contents Acknowledgements............................................................................................................................................................................................ 1 Foreword................................................................................................................................................................................................................ 2 Digitizing Agriculture: Evidence from E-Voucher Programs in Mali, Chad, Niger, and Guinea................................................. 5 Special Topic: Improving projects performance with cost-effective Iterative Beneficiary Monitoring.......................... 17 Macroeconomic Indicators of AFCW3 Countries at a Glance, 2014-2019..................................................................................... 38 Chad...................................................................................................................................................................................................................... 39 Guinea................................................................................................................................................................................................................... 42 Mali........................................................................................................................................................................................................................ 45 Niger...................................................................................................................................................................................................................... 48 AcKNOWLEDGEMENTS This report is a collective effort prepared under the direction of José López-Calix, Lars C. Moller, Marianne Grosclaude and Andrew L. Dabalen, with the written contributions of an interdisciplinary Bank team composed of Amadou Ba, Rhoda Rubaiza, Andre-Marie Taptue, Aly Sanoh, Aboudrahyme Savadogo, Olivier Béguy, Susana Sanchez, Boulel Touré, Luc Razafimandimby, Marcel Nshimiyimana, Markus Kitzmuller, Olanrewaju Malik Kassim and Irum Touqeer. Soukeyna Kane, Christophe Lemiere, Madio Fall, Francois Nankobogo, Joelle Dehasse, Rachidi B. Radji, Michael Hamaide and Boubacar Sidiki Walbani provided strong encouragement, helpful guidance, information and comments. The analysis of e-vouchers pilot experiences and first set of iterative beneficiary monitoring (IBM) were prepared as part of global work programs on agricultural and poverty issues Bankwide. Special thanks are due to Flore Martinant, Irina Shuman who peer reviewed the paper on e-vouchers; and to Kristen Himelein who did so on the paper on IBM. Compiling the report would not have been possible without the editorial assistance of Maude Jean-Baptiste, media support from Anne Senges, Dasan Bobo and Edmond Bagde Dingamhoudou, and administrative support from Micky Ananth, Mariama Diabate-Jabbie and Jeanne D’Arc Edima. Precious editorial and composing work was timely executed by Valerie Bennett, Maria Deverna, Pia Decarsin and their translation and editing colleagues at JPD systems. 1 foreword The agricultural sector is a significant contributor to the economies of Guinea, Mali and Niger. In terms of share of GDP, agriculture represents 16 percent, 38 percent, and 40 percent respectively. It is also a primary source of employment for most of the population (68 percent, 57 percent, and 75 percent respectively). Unfortunately, such critical sector is characterized by low yields due to low input use and quality. The International Fertilizer Development Center (IFDC) estimated the average cereal yields to be 1.6 metric tons per hectare in 2014 in Mali, which is above the sub-Saha- ran average (1 metric ton/ha) but well below potential yields. With a majority of their land area lying in semi-arid and arid zones of the Sahel and the Sahara desert, increasingly frequent crises arising from the Sahel region’s high vulnerability to climate change--characterized by recurrent extreme weather events such as floods and droughts—low productivity has adverse effects on farmers’ incomes and undermines household food security. Hence the main article in the present volume is devoted to the innovative effort of introducing e-vouchers schemes, supported by digital means, in Guinea, Mali and Niger. The e-voucher program is built around three key components: a digital platform for SMS messages, a reliable database of electronically-registered farmers in selected regions, and a directory of agro-dealers. In so doing, fertilizers (or seeds) distribution becomes transparent, ensures high quality and as it unfolds, fosters private sector participation. Based on pilot practices in the sub-region, four major lessons are learned. First, targeting is a key determinant of scope and success. In general, targeting depends on the main objective of the program: either poverty reduction, which would aim at farmers located at the lowest deciles of the income distribution, or agricultural productivity, which would also aim at those located a few deciles above. A clear decision in this regard should be taken upfront. Second, in rural populations with high levels of illiteracy, digital technologies should be adapted to their needs, with voice messaging working better than SMS messages. Third, effectiveness of these programs relies on the efficiency of public procurement. In many cases, delays in timely providing agricultural input result from bottlenecks at the procurement stage. As AFCW3 governments become strongly involved in upscaling these approaches, it is my hope that this report will provide further insights to help them in their successful implementation. The second special article examines the pilot experiences at introducing Iterative Beneficiary Monitoring (IBM) as an innovative and cost-effective monitoring tool in Bank projects. An IBM is designed as a light, low-cost, independent, rapid, and iterative feedback loop approach. Based on a simple questionnaire and short field missions, it collects information on project performance directly from beneficiaries. The outcome is a short and sharp report on issues that must be addressed by project teams. In this way, IBM improve both project team proactiveness and beneficiary engagement and satisfaction through positive, self-reinforcing cycles of improvement. IBM were first developed in Mali, initially to deal with limited access to project activities due to insecurity, but due to its agile nature, it also became 2 attractive and applicable in secure settings. The range of projects applying IBM includes, inter alia, those on school meals programs, e-voucher schemes, and health cash-transfer cards distributed among the extremely poor. Results from monitoring these initial projects were shared with the project managers. The World Bank is currently scaling up the IBM approach to various projects in Mali, Chad, Guinea, Niger, Benin, the Central African Republic and Nigeria. In terms of the economic outlook, I am especially pleased to report that all countries are projected to experience solid positive growth rates in 2019, with Niger and Guinea at 6.5 and 5.9 percent respectively, Mali at 5.0 percent and Chad approaching 3.4 percent. Chad’s positive growth also follows the successful restructuring of its commercial debt. Macroeconomic stability, including low inflation rates, is also expected to be preserved through growth recovery in 2019. This is relevant in the context of rising oil prices. Inflation rates will remain below 3 percent and, in the case of Guinea, in single digits. In the meantime, downside risks originating either from domestic or external shocks or policy reversals will remain significant. In response, all governments keep implementing fiscal consolidation and growth- enhancing policies supported by International Monetary Fund (IMF) and World Bank budget support programs currently on track or under preparation. Finally, I want to remind our readers that this is the seventh edition in a series of reports dealing with key development issues in Chad, Guinea, Mali, and Niger. The AFCW3 Economic Update series is intended to foster public debate about key macroeconomic and structural developments in support of poverty reduction. The series also provides a broad analysis — even if the findings are preliminary and less than fully polished. In short, this series represents an innovative knowledge-sharing vehicle for the World Bank, and our AFCW3 region. Indeed, it is used to approach the media, civil society, universities and the public at large to discuss priority policy reforms introduced or debated in these countries.1 Hence, I would like to again express my gratitude to our governments and technical and financial partners for their cooperation and many joint contributions over the past few months. Their encouragement, inputs and technical advice have created an environment particularly well suited to a rich and regular exchange of views on development policy. I hope that this series will make it possible to deepen these discussions and move them into the public space to better inform and enable citizens to express their own views. Soukeyna Kane World Bank Director of Operations Chad, Guinea, Mali, and Niger It should be noted that the findings, interpretations, and conclusions expressed in this report are entirely those of World Bank staff, and do not necessarily represent 1 the views of the World Bank Group and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. 3 4 digitizing agriculture Evidence from E-voucher Programs in Mali, chad, Niger, and Guinea Amadou Ba (Sr. Agriculture Economist) and Rhoda Rubaiza (Sr. Agriculture Economist) March 2019 Abstract This paper proposes to review evidence emerging from the design and implementation of an electronic agriculture subsidy pilot system based on vouchers in West Africa. It presents the characteristics of the e-voucher programs, describes such programs in Guinea, Mali, and Niger, and draws lessons and recommendations from this experience. 1. INTRODUCTION Under their agricultural sector capacity-building programs, the governments of Guinea, Mali, and Niger distribute sets of improved technological packages in the form of subsidies (fertilizers, seeds, seedlings, small and mid-size equipment, small ruminants, and other inputs) to crop and livestock producers. The subsidies are aimed at assisting these producers in scaling up their farms against a backdrop of vulnerability to climate change, steadily lower soil fertility, and high demand for farm produce due to high population growth. Working under the auspices of their respective governments, the Ministries in charge of agriculture in Guinea, Mali, and Niger have implemented an electronic subsidy pilot system based on voucher codes (e-vouchers) to improve efficiency, transparency, and traceability in allocating these subsidies. Under the e-voucher program, an electronic database of beneficiaries (coding) was developed, and electronic vouchers are sent by SMS to beneficiaries specifying the subsidies granted to them along with information on locations where they will receive the agriculture inputs. The program also monitors how the subsidies are utilized and assesses their impacts on recipients. This pilot program has been supported in all three countries by the World Bank-funded West Africa Agricultural Productivity Program (WAAPP), whose development objective is to generate and accelerate the adoption of improved technologies in participating countries for priority commodities, as outlined in the agricultural policy (ECOWAP) of the Economic Community of West African States (ECOWAS). The objective of this report is to share the experience and lessons learned in setting up an e-voucher program in West Africa with policy makers and development partners looking to design or implement similar programs in order to improve the efficiency, effectiveness, and impact of input subsidy programs. 5 2. CONTEXT Guinea, Mali, and Niger are low-income countries located in West Africa. The agricultural sector is a significant contributor to their economy in terms of share of GDP (16%, 38%, and 40% respectively in 2017)2 and exports as well as a primary source of employment for most of the population (68%, 57%, and 75%, respectively in 2018).3 The agriculture sector in all three countries is characterized by low yields due to low input use. For instance, according to the Fertilizer Statistics Overview, Mali 2013-2016 report,4 although fertilizer consumption is growing, it is still quite low. The report found that 590,655 metric tons of fertilizer were consumed in 2016, which was nearly double the consumption in 2015. With growing input use, productivity has been increasing, but it remains low and below potential. The International Fertilizer Development Center (IFDC) estimated the average cereal yields to be 1.56 metric tons per hectare in 2014 in Mali, which is above the sub-Saharan average (1 metric ton/ha) but below potential yields. FIGURE 1: ANNUAL FERTILIZER CONSUMPTION IN MALI 5 HS Code Fertilizer Name 2013 2014 2015 2016 3102100000 Urea 169,514 128,963 131,562 219,405 3105400000 MAP 64,778 28,587 58,146 87,427 3104200000 MOP 34,513 43,082 54,180 82,905 3102210000 Ammonium sulphate 40,844 23,290 18,897 47,759 3105300000 DAP 5.384 3,875 4,354 39,448 Other Fertilizers 56,952 67,662 33,336 113,711 Total (MT) 371,985 295,459 300,474 590,655 Source: FTWG-MLI FIGURE 2: CEREAL YIELDS IN MALI, IN METRIC TONS PER HECTARE 6 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2 2018 World Development Indicators (WDI) 3 Ibid. 4  TWG-MLI in Africa Fertilizer Organization 2017. Fertilizer Statistics Overview, Mali 2013-2016. https://africafertilizer.org/wp-content/uploads/2017/09/ Source: F Mali-Fertilizer-Statistics-Overview-2016.pdf 5 Ibid. 6 A severe drought was recorded in 2011. 6 Mali and Niger are landlocked, with a majority of their land area lying in semi-arid and arid zones of the Sahel and the Sahara Desert. Increasingly frequent crises arising from the Sahel region’s high vulnerability to climate change, which are characterized by recurrent extreme weather events such as floods and drought, have had adverse effects on farmers’ incomes and undermined household food security. In the aftermath of the 2008 global food crisis, the governments of Guinea, Mali, and Niger launched agricultural development strategies focused on crop production intensification, including the “Nigeriens Feeding Nigeriens” (3N) Initiative, the Rice Initiative in Mali, and the Presidential Initiative on Agriculture in Guinea, to assist households affected by the crisis. These initiatives address low agricultural yields stemming from the inadequate use of farming inputs, including fertilizers and certified seeds, by subsidizing improved farming inputs in order to boost agricultural productiv- ity. Under their respective initiatives, the governments of the three countries developed input subsidy programs to rebuild the capacities of households impacted by the food crisis and recurrent climate shocks. 3. TRADITIONAL INPUT SUBSIDY PROGRAMS Of the three countries, Mali has the largest agricultural subsidy program. In 2018, subsidies granted were estimated at CFAF 45 billion (approximately US$80 million). In response to the global food crisis, the Government launched the Rice Initiative during the 2008–2009 crop year. Its objective was to boost local rice production by providing farmers with subsidized seeds and fertilizers as well as credit facilities for farm equipment and extension services. To further support the development of the sector, the Government enacted the National Rice Development Strategy (NRDS) 2009–2018. In 2009, the Government extended the rice Initiative to maize, wheat, millet, and sorghum in order to increase production, mainly through fertilizer subsidies. However, even though the national fertilizer subsidy program has increased agricultural production, several analyses show that the program has been targeting farmers who are well off instead of those who are poor. The higher subsidy allocation barely improved farming yields from the near-stagnation observed during the period under review, thereby raising the issue of farming input subsidy efficiency. Major issues arising from the implemen- tation of these farming input subsidy programs include: 1) the design and implementation phases of the subsidy program characterized by the predominant role of the public sector, with little to no room for leveraging private sector investments; 2) selection of suppliers of inputs (mostly seeds and fertilizers) through a tender process that was often not transparent; 3) although the main target population consists of smallholder farmers, in practice the targeting system is flawed and does not always guarantee that the subsidized inputs are received by the intended beneficiaries; 4) no formal assessment of performance and impacts of those programs was conducted on a regular basis by independent external bodies; 5) mismatch between quantities of inputs distributed and the surface area of farms owned by beneficia- ry households; and 6) lack of transparency and traceability in input allocation and distribution, thereby discrediting the program and creating delays and bottlenecks in the delivery of inputs to beneficiaries. Ultimately, under the traditional farming input subsidy system, large amounts of public resources were spent but at substantial risk of not achieving the expected outcomes due to a lack of transparency in distribution channels and the absence of a monitoring and evaluation system. Furthermore and of equally great concern, the intended beneficiaries of the subsidy are often underserving as a result of huge losses along the entire distribution chain. Consequently, to improve their input subsidy system, Guinea, Mali, and Niger opted to design a pilot subsidy system based on improved identification of beneficiaries and their needs, information and communication technologies (ICT), and private sector engagement as well as the tracking of operations and outcomes. Of note is the fact that the e-voucher program was launched against a backdrop of steady improvement in rural mobile telephone coverage in Guinea, Mali and Niger. 