Republic of Guatemala DIGITAGRO Investing in digital technology to increase market access for women agri-preneurs in Guatemala Report No: ACS34060 Viviana M.E. Perego, Javier Romero, Katie Freeman, Angela Lopez, Glenn Ortiz, Hugo Salas, Rudy Ramirez, Arianna Locatelli, Danielle Orihuela, Camila de Ferrari DIGITAGRO 2022 Report No: ACS34060 Republic of Guatemala DIGITAGRO: Investing in digital technology to increase market access for women agripenurs in Guatemala. April 2022 SLCAG LATIN AMERICA AND CARIBBEAN 2 DIGITAGRO 2022 Standard Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement: The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@ worldbank.org. Images from Adobe Stock stock.adobe.com 3 DIGITAGRO 2022 Acknowledgments The DIGITAGRO project was financed by the InfoDev Trust Fund and implemented by the Agriculture and Food Global Practice of the World Bank with support from the Gender Innovation Lab for Latin America and the Caribbean of the World Bank (LACGIL). The broader DIGITAGRO team included Viviana Maria Eugenia Perego (Task Team Leader, Agriculture Economist, World Bank), Katie Kennedy Freeman (Senior Agriculture Economist, World Bank), Javier Romero (Economist, World Bank), Angela Rocio Lopez (Consultant, World Bank), Hugo Salas (Consultant, World Bank), Glenn Ortiz (Consultant, World Bank), Rudy Ramirez (Consultant, World Bank), Barbara Coello (Consultant, World Bank), Dennis Escudero (Investment Support Officer, Food And Agriculture Organisation of the United Nations), Karin Estrada (Consultant, Food And Agriculture Organisation of the United Nations), Arianna Locatelli (Intern, World Bank), Danielle Marie Orihuela (Intern, World Bank), Camila De Ferrari (Intern, World Bank), Maria Amalia Cordova (Intern, World Bank). The e-commerce platform was built by Code Solutions Network S.A., whereas the extension videos were produced by Espacio Visual S.A. Surveying and data collection were carried out by Innovations for Poverty Action. The team is grateful for the guidance and support of Preeti Ahuja (Practice Manager, World Bank), Eric Lancelot (Program Leader, World Bank), Marco Scuriatti (Resident Representative, World Bank), Fernando Paredes (Senior Operations Officer, World Bank). Kateryna Schroeder (Agriculture Economist, World Bank), Jacobus Joost De Hoop (Senior Economist, World Bank), and Patricia Van de Velde (Consultant, World Bank) kindly served as Peer Reviewers for this report. Alejandro de la Fuente (Senior Economist, World Bank) and Dahyeon Jeong (Economist, World Bank) provided valuable feedback during the LACGIL Quality Enhancement Review of the experimental design of this project. The team acknowledges advice and brainstorming from Tomás Ricardo Rosada Villamar (Senior Agriculture Economist, World Bank), Michael Morris (Lead Agriculture Economist, World Bank), Lourdes Rodriguez Chamussy (Senior Economist, World Bank), Ana Maria Munoz Boudet (Senior Social Scientist, World Bank), Elizaveta Perova (Senior Economist, World Bank), Rocio Sanchez Vigueras (Digital Development Specialist, World Bank), and members of the LACGIL. The DIGITAGRO project involved a variety of activities and continued dialogue between the World Bank and partner institutions World Food Programme and Food and Agriculture Organisation of the United Nations. In particular, the team wishes to especially acknowledge the close collaboration with Irene del Rio, Andreia Fausto, Lena Schubmann, Karen Kestler, Luis Melgar, Gabriela León, Herbert Canastuj, Allan Solorzano, Marvin Lima from the World Food Programme; as well as Maynor Estrada, Marco Moncayo, Norma Pérez, Milton Orozco, Wenceslao Barrios from the Food and Agriculture Organisation of the United Nations. The team recognizes with gratitude the support and collaboration from the Departmental Interinstitutional School Feeding Technical Commission of San Marcos, Guatemala, and in particular: Leonel de León, Gilben Escobar, Guadalupe Rodas, Aumner Pérez, Marcos López, José García, Ronal García, from Ministry of Agriculture; Guillermo Mazariegos, David Makepeace, Mibzar Fuentes, Elfego López, Víctor Morales, Edgar Mazariegos, from the Ministry of Education; José Rabanales, from the Ministry of Public Health. Invaluable support on field was provided by Ministry of Agriculture extension agents in San Marcos. Valuable insights and advice were gathered from Maria Febres and Lourdes Ortiz (Inter-American Institute for Cooperation on Agriculture), Oscar Grajeda (International Fund for Agricultural Development), Maria José Schaeffer (United Nations), Eugenia Close (United Nations Entity for Gender Equality and the Empowerment of Women), Iván Buitron (Asociación de Exportadores de Guatemala), Gerardo de León (Federación de Cooperativas Agrícolas de Productores de Café de Guatemala). The team expresses its gratitude for logistical and administrative support from Mario Méndez (Team Assistant, World Bank), Sofia Neiva (Team Assistant, World Bank), Jairi Hernandez (Team Assistant, World Bank). Translation services were provided by Sofie Van Renterghem and graphic design by Jaime Sosa. 4 DIGITAGRO 2022 Contents Executive Summary................................................................................................................................................................... 11 Chapter 1. Introduction............................................................................................................................................................... 19 Chapter 2. Family agriculture, child nutrition, and women’s (dis-) empowerment in Guatemala: three related challenges............................................................................................................................................................... 22 2.1 Family Agriculture............................................................................................................................................................ 23 2.2 Women in Rural Guatemala.......................................................................................................................................... 27 2.3 Food and nutrition security.......................................................................................................................................... 29 2.4 COVID-19........................................................................................................................................................................... 31 Chapter 3. The School Feeding Program: an opportunity for family farming................................................................ 32 3.1 Objectives and Structure............................................................................................................................................... 33 3.2. Key SFP institutions........................................................................................................................................................ 35 3.3. Expected benefits of the SFP....................................................................................................................................... 36 3.4 Bottlenecks....................................................................................................................................................................... 37 3.5 The SFP and the COVID-19 pandemic........................................................................................................................ 38 Chapter 4. DIGITAGRO................................................................................................................................................................ 39 4.1 Digital technologies for rural development.............................................................................................................. 40 4.2 DIGITAGRO: Investing in digital technology to increase market access for women agri-preneurs in Guatemala................................................................................................................................................ 42 4.2.1 E-commerce platform................................................................................................................................................. 44 4.2.2 Suite of extension videos........................................................................................................................................... 45 4.2.3 Digital information campaign and impact evaluation......................................................................................... 47 Chapter 5. Evaluating a digital information campaign in San Marcos............................................................................. 52 5.1 The experiment............................................................................................................................................................... 53 5.2 Characterizing women agri-preneurs in San Marcos............................................................................................. 56 5.3 Did the digital information campaign work?............................................................................................................. 60 5.3.1 The intervention increased knowledge about the School Feeding Program and encouraged some women to sell SFP products.................................................................................................................................... 60 5.3.2 The intervention did not convince many women to register in the SFP........................................................ 67 Chapter 6. Why are women not entering the SFP market? An exploration of barriers to participation................. 69 6.1 Product portfolio and land size................................................................................................................................... 70 6.2 Skills needed to join the SFP......................................................................................................................................... 73 6.3 Registration process and institutional trust............................................................................................................. 74 6.4 Prices.................................................................................................................................................................................. 76 6.5 Recommendations: How to promote women’s participation in the School Feeding Program?.................. 78 5 DIGITAGRO 2022 Chapter 7. Discussion and conclusion: Digital technologies and agrifood transformation....................................... 80 References..................................................................................................................................................................................... 85 Annexes........................................................................................................................................................................................... 90 Annex 1. Baseline descriptive statistics and sample balance..................................................................................... 90 Annex 2. Take-up and response rate tables.................................................................................................................. 117 Annex 3. Sample balance at endline and quality of the empirical design.............................................................. 119 Annex 4. Detailed impact evaluation results................................................................................................................. 121 Annex 5. Robustness checks using the Inverse Probability Weighting Estimator (IPWE)................................... 137 Annex 6. Results of focus group discussions and of complementary surveys in San Marcos......................... 137 A6.1 Focus groups with farmers and OPFs at project design stage........................................................................ 137 A6.2 Virtual focus group with OPFs at endline.............................................................................................................. 140 A6.3 Virtual focus groups with MAGA extensionists at endline................................................................................ 143 A6.4 Survey of MAGA extensionists on contacts with treated group participants at endline........................... 146 A6.5 Survey of existing SFP providers............................................................................................................................. 147 List of Tables Table 1. Criteria for classification of households in family agriculture............................................................................. 24 Table 2. Main activities of the WFP and FAO in the joint FAO-WFP project................................................................... 44 Table 3. General household information................................................................................................................................ 90 Table 4. Agricultural engagement information...................................................................................................................... 94 Table 5. Agricultural production.............................................................................................................................................. 101 Table 6. Agricultural sales......................................................................................................................................................... 105 Table 7. Perceptions about the SFP........................................................................................................................................ 111 Table 8. Baseline response rate, by treatment group........................................................................................................ 117 Table 9. Endline response rate, by treatment group......................................................................................................... 117 Table 10. Verified and assisted farmers, by treatment groups........................................................................................ 117 Table 11. Take-up questions for the treatment group...................................................................................................... 118 Table 12. Balance table at endline.......................................................................................................................................... 119 Table 13. Covariates included in all regressions................................................................................................................. 121 Table 14. Treatment effect on information intake about the School Feeding Program............................................ 123 Table 15. Treatment effect on SFP information intake by participation in traditional extension programs........ 124 Table 16. Treatment effect on main information source about the SFP....................................................................... 125 Table 17. Treatment effect on women’s sales...................................................................................................................... 126 Table 18. Treatment effect on women’s sales by agency level (animal products)...................................................... 127 6 DIGITAGRO 2022 Table 19. Treatment effect on sales of SFP animal products, by marital status.......................................................... 128 Table 20. Treatment effect on women’s decision-making around sales of agricultural and animal products, by marital status................................................................................................................................... 129 Table 21. Treatment effect on willingness to participate in the SFP............................................................................... 130 Table 22. Treatment effect on selling barriers..................................................................................................................... 131 Table 23. Adoption of agricultural practices, by SFP registration status....................................................................... 132 Table 24. Treatment effect on registration process awareness and perceptions...................................................... 133 Table 25. Treatment effect on institutional perception (MAGA)...................................................................................... 134 Table 26. Treatment effect on institutional perception (MINEDUC)............................................................................... 135 Table 27. Treatment effect on institutional perception (SAT).......................................................................................... 136 List of Figures Figure 1. Agricultural production among sampled women in San Marcos.................................................................... 57 Figure 2. Produce of animal origin among sampled women in San Marcos.................................................................. 57 Figure 3. Reported barriers to sell............................................................................................................................................ 58 Figure 4. Main reason for being interested in the SFP market.......................................................................................... 59 Figure 5. Main reason for NOT being interested in the SFP market................................................................................ 60 Figure 6. Treatment effect on information intake about the School Feeding Program.............................................. 61 Figure 7. Treatment effect on SFP information intake by participation in traditional extension programs........... 62 Figure 8. DIGITAGRO treatment effect among Indigenous beneficiaries...................................................................... 63 Figure 9. Treatment effect on main information source about the SFP......................................................................... 64 Figure 10. Treatment effect on women’s sales...................................................................................................................... 65 Figure 11. Treatment effect on women’s sales by agency level (animal products)....................................................... 66 Figure 12. Treatment effect on sales of SFP animal products, by marital status.......................................................... 66 Figure 13. Treatment effect on women’s decision-making around sales of agricultural and animal products, by marital status – IPWE results............................................................................................................................... 67 Figure 14. Treatment effect on willingness to participate in the SFP............................................................................... 68 Figure 15. Treatment effect on selling barriers..................................................................................................................... 68 Figure 16. Distribution of farmers’ land size by registration status in the control group........................................... 71 Figure 17. Likeliness of selling one or more products, by SFP registration status....................................................... 72 Figure 18. Share of sales of SFP agricultural products by registration status and reference period...................... 72 Figure 19. Adoption of agricultural practices, by SFP registration status....................................................................... 73 Figure 20. Treatment effect on registration process awareness and perceptions...................................................... 74 Figure 21. Treatment effect on institutional perception..................................................................................................... 75 Figure 22. Treatment effect on contacts with MAGA and SAT........................................................................................... 76 7 DIGITAGRO 2022 Figure 23. Distribution of price differential between SFP and market prices, by registration status...................... 77 Figure 24. Treatment effects on entry and willingness to participate in the SFP........................................................ 137 List of Boxes Box 1. Guatemala’s National Rural Extension System......................................................................................................... 26 Box 2. Reforms to the School Feeding Law (SFL) approved in September 2021......................................................... 35 Box 3. Specific roles of SFP key actors..................................................................................................................................... 36 Box 4. WFP and FAO: Joint actions to link family agriculture and the School Feeding Program in Guatemala.................................................................................................................................................................................. 43 Box 5. Measuring Empowerment Better................................................................................................................................. 49 Box 6. Randomization protocol.................................................................................................................................................. 54 Box 7. Delivering the treatment using digital technologies in rural Guatemala............................................................ 55 Box 8. Did DIGITAGRO promote SFP inclusion of Indigenous Peoples?.......................................................................... 62 Box 9. Guatemala’s digital ecosystem...................................................................................................................................... 83 8 DIGITAGRO 2022 Abbreviations and Acronyms AMER Municipal Rural Extension Agencies (Agencias Municipales de Extensión Rural) CADER Learning Center for Rural Development (Centro de Aprendizaje para el Desarrollo Rural) CCFAS Climate Change, Agriculture and Food Security Program COVID-19 Coronavirus Disease 2019 CRI Climate Risk Index CTIDAE Departmental Interinstitutional School Feeding Technical Commission (Comisión Técnica Interinstitucional Departamental de Alimentación Escolar) CTIMAE Municipal Interinstitutional School Feeding Technical Commissions (Comisión Técnica Interinstitucional Municipal de Alimentación Escolar) ENCOVI National Living Standards Survey (Encuesta Nacional de Condiciones de Vida) ENSMI National survey on maternal and child health (Encuesta Nacional de Salud Materno Infantil) ENEI National Income and Employment Survey (Encuesta Nacional de Empleo e Ingresos) DIGEFOCE General Directorate for Strengthening the Educational Community (Dirección General de Fortalecimiento de la Comunidad Educativa) DIME Development Impact Evaluation group FAO Food and Agriculture Organization of the United Nations FONTIERRAS Land Fund Act GFSI Global Food Security Index GHI Global Hunger Index HDSS Household Dietary Diversity Score IFPRI International Food Policy Research Institute INDC Intended Nationally Determined Contribution INE National Statistical Institute (Instituto Nacional de Estadística) IOM International Organization for Migration IPC Integrated Food Security Phase Classification LAC Latin America and the Caribbean 9 DIGITAGRO 2022 LACGIL Gender Innovation Lab for Latin America and the Caribbean MAGA Ministry of Agriculture, Livestock, and Food (Ministerio de Agricultura Ganadería y Alimentación) MINFIN Ministry of Public Finance (Ministerio de Finanzas Públicas) MINEDUC Ministry of Education (Ministerio de Educación) MSPAS Ministry of Public Health and Social Assistance (Ministerio de Salud Pública y Asistencia Social) NIT Tax identification number (Número de Identificación Tributaria) OLS Ordinary Least Squares OPF School Parent Organization (Organización de padres de familia) PAFFEC Family Farming Program to Strengthen the Rural Economy (Programa de Agricultura Familiar para el Fortalecimiento de La Economía Campesina) Pro-WEAI Project-level Women’s Empowerment in Agriculture Index SAT Superintendency of Tax Administration (Superintendencia de Administración Tributaria) SEGEPLAN General Secretariat for Planning and Programming of the Presidency (Secretaría de Planificación y Programación de la Presidencia) SESAN Secretariat of Food and Nutritional Security (Secretaría de Seguridad Alimentaria y Nutricional) SFL School Feeding Law SFP School Feeding Program SNER National Rural Extension System (Sistema Nacional de Extensión Rural) STEG Union of Education Workers (Sindicato de Trabajadores y Trabajadoras de la Educación de Guatemala) UNDP United Nations Development Programme UNEP United Nations Environment Programme UNFPA United Nations Population Fund USAID United States Agency for International Development USDA United States Department of Agriculture WFP World Food Programme WHO World Health Organization 10 DIGITAGRO 2022 Executive Summary Guatemala’s School Feeding Program (SFP), which aims at improving food and nutrition security for schoolchildren, represents a key opportunity for promoting linkages between local smallholder farmers and public food procurement, since the program requires that 50 percent of food purchases by public schools come from local family farmers. Despite promising results in terms of child nutrition and involvement of family farmers, several challenges still hinder the correct functioning of the program, from low market information, to insufficient incentives for farmers to enroll as registered providers of the SFP, to low production capacity. These barriers are particularly severe for women producers, who face higher information gaps, lower market access, and higher informality than their male counterparts, compounded by restrictive social norms. Yet, school feeding represents a crucial opportunity for women farmers, whose production usually specializes in foods that are in high demand by schools, and it could represent an avenue for improved female entrepreneurship and empowerment. The DIGITAGRO project piloted digital technologies to improve market access for women agri-preneurs in the ambit of the School Feeding Program. The intervention, which operated in the department of San Marcos in Western Guatemala, revolved around three main activities aiming to address the information gaps and asymmetries that preclude the smooth functioning of the SFP, on one hand, and hold back women from taking advantage of the program as a profitable market opportunity, on the other: DIGITAGRO’s Three Main Activities The creation of an e-commerce The production of extension videos targeted platform, in partnership with the to women agri-preneurs, in partnership with World Food Programme (WFP). the Food and Agriculture Organisation of the United Nations (FAO). A digital information campaign, conducted in collaboration with the World Bank’s Gender Innovation Lab for Latin America and the Caribbean (LACGIL). The efficacy of this activity was also assessed through a rigorous impact evaluation following an experimental design. 11 DIGITAGRO 2022 E-commerce platform The platform Alimentación Escolar, which is currently being piloted on the field by the WFP, was developed with a human-centered approach and is composed of several user-friendly modules, including: Demand Supply Generation of accurate, school-specific Registration of local food suppliers as official shopping lists with products for school SFP providers, registration of products offered, meals, based on number of students and supply capacity, unit prices. the schools’ selection of official menus. Linking demand and supply (Admin) Purchase monitoring Matching demand and supply between Crossing data generated by suppliers and schools and local family farmers, performing schools’ shopping lists, monitoring of school transactions, feedback mechanisms on spending and processes. product quality. Considering the high connectivity challenges faced in rural areas of Guatemala, each module is supplemented by an SMS notification tool, which allows each aspect of the platform’s workflow to also operate offline. At the outbreak of the COVID-19 pandemic, a preliminary version of the platform (Mis compras del PAE) supported the Ministry of Education in ensuring the smooth functioning of the SFP in 25,000 schools nationwide. Extension videos The videos targeted to women agri-preneurs aimed to develop a soft extension system to engage them as effective suppliers of the SFP. The videos, available in Spanish and in Mam (the most prevalent Mayan language in San Marcos), were intended to complement the work by extension agents of the Ministry of Agriculture, providing them with an additional tool to convey basic information to large audiences and in a remote fashion (particularly during COVID-19 social distancing), to be then expanded on through dedicated work on the field. In total, the extension suite is composed of five short videos on the following topics: General information about the School Feeding Programme and sensibilisation on its implications for children’s nutrition, market opportunities, local economies, and community development. Practical guidance on how women farmers can access the school feeding market, including information on the online e-commerce platform; Good food hygiene and post-harvest practices, to increase farmers’ knowledge and skills around post-harvest practices and food quality and safety, and to ensure that farmers’ produce matches the quality, hygiene, and safety guidelines enforced at schools. The videos target agricultural products and products of animal origin relevant to the agricultural and cultural context of San Marcos and included in official school meals. 12 DIGITAGRO 2022 Information campaign and impact evaluation The information campaign carried out in collaboration with the World Bank’s LACGIL aimed at proactively encouraging women agri-preneurs to participate in the SFP market within the socially-distanced COVID-19 context. The campaign was entirely carried out by phone by sending an information bundle consisting of a video sent through WhatsApp and a series of text messages. Given the nature of the intervention, adult women with access to a smartphone and working in agriculture were invited to participate. A total of 881 eligible women across 272 villages in San Marcos enrolled in the program. Women received a summary video on their phones with key information on the general features of the SFP, ways for farmers to sell to schools, food quality standards and practices, and SFP functioning under COVID-19. In addition, women received a tailored set of six SMS that included reminders on how to access the SFP, information on products demanded and prices paid by the program, and relevant contact information. An impact evaluation study assessed the effectiveness of the information campaign: it showed that the program had positive impacts on knowledge around the SFP, institutional trust, business-related decision-making, and sales. In particular, the digital information campaign increased the share of women that report knowing key aspects about the SFP, such as the products traded in the SFP market (9 percent) and that they can register as providers (21 percent). The effects were larger for women not reached by traditional extension programs. Figure ES1. Treatment effect on information intake about the School Feeding Program 1.0 0.9 Proportion of women 0.8 0.7 0.6 0.573 0.5 0.4 0.818* 0.750 0.687*** 0.3 0.498* 0.2 0.411 0.1 0.238*** 0.138 0.0 Knows SFP buys Knows that can Knows the steps Knows which products from local farmers register as a provider to register the schools buy Control Treatment In addition, among partnered women, the intervention increased participation in the decision-making process to sell animal products (12 percent), as well as the likeliness to sell chicken meat (10 percentage points) and eggs (15 percentage points). These results are likely a lower bound of the effect of the intervention in light of the COVID-19 pandemic – as mobility restrictions and health constraints, together with a reduced pool of products demanded by the SFP, may have presumably had a negative effect on women’s sales overall. Finally, the information campaign increased trust in some government institutions participating in the SFP, such as the Ministry of Agriculture (5 percent) and the Tax Authority (11 percent). 