FUTURE of FOOD Harnessing Digital Technologies to Improve Food System Outcomes © 2019 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Authors Robert Townsend, Julian Lampietti, David Treguer, Kateryna Schroeder, Mekbib Haile, Armine Juergenliemk, Eva Hasiner, Alexandra Horst, Artavazd Hakobyan with inputs from Panos Varangis, Diego Arias, Luda Bujoreanu, Jeehye Kim, Erick Fernandes, Parmesh Shah, Dina Umali-Deininger, Michael Morris, Flore de Preneuf, Marianne Fay, Jean Saint-Geours, Anup Jagwani, Gene Moses, Ernest Bethe, Marta Bogdanic, Michel Rogy, Jane Treadwell, Raman Krishnan, Katie Kennedy Freeman, Joanne Gaskell, Mercedes Stickler, Ashwini Sebastian, Hyea Won Lee, Nadim Khouri, Li Guo, and Astrid Jakobs de Padua. Overall guidance was provided by Juergen Voegele, Martien van Nieuwkoop, and Simeon Ehui. Images Front cover: Flore de Preneuf/World Bank Title page: Flore de Preneuf/World Bank Interior: 2, Dominic Chavez/World Bank 5, Bart Verweij/World Bank 9, Mai Ky/World Bank 10, Simone D. McCourtie/World Bank 18, Tom Perry/World Bank 30, Neil Palmer/CIAT 31, Neil Palmer/CIAT 33, Maria Fleischmann/World Bank 34, Chhor Sokunthea/World Bank 40, Neil Palmer/CIAT FUTURE of FOOD Harnessing Digital Technologies to Improve Food System Outcomes April 2019 FOREWORD New and expanding information and com- The World Bank Group is strongly commit- munications technologies are bringing ted to expanding digital opportunity for all, far-reaching change across the globe. Yet, including for farmers and agribusiness. And these advances have recently coincided with we can do more to help countries use digital a slowing of global poverty reduction, an technologies to improve their food system increase in global hunger, and persistent nat- outcomes in rural and urban areas. ural resource degradation. How can we better harness the impact of digital technologies to This report looks at the opportunities from tackle the world’s pressing sustainable devel- digital technologies, including better trans- opment challenges? parency of agricultural value chains, smarter farms, and improved public services. It also Improving the performance of the food reviews some of the risks, including an system is a central part of reaching the over-concentration of service provider Sustainable Development Goals. That’s market power, poor data governance, and because the food system provides nourish- exclusion. It presents a set of potential entry ment and supports livelihoods for most of points for public sector action to seize the the world’s poor and better management opportunities that expanding rural network of the system could strengthen the stew- coverage can bring while fostering digital ardship of the world’s natural resources. entrepreneurship and facilitating demand. Improvements in food system performance are associated with dramatic declines This report offers an opportunity to broaden in poverty, hunger and natural resource the discussion on ways to more rapidly spread degradation. digital dividends and drive policy and invest- ments that can deliver positive impacts for Over half of the world’s population now has the food system, people and the environment. access to the internet and over 5 billion peo- ple subscribe to mobile phone services. Most people without access live in rural areas. Where there is coverage, adoption is of- Laura Tuck ten low due to factors such as cost, lack of Vice President, Sustainable Development knowledge and skills, and mistrust. World Bank HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 3 Key messages • Digital technologies—tools that collect, store, analyze, and share information digitally, including mobile phones and the Internet—have significant potential to improve effi- ciency, equity, and environmental sustainability in the food system. They can: significantly reduce the costs of linking sellers and buyers; reduce inequalities in access to information, knowledge, technologies, and markets; help farmers make more precise decisions on re- source management by providing, processing, and analyzing an increasing amount of data faster; and potentially reduce scale economies in agriculture, thereby making small-scale producers more competitive. • A range of digital technologies in the food system are already leading to: better informed and engaged consumers and producers, smarter farms, and improved public services. These technologies range from simple off-line farmer advisory digital videos to complex systems requiring higher levels of mobile phone and Internet connectivity, such as distributed ledger technologies for value chain traceability and some forms of precision agriculture. • Adoption of digital technologies varies significantly across countries, with lower current adoption rates in low-income countries. Increasing adoption of digital technologies in the food system will require addressing supply-side factors—such as rural network cover- age and availability of digital applications—and demand-side factors, including skills and knowledge, trust, affordability, and complementary investments. Addressing these factors will require a range of public policy actions including increasing the space for private sector activity, improving the policy and regulatory environment, and using public invest- ments to help crowd-in private sector investment. • While digital technologies have significant potential they also pose several risks, includ- ing: an overconcentration of service provider market power; lack of data privacy; exclusion and potential job losses for some activities; and cybersecurity breaches. These risks cut across all segments of the economy, including the food system. Addressing these risks calls for public policy to: keep service provider entry barriers low; ensure good data governance; foster inclusion through targeted support to smallholder farmers, youth, women, and other vulnerable groups; and support skills development. • While digital technologies offer significant opportunities to improve food system out- comes, they should not be viewed as a panacea. Other investments are needed to address the multiple constraints farmers face such as roads, energy, post-harvest storage, and lo- gistics that can better link farmers to markets. Countries can also make policy improve- ments to increase incentives for farmers and agribusinesses to invest in farms and across agricultural value chains. These investments and policy improvements may in turn increase demand for digital technologies. 4 FUTURE OF FOOD Why harnessing digital technologies to improve food system outcomes matters The food system needs to play an active role generation from rural activities on farms in accelerating progress on ending pov- and along agricultural value chains. erty and hunger, contributing to growth Lifting this population segment out of and jobs, and better stewardship of the extreme poverty by 2030 will require world’s natural resources if the Sustainable average income gains of at least 60 per- Development Goals (SDGs) are to be cent in Sub-Saharan Africa, and at least achieved. 30 percent in Asia. • Ending poverty: The pace of global • Ending hunger, and improving nu- poverty reduction slowed to 0.6 percent trition and human capital: Between between 2013 and 2015, well below the 2015 and 2017, the number of people 25-year average decline of 1 percent per affected by hunger rose by 36.4 million, year.1 This slowdown puts at risk achieve- to 821 million.3 In addition, more than ment of the SDG target to end poverty 2 billion people are deficient in key vita- by 2030. In 2015, 736 million people, mins and minerals necessary for growth, 10 percent of the world’s population, development, and disease prevention.4 were living on less than US$1.90 per day, Energy and micronutrient deficiency of which 79 percent lived in rural areas contribute to the 151 million children and most relied on agriculture for their under age five who are stunted and can- livelihoods.2 While some of the poor will not grow to achieve their full potential. At migrate to urban areas by 2030, most will the same time more than 2 billion adults not, and will continue to rely on income are overweight and obese,5 increasing the HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 5 risk of non-communicable diseases such pollution is on the rise.12 Under current as Type 2 diabetes, hypertension, heart practices, greenhouse gas (GHG) emis- attacks, and certain cancers. Furthermore, sions from agriculture and land use one person in 10 is impacted by contam- change are projected to represent 70 per- inated food.6 cent of total allowable emissions from all economic sectors to limit global warming • Contributing to growth, providing to 1.5°C by 2050.13 If climate change goes more and better jobs, and boosting unmitigated, the total days under drought shared prosperity: As countries become conditions are projected to increase by richer, per capita expenditures on food more than 20 percent in some regions.14 tend to rise, even as their share in house- Further, food that is harvested but then hold expenditures declines. With rising lost or wasted occupies total arable land incomes, diets change and the share of in- equal in size to China, consumes about come spent on cereals declines relative to 25 percent of all water used in agriculture other fresh, processed, and convenience each year, and accounts for about 8 per- foods. This change leads to increased food cent of global GHG emissions. management and transformation beyond the farm, which have a multiplier effect in creating new enterprise growth op- Characteristics of the food portunities and jobs in the broader food system make it ripe for system. Farming (or agriculture) still em- expanded use of digital ploys more people than any other sector in many countries, accounting for an esti- solutions mated 67 percent of total employment in The food system has several characteristics low-income countries.7 The food system that offer significant potential for digital also accounts for a large share of manu- solutions to have positive impacts: facturing and services jobs. For example, in Malawi and Tanzania, food and bever- • Large and complex, with many actors: ages account for more than 40 percent of The global food system is complex, in- total manufacturing employment.8 Even volving many actors exchanging vast in the European Union (EU), the food amounts of information. It comprises and beverage industry represents a larger about 100,000 enterprises supplying in- share of employment than other manu- puts such as seeds, fertilizers, machinery, facturing sectors.9 animal health services, crop and livestock • Better stewardship of the world’s natu- insurance, finance, and livestock feed ral resources: Global agricultural produc- to 570 million farms around the world, tion practices are currently unsustainable. many in remote areas. And millions of The annual cost of land degradation is informal and formal enterprises move, about US$300 billion per year,10 about process, and sell agricultural outputs to one-third of the world’s largest aquifers 7.5 billion consumers. Digital technol- are being depleted,11 and agricultural ogies have the potential to significantly 6 FUTURE OF FOOD reduce the costs of linking these actors in of the world’s population).15 Sub-Saharan the food system, more efficiently match- Africa was the only region with mobile sub- ing buyers and sellers; and helping to scriber and mobile Internet penetration rates better target poor and vulnerable farmers significantly lower than the global aggre- with support services. gate, with rates of 44 percent and 21 percent, respectively.16 While there has been a sig- • Large inefficiencies: The food system’s nificant expansion of network coverage inefficiencies are reflected in its un- beyond urban areas, the vast majority of the sustainable use of land and water, and 1.2 billion people worldwide not covered by significant food loss and waste. Food a broadband-capable network live in rural production, as a biological process, re- areas.17 Advances in analytics are transform- quires many decision points that affect ing enormous amounts of digital data into production efficiency, such as what and useable form for decision makers. Over the when to plant; and what, how much, past 10 years, agriculture has experienced re- when, and where to apply fertilizer and markable growth in ag-tech18 investments, water optimally, which depend on other with US$6.7 billion invested in the past five factors such as soil type, soil moisture, years,19 most of which has been in developed and weather forecasts. In this data-inten- countries. However, agriculture is currently sive process, farmer decision making on the slowest sector in terms of adopting dig- resource management can be improved ital technologies, according to McKinsey’s by processing and analyzing more data Industry Digitization Index.20 faster, particularly considering the impact of climate change. Digital technologies have the potential • Asymmetric access to technologies, to improve efficiency, equity, nutrition knowledge, information, and markets and health, and sustainability of the food lead to lack of equal opportunity, and con- system. In terms of efficiency, digital tech- tribute to poverty and hunger, inequality, nologies can: (i) improve