7 4. PILOT E-VOUCHER SYSTEM The e-voucher system is a digital platform-based solution for the distribution of farm inputs that instantly sends out electronic coupons (or vouchers) by SMS to the mobile terminals of a database of farmers (beneficiaries) notifying them of subsidies granted to them and where these can be collected. By so doing, it also allows for real-time monitoring and evaluation of input distribution to farmers. 4.1 Program Objectives The e-voucher program was launched in 2015–2016 in Northern Mali (Mopti, Timbuktu, and Gao) under the auspices of the West Africa Agricultural Productivity Program (WAAPP). Thereafter, it was extended to southern regions of the country and to neighboring countries, starting with Niger, where two e-voucher cycles were completed in 2017 and 2018, followed by Guinea in 2018. A feasibility study of the system is currently underway in Chad. In all three pilot countries, the goals of the e-voucher program were as follows: mproving transparency in farm input distribution. In Mali, the e-voucher program was launched against > I a backdrop where the distribution of inputs on the basis of traditional subsidies had revealed numerous limitations. One of several technical reasons advanced for the ineffectiveness of the classical subsidy approach is the involvement of several intermediaries, including village distribution committees, local elected officials, and government officials. In these circumstances, the e-voucher system could be seen as a more transparent solution for establishing direct contact between farmer and agro-dealer. >  Improving the targeting of the most vulnerable population segments in distributing inputs. Surveys conducted prior to e-voucher implementation helped identify the most vulnerable segments of the population as well as their input needs based on prevailing agro-environmental conditions where they operate. In Niger, this approach was implemented in close collaboration with the Bank-funded Social Safety Nets Project, whose database served as a launch pad for the e-voucher program.  ngaging the private sector in distributing high-quality inputs. Under the e-voucher system, a database > E of agro-dealers selected on the basis of stringent technical and commercial criteria was created for each country. > Improving monitoring and evaluation of subsidy utilization and impacts on users. 4.2 How does the e-voucher platform work? The system is built around the introduction and interaction of three key components: > Developing a digital platform; > Creating a reliable database of electronically-registered farmers; and > Creating a directory of agro-dealers. 4.3 What is the structure of the e-voucher platform? In the three countries under study, the Ministry of Agriculture is responsible for coordinating the various actors within the program, including telecom companies, farmers, and agro-dealers. Since the e-voucher system is SMS-based, an awareness and sensitization campaign was conducted to encourage farmers to register and not to delete SMS messages received from the platform telephone number. 8 A farmer receives an SMS with the details of the relevant agro-dealer, the amount and type of inputs the farmer is eligible to receive, and the name and contact number of the agro-dealer. The agro-dealer confirms the farmer’s eligibility and voucher via the system before delivering the inputs. Once the agro-dealer receives confirmation, the voucher becomes void. After the transaction, the farmer receives a message requesting confirmation of the amount and type of inputs received. Figure 3 below illustrates the e-voucher structure. FIGURE 3: E -VOUCHER PLATFORM IN MALI AND GUINEA (Example of Slimtrader model implemented in Mali and Guinea) Ministry in charge of agriculture Telephone Telephone Telephone company A company B company C Farmers > Step 1 Electronic Agro_Dealers Database > Step 3 Database > Step 6 Platform Step 7 > Step 4 > > Step 5 Step 2 > The above steps are described in Annex 1. 5. HOW WAS THE E-VOUCHER IMPLEMENTED IN GUINEA, MALI, AND NIGER? 5.1 Identification of beneficiaries Identification of beneficiaries is the starting point and a key component of the e-voucher implementation process. The main aim is to gather data on farmers, planted areas for each crop, and geo-referencing of farmland. Overall, it is important to note the time and resources needed to carry out surveys necessary for the identification of beneficiaries. In Mali, 250 survey officers were mobilized during the 2017–2018 campaign in order to electronically register farm holdings in a database over a 45-day period for the second e-voucher program. Guinea hired 210 young survey officers in the Upper Guinea region and trained them in survey techniques and electronic registration on the e-voucher platform. In Niger, the survey methodology was to identify the communities targeted following a review of the list of vulnerable communities drawn from the Crop Estimation and Forecast Survey (EPER), which was primarily aimed at identifying vulnerable households and using the data to populate the database of the Unit in charge of Social Safety Nets (CFS) with additional variables that would allow for the proper distribution of farm inputs using electronic vouchers. About 36,000 households were identified in 20 municipalities across Niger. 9 Below is a summary of results from the farmers’ census conducted during the 2018 rainfed crop season. TABLE 1: FARMERS IDENTIFIED FOR THE E -VOUCHER PROGRAM COUNTRY NUMBER OF LISTED FARMERS FEMALE % COMMENTS Guinea 68,520 36% The pilot phase was implemented in Kankan region, specifically in Kankan, Kouroussa, Mandiana, and Siguiri prefectures. Mali 93,054 20% This phase covered two districts in Segou region and two more in Sikasso region. Niger 47,678 21% Farmers in 20 municipalities across 4 regions. 5.2 Agro-dealer selection The same agro-dealer selection process was conducted in all three countries. An advertising campaign was held, followed by agro-dealer selection, which was conducted in line with public procurement procedures. To encourage the establishment of a network of local agro-dealers, suppliers are required to locate their stores within the target municipalities and districts so as to shorten the distance between the village and the place of delivery of their products. For example, the terms of reference for the suppliers lay emphasis on the quality of the inputs as the main requirement in delivering certified seeds in Niger. In the early phase of the program, suppliers were selected and paid under World Bank-funded projects. Following the improvements brought about in the farm inputs distribution process with the introduction of the e-voucher, the governments gradually took over the financing of the inputs in their national budgets. This positive development has been observed in Niger, where the Government financed some of the seeds distributed during the 2018 e-voucher operation from its own resources, as well as in Mali, where inputs are fully funded from the national budget. 5.3 Development of digital platform The initial design of the digital platform as well as codification of farmers and their organizations were carried out in both Mali and Guinea by an international firm specialized in developing such applications. With time, local start-ups in the region began developing similar e-voucher digital platforms in Mali and Niger. However, hosting and maintaining the digital platforms and servers was a thorny issue for these countries. Because some countries prefer that the platforms not be hosted or administered by private firms or outside the country’s borders, the national bodies responsible for implementing the e-voucher programs still lack the required technical capacities. In this respect, Niger and Mali are typical examples. In Niger, the database is administered by the Directorate General for Agriculture (DGA), which still lacks the necessary ICT skills, while in Mali, this responsibility lies with WAAPP, which is implemented by the National Council for Agricultural Research (CNRA). 10 Based on Mali and Niger’s experience, the Guinean authorities opted to entrust the management of the database to a relevant technical unit, namely the Directorate General for Information Technologies and the Digital Economy (DNTIEN). This agency is responsible for the general operation of the e-voucher system, practical utilization of the mobile application, and technical administration of Guinea’s e-voucher. Each of the appointed project administrators has a user name and a password used to log on to the project’s website (https://www.evoucher. gov.gn/fr) and access the farmers’ database, input subsidy data, and provide assistance to stakeholders where needed. 5.4 Contracts with mobile telephone operators and allocation of short codes To ease the sending of SMS messages between the digital platform and the various stakeholders (farming input suppliers and beneficiaries), the government agencies in charge of regulating the telecommunications sector have assigned short codes for the exclusive use of the e-voucher (namely 36026 in Mali and 4042 in Niger). In Niger, only one mobile operator was able to promptly sign a contract with the national entity responsible for implementing the e-voucher. In contrast, the two major mobile telephone operators in Mali signed contracts to incorporate the short code into their SMS network with the aim of delivering SMS messages between the platform and the beneficiaries on the one hand and between the platform and suppliers on the other. During the e-voucher experimental phase, the cost of SMS messages sent was fully borne by WAAPP and varied from CFAF 10 to 25 from one country and one operator to another. In all three pilot countries, issues with network coverage disrupted the smooth delivery and receipt of SMS messages. 5.5 Targeting of beneficiaries Targeting entails setting the eligibility criteria used to select potential beneficiaries of the subsidy program. Since the subsidy cannot cater for the rather high number of registered farmers, the countries came up with a set of selection criteria that can be classified into three major categories:  ulnerability-based targeting criteria: E-voucher programs based on food- or poverty-related vulnerabil- > V ity target regions and groups suffering from acute crop and livestock production shortages compared with the national or multiyear average. Niger’s e-voucher program is a typical example of this targeting approach, targeting regions with the greatest production shortages and focused especially on the most disadvantaged communities, recognized as such in surveys conducted under the Social Safety Nets Program.  riority crop promotion-based targeting criteria: By adopting these criteria, the countries are looking to > P boost productivity in strategic segments of the farming sector. Mali’s government has opted to use the e-voucher to promote grain production (maize, millet, sorghum, and rice).  ender-based targeting criteria: In Mali, the first e-voucher program, which was implemented specifically > G in the northern region, targeted mostly women and small-scale farmers. Guinea took its cue from Mali by applying the same criteria to launch its pilot e-voucher program. 11 Table 2 below outlines the targeting criteria applied in Guinea, Mali, and Niger. TABLE 2: E -VOUCHER TARGETING CRITERIA IN GUINEA, MALI, AND NIGER E -VOUCHER PROGRAM TARGETING CRITERIA North Mali pilot program • Three districts areas affected by extreme weather events 2015–2016 • Be a female, regardless of farming activity • For crop farmers, farm an area of between 0.5 and 5 hectares • For livestock farmers, have at least five (5) head of livestock to be eligible for the subsidy for  small ruminants and own a stock of ten (10) to benefit from the poultry subsidy Mali (Segou and Sikasso) • Be a maize farmer in Koutiala and Yanfolila districts (Sikasso region) 2017–2018 Be a millet, sorghum, irrigated rice, rainfed rice, or lowland rice farmer in Bla and Niono districts •  (Ségou region) Guinea • Focus on the Upper Guinea region, which previously benefited from input subsidy 2017–2018 kits for cotton • Gender sensitivity: 40% women • Smallholders (0 to 2 ha) with poor access to agriculture inputs: 80% of producers Based on the above criteria, three major categories of beneficiaries were identified: • 50% for small-scale farmers, i.e., ≤ 2 ha of rice or maize (of which at least 50% women) • 30% for medium-scale farmers, i.e., 2–4 ha of rice or maize (of which at least 40% women) • 20% for farmers cultivating 5–10 ha of rice or maize, irrespective of gender Niger • Be a resident of a municipality with > 50% of population consisting of vulnerable 2017–2018 households7 • Beneficiaries already selected under the Social Safety Nets Program BOX 1: MALI – E -VOUCHER IMPLEMENTATION, 2017–2018 In Mali, the e-voucher system was implemented in Ségou region (Bla and Niono districts) and Sikasso region (Koutiala and Yanfolila districts). The subsidized crop in Sikasso region (Koutiala and Yanfolila districts) was maize, while in Segou region (Bla and Niono districts), the subsidized crops were millet, sorghum, irrigated rice, rainfed rice, and lowland rice. Only fertilizers were provided to beneficiaries in the Mali e-voucher program. The subsidy was allocated as follows: (i) for farms with an area of 1–3 ha, a fertilizer subsidy was granted for 1 ha; (ii) for an area of 3–5 ha, a fertilizer subsidy was granted for 1½ ha; and (iii) for an area of 5 ha and above, a fertilizer subsidy was granted for 2 ha. A total of 93,000 farmers (of whom 19.3% were women) were registered. After application of the eligibility criteria defined in the of the Ministry of Agriculture’s subsidies procedure manual, 24,583 beneficiaries were deemed eligible for the e-voucher program. After distributing the e-vouchers in all four districts, 18,276 out of 24,583 targeted beneficiaries collected their vouchers from the suppliers, a rate of 74.3%. A total of 10,207 metric tons of fertilizer were distributed across the four districts. 7 Source: Findings of the Crop Forecast Survey (EPER) conducted during the 2017 rainfed crop season by the Directorate of Statistics of the Ministry of Agriculture and Livestock. 12 BOX 2: GUINEA – E -VOUCHER IMPLEMENTATION, 2017–2018 The voucher platform for Guinea enrolled 96,000 farmers, of whom 22.7% were women. The subsidy covered 40% of fertilizer and herbicide expenses for small-scale farmers, 30% for medium-scale farmers, and 20% for large-scale farmers. In terms of assistance, two types of kits were provided, consisting of a rice kit containing 50kg of rice seeds, 150kg of NPK, and 100kg of Urea, and a maize kit containing 35kg of maize seeds, 100kg of NPK, and 50kg of Urea. Overall, each category of beneficiaries was expected to receive 1 kit for small-scale farmers, 2 kits for medium-scale farmers, and 4 kits for large-scale farmers working more than 5 ha. Each kit should therefore allow for the effective farming of roughly 1 ha. For the pilot phase, 5,000 kits consisting of seeds, fertilizers, and herbicides (including 4,000 kits for rice and 1,000 kits for maize) were distributed, for a total of 200 metric tons of rice seeds, 35 metric tons of maize seeds, 700 metric tons of seeds, 15.450 T of urea and NPK fertilizer, and 25.000 L of herbicide. For this phase, three categories of producers were identified according to declared farm size (less than 2 ha, 2–4 ha, more than 5 ha), accounting for 82%, 11%, and 7%, respectively, of beneficiaries. In total, 1,261 out of 3,500, or 36% of beneficiaries, were women. This percentage rose to 39% for smallholders (less than 2ha). In 2019, 20,000 farmers will benefit from e-vouchers. BOX 3: NIGER – E -VOUCHER IMPLEMENTATION, 2017–2018 In Niger, the 2018 e-voucher operation as well as the 2017 pilot experiment, were conducted by the General Directorate for Agriculture (DGA) with financing from WAAPP and the Climate Smart Agriculture Support Project (PASEC). The Safety Nets Unit (CFS) provided technical support in order to better target vulnerable groups. The CFS database of vulnerable households in Niger was used by the DGA by introducing a number of variables related to the agricultural sector. To this end, the DGA carried out a complementary survey to complete the baseline for the Safety Nets census and in order to target vulnerable farming households that could benefit from the distribu- tion of seeds of improved varieties of millet, sorghum, and cowpea through the issuance of electronic coupons. The choice of the beneficiary municipalities was made on the basis of the findings of the Crop Estimation and Forecast Survey (EPER) of the 2017 rainy season conducted by the Directorate of Statistics of the Ministry of Agriculture and Livestock. The e-voucher program was implemented in 20 municipalities in Dosso, Maradi, Tahoua, Tillabéry, and Zinder regions. The e-voucher kit consisted only of seeds (10 kg of cereals and 5 kg of cowpea) to cover at least 1.5 ha of crop. A total of 30,838 households benefited directly from the program, of whom 26% were headed by women. 13 6. LESSONS LEARNED DURING THE IMPLEMENTATION OF THE E-VOUCHER During the pilot phase, a number of challenges were encountered from registration to distribution, from which the following lessons were drawn: 6.1 Targeting is a key determinant of successful e-voucher implementation Governments cannot subsidize the procurement of farm inputs for all farmers and must therefore ensure that the subsidies allocated are used efficiently. Thus, targeting beneficiaries is a key factor in using e-voucher systems to successfully distribute inputs using information and communication technologies (ICT). The three countries under study achieved different levels of success in their ability to reach their intended beneficiaries, which may be linked to the fact that the program objectives were not clear: was it poverty reduction (and thus a safety program), or was the aim to increase productivity? If the objective is poverty reduction, the extremely poor would be the targeted beneficiaries. However, if the objective is to increase productivity, then the targeted beneficiaries would potentially place in the second and third poorest quintiles, where farmers are in transition from subsistence to market-oriented production and would thus be motivated to make the most of the subsidy in order to boost their productivity. 