13 DIGITAGRO 2022 Figure ES2. Treatment effect on sales of SFP animal products, by marital status Partnered Single 0.7 0.6 0.5 0.486*** Proportion of women 0.4 0.366 0.332 0.343 0.3 0.2 0.164*** 0.1 0.066 0.073 0.065 0.0 Eggs Chicken meat Eggs Chicken meat -0.1 -0.2 Control Treatment Although the information campaign had significant impacts on various outcomes, it did not make women more likely to join the SFP market. The impact evaluation study included quantitative and qualitative analysis to shed light on the barriers that prevent women from entering the SFP. Figure ES3. Treatment effect on willingness to participate in the SFP 0.5 0.4 Proportion of women 0.4 0.3 0.3 0.2 0.2 0.326 0.316 0.1 0.193 0.205 0.1 0.0 Interested in registering Sold crops to a registered SFP provider Control Treatment 14 DIGITAGRO 2022 Barriers to joining the SFP and the suitability of digital technologies to address them In order to be able to join the SFP, women agri-preneurs need to overcome a diverse set of barriers (Table ES1). First, they must be aware of the SFP as a business opportunity. Second, they need to fulfil at least three individual- level prerequisites: (i) to be empowered and able to take business-related decisions; (ii) to have the production capacity and portfolio diversity needed to cater to the SFP; and (iii) to have the managerial and technical skills to trade in the SFP market. Third, at the market and structural level, prices need to be competitive, and the institutions involved need to be trusted by all parties involved. Finally, the registration process needs to be simple and understood by prospective candidates. Awareness The information campaign and its impact evaluation showed the potential for digital technologies to raise awareness about the SFP and knowledge about this market. Empowerment The information campaign also increased the participation of women in the decision-making of some agricultural products and their sales. However, further interventions are needed to address women’s empowerment more broadly. In the agricultural sector, for example, the promotion of women’s association, as well as women’s access to land ownership and formal rights can impact intrahousehold decision making and entrepreneurship. Production capacity, portfolio diversity, and skills The analysis also suggests that production capacity, portfolio diversity, and technical and managerial skills are important barriers for women to join the SFP. Households that already participate in the SFP have much higher production capacity than non-registered households. For instance, relative to non-registered farmers, registered providers have more than twice the amount of land, cultivate a higher diversity of products demanded by the SFP, and use more productive techniques, such as specialized inputs, machinery, and infrastructure. In addition, registered providers need managerial skills to pay taxes and deal with the accounting of providing to schools. While digital technologies can certainly help increase technical and managerial skills through e-extension, well-targeted technical assistance is a necessary complement to ensure deeper human capital formation in rural settings. A package of interventions that address more than one barrier at a time, such as information, skills, mentorship, access to productive inputs, and the promotion of rural producers’ groups might be more successful than interventions addressing a single barrier. A promising approach in this sense could be an adaptation to the SFP reality of the Productive Alliances model, implemented in several countries in the LAC region and worldwide, to facilitate the links of producers’ groups with commercial partners and technical assistance for productivity enhancement and business development. 15 DIGITAGRO 2022 Market-level and structural conditions In addition, the analysis shows that conditions such as prices and trust in the institutions involved in the SFP, limit the SFP’s ability to recruit more providers. An analysis of agricultural and animal product transactions suggests that the SFP reference prices are aligned with the average prices women report obtaining in traditional markets. The SFP might need to offer higher prices to producers to account for costs directly incurred for participating in the program, such as food quality standards, taxation, transport to schools in remote areas, cost differences between local varieties of staple crops in different climatic zones, and others. Agrologistics improvement would also help addressing some of the logistic challenges faced by producers, for example in terms of transport. Furthermore, while the information campaign resulted in a mild increase in overall institutional trust, trust issues still appear large, especially as regards the national Tax Authority (SAT): persistent informality in rural areas causes farmers to have never had any contact with this institution before, and producers believe that taxation might rip out the benefits from selling to schools. Registration process Finally, policymakers should consider simplifying the registration process further. The impact evaluation showed that more women know the process to register thanks to the information campaign, but very few consider it a simple process. To register as a provider, agri-preneurs need to be classified as small commercial producers in the MAGA’s Family Agriculture Registry and then register with the SAT to have a tax identification number to issue invoices: this is considered at first sight too complicated and costly by small producers, who often feel they would need to hire expensive professional accountants to follow-up with the required procedures. When they are assisted by field extensionists, however, producers report finding the process rather easy. Close assistance from field extensionists, as well as a more streamlined registration process with one-stop-shops at the local level and online, might help promote SFP registration. 16 DIGITAGRO 2022 Table ES1. Barriers to SFP participation, and suitability of digital technologies to address them Main barriers Potential for Digital Technologies to overcome barriers First Step Program The information campaign raised awareness about the SFP as a business awarness opportunity. It also increased knowledge about the SFP. Individual-level barriers Empowerment and The information campaign increased decision-making and sales for some entrepreneurial women. But additional gender gaps need additional interventions. spirit E-commerce can lower transaction costs for schools of dealing with multiple Production producers, but smallholders might just not have the production capacity capacity & throughout the year to register as SFP providers. Improving coordination portfolio among producers and promoting producers groups would help them leverage economies of scale and scope. Technical and E-extension can help smallholders acquire more skills and adopt better managerial skills agricultural practices, but close technical assistance ensures stronger results. Market and structural-level barriers While the e-commerce platform will ensure price transparency and improve Price competition in the SFP market, the official SFP price formation mechanism depends highly on political considerations. The information campaign increased trust in some institutions involved in the SFP, and more accessible, standard and regular information via digital means Institutional trust could further improve trust. However, a particular effort will be needed to emphasize the benefits of formalization in highly informal rural settings. Final Step The information campaign increased knowledge about the registration process, but close assistance from field extensionistsis is needed to promote SFP Registration registration. One-stop-shops at the local level and on-line could also help streamline the process. High potential for digital technologies Potential for digital technologies in combination with other measures Limited potential for digital technologies if underlying constraints are not removed 17 DIGITAGRO 2022 Lessons learned The findings of the impact evaluation study suggest that continued work is needed to keep aligning the structure of the SFP to the reality of smallholder producers and women. In the current setting, the program seems better suited to comparatively larger producers, with more sophisticated production systems and a higher capacity of supplying a broader pool of products throughout the year. In addition, women still appear to need the support of their families to establish themselves as successful agri-preneurs in the SFP, and low levels of trust in the main institutions involved in the program make such informal mechanisms even more pervasive. Overall, these results call for strengthened and well-targeted extension services and technical assistance to accompany producers in their transition to more commercial agriculture, with an eye to the specific learning needs of women and of other vulnerable populations such as Indigenous Peoples. Thanks to their ability to penetrate among broad audiences and to reduce the cost of remoteness, digital technologies can support this agenda, but to be effective at scale they require a suitable conducive environment in terms of infrastructure, regulatory environment, and human capital. In Guatemala as in many other countries around the world, investing in digital development, strengthening the regulatory environment, increasing rural connectivity and mobile penetration, and promoting digital literacy and skills will have high payoffs. 18 Chapter 1 Introduction 19 DIGITAGRO 2022 Introduction Despite the strong role played by the agri-food sector in Guatemala’s economic performance and employment, reflected in high exports and strong results by larger commercial agri-businesses, small producers face daunting levels of market access, revenue generation capacity, and resilience. At the same time, Guatemala displays worrying food and nutrition outcomes, with the fourth highest rate of chronic malnutrition in the world (the highest in Latin America and Caribbean) and with only 40 percent of Guatemalan families enjoying food security. To contribute towards tackling child malnutrition while supporting the livelihoods of smallholder farmers, Guatemala’s School Feeding Law, approved in 2017, tripled (and later quadrupled) the previous national budgetary allocation for school meals, and mandated that 50 percent of direct food purchases by public schools come from local smallholder farmers. The budgetary increase will help guarantee more nutritious meals and more adequate portions and combinations for schoolchildren aged 6 to 12, based on their specific calorie, protein, and micronutrient requirements. At the same time, the law aims at empowering the country’s close to 2.5 million smallholder farmers, promoting linkages between local food production and more than 33,000 public schools nationwide. Schools in remote areas, however, often lack information on which producer to buy their food from, as well as basic knowledge on safe and hygienic cooking practices. At the same time, unclear administrative procedures discourage many local producers from participating in the school feeding market. Those producers who do, moreover, frequently struggle with low production capacity, and with the fact that schools do not regularly plan their menu and purchase schedule ahead of the school year. These challenges are further exacerbated for women producers, who face higher information gaps, lower market access, and higher informality than their male counterparts, compounded by restrictive social norms and disempowerment. Yet, women who are engaged in agriculture have ample potential to be engaged in the school feeding business, as they tend to specialize in the production of foods that are in high demand by schools (such as seasonal fruit, vegetables, eggs, cheese, poultry). The School Feeding Program (SFP) thus represents a crucial window of opportunity for rural women in Guatemala, and a vehicle for their evolution from invisible farmers to proper agri-preneurs – economic agents in their own right in the agribusiness space. Information diffusion through digital technologies can increase market participation in rural areas and holds promise to enhance the status of women in the business sphere. The World Bank’s DIGITAGRO project, funded by a grant of the InfoDev Trust Fund, piloted digital technologies to improve market access for women agri- preneurs, so they could supply the School Feeding Program in a fair, safe, sustainable, and profitable way – while helping schools improve children’s nutrition. In particular, the intervention was built around three main activities, which were all based on the provision of information and knowledge to support the productive participation of women in the SFP. These activities included: (i) the creation, in partnership with the World Food Programme (WFP), of an e-commerce platform to link the supply (family agriculture) and demand (schools) for food; (ii) the production, with the Food and Agriculture Organization of the United Nations (FAO), of extension videos targeted to women agri-preneurs; and (iii) a digital information campaign, conducted in collaboration with the World Bank’s Gender Innovation Lab for Latin America and the Caribbean (LACGIL). The digital information campaign was complemented by an impact evaluation study, to assess the efficacy of information diffusion through digital technologies towards increasing market participation and the status of rural women in the ambit of the School Feeding Program. The purpose of this report is to describe the DIGITAGRO project and to present the findings of the impact evaluation study on the information campaign, in order to derive lessons on the use of digital technologies to promote market access for rural women, with a specific focus on their inclusion in Guatemala’s School Feeding Program. The intervention increased knowledge about the SFP, especially for women not reached by traditional extension programs, prompted some women to increase their sales, and had some effect on intra-household 20 DIGITAGRO 2022 decision-making. The results of the study suggest that digital technologies can be used in rural Guatemala to convey information about the SFP and encourage women to participate in agribusiness, but that, nonetheless, the SFP still faces challenges that hold small producers back. Barriers such as product mismatches, production capacity, and institutional trust seem to be hampering women’s direct involvement in the SFP, calling for targeted complementary extension services and a better alignment of the SFP structure with local market conditions. This report is intended to reach a broad audience of policy makers, program administrators, development professionals, and academics in Guatemala and in the broader development community. The contents of this report are based on information derived from a wide range of sources. Preparation of the report involved the setting up and analysis of a field experiment conducted among 881 eligible women in 271 villages in Western Guatemala at the height of the COVID-19 pandemic, data analysis from secondary sources, interviews with key informants from the public and private sector as well as development partners, several focus groups with local stakeholders, and literature reviews. The rest of the report is organized as follows. Chapter 2 provides an overview of family farming in Guatemala, including an assessment of the gaps encountered by rural women, and highlights child nutrition issues in the country. Chapter 3 describes the School Feeding Program, highlighting its functioning, the main actors involved, its expected benefits and the challenges it faces. Chapter 4 presents the DIGITAGRO project, providing a rationale for the use of digital technologies in agriculture, describing the main activities of the project, and providing details on the set-up of the impact evaluation study. Chapter 5 presents the experimental setting and main findings of the impact evaluation, whereas the potential mechanisms that could be driving the results are explored in Chapter 6, together with recommendations for promoting participation in the School Feeding Program. Chapter 7 discusses lessons learned and concludes. 21 Chapter 2 Family agriculture, child nutrition, and women’s (dis-) empowerment in Guatemala: three related challenges 22 DIGITAGRO 2022 Family agriculture, child nutrition, and women’s (dis-) empowerment in Guatemala: three related challenges Key messages » Family farming is a key pillar of rural Guatemala’s economy and in food and nutrition security. However, family farmers face several challenges: lack of information, insufficient access to markets, weak infrastruc- ture, low access to productive assets, land degradation, postharvest losses, climate change. These challen- ges increase transaction costs, food price inflation, and market access and growth. » Women play an important role in Guatemala’s agriculture sector, but gender inequalities and discrimina- tion put them at a disadvantage. Rural women face low levels of empowerment, high invisibility as econo- mic subjects, and lower access to productive resources such as land. The latter hampers their access to financial resources and formal participation in agricultural associations, and gives them lower exposure to agricultural extension services. For Indigenous women, these barriers are even more acute. » Despite the centrality of the agriculture sector in Guatemala’s economy, the country experiences severe food and nutrition security challenges, among the highest in the world. Food availability is greatly affected by climate change. » The COVID-19 pandemic has disproportionately affected small-scale producers. The slowdown of econo- mic activity and mobility restrictions hindered their interaction in marketplaces, and, due to widespread informality, only few were eligible for government’s emergency programs targeted to firms. The pandemic also worsened Guatemala’s food and nutrition security conditions. 2.1 Family Agriculture Agriculture is one of the most important economic activities in Guatemala. The country has 7.3 million hectares of land (equivalent to 67.5 percent of national territory) used for agriculture.(1) 12 percent of this land is dedicated to the cultivation of staple products (maize, beans, tomatoes, rice, and potatoes),(2) while 14 percent to coffee, sugar cane, rubber, and cardamom.(3) Agricultural activities predominately take place in the open fields. Therefore, cultivation outputs are heavily influenced by fundamental elements of the production process that include biophysical factors, inputs, and agricultural practices and managements techniques. The agricultural sector’s contribution to Guatemala’s GDP has accounted for an average 10.7 percent over the last six years.(4) The sector also contributes to the generation of employment and income, employing approximately 1.8 million people (30 percent of the population). Agriculture plays an even greater role in the rural areas of the country where, simultaneously, 48 percent of the Guatemalan population resides, and where the majority of the 1 Ministerio de Agricultura, Ganadería y Alimentación, 2016a. 2 World Bank, 2020a. 3 Herrera et al., 2018. 4 Ministerio de Agricultura, Ganadería y Alimentación, 2016a. 23 DIGITAGRO 2022 Indigenous population also resides. According to the 2014 National Survey of Living Conditions (ENCOVI) results, the poverty level in these areas of the country is 76.1 percent versus 42.1 percent in the urban areas.(5) Within this context, agriculture serves as the main avenue for employment and subsistence for rural communities as it employs 31 percent of the rural population.(6) The family farming system, in particular, plays a significant role as a supplier of food for the entire Guatemalan population, as a generator of employment and source of income for rural communities, and contributor to the development of rural communities. Subsistence farming— the provision of agricultural and livestock products for famers and their families in rural communities— is the main function of family farming. This sector of rural agriculture also generates a surplus of the country’s cultivated staple products (corn, beans, rice), native and exotic vegetables, as well as eggs, milk, and meat, though on a smaller scale.(7) According to Guatemala’s 2016 National Income and Employment Survey (ENEI), 48.3 percent of the rural population—approximately 1,433,850 workers— is dedicated to agriculture, livestock, and forestry activities.(8) Other value chains that form part of the family farming sector include aquaculture, artisanal fishing, artisanal crafts and agrotourism, livestock farming and forestry. In Guatemala, the Ministry of Agriculture, Livestock, and Food (MAGA) classifies family farming in three subgroups: infra-subsistence, subsistence, and commercial. Table 1. Criteria for classification of households in family agriculture Criterion Infra-subsistence Family Subsistence Commercial Agriculture Family Agriculture Family Agriculture Capital/land <0.5 manzana 0.5-1 manzana 1-5 manzanas Family labor in agriculture Head of household Head of household Head of household participates 50-70% participates 70-100% participates 100% Head of household Woman/man Woman/man Woman/man > 50% agricultural income < 50% > 50% > 80% Receive remittance Yes/no Yes/no Yes/no Source: Ministerio de Agricultura Ganadería y Alimentación, 2016b. » Infra-subsistence farming households are those that are unable to produce the necessary amount of their food during the year. They have less than 0.5 manzanas of land(9) for farming and do not hire external labor.(10) The harvest from infra-subsistence farming is dedicated to self-consumption. However, annual harvests never meet household food demands. » Subsistence farmers dedicates their harvest to self-consumption. These households have tracts of land between 0.5 manzanas and 1 manzana.(11) While this type of farming does not hire external labor, households in this group have created stable production systems and have consolidated their productive 5 Instituto Nacional de Estadística (INE), 2015. 6 World Bank, 2020a. 7 Ponce and Arellano, 2017. 8 INE, 2017: ENEI 2016. 9 1 manzana = 10,000 square meters. 10 Ministerio de Agricultura Ganadería y Alimentación, 2016b. 11 Ministerio de Agricultura Ganadería y Alimentación, 2016b. 24 DIGITAGRO 2022 practices,(12) allowing for the growth in production performance and the diversification of crops. According to the MAGA’s Family Farming Program to Strengthen the Rural Economy (PAFFEC), subsistence farming is capable of meeting almost 80 percent of self-consumption demands. (13) » Commercial farms possess between 1 and 5 manzanas of land for harvesting their products.(14) Farmers in this group hire laborers for certain tasks such as fertilization or harvesting. Within this structure, they can sell a portion of their products while also allocating part of their harvest for self-consumption. While family farming is a key pillar of rural Guatemala’s economy as well as its food and nutrition security, it faces several crucial challenges. From lack of information, low access to markets, weak infrastructure, low access to productive assets, to land degradation, postharvest losses, and climate change. At one level, smallholder farmers lack crucial market information. For instance, although the MAGA publishes weekly prices for agricultural products, this information is not easily available to all farmers due to remoteness, poor connectivity, and broad accessibility issues. Furthermore, Guatemala’s quality of infrastructure varies greatly across the country, which makes several rural communities isolated due to rugged landscapes and underdeveloped transportation infrastructure. Among other challenges, this constrains information flows between buyers and producers, generating high transaction costs that stifle agri-business growth and productivity as intermediaries (“coyotes”) ultimately capture the majority of profits. Poor connectivity infrastructure also results in high food price inflation, particularly for rural areas that have less accessibility to markets, and leads to high levels of food loss from spoilage and damage due to inadequate roads or long routes. Ultimately, rural farmers’ opportunities to gain access to Guatemala’s growing domestic and export markets are limited, as is their ability to increase sales and improve their welfare. Due to inefficient storage and preservation facilities, moreover, Guatemala also faces the challenge of substantial food losses. According the United Nations Environment Programme (UNEP) 2021 Food Waste Index, household food waste estimates reached 72 (kg/capita/year) / 1,273,466 (tonnes/year).(15) Due to poor post-harvest management, a significant amount of Guatemala’s agricultural produce is ultimately deemed spoiled and unable to sell: food loss and waste stands at approximately 38 percent of the country’s total production annually, with tomatoes reaching loss rates of 54 percent, maize 34 percent, and beans 26 percent.(16) Among the most prevalent causes of post-harvest losses, the FAO reports that more than 35 percent of beans and maize in Guatemala are spoiled due to human damage at selection, followed by plagues and animal (rodents) damage as well as labor damage at harvest.(17) Access to productive assets such as savings, credit, and insurance is also limited for rural producers. Such products have the potential to increase farmers’ resilience by providing them with a buffer in case negative events occur and giving them the ability to invest in risk-reducing technologies (i.e. improved storage facilities) that can prevent losses of food and income. Nevertheless, small farmers have a rate of access to finance that is 13.9 percent lower than the national average.(18) Prohibitive costs for many of such products, such as high premiums for agricultural insurance and the lack of suitable collateral to back formal loans (mostly due to land tenure challenges) prevent smallholder farmers from participating in the formal banking system and in risk-reducing schemes, together with challenges related to data availability, product design, cost, and distribution. In 2017, only 41 percent of the rural population older than 15 possessed an account with a financial institution, and only 10.5 percent reported having borrowed from a financial institution.(19) Furthermore, smallholder farmers are particularly vulnerable to climate change, which is compounded by the 12 Ministerio de Agricultura Ganadería y Alimentación, 2016b. 13 Ministerio de Agricultura Ganadería y Alimentación, 2016b. 14 Ministerio de Agricultura Ganadería y Alimentación, 2016b. 15 United Nations Environment Programme (UNEP), 2021. 16 World Bank, 2020a. 17 FAO, 2019. 18 World Bank, 2020a. 19 World Bank, 2017: The Global Findex Database. 25 DIGITAGRO 2022 adoption of unsustainable agricultural practices, inadequate infrastructure, and high rates of food loss and waste. Guatemala ranks 16th in the world among other countries that are at most risk from climate events such as floods, droughts, and temperature variation.(20) Projections indicate that by 2050, climate change will induce a temperature increase of between 2.5° and 4°C, extension of semi-arid climate regions, longer dry spells, irregular rainfall that would equate to more droughts and floods.(21) These climate-related events threaten the viability of staple crops such as maize and beans, that are projected to experience a 14 percent decrease in yields by 2050.(22) Additional side-effects include greater risk of pest infestation and crop disease, loss of agricultural land, and crop suitability. According to Guatemala’s 2015 Intended Nationally Determined Contribution (INDC) report, by 2050 economic losses—from the agriculture sector alone—from drought, flooding, and other extreme weather events are projected to reduce GDP by 1.3 - 3.7 percent.(23) In November 2020, Guatemala was hit by two hurricanes—Eta and Iota—within a two-week period. These natural disasters affected an estimated 286,000 farmers in the rural regions of the country.(24) Preliminary damage assessments from the United States Department of Agriculture (USDA) reported that of the 165,000 hectares affected by the storms, 101,000 hectares belonged to smallholder farmers.(25) Increase in food prices, particularly in the local markets, reflected the reduced food availability due to the impacts of the storms. Additional impacts materialized in 2021 through lower volume of harvests resulting from floods, soil degradation, and infrastructure losses. Box 1 Guatemala’s National Rural Extension System The National Rural Extension System (Sistema Nacional de Extensión Rural, SNER) has three specific objectives: 1. To train agricultural producers in the adoption of environmentally responsible agricultural processes, based on technological innovation and the sustainable use of natural resources. 2. To create and consolidate the capacities of formal organizations so that agricultural producers can im- prove the production, marketing, and competitiveness of their products, and access credit and financial services. 3. To promote and strengthen the involvement of municipal governments in the development of local communities. The SNER is organized through the Rural Extension Directorate of the MAGA, where the Department of Rural Extension has strategic, coordination and supervision functions. The services provided by the mi- nistry are channeled through Departmental Coordinations, while the Municipal Rural Extension Agencies (Agencias Municipales de Extensión Rural, AMER) are the operational unit of the rural extension service at the local level. In each municipality, the AMER is made up of three people: (i) Extension Agent for Rural Development, who 20 Eckstein et al., 2020. 21 USAID, 2017. 22 World Bank, 2020a. 23 Gobierno de Guatemala, 2015. 24 Tay, 2020. 25 Tay, 2020. 26 DIGITAGRO 2022 coordinates the team and trains and assists the other extensionists, applying a comprehensive rural develo- pment approach; (ii) Extension Agent for Family Farming, who focuses on subsistence and infra-subsistence farmers; and (iii) Extension Agent for Healthy Households, who focuses on improving nutrition and equity in rural families. At the family farming level, rural families are trained through participation in Learning Cen- ters for Rural Development (Centros de Aprendizaje para el Desarrollo Rural, CADER). A CADER is composed of a self-selected group of at least ten rural families, a volunteer promoter (a lea- der agricultural producer selected and appointed by the community who acts as coordinator-faci- litator), and a menu of technological innovations promoted by the SNER, aimed at increasing agri- cultural productivity, the improvement of rural households, and the protection of natural resources. The technological innovations are installed in the different plots and homes of the members of the group, with the purpose of practical training and development of the self-management capacity of members. Fo- llowing a “farmer-to-farmer” approach, the promoter, who receives special training from extension agents, develops activities of demonstration and transfer of agricultural knowledge, home improvement and rural development with the community group. As regards the promotion of formal organizations of agricultural producers, the actions are concentrated on (i) supporting the strengthening of existing formal organizations, in order to achieve their profitability, sustainability and self-management; and (ii) supporting and facilitating the formalization of existing producer groups when they are not yet formally organized. The “graduation” from informal to formal requires that organizations have consolidated their capacities to obtain access to competitive financial services, formally market their products, and manage their accounting and tax procedures. Source: Ministerio de Agricultura, Ganadería y Alimentación, 2013. 