the use of capital, and unstainable use of natural resources. including machinery and equipment, in the Digital technologies can potentially re- food system, thereby increasing its technical duce inequalities in access to information, and allocative efficiency; (ii) facilitate the ac- knowledge, technologies, and markets. quisition of skills and knowledge needed for agricultural production, thereby improving Digital technologies have been spreading labor efficiency and the optimal use of in- rapidly across various sectors and regions, puts; (iii) improve farmers’ decision making driven by lower costs, better connectivity, through accurate, timely, and location-spe- and advanced analytics. Network coverage cific price, weather, and agronomic data and and mobile phone use has increased sig- information that will become increasingly nificantly in recent years. As of 2017, there important in the context of climate change; were 5 billion unique mobile subscribers and (iv) reduce costs associated with match- (66 percent of the world’s population), and ing producers and consumers, which will 3.3 billion mobile Internet users (43 percent help expand output markets and improve HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 7 producer access to inputs. Improved produc- concentration, are addressed in more detail tion decision and production efficiency can in subsequent sections of the paper. improve farmer profits. In terms of improving equity, digital technologies have the poten- • Inequality of access: Digital technologies tial to address unequal access to information, have the potential to reduce inequalities knowledge, technologies, and markets, and in the food system but may also poten- thereby improve relative incomes of poor tially widen inequality. Uneven connec- people. In terms of improving nutrition and tivity and broadband infrastructure lead health, in one recent study, the use of mo- to unequal opportunity. If digital tech- bile phones was associated with increases in nologies require more specific skills, the household income, gender equality, and food benefits may accrue disproportionately to and nutrition security. Positive nutrition ef- those farmers that are positioned to take fects occurred primarily through income and advantage of such opportunities.22,23,24 If gender equality pathways.21 In addition, dig- digital technologies require substantial itally-enabled improved traceability can help investments, small-scale farmers may be increase food safety, with positive health ef- excluded from their adoption and may see fects. As regards environmental sustainability their relative competitiveness decrease. of the food system, digital technologies can Costs of entry for some value chains may improve use of natural capital such as wa- increase. For example, if retailers require ter and land, and improve the use of inputs, suppliers of fresh produce such as leafy such as fertilizers. For example, remote sens- greens to apply real-time traceability us- ing technologies can measure water use and ing blockchain technology to increase monitor net withdrawal of groundwater, response times in case of food-borne ill- which can help determine sustainable use tar- ness, supplier entry costs will rise, and gets for better irrigation water management. those unable to meet those requirements will be excluded. The gender divide may While digital technologies have many po- also widen due to men’s generally stronger tential advantages, they also present risks socio-cultural access to new technologies that will need to be monitored and ad- than women. Even in richer countries, dressed to ensure broad benefits. The such as those in the EU, data show that World Bank’s World Development Report within countries the use of e-government 2016 on the “Digital Dividend” details sev- services is associated with higher educa- eral risks, including: inequality of access and tion, employment, urban residence, being affordability; concentration of market power male, and broadband access. Support to as regards e-commerce platforms, social net- small-scale farmers, particularly women, works, and search engines; issues related to through skills development and entry cost data privacy, ownership and consumer pro- can help reduce inequality risks. tection; and cybersecurity breaches. These risks apply also to the food system. Two • Data privacy: Digital technologies raise risks—inequality of access and data priva- questions about the ownership and use cy—are highlighted here, and with market of digital information acquired by these 8 FUTURE OF FOOD technologies. Providers of digital applica- benefits of digital technologies, especially tions and data storage platforms for actors in developing countries, and to address the in the food system, including farmers, are multiple constraints faced by farmers. These private enterprises that could potentially investments include: (i) complementary in- use these data in different ways, including frastructure such as roads, electricity, and making the data available to third parties. post-harvest storage to improve access to In addition, significant amounts of data markets and value addition; (ii) investment on farming practices are generated off- in skills and knowledge to improve farming farm through remote sensing via satellite practices in general, but also to improve dig- imagery. Rules governing data ownership ital literacy and enable farmers to better take and use are often inadequate, which rais- advantage of digital technologies; (iii) farmer es concerns about data misuse. There are organizations that more effectively link farm- ongoing efforts to develop workable ar- ers to markets; and (iv) access to financial rangements that will give farmers suffi- services to help finance agricultural inputs cient confidence about data privacy and and for savings mobilization, for which digi- which are not overly restrictive for service tal technologies can also help. Countries can providers.25 also make significant policy improvements to increase incentives for investment in food While digital technologies offer an oppor- systems that in turn may increase demand tunity to improve food system outcomes, for digital technologies. Digital technologies they are not a panacea. Complementary in- should be seen as a complement to, rather vestments are required to realize the potential than a substitute for, better policies. HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 9 What digital technologies are helping to improve food system outcomes? A range of digital technologies are current- by improving product traceability, and bet- ly facilitating improved communications, ter informing consumer choice as to price, operations, and transactions across the nutrition, production practices, and envi- food system, that are resulting in better in- ronmental and biodiversity impacts. With formed and engaged consumers with more increased transparency, consumers are transparent value chains, smarter farms, and increasingly influencing how food gets pro- improved public services. This section pro- duced. The range of digital technologies vides some illustrative examples rather than improving value chain transparency in- an exhaustive list of digital technologies in cludes distributed ledger technologies, smart the food system. contracts, food-sensing technologies, and e-platforms. Better informed and Distributed ledger technologies (DLTs) engaged consumers with and smart contracts have significant more transparent value potential to increase efficiency and trans- chains parency in agricultural supply chains by improving product traceability and integ- Consumers are increasingly demanding more rity, contract certainty, proof/verification of transparency in the products they buy, in- geographic origin, and compliance with san- cluding product ingredients and information itary and phytosanitary requirements. DLTs on sources and production methods. Digital and smart contracts also have potential to technologies enable increased transparency improve implementation and monitoring 10 FUTURE OF FOOD of World Trade Organization agreements identified with up to a 99.67 percent success and key provisions relating to agricultural rate contamination in red meat.30 FruitQC, trade. DLTs can ensure that gains from trade a Tellspec application, provides information accrue more directly to producers and con- on quality, ripeness, and flavor of fresh fruit sumers.26 Traceability can reduce food loss in less than 15 seconds.31 By providing infor- in food systems, with estimated gains of up mation on food quality, food sensing could to 30 million tons annually if DLTs were to reduce food waste by 5 to 7 percent.32 With be incorporated in half of the world’s sup- the cost of food-sensing technologies rapidly ply chains.27 For net environmental effects, declining, they could be increasingly used to emissions reductions from decreased food reduce expenses caused by food safety issues, production associated with reduced food which are particularly important in low- and loss would need to be assessed against po- middle-income economies where foodborne tential increases in emissions arising from illnesses cost US$110 billion each year.33 DLT applications’ energy usage. The use of Food-sensing technologies are becoming blockchain in the food system is still at an increasingly accessible through mobile de- early stage, but there have been a number vices for both consumers and supply-chain of successful pilot tests of its use. For exam- stakeholders, and could potentially provide ple, the French supermarket chain Carrefour inspection of a much larger share of com- uses blockchain to provide consumers with mercially available food.34 However, even detailed information on chicken purchas- for the most advanced food-sensing applica- es, including veterinary treatments and tions, the algorithms should be periodically freshness.28 To ensure their scalability and validated and recalibrated by traditional lab- accessibility, however, DLTs solutions re- oratory methods to improve accuracy and quire appropriate ecosystems. While some of precision. the elements of such ecosystems are technol- ogy-specific, they also largely entail enabling Digitally-enabled marketplaces for ag- policy, regulatory and institutional condi- ricultural products (e-platforms) can tions as well as basic requirements relating to potentially shorten agricultural val- infrastructure, literacy, and digital and net- ue chains, provide access to new markets, work coverage.29 reduce transaction costs, improve price transparency, and offer small and medium Food-sensing technologies are increas- enterprises (SMEs) new business opportu- ingly used to identify food quality, and nities. The power of e-platforms lies in the monitor food safety. Food-sensing tech- significantly reduced search cost of matching nologies are using cost-efficient and producers with consumers and lenders with non-invasive approaches such as spectrosco- borrowers, and their capacity to transfer and py and image analysis to provide information distribute risk. This process has the potential on biometric and biochemical features of to sharply reduce past market failures and food (such as size, shape, moisture, protein profoundly reshape value chains. In directly and fat content, and level of contamination). linking producers to consumers, e-platforms For example, hyperspectral spectroscopy has have the potential to shorten value chains HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 11 for some products. Lower search costs technologies include e-extension, precision help farmers bypass middlemen and im- agriculture (remote sensing), matching mar- prove agricultural market performance.35,36,37 kets for mechanization services and other According to the Department of Market inputs, improved access to market price System Development at China’s Ministry of information and finance, and improved cer- Commerce, the total online sales volume of tification for export markets. agricultural products in China reached over RMB 240 billion in 2017 (about US$36 bil- E-extension services can help raise farm lion), representing a 53 percent year-on-year profits and provide a cost-effective way to increase. As of 2017, in China there were reach a greater number of farmers. They can 9.85 million online shops operated by rural provide farmers with relevant and real-time farmers and employing over 280 million peo- access to information on how to resolve both ple. China’s e-commerce giant, Alibaba, has general and specific problems, ranging from launched several initiatives to become a one- sustainable farming practices, climate-smart stop-shop for SMEs conducting business solutions, and market access. Extension online, including online marketplaces, back- agents can use a combination of digital tech- end e-commerce merchant services, and a nologies to reduce the cost and increase the cloud-computing e-commerce platform. frequency of interaction with farmers.39,40 For Nevertheless, according to a recent survey in example, in Bangladesh, farmer-to-farmer rural China, in 2016 only 1.5 percent of rural videos on rice seed production produced for households sold agriculture products, mainly women led to lower production costs from vegetables and fruit, through the Internet.38 lower