6.2 In populations with high levels of illiteracy, technologies used should be adapted to them With only 32%8 of the adult population in Guinea, 33%9 in Mali, and 30.6%10 in Niger being literate, key information sent to the beneficiaries about their subsidy allocation was lost. A significant proportion of farmers deleted their SMS vouchers out of ignorance. Introduction and use of voice messaging would contribute to solving this issue. Poor network coverage in rural areas coupled with the fact that some beneficiaries (mostly women) did not have a personal telephone number or handset further complicated implementation. 6.3 Public procurement of inputs is linked to delays in delivering inputs Evidence from the three countries demonstrates that input subsidy programs in which the inputs are procured through the public sector often experience delays in deliveries. In the countries under study, delivery of farm inputs was delayed in several areas. In some cases, e-vouchers were delivered after the rainy season had begun. The delays were mostly due to bottlenecks in public procurement and organizational challenges faced by some agro-dealers who were not yet used to the new system of inputs distribution. 7. PROSPECTS AND RECOMMENDATIONS 7.1 Clarify program objectives as well as target beneficiaries Clarifying the objectives of the subsidy program is fundamental to its success. At the design stage, governments should define the objective of the e-voucher program so that beneficiaries will be more clearly targeted and supplied with effective kits. 7.2 Use Near-Field Communication (NFC) technology to remove illiteracy and telephone ownership barriers Near-Field Communication (NFC) technology has been used in Ethiopia to increase access to inputs. Beneficiaries receive NFC strips on cards or wrist bands while agro-dealers acquire smart phones with an app that can read the strip. When the NFC tag or card taps the phone, the information it contains is downloaded to the phone, thus eliminating the 8 WDI, 2014 9 WDI, 2015 10 WDI, 2012 14 need for the beneficiary to be able to read, write, or type and send an SMS. In addition, the app could be used offline, the transaction completed, and the information sent later when the phone is in area with good coverage. 7.3. Private sector participation in the program Private sector participation in the program should be increased. Conversely, public sector participation should be reduced. Both are key factors of success. Evidence from Ethiopia, Rwanda, and Mozambique shows that increasing private sector participation in areas such as routing input payments through financial institutions, supporting the private sector’s involvement in the importation and production of inputs, and a defined exit plan to reduce the subsidy allocation will increase transparency and cost effectiveness and improve the timeliness of input deliveries. 7.4 Diversify e-voucher kit contents to include improved seeds and e-extension services Generally, the kits provided to farmers contain only one type of input (seeds in Niger and fertilizers in Mali). In both cases, the impact of the subsidy on productivity may be easy to establish in the absence of another input. The e-voucher kits need to be redesigned into a more consistent package of inputs (improved or certified seeds, fertilizers and counseling or extension services) tailored to the needs of the beneficiaries and capable of considerably increasing yields. In the same vein, the development of e-extension services using the same electronic platform as the e-voucher is an option within the reach of the three countries. 7.5 Improve e-voucher program planning Implementation of the e-voucher program was often delayed. In several villages, inputs were supplied after the rainy season had begun. Improved planning of core activities within the e-voucher program (inventory of inputs, contracts with agro-dealers, creation of digital platform and functionality tests) will go a long way to overcoming this challenge. 7.6 Encourage more gender-oriented interventions Due to the focus on certain crops and the land ownership constraints faced by women, women received very few e-vouchers. Overcoming this weakness will require channeling more support to women, as was the case with the first e-voucher program implemented in northern Mali and Guinea. Countries must reorient the e-voucher program to cater for more crops generally grown by women, such as vegetables, sesame, and rosella. More innovative solutions should be implemented to benefit women without access to land, following the successful experiment in Niger, where kits containing small stock and poultry as initial assets were supplied to poor and landless women. 7.7 Build database management and administration capacities In all three countries, implementation of the e-voucher program exposed the challenges faced by the entities in charge of agriculture in managing the digital platform. These entities often lack the necessary technical skills despite substantial assistance from private service providers, including local start-ups. By opting to entrust the database management to a fully-qualified technical entity of the Directorate General for Information Technologies and the Digital Economy (DNTIEN), Guinea allowed for more effective stakeholder ownership and technical support for proper implementation of a digital farming development program. These national entities need to be strengthened for them to play their role fully. 11 AGRA, 2019. A Review of Mali’s Farm Input Subsidy Program. 15 ANNEX I: E-VOUCHER SYSTEM STEP-BY-STEP Step 1: An SMS is sent from the e-voucher platform (represented by a short code, e.g., 4455) > E.g.: “Your voucher number is: 12345. Please collect 50kg of rice seeds from Agro-dealer ABC, contact number: 74989521.” Step 2: The farmer goes to Agro-dealer ABC with the SMS. Step 3: Agro-dealer ABC requests approval from the platform. > E.g.: “Approve 11203 30 on 4455” Step 4: Agro-dealer ABC receives approval via SMS: > E.g.: “30kg for Voucher No. 11203 is good for delivery to: [Name of farmer; telephone number]” Thereafter, the voucher number is no longer usable. Step 5: Agro-dealer ABC delivers the product to the beneficiary. Step 6: The beneficiary receives a confirmation SMS: > E.g.: “How many kgs of products have you received for voucher No. 11203? Send: Confirm [space] Voucher number [space] Quantity on 4455.” > E.g.: “Confirm 11203 50” Step 7: The beneficiary responds to the SMS (following the instructions in Step 6). Finally, the platform sends the following reply: “Thank you for the confirmation” 16 special topic Improving projects performance with cost-effective Iterative Beneficiary Monitoring André-Marie Taptue, Aboudrahyme Savadogo, and Aly Sanoh, March 2019 Abstract Iterative Beneficiary Monitoring (IBM) is an approach to project monitoring designed as a light, low-cost, independent, rapid, and iterative feedback loop that collects information directly from beneficiaries and produces short reports on challenges that can be addressed by project teams. This approach improves project efficiency and increases beneficiary engagement and satisfaction by creating positive, self-reinforcing cycles of improvement. IBM was first developed in Mali to deal with limited access to project activities due to insecurity, but its low-cost, high-frequency, and rapid nature also makes it attractive and applicable in secure settings. IBM was used to identify shortcomings in a school meals project in Mali and helped reduce time needed to transfer funds to schools and increase the number of schools that offered meals five days a week. It also supported an e-voucher scheme in Mali, which ensured that farmers receive fertilizer of good quality, at the appropriate time, and in the promised quantity. As part of a rural mobility and connectivity project in Mali, IBM engaged early and checked whether the population heard sensitization messages from rural radio stations, they were affected by the work undertaken, and whether they were satisfied. IBM implementation convinced the managers of a health insurance project for the extremely poor in Mali to find a way to enhance distribution of health cards along with cash transfers. Results from monitoring these initial projects were shared with the project managers in the country with a view to solving issues identified while the projects were still underway. The World Bank is currently scaling up the IBM methodology, with various Global Practices applying it to projects in Mali, Chad, Guinea, Niger, Benin, the Central African Republic, and Nigeria. 1. INTRODUCTION Successful project implementation is often hampered by lack of insufficient information from beneficiaries. Even in a secure environment, access to events on the ground changes according to the season and depends on the existence and quality of roads, the cost of data collection, and the capacities of project teams. Particularly in fragility, conflict, and violence (FCV) contexts, physical access can be limited by insecurity and limits to field visits, a combination that obscures information on specific dynamics and the opinions of beneficiaries in those areas. Subsequently, these information gaps effectively inhibit operational engagement, especially in places where development interventions are most critically needed. Moreover, these information gaps cannot always be adequately closed by project monitoring and evaluation (M&E) systems as they use cumbersome approaches, while local capacities and available resources may be inadequate to carrying out large-scale surveys. These systems are expected to track progress and flag potential shortcomings 17 or problems. In practice, most M&E systems do not provide frequent or independent reports but focus instead on producing progress indicators for the midterm and final reviews of the project. Even this reduced role for M&E systems is not always effectively executed and reports often come too late to help projects improve. Moreover, because of their high cost, M&E surveys cannot be repeated frequently, with data collection usually taking place three times or less over a five-year project. Supervision missions offer another source of information on project performance, but the information such missions can obtain is limited because they are less frequent, planned for short periods (usually no more than two weeks), and are often put on hold by project teams in case of security-related events. Objective information about the effectiveness of projects may come from evaluations by non-project staff. Typically, these take the form of randomized controlled trials or large-scale surveys, such as Service Delivery Indicator (SDI) surveys, which measure the quality of service delivery in health and education, or Public Expenditure Tracking Surveys (PETS), which track the flow of resources from central to decentralized level. Though they are reliable, these data-inten- sive approaches are expensive and difficult to conduct in a fragile and insecure environment and cannot therefore be repeated frequently. Moreover, they are time-consuming and rarely deliver quick results, which sometimes becoming available only after project closure. This information gap can be filled to improve project results by designing a distinct approach for frequently gathering information from beneficiaries and other stakeholders. To support project managers in achieving their objectives, a feedback loop system that is iterative and provides unbiased information is needed. This will allow the project team to learn from any difficulties facing project implementation and therefore improve performance. Once action has been taken, the team must assess whether any deficiencies identified have been resolved. To allow for regular feedback, data collection should be affordable and agile so as to yield quick results. Reliable, regular, and inexpensive data are the ideal. This system helps improve project effectiveness and increases beneficiary satisfaction and engagement. To meet these requirements, a beneficiary feedback system has been designed that is simple and inexpensive, focuses on a select set of issues, and is implemented by an independent entity with no stake in the outcome of the project. This approach is known as Iterative Beneficiary Monitoring (IBM). IBM complements supervision missions and project M&E by offering an agile, problem-oriented feedback loop for project management. It provides feedback to project teams through multiple rounds of small-scale data collection that allows project teams to identify implementation issues early and to take corrective actions. It collects data from beneficiaries using fewer questions and smaller samples while remaining informative. This approach focuses on flexibility of design, reduces cost, facilitates timely data analysis, and increases speed of report preparation while focusing on feedback relevant to implementation. In addition to improving project effectiveness, IBM aims at increasing beneficiary satisfaction and engagement. These objectives can be met by ensuring that beneficiaries are reached wherever they are located and receive in timely fashion goods and services that are useful to them. Compliance with the Procedural Manual and identification of any hindrances to the project during the implementation phase leads to improvement in the project. To increase beneficiary satisfaction, the IBM approach enables beneficiaries to provide feedback on the project and to verify whether a remedy is to their satisfaction, thereby increasing their interest and engagement. Applied in Mali since 2015, IBM has improved the effectiveness of projects in the agriculture, education, transportation, and healthcare sectors. It led to increased requests that it be expanded in the rest of the countries in the Mali Country Management Unit (CMU), which is composed of fragile countries such as Chad and Mali itself as well as countries at risk of fragility such as Guinea and Niger. Implementation of projects in these countries faces a number of constraints that limit the achievement of their objectives. While projects experience long delays at several levels during implementa- 18 tion, it has been difficult to identify and quantify those hindrances. In addition, insecurity and drivers of fragility such as economic and regional disparities, uneven development, and poor governance limit access to beneficiaries in order to elicit their opinions about project implementation. Hence, IBM plays an important role in collecting information from beneficiaries and thus enhancing project impact. 2. METHODOLOGY The idea behind an iterative feedback loop is to allow the project team to learn from any hurdles in project implemen- tation and thus improve performance while the project is still ongoing. Once action has been taken, the team must assess whether any identified deficiencies have been resolved. IBM follows a process that moves away from large and infrequent surveys followed by long reports to agile and frequent surveys accompanied by short reports. IBM design follows five steps (see Figure 1), beginning by becoming intimately acquainted with a project and appreciating any challenges project teams are facing. The first step is time-consuming but indispensable to understanding details of project design and to determining what would constitute high-quality monitoring information. Core project documents need to be read, starting with the Project Appraisal Document and the Implementation Manual. These are invaluable in identifying sources of information or standards that can be used to assess the project. Supervision reports, memorandums, and mission reports will also help identify issues of potential concern. Collecting information from beneficiaries and other stakeholders on the front line of service provision (such as teams working in schools, clinics, or farmers’ organizations) is at the heart of the iterative feedback approach and constitutes the second step, when questionnaires and sampling methods are designed. The experience of all stakeholders with the project is what ultimately matters. IBM thus focuses on obtaining direct feedback from these beneficiaries. Identifying what information to obtain and from whom is an important step in the design of a feedback system. For instance, in a project offering school meals, the perspective of parents and guardians is critical because they can ascertain whether their children have eaten. Students can give their views on the quantity and quality of the food and how often they receive it. Head teachers can confirm whether the money to buy the food arrives on time, parent teacher associations can explain whether procedures are being followed, and those who prepare the food are well-placed to report on whether the money they receive is sufficient. It is thus critical that the iterative system be developed in close collaboration with project managers. Managers need to provide access to project files (including beneficiary databases needed for sampling) and to validate the methodology and instruments chosen for data collection. If this is not done with care, project managers may eventually contest the validity of the results, and little follow-up can be expected. Apart from collaborating closely with project managers, the monitoring team will also need to ensure that the identity of respondents and the locations where data are collected are kept confidential. It is important to keep the data collection process simple and to resist the temptation to collect more information than is strictly necessary. A project manager’s capacity is often constrained, and a project team can only handle so many issues at a time. Given that the approach is iterative, new issues can be addressed in subsequent rounds of data collection as not all issues need