2.2 Women in Rural Guatemala While women play an important role in Guatemala’s agriculture sector, gender inequalities and discrimination put them at a disadvantage in comparison with their male counterparts. Guatemala ranks among the lowest in the United Nation’s Development Programme’s (UNDP) Gender Inequality Index, standing at 127th out of 189 countries.(26) Gender gaps exist in all sectors in the country. These gaps are particularly experienced in decision- making relating to the family and community, social and political participation, obtaining access to resources, the uneven distribution of domestic work, and high exposure to gender-based violence. Levels of disempowerment are more acutely felt amongst rural women, and, given Guatemala’s historical exclusion of Indigenous populations, these gaps are further intensified for rural and Indigenous women creating a system of multi-dynamic structural discrimination and inequality. As an example, according to the 2017 project-level Women’s Empowerment in Agriculture Index (Pro-WEAI), in the Polochic Valley and in the Western Highlands, 80 percent to 90 percent of women are disempowered, and an Indigenous woman is three times as likely to be disempowered as the average woman participating in the agriculture sector.(27) One of the main limitations women face in the agriculture sector is invisibility. Women formally constitute 26 United Nations Development Programme (UNDP), 2020. 27 Garbero and Perge, 2017. 27 DIGITAGRO 2022 approximately 10 percent of agricultural employment, but many more execute other agriculture-related activities.(28) Considering the women that participate in unpaid farm work and/or supporting their spouses on the family farms, the portion of women in agriculture reaches almost 40 percent.(29) This discrepancy is largely due to cultural and social norms, which tend to categorize women as secondary agricultural workers or helpers to their husbands, fathers, or brothers, and not as economic subjects or potential entrepreneurs in and of themselves. In addition, women’s involvement in agriculture has been steadily increasing because of male outmigration. According to Guatemala’s 2018 national census, of the total emigrants recorded for that year, 77.5 percent (242,203) were men compared to 22 percent (54,447) that were women.(30) As a result, a greater number of rural women are left behind to take over the activities of their migrant husbands. However, this change in status is seldom reported to the authorities, for fear of repercussions for illegally-migrating husbands. Informality, however, disconnects women from public support systems, and increases their vulnerability to shocks and their dependence on remittances. Of the women that are employed in agriculture, 57 percent do not earn a salary, and of those that do, 97 percent earn below the minimum wage.(31) The 2018 Guatemala National Census (INE) shows that only 53 percent of women in agriculture can read and write, versus 71 percent of men. Women also face barriers in access to productive assets and credit, human capital, and opportunities for entrepreneurship. All together, these factors make rural women and their families more vulnerable to poverty and food insecurity. Gender inequalities in agriculture also limit the development of the sector. According to the Food and Agriculture Organization of the United Nations (FAO), if women worldwide had the same access to productive resources as men, production on their farms would increase from 20 percent – 30 percent.(32) Among the productive resources that rural women in Guatemala have lower access to, land is paramount, as only 7.8 percent of landowners are women.(33) Rarely do families allow women to inherit land. Cultural and traditional norms dictate that while women will live on the land of their husbands, only the men of the family need the land to cultivate. In most cases, a women’s father or husband—sometimes a son— becomes owner of the land title. On the institutional side, the percentage of women in land access programs has also been historically low. While Guatemala has a land fund (FONTIERRAS), whose regulating law formally establishes the inclusion of women on equal terms as men, mechanisms to access land and rules on land ownership and co-ownership by women perpetuate gender gaps: by 2014, only 10.7 percent of women versus 89.3 percent of men benefited from the land access program established by FONTIERRAS,(34) while in 2016 just 29 percent of the beneficiaries of credits and subsidies for buying land were women.(35) Lack of land ownership implies less access to credit, as most women cannot use land as collateral for a loan. According to 2016 estimates, only 28.7 percent of women compared to 71 percent of men benefited from credits and subsidies for land purchases.(36) In terms of credit and subsidies for land leases for production, 33 percent of women had access to these assets compared to 66 percent of men.(37) Besides land tenure, inadequate guarantees, lack of information relating to available credit options and application procedures, fear of rejection, and lack of effective strategies by financial institutions to assist women to understand their processes and procedures contribute to lower access to credit by women farmers. Low credit, in turn, translates into lower 28 FAOSTAT, 2017. 29 INE, 2019: 2018 Guatemala Census. 30 INE, 2019: 2018 Guatemala Census. 31 INE, 2015: ENEI, 2014. 32 FAO, 2011. 33 FAO, 2016. 34 Moreno, 2020. 35 Moreno, 2020. 36 FAO, 2017. 37 FAO, 2017. 28 DIGITAGRO 2022 productive investment. As a result, credit-seeking women are often relegated to invest in smaller-scale activities than men, which creates an artificial specialization of tasks along gender lines. While both men and women perform several tasks along agrifood value chains, men are generally more involved in higher-value-added tasks requiring more capital investment along the transformation chain. Inadequate access to land compounded by cultural norms and gender stereotypes is also an obstacle for formal participation of women in agricultural associations, cooperatives, or rural producers’ organizations. According to the United States Agency for International Development’s (USAID) 2018 Gender Analysis, only 7.8 percent of landowners were women.(38) As land ownership is often a condition to gain membership in associations, most women are either automatically denied or discouraged from joining. When women do join an association, internal rules and cultural barriers (as women are not viewed as direct participants, but rather as wives of associates) often limit their full participation, and women do not have complete access to credit and other benefits to the same extent as men.(39) The fact that, culturally, women are not considered farmers but at most helpers to their male relatives stifles female agricultural entrepreneurship in several ways, including through lower exposure to agricultural technical assistance. According to the MAGA, only 5 percent of women farmers received technical assistance.(40) Furthermore, when agricultural extension services reach women, in practice, traditional approaches are replicated that predominantly assign women to the family farming sphere. This limits women’s potential for greater specialization and opportunities for economic advancement, and hinders their knowledge of productivity-enhancing, climate- smart agricultural practices that could improve their food security, livelihoods, and resilience. Indigenous women face additional barriers to access technical assistance, especially those who live in remote rural areas, as many extension services are only provided in Spanish and not in native languages.(41) 2.3 Food and nutrition security Despite the centrality of the agriculture sector in Guatemala’s economy, the country’s food system faces a precarious situation. In 2018, hunger and undernutrition cost Guatemala more than US $8 million per day, which accounted for around 10 percent of the country’s GDP.(42) By 2019, Guatemala was ranked 68 out of 113 countries on the Global Food Security Index (GFSI) and 72 out of 117 qualifying countries in the 2019 Global Hunger Index (GHI), and suffered from a level of hunger that was considered serious. Only 40 percent of the country’s families experience food security, meaning that they can meet essential food and non-food subsistence needs.(43) Meanwhile, 9 percent of families experience slight food insecurity, another 31 percent face moderate food insecurity, and 20 percent are classified to be in situations of severe insecurity.(44) Guatemala’s population is projected to increase by 33 percent from 17.9 million to 26.9 million over the next 30 years,(45) which will increase pressure on the country’s food system. Guatemala has the fourth highest rate of chronic child malnutrition in the world and the highest in Latin America and the Caribbean.(46) According to the most recent national survey on maternal and child health (ENSMI), 47 percent of all children under the age of 5 were stunted in 2014-2015, and the percentage increased to 66 percent 38 Herrera et al., 2018. 39 Moreno, 2020. 40 FAO, 2017. 41 Moreno, 2020. 42 Prost and Martínez, 2020. 43 World Bank, 2020a. 44 World Bank, 2020b. 45 World Bank, 2020a. 46 FAO, 2019. 29 DIGITAGRO 2022 for Indigenous children in the lowest income quintile.(47) Levels of malnutrition tend to be even higher for rural and Indigenous children. Chronic malnutrition affects up to 53 percent of rural children versus 34.6 percent in urban areas, and 58 percent of Indigenous children compared to 34.2 percent of non-Indigenous children.(48) In the same year, 12.6 percent of Guatemalan children under the age of 5 were underweight, more than 4 times the average for the LAC region.(49) Poor environmental health (with more than 30 percent of households having no access to safe sanitation)(50) and lack of awareness of good nutritional habits and child feeding practices contribute to malnutrition. In 2014-2015, for children between 6-23 months, only 52.1 percent received the minimum acceptable diet that meets the minimum level of diversity and the minimum number of meals per day as per their age group.(51) Due to poor or inadequate diets, Guatemala is one of the three main countries with the highest rates of nutritional deficiencies in the LAC region. Guatemala’s 2020 Nutrition Smart Agriculture Profile reports that 77 percent of families have an inadequate diet that is based on overconsumption of cereals and underconsumption of animal-based products, fruits, and vegetables.(52) As a result, Guatemala ranks low on various health and nutritional indicators among other countries in Latin America and the Caribbean. Iron (32 percent) and zinc (34.9 percent) deficiencies are highest in children under five years old.(53) Food availability in Guatemala is greatly affected by climate change. Guatemala ranks 62nd in the 2021 Global Climate Risk Index(54) with climate risk index score (CRI) of 65.67.(55) The lower the index score, the higher the climate risk countries face. Between 2010-2020, Guatemala experienced 48 occurrences of natural disasters that jointly caused 1,408 deaths, affected 10.5 million people, and amounted to more than US$ 2.2 billion in total estimated damages.(56) Natural and climatic hazards have resulted in reduced harvests or the complete destruction of staple crops as well as of critical agrifood infrastructure, severely jeopardizing the livelihoods and the food and nutrition security of the poor and vulnerable. For example, in November 2020 hurricanes Eta and Iota had devasting effects in terms horticultural losses. Preliminary assessments reported an estimated 24,000 hectares of beans, 630 hectares of coffee, 420 hectares of commercial rice, and 14,500 hectares of flooded palm oil plantations. (57)These conditions exacerbate existing risks and vulnerabilities such as malnutrition and hunger, that are also linked to land atomization and low agricultural productivity. The production of coffee, sugar cane, and palm oil have taken over 44 percent of land that is suitable for staple crops cultivation.(58) Meanwhile, 70 percent of the Guatemala’s food supply comes from smallholder farmers that must operate within small farms that average 0.5 hectares in size, resulting in low crop yields.(59) This situation is even worse for rural Indigenous populations. Overall, although 70 percent of food supply is derived from small farmers, smallholders barely produce enough food for subsistence, and on average their 47 USAID, 2018. 48 USAID, 2018. 49 World Bank, World Development Indicators (WDI). 50 World Bank, 2020b. 51 Ministerio de Salud Pública y Asistencia Social (MSPAS), Instituto Nacional de Estadística (INE), and ICF International, 2017. 52 World Bank, 2020b. 53 Moreno, 2020. 54 The Global Climate Risk Index indicates a level of exposure and vulnerability to extreme weather events. 55 The CRI 2021 is based on the loss figures of 180 countries from the year 2019 and the period 2000 to 2019. This ranking represents the most affected countries. In each of the four categories (number of deaths; number of deaths per 100,000 inhabitants; sum losses in US$ in purchasing power parity (PPP); losses per unit of gross domestic product (GDP) ranking is used as a normalization technique. Each country’s index score has been derived from a country’s average ranking in all four indicating categories, according to the following weighting: death toll, 1/6; deaths per 100,000 inhabitants, 1/3; absolute losses in PPP, 1/6; losses per GDP unit, 1/3 (Eckstein et al., 2021). 56 EM-DAT, 2021. 57 Tay, 2020. 58 World Bank, 2020b. 59 World Bank, 2020b. 30 DIGITAGRO 2022 produce only provides coverage for 4-6 months of household consumption, inducing a dependence on commercially-sourced products.(60) 2.4 COVID-19 In addition to long-standing challenges in terms of stagnating productivity, poverty and inclusion, climate, and food and nutrition insecurity, rural Guatemala is currently still grappling with the impacts of the novel coronavirus (COVID-19) pandemic. The country was the second hardest hit in Central America. According to the World Health Organization’s (WHO) Coronavirus (COVID-19) Dashboard, as of September 2021, there have been 537,987 cases of COVID-19 and 13,185 deaths.(61) Government authorities issued a state of emergency on March 5, 2020, and implemented strong containment measures that included border closures, mobility restrictions, and suspension of activities in the public and private sector. Following a robust growth of real GDP of 3.5 percent a year between 2010 and 2019, the country experienced a 1.8 percent contraction in real GDP in 2020.(62) Small-scale producers have been disproportionately vulnerable to the slowdown of economic activity and restrictions imposed on local and international mobility. For instance, small-scale farmers depended on selling their produce in marketplaces, whose hours of operation where restricted as a result of safety measures during the pandemic. In addition, most small-scale farmers are informal(63) and as such are not part of government registries used to determine the eligibility for emergency programs targeting firms. Economically, the consequences of lockdown were experienced in agricultural and non-agricultural income. Approximately three out of every four households reported a decrease in income.(64) Despite the government’s quick response with support programs, poor households were confined to and depended on limited coping mechanisms to deal with income reductions. Reduced food availability and higher food prices reduced food security and dietary diversity. According to a study conducted by the International Food Policy Research Institute (IFPRI), between the end of 2019 and mid-2020, the pervasiveness of food insecurity doubled.(65) The same study also reported that household dietary diversity experienced a small but statistically significant decrease from 6.9 to 6.4 in the Household Dietary Diversity Score (HDSS).(66) The impacts were even more evident at the rural producer level. According to an October 2020 survey conducted by MAGA in collaboration with Climate Change, Agriculture and Food Security program (CCFAS), of the 20 percent of infra-subsistence farmers surveyed at the national level, 80 percent faced food insecurity issues.(67) Meanwhile, of the subsistence farmers surveyed at the national level (approximately 24 percent), more than 50 percent faced food insecurity.(68) 60 World Bank, 2020b. 61 World Health Organization (WHO), 2021. 62 World Bank, 2021a. 63 Global Jobs Indicators database. 64 Ceballos et al., 2021. 65 Ceballos et al., 2021. 66 The Household Dietary Diversity Score (HDSS) is defined as the number of food groups consumed (ranging rom 0 to 12) in the 24 hours preceding the interview. Ceballos et al., 2021. 67 Leal et al., 2021. 68 Leal et al., 2021. 31 Chapter 3 The School Feeding Program: an opportunity for family farming 32 DIGITAGRO 2022 The School Feeding Program: an opportunity for family farming Key messages » Guatemala’s School Feeding Program (SFP) is a national initiative that aims at promoting children’s healthy eating, food access, and permanence in public schools while promoting rural economies linking family agriculture as the main source of inputs for food preparation. » As 50 percent of products to be used in school meals must originate from family farming in the school’s community or municipal jurisdiction, the SFP is a promising business opportunity for Guatemala’s small- holders. » The operation and effectiveness of the SFP are challenged by schools’ infrastructural deficiencies and in- formation gaps that prevent effectively linking schools with registered SFP providers. In addition, farmers’ low production capacity, unclear administrative procedures, and lack of economic incentives discourage many potential providers, especially women, from participating in the program. » COVID-19 exacerbated existing challenges and created additional difficulties, as schools started delivering food bags to students’ families to cook at home, restricting the list of SFP products to only non-perishable foods. 3.1 Objectives and Structure The School Feeding Program (SFP) is regulated by the School Feeding Law (SFL) Decree 16-2017, whose objective is “to guarantee school meals, promote health and promote healthy eating of the child and adolescent populations that attend public or private school establishments in order for them to take advantage of the teaching-learning process and the formation of healthy eating habits, through food and nutritional education effort and the provision of food to students during the school year”. This program is national in scope and operates in public schools at the pre-primary and primary levels in the 340 municipalities of the country. As such, the governing institution for the implementation of the SFP is the Ministry of Education (MINEDUC), which works in coordination with other Ministries and entities on issues of specific competence. According to recent MINEDUC reports, the Program serves 2.6 million students from 29,469 schools.(69) The SFP is aimed at contributing to the integral development, access, learning, school performance, and formation of healthy eating habits through food and nutrition education and the provision of a nutritious, complementary diet during the school year. First, the program seeks to contribute to children’s access and permanence in public schools with good practices relating to food and nutrition, in dignified and healthy environments, and considering organized community participation, local socioeconomic development, and inter-institutional coordination with common objectives. Given high poverty levels in the country, especially in rural areas, the SFP aims to partially overcome the deficiencies in nutritional matters that exist in most of the homes of children who attend pre- primary and primary schools. At the same time, it aims to promote rural economies through the linking of family agriculture as the main source of inputs for the preparation of healthy foods. 69 Congreso de la República de Guatemala, 2021a. 33 DIGITAGRO 2022 In practice, the key objectives of the SFP are to promote school feeding as well as health and sanitation in schools, empower small agricultural entrepreneurs, and increase access to education. In 2017, the SFL allocated resources from the national budget to feed children in schools, for a total amount three times higher than what was allocated before. For the years 2019-2021, the resources delivered to the schools were in the amount of Q.4.00 (about USD0.52) per student per day, with disbursements made to schools every 40 or 50 days depending on the provisions made by MINEDUC. This has allowed the coverage, quality, and quantity of meals served to students to be expanded. In particular, MINEDUC, in coordination with the Ministry of Agriculture (MAGA), Ministry of Public Health and Social Assistance (MSPAS) and Secretariat of Food and Nutritional Security (SESAN), has devoted special attention to the definition of varied, nutritious, safe and culturally appropriate school menus, adapted to the productive and cultural reality of the various departments of the country.(70) Each menu describes exactly how many ingredients each preparation needs, and has detailed instructions for food preparation standardizing processes and emphasizing food quality and safety. A key aspect of the SFP is that the products to be used in the preparation of school meals must originate from local family agriculture. The SFL establishes that the products delivered to schools must come at least by 50 percent from family farming producers in the corresponding community or municipal jurisdictions, while the remaining 50 percent can be purchased in local businesses, such as stores, grocery stores, or private sellers registered with the Municipal School Food Commissions. To be registered as SFP Providers, family agriculture producers must meet two fundamental requirements: (i) be classified as small commercial producers in the official MAGA Family Agriculture registry;(71) and (ii) be registered with the Superintendency of Tax Administration (SAT) to have a tax identification number (NIT) and be able to issue invoices. Groups of eligible producers can also jointly register as providers. Importantly, although providers do not have to directly produce all the food demanded by the schools, in many cases they commit to deliver it all. In such cases, providers can buy the products that they do not have from other producers in the community, and then sell these products to the school. MINEDUC and MAGA jointly work on establishing referential SFP food prices for each municipality. These prices are revised periodically (mostly with monthly frequency) based on average prices in local markets, and the price list is made public and is shared with every school. The list then serves as the basis for bargaining between schools and providers, who can deviate slightly from the referential prices but then have to commit to the negotiated price for the duration of the disbursement period. On the schools’ side, the SFL foresees an important participation of mothers and fathers of school-age children in the administration of the SFP, to extend the scope of food and nutrition education beyond the school. Most schools have Parent Organizations (OPFs), which are elected every 2-3 years by the school parents’ assembly and usually comprise four people (including a president and a treasurer) that are charged with purchasing the food and organizing the preparation of school menus – often with the support and direction of teachers and/ or the school principal. In particular, the OPF is in charge of choosing the menus that are prepared daily (either by volunteering school mothers taking turns or by cooks paid by fees collected among parents), of maintaining the direct relationship with the food provider, and to keep a number of administrative records(72) which are periodically monitored by MINEDUC officials. To strengthen the administrative capacity of OPFs, MINEDUC through its General Directorate for Strengthening the Educational Community (DIGEFOCE) delivers periodic trainings to OPFs on issues such as records, verification of purchases, invoices, cash flows, quotes. 70 In total, schools can choose among 23 menu options: 10 national menus, 3 regional menus, and 10 departmental menus. 71 Criteria for MAGA classification pertain to land size, percentage of family labor in agriculture, and percentage of family income derived from agriculture (cf. Chapter 2 for more details). Being registered in MAGA’s registry also allows farmers to receive dedicated technical assistance from MAGA municipal extensionists. 72 Specifically, each school has to keep: (i) a Minute Book, which records all the agreements relating to the organization of the PAE; (ii) a Cash Book, recording all payments made to providers and specifying all the products that purchased; and (iii) a Warehouse Book, which keeps track of all food entering the school with each purchase and of available food stocks. 34 DIGITAGRO 2022 Box 2 Reforms to the School Feeding Law (SFL) approved in September 2021 In September 2021, the Congress of the Guatemala through Decree 12-2021 introduced a number of refor- ms to the School Feeding Program, expected to come into effect for the 2022 school year. The main changes introduced by the Decree are as follows: » 50 percent increase from Q4 (USD0.52) to Q6 (USD0.78) in the daily allocation per student in the pre-pri- mary and primary levels. » Mandate for the Ministry of Public Finance to guarantee the monetary resources to comply with the SFP each fiscal year. » Expansion in coverage in year 2023 to students 3-5 years old and 12-17 years old (the latter with an allo- cation of Q4 per day). With these additions, the SFP is expected to reach more than one million students nation-wide. » Creation of school gardens to involve the entire school community in the implementation of the School Feeding Program. » Legal provision for food delivery to occur even in the event of closure of public establishments, as was the case during the COVID-19 pandemic. » Promotion of programs for the strengthening of family agriculture. » Creation of an Interinstitutional Commission for School Feeding to (i) carry out analysis and actuarial stu- dies of the program and its financing; (ii) prepare and propose budget forecasts; and (iii) perform annual evaluations of the SFP on issues such as school enrollment, students’ weight and height, budget execution, and input price. Source: Congreso de la República de Guatemala, 2021b. 3.2. Key SFP institutions As established by the SFL, the governing body of the SFP is the MINEDUC. The MINEDUC presides the national Inter-Institutional School Feeding Commission, which has a broad coordination role of the program and is composed of high-level representatives of the Ministry of Public Finance (MINFIN), MSPAS, and MAGA, as well as delegates from the Secretariat of Food and Nutritional Security (SESAN) and the General Secretariat for Planning and Programming of the Presidency (SEGEPLAN). A National Interinstitutional Technical School Feeding Commission, made up of technical representatives from the aforementioned institutions, proposes manuals, guides, plans and actions to be developed for the effective functioning of the SFP. At the local level, each of the 22 Departments of the country has a Departmental Interinstitutional School Feeding Technical Commission (CTIDAE). The CTIDAEs are coordinated by delegates of MINEDUC and see the participation of delegates from MAGA, MSPAS, SESAN, and SEGEPLAN, on the government side, as well as by the Union of Education Workers (STEG) and representatives of international organizations, like FAO and WFP. The CTIDAEs meet at least once a month to implement actions related to the SFP, which range from determining school menus to the process of disbursing resources to the OPFs. These commissions also verify food delivery to children 35 DIGITAGRO 2022 and the quality of family farming products sold to schools. In the context of the COVID-19 pandemic restriction, the CTIDAEs have been overseeing the delivery of bags of non-perishable food to children’s families (see infra). The CTIDAEs also evaluate work plans of the Municipal Interinstitutional School Feeding Technical Commissions (CTIMAEs), which oversee the implementation of the SFP at the municipal level. Box 3 Specific roles of SFP key actors MINEDUC. Governing body of the SFP, it works in coordination with other Ministries and entities on issues of specific competence. It is responsible for the regulation, planning, and coordination of all public and pri- vate activities related to the SFP. MAGA. It maintains the registry of family farmers at the national level, and provides agricultural extension and entrepreneurship services to family farmers. MSPAS. It promotes health protection actions within the SFP, specifically related to the safety of the food served to children in schools. CTIDAEs. They design and implement a departmental interinstitutional work plan for the operation of the SFP, and develop support, monitoring, and technical evaluation actions for the SFP. CTIMAEs. Their functions are similar to the CTIDAE, but at the municipal level. 3.3. Expected benefits of the SFP The potential contributions of the SFP policy are manifold and reach beyond its primary purpose of increasing food and nutrition security. These include better fed school children that are more attentive and faster learners; an education community that learns about food, nutrition, and health issues; and the strengthening of the local economy. First and foremost, the program is set to increase children’s food security by giving them access to nutrient-rich, varied foods from a reliable source. The SFP aims at directly impacting school-aged children’s and adolescent’s nutrition, by providing them with the necessary proteins, vitamins, fats, carbohydrates, and minerals for physical and intellectual development. Nutritious, safe, and sufficient school meals, in turn, are expected to help keep food-insecure children in school and to ensure their proper cognitive and physical development, with positive spillovers on their learning capacity and school performance. Furthermore, food and nutrition education coupled with tasty school meal preparations aim at contributing to change the eating habits of children and their families for the better, giving them a taste for healthy food. Relatedly, a pre-pandemic study undertaken by MINEDUC recorded positive reactions to the food served in schools, as 95 percent of students considered the meals to their liking.(73) In addition, the SFP is explicitly targeting the strengthening of the country’s small-scale agriculture industry and empowerment of rural economies. The 1,869.2 million Quetzales (the equivalent of USD 245 million per year)(74) committed by the program in purchases from family farmers opens a sizeable market for the country’s close 73 Ministerio de Educación, 2019. 74 Alvizures, 2020. 36 DIGITAGRO 2022 to 2.5 million small-scale farmers,(75) who would benefit from greater opportunities to sell crops or products of animal origin in a profitable, safe, and sustainable way, as well as to improve their productive capacities. This is even more true for women farmers, who, apart from being involved in support and commercialization activities of their families’ produce, are also in charge of keeping family gardens and raising small animals, thus producing foods (such as fruits, herbs and vegetables, eggs, chicken meat, or fresh cheese) that are in high demand for preparing school menus. In addition, since SFP prices are public and suggested by the government, all producers participating in the SFP (selling their produce either directly to schools, or indirectly to a registered provider) are granted to receive a fair, stable, and transparent price, unlike what usually happens when they sell their products in local markets or to intermediaries/coyotes. 