seeding rates, and a 15 percent increase Lack of knowledge and skills, and low trust in rice yields.41 Plantix, a mobile crop advi- in online transactions were the main reasons sory application for farmers, provides a quick for low adoption. Storage and transportation diagnosis of plant diseases, pests, and nutrient of fresh produce present another challenge deficiencies based on a photograph taken by to growing remote e-commerce sales. If a smartphone. Digital Green, which works in these aspects are addressed, there is potential South Asia, Latin America, and Sub-Saharan for significant growth. Africa, has produced and disseminated over 5,000 locally relevant videos in more than 50 languages, enabling farmers to share knowl- Smarter farms edge on agricultural production practices with one another42 that provides a relatively Digital technologies can help improve farm cost-effective way of helping increase adop- profits by improving productivity and resil- tion of improved production practices.43 ience through more efficient use of land and These videos are primarily screened off-line in water, improved access to capital, reduced communities that have limited electricity and product price dispersion across markets, Internet connectivity. While there are various and in some circumstances may lead to forms of e-extension, many are fairly recent, higher product prices and a higher share and evaluation is still ongoing as to which of retail prices for farmers. Relevant digital forms of agricultural e-extension services 12 FUTURE OF FOOD work best, and under what conditions.44 In productivity gains and cost savings, preci- general, for successful results, e-extension sion farming via satellite technology enables services need to provide timely, localized, governments to study how agricultural prac- and customized information that addresses tices affect the ecosystem, develop better specific farming concerns in a comprehensi- regulations,49 and enforce sustainable land ble format and appropriate language.45,46 At management practices, as in Uruguay.50 Even the same time, digital technologies are often though many high-tech precision tools are a complement to field advisory visits, rather more accessible for large-scale farms that can than a substitute. Used in combination with afford significant investments in technology, other more traditional forms of extension the situation is changing as access to tech- may make e-extension services even more nologies and their delivery become cheaper useful to farmers. and more affordable for smallholders. For ex- ample, the International Maize and Wheat Precision agriculture can improve the Improvement Center (CIMMYT) has been quantity and quality of agricultural out- testing variable rate fertilizer application kits put while reducing input usage (such as for smallholder farmers.51 water, energy, fertilizers, and pesticides), thereby generating climate benefits, while Matching markets can improve small- also increasing time efficiency by perform- holder farmers’ access to mechanization ing farming practices remotely. Reducing services and, in turn, reduce unit costs and the use of inputs such as fertilizers and pes- increase competitiveness. Physical capital, ticides offers positive environmental effects. such as farm machinery, is a key driver of on- Precision farming uses data received from farm productivity. However, the high upfront global positioning systems, satellite and aeri- cost of farm machinery often puts it out of al imagery, and sensors (for example, sensors reach of smallholder farmers. Digital plat- for soil conditions, ground water levels, and forms can create new markets for machinery precipitation detectors) to enable a range rentals by: (i) providing more affordable of precision agriculture applications. In the access to physical capital for smallhold- United States, precision agriculture tech- er farmers, and (ii) putting underused assets nologies were used on 30 to 50 percent of to work by matching suppliers of machin- corn and soybean acreage in 2010–2012. ery rental services with farmers wanting Impacts on farm profits were positive but these services, just as Uber matches driv- small, with adoption more likely on larger ers with customers for taxi services. As the farms.47 As precision agriculture technolo- marginal cost of matching buyers and sellers gies become easier to implement they could through digital platforms is extremely low, help improve incomes on smaller farms. The this service has the potential to reduce unit Nano Ganesh system in Pune, India, uses costs of machinery rental services through digital applications for irrigation systems to saved transaction and search costs. A study allow control of irrigation pumps remotely, in Zambia showed that hiring tractor ser- by mobile phone, which saves farmers water, vices leads to higher farm profit, by enabling energy, and time.48 In addition to potential farmers to expand their cultivated area and, HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 13 consequently, increase their income.52 Well- (ii) the specific content of the information known examples include: TroTro Tractor, has greater impact as mobile phone penetra- Ghana’s platform to connect tractor opera- tion and access to information increases.58 tors with farmers; Trringo, India’s foremost In these situations, content customized to tractor and farm equipment rental service; higher value commodities and varieties pro- and Hello Tractor, which matches tractor duced by farmers in specific locations had owners with smallholder farmers in need of higher impact.59 tractor services via simple text messages in Ghana, Kenya, and Nigeria. The digitization of transactions and pay- ments, and data analytics, can improve Under certain conditions, improved access access to finance by smallholder farmers. to market information via mobile phone Agricultural finance can facilitate farmer ac- can help increase farmers’ sales and prices, cess to improved inputs and technologies. and reduce price dispersions across mar- Yet, less than 10 percent of smallhold- kets. Increased access to information about er farmers worldwide have access to formal market prices via mobile phone can increase credit.60 Digital technologies improve access farmers’ bargaining power vis-a-vis traders to finance through a combination of lower- on farm gate product sales. This process, in ing operating costs and offering better ways turn, can help increase farmers’ shares of re- to assess weather, market, and credit risk. tail prices and farmer sales.53,54 For example, Financial institutions and agribusinesses can analysis shows that access via mobile phone use digital transaction records to process and short message service (SMS) to market price facilitate credit. Mobile phone-based dig- information on local crops in the central ital financial services have the potential to highlands of Peru increased farmers’ sales reach formerly unbanked populations as mo- prices by 13 to14 percent relative to coun- bile phone and Internet usage expands. In terparts without access to this information.55 addition, numerous studies show that dig- For sales to markets, market information ital financial solutions have a positive and via mobile phones can better inform farm- significant effect on annual household in- ers and traders where prices are highest. put use, agricultural commercialization, The latter market arbitrage can help reduce and household income.61,62 New FinTech price dispersions across markets,56,57 and re- solutions are developing rapidly. M-PESA duce relative oversupply to specific markets represents a prominent example—the thereby potentially lowering food loss and Kenya mobile phone-based payment system waste. Not all studies show that these effects was introduced in 2007 and, by 2009, had are significant. Impacts on farmer pric- reached 65 percent of Kenyan households. es from increased market information via Furthermore, it is estimated, that M-PESA mobile phones tend to be dependent on lo- has lifted 2 percent of rural Kenyan house- cal conditions, for example: (i) impacts tend holds out of poverty.63 While FinTech firms to be more significant, particularly for me- still play a negligible role in global com- dium-to-high value commodities, where mercial lending, digital technologies in the mobile phone penetration is lower; and global financial sector will gain growing 14 FUTURE OF FOOD importance over the coming years, including payment system, the government reached in the food system.64 For example, big data65 4.3 million smallholders at a cost of approx- and advanced analytics can significantly imately US$96 million.68 This system also reduce the cost of establishing creditwor- facilitated financial inclusion of farmers and thiness of farmers and assessing insurance expanded private sector opportunities.69 A risk. These lower costs can potentially low- recent review of pilot e-voucher programs er interest rates and insurance premiums for for subsidized farm inputs in Guinea, Mali, farmers thereby increasing access to these fi- and Niger highlighted several lessons for nancial services. Farmdrive in Kenya, and implementation of these types of programs. Harvesting in Uganda, are examples of com- These lessons include the following: the im- panies using data analytics to assess farmers’ portance of mobile network coverage; actual creditworthiness for financing. possession of mobile phones by intended beneficiaries, particularly women; promot- ing literacy and knowledge about how the Improved public services program works; and organization of procure- ment and agro-dealers to ensure availability Digital technologies can help improve public of farm inputs at the right time of the agri- services in the food system. Beyond e-exten- cultural season.70 sion services detailed in the previous section, such services include better targeted gov- Remote sensing and big data applications ernment-to-person payments, improved are being increasingly used for enhanced risk management, and monitoring and management and efficiency, and also to evaluation. mitigate risk. These applications include au- tomated early warning systems for crop or Digital technologies can be used to livestock health, related to weather, pests better target government-to-person pay- and diseases that can facilitate proactive and ments, improving distribution efficiency timely management responses. Advances and transparency. Through the introduc- in data management and machine learn- tion of a digital system in Estonia, the time ing71 now make it possible to integrate and spent applying for agricultural subsidies at analyze billions of data points (analog and the Agricultural Registers and Information digital, ground-based and remotely sensed) Board decreased from 300 minutes to at high-spatial and temporal resolution. 45 minutes.66 In Nigeria, the government For example, the World Bank Agriculture reached a higher number of recipients with Observatory is accessing ground-based hy- lower cost and lower leakage through an drometeorological data (very sparse in e-wallet program for subsidized fertilizers. most developing countries) and high-den- Prior to this program, in 2011, the govern- sity weather satellite radar platforms, and ment spent approximately US$180 million applying machine learning to generate a for 600,000–800,000 smallholders, most of continuous weather surface across the earth’s which never reached the intended benefi- croplands. The resulting weather data com- ciaries.67 In 2013, with the e-wallet digital prise 7 billion data points generated daily by HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 15 1.5 million virtual weather stations across digital survey tools have long been used to the world and updated four times a day, to lower the cost of project-level data collec- generate “real-time” weather data at intervals tion, more advanced technologies such as of nine kilometers across global agricultural remote sensing can bring additional efficien- land. This data surface represents a disrup- cies to M&E. For example, satellite earth tive approach to the traditional method of observation provides unbiased, consistent, relying solely on ground-based hydromete- and timely information on whether invest- orological stations that need to be procured, ments in agricultural development are taking installed, calibrated, and maintained, even place in a sustainable and effective manner. before the data are accessed, analyzed, and Satellite data can thereby promote better synthesized. By coupling the high resolution transparency, responsibility, and account- weather surface with cropping calendars, it ability to impact assessments of agricultural is possible to make real-time assessments investments on crop production and water of crop performance and to take proactive management. The World Bank support- management risk-mitigation interventions ed the Resilient Agriculture and Integrated and decision support across the entire value Water Resources Management Project in chain, comprising farmers, input suppliers, the Dominican Republic, which used drones logistics providers, markets, and policy mak- to monitor changes in the physical condition ers. Because early projections of crop yield of irrigation infrastructure in 1,200 hect- anomalies are possible