to be investigated in the first iteration. This gives the project team the option to prioritize what is most critical or most easily addressed. As the data collection process is kept to a minimum, the design of data collection instruments is relatively straightforward. Nonetheless, validation of the data collection instruments by project managers remains an essential step. This includes pretesting in a real-life setting and discussing the instruments with key project staff to ensure that the right issues are captured in an appropriate way. 19 FIGURE 1: THE FIVE STEPS OF THE IBM APPROACH 1ST STEP 2ND STEP 3RD STEP 4TH STEP 5TH STEP Get Design acquainted low cost data Share with project; collection Data results identify system and analysis with TTL, Adapt challenges collect information and preparation CMU and jointly with from of short reports project TTL beneficiaries managers SUBSEQUENT ROUNDS Small samples are not a problem in themselves. When project-related issues are widespread or when standards or deadlines must be met (as set out in the Implementation Manual), a small number of observations may pinpoint a problem. Irrespective of sample size, sample design is critical to ensuring that results are representative. This implies identifying a reliable database from which the sample can be drawn. This is usually not a problem as most projects maintain a database of beneficiaries or can build one quickly. However, additional decisions may need to be made with project teams regarding cost-saving methods. For instance, the team may propose to sample from one small geograph- ic area only, which may be acceptable if the area reflects an upper bound, where the effects of any of the project’s shortcomings are likely to be worse than in other areas. However, it may constitute an unacceptable measure when the involvement of area managers, such as the highest-ranking district official, is critical to ensuring project outcomes. For example, suppose that money transfers to schools close to the capital are delayed; then it is plausible to assume that the situation is worse in more remote areas. Technology can be used to enhance efficiency and reduce sampling costs. If projects collect beneficiaries’ phone numbers, information can be elicited rapidly and cost-effectively through phone surveys, even when sample size increases. This is particularly important in an insecure context or when the population may be hostile to the authorities and their activities. Mobile phone-based data collection is also a solution when beneficiaries are themselves mobile, as is the case for displaced populations or nomads. Because collecting data over the phone is inexpensive, collecting phone numbers simplifies the creation of an iterative feedback loop. However, in the absence of a database of beneficiaries with phone numbers, data can be collected using face-to-face (F2F) interviews, though these tend to be expensive due to high transportation and accommodation costs and are sometimes risky for enumerators. Therefore, F2F samples need to be kept to a minimum. While the risk for enumerators in insecure areas is mitigated by introducing them to local authorities to avoid confusion with other agents, in all cases, respondents are also protected by all information likely to identify them being kept confidential. In comparison, data collection itself is relatively inexpensive. The principle behind IBM is that each round of data collection should cost less than US$5,000. Given this cost structure, the iterative feedback loop differs fundamentally from typical surveys, where data collection is the costliest part of the process. Keeping data collection costs low is important for the success of IBM because since frequent data collection would not be otherwise possible, its iterative character may be lost. 20 Once collected, data are analyzed and offered as feedback to project managers in the Government and to World Bank team leaders at Steps 4 and 5 of the system. Given that the number of questions is kept small in each iteration, data analysis is rapid. IBM reports are specific, factual, and short: typically, less than 10 pages. As reports are likely to reveal the project’s shortcomings, care needs to be taken to ensure the highest standards of accuracy. Often, results are discussed with those responsible for the project in the client government. These authorities may therefore request that the project team take the required steps to address any issues, but this is rarely needed as project teams tend to be highly responsive to IBM findings and promptly work to address the issues identified and to overcome any shortcomings in order to complete Step 5 of the system. Another round of data collection then follows (generally after a few months) with the aim of measuring improvement and sometimes to identify new issues that might have arisen since the previous round. The reporting process is the same as for the previous round. This cycle is then repeated on a regular basis until the end of the project. To date, IBM reports have been produced for internal use by project teams in client governments and the World Bank). This is because wider public disclosure could lead to unintended consequences. Media and NGO experience with water price monitoring in Tanzania is illustrative in this regard. In this case, light monitoring principles were applied, but instead of working to address the issue with the regulator, those in charge of the monitoring process sought media attention. While public pressure and parliamentary questions did lead to corrective action, these were somewhat ad hoc and symbolic in nature. Moreover, some outcomes proved perverse as some water kiosks were closed because they had been overcharging, leaving those dependent on these kiosks with even fewer options than they had previously. Moreover, following initial media interest, there was no systematic follow-up of the issue, and overcharging continued unabated. 3. IBM IN AFCW3 COUNTRIES The IBM approach started in Mali in 2015, first as part of an education project (school meals), then with an agriculture project (e-vouchers) in Mali and Niger, followed in Mali by a cash transfer project, a rural mobility project in the transportation sector, and more recently government projects such as health insurance for the extremely poor and land commissions, both of which constitute a trigger for the Development Policy Operation (DPO) program. 3.1 IBM applied to education projects In the case of the school meals project in Mali, the project team leader expressed a concern to the IBM team that only part of the money allocated to the project component was being used. The two teams decided to look further into the issue and agreed on a clear division of tasks. The IBM team would take charge of all issues related to data collection and reporting, while the project team would facilitate all interactions with the Ministry of Education and the Project Implementation Unit (PIU). To understand the challenges involved in implementation, the National Center for School Cafeterias in the Ministry of Education shared the database of schools benefiting from the school meals program. This database was used to draw a sample of beneficiary schools. To ensure ownership and accuracy, officials from the Ministry and the Center actively participated in the preparation and validation of the survey methodology and tools but were not provided the list of schools included in the sample. The first round collected data in 20 randomly selected schools.12 Two enumerators were trained and traveled to each school to carry out F2F interviews with head teachers, school cafeteria managers, and a subsample of parents. It cost less than US$5,000 to complete the data collection process, and the report took little time to prepare as information was collected only on a limited set of issues. Officials from the National Center for School Cafeterias and the project team leader were informed of the main results. Results were also shared with the Country Director and the Minister of Education. In general, the size of the sample in an IBM design is guided by three factors: time needed for data collection (usually less than two weeks), the budget, and 12  the length of the questionnaire. A pilot test yields the estimated time needed to complete an interview as well as the number of beneficiaries likely to be interviewed within two weeks. This is adjusted if necessary to comply with budget constraints. 21 Results showed that it took more than four months to transfer money from the Ministry of Education to schools. Consequently, much of the money for school meals arrived after the school year had started, thus jeopardizing one of the objectives of the program, which is to increase enrollment rates. Moreover, the amount of money sent to schools was insufficient to feed all students during the envisaged period, and some schools were forced to offer meals less than five days a week, thus reducing the incentive for students to remain in school. FIGURE 2: REGULAR FOLLOW- UP IMPROVED SCHOOL MEALS PERFORMANCE Average time for transfer (days) Percentage of schools that offer food less than 5 days per week 1st Transfer 2nd Transfer 3rd Transfer 4th Transfer First round of Second round of follow-ups follow-up First round of follow-ups Second round of follow-ups Source: Authors’ calculations based on IBM data While transfers were expected to be made every quarter, their actual frequency was much lower. Moreover, the Procedur- al Manual was not followed. Whereas the amounts transferred were supposed to reflect enrollment rates, they were often much higher or much lower than the expected amount. The monitoring report was discussed with the project team leader, project staff, and the Minister of Education, who responded by sending letters to project officials to make them aware of the issue. Additional supervision missions were requested, and the Minister requested accurate information on school enrollment to rectify problems with the transfer amounts. Six months later and one year before closure of the project, a second round of data collection was conducted in 30 schools randomly drawn from the updated list provided by the National Center for School Cafeterias, excluding those interviewed in the first round. Results showed that it now took much less time to transfer money to schools. Most schools received close to the exact amount they expected, and all the money disbursed by the Ministry reached the schools. Despite this, some schools were still offering meals less than five days a week, particularly those that had not received the money required to feed all students. The second report showed significant improvements in project implementation, although certain issues persisted (see Table 1). Both positive and negative findings were shared with the project team leader, project managers in the Government, and the Minister of Education. 3.2 IBM applied to agriculture projects The success of IBM in the school meals project in the education sector increased interest from other project managers. Hence, the approach was expanded to an agriculture project that distributed subsidies in the insecure north of the country using electronic vouchers (e-vouchers). Under the e-voucher system, beneficiaries were counted, and their phone numbers and core characteristics captured in a database. This information was used to send them e-vouchers by text message. Upon receipt of their vouchers, beneficiaries could go to a store to collect their products, typically fertilizer or livestock products. 22 TABLE 1: RESULTS OF TWO ROUNDS OF ITERATIVE FEEDBACK ON A SCHOOL MEALS PROJECT IN MALI 1ST ROUND ACTIONS TAKEN 2ND ROUND: 6 MONTHS LATER Sample 20 schools Report discussed with 30 schools not included in Minister of Education initial sample Duration and method 10 days of F2F interviews 10 days of F2F interviews for data collection Cost of data collection < US $5,000 < US $5,000 Preparation and analysis 5 staff weeks 2 staff weeks Source of financing Poverty monitoring Poverty monitoring Issues Findings Findings 1. Time to transfer More than 3 months Sending of awareness Reduced by two-thirds money to schools letters by Minister to project managers 2. Does the total amount Yes Yes sent by the central government reach schools 3. Does money arrive on time? No, money arrives long Transfer delays reduced considerably after classes have resumed 4. Number of transfers per year 1 out of 4 planned 3 out of 4 planned 5. Number of days covered 50% of schools cover less Setting-up of supervision Reduced to 40% by amounts sent to schools than 40 days, as requested missions by Minister 6. Number of days per week 25% of schols offer meals less Reduced to 13% meals are offered to students than 5 days a week 7. Do transferred amounts Transfers do not account Improved, but a gap persists reflect enrollment rates? for school size, as required between school size and figures used in the Ministry. Project management expressed concern about the limited uptake of the subsidies. A supervision mission reported that during the first wave of input distribution, only 41 percent of beneficiaries who had been sent an e-voucher collected their products, even though these were free of charge. This suggested that there were problems with the input distribution system or lack of interest among beneficiaries in the products on offer. Identifying the exact nature of the problems was clearly important for the success of the project. The key aim of implementing IBM was then to confirm the percentage of farmers who did collect their products and to check why others did not. Because the project relied on e-vouchers, there existed a database of beneficiaries’ phone numbers, and as the areas of intervention remained insecure, the team opted to use phone interviews for data collection. Project managers shared the database of beneficiaries with the analysts and participated in working sessions designed to validate the methodol- ogy and survey instruments and to select a representative sample of 100 beneficiaries who were interviewed by phone. Analysis of the shared database revealed the presence of many duplicate phone numbers allocated to different people in different villages. While the Procedural Manual permits different beneficiaries to use the same phone number since not everyone owns a phone, they would all be expected to live in the same village. However, the duplicates identified 23 in the database were not living in the same location. After survey instruments were validated, two enumerators were trained to collect data over the phone. After four attempts to call each respondent, only 40 percent were reached, raising questions about network coverage in villages where beneficiaries live, the accuracy of the phone numbers in the database, or the location of beneficiaries as some may have left their initial location due to insecurity. The initial results showed that all the beneficiaries who had received e-vouchers had collected their products, suggesting that low uptake of products was not due to lack of interest. As a significant proportion of beneficiaries could not be reached by phone, it was not possible to know whether all the e-vouchers had been successfully delivered. It seemed plausible that, as with the failed phone interviews, many e-vouchers had failed to reach their intended beneficiaries, suggesting a communication problem between the e-voucher platform and beneficiaries. Finally, many beneficiaries indicated not having received the full quantity of (free) products indicated on their vouchers, nor were they compensated for any items not received. Following these results, the project team and telecom providers were contacted to discuss the findings and to address certain issues, including the number of duplicate phone numbers in the database, the inability to send a high number of text messages per second, and the absence of a “text message received” message. The report was shared with the Ministry of Agriculture and the project team, which took action to clean up the database, asked controllers to ensure that farmers received the full quantity of the products due to them, and reported on their findings. TABLE 2: RESULTS OF TWO ROUNDS OF ITERATIVE FEEDBACK ON FERTILIZER DISTRIBUTION USING E -VOUCHERS 1ST ROUND 2ND ROUND: 5 MONTHS LATER Sample size 100 beneficiaries 850 beneficiaries Duration and method for 5 days by phone 10 days by phone data collection Cost of data collection < US $5,000 < US $5,000 Preparation and analysis 3 staff weeks 1 staff weeks Source of financing Agricultural project Budget support operation ISSUES ACTIONS TAKEN ISSUES ACTIONS TAKEN 1. Are beneficiary localities One telecom firm provides 1. E-vouchers distributed two DPO delayed till issues of coverage and covered by telephone network? information on its network months after the start of the timelliness of the e-voucher system project. planting season. are addressed. 2. Are vouchers successfully Meetings organized with 2. Only 15% of beneficiaries delivered? telecommunication firms who collected their fertilizer as improved the number of text vouchers had been sent late. messages that can be sent per Approach to identifying beneficiaries 3. Only 40% of beneficiaries can second and agreed to send text- 3. Only 8% of beneficiaries in changed. be reached by phone for receipts. In second round 64% data base are women. interview. could be reached. 4. 13% of beneficiaries have Database cleaned, duplicates 4. Large price difference Fertilizer now procured using a duplicates in databases. reduced to less than 5%. between official price of competitive international procedure. fertilizer and market price (of up to $9 per bag). 5. 43% of beneficiaries receive Measures taken to improve less fertilizer than expected. oversight at the delivery of inputs; 30% report receiving less than the expected quantities. 