3.4 Bottlenecks Despite its promise to tackle children’s food and nutrition insecurity and to revitalize rural economies, several bottlenecks affect the operation and effectiveness of the School Feeding Program. On one hand, school buildings are plagued by infrastructural deficiencies, especially in remote areas. According to a diagnostic undertaken in year 2019 by MINEDUC on 1,167 schools nation-wide, 88 percent of schools in the country do not count with a canteen where to serve food, only around 12 percent possess basic kitchen utensils and a refrigerator, and less than two-thirds have daily access to water for cooking children’s food (16 percent do not have access to water at all).(76) In rural areas, more than half of schools still use traditional stoves for cooking, and more than a fifth only have a dirt floor. On the other hand, a number of constraints prevent family farming production to account for the intended 50 percent of schools’ food purchases. The 2019 MINEDUC study cited above reports that almost 30 percent of schools are not linked with any registered SFP provider, and only as few as 5 percent of schools are able to meet their full food demand through family farming: in many cases, food products come from grocers or traditional merchants who are not farmers. This is due in part to information gaps on the part of schools and farmers. Schools often do not know which producer to buy their food from, nor their effective production capacity. At the same time, local producers do not know that they can participate in the SFP, and unclear administrative procedures discourage many of them from participating in the school feeding market. Smallholders lack incentives to enroll in official SFP registries, due to lack of information on accounting management, issuance of invoices and accountability, as well as to generic fear of the tax administration authority. In addition, despite efforts to ensure fair and stable food prices, there are often price-setting difficulties at the municipal level, as MAGA and MINEDUC respectively lobby for higher and lower prices, each in support of their respective beneficiaries (schools on one hand and farmers on the other): this interinstitutional bargaining process in turn results in prices that producers consider too low to justify their participation in the program. Those producers who are interested in the school feeding market, moreover, frequently struggle with low production capacity, and with the fact that schools do not regularly plan their menu and purchase schedule in sync with harvest cycles. Furthermore, a significant amount of agricultural products is ultimately lost or deemed unfit for children’s consumption and rejected, due to a combination of inefficient transport and agrologistics infrastructure, lack of knowledge on good post-harvest practices, and low awareness of basic food quality and safety standards. These challenges are all the more true for women producers, who face higher information gaps, lower market access and lower-scale production, and higher informality than their male counterparts, compounded by restrictive social norms. Overall, traditional gender roles and institutional oversight have created a cycle that limits women to the domestic sphere, creating a lasting perception of women as helpers to their husbands and/or male relatives that prevents them from being seen, or perceive themselves, as autonomous economic agents. Cultural, historical, and language barriers limit the inclusion and empowerment of Indigenous women even further. 75 FAO, 2018. 76 Ministerio de Educación, 2019. 37 DIGITAGRO 2022 3.5 The SFP and the COVID-19 pandemic The lockdown and social distancing measures enacted in response to the emergence of the COVID-19 pandemic in March 2020 caused a number of additional challenges threatening the fulfillment of the SFP objectives, in particular as schools closed and public institutions linked to the SFP reduced face-to-face activities to a minimum. In response to school closures, the government decided to keep the SFP running so that students can continue to receive school meals during quarantine. Throughout the pandemic, schools have been providing bags of nutritious non-perishable foods (such as beans, rice, cereal, sugar, corn flour, and oil) for a value equivalent to the resource allocation per student, which are distributed to children’s families by the OPFs every 20, 25 or 30 days (depending on specific arrangements by MINEDUC). This has made the SFP one of the most important food security instruments in the country, ensuring at least one daily meal for the families of school-age children. At the same time, however, the amount of food that was originally intended for a single student is now being given to an entire family, which might have important repercussions on children’s nutrition security. Similarly, it has become more difficult to ensure the quality of food purchased by schools: while both the CTIDAEs and CTIMAEs normally carry out regular inspection and monitoring of the food delivered by producers, these activities have been very limited due to mobility restrictions. In practice, this means that products do not always meet adequate quality standards required by the SFP, delaying the delivery of safe and healthy products to students and their families. Family farmers engaged in the SFP, on the other side, have also been severely affected by the pandemic, primarily because the list of products purchased by schools has been reduced to only non-perishable foods. In addition, farmers have been suffering the unfair competition of grocers illegally selling agricultural products to schools, due to reduced monitoring by the CTIDAEs and CTIMAEs and to the difficulty for schools to find a registered SFP provider in the quarantine setting. Furthermore, reduced in-person activities have resulted in lower provision of agricultural extension by MAGA extensionists, depriving smallholder farmers of the training and technical assistance necessary to improve their capacity and resilience, as well as of the needed assistance in the registration process with the SFP. Recently, however, MAGA field activities have slowly resumed, taking the pertinent biosafety measures and where possible relying on the use of mobile telephones and other technologies for remote communication – despite obvious challenges in access to technology, digital literacy, and connectivity. 38 Chapter 4 DIGITAGRO 39 DIGITAGRO 2022 DIGITAGRO Key messages » The DIGITAGRO project piloted digital technologies in the department of San Marcos to improve market access for women smallholders, so they could supply the School Feeding Program in a fair, safe, sustai- nable, and profitable way – while helping schools improve children’s nutrition. Partners in the implemen- tation of these activities included the World Food Programme, the Food and Agriculture Organization of the United Nations, and the World Bank’s Gender Innovation Lab for Latin America and the Caribbean, in coordination with local authorities. » The intervention comprised three main activities: » The creation of an e-commerce platform, which aimed to close information gaps between schools and producers on prices and production; to ensure transparency of transactions; and to promote efficiency in food demand from schools. » The production of extension videos targeting women agri-preneurs, to address lack of information on how to access the SFP and on basic agriculture, livestock, and food safety and hygiene good prac- tices necessary to comply with SFP food quality standards. » A digital information campaign, to promote the SFP as a market opportunity for women and provide essential product, price, and contact information. The efficacy of this activity was assessed by an impact evaluation study following an experimental design. 4.1 Digital technologies for rural development Worldwide, digital technologies (tools that collect, store, analyze, and share information digitally, including mobile phones and the Internet) are being increasingly recognized for their broad potential to lower the cost of economic and social transactions for firms, individuals, and the public sector.(77) Notably, this potential can translate into improved efficiency, equity, and environmental sustainability of the agrifood system:(78) digital technologies hold great promise to enhance the efficiency of agrifood markets and production systems, to reduce challenges and inequalities in access to information, technologies, and markets, and to improve the delivery of public services to actors all along the agrifood value chain. The advantage of digital technologies stems from their ability to generate and transmit large amounts of data, as well as to aggregate multiple and diverse economic agents at the same time, at negligible marginal cost. This, in turn, makes it possible to optimize information flows and reduce transaction costs across input, output, and financial markets in the agrifood system.(79) On the farm, digital technologies can provide timely and precise information on input use and help the process of farm decision-making for resource allocation and management, thereby boosting production efficiency. By delivering information on agricultural practices, tools, and inputs to a wide range of producers at lower cost than 77 World Bank, 2016a. 78 World Bank, 2019. 79 Schroeder et al., 2021. 40 DIGITAGRO 2022 traditional extension services,(80) electronic extension systems (e-extension) involving combinations of software, platforms, and devices with varying levels of sophistication can increase the spread and rates of adoption of good agriculture practices and technologies.(81) By reducing the cost of remoteness and making advisory services and technical assistance available to a broader range of producers, moreover, e-extension can also contribute to farmers’ inclusion and promote equity.(82) Off-farm, on the other hand, digital technologies can lower information-related transaction costs associated with farmers’ access to upstream and downstream markets, leading to improved allocative efficiency and equity in agrifood markets. Notably, digital technologies facilitate the transmission of market information, allowing producers to overcome persistent information asymmetries and reliance on market intermediaries.(83) A number of studies have shown that, in many cases, this can result in better farmgate prices(84) and in higher competition and lower price dispersion on local agrifood markets(85) in various parts of the world. Furthermore, by reducing search costs and increasing transparency and trust in transactions (for instance thanks to digitally-enabled marketplaces, also known as e-platforms) that link buyers and sellers along the agrifood value chain, digital technologies can reduce input costs(86) and promote market access for small-scale and other marginalized producers,(87) also through improved product differentiation and positioning on international markets.(88) Certainly, emerging technologies should not be viewed as a cure-all solution to development challenges: if improperly managed, they could exact significant costs, especially in terms of inequality, market power, and data privacy and cybersecurity;(89) and in many instances, their benefits have failed to materialize evenly and at large scale.(90) Importantly, the potential benefits of digital technologies, especially in rural areas, cannot be fully realized in the absence of complementary analog investments and policies – in particular in terms of infrastructure (including improving connectivity), creation of skills (including digital literacy) and social capital, market structure, and access to finance.(91) Nonetheless, the potential of leveraging digital technologies as a complement of broader initiatives for rural development is high, and can be a catalyst of efficiency, inclusion, and equity along economic, spatial, and social lines. In particular, some applications appear to be simple yet potentially impactful tools in support of the development objectives of Guatemala’s School Feeding Program. E-platforms for agricultural products can shorten agricultural value chains, provide access to new markets, reduce transaction costs, improve price transparency, and promote market efficiency. E-extension services provide a cost-effective way to reach a greater number of producers, and can help raise farm resilience and profits. And simple diffusion of market information, even with unsophisticated methods such as via simple text messages, can help increase farmers’ sales and revenues, and reduce price dispersions across markets.(92) The rest of this section describes how the adoption of these technologies was piloted in the DIGITAGRO project to encourage the participation of women smallholders in the SFP. 80 Aker, 2011. 81 Al-Hassan et al., 2013; Cole and Fernando, 2012; Gandhi et al., 2009; World Bank, 2011. 82 Deichmann et al., 2016. 83 Deichmann et al., 2016. 84 Hildebrandt et al., 2020; Labonne and Chase, 2009; Mitra et al., 2018; Svenson and Yanagizawa, 2009. 85 Aker and Fafchamps, 2015; Beuermann et al., 2012; Jensen, 2007. 86 Raj et al., 2011, Sawant et al., 2016. 87 Aker et al., 2016; Einav et al., 2016; Kumar, 2004. 88 Foster et al., 2018; Qiang et al., 2012. 89 Morris et al., 2020. 90 Deichmann et al., 2016; World Bank, 2016a. 91 Shroeder et al., 2021; World Bank, 2019. 92 World Bank, 2019. 41 DIGITAGRO 2022 4.2 DIGITAGRO: Investing in digital technology to increase market access for women agri-preneurs in Guatemala In line with the conceptual framework outlined above, the DIGITAGRO project piloted digital technologies to improve market access for women smallholders, so they could supply the School Feeding Program in a fair, safe, sustainable, and profitable way – while helping schools improve children’s nutrition. The intervention revolved around three main activities, which had in common the aim to address, from a variety of angles, the information gaps and asymmetries that preclude the smooth functioning of the School Feeding Program, on one hand, and hold back women from taking advantage of the program as a profitable market opportunity, on the other. These activities, described in detail in the rest of this chapter, are as follows: 1. The creation of an e-commerce platform, in partnership with the WFP, to close the information gap between schools and producers on who produces and buys what and at what price and to ensure transparency of transactions, as well as to promote efficiency in food demand from schools based on number of pupils and chosen menus. 2. The production of extension videos targeted to women agri-preneurs, in partnership with FAO, to address the lack of information on the part of women producers on how to access the SFP and on basic agriculture, livestock, and food safety and hygiene good practices necessary to comply with SFP food quality standards. 3. A digital information campaign, conducted in collaboration with the World Bank’s Gender Innovation Lab for Latin America and the Caribbean (LACGIL), to promote the SFP as a profitable market opportunity for women and to provide essential product, price, and contact information through a short video and SMS reminders. The efficacy of this activity was also assessed through a rigorous impact evaluation following an experimental design. The pilot operated in the department of San Marcos, to exploit synergies with ongoing activities in support of the School Feeding Program carried out by development partners WFP and FAO (cf. Box 4). The department of San Marcos is located in the southwestern part of the country, and it is bounded by the Pacific Ocean to the south and Mexico to the west. The geography of the department sees three different topographic regions with distinct climatic features and production systems. The plateau (altiplano) in the northern part is characterized by steep mountainous land and volcanoes and a cold and rainy climate. The central part (bocacosta), on the slopes of the mountain range, has the heaviest rainfall in the country and a semi-warm climate with no well-defined cold season. The southern part (costa) forms a band of alluvial plain along the Pacific coast and has a warm climate with lower rain ranging from dry winters to humid summers. Accordingly, agricultural production sees a prevalence of vegetables and legumes in the highlands, and tropical fruits on the coast. With a population of more than 1.03 million inhabitants in 2018, San Marcos accounts for 6.9 percent of the Guatemalan population, and is the fourth most populous department of the country.(93) Around 31 percent of the population is Indigenous of Mayan descent, of whom 92 percent belong to the Mam community and live predominantly in the plateau area.(94) Almost 75 percent of the department’s population lives in rural areas, and 40 percent of the economically active population belong to the agriculture, livestock, forestry, and fishery sector.(95) In 2014, 60.2 percent of the department’s population was considered poor (of these, 36.5 percent were extremely poor),(96) whereas 82.6 percent suffered from multidimensional poverty (the fourth highest rate of the 93 INE, 2019. 94 INE, 2019. 95 INE, 2019. 96 Banco de Guatemala, 2019. 42 DIGITAGRO 2022 country).(97) In the same year, 55 percent of children under the age of 5 suffered from chronic malnutrition,(98) while in 2018 45 percent of the population was in a situation of moderate or severe food insecurity.(99) San Marcos, which shares a border with Mexico, is also among the areas most affected by international outmigration: in 2018, it hosted the third-highest number of households reporting either an emigrated family member or to be receiving remittances, accounting for 10.8 percent of the national total.(100) While remittances constitute a significant complement to the earnings of migrants’ households, it is not uncommon that they represent their only source of income, in a department where 62 percent of persons of working age are either unemployed or inactive. As migrants are overwhelmingly male,(101) this means that a high number of women are left behind in a condition of vulnerability, especially in rural areas.(102) Box 4 WFP and FAO: Joint actions to link family agriculture and the School Feeding Pro- gram in Guatemala Both the FAO and WFP have more than 50 years of global experience working in support of school feeding and local agricultural production. In Guatemala, where both agencies have more than 40 years’ presence, WFP and FAO joined forces with the International Fund for Agricultural Development (IFAD) to support the Government of Guatemala in the implementation of the new School Feeding Law, to contribute to the effec- tive participation of family farming organizations in the SFP as local providers of healthy and nutritious food with cultural relevance for school children. Between 2018 and 2022, the joint FAO-WFP-IFAD project has been providing support in the following areas: 1. Linking family farming to the SFP, by strengthening the capacities of small farmers in terms of producti- vity, market access, environmental sustainability, and financial management. 2. Creation of healthy, culturally appropriate menus with local products, and the improvement of school parents’ organizations (OPFs) managerial and food preparation capacity. 3. Support to the operational base to take the SFP to scale through effective coordination of the parties involved in the program and improvement of monitoring and evaluation mechanisms. 97 UNDP, 2016. Multidimensional poverty is an aggregate indicator combining health, education, and standards of living. 98 ENSMI, 2017. 99 IPC, 2018. 100 IOM and UNFPA, 2021. 101 At the national level, around 80 percent of migrants are men (IOM and UNFPA, 2021). 102 At the national level, 14.9 percent of rural households with migrant relatives or receiving remittances are in a condition of occupational precariousness unmet basic need, defined as the intersection of (i) household head with no education, (ii) number of household members equal or above 4; and (iii) all other household members are unemployed or inactive (IOM and UNFPA, 2021). 43 DIGITAGRO 2022 Table 2. Main activities of the WFP and FAO in the joint FAO-WFP project Main activities WFP Main activities FAO » Support in the articulation of the national » Support in the articulation of the national, departmental, and departmental technical commissions for and municipal technical commissions for School Feeding. School Feeding. » Knowledge transfer to the MINEDUC, MAGA and » Technical and financial support in the Municipalities, to generate sustainability. socialization of school menus among OPFs, » Technical and financial support for the strengthening of teachers, and school and staff. productive, associative and administrative capacities of » Technical and financial support in the training family farming organizations and producers. of personnel who prepare food. » Technical support to enabling a formal economy » Support in the provision of necessary environment, introducing a tax culture among family supplies and utensils in prioritized schools. farming organizations. » Nutritional and gender awareness » Support in the knowledge exchange on experiences campaigns. in the implementation of public purchases from family » Technical and financial support for women’s farming in the Latin America region. participation and economic empowerment in » Support in the purchase of equipment for food the school feeding value chain. transformation processes. Among several achievements, the project has successfully supported the design and implementation, on the part of MAGA, of mechanisms for linking family farming to the SFP, and it has contributed to the formalization and operation of a public procurement system for school meals that reflects the specificities of the family farming value chain. 4.2.1 E-commerce platform The SFP e-commerce platform produced in partnership with the WFP had the objective to link schools and eligible producers to information on food demand and supply for school meals, on one hand enabling producers to seize promising market opportunities, and on another hand allowing schools to have more complete information about existing supply, while enhancing the overall transparency of public procurement systems. The activity was developed based on a mapping of the SFP processes and information flows, analyzing the role and bottlenecks of different actors involved in the school meals value chain, and validating each step in an iterative process with local actors and central government authorities. While the ultimate aim of the project was to develop a fully-functional digital marketplace for school meals and pilot it among San Marcos farmers and schools, the COVID-19 outbreak called for some mid-course corrections and a re-prioritization of activities to support the MINEDUC in ensuring the smooth functioning of the SFP amidst the disruptions caused by the pandemic. As such, the bulk of DIGITAGRO funds were directed towards the rapid development of a preliminary simplified version of the platform that allowed the MINEDUC to monitor the food purchases of schools despite mobility restrictions and social distancing. The resulting online platform Mis compras del PAE (in English “My SFP purchases”) became operational in August 2020 and is currently being used by 24,937 schools nationwide to register the invoices of their food purchases from around 45,000 providers (both family farmers and regular shops). For each product purchased, the platform allows the user to register the type of 44 DIGITAGRO 2022 product, the provider, the price paid, and the date of the purchase. In addition, the platform automatically reports on each school’s SFP funds balance and generates the school’s accounting reports required by the MINEDUC. In parallel, DIGITAGRO has been supporting the conceptualization of the final e-commerce platform that is currently being developed by the WFP, through participation in a Technical Committee for the School Feeding Platform composed of representatives from the MINEDUC, MAGA, WFP, and the World Bank. The related platform Alimentación Escolar (in English, “School Feeding”) is currently in its final phases of development, both as a web tool and as a smartphone app. The platform is composed of several user-friendly modules, including: » Demand: Generation of accurate, school-specific shopping lists with products for school meals, based on each school’s number of students and the schools’ selection of official menus. » Supply: Registration of local food suppliers as official SFP providers, registration of products offered, supply capacity, unit prices. » Linking demand and supply: Matching demand and supply between schools and local family farmers, performing transactions, feedback mechanisms on product quality. » (Admin) Purchase monitoring: Crossing data generated by suppliers and schools’ shopping lists, monitoring of school spending and processes. Considering the high connectivity challenges faced in rural areas of Guatemala, each module is supplemented by an SMS notification tool, developed with DIGITAGRO funds, which allows each aspect of the platform’s workflow (orders, offers, invoices, receipts, direct communication between schools and providers) to also operate offline in remote areas. The SMS tool also feeds into the admin module so as to enable reporting on the notifications sent for each workflow step and user type. The platform is fully compatible with the systems and software of the MINEDUC, to which it will be transferred upon completion. Ideally, the platform could in future also be linked to other official platforms beyond school procurement (e.g. MAGA registries of family agriculture, registry of beneficiaries of social security schemes). The entire tool has been designed following a human-centered approach that aims to enhance effectiveness, efficiency, and accessibility by focusing on the user’s needs and requirements to make systems usable. In practice, this meant that each aspect of the platform (language, graphics, logic sequencing of steps) has been focused on the practical usability of the end user, based on studies of user behavior and empathy and building upon participatory action through community brainstorming, validation and feedback. These activities were carried out through several workshops and focus groups with local family farmers, schools, and institutional counterparts in the department of San Marcos. The piloting of the platform will be carried out by the WFP throughout 2022 in at least 100 schools in San Marcos and five other Guatemalan departments to be defined in consultation with government counterparts. The piloting will be accompanied by an impact evaluation study that will be carried out by the World Bank’s Development Impact Evaluation (DIME) group following an experimental design. 4.2.2 Suite of extension videos The videos produced in partnership with FAO aimed to develop a soft extension system targeted to women agri- preneurs, to engage them as effective suppliers of the SFP. In particular, while the SFP is a widely known government program that most people are aware of for its contribution towards child nutrition, the videos seek to specifically portray the SFP as a good market opportunity for women producers. Apart from the sensibilization around the potential profitability of selling to the SFP, moreover, the videos provide practical guidance on the easy steps that need to be followed to participate in it, and on the basic practices that ensure that produce to be sold to schools can meet the SFP food quality, hygiene, 45 DIGITAGRO 2022 and safety standards. The videos combine images, footage, graphics, animations, narration, and motivational messages from existing SFP providers.(103) It is important to specify that the videos are by no means to be considered substitutes to official extension channels. Rather, they are intended to complement the work by regular MAGA extension agents, providing them with an additional tool to convey simple, basic information to large audiences and in a remote fashion (particularly crucial during COVID-19 social distancing), to be then expanded on through dedicated work on the field. In fact, the content of the videos was validated by local MAGA officials and extensionists in San Marcos, and the videos themselves always refer the viewer to the nearest MAGA and MINEDUC field offices and agents for further information and assistance. In total, the extension suite is composed of five short videos (one motivational, one informative, two educational, and a summary), for a combined duration of around 30 minutes: » Video 1: Motivational video on the School Feeding Program in Guatemala. This video aims to be broad in coverage and to work as a sort of general commercial of the SFP to be shown to both female and male audiences. The video exposes the general features of the SFP, focusing on its main actors, its functioning, and its engagement of family agriculture, and it raises awareness on the program’s beneficial implications in terms of children’s nutrition and educational attainment, market opportunities and fair prices for smallholder farmers, local economies and community development, and overall transformation of the rural society. » Video 2: Information video with practical guidance on how women farmers can access the SFP market. The video consists of 3 parts: (i) different ways to join the SFP either as a registered provider (explaining the necessary steps to register with MAGA and SAT) or by selling products to an existing supplier; (ii) a recap of the basic features of the SFP and of the type and quality of produce in demand by schools; (iii) a basic introduction to the concept of an e-commerce platform that connects the demand of schools with the supply of family agriculture, and information on the advantages of the platform developed under the pilot. » Video 3: Educational video on good practices for agricultural produce. The video presents in a didactic way simple good practices recommended for four groups of crops (vegetables, fruits, and herbs; cereals; legumes; roots and tubers) demanded by schools, stressing the importance of adopting such practices in order to access high-value markets such as the SFP. The featured agricultural products are relevant to the agricultural and cultural context of San Marcos and are those included in official SFP menus. While the video briefly touches on the pre-harvest and harvest phases, it especially focuses on post-harvest practices (further divided into collection, cleaning and disinfection, selection, packaging, storage, distribution), as well as on food safety, hygiene, and quality standards. » Video 4: Educational video on good practices for produce of animal origin. The video presents in a didactic way simple good practices for products of animal origin (meat, dairy, eggs) demanded by schools, again stressing the how such practices are instrumental to accessing the SFP and other high-value markets. As in Video 3, the featured products of animal origin are relevant to the agricultural and cultural context of San Marcos and are those included in official SFP menus. The video features good practices of benefit, dressing, manufacture, selection, packaging, storage, and distribution, as well as on food safety, hygiene, and quality standards. » Video 5: Summary video. The video summarizes the key content of Videos 1-4 on the opportunities offered by the SFP and the good practices required to sell produce to schools. The video thus consists of three parts: (i) the existence and importance of the SFP; (ii) how the SFP works, and different ways of joining 103 To comply with COVID-imposed mobility restrictions and social distancing, live action was replaced by digital animations whenever filming on site was not possible. Given the impossibility of shooting a full interview with SFP providers, identified testimonial women were asked to record themselves with their phones so that their messages could be embedded in the videos. 