2, is not a cointegrating relationship. Engle and Granger suggested estimating the above relation via OLS and then testing for the presence of a unit root on the estimated residuals ii, in order to test the null of no cointegration. Johansen (1988, 1991) Consider at VAR of two variable p,,and p,, and two lags, i.e. a VAR(2) 21 This VAR(2) has Vector Error Correction (VECM) representation n1 (A,- A,- I1) l Pv 2, t- I (V2 The rank of the matrix (A - A2 1) is equal to the number of cointegrating vectors. If it is 2, both variables are stationary. If it is 0 the non cointegration relation is present. If it is equal to 1 then the variables are cointegrated. With two price variables, cointegration can be assessed by testing the significance of the characteristic roots or eigenvalues of (A, - A2 - I). If the variable are cointegrated then 0 < , Midlands No No 0.86 MfNorIh Midlands No No 0.83 MatSouith Midlandå No No 0,91 26 Cowpeas Loention Long-run RelaionsMllp Adjesment MCombined Av eage Market __ _Mar_ket 2 Grge Jolanseni Long-term Elasticity Adostmet HafLife Caintegrated magrkcels -A... Midlands Manicaland No Yes 0.87 0.56 1.38 Midlands MatNorth No Yes 0.75 0.51 2.46 MashkCentral Musviiga No Yes 1 0.35 3.66 MashCentral MatNorth No Yes 0.86 0.32 L49 MatNorth Masvingo No Yes 1,17 0.15 2.83 Midlands Masvingo No Yes 6.81 0.06 2,65 Notcointegrated Mashecntral Midlands No No 1.13 MashCentral Manicaland No No MatNorth Manicaland No No Ll.6 Masvingo Manicaland No No 0.96 Sugar beans Locaton Long-Run Relationship Adjustment 1 h Combined Average Markek 1akt2grner Johlansen Logtr lsticity Aditstmlent Infillife Cointegrated fiarkets Harare MashWest No Yes 1.03 0.68 0.87 MashCentral Masvinga No Yes 0,98 0,52 2.6 MashEast Masvingo No Yes 0.98 0.51 5.09 Hware Midlands No Yes [.01 0.39 1.62 MashEast MatNorth No Yes 0.94 0.34 4,13 MashCentral MatNorth No Yes 0.94 0.31 2.2 MasCentral MashWest No Yes 1 0,24 5.74 1Harare MatNortli No Yes 0.97 0.23 3.06 MastEast Midlands No Yes 0.98 0.22 6.02 HTarare MashEast No Yes 1.03 0.11 16.72 Midlands MatNorth No Yes 0.95 0.01 52,02 27 Notcointegrated Harare MashCentral No No 1.02 Harare Masvingo No No 1 MashEast Mashceontral No No Maslast MashWest No No MashWest Midlands No Nu 0.98 MashWest MatNorth No No 0.93 MashWest Masvingo No No 0.97 Midlands Masvingo No No 0.98 MatNorth Masvingo No No 103 MashCentral Midlands No No 0.98 Groundnuts location Long-Run Relationship Adjustmient Maret LoaEng 2Le -a RainsiCominaed Average Marketr Markel Grangle Jolinsen Long-term E[ästiity Adustment malfLifee Coiniegrated nfmlcets Hurare MatNorth No Yes 031 0,67 172 Harare Manicaland No Yes 0.82 0.63 1.32 Harora Harare No Yes 0.96 0.51 1.08 MatNorth Manicaland No Yes 1,13 0.44 3.3 Midlands Maniealand No Yes 1 0.34 2.24 Vlarare Midlands No Yes 0,82 014 5.02 Not cointegrnted ¯ l . Harare Midlands No No 0.8 Harare Manicaland No No 0.85 Midlands MatNorth No No 0.89 Sorghum Location Long-r,un Relationship A djuis;tment Eigel iComined Average Market 1 Market2 «rgl er Johansen Loig-termnisticily Adjustment Halflife Cointegrated markets Masvingo No Yes L08 0.24 5.68 Not colntegrated MashWest Midlands No No 0.9 MashWest Masvingo No No 0.91 MashWest Manicaland No No 1.01 Midlands Masvingo No No 1.02 Midlands Manicu[and No No .1 28 Zimbabwe maize prices integration with regional and world prices Lcation Long-Run Relationship Adjustnent Engle- Combined Average Mfrket Arket2 Grnnger o I ng-termELISty Adjustment Hale-Life Cnitegrated markets - South.africa Zim.grain No Yes -0.47 0.12 12.2 SouctIafrica Zim.grain,surplus Yes No -0.64 0.23 4.1 World.grain Zim.grain No Yes 1.3 0.02 72.6 Notcolnigrnted World.grain Zim.grain.surplus No No 1.67 Zimbabwe.grain Zim.grain,deficit No No 1.76 Zimbabwe.grain Zim.grain.surplus