at the sub-national, ares (ha) of rice fields. The pilot study was national, regional, and even global level, the designed to understand the value, includ- Agriculture Observatory platform can pro- ing cost-effectiveness and feasibility, of using vide early warning of potential food shocks drones for M&E in infrastructure develop- several months in advance of normal har- ment interventions relative to other methods vest periods. Information and analysis from such as satellite imagery, aerial photography, the Agriculture Observatory are being used and physical inspection. Drones provid- to guide the design and implementation of ed higher resolution, easy deployment, and agriculture projects. National Agriculture eliminated problems relating to cloud cov- Observatories are in operation or are being er interference that occur with satellite set up in Ethiopia, Kenya, Russia, Zambia, imagery and aerial photography. However, and Zimbabwe, in close collaboration with as drones cover relatively smaller areas as national agricultural and meteorological they can only make a finite number of flights agencies. per day, and have a higher regulatory bur- den, they are more cost-effective for smaller Digital technologies offer a broad spec- areas. Another tool for monitoring and eval- trum of tools and data to enhance uation is the Geo-Enabling Initiative for monitoring and evaluation of outcomes in Monitoring and Supervision (GEMS). agriculture. The uptake of digital tools and GEMS is a smart-phone-based system that consequent data generation in agriculture enables monitoring agents to enter data on can help enable more cost-effective monitor- implementation progress, including photos, ing and evaluation (M&E) of results. While and is automatically time stamped with GPS 16 FUTURE OF FOOD FIGURE 1: Continuum of digital technology opportunities (illustrative) High mobile coverage Mobile payments, including government-to-person payments Distributed ledger technologies for value chain traceability Food-sensing technologies to E-platforms monitor food quality and safety for agriculture E-Platform for mechanized products Precision agriculture services Market price information E-extension using video tools E-agriculture statistics Low mobile coverage Low Internet connectivity High Internet connectivity coordinates. The information can be entered connectivity. Mobile-phone-based market off-line and, when a mobile connection is price information, e-platforms for mecha- available, is uploaded to a central database. nized services (such as Hello Tractor), and GEMS is being increasingly used to mon- government-to-person payments require ac- itor implementation progress of projects cess to mobile networks but low Internet in fragile situations, such as in Democratic connectivity, while e-statistics require Republic of Congo and the Central African Internet connectivity but low mobile cov- Republic. erage. Additionally, some technologies are best suited to environments with both high The set of digital technologies across val- mobile coverage and high Internet connec- ue chains, farms, and public services tivity such as distributed ledger technologies highlighted in this section require differ- for value chain traceability. The continuum ent levels of mobile phone and Internet of digital technologies offers opportunities connectivity. For example, agricultur- across a range of mobile phone and Internet al extension and farmer-to-farmer learning connectivity (figure 1). Skills needs also vary provided by digital videos can be provided across this continuum, an aspect that is ad- off-line with no need for mobile or Internet dressed in the next section of the paper. HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 17 What public actions can facilitate broader adoption of digital technologies and harness their impacts on food system outcomes? The adoption of digital technologies in ag- coverage and number of mobile applications gregate is affected by both supply factors developed per person). As network coverage (network coverage and digital content), and increases, so do potential users of digital ap- demand factors (including skills/knowl- plications, which increases the incentives of edge). These factors have strong positive entrepreneurs to develop digital applications correlations with country level digital adop- or content. Knowledge/skills impact digi- tion (box 1). While data on these correlates tal adoption in two primary ways. First, the are not available for separate geographies or knowledge/skills of end users, such as farm- sectors, and represent aggregate countrywide ers, affect demand for digital technologies. data, a similar general pattern is likely for For example, farmers have to know that a the food system. However, adoption rates in certain technology exists, believe it will help agriculture have been lower than other sec- them (demonstration of effectiveness), and tors according to the McKinsey Industry learn how to use it. Second, the knowledge/ Digitization Index. This situation is likely skills of entrepreneurs that develop digital reflective of the lower mobile coverage and technologies impacts digital adoption (the Internet connectivity in rural areas where correlation coefficient of basic skills and mo- farmers and many agribusinesses operate, bile applications developed per person was and the lower levels of skills in rural areas. 0.89 using 2017 cross-country data). Digital The development of digital applications or technologies can be designed in a way that content, indicative of digital entrepreneur- requires low-level skills and literacy for its ship, is correlated to network coverage (0.84 use, for example with voice and touch screen correlation coefficient between network functionality. 18 FUTURE OF FOOD BOX 1: Digital adoption correlates Network coverage Digital content 1.0 1.0 Correlation coe cient = 0.85 Correlation coe cient = 0.90 Digital Adoption Index Digital Adoption Index 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0 20 40 60 80 100 0 20 40 60 80 100 Network coverage Number of mobile apps developed per (normalized score 0–100) person (normalized score 0–100) Basic skill 1.0 Correlation coe cient = 0.87 Digital Adoption Index 0.8 0.6 0.4 0.2 0.0 0 20 40 60 80 100 Basic skills (normalized score 0–100) Sources: World Bank and GSMA 2017 country level data. Even with no or low mobile phone cover- broader adoption of digital technologies age and Internet connectivity, there are and harness their impact on food sys- still opportunities for digital technologies. tem outcomes. MFD looks for ways to Digital technologies are advancing rapid- crowd-in private resources to help achieve ly and there are now off-line technologies development goals, while optimizing the that can help poor and even illiterate farm- use of scarce public resources. As digital ers improve production practices with no or technologies are primarily generated by the limited mobile phone coverage and Internet private sector, and as the farmers and agri- connectivity. businesses adopting these technologies are also private actors, MFD can help identify The Maximizing Finance for Development entry points for public sector action to facili- (MFD) framework can help identi- tate broader adoption of digital technologies fy public actions needed to facilitate and harness their impacts on food system HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 19 outcomes.72 A sequence of questions about Expand rural network rural network coverage, digital applications and content, and private end-user adop- coverage tion can help identify appropriate public actions. These questions include: (i) Is the Rural network coverage is often relatively private sector investing in digital technolo- low but varies significantly across countries gies? (ii) If not, is it because there is limited with similar income levels: While there are space for private sector activity as a result off-line digital technologies, as discussed in of restrictions to competition, public sector the previous section, network coverage in- dominance, or private sector monopolies? creases the opportunity set of digital mobile (iii) If not, is it because of policy and reg- technologies for people residing in rural ar- ulatory weaknesses or gaps? (iv) If not, can eas. In addition, the type of network available public investment help crowd-in private in- influences the type of digital application that vestment? (table 1). While these questions can be used. For example, second genera- proceed in sequence, the intention is not tion (2G) networks are more suited for voice to stop if the answer to a question is affir- and text messaging, while third generation mative, but to answer all the questions and (3G) and fourth generation (4G) networks ensure a complete assessment and identifi- allow for broader applications and use of cation of public policy entry points. Once smart phones. While, on average, more than these have been identified, implementation 90 percent of the population in high income of public policy and regulatory changes, and countries have network coverage, only about public investments need not be sequential. 50 percent of the population in low-in- Indeed, they are likely to be simultaneous come countries do, on average. For example, or multi-staged. Cutting across these as- about three-quarters of the population in pects are considerations for public policies Chad have no network coverage, whereas in to address efficiency, equity, and environ- Mozambique and the Democratic Republic mental sustainability aspects, as well as of Congo, about 60 percent have no cov- addressing the risks of digital technologies erage.73 While more than 60 percent of highlighted in the earlier section, such as Burkina Faso’s population have 2G coverage, data privacy, inequality, and concentration only 25 percent have 3G coverage or faster.74 of market power. Table 1 provides a brief The extent of network coverage across coun- assessment of the first question, and if the tries, among other factors, is likely affected by answer to one of the other questions is yes, GDP per capita levels, country size, popula- it provides potential entry points for pub- tion distribution, and policies.75 All else held lic sector actions. Factors included in table equal, countries that are richer, smaller, and 1 are expanded on in the subsequent text. with more concentrated populations tend to The intent is not to provide an exhaustive have mobile network coverage rolled out to a treatment of each factor, but rather provide larger share of the population.76 By contrast, a range of potential entry points for action, countries with more dispersed populations, and in the process, refer to ongoing and rel- as reflected in a relatively higher share of the evant World Bank work. population living in rural areas, tend to have 20 FUTURE OF FOOD TABLE 1: Potential entry points for public-sector actions to facilitate broader adoption of digital technologies and harness their impacts on food system outcomes Supply-side factors Demand-side factors Expand rural network Develop relevant digital Facilitate demand for coverage applications digital technologies in the MFD cascade (expand digital (foster digital food system (particularly of questions: infrastructure) entrepreneurship) smallholder farmers) Is the private ¾¾ >90% coverage in high ¾¾ Increase in ag-tech ¾¾ Higher farmer adoption of sector doing it? income countries investment over last 10 digital technologies in high- ¾¾ 50% coverage in low- years; growth in ag-tech income countries relative to income countries (even start-ups in Africa over the low-income countries; and lower for 3G or faster past two years higher adoption rates on networks) ¾¾ Significant variation across larger farms countries If not, then is ¾¾ Foster competition ¾¾ Lower entry costs to it because of among telecoms facilitate competition limited space (competition is associated among digital platforms for private with more extensive rural sector activity? coverage) If not, then To help lower the cost of ¾¾ Improve the enabling ¾¾ Improve farmers’ incentives is it because providing rural coverage: environment for business to invest of policy and development ¾¾ Develop data governance ¾¾ Adopt a spectrum policy regulatory ¾¾ Design digital regulations arrangements that build users’ that boosts connectivity gaps and around functionality confidence and trust in digital ¾¾ Lower infrastructure weaknesses? ¾¾ Clarify data ownership technologies taxes/duties ¾¾ Develop governance ¾¾ Allow infrastructure arrangements for open sharing data ¾¾ Ensure consistency/ ¾¾ Invest in open data streamline local level that have public good regulations characteristics ¾¾ Reduce policy/regulatory uncertainty If not, then ¾¾ Invest in complementary ¾¾ Support skills ¾¾ Support skills development can public infrastructure development for vulnerable groups investment help ¾¾ As a last resort, subsidize ¾¾ Improve access to ¾¾ Develop relevant, customized crowd-in private service providers to offset finance for start-up and tools investment? higher costs of rolling out early maturity ag-tech ¾¾ Reduce costs of technology rural coverage enterprises adoption ¾¾ Support increased use of ¾¾ Improve access to finance digital payments ¾¾ Invest in complementary infrastructure ¾¾ Support development of digital farmer identification lower network coverage (figure 2). While have managed to achieve relatively high lev- coverage in rural areas tends to be lower than els of rural coverage. For example, more than in urban areas, some low-income countries 90 percent of Rwanda’s population have 3G HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 21 FIGURE 2: Network Coverage policy alone is unlikely to ensure parity in 100 urban and rural costs and coverage.80 90 80 Policy and regulatory entry points to low- Digital Adoption Index 70 er costs: At lower population densities, unit 60 50 costs of telecom service providers per user 40 are higher. In addition, rural areas often 30 have higher installation and maintenance 20 costs, greater distances from main roads, 10 0 more uneven terrain,81 and lack of electric- 0 20 40 60 80 100 ity. This situation results in higher prices Percent of population in rural areas in rural areas, or a lack of services if de- Sources: World Bank and GSMA country-level data. mand is insufficient to cover costs. Potential public policy entry points to expand rural coverage and 85 percent of the population in coverage in developing countries include: Kenya have 3G coverage. The rural popula- (i) ensuring sufficient release of spectrum tion comprises 83 percent and 73 percent of (frequencies allocated for communications the total population in Rwanda and Kenya over the airwaves) and designing associat- respectively, suggesting high rural coverage.77 ed spectrum policy with the aim of boosting Carefully designed policies and regulations rural connectivity;82 (ii) lowering infrastruc- can promote greater expansion of networks ture taxes and/or import duties to lower to rural areas. costs; (iii) making consideration for infra- structure sharing, including mast or tower Expand the space for private sector activi- sharing, as a way to lower costs, and allow- ty: Evidence from more than 200 countries ing customers to roam between networks; over 15 years shows that network competi- and (iv) ensuring consistency and stream- tion has helped expand network coverage.78 lining of regulations at the local level on site Taking into account other factors such as deployments.83 GDP per capita; and population size, den- sity, and distribution, analysis shows that Reduce policy uncertainty: Reducing population coverage was, on average, 12 macroeconomic and political stability, and en- percentage points higher in countries with suring more reliable policies reduces private network competition compared to coun- sector investment risks. In addition, regula- tries served by a single network provider; tory systems that provide greater certainty population coverage for 3G was, on average, for investment serve as a necessary condition 36 percentage points higher, and popula- for the expansion of digital infrastructure in tion coverage increased three times as fast rural areas. According to the World Bank’s in countries with network competition.79 Enabling the Business of Agriculture report,84 While increased competition in the telecoms countries with higher quality ICT regu- sector is associated with increased rural cov- lations85 tend to also perform well on the erage in low-income countries, competition GSMA Mobile Connectivity Index. In 22 FUTURE OF FOOD addition to high transaction costs, arbitrary Development of digital applications is in- regulatory changes contribute to high prices fluenced by a broad set of country condi- for end-users in rural areas.86 tions. These conditions include supply-side factors, such as: digital infrastructure, includ- Use public investment to help crowd-in ing network coverage; other infrastructure private investment: Complementary invest- such as roads and electricity; the enabling ments such as rural electrification, including environment for entrepreneurship; skills de- local solar energy sources, can address power velopment; and access to financing. Also im- supply constraints in rural areas. Countries portant are demand-side factors, including such as the United States have also provided support for smallholder farmers and agri- subsidies to services providers to offset the businesses to adopt digital technologies. A higher cost to extend broadband coverage to vibrant innovation ecosystem also requires rural areas.87,88 collaboration among companies, inves- tors, governments, and development part- ners. Understanding the binding constraints Foster digital among these sets of conditions can help in- entrepreneurship form what the public sector can do to improve the innovation ecosystem for ag-tech entre- Digital entrepreneurship varies signifi- preneurs. While some of these aspects are cantly across countries, with recent growth relevant for all entrepreneurship in the food in ag-tech startups. There is a positive cor- system, others are specific to digital develop- relation between the number of digital ment. For example, there are ongoing ques- applications developed per person and net- tions about the risks associated with market work coverage,89 and this correlation is likely concentration in digital platforms, data pri- indicative of the importance of an expanding vacy, and consumer protection. Digital regis- potential client base for developers. However, tries are also one solution for better targeting there is a fairly large variation across coun- services, and better informing development of tries in the number of digital applications digital applications. Per the MFD framework developed per person, per level of network in table 1, this section discusses the space for coverage. Part of this difference may be due private sector activity, including policies and to differing enabling environments for digi- regulations, as well as public investments to tal entrepreneurship in these countries. There crowd-in private investment in digital appli- has been significant growth in ag-tech invest- cation and content with a particular focus on ments in the past 10 years,90 and in the past addressing such risks as market concentration two years, significant growth in ag-tech start- and data governance. ups in Africa.91 This section highlights some public sector entry points to foster digital en- Lower entry costs for private sector com- trepreneurship and investment in the food petition in digital platforms. There is a risk system, and to address some of the risks. The that digital technologies can increase the World Bank has ongoing work on this topic concentration of market power. Prominent focusing on the Africa region. examples include Facebook among social HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 23 network providers, Amazon among e-com- operators. The cost of money transfers sub- merce marketplaces, and Google among sequently declined.93 search engines. There has been much debate on this topic which is beyond the scope of Improve the enabling environment for this paper to try to resolve. Nevertheless, a business development. Macroeconomic sta- key goal is to keep market entry costs low bility and peace are key conditions for private for new entrants and switching costs low enterprise development. In addition, an en- for consumers.92 Low entry costs allow for vironment characterized by unclear property dynamic development and competition, rights, frequent policy change and reversal, so that new players with superior business uncertain contract enforcement, and high cor- models and innovation can enter a specif- ruption translates into lower investment and ic market and disrupt the existing market growth.94 Supportive policies such as making leader. However, developers of new digi- it easy to start a business, business-friendly tal solutions often face high up-front costs tax policy, and strong patent protection can associated with software, data storage, ana- also help facilitate entrepreneurship. lytics, and security, which deter competition. In addition, network effects and switching Design digital regulations around func- costs can create barriers to scale for new en- tionality and contestability. The growing trants; the more users connected via one complexity of digital ecosystems underpin- network (such as for buyer-producer plat- ning digitization of food systems, as well forms within a value chain), the smaller the as an exponential increase in the volume incentive for those users to switch to anoth- and speed with which data are collect- er network. These aspects make regulation a ed and analyzed add to the challenges of significant challenge. In some cases, regula- designing appropriate policy and regulato- tions have been expanded to curb excessive ry systems. Growing innovation and rapid market concentration, as was the case for market changes make regulations both more M-Pesa, a highly successful money trans- complex and more prone to quickly be- fer system rolled out in Kenya by Safaricom, come obsolete. On the other hand, existing a mobile services provider. Initially, banking policy and regulatory frameworks are of- regulators’ hands-off approach contribut- ten unsuited to addressing concerns, such ed to the rapid growth of M-Pesa. Through as data privacy and ownership issues, arising exclusivity arrangements, M-Pesa agents from rapid digitization. All this results in a could only offer products and services with- higher likelihood of regulatory uncertainty, in the M-Pesa network, which locked agents higher compliance costs, and lower levels of and most users into a single network. Given technology adoption. According to a recent Safaricom’s dominant market share, the ex- PwC survey,95 regulatory uncertainty around clusivity contracts posed an entry barrier blockchain-based solutions was identified for other telecom operators, and the Kenya as a major scale-up challenge across various Competition Authority expunged the ex- sectors. For the new regulatory frameworks clusivity agreements in 2014, permitting to address these challenges, regulations and M-Pesa agents to work with other mobile regulatory institutions should be re-designed 24 FUTURE OF FOOD around the principles of contestability—that different approaches to address data own- is, to address risks associated with market ership issues. For example, U.S. and New concentration—and functionality focus- Zealand agricultural sectors have been using ing on the ultimate policy objectives (for voluntary industry standards to establish an example, cost effectiveness or privacy pro- understanding between farmers and service tection), rather than on the technologies. providers that use farm data on data owner- Regulations should be dynamic and focus ship.97 In 2018, an EU coalition of agri-food on principles that allow ex-post enforcement associations introduced a joint EU Code of of broad rules rather than detailed ex-ante Conduct on agricultural data sharing.98 prescriptions. Lastly, new regulatory frame- works should recognize that many current Develop governance arrangements for regulations may be outdated, and the new open data. Policy and regulatory frameworks digital economy may require a clean slate ap- guiding aspects of data privacy and owner- proach to re-evaluate existing and/or new ship also serve as major factors in enabling regulations.96 access and use of open data in agriculture, this includes government-to-private and Clarify data ownership. The economic value private-to-private data sharing. Advantages of data in food systems increases as they are of open data are numerous—they can ad- aggregated, creating important policy and vance innovation, serve as a platform for regulatory issues related to individual versus entrepreneurship, and increase transparen- aggregate data ownership and use rights that cy and accountability in the food system. are important for both developers of digital To reap such benefits, however, certain chal- applications and farmers. Furthermore, de- lenges must be addressed,99 including both fining data ownership and data use rights technical and legal. To facilitate the use of becomes increasingly complicated as the open data, standards need to be designed to number of data users increases. In the event ensure privacy of individual data through ef- that data are generated using equipment fective anonymization while also ensuring its owned by a farmer, defining data ownership interoperability. Also, a certain level of infra- rights is a straightforward task. However, structure needs to be in place to enable the this is not the case in a situation in which scaling up of open data initiatives. Properly these data are aggregated and transmitted to designed data governance frameworks serve cloud storage that can subsequently transfer an integral role in enabling open data avail- them to a third party. Across the developed ability and use. While there is no single world, legal and regulatory frameworks governance solution for open data in agricul- around agricultural data ownership remain ture, Global Open Data for Agriculture and piecemeal and ad hoc. For example, in the Nutrition identifies four possible governance United States and Canada, existing law does strategies, namely: i) inter-institutional co- not recognize agricultural data as physical operation to build consensus about data or intellectual property. As such, data own- ownership; ii) model frameworks adopted at ership