24 BOX 1: IBM ON E -VOUCHER SCHEME: SUMMARY A second round of data collection was carried out five months later with a larger sample. At this stage, there was a need to assess how well the approach has worked since successful implementation of the e-voucher scheme was a precondition for budget support to the Government of Mali. More information was needed than a simple understand- ing of whether the approach was working, and evidence had to be collected about the percentage of beneficiaries in each district as well as the application of targeting criteria. The second round showed that the management of the system had improved. The database was cleaner, more respondents could be reached, more messages could be sent per second, and sent messages were now received. However, the results also showed that the rollout of the scheme still left much to be desired. Not all the agreed zones were covered, and e-vouchers were sent late, typically three months after the start of the agricultural season. Moreover, e-vouchers were distributed for fertilizer that could not be used for the current growing season. Finally, fertilizer suppliers turned out to have been selected using a noncompetitive method. These findings led to high-level discussions between Work Bank managers and the Malian authorities. The results of the two rounds of beneficiary monitoring are presented in Table 2 and the summary in Box 1. In Niger, IBM was implemented as part of the e-voucher scheme for the West Africa Agricultural Productivity Program (PPAAO) and the Community Action Project (PAC 3). The e-voucher scheme in Niger operates in a similar manner to the one in Mali, as described above. PAC 3 aims to: (a) strength beneficiaries’ capacities as part of planning and monitoring of local development; and (b) improve vulnerable populations’ access to social and economic services. The project supports micro-projects implemented by community associations in line with the development plans of client governments. Beneficiaries are members of associations who are experiencing difficulties related to substantial shortfalls in cereal production, severe levels of food insecurity, high household indebtedness, major livestock losses, and so on. IBM was implemented on a sample of 455 beneficiaries, including 202 livestock breeders for herd rebuilding and fattening, 220 for sustainable land and water management, and 33 for market gardening. IBM results show that 52 percent of project beneficiaries were women. However, about 10 percent of women did not receive training for improving their activities while only 1 percent of men missed out. In practice, the Management Committee supposed to follow up on the activities of beneficiaries paid more attention to men’s activities compared to women’s (with 94 percent of men receiving follow-up activities against only 56 percent of women). One of the main weaknesses of project implementation highlighted by IBM was delays in providing support to beneficiaries, with about 50 percent 25 of beneficiaries receiving support from the project at least three months after transmitting their financial contribution for the co-funding of their project. On average, men experience a delay of 2.5 months while women waited 3.6 months before receiving support from the project. A grievance mechanism is in place for project implementation to allow beneficiaries to report constraints and problems they face. However, 39 percent of beneficiaries were not aware of this mechanism. Results of the monitoring were shared with the project teams in the World Bank and the Government. The report has just been completed, and the next round of IBM will assess any actions taken as well as their impact. 3.3 IBM applied to healthcare projects IBM is being applied to support for the Government’s healthcare assistance program (RAMED), which aims to provide free healthcare to the extremely poor in Mali. It aims at helping the extremely poor in Mali receive free healthcare when they get sick. To achieve this aim, the extremely poor are identified and provided with health insurance cards to present at the hospital. Discussion with the project team raised a concern about the distribution of health insurance cards and their acceptance in clinics. Hence, the entry point for the implementation of IBM in this long-term project was to assess the extent to which insurance cards were distributed and whether clinics readily accept those cards. Two rounds of IBM on 700 beneficiaries each were designed for the project to last until December 2018. Having validated the survey materials with the Government’s project team, the IBM team obtained the list of potential beneficiaries along with their phone numbers. This provided the option to collect data by phone as many phone numbers were functional. Each person selected in the sample and not accessible had to be called four times before confirmation that they were not reachable over the phone. The first round of IBM exhibited several issues hampering the operation of the project. Contrary to the project manager’s views about large-scale distribution of insurance cards, the first round of IBM revealed that only 39 percent of beneficiaries received their cards. In fact, although cards were printed in Bamako and sent out to mayors in beneficiary municipalities, mayors were given no means of distributing the cards. In addition, even among those who received their cards, some had to pay for consultation and medication at the hospital despite showing their health cards. Talking to hospital doctors revealed that awareness campaigns do affect beneficiaries’ use of cards. However, hospitals were not confident that they would be reimbursed if they provided free consultation and medication to patients holding health certificates. Finally, some beneficia- ries had not been made aware of the importance of the cards and did not know why they have been granted those cards. Dissemination of the report led to two main initiatives related to the project: (a) sending the information mission to clinics to reassure them of payment and explain the method for claiming reimbursement when offering free healthcare to patients FIGURE 3: PERCENTAGE OF PATIENTS WHO PAID FIGURE 4: PERCENTAGE OF PATIENTS BY ACCEPTANCE FOR MEDICATION AND CONSULTATION OF RAMED CARDS AT HOSPITALS IN MARCH 2018 % of beneficiaries having paid for consultation % of beneficiaries having paid for medication The hospital did not The hospital accepted accept the RAMED card the RAMED card 26 BOX 2: IBM ON HEALTH INSURANCE PROJECT: SUMMARY showing a health card and to inform beneficiaries regarding the use of the cards; and (b) promoting the distribution of cards. The project combined card distribution with payments of cash transfers since RAMED beneficiaries are also beneficiaries of cash transfers. This increased card distribution from 39 percent to 52 percent as measured during the second round of IBM, which took place three months after the first round. However, clinics still refuse to provide free healthcare to beneficiaries, and half of those who went to hospitals with a card paid for services. One suggestion was to set up a helpline beneficiary who faced resistance at the hospital could call so that a project manager could explain the mechanism and convince the hospital. 3.4 IBM applied to transportation projects The Rural Mobility and Connectivity Project and its Citizen Engagement component in Mali also use IBM. The project aims at rehabilitating rural roads and bridges in rural municipalities in two regions of Mali: Koulikoro, and Sikasso. While works were still under way, the project leader and the project coordinator in the Government paid attention to several issues, including: (a) whether awareness messages through rural radio stations were heard by the population; (b) high-quality bypasses were constructed; (c) the grievance redress mechanism (GRM) was in place and the population knew how it operated; (d) the completed infrastructure was of good quality or instead became flooded and blocked during the rainy season; and (e) the population was satisfied with the infrastructure. Those concerns constituted challenges to be addressed at the onset of IBM implementation. People living around the roads and bridges under construction or rehabilitation were expected to provide answers to these questions. Two enumerators were trained and sent to the two regions to collect data. IBM design thus went through an F2F survey for data collection and targeted populations living around road sections and bridges eligible for the project. In the absence of lists of such people, the enumerators were asked to randomly choose households following the “random walk” method. Starting from a well-known point in the village (school, clinic, pharmacy, bakery, and so on), the first household is chosen. The enumerator then moves forward, skips five households, and selects a second household. The process continues to cover the entire village until the enumerator reaches the number of households identified in the village. Overall, 90 households were chosen, and their heads were interviewed for the survey. Because of enumerators traveling to the project areas and visiting households, it was possible to map the position of households and to confirm that they lived close to relevant infrastructure. 27 POSITION OF VILLAGES VISITED AS PART OF IBM During this first phase of data collection in the field, the enumerators collected the phone numbers of household heads, which will be used for subsequent rounds of IBM. The phone survey could also be used to collect data on issues such as gender-based violence (GBV), as requested by the project leader. On this last point, the idea is to conduct a phone survey that guarantees the confidentiality of responses and asks respondents whether they noticed any signs of GBV between a member of their household and a worker on the project. A questionnaire for this type of IBM will be designed by gender specialists. Despite delays, awareness messages were sent to beneficiary villages using rural radio stations. Overall, 90 percent of households heard a message related to the project before or during the project. In addition, more than half of household heads participated in a sensitization meeting on the project. For the project leader, the Government project coordina- tor, and those responsible for the citizen engagement component, this result indicates the potential of sensitizing the population through rural radio stations since they aim to ensure that the degree of information does not decrease in future given that the project is based on a sequential approach, with each sequence benefitting from sensitization activities. Future IBM reports will assess these activities, compare results, and send findings to stakeholders. The IBM report confirmed the operation of the GRM. More than 80 percent of households were aware that they can file a grievance if they were negatively affected by the project. In addition, 90 percent of those households knew where to file such a grievance. Furthermore, all those who said that they were affected by the project filed a grievance with the GRM office. However, more than half of those who did so received no answer within 40 days, as stipulated by the Procedural Manual. Subsequent rounds of IBM will ensure that people affected by the project continue to file grievances and that these are answered on time. Works started in all villages but made varying progress in different villages. The photos below show a bridge and a portion of a road rehabilitated by the project. These roads are almost finished or are the most advanced. In some municipalities, households estimated that less than half of the work was completed, while in others, households claimed that not much has been done. Monitoring the progress of works by municipalities is another focus of IBM. As the phone numbers of households surrounding less advanced infrastructures have been collected, IBM will be designed to remotely follow the progress of these works. 28 BRIDGE AND ROAD UNDER REHABILITATION IN KOULIKORO When rehabilitating a bridge or a portion of a road, firms are requested to first construct a bypass to serve the population during the period of works. While they are still working, it should be checked whether bypasses have been construct- ed and are useable. Otherwise, firms should be required to build bypasses to facilitate movement by the population. The photo below shows a bridge under construction along with a bypass. More than 90 percent of households stated that firms built bypasses when they rehabilitated a bridge or a road, though opinions diverged about quality. In some municipalities, more than 90 percent of households confirmed that bypasses were of good quality, though this figure falls to 85 percent in some municipalities and as low as 34 percent in others. BRIDGE UNDER REHABILITATION, WITH BYPASS 29 IBM also assessed beneficiary satisfaction about the quality of rehabilitated infrastructures. More than 90 percent of households were satisfied with the project outcomes, which they thought were of good quality. However, almost 20 percent of households stated that rehabilitated bridges and roads were flooded during the rainy season and interrupted population movement. This last result calls into question the population’s qualifications for judging the quality of the built infrastruc- ture. It also points to the limits of the information that can be collected by IBM from beneficiaries. For example, beneficiaries’ views about the quality of infrastructure such as roads, bridges, classrooms, buildings, and so on should not be mistaken as approval of the technical specifications. 3.5 Mainstreaming IBM in AFCW3 countries: What next? The general approach is to support projects representing the various pillars of the Country Partnership Frameworks (CPF) and to provide opportunities for benchmarking across the CMU for similar operations. Extending IBM in AFCW3 countries will mean introducing it in other sectors and projects in Guinea and Chad in addition to Mali and Niger. Projects common to all or some of these countries will be targeted and monitored using the same questionnaires. For instance, the Rural Mobility and Connectivity Project is being implemented in Mali, Niger, and Guinea and managed by the same team from the World Bank. Lessons learned from IBM on that project in Mali will be used to design projects in the other countries using the same questionnaire, adapted as necessary. Regional projects in these countries are also targeted. This is the case of the Regional Project for Support to Pastoralism in the Sahel (PRAPS) in the agriculture sector. This project aims at improving access to essential productive assets, services, and markets for pastoralists and agro-pastoralists in selected cross-border areas and along transhumance axes across six Sahel countries, including Chad, Mali, and Niger. The project started its activities in these countries two years ago. Using the same questionnaire, IBM will be applied in order to identify shortcomings that hinder its effectiveness and benchmarking across countries. 4. IBM AND GENDER SENSITIVITY Because it collects evidence directly from beneficiaries, IBM has been highly effective at monitoring gender-related outcomes of projects. Generally, projects define the percentage of women to be targeted among beneficiaries, even though in many cases this percentage is not achieved. In several instances, alarming gender biases were uncovered by IBM reports. Beneficiaries of a cash transfer program turned out to be mostly men, as were the beneficiaries of the e-voucher program. In the former, the project transferred money to household heads. However, in Mali, 90 percent of household heads are men. This link between the criteria for selecting beneficiaries and local custom was disclosed by IBM. In the case of the e-voucher program, it chose to register land owners as beneficiaries of fertilizer instead of farmers who actually work the land. This criterion excluded women who work family land without being owners. In addition, agriculture products eligible for fertilizer were cereals, which are mostly produced by men. Vegetables and gardening products grown by women were not eligible. This choice excluded women as beneficiaries of the program. In another government project aiming at creating land commissions in each municipality to deal with land issues locally, it turned out that land commissions had almost no female members when in fact women are mostly affected by land issues. To be a member of a land commission, one should be a leader of a local association. However, in each municipality, there are hardly any women’s associations, while there are men’s associations aplenty. Therefore, few women were elected as members of land commissions. FIGURE 5: SELECTED GENDER OUTCOMES UNCOVERED BY DIFFERENT IBM ACTIVITIES CASH TRANSFER BENEFICIARIES E -VOUCHER BENEFICIARIES LAND COMMISSIONS MEMBERS Female, 22% Female, 8% Female, 6% Male, 79% Male, 92% Male, 94% 30 The adverse gender-related results uncovered by IBM were not the consequence of bad intentions. Projects were often designed with gender in mind and, in some instances, even employed gender specialists. Invariably, World Bank staff responded positively to the findings. Yet, a positive attitude alone is insufficient to ensure that gender biases are not perpetuated through project design and implementation. In some instances, the lack of gender sensitivity was a genuine oversight, and in the case of the e-voucher system, the approach to beneficiary registration was changed, and women were registered as potential beneficiaries along with household heads. As a result, the percentage of women beneficia- ries increased even if it did not reach the agreed 40 percent. In 2019, IBM will assess whether this figure has changed. FIGURE 6: PERCENTAGE OF WOMEN AMONG BENEFICIARIES OF FERTILIZER IN MALI Upon further reflection with managers of cash transfer projects, it was agreed that the issue could be addressed by reframing cash transfers as support to women as opposed to households, and additional financing for the social protection program under preparation will take this approach. Managers are also committed to making gender an agenda item during project implementation for each concept note and decision review for new projects and will continue to encourage the IBM team to collect information on gender outcomes from ongoing projects. 