No No 1.3 Zimbahive,grain.d Zim.grain,surplus No No 0.54 eficit Zambia South.africa No No 0.09 Zambia Zim.grain No No -0.19 Zambia Zim.grain.defcit No No -0.33 Zambia Zim.grain.surplus No No -0,24 Southafrica Zim.grain.deficit No No -0.5 29 III: Cointegration maps Groundnuts Cowpeas Sorghum Sugar Beans 30 IV: Regression results Malle Graink Maze mnal Cointegration Speed of Cointegration Speed of adjustment adjustnent Distance 0.01183* -0.00049 Distnce 0.4306 -0,00798 (-2,362) (-1.429) (0.5587) (-0.167) Population -2.025 -0.04438 Population -3.2429 -0.42239 {.0.335) (-0.088) (07556) (-0.1.577) Production -1.96813 -0.14258 Production -2.5827 -0.25359 (-0.755) (-0.719) (-0,6267 (-1.069) Constlant 23.2501*** 0.81198*** Constant 20.9293*** 0.9582*** (9.933) (5.560) (6.1070) (4.391) R Square 0.1518 0.2261 R Square 0,1304 Adjused R Squarc 0,1009 0.06021 Adjusted R Square 0.03381 -sltat 2.982 1.363 F-sts 1.35 31 V: Surplus/ deficit regions (2008/09 to 2013/14 average) eProvince e..CMaize ProutIon .Ince Surplus Deficit Mashonaland West 260.661.20 164,962,00 95,699,20 SurpuS Mashnaland Central 219,499.00 128,674,00 90,825.00 Surplus Midlands 204,144.40 182,060.00 22,084.40 Surplus Mushonaland Eust 155,736.20 151,522.00 4,214.20 Surplus Matabelciand North 63,981A0 82,646.00 (18,664.60) Deficit Matabeleland Souith 51,960.40 76,111.00 (24,150.6 Deficit Manicaland 152,439.40 197,321.00 (44,881.60) Deficit Masvingo 88,717,20 167,475.00 (78,757.80) Deficit 32 VI: Trader/ buyer registration fees Agricultural Marketing Authority (Grain, Oilseed and Products) By-laws, 2013 SECOND SCHODULe (Section 6) FEES Form No. Form description Fee US$ AMAG Application for registration as Grain Industry Stakeholder Association On late registration 500,00 AMAG 2 Application for registration as a contractor or 1000.00 processor On late registration 2000,00 Application for registration as a buyer, AMAG 3 1000,00 broker or trader On late registration 2000,00 AMAG 4 Application fbr registradon as a grower 1,00 On late registration 2,00 AMAG 5 Application for service hammer miller 200 On late registration 5,00 33 VII: Natural regions of Zimbabwe25 Mbpite /fr LPpane Nky lays MatakeWtea North . Legend ' i Nature regions uili f ttec productrioin.esm rsna II Dstrk budr Iebrkie '"' M ',Aas RIgo JIM ncal onidat h rr7010m prnu Sr eaieylw h eini uitable for intesry farming, suchrestry, teabacffee,fruittbee and livetzek Region IIA1: Rainfall confined to summer and is moderately high, ranging from 700-1050 mm per annum. Howevert the region experiences shoit rainy seasons with severe dry spells during the rainy season. Suitable for intensive farming based on maize, tobacco, cotton and livestock. Region 1I: Rainfall is moderate, ranging rom 500 - 800 mm per annum. In addition, the region experiences are relatively high temperatures and infrequent heavy rainfalls. Subject to seasonal droughts and severe mid-season dry spells. This region is a semi-intensive farming region suitable for livestock production, fodder craps and cash crops under good farming management. Region IV: The region experiences fairly low rainfall in the ranges of 450-650 mm per annum. Seasonal droughts severe dry spells during the rainy season are frequent in this region. Suitable for semi - extensive faring systems based on livestock, resistant fodder crops, forestry, wildlife and tourism. Region V: Receive low and highly er rati infall below 450 mm per annum. The topography and soils of the region are poor. The region is suitable for extensive cattle ranching, forestry, wildlife and tourism. Zambezi Valley was infested with tsetse fly, which is largely eradicated. 25 Source: WFP. Zimbabwe: Results of exploratory food and nutrition security analysis, October 2014. 34