is not clearly defined. Nevertheless, the local, national, and regional levels; iii) so- countries have been experimenting with cial certification schemes that leverage the HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 25 power of ethical consumption, and iv) reach- Twiga Foods in Kenya, a five-year-old com- ing an international agreement on ownership pany that uses a technology platform to of open data. improve the supply chain from farmers to markets, has effectively benefited from men- Invest in open data that have public good torship programs (Google Launchpad and characteristics. Invest in agricultural statis- GSMA Ecosystem). The Africa Agriculture tics and agroclimatic data that can be used by Incubators Network (AAIN) is a network of all, including creative entrepreneurs. Making companies that supports agriculture start- these data available would reduce the cost of ups via technical support and mentorship, development and scaling up of novel dig- including help in preparing business propos- ital applications, as developers would not als to pitch to investors. Entrepreneurship have to devote resources to data collection. programs that combine interventions (men- Public investment in agricultural statistics toring or coaching, finance, and access to and agroclimatic data is likely an under-ap- markets), addressing the multiple constraints preciated driver of the development of that entrepreneurs face, tend to be more ef- new digital applications to better meet the fective than single-intervention programs.102 needs of smallholders. It is an open question While a recent global analysis of youth-tar- whether there is a need for direct subsidies geted interventions found that only about to entrepreneurs to develop applications for one-third showed a significant positive im- low-income farms and agri-SMEs, as such pact on employment or earnings, in low- and activities will become more profitable once middle-income countries such programs low-income owners of farms and firms are have been more successful, and skills training digitally connected, and entrepreneurs are and entrepreneurship programs seem to have able to develop more relevant applications had a higher impact.103 Involving the private with increased availability of information.100 sector in program delivery is associated with improved impacts. Support skills development. The number of digital applications developed per person is Improve access to finance for start-up highly correlated with basic skills (in 2017, and early maturity AgTech enterprises. the cross-country correlation coefficient Access to finance is an important ingredient was 0.89).101 Skills development can fos- in successful start-up and scaling of digital ter digital entrepreneurship. Incorporating technology enterprises. Early maturity enter- more entrepreneurial and digital technol- prises need to rely mainly on financing from ogy content in the curricula of agricultural family or friends, venture capital, or com- universities and training institutes, togeth- mercial finance blended with concessional er with associated teaching staff, could help funding provided by a development partner. develop skill sets needed to foster digital de- Venture capital investments in digital agri- velopment in the food system. Mentoring culture have increased in the past five years, and ongoing business advisory programs for with significant investments in precision ag- enterprise development tend to be more ef- riculture, agriculture marketplaces, FinTech, fective than a one-off training. For example, AgTech imagery, and indoor agriculture.104 26 FUTURE OF FOOD While much of the focus is on developed erode confidence and trust among users. countries, particularly the United States, Opening a bank or mobile money account venture investment in developing coun- usually requires some form of official govern- tries has been increasing. The International ment identification, such as an identification Finance Corporation (IFC) of the World card, which many people in rural areas lack. Bank has an important role to play in this Digital identification can help bridge this space. IFC has a significant venture port- gap. Digital payment systems are only suc- folio that it invests on commercial terms in cessful if there is sufficient interest and trust emerging markets, including food systems. from both entrepreneur and customers to use Examples include Chaldal in Bangladesh, this form of payment. An appropriate con- an online grocery delivery service, and Big sumer protection framework, robust digital Basket in India, which has added groceries networks, and banking and telecoms policies to its portfolio of products and is sourcing that support digital financial services are all from farmers.  IFC also invests indirect- important components of a functioning dig- ly through funds into companies, such as ital payment system.105 In addition, targeted Agrofy in Argentina, which provides an on- efforts are needed to support inclusion of the line agriculture marketplace, and Agrostar poor and elderly in these payments systems in India, which is an online marketplace for to prevent widening inequality.106 When agricultural inputs. Leveraging blended fi- digital applications require digital payments, nance, IFC has also directly invested in early the unenrolled get locked out. stage AgTech companies such as Kenya’s Twiga Foods. Blended finance could po- Support development of digital identifi- tentially help provide financial support to cation for farmers. Digital identification high-development impact investments that provides a critical platform for digital ag- would otherwise not easily attract financing riculture service providers to scale their on strictly commercial terms because risks businesses because startups and digital ag- are high and the returns are either unproven ricultural solution providers spend about or not commensurate with the level of risk. 50 percent of their initial business develop- ment efforts profiling and identifying target Support increased use of digital payments, farmers. In addition, digital identification, particularly for the poor and vulnerable especially when linked to land and livestock groups. Digital payments can be an im- assets, is a powerful bridge for farmers to ac- portant element of entrepreneur success. cess financial services, to reduce fraud and For example, digital payments through improve efficiency in the delivery of goods e-commerce platforms can help broaden an and services, and for governments to bet- entrepreneur’s client base and enable farm- ter target agricultural support. For example, ers to more easily sell or purchase products Estonia put 99 percent of its public services on these platforms. Expanding network online and enabled online identity verifica- coverage can help increase scope for digital tion and authentication via a platform that payments. Poor network quality and cover- connects various registries to the country’s age can lead to transaction failures that can foundational digital ID system. As a result, HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 27 Estonia’s farmers spend 45 minutes, on av- of government-to-farmer payments. Where erage, on support applications, down from there is demand for better outcomes that dig- 300 minutes using the previous paper-based ital technologies can help deliver, then several system. They also receive their payments factors can boost demand for digital technol- more rapidly as transfers are directly linked ogies, including: support for knowledge and to their bank accounts. Furthermore, farm- skills development; ensuring development of ers can register land and cattle online and relevant, customized digital tools in a suit- access detailed geographic and soil-related able format and relevant languages; reducing information through the platform. Another the cost of adoption and facilitating access to example is Uruguay, which has become a finance if needed for adoption; and building leading exporter of meat since implementing trust in digital applications. Complementary a livestock traceability system. The system infrastructure investments such as roads, en- setup was initially motivated by an out- ergy, and post-harvest storage can help link break of foot-and-mouth disease in 2000 farmers to markets that would make several and 2001, which led to the overhaul of a digital technologies more relevant to them. paper-based system that had been in place for 30 years. Each head of cattle is associat- Support farmer knowledge and skills, par- ed with an individual farmer, based on his or ticularly those of poor farmers, women and her national ID. other vulnerable groups. There is a high cor- relation between digital adoption and basic skills. Investments to increase digital literacy Facilitate demand for and knowledge can help farmers take ad- digital technologies in the vantage of digital technologies.107 Targeting food system support to poor farmers, especially wom- en, can help reduce disparities in adoption. In the absence of demand for digital tech- Larger scale, better educated, and wealthier nologies, actions to facilitate supply are farmers are better positioned to take advan- unlikely to increase adoption. Digital tech- tage of digital technologies.108 In addition, nologies are not an end in themselves, but in low- and middle-income countries few- a means to help improve food system out- er women than men own a mobile phone or comes. If food consumers are not demanding use the Internet.109 Lack of digital literacy more transparency, food distributors will contributes to the gender gap in ICT us- have little incentive to adopt digital technol- age.110,111 Support to farmers that are at risk ogies to ensure traceability. If farmers retain of exclusion from digital technologies could their production for subsistence consump- be provided through extension and adviso- tion, they will likely have little use for mobile ry services, with various forms of learning phone-based systems that facilitate price having different impacts on rural wom- discovery. If governments do not promote en.112 Agricultural extension systems can policies that involve payments to farmers, also take advantage of digital technologies they will likely have little interest in tech- to disseminate information and knowledge nologies that allow more accurate targeting as discussed earlier in this paper. In addition, 28 FUTURE OF FOOD public-private partnerships can help farmers Reduce the costs of digital technology gain a presence on e-commerce platforms. adoption. The fees that farmers pay for digital For example, technical support was provid- services seem to influence adoption rates.117 ed to melon farmers in the Xinjiang Region The public good nature of a particular digi- in China to improve the quality of their pro- tal service needs to be reflected in its price for duce, support online promotion, and manage farmers, particularly for women farmers. For logistics shipments to clients. example, if the objective of mobile-phone- based market information programs for poor Develop relevant customized tools in a rural areas is to improve equity, shouldn’t this suitable format and relevant languages. information be provided for free as a pub- For example, in Ghana, the Talking Book, a lic service cost to government? These types of behavior change initiative, experienced high public subsidies should not displace private adoption rates by providing 140 hours of au- sector activity, but in areas of high poverty dio content about agriculture and other rural with low ability to pay, displacement may not issues in local languages to illiterate people be a concern. Another policy that can help via a low-cost audio computer. From 2008 to reduce the cost of digital services is to low- 2015, the number of people who used Talking er the cost of expanding network coverage to Books increased from 1,000 to 175,000 in rural areas, as noted earlier in the paper. Ghana alone. In 2012, the average harvest of a farmer that had used a Talking Book in- Improve access to financial services. creased by 36 percent after one year.113 In Finance for agriculture through commercial Rwanda, an e-wallet initiative for farmers banks remains limited; for example, about experienced low uptake because SMS mes- 1 percent of commercial credit in Africa goes sages were sent in English rather than in to agriculture. Microfinance institutions in Kinyarwanda.114 To identify potential users’ several countries do serve rural households needs correctly it is crucial to include par- but at a relatively high cost, and governments ticipatory approaches in digital agriculture often step in through policies, regulations, initiatives.115 A needs assessment conducted and state financial institutions. Despite good for the Kubere Centre in Uganda indicat- intentions, some of these measures do not ed that women were primarily interested in achieve the intended outcomes, and may in- farming techniques, market prices for farmer stead hinder the sustainable provision of produce, and health and education issues.116 financial services for agriculture, while also Consequently, this information was provid- failing to crowd-in private sector participa- ed through radio and mobile phones used tion. In most countries, despite government by