5. OTHER USES OF IBM 5.1 IBM to take over Enhanced M&E IBM can be used to pursue supervision activities undertaken through the Enhanced M&E (formally Third-Party Monitoring) mechanism, which is costly and difficult to repeat. This approach is being piloted in Gao region (Mali) on the Reconstruction and Economic Recovery Project (PRRE). This was introduced in a context where supervision of operations supported by the World Bank in Mali was constrained by protracted insecurity, particularly in areas directly affected by the 2012 conflict. In this context, the regular project supervision mechanism so far applied by the World Bank revealed limitations in terms of allowing the Bank’s teams to support the country in timely and effective fashion and to ensure that operations yielded expected outcomes on the ground. Hence, the Mali CMU engaged over time in discussions with specific Global Practices to initiate alternative approaches in support of more effective project supervision of operations by World Bank teams. One of those initiatives is Enhanced Monitoring and Evaluation (Enhanced M&E). 31 A pilot phase of Enhanced M&E is being conducted in Gao for the country’s Reconstruction and Economic Recovery Project (PRRE). The proposed Enhanced M&E arrangement is articulated around the combination of two complemen- tary instruments: one involving noncomplex activities targeting infrastructure projects that have already been completed, the other requiring relatively complex activities for the monitoring of ongoing or planned infrastructure investments. The first instrument relies on the Economic Statisticians Scientific Interest Group (GISSE), a firm that conducted a single field mission to collect data from all targeted sites where infrastructure works have already been completed in Gao. The second instrument is built around a non-governmental organization (NGO), the Action Research for Development Association of Mali (AMRAD), which conducts periodic missions to targeted sites as infrastructure works evolve to collect required data and information on both technical and social dynamics. The World Bank provides funding for the project and coordinates Enhanced M&E activities. The Project Implementation Unit (PIU) set up by the Government implements the project, signs contracts for third parties, and supervises them. FIGURE 7: DIAGRAMMATIC REPRESENTATION OF ENHANCED M&E IN MALI However, Enhanced M&E is expensive (see Table 3 for examples) due to the costs associated with insecurity risks and therefore cannot be repeated. It is funded through project resources that would otherwise be allocated to other activities. Thus, projects cannot afford repetition of the Enhanced M&E approach, and Enhanced M&E is usually a one-time activity that provides a snapshot of the project but cannot follow up on recommendations. To maintain permanent oversight of projects covered by third parties, these are requested to keep a record of the phone numbers of beneficiaries and all stakeholders and to take photos of infrastructure. Later, IBM is implemented to follow up on the recommendations from the Enhanced M&E and to assess changes using the beneficiary database constituted during Enhanced M&E visits on the ground. Beneficiaries are asked to send new photos of the infrastructure, which are then compared to earlier ones to confirm that requirements were met. For instance, during the Enhanced M&E of the PRRE project in Gao, the NGO and the private firms contracted provided photos of student desks in bad shape, and the procurement specialist required that those desks be replaced before acceptance of delivery and payment to the providers. By the time the desks were replaced, the Enhanced M&E activities had been completed, and the desk replacement was verified by IBM, which requested that school principals confirm the same with photos. 32 TABLE 3: COST OF ENHANCED M&E ON WORLD BANK PROJECTS COUNTRY PROJECT COST COST OF ENHANCED M&E (US$ thousands) (US$ millions) Cameroon Transportation Project 100 1,300 Mali Reconstruction and Economic Recovery Project 100 639 Somalia Reconstruction and Sustainable Development Project 250 12,000 South Sudan Migration Project 260 2,000 Afghanistan Reconstruction Project 3,300 30,500 Pakistan Crisis Recovery 200 1,600 Iraq Emergency Project on Education, Development, 955 1,800 and Transport 5.2 IBM in support of Development Policy Operations IBM has been used in Mali to consolidate evidence provided by the Government as part of the budget support program. A series of Development Policy Operations (DPO) has been implemented in the Mali CMU since 2016, focusing on poverty reduction and inclusive growth while providing support to the budget. While the old DPO series focused on strengthening public financial management (PFM) systems and governance, the ongoing series orients structur- al reforms toward specific sectors as well as inclusion. This series aims at unleashing the potential of key economic sectors such as agriculture, health, power, telecommunications, and roads. In so doing, it also aims at supporting the Government’s efforts to ensure inclusive and resilient growth through increased transfers to the poorest and most vulnerable population and the extension of social protection coverage. In Mali, IBM has been used as a tool to validate evidence of prior DPO actions completed by the Government. It was applied to assess the evidence for the e-voucher scheme, the land commissions, and the health insurance program for the extremely poor. In Guinea and Chad, IBM will support the DPO series based on indicators related to social projects in the agriculture and education sectors. As regards the e-voucher scheme in Mali, IBM confirmed the claim by the Government that the protocol for the distribution of fertilizer was followed. Sensitization messages needed to be sent at a specific moment before the onset of distribution. IBM used its call center in Mali to confirm with farmers that they received those messages as claimed by the Government. In addition, it was possible to assess the role of those messages’ usefulness in the success of the e-voucher scheme. Following the sensitization message, the DPO trigger requested that e-vouchers be sent at a specific moment to all farmers. As the lawyers argued that a simple letter from the Government was not enough, IBM was introduced in order to phone a sample of farmers to find out when they received their e-vouchers. Results of that enquiry, including the percentage of farmers who confirmed receipt of e-vouchers and the date at which they were received, were sent as evidence to the lawyers who validated the information. Regarding the health insurance program for the extremely poor, IBM confirmed that insurance certificates were distribut- ed and further compared the percentage of beneficiaries who received them to the target in the DPO matrix as well 33 as the claim by the Government. The rapid system confirmed whether beneficiaries received free healthcare when they got sick and went to the hospital as well as whether clinic managers were reimbursed by the project. A repetitive system was implemented to ensure there were no shortcomings at any stage in the project that could undermine its efficiency. Regarding land commissions, the lawyers wanted evidence that they had been created and were functional. After gathering this information, IBM was used to assess whether the land commissions organized meetings to address land issues in their municipalities by phoning a representative sample of members in each commission. Results were accepted as evidence by the lawyers. Since the operations of these commissions should be permanent, this frequent and rapid survey of commission members has become regular. 6. CONCLUSION IBM proved to be an effective tool that can enhance project impact. It was implemented in different sectors to help projects achieve their objectives effectively. However, IBM implementation faces a number of risks and challenges. The main risks are related to obtaining lists of project beneficiaries with their contact information, delays in addressing issues identified during the first round of IBM, network-related issues, and insecurity. Since these risks might undermine the implementation of the system, the success of IBM relies on the collaboration and engagement of those project managers who should make available the lists of beneficiaries. Given the positive effects of IBM in Mali and Niger, the high interest expressed by project leaders, and the involvement of project coordinators, the risk of non-collaboration is low. To mitigate the risk of delays in addressing issues identified and triggering subsequent IBM rounds, the IBM team can join the project team in finding solutions after dissemination and ensuring follow-up. To mitigate the impact of network-related issues, phone and field surveys can be combined. In the absence of beneficiaries’ phone numbers, enumerators are sent to the project area for data collection despite exposure to insecurity. In that case, data collection is kept to a minimum, and enumerators are introduced to local authorities to avoid confusion with other agents. During their first visit to the project area, enumerators collect phone numbers of beneficiaries and stakeholders for subsequent rounds of data collection. Challenges come from different sources. IBM’s iterative feedback approach is relatively straightforward, but applying it successfully requires care. Building a good rapport with a project team is critical because no one likes to receive negative feedback, although this is precisely what an iterative feedback system is meant to do. Confidentiality, good relations with project staff and the Government, and agreement on the shared objectives of the monitoring process are essential. Once the objectives of the monitoring process are clarified and aligned with those of those responsible for project implementation, reticence typically disappears. Starting IBM early in the project’s life increases positive impact. Hence, integrating an iterative monitoring approach into the project design has the benefit of identifying options for beneficiary monitoring early on. Small changes in the project design or in the procedural manual can greatly facilitate iterative monitoring. For instance, it makes a difference when the procedural manual stipulates that phone numbers and core characteristics of beneficiaries need to be captured in an electronic database that can be accessed for sampling and (anonymized) monitoring. In addition, when the procedural manual stipulates that certain benefits need to be distributed by a certain date, it offers a clear point in time at which progress toward project objectives can be measured. Even if an iterative monitoring approach is only designed during the project implementation phase, ways can be found to make follow-up monitoring easier. Registering the phone numbers of respondents during F2F interviews allows for easy follow-up. During each round of the IBM process on school meals, the phone numbers of respondents (cafeteria managers, head teachers, and household heads) were collected for future follow-up. Sometimes, feedback is offered voluntarily, with beneficiaries providing information to the project team, often by text message, about instances when the money for school feeding was exhausted before the expected date, whether the money arrived on time, or any other 34 issues affecting the operations of the cafeteria. When such information is received and deemed relevant, the project team can use the phone numbers of other beneficiaries to verify whether what was reported is a unique case or a more generalized problem. Another issue for consideration is who should conduct the monitoring. In Mali and Niger, staff from the Poverty and Equity Global Practice are responsible for data collection, while sector staff facilitate dialogue with the project teams. Working with Poverty and Equity Global Practice staff has major advantages since they have ample experience with sampling, designing instruments for data collection, training enumerators, and executing primary data collection activities as well as with data analysis and reporting. Moreover, its staff is familiar with the World Bank and its operations. Local presence is another important element for success. Presence facilitates building trust with the project teams and an understanding of how the project operates and makes it much easier to have discussions about results and corrective actions. Presence close to the location of project implementation also increases responsiveness, which is important when issues need to be identified and addressed quickly since lost days cannot be made up, missed meals cannot be replaced, and agricultural inputs distributed late are of little use to farmers. Familiarity with project procedures and staff facilitates the design of an iterative feedback loop, and outsourcing the approach in the same way as financial audits are outsourced is likely to be a challenge. However, an intermediate approach could work. An IBM specialist could be hired within projects and operate independently similar to procurement and financial specialists. Designing instruments and reporting could be left to staff familiar with household survey design and analysis, and dialogue with the client could be left to those responsible for the project, while data collection could be outsourced. This setup is feasible within the World Bank’s project architecture as staff time can be funded out of supervision budgets while data collection can be funded out of the M&E budget of each project. This institutional setup underscores the respective responsibilities of both the recipient government and the World Bank for project implementation and supervision while guaranteeing sufficient separation of functions to avoid reporting bias. 35 ANNEX 1. SAMPLING METHOD13 The sampling for IBM surveys varies from one project to another. There is no single methodology that can fit all monitored projects. In general, sample size depends on the budget and time constraints. Budget allocated for data collection should not exceed $5,000, which mostly pays for enumerators and credit for phone surveys. To maintain rapidity and ensure high data quality, IBM deploys enumerators in the field or uses phone calls for no more than two weeks. When the number of beneficiaries is less than 1,000, all of them can be sampled if time and budget allow it. This option was used in Niger on the e-voucher project, which had less than 500 beneficiaries. When project size exceeds 1,000 beneficiaries, IBM applies a formal sampling methodology. For instance, applying IBM to school meals, e-vouchers, and healthcare in Mali followed a probabilistic approach to sampling. To identify shortcomings hampering a project, it is not necessary to always target all beneficiaries or all regions where beneficiaries live. Everything depends on the challenges. At the first attempt, the team may consider a region close to the capital in order to assess the time taken for providing services or goods or for transferring money to beneficiaries. If this time is found to be excessive, it can be inferred that the situation is even more serious for remote areas. If not, subsequent rounds of IBM can go further in selecting the sample. From one round of IBM to another, it is advisable that the project team draw new samples, which may help reduce selection bias (if any) during sampling. As regards the school meals project in Mali, the first round of IBM focused on Koulikoro region, the nearest to Bamako. We expected that proximity to the capital would facilitate supervision and limit money transfer times. A total of 68 schools in that region benefited from the project. Given the number of questionnaires (school principals, school meals managers, and parents), the budget, and the time constraint, we arrived at a size of 20 schools to be interviewed in F2F mode. After numbering the schools, the sample of 20 was randomly selected using the systematic sampling approach. In each selected school, all principals and school meals managers were interviewed. In each class, five students were randomly selected, and their parents also participated in the survey. Based on the results, which showed serious problems in that region, the second round of IBM applied on 30 schools in addition to the first 20 still focused on in the same region. With positive results noted in the second round, the third round expanded to Gao and Mopti, which are far from the capital as well as affected by insecurity. For the e-voucher program as well as the health insurance program for the extremely poor in Mali, the lists of beneficia- ries was available with their phone numbers. Accounting for the time and budget constraints, we determined sample sizes (100 for the first round and 800 for the second round for the e-voucher program, and 700 for the health insurance program), with the samples selected using the systematic sampling method and covering all regions where the projects were implemented. About the rural mobility and connectivity project, the request from the project team was to measure the impact on populations living close to infrastructures being rehabilitated. In the absence of a list of those households, the sample (90 households) was selected in the field using the random walk method. Starting from a well-known point in the village (school, clinic, pharmacy, bakery, etc.), the first household is chosen. Then the enumerator moves forward, skips some households, and selects the second one. The process continues to cover the entire village until the enumerator reaches the number of households as defined in the village. The second round of IBM under preparation, which targets users of rehabilitated roads and bridges, plans to use the quota method. The universe populations for IBM in Mali were as follows: School meals (first and second rounds: 68 schools); E-vouchers (first round: 252,995 farmers; second 13  round: 97,476 farmers; third round: 92,792 farmers and 106 suppliers); Healthcare insurance (first and second rounds: 4,035 beneficiaries). In Niger, the universe populations were: 300 for the PPAAO, 12000 for the Safety net project, and 455 for PAC3. 