46 DIGITAGRO 2022 the SFP as an agri-preneur; (iii) the importance of adopting high food-quality standards. The video also includes simple messaging (as advised by MAGA) about best practices for food handling to be adopted in the COVID-19 pandemic context, as well as basic information on the modalities of SFP operation under COVID-19. The entire video suite is designed to be specifically relatable for women agri-preneurs. All producers shown in the videos are women, as are the featured SFP providers interviewed. Furthermore, the information, educational, and summary videos are narrated by a female voice, directly address female viewers, and predominantly show the products that women in San Marcos specialize in (e.g., leafy vegetables, poultry, cheese, eggs). To boost inclusion of women from underserved communities, all videos consistently show women dressed both in Indigenous and non-Indigenous clothes. Moreover, the suite is available both in Spanish and in Mam language, and the latter features an interview with a Mam SFP provider. The videos were produced in various formats so as to be available to women through a variety of channels including internet, social media, WhatsApp, individual or community projections organized by MAGA extensionists, among others. Video 5 was included in the information campaign described in the next section, and was also shown in loop in waiting rooms at local public spaces in San Marcos to improve visibility and dissemination. 4.2.3 Digital information campaign and impact evaluation The information campaign carried out in collaboration with the World Bank’s LACGIL aimed at proactively encouraging women agri-preneurs to take part in the SFP market within the socially-distanced COVID-19 context. The campaign was entirely carried out by phone, by sending an information bundle consisting of a video and a series of text messages to 881 eligible women (identified in collaboration with MAGA) across 272 villages in San Marcos. Eligibility was defined by being an adult (aged 18+) woman having sold crops or products of animal origin in the previous year and having access to a smartphone. Women participating in the activity received Video 5 summarizing the main highlights of the extension suite, including key information on the general features of the SFP, ways for farmers to sell to schools, food quality standards and practices, and SFP functioning under COVID-19. In addition, women received a tailored set of six SMS that included reminders on how to access the SFP, information on products demanded and prices paid by the program, and relevant contact information: » SMS 1 listed the main products bought by schools in the program. » SMS 2 contained a reminder that the receiver could register as SFP provider and shared the contact information of the MAGA extensionists serving the village of the receiver. » SMS 3 contained a reminder that the receiver could also participate as a support farmer selling produce to registered providers and shared again the contact information of local MAGA extensionists. » SMS 4 detailed the SFP reference prices of the products demanded by schools in the municipality of the receiver.(104) » SMS 5 shared the contact information of SFP providers in the same or nearest community as the receiver who could buy products from local producers, identified in collaboration with MAGA officials. » SMS 6 contained the complete list of all SFP providers in the municipality of the receiver. Throughout the intervention, it was assumed that knowledge provision through the information campaign would result in greater awareness about the SFP market and how to participate in it, positive changes in the attitudes and behaviors towards supplying the SFP (including formally by registering as official providers), and women’s 104 Information for the price list was provided by MAGA officials based on the invoices of providers’ sales to schools in April 2021. 47 DIGITAGRO 2022 empowerment through improved status in the business sphere. To test these hypotheses, the activity was complemented by an impact evaluation carried out between January and November 2021. The information campaign targeting women was rolled out following an experimental design whereby villages were assigned to a treatment group and a control group.(105) While women in the treated villages received the information on the SFP, women in the control villages received a placebo bundle of similar structure and length, consisting of (i) a video on local handicrafts and biodiversity in a different Guatemalan department and (ii) six SMS with reminders on simple COVID prevention measures such as washing hands, using a facemask, social distancing. Importantly, all control-group participants were granted access to the full SFP information available to the treatment group at the end of the evaluation period. The impact evaluation study comprised activities of baseline and endline data collection with participant women, and was complemented by additional surveys and focus groups with registered SFP producers, schools, and MAGA extensionists. To comply with pandemic safety standards, all these activities were carried out by phone – through calls, WhatsApp, and text messages. To make sure that participating in the study did not result in a cost for beneficiaries, airfare or a prepaid internet data plan were provided to each participant for free. Moreover, in a context where it was more difficult to ensure the privacy of phone interviews (given quarantine and the fact that the enumerator could not see the place where the interviewee was answering the phone), an add-on study was conducted to test the efficacy of alternative methods to address potentially sensitive questions in remote interview settings (see Box 5). Details on the experimental design, as well as the results of the impact evaluation study, are presented in the next chapter of this report. 105 In total, 445 women in 130 villages were assigned to the treated group, and 436 women in 142 villages were assigned to the control group. 48 DIGITAGRO 2022 Box 5 Measuring Empowerment Better Relative to men, women face more acute challenges in agriculture and rural settings. Women have less access to land, credit, technology, agricultural information, and higher levels of informality, all compounded by discriminatory and restrictive social norms. Usually, these multi-faceted limitations are at the same time reinforced by a situation of high disempowerment of rural women, where “empowerment” is defined as the expansion in people’s ability to make strategic life choices. While the immediate purpose of DIGITAGRO was to facilitate women’s involvement in the SFP, it was expec- ted that, by providing greater awareness about the school feeding market, increased knowledge on how to participate in it, and information on how to achieve higher-quality agricultural produce, the pilot would improve women’s entrepreneurship and their status in the business sphere, which might in turn translate into enhanced empowerment and agency in their households and communities. To test for this hypothesis, the data collection included a specific module on women’s agency and decision-making ability. In this modu- le, women were asked a select a subsample of questions from the Project-level Women’s Empowerment in Agriculture Index (Pro-WEAI), a survey-based index for measuring empowerment, agency, and inclusion of women in the agricultural sector in a standardized manner, guided by recent evidence on the most relevant questions on agency to be asked on short questionnaires.(106) Questions spanned topics such as business decisions (Who contributes the most to the decision to buy new assets (hoe or fumigation pump)?), money management (Do you have to ask permission to buy vegetables or fruits; clothes for yourself; medications or personal supplies?), and freedom of movement (Who usually decides if a woman can go visit a friend or friend/ neighbor’s house?). As the COVID-19 pandemic struck, however, the shift to phone-based surveying demanded by social dis- tancing protocols and national mobility restrictions gave rise to several concerns as to the reliability of re- motely-collected data on these issues. In fact, as empowerment-related questions tend to be perceived as sensitive by respondents, there is always some degree of risk that they are met with some hesitancy, and even that respondents may provide inaccurate responses if they do not feel comfortable with the in- terview mode and setting. Women may not feel comfortable speaking openly in response to questions on topics such as decision-making over resources, productive inputs, or the income derived from their own work. In addition, depending on their particular, familial and social circumstances, women may also be afraid of incurring in retaliation from their family or from members of their community if they fear that their responses will not remain private. While in-person interviews can be executed in accordance with procedures that ensure privacy, however, a phone call runs the risk of being conducted in the pre- sence of family members or other third parties – for example if women are not the primary owners of the telephone, or if they are less tech-savvy and need assistance while using it, or simply because the enumerator cannot make sure that the interview can happen where other people cannot eavesdrop. This, in turn, may make it almost impossible for the respondent to answer certain questions aloud, po- tentially leading to under- or over-reporting of sensitive topics such as empowerment and agency. 106 Jayachandran et al., 2021. 49 DIGITAGRO 2022 To test whether this was the case, a survey experiment was conducted during the endline survey of the DIGITAGRO impact evaluation, with funding from IPA’s Advanced Research Methods Award. The experiment incorporated additional layers of privacy to the empowerment module of the DIGITAGRO questionnaire, to evaluate the effect of privacy changes on women’s reporting bias on sensitive topics. 1,183 women were randomly assigned to either of two treatment groups or a control group. Individuals in the control group received the questions directly from the enumerator. The first treatment group received the same set of questions, but women in this group responded using code words. For example, when the enumerator as- ked, “Who usually decides whether you can go to the market?”, the women in the control group could reply with the following responses: “myself” or “my husband.” Meanwhile, women in the treatment group would respond with code words, such as, “one” or “two” that corresponded to one of the aforementioned response options. This procedure enhanced privacy as respondents did not risk disclosing potentially sensitive infor- mation to individuals in their surroundings. The second treatment group was granted an additional layer of privacy. Instead of responding with codewords, women in this group would listen to response options then dial telephone keys that corresponded to a particular option. For example, in response to certain questions, women could dial 01 if their answer was “myself” or dial 02 if their answer was “my husband.” It is important to note that assignment to these experiment groups was orthogonal to the sample composition of the original DIGITAGRO impact evaluation (i.e., treated individuals of the survey experiment could indifferently belong to either the treatment or control group of the DIGITAGRO impact evaluation, and similarly for the control individuals). This survey experiment was complemented by qualitative research methods to shed light on the mecha- nisms behind the results. A sample of 121 women received semi-structured interviews with exploratory questions on agency, disclosure, privacy, trust, and confidentiality, and the contents of these surveys were then coded to create a quantitative index of agency against which to benchmark the quantitative results of the survey experiment. If the hypothesis of this survey experiment were confirmed, the results would esta- blish precedent for the development and implementation of similar surveys surrounding topics of women’s empowerment and agency. Additionally, the results could reinforce the credibility of phone surveying as a reliable data collection tool for related topics. By extension, the results of this project would be relevant and applicable to other communities, including informal and vulnerable groups worldwide. Lessons Learned Why do women disclose certain information? Although the quantitative analysis is ongoing, the qualitative analy- sis already suggests that there are certain elements that prompt disclosure of information. However, there are also risks associated to disclosing information. Women were asked questions focused on their experiences in answering phone surveys, the feelings that were prompted throughout the process, and the factors that could impact the reliability of collected data. Gender norms, agency, privacy, and trust certainly have an influence on what information women decide to disclose. Therefore, it is critical to integrate phone interview methods that create an environment in which women feel comfortable to provide reliable responses to questions. 50 DIGITAGRO 2022 One of the most predominant themes across the interviews was asking for permission. Some interviewees said that it is both common and expected that all decisions and actions taken by a woman should be com- municated with her spouse. Asking for permission is based more on notions of respect that intersect with notions of homemaking. Essentially, it serves as a method to communicate with their spouse and take their ideas into consideration. For other interviewees, instead, report that they would feel more confident and outspoken in different environments outside of their community in which, because of a sense of social pres- sure and the fear of judgment from members of their own community. For some women, privacy was a key determinant of whether they decide to disclose certain information. Based on the findings from the questionnaire, interview environment and the presence of spouses or other individuals during a phone survey were highly influential to the provision of information. Control over the interview environment is not guaranteed in phone interviews. There are times that women want to discuss private matters but refrain from doing so due to the possibility of other individuals in their surroundings. Women described their spouses’ suspicion about who the women were talking to as well as being scolded by their spouses for devoting too much time to the phone call. Women also refrain from discussing private matters for fear that individuals or third parties in their surroundings misconstruing their conversations and disseminating this information to other community members. If a woman is given a private space and/or spoke to someone that could assure their privacy, she would be more open to expressing what is happening in her life that she may not otherwise divulge under different circumstances. 51 Chapter 5 Evaluating a digital information campaign in San Marcos 52 DIGITAGRO 2022 Evaluating a digital information campaign in San Marcos(107) Key messages » DIGITAGRO’S information campaign was evaluated through an impact evaluation study based on an expe- rimental design, which involved 880 adult women engaged in the production and commercialization of a variety of agricultural-animal products. » The study reveals that digital information campaign increased awareness among rural women about the SFP as an economic opportunity. The delivery of information proved especially relevant for individuals that had not been reached by traditional extension programs. » Furthermore, the intervention had an effect on women’s selling decisions, increasing the likeliness of se- lling animal products demanded by the SFP, and changed women’s ability to participate in household decision-making processes. These effects are stronger for partnered women. » Despite the success at delivering information about the SFP and at encouraging women to increase their sales, the information campaign did not have an effect on the willingness of participants to join the SFP. The intervention, in fact, was not constructed to address several analog challenges faced by producers when selling their products. 5.1 The experiment The experimental design of the information campaign comprised five key activities, including sample design, data collection (baseline and endline), treatment delivery, and complementary information collection and analysis. The sample design was built using administrative records provided by MAGA of the CADER(108) program, as well as the lists of farmers that already supply to registered providers in the SFP market. As the digital information campaign had to rely on video delivery by WhatsApp, villages with no record of WhatsApp users were dropped from the sample, while the remaining villages were randomly allocated to either the treatment or control group (208 villages were assigned to treatment and 201 selected to control). Similarly, individuals that did not have access (either directly, or through a family member) to WhatsApp were considered not eligible to participate in the intervention. After this first screening, the pool of potential participants comprised 3042 individuals across 409 villages (see Box 6 for technical details of the randomization protocol). 107 Chapter 5 summarizes the main results of the impact evaluation of DIGITAGRO’s information campaign. The companion technical paper Investing in Digital Technologies to Increase Market Access for Women Agri-preneurs in Guatemala (Lopez et al. 2022) explains the empirical design, fieldwork, intervention, and results in more detail. Tables in Annexes 1 through 5 draw heavily from the technical paper. 108 The CADER (Learning Center for Rural Development) is an extension service offered by the MAGA that comprises a voluntary group of farmers in a village with a leader trained by the MAGA in agricultural practices. Each leader is compelled to extend the training among the group members. 53 DIGITAGRO 2022 Box 6 Randomization protocol The DIGITAGRO randomization protocol followed a cluster trial methodology, where villages (clusters) were allocated randomly into the treatment or control group and stratified based on a set of characteristics. Spe- cifically, the strata were the municipality and a variable to account for villages with high rates of WhatsApp adoption (above 95 percent). This last was constructed based on the percentage of WhatsApp users within each village. So, the registry of CADER participants was ordered into municipalities, and then the clusters were further divided into high and low WhatsApp adoption rates for the stratification. To maximize the probability of achieving balance, the randomization process was iterated 500 times with different random seeds. Then, the treatment-control allocation that yielded the highest p-values for a set of predetermined village-level characteristics was chosen for the experiment. As an additional list of farmers was received from MAGA to maximize participation beyond the CADER fra- mework, this new information resulted in the randomization being performed twice: first, with the initial batch of data available (343 villages from the CADER registries), and second with the new batch of data (66 new villages obtained from MAGA). The baseline survey gathered information from eligible women on agriculture, household characteristics, empowerment markers, agricultural production, and previous knowledge and interaction with the SFP program. Only women participating in agriculture and whose households had sold any agricultural product during the last year before the intervention were considered eligible to participate in the intervention. If the woman was considered eligible, the baseline survey followed immediately after the screening. Overall, 881 eligible women across 272 villages were successfully interviewed from the total original pool and invited to receive the video and SMSs.(109) The baseline survey was collected in April and May of 2021. Treatment delivery (video + SMS) happened after the baseline survey, using WhatsApp and text messages.(110) Once the participant finished answering the baseline interview, she agreed with the enumerator on a specific day when they could watch the summary video on the SFP. On that day, the farmer’s phone would receive an internet package equivalent to approximately 1 USD, and the farmer would receive the video through WhatsApp.(111) This internet package was included as an incentive for participation and allowed the interviewee to watch and download the video without incurring any personal cost. After sending the video, a protocol would verify that the participant had been able to watch the video and offered help in case of technical difficulties if needed. Once the verification process was finalized, the farmers received the SMS suite, with two-day intervals between each message to avoid saturation and fatigue.(112) Box 7 provides further details about the implementation of the treatment.  109 See Annex 1 for baseline descriptive statistics and balance tables. 110 See Annex 2 for take-up and response rates statistics. 111 WhatsApp was the preferred given its high usage rates in Latin America. The application is pre-installed on all the phones in Guatemala, making it ubiquitous across smartphone users and carriers. 112 Nonetheless, SMSs were also sent to farmers for whom it was not possible to verify whether they had watched the video or not, in order to maximize information diffusion. 54 DIGITAGRO 2022 Box 7 Delivering the treatment using digital technologies in rural Guatemala In rural San Marcos, lack of broadband internet coverage coupled with low smartphone and techno- logical skills required special logistical considerations to deliver the intervention. Thus, a five-step verifi- cation process was created to verify that farmers could watch the video and assist farmers who needed further support: 1. The participant was asked if she had watched the video in a verification call. 2. If she had not, the enumerator asked for the reason. 3. If possible, the interviewee was prompted to search for assistance in the household while on the phone. 4. If needed, further technical support was given. 5. Finally, the process repeated until successful verification, or until the farmer refused to be called again. Fieldwork personnel was trained to help farmers with a series of technical difficulties. For example, enume- rators could assist farmers in opening WhatsApp if they did not know how to do it or provide technical help if the device memory was full or the internet was not enabled. Help was not provided if the farmer had no phone signal, ran out of data, or could not use the household phone at that moment. In this last case, ano- ther phone call was scheduled for a later, more suitable moment. Overall, it was verified that 93.8 percent of farmers in the baseline were able to watch the video. From a total of 881 interviewees, most of them (827) gave verbal confirmation that they had watched the video to the enumerators, and only 1.8 percent needed technical assistance to watch it. Participants were made aware that they had the possibility to opt out of the intervention at any moment if they so wished: overall, only 30 women opted out of the video campaign (21 in the treatment group and 9 in the control group), and a tiny six asked to stop receiving text messages (three in the treatment group and three in the control group). The endline survey was conducted two months after the treatment (in August of 2021) to analyze the effect of the intervention, allowing for one cycle of school food purchases between the baseline and endline. Attrition between baseline and endline was not negligible but remained within acceptable levels for phone interview recontact rates: 73 percent of the individuals interviewed at baseline could be recontacted at endline, and evidence suggest the two groups are comparable.(113) The endline gathered information about the SFP, harvesting, sales of agricultural and animal products, trust in institutions involved in the SFP, farmers’ interactions with the SFP market, and empowerment markers. To complement the main data collections, several complementary qualitative and data collection activities were conducted with key actors of the program (existing SFP providers, MAGA officials, and School Parents’ Organizations) to contextualize the intervention results.(114) At design phase, focus groups were held with 113 See Annex 3 for sample balance tables. 114 After the outbreak of COVID-19, all surveying activities and focus groups were done through phones to prevent any potential health risks in the context of the COVID-19 pandemic. See Annex 2 for take-up and response rates. 55 DIGITAGRO 2022 School Parents Organizations (OPFs) as well as with farmers and SFP providers to understand the challenges of the SFP and the role of women in San Marcos agriculture. Following the endline data collection, SFP providers were engaged through survey to elicit their perspective on the SFP and on the contents of the digital information campaign. MAGA officials participated in focus groups aimed at disentangling the difficulties in engaging new producers in the program, learning about the agricultural extension efforts during the pandemic, understanding certain characteristics of the SFP, and getting further feedback on the usefulness of the digital information campaign. MAGA officials also participated in a self-administered online survey about contacts with producers after the information campaign. Lastly, OPFs participated in focus groups to understand the relationship between OPF and SFP providers, the OPF role in the SFP market dynamics, SFP accounting methods, and schools’ adaptation to the pandemic. In-depth findings from these activities are reported in Annex 6. 5.2 Characterizing women agri-preneurs in San Marcos As mentioned in Section 5.1, over 880 adult women engaged in family farming in San Marcos (of whom 319 in the plateau area, 183 in the central area, and 378 in the coastal area) participated in the baseline data collection for the digital information campaign.(115) This sample is primarily composed of women in their prime age (38 years old on average), who tend to be partnered (71 percent), have on average 2 children, and live in large families with high dependency rates (over an average of six household members, two are on average children aged less than 15 and two are elderly people older than 65). Although 17 percent of interviewed women indicate they are the head of their household, families tend to have a predominantly patriarchal structure: in 59 percent of the cases, the husband or partner is identified as the household head and in 19 percent of cases it is the woman’s parents who are. These households are also overwhelmingly characterized by very low levels of education: overall, only 15 percent of household heads went beyond the basic cycle of mandatory schooling consisting of primary and lower secondary education. In 61 percent of the sampled households, the head only attended some primary education (of these, less than two-fifths completed it), and in 13 percent of cases the household head did not receive any education at all. Percentages are very similar for interviewed women, with 19 percent going beyond the mandatory schooling, 55 percent attending some primary education, and 11 percent not receiving any education at all. While no data on ethnicity was collected, 21 percent of the sample opted for the interview to be held in Mam (the Mayan language spoken in San Marcos) rather than in Spanish, indicating that Indigenous women represent at least a fifth of the sample: 92 percent of these women live in the plateau area, where they make up more than half of the sample. As potential suppliers of the School Feeding Program, all interviewed women are engaged in commercial family farming, or at least in subsistence farming with potential to transitioning to commercial farming, according to the official classification of the Ministry of Agriculture (cf. Section 2 of this report). As such, all women in the sample are involved in the production and commercialization of a variety of agricultural products (see Figure 1, panel a) and products of animal origin (see Figure 2). Overall, women report working on farmland of about 2.4 manzanas of extension (approximately 24,000 square meters) on average, although only one out of two (54 percent) report that either they or their family own some of this land. There appear nonetheless to be strong differences in production systems along geographic lines. Expectedly given the climatic variation across the department, the range of agricultural products harvested differs substantially between the plateau and central areas, on one hand, and the coastal area, on the other (Figure 1, panels b-d): despite the ubiquitous prevalence of food-security crops such as corn and beans production, in fact, smallholder women in the plateau and central areas produce a wide range of vegetables, roots and tubers, whereas the coastal area is dominated by the production of fruits such as banana, lemon, and tomato. Similarly, while the average land size is 1.86 and 1.87 manzanas in the plateau and central areas, it is substantially larger at 3.35 manzanas in the coastal area. 115 Detailed tables with descriptive statistics are available in Annex 1. 56 DIGITAGRO 2022 Figure 1. Agricultural production among sampled women in San Marcos (a) Overall (b) Plateau area carrot chard carrot chard broccoli broccoli tomatoes tomatoes onion onion cabbage cabbage cilantro cilantro potatoes potatoes black beans black beans corn corn (c) Central area (d) Coastal area chard tomatoes carrot broccoli black beans tomatoes onion cabbage cilantro platano lemons spinach potatoes black beans corn corn Note: Estimates as simple means of women in the baseline. Figure 2. Produce of animal origin among sampled women in San Marcos Cow meat 2% Other animal products 1%* Other animal meat 6% Cow Cheese 4% Cow milk 9% Chiken eggs 63% Chiken meat 16% *Other animal products include sausages, wool, and butter Note: Estimates as simple means of women in the baseline. 57 DIGITAGRO 2022 Despite their strong engagement in agriculture, the majority of interviewed women are working in a condition of informality, which increases vulnerability to shocks and poverty in the face of crises: only in 44 percent of cases does anyone in the household have a tax identification number (NIT, for its Spanish acronym), and a meager 24 percent can issue invoices. Three-quarters of women, in addition, report facing a broad range of challenges in selling their produce (see Figure 3). The most recurring problem is transportation of products, including distance from the selling place. In fact, women who do not sell their produce directly on their farm have to travel long distances to reach their selling place, most often by pick-up (used by 27 percent of women), car (21 percent), and even walking (another 21 percent). While the trip takes less than 45 minutes for 72 percent of women, in the remaining cases travelling time exceeds one hour, and 7.5 percent of women have to travel for more than two hours to reach their destination. In general, the percentage of women reaching their selling place by walking decreases when travel time increases, meaning that those who travel long hours have to reach really distant locations to be able to sell their produce. Other notable prevalent barriers to selling are low agricultural prices, the inability of finding a place to sell the products, the lack of water, product damages and substandard quality of produce, and lack of financing. While less than 4 percent of women indicate access to inputs as a significant constraint, the use of agricultural inputs is rather mixed in the sample. While most women use fertilizers, both natural (89 percent) and chemical (70 percent), almost half of the sample (46 percent) does not employ pesticides or herbicides in their production, and only 51 percent use improved seed. When it comes to the use of agricultural machinery (such as tractors, planters, weeders), the percentage drops to 4.5. Consistent with the women’s record of concerns with lack of water, only 29 percent of the sample report using a technified irrigation system (e.g. spray, exuding, or drip irrigation system), which leaves the rest substantially exposed to climate hazards. Due to their agricultural engagement, women producers would in theory be eligible to receive agricultural extension services by the Ministry of Agriculture. However, this is not always the case in practice, due to various Figure 3. Reported barriers to sell problems in transporting products/high… low selling price not able to find a place to sell lack of water for agriculture products damages lack of financing covid-19 increase in input prices competition of imported products limited/infertile land not able to get input lack of personnel low production social conflicts local competition lack of training/orientation about harvesting unstable prices little children to care other (lack of time/sickness/no answer) 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0% Percentage of incidence (over tot women reporting problems) Note: Estimates as simple means of women in the baseline. 