women. Digital technologies can be de- intervention, agriculture attracts credit well signed to enable smallholder farmers with below its contribution to GDP. Among key low cognitive skills to learn and upgrade issues for preventing the spread of financial their skills as they use them. This process en- services in rural areas and agriculture are a courages inclusion rather than displacement high cost to serve clients in remote areas (with of low-skilled smallholders and helps raise seasonal and small transactions), and percep- their productivity. tions of risk due to lack of agroclimatic data HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 29 and financial records. Digital technologies culture.118 For instance, it is difficult to sell can help in providing solutions to both these products on e-commerce platforms if there issues. Digital financial services, through mo- are no roads to markets, or to sell high-qual- bile phones, ATMs, Internet-banking, and ity fruits to online customers if there is no agent banking, can reach smallholder farm- storage to preserve their freshness for distant ers in remote areas at very low cost. In terms markets. A recent study in China indicated of risks, digital technologies can provide in- that poor storage and transportation, par- formation on agroclimatic risks, enabling ticularly for perishable products, was a fac- providers of financial services to assess the tor in decisions not to sell products online.119 riskiness of production that their clients In addition, rural electrification can help fa- face. In addition, digital payments and other cilitate rural-based food processing and val- transactions that use digital platforms gener- ue addition. Some of these sectors (roads, ate financial records that can help providers electrification, and storage) may be able to of financial services assess credit risks for crowd-in private investment, and could be these clients where other forms of financial the subject of their own “cascade” consider- records do not exist. ations of public-private investment (see ta- ble 1) but that is beyond the scope of this Invest in complementary infrastructure paper. In addition, there are also significant and policies that increase incentives to policy improvements countries can make to invest. Lack of complementary rural in- increase incentives of farmers and agribusi- frastructure, such as roads, energy, post-har- nesses to invest in farms and across agri- vest storage, and logistics can limit adoption cultural value chains,120 which, in turn, can and impact of digital technologies in agri- increase demand for digital technologies. 30 FUTURE OF FOOD Implementation considerations Implementation using an MFD approach to more remote rural areas; supporting requires a more private sector-oriented youth and women entrepreneurs with perspective and public-private dialogue in- mentoring and technical support; improv- formed by ex-ante analytics and data on ing the skills of smallholder farmers to constraints and opportunities121 to enhance better take advantage of digital technol- digital technologies to improve food sys- ogies; and fostering development of dig- tem outcomes. This approach can help guide ital technologies, including government the prioritization of public sector actions, e-services, customized to the needs and particularly on efficiency, equity, and en- general skills levels of farmers, delivered vironmental impacts, and to address risks in suitable formats and relevant languages. associated with digital technologies such as Women are particularly disadvantaged as potential exclusion, lack of data privacy, mar- regards access to, use and control of digital ket concentration of service providers, and tools, especially in developing countries.122 cybersecurity breaches. Targeting public sup- Programs can be designed to have signif- port and efforts to improve data privacy are icant impacts on women. For example, two aspects worth reiterating. recent support connected women farm- ers with agricultural market platforms in • Targeting public support to foster digital Bihar, India. The project organized wom- inclusion in the food system should be a en farmers into their own producer com- key consideration, with particular atten- pany and supported them with a host of tion to disadvantaged groups. This pro- services: daily commodity price informa- cess includes: expanding network coverage tion via mobile phones; digital scales and MAXIMIZING FINANCE FOR DEVELOPMENT IN AGRICULTURAL VALUE CHAINS 31 electronic moisture meters to challenge culture. While implementation will require manipulated equipment of traders; access collaboration and action across government to an online commodity exchange allow- ministries—including communication, fi- ing them to sell anywhere in the country; nance, education, rural development, and and improved access to storage and ware- transportation—agriculture ministries need housing so they could delay selling until to play a more prominent role in enhanc- prices improved. These interventions re- ing digital technologies to improve food sys- sulted in a 20 percent increase in the prod- tem outcomes. To achieve this, agriculture uct prices the women farmers received. ministries need to: (i) engage with the pri- Beyond individual programs, digital inno- vate sector to develop a clear and mutual un- vations should be mainstreamed into na- derstanding of constraints and opportuni- tional strategies in a gender-sensitive way ties to enhancing digital technologies in the to help advance the processes of women’s food system; (ii) engage with other minis- social inclusion. There is also a need to col- tries and provide a voice to policies and in- lect gender-disaggregated data to analyze vestments needed in the food system such how women and men access digital tools as digital and other infrastructure, invest- and use them for agriculture-related ac- ment climate, digital skills development, and tivities, and how access for women can be entrepreneurship programs, as well as facil- improved. itate government-to-government data shar- ing and linkages; (iii) provide direct support • Data privacy considerations. Efforts to through e-extension to farmers, digital skills improving data privacy should give con- development programs, market information sideration to three underlying principles: services, matching grants for facilitating dig- (i) transparency in data collection, in ital entrepreneurship, and programs to sup- other words, individuals should know if port women farmers, such as the Bihar exam- someone is collecting their data; (ii) in- ple cited above; and (iv) digitize some of their dividuals need to know and have a voice own systems including government-to-farm- in how their data are being used; and er payments (e-vouchers), early warning and (iii) data sharing models need to work risk management systems, and M&E. for both data suppliers (individuals) and users (enterprises/companies). Data gov- Countries need to tailor actions to spe- ernance arrangements should serve to cific local conditions. Prioritizing actions build confidence and trust of users of across supply-side constraints (such as a lack digital technologies, such as farmers and of network coverage and limited digital con- agribusinesses, and help facilitate devel- tent), and demand-side constraints (such as opment of digital applications that can lack of knowledge or complementary in- benefit these farmers and agribusinesses, vestments) will obviously depend on which such as improving access to finance, as of these aspects is most pressing in a spe- discussed earlier in the paper. cific location. For example, some countries Multi-agency and multisectoral collabo- have lower digital applications per person ration, and the role of ministries of agri- than their network coverage would seem 32 FUTURE OF FOOD to suggest, based on cross-country com- to smallholder farmers, youth, women, and parisons, indicating perhaps that relatively other vulnerable groups; and support skills more attention should be given to entrepre- development and training. Public policy neurship to develop digital applications and should also focus on creating an enabling content in these cases, especially for small- environment that will create effective de- holders. While other countries have higher mand for digital technologies that deliver digital applications per person than their improved food system outcomes. Through network coverage would seem to suggest, in- monitoring uptake of these technologies, dicating perhaps relatively more attention public actions can focus on resolving bot- should be given to expanding network cov- tlenecks that prevent the market from erage relative to entrepreneurship. responding to demand, such as: investing in digital entrepreneurship, human capital de- Digital technologies have the potential to velopment, and supporting infrastructure; transform the food system through better encouraging private sector competition to informed and engaged consumers, smart- improve the availability of digital technolo- er farms, and improved delivery of public gies and lower costs of adoption; regulating services. Digital technologies also pose sev- the use of digital technologies to reduce po- eral risks, including exclusion, lack of data tentially harmful uses; and in some cases, privacy, cybersecurity breaches, and over- subsidizing the use of digital technologies concentration of service provider market by disadvantaged groups to ensure that they power. Addressing these risks calls for pub- do not fall even further behind. These ac- lic policies that: keep service provider entry tions can all harness digital technologies barriers low; ensure good data governance; to provide much-needed improvements in foster inclusion through targeted support food system outcomes. HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 33 34 FUTURE OF FOOD Endnotes 1. World Bank, Poverty and Shared Prosperity 2018: a Sustainable Future: A Menu of Solutions to Feed Piecing Together the Poverty Puzzle (Washington, Nearly 10 Billion People by 2050 (Washington, DC: DC: World Bank, 2018). World Resources Institute, 2018). 2. Ibid. 14. Ibid. 3. FAO, IFAD, UNICEF, WFP, and WHO, The State 15. GSMA, The Mobile Economy 2018 (London: GSMA of Food Security and Nutrition in the World 2018: 2018). Building Climate Resilience for Food Security and 16. Ibid. Nutrition (Rome: Food & Agriculture Org., 2018). 17. GSMA, Enabling Rural Coverage: Regulatory and 4. FAO (Food and Agriculture Organization of Policy Recommendations to Foster Mobile Broadband the United Nations), The State of Food and Coverage in Developing Countries (London: GSMA, Agriculture 2013: Food Systems for Better Nutrition 2018). (Rome: FAO, 2013). 18. Digital technologies are a subset of agricultural 5. Development Initiatives, 2018 Global Nutrition technologies. Report: Shining a Light to Spur Action on Nutrition (Development Initiatives Poverty Research Ltd, 19. Finistere Ventures, “2018 Agtech Investment Bristol, 2018). Review,” available at: https://pitchbook.com/news/ articles/finistere-ventures-2018-agtech-investment- 6. WHO, WHO Estimates of the Global Burden of review. Foodborne Diseases: Foodborne Disease Burden Epidemiology Reference Group 2007–2015 (Geneva, 20. McKinsey Global Institute Analysis. Switzerland: WHO, 2015). 21. H. Sekabira and M. Qaim, “Can Mobile Phones 7. ILO (International Labor Organization) data. Improve Gender Equality and Nutrition? Panel Data Evidence from Farm Households in Uganda,” 8. UNIDO (United Nations Industrial Development Food Policy 73 (2017): 95–103. Organization) data. 22. D. Acemoglu, “Why Do New Technologies 9. Based on 2013 Eurostat data that separates Complement Skills? Directed Technical Change and manufacturing into 24 subsectors; Food Drink Wage Inequality, The Quarterly Journal of Economics Europe, “Data and Trends: EU Food and Drink 113, no. 4 (1998): 1055–89. Industry 2016”. 10. E. Nkonya, A. Mirzabaev, and J. Von Braun, eds., 23. C. Goldin and L. Katz, The Race Between Education Economics of Land Degradation and Improvement: and Technology (Cambridge: Harvard University A Global Assessment For Sustainable Development Press, 2009). (Cham, Switzerland: Springer International 24. S. Cole and A. Fernando, “‘Mobile’izing Agricultural Publishing, 2016). Advice: Technology Adoption, Diffusion and 11. A. Richey, B. Thomas, MH. Lo, et al., “Quantifying Sustainability,” Harvard Business School Finance Renewable Groundwater Stress with GRACE,” Working Paper 13–047 (2016). Water Resources Research 51, no. 7 (2015): 5217–238. 25. M. Keogh and M. Henry, “The Implications 12. E. Cassou, S. Jaffee, and J. Ru, The Challenge of of Digital Agriculture and Big Data for Agricultural Pollution: Evidence from China, Australian Agriculture” (Australian Farm Institute, Vietnam, and the Philippines (Washington, DC: Sydney, 2016). World Bank, 2018). 