36 2. QUESTIONNAIRE IBM questionnaires are specific to each project, they depend on challenges met under monitoring, and are adaptable from one round to another within the same project. There exists no standard questionnaire usable for IBM on all projects. However, IBM questionnaires focus on a maximum of five main challenges, and questions are phrased to monitor these. Questionnaire design is also based on the type of beneficiaries of the project (individuals, communities, NGOs, etc.). When respondents are individuals, the first questions capture the socio-demographic characteristics after information on geographic localization for spatial analysis. When the respondents are NGOs, municipalities, or other entities, question- naires start with geographic localization. Following those questions, which might be transferred from one project to another, the remainder of the questionnaire contains questions specific to the project and the challenges under consider- ation. Questionnaires used so far for IBM may be shared upon request. 37 MACROECONOMIC INDICATORS OF AFCW3 COUNTRIES AT A GLANCE, 2014-2019 Growth recovery consolidated in 2018, and ...with inflation rates marginally increasing with prospects for this year are positive... recovery, but remaining in single digits. Chad Mali Guinea Niger Chad Mali Guinea Niger GDP growth rate (%) Annual, average 2014 2015 2016 2017 2018e 2019p 2014 2015 2016 2017 2018e 2019p High imports related to mining kept current ...while terms of trade appear improving, but account deficits high in Guinea and Niger... not as good as last year for Chad. 2014 2015 2016 2017 2018e 2019p 2014 2015 2016 2017 2018e 2019p Annual, % change % of GDP Chad Mali Guinea Niger Chad Mali Guinea Niger Fiscal deficits were high in 2018, except for ...even though they are projected to be Chad, and need further consolidation... supported by increased revenue. 2014 2015 2016 2017 2018e 2019p Chad Mali Guinea Niger Including grants, cash basis, % of GDP Excluding grants, % of GDP Chad Mali Guinea Niger 2014 2015 2016 2017 2018e 2019p Source: IMF and World Bank staff estimates; IMF AIV, several years. Data for 2018 are projections and estimate in some cases. Data for 2019 are projections. Fiscal deficits include grants and are on cash basis Note:  (except for Niger – on commitment basis); they may slightly differ from those reported in the text done on a commitment basis. 38 chad The economic situation improved, but recovery remains weak. Real GDP growth reached 2.6 percent in 2018 due to higher oil prices and agricultural output. Stronger revenue collection and a declining wage bill resulted in a fiscal surplus. In the medium term, oil exports could support growth acceleration to about 5 percent while strengthening fiscal and current account balances. However, oil price volatility, insecurity, and banking sector vulnerabilities pose downside risks. Poverty is expected to decline, though at a slower pace due to high population growth. RECENT DEVELOPMENTS The economy recovered in 2018 following two years of severe recession. Growth is estimated at 2.6 percent in 2018, thanks to increases in oil prices, oil production, and agricultural output. The substantial negative output gap is beginning to close as GDP growth converges to its potential rate (1 percent). The narrowing output gap is consistent with an acceleration in CPI inflation, which reached 2.1 percent in 2018, up from -0.7 percent in 2017. The primary sector (mainly agriculture and oil sub-sectors) contributed about 2 percentage points to 2018 headline growth. In contrast, the contributions of the secondary and tertiary sectors stood at 0.1 and 0.7 percentage points, respectively. The improvement in the industrial sector indicates a slow rise in capital investment, while services benefited from strong primary sector activity and arrears repayment by the central government. The external current account deficit fell from 5.1 percent in 2017 to 4.2 percent in 2018 thanks to stronger oil export performance. Following a contraction in 2017, imports grew by 1.0 percent as private consumption and capital investment started to pick up. The financial account also improved as a result of lower debt servicing to Chad’s main private creditor (Glencore). The Government continued to pursue fiscal consolidation by mobilizing revenues and containing recurrent spending. Total revenues increased from 14.2 percent of GDP in 2017 to 15.5 percent in 2018 thanks to a boost in oil revenues. The collection of non-oil revenues increased as taxes were paid through commercial banks instead of as direct payments to the Treasury. Total spending remained stable at around 14.6 percent of GDP in 2018, supported by a significant drop (rise) in the wage bill (capital expenditure) from 6.5 (3.6) percent of GDP in 2017 to 5.6 (4.4) percent in 2018. The overall fiscal deficit is estimated to decline from -1.4 percent of GDP in 2017 to a surplus of 0.8 percent in 2018. In June 2018, Chad finalized the restructuring of its oil-collateralized debt with Glencore. The agreement includes a cash sweep mechanism linking interest and amortization payments directly to oil revenue performance. As a result, public debt returned to a sustainable path characterized by a significantly lower debt service-to-revenue ratio. Public debt stood at 49.2 percent in 2018. The risk of external debt distress remains high. 39 As a member of the Central African Economic and Monetary Union (CEMAC), Chad’s monetary policy is conducted by the regional Central Bank (BEAC). Following the oil price shock and subsequent plunge in foreign exchange reserves to 2.7 months at end-2017, BEAC implemented tighter monetary policies to rebuild regional reserve buffers and ensure financial sector stability. The Central Bank eliminated statutory advances and increased its policy rate from 2.95 percent in March 2017 to 3.5 percent in October 2018. Poverty and vulnerability are pervasive in Chad. According to the latest national household survey (2011), 29 percent of the population falls below the food poverty line, 47 percent below the national poverty line, and 68 percent are considered vulnerable. OUTLOOK Oil exports should remain a key driver of growth in the medium term. In addition, the privatization of the cotton public enterprise is expected to significantly improve the contribution of agriculture to growth. Lower than expected oil prices imply more moderate growth acceleration in 2019. However, the expected boost in oil production should ignite investment and exports in 2020, yielding about 5.6 percent real GDP growth. Convergence to the potential growth rate should lead to a slowing rate of 4.8 percent by 2021. Import growth should accelerate to 4.1 percent by 2021 thanks to higher consumption and investment, albeit outweighed by stronger export growth. Consequently, the current account deficit should stabilize at around 5 percent of GDP by 2021. CEMAC’s regional reserves are estimated at about 4.2 months of imports by 2021 as the BEAC continues to tighten monetary policy and exceptional financing. The Government is expected to further rationalize current expenditures while strengthening revenue mobilization efforts. Thus, the overall fiscal balance should remain in surplus and could rise to about 1.2 percent of GDP in 2021. The Glencore agreement and expected clearance of arrears will decrease the public debt-to-GDP ratio from 49.2 percent in 2018 to 37.0 percent in 2021. With solid real GDP growth over 2019–2021, poverty (using the international poverty line of US$1.90 a day in PPP terms) is expected to decline from 40 to 38 percent. However, with annual population growth at 3.3 percent, the absolute number of poor remains unchanged at about 6.3 million in 2021. RISKS AND CHALLENGES Chad’s economic recovery remains fragile and subject to significant risks. Oil price volatility poses both upside and downside risks to the economy. The cash sweep mechanism with Glencore dampens the fiscal impact of such volatility as debt repayments increase with rising oil prices (and vice versa). A further decline in bank liquidity and the potential for additional domestic arrears increase financial sector vulnerabilities. A potential rise in the wage bill and total debt stock may shrink fiscal space in the medium term. Lastly, general elections potentially scheduled in Q3 of 2019 could delay important policy reforms. To mitigate these risks and reduce poverty, Chad needs to invest oil revenues strategi- cally in key sectors such as infrastructure, agriculture, education, and health. 40 KEY ECONOMIC AND FINANCIAL INDICATORS: 2015–2021 2015 2016 2017 2018(e) 2019(p) 2020(p) 2021(p) Real Economy (annual percentage change unless otherwise specified) Real GDP 2.8 -6.3 -3.0 2.6 3.4 5.6 4.8 Oil GDP 32.1 -11.2 -16.2 10.9 12.6 19.4 9.0 Non-oil GDP -2.9 -6.7 -0.5 1.0 1.5 2.4 3.7 Per Capita GDP (US$) 962.7 874.8 823.1 820 823.2 844.4 859.5 GDP Deflator (level) -8.0 -1.2 1.1 1.2 1.3 1.3 1.3 Consumer Price Inflation (average) 3.7 -1.1 -0.7 2.1 2.6 3.1 3.0 Oil Prices WEO (US$/barrel) 50.8 42.8 52.8 68.3 61.8 61.5 60.8 Chad Price (US$/barrel)13 39.9 36.2 49.4 63.3 57.8 57.5 56.8 Oil Production (millions of barrels) 47.5 44.4 35.9 39.8 44.8 53.5 58.5 Fiscal Accounts (percentage of non-oil GDP unless otherwise specified) Expenditures (total) 22.9 18.0 18.0 18.3 18.3 18.4 18.0 Revenues and Grants (total) 17.1 14.9 17.1 19.4 18.5 20.0 19.5 General Government Balance (incl. grants, commitment basis) -5.8 -3.0 -1.0 1.0 0.2 1.6 1.5 Overall Balance (including grants, cash basis) -4.5 -5.2 -2.5 -0.6 -0.4 0.9 0.8 Non-oil Primary Balance (commitment basis, excluding grants) -9.7 -4.4 -3.8 -4.4 -4.1 -3.7 -3.0 Selected Monetary Accounts (annual percentage change unless otherwise specified) Base Money -4.7 -7.7 -4.3 7.2 … ... ... Credit to the Private Sector 1.1 -2.7 -1.7 1.0 … ... ... Interest (BEAC key policy rate) 2.45 2.45 2.95 3.50 3.50 2.95 2.95 External Sector Exports of Goods and Services (GNFS) USD) -33.3 -23.1 25.0 14.6 7.7 10.7 9.1 Imports of Goods and Services (GNFS) USD) -24.7 -12.0 4.1 2.9 5.2 5.6 5.7 Terms of Trade -38.3 -6.9 28.1 25.8 -0.1 -5.1 -4.2 Balance of Payments (percentage of GDP unless otherwise specified) Current Account Balance (incl. transfers) -11.3 -13.0 -5.1 -4.2 -5.7 -4.9 -5.1 Gross Reserves (US$ billions, EOP) 0.4 -0.3 0.0 0.1 0.3 0.5 0.7 Gross Reserves (regional, months of imports 4.3 2.3 2.7 3.0 3.7 3.9 4.5 of goods and services) External Debt 25.0 27.2 27.3 26.2 25.0 22.2 19.6 Exchange Rate (period average) in USD/CFAF 591.2 592.7 592.7 580.9 — — — Memorandum Items: Nominal Non-oil GDP (CFAF billions) 5,184 4,838 4,829 5,011 5,283 5,631 6,028 Nominal GDP (CFAF billions) 6,474 5,984 5,746 6,079 6,455 7,103 7,669 Sources: World Bank MFMOD, IMF, and Chadian Authorities. Note: (e) = estimated; (p) = projected 14 The Chad oil price is the Brent price minus a quality discount. 41 GUINEA Growth slowed to 5.8 percent in 2018 as the recent mining sector boom cooled. The external current account deficit increased as export growth slowed. The outlook is positive, supported by substantial mining-related FDI and infrastructure investments. The extreme poverty rate is expected to decline further. Declining commodity prices, lower-than-forecast mining production, and election-related fiscal slippages are downside risks to the outlook. RECENT DEVELOPMENTS Growth slowed from 9.9 percent in 2017 to 5.8 percent (3.1 percent in per capita terms) in 2018, though GDP remains well below potential output. Despite strong FDI, mining sector growth slowed from 52.3 percent in 2017 to 6.7 percent in 2018. However, non-mining growth increased from 4.1 percent in 2017 to 5.6 percent in 2018 thanks to investments in infrastructure. Inflation approached double digits at 9.9 percent in 2018 driven by an increase in fuel prices and electricity tariffs in the context of a positive output gap. The external current account deficit increased from 6.8 percent of GDP in 2017 to 16.1 percent in 2018, in line with its historical average. A slowdown in export growth (7.3 percent) and faster expansion of FDI-induced imports (19.8 percent) drove this increase. FDI flows of 13.2 percent of GDP helped finance the current account deficit. Gross international reserves increased to 3.2 months of imports in 2018 (up from 2.4 months in 2017), reflecting larger than expected external project loans and budget support. The fiscal deficit increased slightly from 2.0 percent of GDP in 2017 to 2.1 percent in 2018. However, the primary deficit declined from 1.2 percent of GDP in 2017 to 0.9 percent in 2018. This improvement resulted from a reduction in energy and fuel subsidies and caps on public sector recruitment and promotions. Tax revenues increased slightly thanks to stronger mining tax revenues amidst weaker tax collection from international trade. Public debt declined from 39.9 percent of GDP in 2017 to 38.9 percent in 2018 aided by strong nominal GDP growth. External non-concessional commitments increased with the signing of the loan for the Souapiti dam (US$1.2 billion). Two other non-concessional loans were signed for the rehabilitation of the RN1 national highway and urban roads in Conakry (US$598 million) in 2018, which were collateralized with mining tax revenues. The risk of external debt distress is moderate. The Government took important measures to improve monetary policy. The Central Bank reduced its financing of the Government within statutory limits. The Government responded with higher borrowing from commercial banks, which grew 28 percent at end-September (year-on-year). Nevertheless, reserve money growth slowed to 6 percent at end-September. Furthermore, the Central Bank strengthened its monitoring and reporting system for Government financing by providing its board with regular reports. The recapitalization of the Central Bank (about US$300 million) has strengthened its autonomy. The Central Bank targets base money in line with the IMF’s program to preserve moderate inflation. Credit to the private sector picked up, reaching 8.8 percent at end-September (year-on-year). 42 Extreme poverty in Guinea is estimated to have declined gradually. Despite the robust recovery, food price inflation has put downward pressure on household purchasing power. Projections based on per capita GDP growth suggest that an estimated 26 percent of the population lived below the international poverty line (US$ 1.90 per day at 2011 PPP) in 2018. This reflects 3.3 million poor living in extreme poverty. Non-monetary poverty remains high. OUTLOOK The medium-term economic outlook is positive. Growth is expected at about 6 percent over the period 2019–2021, with an output gap remaining significantly positive throughout the period. Mining and mining-related infrastructure investment should continue to drive growth, financed by FDI inflows. Construction is also projected to grow strongly, with public and private investment into energy and transportation infrastructure. Agriculture productivity will improve thanks to better provisioning of agriculture inputs and institutional and infrastructure improvements. The fiscal deficit (including grants) is expected to decline to about 1 percent of GDP by 2021, driven by higher tax revenues, which will increase from 13.5 percent of GDP in 2018 to 15.1 percent in 2021. The Government will continue to implement tax revenue mobilization measures, such as better rationalization of ad-hoc tax exonerations, simplifi- cations to the tax code, a review of international tax rules and provisions, a clean-up of the taxpayers’ databases, and setting up a new organizational structure for the Directorate General of Taxes. Over 2019–2021, exports are projected to grow by 7.9 percent, driven by new mining production, while imports will grow by 4.9 percent, driven by intermediate and equipment imports. As a result, the external current account deficit is projected to decline to 12.9 percent of GDP by 2021. FDI inflows will meet over 70 percent of financing requirements between 2019 and 2021, with long-term loans meeting the rest. The Central Bank plans to pursue a more prudent monetary policy and contained financing of the Government. The targeting of reserve money growth will be supported by better liquidity management. Thus, inflation is expected to decline slightly to 8 percent in 2021, or about 1 percentage point lower than average annual inflation rate over the last five years. The independence of the Central Bank will continue to be strengthened as part of the IMF program. The extreme poverty rate is projected to decrease further from 26 to 22 percent by 2021. RISKS AND CHALLENGES Downside risks to the economic outlook prevail. A growth slowdown in China or advanced economies or weaker bauxite demand owing to US tariffs on China’s aluminum output could reduce investment in Guinea as a result of weaker demand for its mineral resources. Lower commodity prices (especially for bauxite and gold) or slow progress on infrastructure development could also lower growth. Tighter or more volatile global financial conditions, such as a surge in the US dollar, could impair competitiveness and strain reserve buffers. Conversely, a faster increase in mining production capacity or infrastructure or higher commodity prices would support higher growth. In addition, fiscal slippages (from lower revenue mobilization, weak prioritization of public investment projects, or higher current expenditures) could lead to inflationary financing by the Central Bank, increased borrowing, and lower medium-term debt sustainability. A greater-than-forecast recourse to debt, especially non-concession debt, would also worsen debt metrics. Increased spending during the up-coming election period or in response to political protests or labor union activity poses additional risks. Future poverty reduction will benefit from higher productivity in agriculture and from creating fiscal space to finance investments into human capital and social protection. 