58 DIGITAGRO 2022 access barriers and cultural constraints, coupled with a reduced field presence by professional extensionists in the wake of the COVID-19 pandemic. As a matter of fact, almost half of the women in our sample report not having received any agricultural training in the year prior to the interview, and a quarter of the sample declare having received some training with only a lower than monthly frequency. Notwithstanding these challenges, however, the interviewed women maintain a dynamic entrepreneurial spirit, as well as a bold and confident attitude towards taking on new challenges. When asked about their ability to perform difficult tasks, 80 percent of them claim they feel confident they can fulfil them, and 91 percent believe they can be successful in obtaining every objective they aim at. Within their households, moreover, almost two-thirds of interviewed women participate in the decision-making process about sales and marketing of their produce. Women’s produce is overwhelmingly sold in marketplaces or directly to the final consumer (often neighbors). Roughly 16 percent of women also report selling either to intermediaries (coyotes) or to merchants and dealers. Despite the fact that 92 and 71.5 percent of women report harvesting and selling, respectively, products that are included in the official menus of the School Feeding Program, only a tiny 8 percent have commercial relations with schools or SFP providers. This does not seem to be driven by an unawareness of the existence of the SFP: in fact, 77 percent of women have heard about the program (mostly through schools themselves, the MAGA, or word of mouth), and almost a quarter of these have contacts (either personally or indirectly through a family member) with at least one SFP provider. When asked whether they would be interested in selling to the SFP in future by registering as a provider, however, only 28 percent of women respond positively, citing as expected benefits (see Figure 4) the match of their production with schools’ food demand (in 48 percent of cases), good prices or earnings prospects (31 percent), and the possibility to contribute to children’s nutrition (15 percent). Among the reasons for not being interested in registering as SFP provider (see Figure 5), 40 percent of women cite thinking not being able to supply the schools in a continuous way because of insufficient quantities of product harvested or because of harvesting only seasonal products, and an additional 7 percent erroneously think the SFP does not demand their produce. Furthermore, another 33 percent of women mention having concerns about the process: some do not know it, some think it is too difficult or that they do not have enough time, some fear it entails a too high commitment on their side. These concerns are compounded by low levels of trust in the institutions that are most involved in the School Feeding Program: only 60 and 64 percent of interviewed women feel they can trust respectively the Ministry of Agriculture and the Ministry of Education, and the percentage halves when it comes to the Tax Administration Superintendency (which oversees the granting of the NIT and of the ability to issue invoices needed to register as SFP provider). Figure 4. Main reason for being interested in the SFP market SFP buys produced varieties good prices or earnings prospects possibility to contribute to children’s nutrition other reasons 0 5 10 15 20 25 30 35 40 45 50 Percentage of incidence (over tot women interested in SFP) Note: Estimates as simple means of women in the baseline. 59 DIGITAGRO 2022 Figure 5. Main reason for NOT being interested in the SFP market lack of products/only seasonal products SFP does not buy their products lack of time not knowing how to register difficult registering process too much responsability/commitment lack of money already sell to other buyers need to ask to husband/father lack of water Other 0 5 10 15 20 25 30 35 40 45 Percentage of incidence (over tot women not interested in SFP) Products Process Others Note: Estimates as simple means of women in the baseline. 5.3 Did the digital information campaign work? This section reports the impact of the digital information campaign, as evaluated through the randomized field experiment. Simple regression analysis was used to compare a range of outcomes of interest across the treatment and control group. By design of the experiment, the randomization protocol should ensure that any statistically significant differences observed between the two groups are the result of the digital information campaign. In order to improve the precision of the estimates, and in some pieces of analysis to correct for potential bias induced by observable confounders, the analysis however also systematically controlled for a broad range of socio-demographic, time, location, and product covariates, depending on the specific outcome considered.(116) The rest of this section presents the main findings of the analysis in terms of information delivery, sales decisions, SFP participation, and women’s empowerment, and explores the likely mechanisms that could be driving such results. 5.3.1 The intervention increased knowledge about the School Feeding Program and encouraged some women to sell SFP products The main finding from the impact evaluation study is that the digital information campaign increased awareness among rural women about the SFP as an economic opportunity. Furthermore, it had an effect on women’s selling decisions, despite the short time horizon between treatment and endline survey. Moreover, the intervention also changed women’s ability to participate in household decision-making process. SFP Information As the SFP is a very popular food-security government program, knowledge about its existence as such is very widespread across Guatemala, and the information campaign did not have much margin for improving its already high popularity. In the DIGITAGRO sample, in fact, the percentage of women in the control group who knew about 116 Technical details on the analysis and full results tables are presented in Annex 4. 60 DIGITAGRO 2022 its existence even without being exposed to the information campaign is already 88 percent (cf. the first green bar in Figure 6): while among women participating in the information campaign this percentage increases to 92 percent (first gray bar in Figure 6), this difference is not statistically significant. What the information campaign seems to have successfully impacted, instead, is participants’ awareness of the program as a business opportunity. With respect to the control group, women agri-preneurs in the treatment group are 19.9 percent more likely to report knowing that the SFP buys agricultural products from local farmers, and 21.2 percent more likely to know that farmers can register as providers of the SFP (cf. always Figure 6). In addition, the information campaign improved women’s knowledge of the basic features of the program: participants in the treatment group are almost twice as likely to report that they know the steps to register as a SFP provider, and they appear more likely to know which agricultural products the SFP buys as well as to have the contact information of a registered SFP provider.(117) These results are consistent with participants’ assessment of the main messages of the information campaign. When asked what they remember the most about the intervention video, 27 percent of women in the treatment group mention that they remember how to sell to the SFP, 25.5 percent mention the products bought by schools, and 32 percent mention some of the video’s recommendations on how to improve product quality – while only 5.5 percent indicate that they remember having learned what the SFP is (cf. results reported in Annex 4). Figure 6. Treatment effect on information intake about the School Feeding Program 1.2 1.0 Proportion of women 0.8 0.818* 0.6 0.880 0.920 0.4 0.750 0.332 0.687*** 0.573 0.424* 0.498* 0.2 0.411 0.238*** 0.138 0.0 Aware of the SFP Knows SFP buys from Knows that can register Knows the steps to Knows which products Has contact info local farmers as a provider register the schools buy of a provider Control Treatment Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. The delivery of information proved especially relevant for individuals that are not reached by traditional extension programs, which highlights the potential of digital tools in promoting information and knowledge diffusion to remote and marginal populations. Among those women who reported not receiving agricultural training or technical assistance over the year prior to the interview (Figure 7, panel a), the information campaign increased 117 Although knowledge about the SFP is self-reported, several pieces of evidence suggest that social desirability bias (i.e., that women are not answering favorably to sympathize with the interviewer) is not the main driver of the results. Results in this section indicate that: (i) participants in the treatment group remember specifics of the video when prompted open-ended questions about it; (ii) other self-reported answers, such as willingness to participate in the SFP, were not answered favorably by participants; (iii) the intervention also affected behavioral margins such as sales. 61 DIGITAGRO 2022 awareness about the existence of the SFP by 10.2 percentage points, increased the likeliness of being aware that the SFP buys products from local farmers by more than 60 percent, raised awareness that farmers can register as providers by 15.3 percentage points, and almost doubled knowledge about the registration steps. For those already reached by traditional extension mechanisms (Figure 7, panel b), however, the information campaign still had a statistically significant effect on the likelihood of knowing the steps to register as a provider (which improved by almost 60 percent). This result shows that, while certainly not a substitute for, digital tools can still be a useful complement to traditional extension services, supporting the more efficient dissemination of simple messages among large audiences. Figure 7. Treatment effect on SFP information intake by participation in traditional extension programs a. Did not participate b. Participated 1.2 1.0 Proportion of women 0.8 0.880 0.695 0.6 0.686 0.445 0.926* 0.924 0.4 0.824 0.709*** 0.526** 0.200* 0.485 0.2 0.441 0.373 0.172 0.274* 0.102 0.0 Knows about Knows SFP buys Knows that can Knows the steps Knows about Knows SFP buys Knows that can Knows the steps SFP market from local farmers register as SFP to register SFP market from local farmers register as SFP to register Control Treatment Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. Box 8 Did DIGITAGRO promote SFP inclusion of Indigenous Peoples? Indigenous Peoples represent an essential part of the demographics in rural San Marcos. According to the 2018 Guatemalan national census, 36 percent of San Marcos inhabitants recognize themselves as Maya, and 93 percent of these pertain to the linguistic Mam community: accordingly, around 23 percent of the rural po- pulation of San Marcos speaks Mam. In particular, the Mam community is predominant in the municipalities of Concepción Tutuapa, Comitancillo, and San Miguel de Ixtahuacán, where on average almost 89 percent of the population speaks Mam. Also, Tajumulco, Rio Blanco, Ixchiguan and San Lorenzo have a considerable population of Mam speakers (37 percent on average). While it is not uncommon for Mam speakers to also understand Spanish, DIGITAGRO videos and inter- views were made available in Mam to encourage the participation of the Indigenous community (cf. sec- tion 4.2 for more detail). Overall, around 20 percent of participating farmers expressed a preference for the baseline and endline interview to be conducted in Mam and for receiving DIGITAGRO videos in Mam. Even though the sample was not designed to be representative of the ethnic composition of the population, it is thus possible to qualitatively estimate the effects of the pilot over the Indigenous popu- lation by using a farmer’s request to be interviewed in Mam as a proxy for self-identification as Mayan. 62 DIGITAGRO 2022 When comparing treated and control individuals in the Indigenous sub-sample (Figure 8), the data point to a positive although statistically insignificant impact on knowledge of the existence of the SFP, that local farmers can sell to it, and that they can register to do it. Similarly, farmers in the treatment group seem to have acqui- red positive but statistically insignificant information, vis-à-vis the control group, about the products bought by the program or about the contact of registered providers. On the other hand, the information campaign had a positive, strong, and significant effect among Indi- genous participants on the knowledge of the registration process in the SFP. The intervention signi- ficantly increased the perception about the difficulty of the process: while no Indigenous person in the control group thinks that the registration process is easy, in the treatment group almost 7 percent of In- digenous farmers do. Moreover, the information campaign more than tripled the likelihood of knowing the steps to register in the SFP. This effect is twice as high in magnitude than the one obtained in the overall sample: considering that the baseline knowledge about the registration process is substantia- lly lower among Indigenous Peoples than in the overall sample (among Indigenous, only 8 percent of the control group possesses this knowledge, whereas in the overall sample’s control group it is 14 percent), it could be argued that the intervention might be a driver of convergence of information availability be- tween Indigenous and non-Indigenous farmers, contributing to levelling the playing field in rural areas. Figure 8. DIGITAGRO treatment effect among Indigenous beneficiaries 1.2 1.0 Proportion of women 0.8 0.6 0.754 0.4 0.818 0.869 0.333 0.274** 0.682 0.519 0.2 0.559 0.404 0.435 0.068** 0.077 0.382 0.000 0.0 Aware of the SFP Knows SFP buys from Knows that can Knows the steps Knows which products Has contact info Thinks registering local farmers register as a provider to register the schools buy of a provider is easy Control Treatment Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level In fact, the whole information campaign was designed to enhance complementarity between digital technologies and traditional extension services, as both the video and text messages mentioned the MAGA and its field official multiple times and consistently encouraged participants to reach out to the MAGA to learn more about the specifics of the SFP, even providing the contact information of local MAGA extensionists. The results of the impact evaluation show that this approach was successful, as participants in the treatment group reported changes in the main source of their information about the SFP: while 5.2 percent of the treated sample reported the 63 DIGITAGRO 2022 information campaign among their main sources of information about the program(118), almost a third of the treated group report it is through the MAGA that they know about the SFP, an increase of over 38 percent vis-à- vis the control group (Figure 9). Importantly, the treatment group is also 20 percent less likely than the control group to report that they know about the SFP through schools, OPFs, or teachers, who were the most prominent pre-treatment information channels. Taken together, these results are noteworthy as they indicate a shift in the main information source and that the intervention is helping farmers receive the correct information from the entity formally in charge of their participation in the SFP, promoting competition and transparency in a context of strong information asymmetries. Figure 9. Treatment effect on main information source about the SFP 0.7 0.6 Proportion of women 0.5 0.4 0.3 0.275** 0.576 0.458*** 0.199 0.2 0.354 0.350 0.316*** 0.052*** 0.1 0.199 0.000 0.0 School/Teacher/OPF Other sources MAGA or Videos&SMS MAGA Videos&SMS Control Treatment Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level Sales Other than on their knowledge about the SFP, the information campaign had an effect on women agri-preneurs’ selling decisions. Having been identified as potential suppliers of the SFP, all participants in the experiment were engaged in some sort of commercial agriculture or agribusiness the year prior to the interview. As such, the percentage of women in the control group that reported selling their products in the two months prior to the interview was already high (83.8 percent in the control group), and the intervention did not have an impact on the likeliness of reporting selling agricultural or animal products per se (Figure 10). Results suggest, however, that the intervention prompted a 20 percent increase in the likeliness of selling animal products demanded by the SFP (mostly those demanded by the program in its traditional, pre-pandemic format) during the months between baseline and endline data collections.(119) This result can arguably be considered a lower bound of the intervention’s potential effect, considering that the COVID-19 pandemic imposed mobility restrictions and health constraints, and reduced the pool of products demanded by the SFP, which is very likely to have negatively affected the sales made by agri-preneurs. 118 No participants in the control group reported the videos as a source of information about the SFP. This is important as it suggests little scope for contamination of the control group. 119 As described in Chapter 3, the SFP took several measures to adapt to the COVID-19 pandemic. The change of the SFP to only buy semi- perishable products resulted in reducing the number of items that schools could purchase (from 67 to 20 options). For example, highly perishable fruits like pineapple, apples, and melons were left out, as well as vegetables like celery or spinach. Also, except for eggs, the SFP suspended buying chicken meat and livestock products like cheese. The series of text messages listed only COVID-19 SFP agricultural products along with their prices. 64 DIGITAGRO 2022 These findings are encouraging, as they show that light-touch digital interventions might change business-related decisions. On the other hand, the fact that effects are detected for animal and not agricultural products is quite intuitive: in light of the short time interval between treatment and endline interview, one could expect faster adaptation in animal products given their availability and readiness to be sold, vis-à-vis agricultural products that necessarily need to follow the rhythms of agricultural cycles. Considering that women in San Marcos tend to specialize precisely in the production of animal products such as eggs and chicken meat (cf. also section 5.2), traditionally demanded by the SFP, it is also likely that the intervention had an effect in encouraging them to capitalize on their work on products where they already had a comparative advantage. Figure 10. Treatment effect on women’s sales 1.0 0.9 0.8 Proportion of women 0.7 0.6 0.5 0.4 0.838 0.805 0.3 0.457* 0.2 0.364 0.380 0.404 0.347 0.341 0.282 0.314 0.1 0.0 Any agricultural or Traditional SFP Traditional SFP animal COVID-19 SFP COVID-19 SFP animal product agricultural product product agricultural product animal product Control Treatment Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. These effects are more pronounced for women who are more empowered at taking business decisions. Looking at the animal products whose sales increased as a result of the intervention, stronger effects are observed among women that at baseline claimed they were participating in their household’s decision-making process about sales and marketing of their produce. The evidence suggests that both less empowered and more empowered women are more likely to sell eggs, although the estimates are not statistically significant. Nevertheless, treated women with more agency in household decision-making are 7 percentage points more likely to report an increase in chicken meat sales than in the control group. For those not involved in decision making at baseline, the estimate of the effect is three times lower (see Figure 11, comparison between panels a and b). The observed impacts on sales are also driven by partnered women. While the digital campaign increased by 48 percent the likeliness of selling SFP animal products among partnered women (cf. the difference between treated and control women in Figure 12, panel a), there is no statistically significant evidence of sales increases for single women (panel b of Figure 12). The effect for partnered women is particularly strong for eggs and especially chicken meat, whose sales more than doubled as a result of the intervention. A likely explanation for this pattern is that, in general, partnered women might have a comparative advantage relative to single farmers, due to their traditional role as helpers in their partners’ agriculture businesses, which might make them more familiar with the production processes and selling mechanisms constituted by their partners. Conversely, single women may have less experience with agribusiness and need more time and effort to establish themselves as entrepreneurs, in the absence of the support from a male relative. 65 DIGITAGRO 2022 Figure 11. Treatment effect on women’s sales by agency level (animal products) a) Participates in decision-making b) Does not participate in decision-making 0.7 0.6 0.5 Proportion of women 0.4 0.418 0.3 0.2 0.353 0.431 0.125** 0.318 0.114 0.1 0.032 0.080 0.025 0.055 0.005 0.001 0.093 0.047 0.000 0.000 0.0 Eggs Chicken meat Cow cheese Cow meat Eggs Chicken meat Cow cheese Cow meat -0.1 Control Treatment Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. Figure 12. Treatment effect on sales of SFP animal products, by marital status a) Partnered b) Single 0.7 0.6 0.5 Proportion of women 0.4 0.3 0.543 *** 0.486 *** 0.415 0.2 0.367 0.366 0.332 0.333 0.343 0.1 0.066 0.164 *** 0.042 0.031 0.041 0.004 0.000 0.073 0.065 0.037 0.000 0.000 0.0 SFP animal Eggs Chicken meat Cow Cheese Cow meat SFP animal Eggs Chicken meat Cow Cheese Cow meat -0.1 product product -0.2 Control Treatment Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 66 DIGITAGRO 2022 Importantly, partnered women are more likely to be empowered and increase their participation in household decision-making around sales of animal products by 10 percentage points as a result of the intervention (Figure 13). (120) (121)However, they do not participate more in the decisions on sales of agricultural products. This is consistent with the evidence that sales of agricultural products remained unchanged by the intervention. Figure 13. Treatment effect on women’s decision-making around sales of agricultural and animal products, by marital status – IPW results a) Total a) Partnered 1.2 1.0 0.8 Proportion of women 0.6 0.932 0.938 ** 0.4 0.800 0.836 0.836 0.833 0.654 0.662 0.659 0.684 0.618 0.637 0.2 0.0 Agricultural Animal Agricultural Animal Control Treatment (complete data) Treatment (IPW) Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 5.3.2 The intervention did not convince many women to register in the SFP Despite these positive outcomes, however, the effect of the intervention on promoting women’s participation in the SFP was limited. Women also kept reporting a number of barriers to participation in the SFP, which were not affected by the information campaign. Willingness to participate in the SFP market Despite the success at delivering information about the School Feeding Program and encouraging women to increase their sales, the intervention did not have an effect on the willingness of participants to join the SFP. In Figure 14, there are no statistically significant differences between treated and control women in the likeliness of 120 Considering that questions on these outcomes were only posed to women with sales in agricultural and animal products, and the increase in sales of animal products was driven by partnered women and women already participating in decision making at the baseline, there is a risk that these estimates may suffer from selection bias. To account for this possibility, the effects on decision-making outcomes were re-estimated using an Inverse Probability Weighted Estimator (IPWE) that weight observations on the outcome variable by the inverse of the probability that it is observed to account for the missingness process (for a detail explanation see Wooldridge, 2007). The results of the exercise suggest that missing data do not seem to introduce significant selection bias for estimating the treatment effect on women’s participation on sales decision-making. In fact, there appears to be a significant increase percent in the likelihood of participating in selling decisions after treatment (Figure 13). 121 Apart from women’s participation in sales and marketing decision-making, the survey collected also information for other six empowerment markers in terms of women’s participation on the household’s assets purchase, children’s healthcare, and relative’s support and aspects of women’s freedom on mobility, spending and socialization. The intervention did not affect these markers. 67 DIGITAGRO 2022 being registered as SFP providers, of selling to a registered SFP provider, nor of planning to speak in the future with one.(122) Although not participating in the SFP might be explained by the little time elapsed between the treatment and endline survey (2 months), lack of willingness to participate in the future suggests that this is not the only factor driving the results. Figure 14. Treatment effect on willingness to participate in the SFP 0.5 0.4 0.4 Proportion of women 0.3 0.3 0.2 0.316 0.326 0.2 0.202 0.1 0.193 0.205 0.176 0.1 0.0 Interested in registering Sold crops to a registered SFP provider Plans to speak to a registered SFP provider Control Treatment Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. Importantly, the intervention did not affect the perception of the main challenges that producers face when selling their products. As reported in Figure 3, the most important problems faced by women at baseline were transporting the harvest, not finding where to sell their products, and low prices. In Figure 15, it is apparent that the intervention did not affect the perception on any of these challenges, as women in the treatment group are statistically equally likely to report these barriers as women in the control group. In practice, this suggests that, even after receiving information on the SFP, women farmers do not see the SFP as a way of easing the obstacles they face when trying to sell their produce. The next section examines the main challenges farmers face to join the SFP. 122 No effect was found either after re-estimating these outcomes using the IPWE (see Error! Reference source not found.). Figure 15. Treatment effect on selling barriers 0.3 0.3 0.2 Proportion of women 0.2 0.212 0.1 0.192 0.185 0.166 0.142 0.1 0.104 0.0 Low sale price Problems transporting Could not find where to sell the harvest their products Control Treatment 68 Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. Chapter 6 Why are women not entering the SFP market? An exploration of barriers to participation 69 DIGITAGRO 2022 Why are women not entering the SFP market? An exploration of barriers to participation Key messages » An important barrier hampering farmers’ incentives to register as SFP providers is the need to supply a diverse portfolio of products throughout the year. » Another challenge has to do with managerial and agricultural skills: aspiring providers are often not aware of food quality standards and proper product storage conditions, and display low levels of technification in production. » The SFP registration process is still perceived as difficult, and this concern is compounded by low levels of trust in the institutions involved in the program. Aspiring providers fear especially the Tax Authority, due to persistent informality in rural areas, and a belief that taxation might be too onerous. » While reference SFP prices tend to be aligned with local market prices, they may not be enough to com- pensate for the extra costs (product quality, continuous supply) that women would incur by selling to the SFP. The mechanics of the SFP institutional life-cycle also sometimes give rise to price speculation on local markets against smaller producers. » Recommendations for promoting women’s participation in the SFP include a blend of digital and analog strategies, to overcome barriers in terms of awareness, empowerment, agricultural production and skills, market structure, institutional challenges. To understand why women are not interested in participating in the School Feeding Program, despite receiving information on how to do so, this chapter studies the main barriers encountered by farmers to participate in the program. The analysis focuses on production capacity, technical skills and agricultural practices, trust in the program and government institutions, and prices. In particular, to shed light on the barriers faced by farmers and study the differences between those who choose to participate and those who do not, the analysis compares the characteristics of farmers who are registered in the SFP with those who are not registered. In order to assess the status quo of these barriers abstracting from any potential impact of the information campaign, the analysis in this chapter is mostly performed on the control group unless otherwise specified. 6.1 Product portfolio and land size An important barrier hampering farmers’ incentives to register as SFP providers seems to be the need to supply a diverse portfolio of products throughout the year. While there is no established rule on the number of providers that can cater to each school, focus groups with school parents organizations (OPFs)(123) reveal that schools prefer to work with one single registered provider with the ability to cater to the entire school and meet the school demand for products. Qualitative evidence suggests that this simplifies the process for schools and reduces transaction costs. 123 See Chapter 3 for an in-depth description of the program’s organizational structure and main actors. 70 DIGITAGRO 2022 As a matter of fact, looking at farmers in the control group in Figure 16, it seems that the scale of registered SFP providers is higher than that of regular producers, as can be inferred by the distribution of land sizes among registered providers lying to the right of that of unregistered producers (panel a). The median land size of registered households (7 cuerdas) more than doubles that of unregistered producers (3 cuerdas), suggesting that their production capacity might be higher than that of regular farmers. The land distribution remains favorable to registered providers even when controlling for the different climatic areas of the San Marcos department (Figure 16, panel b). Figure 16. Distribution of farmers’ land size by registration status in the control group a) Overall San Marcos b) By climatic areas .15 Coastal-central area Plateau area .15 kdensity Land size (cuerdas) kdensity land size (cuerdas) .1 .1 .05 .05 0 0 0 10 20 30 0 10 20 30 0 10 20 30 Land size of control group Land size of control group Not Registered Registered Not registered Registered Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. The hypothesis that registered providers may have a larger scale and productive capacity than unregistered ones is also supported by Figure 17, where the former are far more likely than the latter to sell more than one product. In addition, comparing the sales of agricultural SFP products reported by the two categories of producers in Figure 18, registered providers appear to count with a broader availability of products throughout the year, as can be observed by their higher probability of having sold any of the listed SFP products both during the year before the baseline interview (left panel) and during the two months before the endline survey (right panel). Other than the mix of products, in fact, there seem to be challenges for unregistered producers also in terms of product availability and readiness through the year. This is particularly evident, for instance, in the case of corn, a product in high demand by schools and an integral part of the Guatemalan diet: while unregistered and registered producers were equally likely to have sold the crop over the course of the year prior to baseline (cf. the almost identical percentages in the left panel of Figure 18), in the two months prior to endline the SFP providers were six times more likely to have sold it than their unregistered counterparts (12 percent vs. 2 percent in the right panel of the figure). Arguably, for farmers with low production levels or limited storage capacity, the inability of selling the entire range of foods to schools in a timely manner may represent an important impediment to joining the program. 