26. M. Jouanjean, “Digital Opportunities for Trade in 13. T. Searchinger, R. Waite, C. Hanson, J. the Agriculture and Food Sectors” OECD Food, Ranganathan, P. Dumas, and E. Matthews, Creating Agriculture and Fisheries Papers no. 122 (2019). HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 35 27. World Economic Forum, Innovation with a Purpose: Village and Household Surveys.,” Background Paper The Role of Technology Innovation in Accelerating Food (World Bank, Washington, DC, 2018). Systems Transformation (Geneva: World Economic 39. U. Deichmann, A. Goyal, and D. Mishra, “Will Forum, 2018). Digital Technologies Transform Agriculture in 28. M. Jouanjean, “Digital Opportunities for Trade in Developing Countries?” World Bank Policy Research the Agriculture and Food Sectors” OECD Food, Working Paper no. 7669 (2016). Agriculture and Fisheries Papers No. 122 (2019). 40. E. Nakasone and M. Torero, “A Text Message 29. M. Tripoli and J. Schmidhuber, “Emerging Away: ICTs as a Tool to Improve Food Security,” Opportunities for the Application of Blockchain in Agricultural Economics 47, no. S1 (2016): 49–59. the Agri-food Industry” (FAO and ICTSD: Rome 41. World Bank, ICT in Agriculture: Connecting and Geneva, License: CC BY-NC-SA 3 2018). Smallholders to Knowledge, Networks, and Institutions 30. V. Martchenko, J. Cook, and M. Shirazi, (Scott (Washington, DC: World Bank, 2017). Automation and Robotics), “Hyperspectral Food 42. Digital Green. Annual Report 2017. Safety Inspection System: Final Report,” Meat and Livestock Australia (2016). 43. K. Toyama, R. Gandhi, R. Veeraraghavan, and V. Ramprasad, “Digital Green: Participatory Video and 31. Tellspec, “How Ripe is My Fruit?” 2018, http:// Mediated Instruction for Agricultural,” Information tellspec.com/wp-content/uploads/2018/08/ Technologies and International Development 5, no. 1 Tellspec_FruitQC.pdf. (2009): 1–15. 32. World Economic Forum, Innovation with a Purpose: 44. E. Nakasone and M. Torero, “A Text Message The Role of Technology Innovation in Accelerating Food Away: ICTs as a Tool to Improve Food Security,” Systems Transformation (Geneva: World Economic Agricultural Economics 47, no. S1 (2016): 49–59. Forum, 2018). 45. Ibid. 33. S. Jaffee, S. Henson, L. Unnevehr, D. Grace, and E. Cassou, The Safe Food Imperative: Accelerating 46. World Bank, ICT in Agriculture (Updated Edition): Progress in Low-and Middle-Income Countries Connecting Smallholders to Knowledge, Networks, and (Washington, DC: World Bank, 2018). Institutions (Washington, DC: World Bank, 2017). 34. YZ. Feng and DW. Sun, “Application of 47. D. Schimmelpfennig, “Farm Profits and Adoption Hyperspectral Imaging in Food Safety Inspection of Precision Agriculture,” USDA Economic Research and Control: A Review,” Critical Reviews in Food Report 217. (2016). Science and Nutrition 52, no. 11 (2012): 1039–58. 48. M. Tulsian and N. Saini. “Market– 35. J. Aker and M. Fafchamps, “Mobile Phone Coverage driven innovations in rural marketing in and Producer Markets: Evidence from West Africa,” India.” International Journal of Scientific & The World Bank Economic Review 29, no. 2 (2015): Engineering Research 5, no. 5 (2014): 1439–45. 262–92. 49. World Bank, ICT in Agriculture (Updated Edition) 36. E. Nakasone, M. Torero, and B. Minten, “The Power 2017. of Information: The ICT Revolution in Agricultural 50. K. Kennedy Freeman and R. Orejas, “Information Development,” Annu. Rev. Resour. Econ. 6, no. 1 Systems for Sustainable Agriculture, Uruguay.,” (2014): 533–50. in M. Sewadeh and S. Jaffee (ed.), Shades of 37. T. Mitchell, “Is Knowledge Power? Competition and Green: Multi-Stakeholder Initiatives to Reduce the Information in Agricultural Markets,” No. iiisdp456. Environmental Footprint of Commercial Agriculture IIIS, (2014). (EcoAgriculture Partners 2015). 38. J. Huang and H. Zhi, “The Uses of ICT in Agriculture 51. J. Van Loon, A. Speratti, and B. Govaerts. “Precision and E-commerce in Rural China: The Evidence from for smallholder farmers: a small-scale-tailored 36 FUTURE OF FOOD variable rate fertilizer application kit.” Agriculture 8, 63. Ibid. no. 4 (2018): 48. 64. World Bank, Global Financial Development Report 52. F. Adu-Baffour, T. Daum, and R. Birner, “Can Big 2017/2018: Bankers without Borders (Washington, Companies’ Initiatives to Promote Mechanization DC: World Bank, 2018). Benefit Small Farms in Africa? A Case Study from 65. “Big data” is an evolving term that describes Zambia” ZEF-Discussion Papers on Development a large volume of structured, semi-structured and Policy no. 262 (2018). unstructured data that have the potential to be 53. P. Courtois and J. Subervie, “Farmer Bargaining mined for information and used in machine learning Power and Market Information Services,” American projects and other advanced analytics applications. Journal of Agricultural Economics 97, no. 3 (2014): 66. E. Kärner, “The Future of Agriculture is Digital: 953–77. Showcasting e-Estonia,” Frontiers in Veterinary 54. E. Nakasone, “The Role of Price Information in Science 4: 151 (2017). Agricultural Markets: Experimental Evidence from 67. J. Grossman and M. Tarzai, “Serving Smallholder Rural Peru,” Paper presented at the Agricultural Farmers: Recent Developments in Digital and Applied Economics Association (AAEA) & Finance” (Consultative Group to Assist the Poor, Canadian Agricultural Economics Association Washington, DC, 2014). (CAES) Joint Annual Meeting. Washington DC. 4–6 August 2013. 68. Ibid. 55. Ibid. 69. IFAD, “Statement by Dr. Akinwumi Adesina, Honourable Minister of Agriculture and Rural 56. J. Aker, “Does Digital Divide or Provide? The Development, Federal Republic of Nigeria, Impact of Cell Phones on Grain Markets in Niger,” Governor from Nigeria, at the 36th Session of the Center for Global Development Working Paper 154 IFAD Governing Council” (2013). (2008). 70. World Bank, “Promoting Digital Agriculture: Case 57. J. Aker and M. Fafchamps, “How Does Mobile of E-voucher Program Implementation in Three Phone Coverage Affect Farm-Gate Prices? Evidence West African Countries,” World Bank Memo from West Africa” (Department of Economics and (World Bank, Washington, DC, 2019). the Fletcher School, Tufts University, Massachusetts, 2010). 71. Machine learning is a method of data analysis that automates analytical model building. It is a branch 58. E. Nakasone and M. Torero, “A Text Message of artificial intelligence based on the idea that Away: ICTs as a Tool to Improve Food Security,” systems can learn from data, identify patterns, and Agricultural Economics 47, no. S1 (2016): 49–59. make decisions with minimal human intervention. 59. Ibid. 72. R. Townsend, L. Ronchi, C. Brett, and G. Moses, 60. P. Varangis, M. Kioko, M. Spahr, et al., Access to “Future of Food: Maximizing Finance for Finance for Smallholder Farmers: Learning from the Development in Agricultural Value Chains.” (World Experiences of Microfinance Institutions in Latin America Bank, Washington, DC, 2018). (English) (Washington, DC: World Bank, 2014). 73. GSMA data. 61. O. Kirui, J. Okello, R. Nyikal, and G. Njiraini, 74. Ibid. “Impact of Mobile Phone-Based Money Transfer Services in Agriculture: Evidence from Kenya,” 75. Frontier Economics, “Benefits of Network Quarterly Journal of International Agriculture 52, no. Competition and Complementary Policies to 892–2016–65177:141 (2013). Promote Mobile Broadband Coverage,” Prepared for GSMA (GSMA, London, 2015). 62. W. Jack and T. Suri, “Mobile Money: The Economics of M-PESA,” NBER Working Paper no. 16721 (2011). 76. Ibid. HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 37 77. GSMA data and World Development Indicators. 91. Disrupt Africa, “Agrinnovating for Africa: Exploring the African Agri-Tech Startup Ecosystem 78. Frontier Economics, “Benefits of Network Report 2018” (Disrupt Africa, 2018). Competition and Complementary Policies to Promote Mobile Broadband Coverage,” Prepared for 92. World Bank, World Development Report 2016. Digital GSMA (GSMA, London, 2015). Dividends (Washington, DC: World Bank, 2016). 79. Ibid. 93. Ibid. 80. P. Buys, S. Dasgupta, T. Thomas, and D. Wheeler, 94. A. Brunetti, G. Kisunko, and B. Weder, “Credibility “Determinants of a Digital Divide in Sub-Saharan of Rules and Economic Growth: Evidence from a Africa: A Spatial Econometric Analysis of Cell Worldwide Survey of the Private Sector,” The World Phone Coverage,” World Development 37, no. 9 Bank Economic Review 12, no. 3 (1998): 353–84. (2009): 1494–505. 95. PwC, “Global Blockchain Survey 2018” (PwC, 81. Ibid. 2018). 82. World Bank, Enabling the Business of Agriculture 96. GSMA, A New Regulatory Framework for the Digital 2017 (Washington, DC: World Bank, 2017). Ecosystem (London: GSMA, 2016). 83. GSMA, Enabling Rural Coverage: Regulatory 97. M. Keogh and M. Henry, “The Implications of Digital and Policy Recommendations to Foster Mobile Agriculture and Big Data for Australian Agriculture” Broadband Coverage in Developing Countries (Australian Farm Institute, Sydney, 2016). (London: GSMA, 2018). 98. EU Code of Conduct on Agricultural Data Sharing 84. World Bank, Enabling the Business of Agriculture by Contractual Agreement. 2017. 99. J. De Beer, “Ownership of Open Data: 85. The Enabling the Business of Agriculture ICT Governance Options for Agriculture and indicator focuses on the licensing framework for Nutrition” (Wallingford: Global Open Data for mobile operators and assesses the type of licensing Agriculture and Nutrition, 2016). regime used in a country, the validity of the 100. M. Trajtenberg, “AI as the Next GPT: a Political- operating license, the public availability of operating Economy Perspective,” NBER Working Paper no. license costs, and spectrum allocation. 24245 (2018). 86. R. Samarajiva and A. Zainudeen. “Regulatory 101. Using GSMA data Reform and Rural Roll-Out of Information and Communication Technologies,” in A. Jhunjhunwala 102. FAO, “FAO Private and Public Partnership Model (ed.), Information Technology and Communication for Youth Employment in Agriculture. Experiences Resources for Sustainable Development (UNESCO from Malawi, Tanzania Mainland, and Zanzibar in partnership with Encyclopedia Life Support Archipelago,” Rural Employment Case Studies Series 4 Systems, 2010). (FAO, Rome, 2014). 87. S. Parsons and J. Stegeman, “Rural Broadband 103. J. Kluve, S. Puerto, D. Robalino, et al., “Do Youth Economics: A Review of Rural Subsidies” Employment Programs Improve Labor Market (Cincinnati: CostQuest Associates, 2018). Outcomes? A Quantitative Review,” World Development 114 (2019): 237–53. 88. J. Genachowski, “Bringing Broadband to Rural America: Update to Report on a Rural Broadband 104. Finistere Ventures, 2017 Agtech Investment Review. Strategy,” Federal Communications Commission, 105. L. Klapper, “How Digital Payments Can Benefit GN Docket 11–16 (2011). Entrepreneurs,” IZA World of Labor 396 (2017). 89. Using GSMA data. 106. R. Zong, “China Can’t Afford a Cashless Society. 90. Digital technologies are a subset of agricultural A Mania for Mobile Payments is Leaving the Poor technologies. Behind,” Foreign Policy, September 11 (2018). 38 FUTURE OF FOOD 107. World Bank, World Development Report 2016. Digital in the Use of Information and Communication Dividends (Washington, DC: World Bank, 2016). Technologies (ICTs) for Agriculture and Rural Development (Rome: FAO, 2018). 108. World Bank, “The Role of Agriculture on Romania’s Path to Prosperity,” Input to Romania Systematic 116. IFAD (International Fund for Agricultural Country Diagnostic (World Bank, Washington, Development), Gender in Agriculture Sourcebook DC, 2018). (Washington, DC: World Bank, 2009). 109. GSMA, Connected Women. The Mobile Gender Gap 117. G. Palloni, J. Aker, D. Gilligan, M. Hidrobo, and N. Report 2018 (GSMA: London, 2018). Ledlie, “Paying for Digital Information: Assessing Farmers Willingness to Pay for a Digital Agriculture 110. J. Aker, I. Ghosh, and J. Burrell. “The and Nutrition Service in Ghana” Paper presented promise (and pitfalls) of ICT for agriculture at International Association of Agricultural initiatives.” Agricultural Economics 47, no. S1 (2016): Economists Conference. Vancouver, British 35–48. Columbia. 28 July-2 August 2018. 111. GSMA, Connected Women. The Mobile Gender Gap 118. U. Deichmann, A. Goyal, and D. Mishra, “Will Report 2018 (GSMA: London, 2018). Digital Technologies Transform Agriculture in 112. K. Vasilaky and A. Islam, “Competition or Developing Countries?” World Bank Policy Research Cooperation? Using Team and Tournament Working Paper no. 7669 (2016). Incentives for Learning among Female Farmers in Rural Uganda,” World Development 103 (2018): 119. J. Huang and H. Zhi, “The Uses of ICT in 216–25. Agriculture and E-commerce in Rural China: The Evidence from Village and Household Surveys.,” 113. S. Treinen, A. Van der Elstraeten, and C. Background Paper (World Bank, Washington, DC, Pedrick, Gender and ICTs: Mainstreaming Gender 2018). in the Use of Information and Communication Technologies (ICTs) for Agriculture and Rural 120. R. Townsend, L. Ronchi, C. Brett, and G. Moses, Development (Rome: FAO, 2018). “Future of Food: Maximizing Finance for Development in Agricultural Value Chains.” (World 114. J. Grossman and M. Tarazi, “Serving Smallholder Bank, Washington, DC, 2018). Farmers: Recent Developments in Digital Finance,” CGAP Focus Note 94 (2014). 121. Ibid. 115. S. Treinen, A. Van der Elstraeten, and C. 122. ITU (International Telecommunication Union), ICT Pedrick, Gender and ICTs: Mainstreaming Gender Facts and Figures 2017(Geneva: ITU, 2017). HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 39 40 FUTURE OF FOOD HARNESSING DIGITAL TECHNOLOGIES TO IMPROVE FOOD SYSTEM OUTCOMES 41 1818 H Street, NW Washington, DC 20433 USA Telephone: 202-473-1000 42 FUTURE OF FOOD