43 KEY ECONOMIC AND FINANCIAL INDICATORS: 2016–2021 2016 2017 2018(e) 2019 (f) 2020(f) 2021(f) (annual percentage change, unless otherwise specified) National Accounts and Prices GDP at Constant Prices 10.5 9.9 5.8 5.9 6.0 6.0 GDP at Current Prices 17.8 22.0 15.9 15.3 14.5 14.5 GDP Deflator 6.7 11.0 9.5 8.8 8.0 8.0 Consumer Prices Annual Average 8.2 8.9 9.9 8.9 8.3 8.0 End of Period 8.7 9.5 9.6 8.6 8.0 7.9 External Sector Exports (US$ terms) 16.2 47.4 0.1 -0.7 14.0 8.9 Imports (US$ terms) 82.4 -8.7 19.4 9.5 5.0 0.1 Money and Credit Net Foreign Assets 7.3 9.6 5.8 6.3 6.7 6.4 Net Domestic Assets 2.7 6.2 6.3 6.1 6.3 6.1 Net Claims on Government 1.9 5.0 -0.3 -0.3 -0.7 -0.5 Credit to Non-government Sector 2.4 0.9 6.5 6.4 7.0 6.6 Reserve Money 15.5 10.3 14.5 9.9 10.9 10.8 Broad Money 9.9 15.8 12.0 12.4 13.0 12.5 Central Government Finances (as percentage of GDP) Total Revenue and Grants 15.8 15.2 15.6 15.6 16.4 16.9 Revenue 13.2 13.2 13.5 13.9 14.7 15.1 Grants 2.1 1.5 1.2 1.1 1.1 1.2 Total Expenditure and Net Lending 16.0 17.2 17.7 17.9 18.1 17.9 Current Expenditure 11.6 11.3 11.6 10.5 11.3 11.3 Capital Expenditure 4.7 5.7 6.1 7.6 7.1 6.6 Overall Budget Balance Excluding grants -0.2 -2.0 -2.1 -2.3 -1.7 -1.0 Including grants -2.3 -3.5 -3.3 -3.4 -2.8 -2.2 Current Account Balance (as percentage of GDP) Excluding official transfers Including official transfers -31.6 -6.8 -16.1 -20.2 -17.1 -12.9 Overall Balance of Payments 0.8 0.7 1.6 1.4 1.2 1.6 Gross Reserves (in months of imports) 2.4 2.4 3.2 3.5 3.8 4.0 External Public Debt 21.6 19.2 21.3 31.1 32.4 31.7 Nominal GDP (GNF billions) 77,899 94,491 109,610 126,271 144,706 165,689 Source: World Bank MFMOD, IMF, and Guinean authorities Note: (e) = estimated; (p) = projected 44 mali Growth slowed further to 4.9 percent in 2018 amidst fiscal slippages and election-related uncertainty. Poverty declined thanks to a substantial increase in agricultural production. The fiscal deficit widened due to a substantial decline in revenue while the external position worsened following a deterioration in the terms of trade. The outlook is clouded by downside risks related to insecurity, fiscal risks from the energy utility, and potential exogenous shocks. Going forward, Mali needs to restore security, improve tax collection, and further diversify its economy. RECENT DEVELOPMENTS Growth declined for a third consecutive year to 4.9 percent (1.9 percent in per capita terms) in 2018 as the economy reverted to its potential rate of growth. Slower growth is partly due to the spread of insecurity in central and southern regions. Uncertainty in the run-up to the presidential election also played a role. On the demand side, total investment fell by more than 2 percent of GDP as the Government cut public investment in response to a steep shortfall in revenue. On the other hand, reduced revenue collection helped support private domestic demand. Inflation increased slightly from 1.8 percent in 2017 to 2.0 percent in 2018. Despite the surge in international oil prices (23.2 percent), domestic prices only increased 3 percent. This helped contain inflation but came at a fiscal cost as budgetary subsidies increased. The external account deficit widened from 5.0 to 7.4 percent of GDP in 2018 as the rise in oil-related imports outweighed solidly-performing cotton and gold exports. Gold output is estimated to have risen by 21 percent, and the country harvested an estimated record crop of 705,000 tons of cotton (25 percent growth). The current account deficit was financed by a combination of FDI (60 percent) and external borrowing (40 percent). The fiscal deficit increased from 2.9 percent of GDP in 2017 to 4.8 percent in 2018 due to a substantial and unexpected shortfall in revenue. Tax revenue declined from 15.2 percent in 2017 to 11.9 percent in 2018. This weak performance is primarily due to severe non-compliance by taxpayers ahead of the presidential election. Revenue losses due to the limited effect of international oil price increases on domestic prices and increases in tax exemptions also played a role. Higher tax avoidance was also associated with growing insecurity and recurrent terrorist attacks, which led to the closure of customs offices in affected areas. In response, the Government undertook spending cuts of more than 2 percent of GDP. Expenditure cuts affected mainly public investment across all sectors, including health and education. The fiscal deficit was mainly financed by issuing regional bonds as the country also endured a drop in external financing. Public debt increased from 36.1 to 38.3 percent of GDP in 2018. The risk of external debt distress remains moderate. Mali’s monetary and exchange rate policies are managed by the Central Bank of West African States (BCEAO), which maintains a fixed peg between the CFA Franc and the Euro. The BCEAO’s international reserves stabilized in 2018, supported by significant Eurobond issuances by Côte d’Ivoire and Senegal. Reserves are estimated to have reached 4.1 months of prospective imports of goods and services at end-2018, up from 3.8 months at end-2017. Following tighter 45 monetary policy between early 2017 and end-June 2018, The BCEAO reduced its refinancing to banks by 24 percent while regional liquidity pressures were temporarily alleviated by Eurobond issuances. The real effective exchange rate (REER) remained broadly stable and in line with fundamentals, with movements generally following those of the Euro/ USD exchange rate. Continued fiscal consolidation among member countries is needed to support regional reserves. The extreme poverty rate is estimated to have declined from 46.3 percent in 2015 to 42.7 percent in 2017. Strong agricultural production, including cotton, and tertiary sector expansion likely increased consumption among rural households and induced a further decline in poverty. OUTLOOK Growth is projected at around 5 percent over the medium term, in line with the potential growth rate. Primary sector growth should slow to its long-term trend but remain high and contribute substantially to growth. The tertiary sector will be one of the main drivers of growth, supported by telecommunications and transportation and to a lesser extent the development of trade and financial services. The external current account deficit is projected to remain above 7.0 percent of GDP over the medium term. The deficit is expected to be financed by both FDI and public borrowing. The fiscal outlook is challenging, and strong consolidation efforts are needed in the short term, especially on the revenue side. The weaker performance experienced in 2018 will make it challenging to reach the WAEMU fiscal deficit target of 3 percent of GDP. The poverty rate is projected to decline steadily provided the robust expansion of Mali’s economy continues over the period 2017–2019 and the security threat does not spread further south. Under those assumptions, per capita GDP will rise, with a concomitant reduction in the poverty rate to about 40.2 percent in 2019. RISKS AND CHALLENGES The most significant and unpredictable risks remain the fragile security situation. Preventing the spread of insecurity toward the southern regions of the country is critical as such a development would come at a significant cost, affecting the agricultural and mining sectors and inducing an economic downturn. A negative weather-related shock would reduce agricultural growth, aggravate food insecurity, and create inflationary pressures, thereby raising social spending needs. Given Mali’s limited fiscal buffers, such risks could lead to an under- execution of public investment projects and an accumulation of expenditure arrears. An unexpected strong deterioration in the terms of trade would further worsen the fiscal and external imbalances and dampen growth. Mali remains resource-dependent, and further diversification of exports is critical to achieving macroeconomic stability. Other external risks arise from potential reductions in donor funding and falling FDI. Finally, at the regional level, further monetary policy tightening by the BCEAO could slow credit to the private sector and increase the cost of domestic debt. The energy utility (EDM) constitutes a main fiscal risk, while the surge of domestic debt could weigh on public finances. 46 TABLE 1: KEY MACROECONOMIC INDICATORS, 2016–2022 2016 2017 2018(e) 2019(p) 2020(p) 2021(p) 2022(p) Real Economy (annual percentage change unless otherwise specified) GDP (nominal, CFAF billions) 8,308 8,929 9,557 10,221 10,950 11,727 12,564 Real GDP 5.8 5.4 4.9 5.0 4.9 4.8 4.8 GDP Deflator 1.4 2.0 2.0 1.9 2.1 2.2 2.2 Consumer Price Inflation (average) -1.8 1.8 2.0 2.0 2.1 2.2 2.2 Fiscal Accounts (percentage of GDP unless otherwise specified) Total Expenditure 22.3 22.9 20.0 23.5 24.1 24.4 24.6 Total Revenue 16.7 18.4 14.2 18.5 19.1 19.3 19.6 Grants 1.6 1.6 1.0 2.0 2.0 2.1 2.1 General Government Balance -3.9 -2.9 -4.8 -3.0 -3.0 -3.0 -3.0 Public Debt 35.9 35.4 37.3 37.5 38.1 38.7 39.3 Domestic Debt 11.0 11.0 12.5 13.3 14.1 14.9 15.5 Selected Monetary Accounts (contribution to broad money growth) Credit to the Government 10.4 3.9 7.4 5.0 — — — Credit to the Economy 13.7 6.3 9.4 8.6 — — — Broad Money (M2) 7.3 7.9 17.0 7.7 — — — Balance of Payments percentage of GDP unless otherwise specified) ( Current Account Balance -7.2 -5.9 -7.4 -7.2 -7.4 -7.2 7.0 Imports 40.3 38.9 38.9 38.1 36.7 35.3 34.0 Exports 23.5 23.1 23.1 21.5 19.9 18.9 18.0 Foreign Direct Investment 1.8 2.7 2.7 2.7 2.7 2.7 2.7 External Debt 25.0 24.4 24.8 24.2 24.0 23.8 23.9 Terms of Trade 15.5 -1.3 -4.9 -3.1 0.2 1.1 0.9 Other memorandum items GDP nominal in US$ (billions) 13.3 16.1 17.4 18.4 19.8 21.2 22.7 Source: Ministry of Finance, IMF, Bank staff estimates (2016-2018) and projections (2019-2022), December 2018 Note: (e) = estimated; (p) = projected 47 niger Growth trended upward, reaching 5.2 percent in 2018. The fiscal deficit declined, while the external current account rose. The outlook is positive, with growth projected to average 6 percent over 2019–2021. Poverty is expected to decrease between 2018 and 2020. Downside risks include commodity price fluctuations, climatic vagaries, and insecurity. Gender inequality, high population growth, and low human development remain key challenges. RECENT DEVELOPMENTS Growth reached 5.2 percent in 2018 (1.4 percent in per capita terms), up from 4.9 percent in 2017, performed slightly above its potential growth rate of 4.9 percent. Growth was mainly driven by services, construction, and agriculture. Aggregate demand was supported by a surge in private consumption and public and private investment as Niger prepared to host the 2020 African Union Summit. Inflation increased from 2.4 percent in 2017 to 2.7 percent in 2018. Slightly higher inflation reflected administrative price and tax hikes as well as opportunistic price increases amid limited competition and the introduction of new taxes. The external current account deficit (including grants) widened from 16.2 percent of GDP in 2017 to 19.2 percent of GDP in 2018. This deterioration was driven mainly by a surge in both private and public investment-related imports. The current account deficit was mostly financed by foreign private investments and project loans. The fiscal deficit (on a commitment basis, including grants) declined from 5.7 percent in 2017 to an estimated 4.4 percent of GDP in 2018. This was the result of an increase in total revenues and grants of 2.8 percent of GDP. Tax revenue increased by 2.0 percent of GDP. Non-tax revenue also increased thanks to one-off factors, including the sale of telecom licenses and an oil contract signing bonus. The increase in total revenue offset a total expenditure increase of 1.8 percent of GDP, resulting largely from an increase in foreign-financed projects, which also drove an increase in imports, as mentioned above. Recurrent expenditure declined by 0.5 percent of GDP thanks to sustained efforts to control wages and salaries. Public debt increased from 49.2 percent of GDP in 2017 to 50.5 percent of GDP in 2018, of which 34.0 percent was external debt. Niger’s risk of external and overall public debt distress is moderate. Niger’s monetary and exchange rate policies are managed by the Central Bank of West African States (BCEAO), which maintains a fixed peg between the CFA Franc and the Euro. BCEAO’s international reserves stabilized in 2018, supported by significant Eurobond issuances by Côte d’Ivoire and Senegal. Reserves are estimated to have reached 4.1 months of prospective imports of goods and services at end-2018, up from 3.8 months at end-2017. Despite tighter monetary policy between early 2017 and end-June 2018, with the BCEAO reducing its refinancing to banks by 24 percent, regional liquidity pressures were temporarily alleviated by Eurobond issuances. The real effective exchange rate (REER) remained 48 broadly stable and in line with fundamentals, with movements generally following those of the Euro/USD exchange rate. Continued fiscal consolidation among member countries is needed to support regional reserves. Niger made progress in reducing poverty, from about 50.3 percent to 45.7 percent (using the international poverty line of US$1.90 per day) between 2005 and 2014. This suggests that extreme poverty may have worsened over the same period, with the bottom 10 percent of the population experiencing negative growth in consumption. Meanwhile, the Gini coefficient deteriorated markedly from 28.6 percent in 2005 to 33.6 percent in 2014. OUTLOOK The outlook remains broadly positive. GDP growth is projected to average 6 percent during 2019–2021, driven mostly by the agriculture sector thanks to additional Government efforts at enhancing productivity. Continued investment aimed at reducing the infrastructure gap in electricity and ICT, in which the private sector would play a larger role, including through PPP and foreign-direct investment, will also support growth. Finally, the construction of an oil pipeline to export crude oil will drive up investment. With temporary supply side pressures fading and given efforts to preserve the stability of monetary policy within the WAEMU framework, inflation is expected to settle at around 2 percent. WAEMU reserves are projected to reach about 4.6 months of imports by 2022 as member countries implement fiscal consolidation measures and external competitiveness improves. The external current account deficit (including grants) should remain high, averaging 20 percent of GDP over 2018–2021. Although export volumes are expected to register good performances as official exports to Nigeria and oil and non-oil exports continuing to improve, this should be offset by public and private investment with high-import content. The current account balance is only expected to improve in 2022 as crude oil exports begin and project-related imports are progressively phased out. The current account deficit should continue to be primarily financed by FDI and project loans. The overall fiscal deficit is projected to reach the WAEMU convergence criterion of 3 percent of GDP by 2020, one year later than other member countries. Revenue mobilization will remain the main driver, but recurrent expenditure rationalization will also contribute, and expenditure restraints, especially on capital spending, will serve as a second line of defense should revenue fall short. Planned investments in boosting agricultural productivity and expanding access to energy and digital technologies in rural areas is expected to lead to poverty reduction. Using the international poverty line (US$1.90 a day in PPP terms), poverty is expected to decline by 2.5 percentage points over 2018–2020. RISKS AND CHALLENGES Downside risks to the economic outlook prevail. The economy remains subject to the vagaries of weather conditions and commodity price fluctuations. The regional security situation is a concern. Niger’s economy therefore remains vulnerable to a prolonged downturn should these shocks materialize. Slow structural reform implementation would not allow Niger to move at a steady pace toward stable growth and macroeconomic stability. Shortcomings in carrying out productive investments in transportation, energy, telecom and agriculture sectors would affect growth negatively in the medium term. Similarly, failure to broaden the tax base, contain expenditures, and enhance the efficiency of public investment would jeopardize fiscal and debt sustainability. Gender inequality, high population growth, and low human development remain key challenges. 49 KEY MACROECONOMIC AND FINANCIAL INDICATORS, 2015–2021 2015 2016 2017 2018(e) 2019(p) 2020(p) 2021(p) Real Economy (annual percentage change unless otherwise specified) Real GDP 4.0 4.9 4.9 5.2 6.5 6.0 5.6 Non-resource GDP 4.5 4.9 4.5 5.3 5.4 6.0 6.1 Export Volume -4.5 -2.0 13.5 2.1 11.8 10.3 5.4 Real per Capita GDP 0.1 1.0 1.0 1.3 2.6 2.1 1.5 Import Volume 7.3 -14.1 10.2 12.7 16.4 13.8 -5.8 GDP Deflator 0.5 0.2 -0.1 3.3 2.4 1.9 2.0 CPI Annual Average 1.0 0.2 2.4 2.7 2.4 2.1 2.0 CPI End-of-period 2.2 -2.2 4.8 2.9 2.2 2.0 2.0 Fiscal Accounts percentage GDP unless otherwise specified) ( Total Revenue and Grants 23.4 20.5 21.4 22.7 24.3 25.2 24.9 Total Expenditure and Net Lending 32.5 26.3 26.8 28.5 28.4 28.0 27.1 Current Expenditure 15.5 14.0 14.1 13.6 13.5 13.1 12.8 Capital Expenditure 17 12.3 12.7 14.9 14.9 14.9 14.4 Overall Balance -9.1 -6.1 -5.7 -4.4 -4.5 -3.0 -2.7 (commitment basis, including grants) Selected Monetary Accounts (annual change in percentage of beginning-of-period broad money) Broad Money 4.6 8.7 -4.9 12.0 7.4 8.5 7.5 Credit to Non-government 6.3 3.8 5.3 13.1 12.5 11.7 11.2 Balance of Payments percentage of GDP unless otherwise specified) ( External Current Account Balance -20.5 -15.5 -16.2 -19.2 -21.1 -23.3 -19.4 (including grants) Imports 27.4 22.6 24.7 26.3 28.4 30.2 27.0 Exports 15.1 13.6 14.6 14.3 14.7 14.8 14.9 Foreign Direct investment 6.9 3.7 3.7 5.3 6.9 9.1 6.7 Total Public and Publicly Guaranteed Debt 41.6 43.7 49.3 50.5 50.8 49.8 48.3 Public and Publicly Guaranteed External Debt 30.3 32.6 32.4 34.0 34.9 36.1 36.6 Public Domestic Debt 11.4 11.1 16.8 16.5 15.8 13.6 11.8 Terms of Trade (percentage change) -7.5 -4.4 -4.2 -1.7 -0.7 -1.7 0.1 Memorandum Items GDP (Nominal-local currency) 4,269 4,511 4,726 5,135 5,600 6,051 6,518 Source: Ministry of Finances, Niger; World Bank (2019) Note: (e) = estimated; (p) = projected 50