71 DIGITAGRO 2022 Figure 17. Likeliness of selling one or more products, by SFP registration status 90 Percentage of women control group 82 80 70 61.2 60 50 38.8 40 30 18 20 10 0 One More than one Number of products Registered Not Registered Note: Estimates as simple means of women in the baseline in the control group. Figure 18. Share of sales of SFP agricultural products by registration status and reference period a) Last year (baseline) a) Last two months (endline) Not Registered Registered Not Registered Registered Limes 0.034 0.080 Limes 0.038 0.020 Black Bean 0.121 0.080 Black Bean 0.019 0.080 Corn 0.159 0.160 Corn 0.023 0.120 Onion 0.053 0.160 Onion 0.023 0.080 Tomato 0.045 0.200 Tomato 0.023 0.080 Carrots 0.087 0.260 Carrots 0.034 0.120 Potato 0.125 0.320 Potato 0.045 0.240 0.200 0.100 0.000 0.100 0.200 0.300 0.400 0.100 0.050 0.000 0.050 0.100 0.150 0.200 0.250 0.300 Percentage of women in the control group Percentage of women in the control group Note: Estimates as simple means of women in the control group. 72 DIGITAGRO 2022 6.2 Skills needed to join the SFP On top of the issues of product mix and production capacity, women considering registering as a SFP provider seem to also encounter barriers in terms of managerial and agricultural skills. Focus groups with MAGA officials and extensionists in San Marcos suggest that aspiring providers often lack certain managerial skills that are indeed necessary to succeed in the program. Most farmers do not have a clear understanding of the SFP supply chain and lack basic planning and bargaining skills. The perception of MAGA’s extensionists is that farmers would need to improve their administrative and marketing skills to successfully participate in the SFP. Furthermore, besides administrative skills, the SFP also enforces specific food quality(124) and safety standards which require a certain level of technical skills and adoption of good agricultural practices. At first glance, farmers registered in the SFP do appear to be more aware of food quality standards and proper product storage conditions. Before the intervention, 81.6 percent of registered farmers in the baseline reported they stored their products in clean and dry places, against 67.1 percent of non-registered women. When women in the control group were asked about how they packaged eggs, 91.8 percent of registered women replied one of the correct methods (secure packaging in cardboard, baskets, or boxes to avoid product loss), vis-à-vis 85.2 percent of those non-registered.(125) Figure 19. Adoption of agricultural practices, by SFP registration status 1.0 0.9 0.8 Proportion of women 0.7 0.6 0.5 0.92 0.89 0.4 0.76 0.68* 0.71 0.3 0.6 0.5*** 0.48** 0.2 0.42 0.26*** 0.1 0.1 0.03** 0.02 0.02 0.0 Natural Chemical Pesticides/ Improved Technified Machinery None fertilizers fertilizers herbicides Seed irrigation system Registered No registered Note: T-test of the differences in the means across the groups in the baseline. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 124 For example, the fruits and vegetables that the schools buy must be of uniform and shiny color, fresh scent, good texture, and uniform size. Also, eggs must be clean with a good smell, not show any crack, and be packed using cardboard. Finally, to comply with packaging, crops must be put in net-type mesh sacks in a specific order (grains at the bottom, tubercles at the middle, and fruit and vegetables on top. 125 Although the summary video shown during the information campaign did not focus extensively on the agricultural practices to comply with SFP standards (which were instead addressed in the digital extension suite), it depicted certain practices and general information regarding the quality standards demanded by the program to improve the overall quality of the products. These recommendations included simple messaging on the requirement that crops must not have bruises or signs of decomposition or infection, nor contain impurities or damage by plagues or diseases, and should be stored in clean and dry spaces. Also, the video pointed out that eggs must be clean and stored in proper packaging such as carboard, baskets or boxes. However, the impact evaluation analysis did not find evidence that these simple messages changed farmers’ agricultural practices, pointing to the need of complementing that information with more specific guidance and instructions. 73 DIGITAGRO 2022 Evidence also suggests that registered women enjoy higher levels of technification in their agricultural activity compared to unregistered farmers. At baseline, women registered in the SFP reported to use more specialized inputs, machinery and technical infrastructure than their unregistered counterparts (Figure 19). The sharpest differences can be observed in the use of pesticides/herbicides, improved seeds, and technified irrigation systems: registered farmers are 21 percentage points more likely to use pesticides or herbicides in their agricultural process, 12 percentage points more likely to use improved seeds, and 16 percentage points more likely to use technified irrigation systems. Although the use of machinery is very low in the entire sample, registered farmers are three times as likely to use it than unregistered producers. Overall, registered women seem to be those that already have a set of technical skills and adoption of agricultural practices that facilitates their participation in the program. 6.3 Registration process and institutional trust Alongside questions of production capacity and skills set, aspiring SFP providers also run into a number of institutional roadblocks. Firstly, while the information campaign was effective at delivering information about the registration process, as outlined above, in Figure 20 it is obvious that, even after the treatment, very few farmers perceive that the registration process is easy. The intervention in fact did increase the share of individuals that report knowing how to register as a provider (from 13.8 percent in the control group to 23.8 percent in the treatment group), and it more than doubled the share of individuals who consider that registration is easy (from 5.7 percent in the control group to 12.6 percent in the treatment group): nonetheless, these percentages are still pretty low, speaking to an intrinsic difficulty of the process for the vast majority of women. Figure 20. Treatment effect on registration process awareness and perceptions 0.4 0.3 0.3 Proportion of women 0.2 0.2 0.238 *** 0.1 0.138 0.126 ** 0.1 0.057 0.0 Knows the steps to register Thinks registering is easy Control Treatment Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. As mentioned in Chapter 3, to be registered as SFP providers, producers must meet two fundamental requirements: (i) be classified as small commercial producers in the official MAGA Family Agriculture registry; and (ii) be registered with the Superintendency of Tax Administration (SAT) to have a tax identification number (NIT) and be able to issue invoices. During focus groups, producers also expressed their concern about the potential cost of the process, as many would rely on professional accountants to take care of the related paperwork and follow-up. To promote registration, the process might need to be further simplified and streamlined, and farmers may need closer assistance from field extensionists to overcome their discomfort with institutions they know little about. 74 DIGITAGRO 2022 According to an ad hoc survey carried out among registered providers in San Marcos, 65 percent of interviewed providers mentioned that the registration process was easy once they obtained detailed registration support from MAGA extensionists. During focus groups with MAGA officials, extensionists confirmed that a high share of farmers do not understand the process nor know the main government institutions involved in the SFP (MAGA, MINEDUC, and SAT), with particularly acute challenges related to the understanding of taxation issues and fear of the Tax Authority. In effect, women in the control group report lower levels of trust in the SAT than in the MAGA and MINEDUC: from an index that ranks trust in each institution from 1 to 4 (with 1 meaning no trust at all and 4 meaning very much trust), the MAGA and MINEDUC score around 2.9 on average, while the SAT scores 2.4 (see green bars in Figure 21, panel d). Interestingly, more than 39 percent of the control group preferred not to answer the question whether they trusted the SAT, against only 10.7 and 7.5 percent for MAGA and MINEDUC, respectively (Figure 21, panels a-c). According to focus groups and field interviews, aspiring providers fear the Tax Authority because, due to persistent informality in rural areas, they have never had any contact with it before, and believe taxation might rip out the benefits from selling to schools. Figure 21. Treatment effect on institutional perception a) MAGA b) MINEDUC 0.6 0.6 0.5 Proportion of women 0.5 0.4 0.4 0.3 0.3 0.2 0.405 0.4 0.2 0.3 0.273 0.305 0.305 0.314 0.141* 0.262 0.1 0.2 0.1 0.2 0.189 0.047 0.107 0.149 0.055 0.068 0.081 0.075 0.144*** 0.0 0.0 Very much Somewhat Little None Do not know Very much Somewhat Little None Do not know c) SAT d) Trust Index¹ 0.6 4.0 3.5 Proportion of women 0.5 3.0 0.4 2.5 3.084* Score 2.862 0.3 2.0 0.394 2.574* 1.5 2.935 2.860 0.2 0.377 2.359 1.0 0.1 0.208 0.213 0.159 0.5 0.13 0.141* 0.13 0.146*** 0.108*** 0.0 0.0 Very much Somewhat Little None Do not know MAGA Trust index MINEDUC Trust index SAT Trust index Control Treatment 1 The index does not include the category don’t know/no response and may be underrepresenting population with an education level lower than primary (54 percent) and in the plateau area (51 percent). Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 75 DIGITAGRO 2022 As such, promoting trust in the key institutions involved in the SFP might be an important driver of registration. Importantly, at an aggregate level the digital information campaign seems to have slightly improved women’s self-reported trust levels in the MAGA and SAT, with promising results especially for the SAT (Figure 21, panels a, c, and d). However, this seems not to have been sufficient to induce women to initiate contact with these two institutions to register in the SFP: in Figure 22, women in the treatment group are not more likely than those in the control group to have spoken with a MAGA official, registered to get a Tax Identification Number (NIT), nor become able to issue invoices. Figure 22. Treatment effect on contacts with MAGA and SAT 0.6 0.5 Proportion of women 0.4 0.3 0.506 0.469 0.2 0.263 0.282 0.1 0.193 0.205 0.0 • ation Number Has Tax Identi• c Can issue invoices Sold crops to a registered SFP provider Control Treatment Note: Estimates from regression analysis. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 6.4 Prices A final angle worth looking at to describe women’s low incentives towards SFP participation has to do with the expected profitability of the SFP as a market opportunity. In fact, although prices do not feature prominently among the major barriers to SFP registration mentioned by farmers, the information campaign (which disseminated information on SFP reference prices(126) set by the MINEDUC and MAGA) resulted in 5 percent of producers in the treatment group mentioning low prices as a significant obstacle. This raises the question whether those reference prices might be considered by prospective SFP applicant women so low as to discourage them to register in the program. At first glance, nevertheless, it seems unlikely that the reference prices of the SFP could be the main disincentive to participate in the program. Note that those reference prices are only one of three criteria considered by providers to determine the actual prices charged to schools for each sale, which also include agreement with other providers in the same municipality as well as some adjustment to local market prices. (127)In addition, reference prices per se do not seem to be bad compared to the prices that farmers could get on the market. 126 As described in Chapter 4, the information on the reference price for each product sent via SMS to the treatment group was mainly collected using administrative records of receipts paid by the schools to different providers in different municipalities of the area of study. For those municipalities where it was not possible to collect this information, the average price of the zone where the municipality is located was used. 127 More than half (62 percent) of interviewees in the ad hoc SFP providers’ survey report that they agree with other providers on the price charged to schools, and 82 percent say they speak with MAGA to decide prices. 76 DIGITAGRO 2022 Taking the example of eggs, SFP reference prices are aligned to the average prices women in the control group reported obtaining on the market during the period of the intervention: 52.2 percent of registered providers and 44 percent of unregistered producers actually obtained lower prices in the market (Figure 23, panel a). A similar pattern can be observed when looking at other SFP agricultural products (Figure 23, panel b). Thus, it is unlikely that low reference prices alone would explain the lack of interest in registration among non-registered farmers. Figure 23. Distribution of price differential between SFP and market prices, by registration status a) Eggs b) Agricultural products No registered Registered Not registered Registered .4 2 kdensity price differentials kdensity price differentials 1.5 .2 1 .5 0 0 -1 0 1 -1 0 1 -4 -2 0 2 4 -4 -2 0 2 4 Differences in unit prices of eggs. SFP vs Market Differences in unit prices of agri products*. Graphs by HH registered to sell crops to the SFP SFP - Market prices * Agricultural products include black beans, carrots, potatoes, and onions. Note: Estimates as simple means of women in the control group. Nevertheless, higher prices would increase the benefits to registration, especially in the presence of other barriers such as those highlighted earlier in this section. Focus group interviews point out that low prices might indeed hinder incentives to join the SFP program. As mentioned, the reference price formation process is the outcome of bargaining between MAGA and MINEDUC, but existing frictions between institutions might push the resulting price down, diminishing the SFP profitability for providers. For example, schools have the clear incentive to obtain the most out of their SFP budget, which brings the MINEDUC to lobby for lower prices. According to MAGA officials, in addition, the reference prices consider the prices charged by producers in the local market as a benchmark for the SFP, but these do not account for the food safety, transportation, and taxation costs borne by faraway producers. A related concern is that reference prices do not consider the cost differences between local varieties of staple crops in different climatic zones. For instance, corn and beans in the highlands have higher production costs than in the coast and central area but are sold in local markets alongside products obtained from the other two areas: thus, considering average local market prices decreases the profit margins from selling higher-cost local varieties. Finally, for given reference prices, the squeezed time frame within which schools demand food reduces producers’ profit margins by increasing their costs (especially transport costs), as recognized by some interviewed OPFs from rural schools. In addition, the cyclicality of schools’ food procurement processes means that schools are often all buying on the same dates. This can in turn be problematic for providers that have to buy and resell products from other producers to procure the entire food range demanded by schools, as there is often not enough produce to serve all of them: given fixed SFP reference prices, this gives rise to price speculation on local markets against providers, and some interviewed MAGA officials and OPFs shared the concern that providers might be often taking prices at a loss. Consistently with this perception, 25 percent of interviewees in the providers’ survey 77 DIGITAGRO 2022 declared that high prices for produce they had to buy from the market were one of their main concerns when selling to the SFP. 6.5 Recommendations: How to promote women’s participation in the School Feeding Program? The analysis in this chapter shows that, in order to be able to join the SFP, women agri-preneurs need to confront a diverse set of barriers that may or may not be overcome by information diffusion and digital technologies alone. An obvious first obstacle is that women tend to ignore that the SFP can be a business opportunity for them, which the digital information campaign proved to be an effective tool at generating awareness about. Business opportunities, however, cannot be leveraged if women lack empowerment and are deprived of decision- making ability in the economic sphere. While the information campaign increased women’s sales and had an effect on their agency with respect to agricultural production, further interventions are needed to address women’s empowerment more broadly. Gender-transformative approaches that look at changing behavior around gender can have an effect on power dynamics and choices of women producers. In the agricultural sector, for example, the promotion of women’s access to land ownership and formal rights can impact intrahousehold decision making and entrepreneurship.(128) The promotion, formalization, and strengthening of women’s associations, as well as the support to women’s participation in mixed organizations (including in leadership positions), can also be promoted as a catalyst of improved conditions and opportunities for rural women. Sensibilization campaigns carried out at the community or family level and involving male co-villagers or relatives, in turn, can raise awareness about women’s contributions in production and at home, as well as about the benefits of strengthening their participation in productive activities. Beyond empowerment, the analysis in this chapter reveals that many of the issues that stifle women’s engagement in the SFP have also much to do with the challenges broadly faced by smallholder producers in general, in terms of productive structure and skills. Relative to non-registered farmers, SFP providers have more than twice the amount of land, cultivate a higher diversity of products, and use more productive techniques, such as specialized inputs, machinery, and infrastructure. In addition, registered providers seem to be more aware of good food quality and hygiene practices, as well as to possess less rudimentary managerial skills. While digital technologies can certainly help increase technical and managerial skills through e-extension, well- targeted technical assistance is a necessary complement to basic training to ensure adoption of good practices and technologies in rural settings.(129) Moreover, a package of interventions that address more than one barrier at a time, such as information, skills, mentorship, access to productive inputs, might be more successful than interventions addressing a single barrier.(130) A promising approach in this sense could be strengthening the work of the national extension system to promote the formation and strengthening of rural producers’ groups, so as to facilitate their links with commercial partners and technical assistance for productivity enhancement and business development: this could be achieved for instance through an adaptation to the SFP reality of the Productive Alliances model, implemented in several countries in the LAC region and worldwide.(131) Compounding individual-level constraints, there seem to be a number of additional barriers that originate from the very organization of the SFP itself, which limit the ability of the program to recruit more providers. For example, the SFP reference prices seem to be broadly in line with local market conditions, and as such they do not 128 World Bank, 2022. 129 Cf. recent experimental evidence on adoption of climate-smart practices in Bragança et al., 2022. 130 Wold Banck 2022. 131 The Productive Alliances model involves a group of smallholder producers, buyers, and the public sector, and revolves around the implementation of a business plan proposed by producers. Productive investments, technical assistance, and business development are financed through public grants, which are matched by producers’ contributions (which could be in-kind) and in some cases also by buyers or by participating private financial institutions. Cf. World Bank, 2016b for more information 78 DIGITAGRO 2022 account for the extra premium that would be required to compensate for higher food quality standards, taxation, cost differences between local varieties of staple crops in different climatic zones, and transport to schools in remote areas. While the price formation mechanism is a delicate political process involving bargaining between different ministries, certain measures to improve efficiency and competition of the SFP market, such as the SFP e-commerce platform, might improve transparency and price setting. Additionally, actions can be undertaken towards reducing the cost to farmers of participating in the SFP: for example, agrologistics improvement would help addressing several challenges faced by producers, including transport hurdles. Another structural barrier hampering incentives towards registration is the low trust in the institutions involved in the SFP, especially as regards the SAT. While the information campaign resulted in a mild increase in overall institutional trust, the reluctance to formalize and the conviction that taxation would outweigh any benefits from participating in the program are concerning. There seems to be a need for clearer communication on the benefits from formalization (in terms of market access, access to financial products such as credit and insurance, eligibility for government social safety net programs, among others), as well as an effort to strengthen institutional field presence and improve closeness and accessibility to rural communities. Relatedly, there could be scope for further simplifying the SFP registration process. If, on the one hand, the information campaign allowed more women to learn about the process to register, on the other hand still very few consider it to be simple, and some fear it would also be expensive. As perceptions reportedly change when producers are assisted by field extensionists, closer assistance on the ground might help promote SFP registration. Moreover, a possibility to streamline processes could be to institute SFP one-stop-shops counters on the territory led by local authorities such as the CTIMAEs, where producers can carry out all necessary procedures. In parallel, processes could also be digitized in a one-stop-shop web portal directly linked to the e-commerce platform, to avoid multiple registrations and at the same time promote participation in the e-commerce facility. 79 Chapter 7 Discussion and conclusion: Digital technologies and agrifood transformation 80 DIGITAGRO 2022 Discussion and conclusion: Digital technologies and agrifood transformation Key messages » Digital technologies in the agrifood system can serve as a tool to improve market and production efficien- cy, equity, sustainability, and resilience. Their benefits are even more relevant as economies worldwide build forward better in their recovery from the COVID-19 pandemic. » The DIGITAGRO project showed that digital technologies can be leveraged to improve market access for women and improve the efficiency of the national School Feeding Program. However, these technologies need to be accompanied by an effort to rationalize the School Feeding Program itself, and can deploy their full development potential only when complemented by additional investments for a suitable digital-en- abling environment. » Especially in rural areas, digital technologies need complementary investments and policies in terms of infrastructure, creation of skills and social capital, market structure, and access to basic services. Smallholder farmers in Guatemala face information barriers that hinder market opportunities and business productivity, especially for women who overall face more challenging circumstances and lower empowerment than their male relatives and co-villagers. The DIGITAGRO project used digital technologies to provide information about the national School Feeding Program, to promote market access among rural women while supporting the smooth functioning of the program and contributing to improving food and nutrition security for children. While digital technologies have long been regarded as a useful complement to development strategies in the agrifood sector, their effective deployment becomes all the more crucial in a situation where the disruptions of the COVID-19 outbreak have been reverberating all along the food chain and in particular among the most vulnerable – in Guatemala and globally. Especially in countries long plagued by pervasive hunger and nutrition issues, it is of utmost importance that agricultural value chains remain functioning and resilient, and that food keeps flowing from production to consumers: digital technology, if done well, can become a powerful tool in support of food security, food safety, and farmers’ livelihoods. In Guatemala, the e-commerce platform developed by DIGITAGRO and the World Food Programme has been key to ensuring the effective functioning of the School Feeding Program at the height of the pandemic, despite school closures and social distancing, enabling food access on the part of schoolchildren and their families and protecting farmers’ sales and revenue. Even past the emergency phase, the platform will now give farmers the opportunity to access a broader network of customers, and at the same time provide schools with a comprehensive database of agricultural producers that can guarantee a reliable source of safe and nutritious foods for their pupils. Similarly, the extension videos developed by DIGITAGRO and the Food and Agriculture Organization of the United Nations have contributed to the dissemination of rules and standards to promote food safety and avoid food loss and waste, including guidelines on handling, processing, packaging, and storage. Other than for their direct contribution to food safety and preservation, the videos will be crucial for farmers to learn good practices that will help them increase the quality of their produce, enabling them to reach higher-value markets and obtain more favorable prices. The DIGITAGRO digital information campaign, finally, has been instrumental in reaching women farmers in remote areas despite the challenging context of social distancing and mobility restrictions, with simple yet effective messaging on school feeding, local contacts, and market prices. In future, improving the 81 DIGITAGRO 2022 dissemination of market information promises to allow for more efficient production planning, lower production costs, and improved market competition and efficiency. In light of these promising opportunities, the findings from the impact evaluation of the digital information campaign are crucial, in that they allow for a rigorous assessment of the intervention, as well as of the mechanisms that may be driving the observed results. In particular, the impact evaluation study reveals that tools such as the digital information campaign can be used in rural Guatemala to effectively disseminate market information and spur women’s entrepreneurship in agribusiness: women participating in the digital information campaign improved their knowledge about the SFP and started seeing it as an economic opportunity, and the intervention also encouraged them to increase their sales of products of animal origin. Notably, the intervention had a higher impact on SFP awareness among women that had not been previously reached by traditional extension services, in practice closing the information gap between receivers and non-receivers of agricultural extension – thus highlighting the relevance of digital technologies at delivering information to broader audiences unattended by official programs. The study also provides useful evidence on the link between information and empowerment, as women in the treatment group saw an increase in their participation in household decision-making on agricultural sales. These encouraging results, however, come with some important caveats: despite all their newly acquired knowledge and their likely improved economic status, women who received the digital information campaign were still overly cautious about getting involved in the SFP. At closer inspection, it appears that the program is better suited to comparatively larger producers, with more sophisticated production systems and a higher capacity of supplying a broader pool of products throughout the year: it is overall harder for smaller women producers to succeed in their transition to commercial agriculture, even in a theoretically enabling setting such as the school feeding market. While SFP prices are generally not too bad compared with local market prices, the price formation mechanism also seems to be working in favor of larger producers who are able to leverage economies of scale and scope rather than fetching produce from third parties to meet the full demand of schools. The fact that the digital information campaign had a stronger effect on partnered women and on women with higher agency in household selling decisions, in addition, further confirms the persisting barriers still faced by rural women, if direct or implicit support from their families is necessary to establish themselves as successful entrepreneurs in the agrifood space. This is even more relevant in a context where institutional trust remains low, notwithstanding positive results of the intervention at improving perceptions about the MAGA and Tax Authority, as women are left to rely on less-than-equitable informal channels and mechanisms to support their entrepreneurial aspirations. Jointly considered, these findings suggest that continued work is needed to keep aligning the structure of the SFP to the reality of smallholder producers and women, if the program is to include them more systematically. In particular, these results reaffirm the view that well-targeted extension services and technical assistance are needed to follow up more closely with producers whose capacities and skills may not yet be up to standard for an involvement in the school feeding market. Importantly, producers who do not have access to even traditional extension services face much lower knowledge about basic SFP features: for example, women who receive extension services are 1.6 times as likely, with respect to those who do not, to be aware that the SFP buys from local producers, 1.2 times as likely to know that they themselves can register as SFP providers, and 1.7 times as likely to grasp the registration process (cf. supra Figure 7). Extending the coverage of these services, as well as streamlining their content, will be key to ensuring more sustained participation in the SFP by smallholder producers, and women in particular. As mentioned, digital technologies can certainly help this agenda, thanks to their ability to penetrate among broad audiences and to reduce the cost of remoteness. As a final word of caution, however, it is important to note that any digital solution requires a suitable conducive environment to be effective at scale. In Guatemala, rural electrification rates are very low to start with: 16.3 percent of the rural population lacks access to electricity, 82 DIGITAGRO 2022 four times as much as the urban population and one of the highest shares among peer countries.(132) Moreover, while 62 percent of the country has access to mobile telephones, only 29 percent has internet access, and just 21 percent has access to a computer.(133) On top of already low rates of internet penetration, the country sees a broad digital divide between urban and rural areas and across ethnic groups: departments where the concentration of rural and indigenous population is high present very low rates of internet access, around a third of those observed in large urban areas.(134) Promoting digital skills, especially among women, is also a key priority: according to the 2018 Guatemala National Census, only 67 percent of men and 46 percent of women use a mobile phone at least once in three months, and as few as and 10.5 percent of men and 5 percent of women use the internet. In 2019, the perception of Guatemalan business executives on the level of digital skills in the population placed the country at the bottom half of the LAC region and only fourth among Central American countries, with scores decreasing since 2017 and 2018.(135) Specific digital literacy training would also be needed for Indigenous Peoples, accounting for their language barriers. Finally, while the country has been making progress in strengthening its regulatory environment and enhancing trust in the digital ecosystem, more work needs to be done to be on par with other countries in the Latin America region and in industrialized economies worldwide (cf. Box 9). Overall, digital technologies can support the transformation of food systems, contributing to their inclusiveness, sustainability, and resilience. The DIGITAGRO project showed how they can be leveraged to improve market access for women and improve the efficiency of the School Feeding Program. However, these technologies need to be accompanied by an effort to rationalize the School Feeding Program itself, and can deploy their full development potential only when complemented by additional investments for a suitable digital-enabling environment to flourish. In Guatemala as in many other countries around the world, investing in digital development, strengthening the regulatory environment, increasing rural connectivity and mobile penetration, and promoting digital literacy and skills will have high payoffs. Box 9 Guatemala’s digital ecosystem Guatemala faces challenges in terms of internet regulatory environment, access, affordability, and quality of services, especially in rural areas. Guatemala’s digital regulatory framework is generally weak, with the country ranking 162 in the ICT Regulatory Tracker out of 192 countries. The telecommunications sector, regulated by the Superintenden- ce of Telecommunications, is governed by an outdated General Law of Telecommunications (1996). More recently, the government has passed a decree that regulates digital signatures (2008), an access to infor- mation law (2008), and an Open Data Policy (2018), and has drafted a law against cybercrime currently under Congress review, after launching a national cybersecurity strategy in 2018. A national broadband plan is still work in progress, leaving the current provision of high-speed internet services vastly unregu- lated. Currently, the national development plan K’atun, Nuestra Guatemala 2032 (K’atun, Our Guatemala 2032) and the digital agenda Agenda Nación Digital 2016-2032 (Digital Nation 2016-2032) are the main planning instruments for the digital transformation of Guatemala. The former establishes a commitment to: (i) close the digital gap within public institutions to improve and speed up processes and transactions and generate knowledge within society; and (ii) design, approve and implement policies for digital inclusion. 132 de la Fuente and Gomez, forthcoming. 133 INE, 2019. 134 INE, 2019. 135 World Bank, 2021b. 83 DIGITAGRO 2022 The latter aims to leverage information and communications technology to contribute to the country’s tech- nological, social and economic development. However, there appears to be ample room for improvement on aspects related to cybercrime (in 2020 Guatemala ranked 150 out of 182 countries globally and 24 out of 33 countries in LAC in terms of the Global Cybersecurity Index), privacy, data protection, and competition. Overall, the percentage of households with access to a fixed internet connection is below 10 percent in 229 out of the 341 municipalities of the country, and only 3.52 percent of rural households have access to it. Although 65 percent of the Guatemalan population was using the internet in 2017, there are still high divides in internet usage with a 23 percent gender gap (one of the largest in the LAC region) and high heterogeneity between urban and rural areas (41.52 vs. 14.35 percent). Among internet users, only 44.5 percent of households have fixed broadband (below the 48 percent re- gional average for LAC), whereas the mobile broadband penetration rate is 90.2 percent (below the 95.6 percent average in Central America). With respectively 95 and 88 percent of the population within range of a 3G and 4G mobile-cellular signal (irrespective of whether or not they are subscribers), mobile broadband has become the most common way through which people in Guatemala access the internet. Affordability of services is a major barrier to internet access. The cost of a 1.5 GB subscription in Guate- mala is 8 percent of Gross National Income per capita (the highest in Central America) for mobile broadband and 5.99 percent for fixed broadband, much higher than the 2 percent target recommended by the Broad- band Commission for Sustainable Development. These costs are too onerous for people living in poverty: for a household from the lowest quintile, fixed broadband prices could represent 7.5 percent of monthly income, and mobile broadband 4.1 percent. The cost to acquire a broadband-enabled device is another factor hindering broadband access: the cost of a smartphone in Guatemala represents more than 30 per- cent of the average income, way above neighboring countries like Costa Rica where it stands at 6 percent. In addition, only 21.3 percent of Guatemalan households have a computer, a percentage that plummets to less than 7 percent in rural areas. In terms of quality of services, Guatemala lacks adequate infrastructure and download speed is very low compared to regional leaders, hampering accessibility despite having access to services. This is mainly be- cause of the high market concentration in both mobile and fixed broadband, with two operators with more than 95 percent of the market share in both segments, which stifles incentives for market shareholders to deploy infrastructure. Sources: World Bank, 2021b; OECD, 2020; ITU, 2021. 84 DIGITAGRO 2022 References » Aker, J. and Fafchamps, M. 2015. Mobile phone coverage and producer markets: Evidence from West Africa. The World Bank Economic Review, 29(2): 262-292. » Aker, J. C., Ghosh, I., and Burrell, J. 2016. The Promise (and Pitfalls) of ICT for Agriculture Initiatives. Agricultural Economics, 47(S1): 35-48. » Al-Hassan, R., Egyir, I., Abakah, J., 2013. Farm household level impacts of information communication technology (ICT)-based agricultural market information in Ghana. Journal of Development and Agricultural Economics, 5(4): 161-167. » Alvizures, A. 2020. Mineduc prioriza alimentación escolar para aumentar la cobertura en 2021. TGW La voz de Guatemala (radio). 8 July 2020. » Ponce, A. and Arellano, H. 2017. Lineamentos estratégicos para el Fortalecimiento de la agricultura familiar y la inclusión en Guatemala. Inter-American Institute for Cooperation on Agriculture. » Banco de Guatemala. 2019. Guatemala en Cifras. Departamento de Estadísticas Macroeconómicas, Banco de Guatemala. » Beuermann, D., McKelvey C., and Vakis R. 2012. Mobile phones and economic development in rural Peru. The Journal of Development Studies, 48(11): 1-12. » Bragança, A.A., Newton, P., Cohn, A.S., Assunção, J., Camboim, C., de Faveri, D., Farinelli, B., Perego, V.M.E., Moraes Tavares, M., Santos Resende, J.C., Almeida Filgueira de Medeiros, S., and Searchinger, T.D. 2022. Extension Services Can Promote Pasture Restoration: Evidence from Brazil’s Low Carbon Agriculture Plan. Proceedings of the National Academy of Sciences of the United States, 119(12). » Ceballos, F, Hernandez, M, Paz, Cynthia. 2021. Survey: Short-term impacts of COVID-19 in rural » Guatemala call for a closer, continuous look at the food security and nutritional patterns of vulnerable families. Washington, D.C.: International Food Policy Research Institute. » Congreso de la República de Guatemala, 2021a. Cobertura Del Programa De Alimentación Escolar Llegaría A 3.6 Millón De Estudiantes. Guatemala: Congreso de la República de Guatemala. https://www.congreso.gob.gt/ noticias_congreso/7233/2021/4#gsc.tab=0 » Congreso de la República de Guatemala, 2021b. Reformas a la Ley de Alimentación Escolar, Decreto número 16-2017 del Congreso de la República. Decreto 12-2021. Guatemala: Congreso de la República de Guatemala. » Cole, S. and Fernando, A. 2012. The value of advice: Evidence from mobile phone-based agricultural extension. Working Paper 13-047, Harvard Business School, Harvard University. » de la Fuente, A., and Gomez, C. forthcoming. Multidimensional Poverty in Guatemala, 2014 2018. Poverty and Equity Global Practice. Washington, DC: The World Bank. » Deichman, U., Goyal, A., and Mishra, D. 2016. Will digital technologies transform agriculture in developing countries? Agricultural Economics, 47(S1): 21-33. » Eckstein, D, Künzel, V, Schäfer, L, 2020. Global Climate Risk Index 2020. Germanwatch. 85 DIGITAGRO 2022 » Eckstein, D, Künzel, V, Schäfer, L, 2021. Global Climate Risk Index 2021. Germanwatch. » Einav, L., Farronato, C., and Levin, J. 2016. Peer-to-Peer Markets. Annual Review of Economics, 13(52): 615-35. » EM-DAT. 2021. EM-DAT: The International Disaster Database. EM-DAT, Centre for Research on the Epidemiology of Disasters (CRED), Université Catholique de Louvain. Brussels, Belgium. www.emdat.be » FAO. 2011. The state of food and agriculture 2010-2011. Women in agriculture. Closing the gender gap for development. Rome: Food and Agriculture Organization of the United Nations. » FAO, 2016. Guatemala: Voz a las Mujeres sobre Seguridad Alimentaria y Nutrición. Guatemala: Food and Agriculture Organization of the United Nations. https://www.fao.org/in-action/gender-equality-guatemala/es/ » FAO. 2017. Inicia proceso para mejorar igualdad de género en la área rural. Guatemala: Food and Agriculture Organization of the United Nations. https://www.fao.org/guatemala/noticias/detail-events/fr/c/1024621/ » FAO. 2018. Guatemala’s school-feeding law prioritizes child nutrition and family farming. Guatemala: Food and Agriculture Organization of the United Nations. https://www.fao.org/guatemala/noticias/detail-events/ en/c/1103375/ » FAO. 2019. The State of Food and Agriculture 2019. Moving forward on food loss and waste reduction. Rome: Food and Agriculture Organization of the United Nations. » FAOSTAT (Food and Agriculture Organization Statistics). Various years. FAOSTAT Statistics Database. Rome: Food and Agriculture Organization of the United Nations. » Foster, C., Graham, M., Mann, L., Waema, T., and Friederici, N. 2018. Digital Control in Value Chains: Challenges of Connectivity for East African Firms. Economic Geography, 94(1): 68-86. » Gandhi, R., Veeraraghavan, R., Toyama, K., and Ramprasad, V., 2009. Digital Green: Participatory video and mediated instruction for agricultural extension. Information Technologies and International Development, 5(1): 1-15. » Garbero, A. and Perge, E. 2017. Measuring women’s empowerment in agriculture. Rome, IT: International Fund for Agricultural Development, 2017. » Global Jobs Indicators database. Various years. The World Bank. » Gobierno de Guatemala. 2015. Contribución Prevista y Determinada a Nivel Nacional. Ministerio de Ambiente y Recursos Naturales. Guatemala. » Guatemala National Census 2018. XII Censo Nacional de Población y Vivienda – Resultados Censo 2018. Instituto Nacional de Estadística. Guatemala. » Herrera, Raul, Quintana, Magali, and Ugarte, Ana. 2018. USAID/Guatemala Gender Analysis Final Report. September 2018.Washington, DC: USAID, 2018. » Hildebrandt, N., Nyarko, Y., Romagnoli, G., and Soldani, E. 2020. Price Information, Inter-Village Networks, and ‘Bargaining Spillovers’: Experimental Evidence from Ghana. NYU Stern School of Business Forthcoming. Available at SSRN: https://ssrn.com/abstract=3694558 or http://dx.doi.org/10.2139/ssrn.3694558 » INE. 2015. República de Guatemala: Encuesta Nacional de Empleo e Ingresos 2-2014. ENEI 2014. Guatemala: Instituto Nacional de Estadística. 86 DIGITAGRO 2022 » INE. 2015. República de Guatemala: Encuesta Nacional de Condiciones de Vida 2014. ENCOVI 2014. Guatemala: Instituto Nacional de Estadística, 2015. » INE. 2017. República de Guatemala: Encuesta Nacional de Empleo e Ingresos 3-2016. ENEI 2016. Guatemala: Instituto Nacional de Estadística. » INE. 2019. XII Censo Nacional de Población y Vivienda – Resultados Censo 2018. Guatemala: Instituto Nacional de Estadística. » IOM and UNFPA. 2021. Caracterización de la migración internacional en Guatemala (Censo 2018). International Organization for Migration and United Nations Population Fund. Guatemala. » IPC. 2018. Situación de la Inseguridad Alimentaria Crónica en Guatemala. Integrated Food Security Phase Classification. Guatemala. » ITU. 2021. Global Cybersecurity Index 2020: Measuring commitment to cybersecurity. Geneva: International Telecommunication Union. » Jayachandran, S., Biradavolu, M. and Cooper, J., 2021. Using machine learning and qualitative interviews to design a five-question women’s agency index (No. w28626). National Bureau of Economic Research. » Jensen R. 2007. The digital provide: Information (technology), market performance, and welfare in the South Indian Fisheries Sector. Quarterly Journal of Economics, 122(3): 879-924. » Kumar, R. 2004. eChoupals: A Study on the Financial Sustainability of Village Internet Centers in Rural Madhya Pradesh. Information Technologies and International Development, 2(1): 45-73. » Labonne, J. and Chase, R. 2009. The power of information: The impact of mobile phones on farmers’ welfare in the Philippines. World Bank Policy Research Working Paper 4996. Washington, DC: The World Bank. » Leal, M, Martinez, B, Martínez, J, 2021. COVID-19 y variabilidad climática, una combinación crítica para el sector agropecuario de Guatemala. Documento de Trabajo CCAFS No. 354: Programa de Investigación de CGIAR en Cambio Climático, Agricultura y Seguridad Alimentaria (CCAFS). » Lopez, Angela, Perego, V., Romero J. 2022. Investing in Digital Technologies to Increase Market Access for Women Agri-preneurs in Guatemala. Working Paper. » Ministerio de Agricultura, Ganadería y Alimentación. 2013. El nuevo Sistema Nacional de Extensión Rural, SNER (2013). Un aporte para el desarrollo socioeconómico de las familias campesinas. Documento Técnico No. 2. Guatemala: Ministerio de Agricultura, Ganadería y Alimentación. » Ministerio de Agricultura, Ganadería y Alimentación. 2016a. La Política Agropecuaria 2016-2020. Guatemala: Ministerio de Agricultura, Ganadería y Alimentación. » Ministerio de Agricultura Ganadería y Alimentación. 2016b. Programa De Agricultura Familiar Para el Fortalecimiento de La Economía Campesina (PAFFEC 2016-2020). » Ministerio de Educación. 2019. Diagnóstico Técnico del Programa de Alimentación Escolar. Gobierno de la Republica de Guatemala: Guatemala, 2019. » MSPAS, INE, and ICF. 2017. Encuesta Nacional de Salud Materno Infantil 2014-2015. Informe Final. Guatemala, Ministerio de Salud Pública y Asistencia Social, Instituto Nacional de Estadística. 87 DIGITAGRO 2022 » Mitra, S., Mookherjee, D., Torero, M., and Visaria, S. 2018. Asymmetric Information and Middlemen Margins: An Experiment with Indian Potato Farmers. The Review of Economics and Statistics, 100(1): 1-13. » Moreno, A.L. 2020. Cadenas Agroalimentarias Modernas y Resilientes- Análisis de Brechas de Género. Unpublished. The World Bank. » Morris, M., Sebastian, A.R., Perego, V.M.E., Nash, J., Díaz-Bonilla, E., Piñeiro, V., Laborde, D., Thomas, T., Prabhala, P., Arias, J., and Centurion, M. 2020. Future foodscapes: Re-imagining Agriculture in Latin America and the Caribbean. Washington, DC: The World Bank. » OECD. 2020. Latin American Economic Outlook 2020: Digital Transformation for Building Back Better. Paris: Organisation for Economic Co-operation and Development. » Prost, M.A. and Martínez, R. 2020. El costo de la doble carga de la malnutrición. Impacto social y económico en Guatemala. World Food Programme and United Nations Economic Commission for Latin America and the Caribbean. » Qiang, C.Z., Kuek, S.C., Dymond, A. and Esselaar, S. 2012. Mobile Applications for Agriculture and Rural Development. ICT Sector Unit Report 96226-GLB. Washington, DC: The World Bank. » Raj, D.A., Poo Murugesan, A.V., Aditya, V.P.S., Olaganathan, S., and Sasikumar, K. 2011. A Crop Nutrient Management Decision Support System: India. In “Strengthening Rural Livelihoods. The Impact of Information and Communication Technologies in Asia”, edited by D.J. Grimshaw and S. Kala, 33-52. Rugby, Warwickshire, UK: Practical Action Publishing. » Sawant, M., Urkude, R., and Jawale, S. 2016. Organized Data and Information for Efficacious Agriculture Using PRIDE™ Model. International Food and Agribusiness Management Review 19(A): 115-30. » Schroeder, K., Lampietti, J., and Elabed, G. 2021. What’s Cooking: Digital Transformation of the Agri-food System. Washington, DC: The World Bank. » Svenson, J. and Yanagizawa, D. 2009. Getting the prices right: The impact of market information service in Uganda. Journal of the European Economic Association, 7(2-3): 435-445. » Tay, K, 2020. Preliminary Assessment of Eta and Iota Tropical Depressions Impact on Guatemalan » Agriculture. GT2020-0022. United States Department of Agriculture: Washington, D.C., 2020 » UNDP. 2020. Gender Inequality Index. United Nations Development Programme. » UNDP. 2016. Más allá del conflicto, luchas por el bienestar. Informe Nacional de Desarrollo Humano 2015/2016 - Guatemala. United Nations Development Programme. Guatemala. » UNEP. 2021. Food Waste Index Report 2021. United Nations Environment Programme. Nairobi » USAID. 2017. Climate change Risk Profile: Guatemala. Washington, DC: USAID. » USAID. 2018. Guatemala: Nutrition Profile. Washington, DC: USAID. » WHO. 2021. COVID-19 Dashboard. World Health Organization. https://covid19.who.int/ » Wooldridge, J. M. 2007. Inverse probability weighted estimation for general missing data problems. Journal of Econometrics, 141(2), 1281–1301. 88 DIGITAGRO 2022 » World Bank, 2011. ICT in Agriculture: Connecting Smallholders to Knowledge, Networks, and Institutions. Report Number 64605. Washington, DC: The World Bank. » World Bank. 2016a. World Development Report 2016: Digital Dividends. Washington, DC: The World Bank. » World Bank, 2016b. Linking Farmers to Markets through Productive Alliances: An Assessment of the World Bank Experience in Latin America. Washington, DC: The World Bank. » World Bank. 2017. The Global Findex Database 2017. Washington, DC: The World Bank. » World Bank. 2019. Future of Food: Harnessing Digital Technologies to Improve Food System Outcomes. Washington, DC: World Bank. » World Bank. 2020a. Guatemala: Food Smart Country Diagnostic, Washington, DC: The World Bank. » World Bank. 2020b. Nutrition Smart Agriculture in Guatemala. Nutrition Smart Agriculture Country Profiles, Washington, DC: World Bank Group. » World Bank, 2021a. Macro Poverty Outlook. Washington, DC: The World Bank. » World Bank, 2021b. Central America Digital Economy: Digital Economy Enabling Environment Assessment. Report No: AUS0002020. Country Annex 3: Guatemala Digital Economy Light Assessment. Washington, DC: The World Bank. » World Bank, 2022. Increasing Women’s Ownership and Control of Productive Assets. LAC Gender Notes. Washington, DC: The World Bank. » World Development Indicators. Various years. Washington, DC: The World Bank. 89 DIGITAGRO 2022 Annexes Annex 1: Baseline descriptive statistics and sample balance Table 3. General household information     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) Age 445 37.831 435 37.506 880 37.670 0.326 0.719 [13.016] [13.620] [13.212] Fraction of households that sold their products in the 445 1.000 435 1.000 880 1.000 N/A N/A past 12 months [0.000] [0.000] [0.000] # household members 445 6.184 435 6.395 880 6.289 -0.211 0.324 [3.187] [3.092] [3.095] # household children (<15yo) 445 2.231 435 2.306 880 2.268 -0.074 0.568 [2.100] [1.678] [1.868] # household men (15-64yo) 445 1.681 435 1.713 880 1.697 -0.032 0.702 [1.295] [1.163] [1.223] # household women (15-64yo) 445 2.083 435 2.131 880 2.107 -0.048 0.589 [1.394] [1.194] [1.282] # household elderly people (>65yo) 445 0.189 435 0.246 880 0.217 -0.057* 0.096* [0.445] [0.551] [0.496] The head of the household is: 1. Interviewed woman 445 0.187 435 0.138 880 0.163 0.049* 0.064* [0.395] [0.382] [0.394] 2. Husband/Partner 445 0.596 435 0.579 880 0.588 0.016 0.633 [0.527] [0.471] [0.497] 3. Son/Daughter 445 0.000 435 0.016 880 0.008 -0.016** 0.014** [0.000] [0.135] [0.098] 90 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 4. Son-in-law/Daughter-in-law 445 0.000 435 0.005 880 0.002 -0.005 0.155 [0.000] [0.067] [0.048] 5. Grandson/Granddaughter 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 6. Father/Mother 445 0.166 435 0.191 880 0.178 -0.025 0.397 [0.436] [0.416] [0.426] 7. Father-in-law/Mother-in-law 445 0.031 435 0.060 880 0.045 -0.028* 0.066* [0.195] [0.255] [0.229] 8. Brother/Sister 445 0.004 435 0.005 880 0.005 -0.000 0.982 [0.068] [0.067] [0.067] 9. Brother-in-law/Sister-in-law 445 0.002 435 0.000 880 0.001 0.002 0.316 [0.047] [0.000] [0.034] 10. Other relative 445 0.013 435 0.007 880 0.010 0.007 0.312 [0.110] [0.082] [0.097] 11. Other not relative 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] The highest education degree of the interviewed woman is: 1. none 445 0.110 435 0.101 880 0.106 0.009 0.674 [0.331] [0.306] [0.321] 2. Preprimaria 445 0.011 435 0.002 880 0.007 0.009 0.105 [0.105] [0.048] [0.082] 3. Primaria incompleta 445 0.339 435 0.324 880 0.332 0.015 0.648 [0.475] [0.508] [0.491] 4. Primaria completa 445 0.196 435 0.234 880 0.215 -0.039 0.167 [0.375] [0.458] [0.422] 5. Básico incompleto 445 0.036 435 0.044 880 0.040 -0.008 0.570 [0.195] [0.208] [0.202] 91 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 6. Básico completo 445 0.117 435 0.097 880 0.107 0.020 0.370 [0.356] [0.309] [0.331] 7. Diversificado incompleto 445 0.020 435 0.016 880 0.018 0.004 0.619 [0.126] [0.120] [0.124] 8. Diversificado completo 445 0.137 435 0.159 880 0.148 -0.022 0.418 [0.382] [0.423] [0.411] 9. Superior incompleta 445 0.018 435 0.009 880 0.014 0.009 0.289 [0.128] [0.118] [0.124] 10. Superior completa 445 0.016 435 0.009 880 0.013 0.007 0.382 [0.124] [0.095] [0.110] 11. Maestría o doctorado 445 0.000 435 0.005 880 0.002 -0.005 0.159 [0.000] [0.068] [0.048] The highest education degree of the head of the household is: 1. none 445 0.133 435 0.122 880 0.127 0.011 0.673 [0.392] [0.364] [0.378] 2. Preprimaria 445 0.007 435 0.007 880 0.007 -0.000 0.978 [0.083] [0.083] [0.083] 3. Primaria incompleta 445 0.362 435 0.379 880 0.370 -0.018 0.612 [0.525] [0.498] [0.509] 4. Primaria completa 445 0.225 435 0.248 880 0.236 -0.024 0.447 [0.457] [0.472] [0.469] 5. Básico incompleto 445 0.029 435 0.023 880 0.026 0.006 0.581 [0.180] [0.153] [0.167] 6. Básico completo 445 0.083 435 0.080 880 0.082 0.003 0.896 [0.289] [0.307] [0.289] 7. Diversificado incompleto 445 0.018 435 0.011 880 0.015 0.006 0.465 [0.155] [0.104] [0.133] 92 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 8. Diversificado completo 445 0.121 435 0.106 880 0.114 0.016 0.513 [0.370] [0.333] [0.348] 9. Superior incompleta 445 0.007 435 0.014 880 0.010 -0.007 0.300 [0.081] [0.117] [0.100] 10. Superior completa 445 0.016 435 0.005 880 0.010 0.011 0.168 [0.139] [0.096] [0.120] 11. Maestría o doctorado 445 0.000 435 0.005 880 0.002 -0.005 0.153 [0.000] [0.067] [0.047] Fraction of women with phone signal at home 445 0.908 435 0.890 880 0.899 0.018 0.423 [0.357] [0.317] [0.341] The interviewed woman is using a NON personal phone 445 0.130 435 0.117 880 0.124 0.013 0.606 [0.376] [0.365] [0.365] The interviewed woman is using a personal phone 445 0.870 435 0.883 880 0.876 -0.013 0.606 [0.376] [0.365] [0.365] The interviewed woman (not using a personal phone) 58 0.379 51 0.314 109 0.349 0.066 0.506 does not own a personal phone [0.492] [0.511] [0.490] The interviewed woman (not using a personal phone) 58 0.621 51 0.686 109 0.651 -0.066 0.506 owns a personal phone [0.492] [0.511] [0.490] Overall portion of women owning a personal phone 445 0.951 435 0.963 880 0.957 -0.013 0.390   [0.229]   [0.203]   [0.213]     The value displayed for t-tests are p-values. Standard deviations are robust. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 93 DIGITAGRO 2022 Table 4. Agricultural engagement information     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) Cultivated Land in cuerdas 444 6.164 430 6.564 874 6.361 -0.400 0.608 [10.109] [12.787] [11.520] Fraction of households having used in the past year: 1. natural fertilizers 445 0.892 435 0.887 880 0.890 0.005 0.843 [0.347] [0.377] [0.366] 2. chemical fertilizers 445 0.701 435 0.685 880 0.693 0.016 0.709 [0.632] [0.642] [0.636] 3. pesticides/herbicide 445 0.557 435 0.520 880 0.539 0.038 0.340 [0.552] [0.617] [0.587] 4. improved seed 445 0.503 435 0.522 880 0.512 -0.018 0.596 [0.540] [0.499] [0.523] 5. technified irrigation system 445 0.303 435 0.269 880 0.286 0.034 0.374 [0.595] [0.551] [0.578] 6. machinery 445 0.036 435 0.055 880 0.045 -0.019 0.241 [0.194] [0.281] [0.239] 7. nothing 445 0.018 435 0.021 880 0.019 -0.003 0.757 [0.135] [0.126] [0.131] Fraction of households selling mainly in their land 445 0.124 435 0.136 880 0.130 -0.012 0.635 [0.373] [0.382] [0.380] Time to reach the most important selling place for households 390 36.813 376 45.814 766 41.231 -9.001*** 0.009*** not selling in their land [39.269] [53.855] [49.187] Used means of transportation: 1. walking 390 0.197 376 0.231 766 0.214 -0.034 0.338 [0.447] [0.541] [0.504] 2. car 390 0.200 376 0.229 766 0.214 -0.029 0.388 [0.448] [0.466] [0.452] 94 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 3. van 390 0.079 376 0.096 766 0.087 -0.016 0.506 [0.308] [0.365] [0.338] 4. animal 390 0.008 376 0.003 766 0.005 0.005 0.333 [0.088] [0.051] [0.072] 5. pick-up 390 0.272 376 0.277 766 0.274 -0.005 0.917 [0.557] [0.710] [0.637] 999. other (bus/taxi/tuc tuc/moto/bicycle) 390 0.244 376 0.165 766 0.205 0.079** 0.031** [0.571] [0.426] [0.530] Fraction of households with a NIT 445 0.456 435 0.430 880 0.443 0.026 0.481 [0.555] [0.581] [0.582] Fraction of households without a NIT 445 0.506 435 0.531 880 0.518 -0.025 0.499 [0.570] [0.567] [0.581] Fraction of women not knowing if a household member has a NIT 445 0.038 435 0.039 880 0.039 -0.001 0.946 [0.195] [0.188] [0.191] Fraction of households able to issue invoices 445 0.240 435 0.248 880 0.244 -0.008 0.813 [0.478] [0.521] [0.508] Fraction of households NOT able to issue invoices 445 0.733 435 0.736 880 0.734 -0.003 0.929 [0.476] [0.547] [0.517] Fraction of women not knowing if a household member can issue 445 0.027 435 0.016 880 0.022 0.011 0.283 invoices [0.159] [0.138] [0.148] Fraction of women with a travelling time to sell equal to: 0 min 445 0.124 435 0.136 880 0.13 -0.012 0.635 [0.373] [0.382] [0.380] <=10 min 445 0.126 435 0.129 880 0.127 -0.003 0.905 [0.322] [0.393] [0.358] (10, 20] min 445 0.227 435 0.168 880 0.198 0.059* 0.083* [0.551] [0.438] [0.510] 95 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) (20, 30] min 445 0.263 435 0.209 880 0.236 0.054* 0.091* [0.472] [0.476] [0.483] (30, 45] min 445 0.074 435 0.057 880 0.066 0.017 0.396 [0.299] [0.282] [0.289] 1 hour 445 0.099 435 0.159 880 0.128 -0.060** 0.035** [0.346] [0.470] [0.419] (60, 90] min 445 0.034 435 0.067 880 0.05 -0.033* 0.098* [0.183] [0.370] [0.292] > 2 hours 445 0.054 435 0.076 880 0.065 -0.022 0.27 [0.261] [0.326] [0.297] Frequence of participation to agricultural trainings during the past year: 1. every day 445 0.002 435 0.005 880 0.003 -0.002 0.554 [0.048] [0.068] [0.059] 2. every week 445 0.013 435 0.028 880 0.020 -0.014 0.176 [0.110] [0.185] [0.152] 3. every 2 weeks 445 0.027 435 0.039 880 0.033 -0.012 0.374 [0.200] [0.203] [0.200] 4. every month 445 0.193 435 0.234 880 0.214 -0.041 0.185 [0.417] [0.502] [0.462] 5. less than every month 445 0.288 435 0.209 880 0.249 0.078** 0.017** [0.514] [0.452] [0.499] 6. never 445 0.476 435 0.485 880 0.481 -0.009 0.838 [0.696] [0.549] [0.624] During the year before the pandemic, the total income from selling the harvest was: 1. Higher 445 0.337 435 0.322 880 0.330 0.015 0.663 [0.542] [0.500] [0.523] 96 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 2. Equal 445 0.209 435 0.255 880 0.232 -0.046* 0.086* [0.381] [0.397] [0.390] 3. Lower 445 0.425 435 0.398 880 0.411 0.027 0.479 [0.609] [0.511] [0.564] -8. Don’t know/no answer 445 0.029 435 0.025 880 0.027 0.004 0.729 [0.174] [0.164] [0.169] Fraction of women with the following problems in selling products: 1. not able to get input 445 0.036 435 0.030 880 0.033 0.006 0.612 [0.181] [0.175] [0.179] 2. increase in input prices 445 0.031 435 0.057 880 0.044 -0.026** 0.044** [0.157] [0.220] [0.193] 3. lack of personnel 445 0.018 435 0.030 880 0.024 -0.012 0.289 [0.153] [0.177] [0.165] 4. not able to find a place to sell 445 0.162 435 0.126 880 0.144 0.035 0.198 [0.437] [0.362] [0.400] 5. lack of financing 445 0.085 435 0.103 880 0.094 -0.018 0.357 [0.298] [0.278] [0.284] 6. competition of imported products 445 0.036 435 0.032 880 0.034 0.004 0.765 [0.194] [0.180] [0.187] 7. social conflicts 445 0.016 435 0.009 880 0.013 0.007 0.360 [0.114] [0.096] [0.106] 8. lack of water for agriculture 445 0.094 435 0.101 880 0.098 -0.007 0.745 [0.287] [0.334] [0.313] 9. problems in transporting products-high distance 445 0.229 435 0.218 880 0.224 0.011 0.755 [0.459] [0.553] [0.503] 10. products damages 445 0.092 435 0.097 880 0.094 -0.004 0.834 [0.333] [0.279] [0.301] 97 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 11. low selling price 445 0.182 435 0.154 880 0.168 0.028 0.312 [0.416] [0.397] [0.405] 999 Other: limited/infertile land 445 0.029 435 0.037 880 0.033 -0.008 0.508 [0.174] [0.167] [0.170] covid-19 445 0.043 435 0.051 880 0.047 -0.008 0.586 [0.198] [0.229] [0.215] little children to care 445 0.000 435 0.005 880 0.002 -0.005 0.160 [0.000] [0.068] [0.048] local competition 445 0.007 435 0.016 880 0.011 -0.009 0.193 [0.082] [0.125] [0.105] unstable prices 445 0.011 435 0.002 880 0.007 0.009* 0.099* [0.103] [0.048] [0.082] lack of training/orientation about harvesting 445 0.009 435 0.009 880 0.009 -0.000 0.974 [0.092] [0.095] [0.094] low production 445 0.016 435 0.011 880 0.014 0.004 0.598 [0.134] [0.103] [0.120] other (lack of time/sickness/no answer) 445 0.038 435 0.083 880 0.060 -0.045*** 0.002*** [0.181] [0.226] [0.208] 1000. none 445 0.234 435 0.239 880 0.236 -0.005 0.867 [0.518] [0.420] [0.469] Fraction of women with the following main problem in selling: 1. not able to get input 445 0.011 435 0.002 880 0.007 0.009 0.151 [0.122] [0.048] [0.094] 2. increase in input prices 445 0.007 435 0.018 880 0.013 -0.012 0.138 [0.081] [0.143] [0.117] 3. lack of personnel 445 0.004 435 0.016 880 0.010 -0.012* 0.088* [0.068] [0.125] [0.101] 98 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 4. not able to find a place to sell 445 0.133 435 0.080 880 0.107 0.052** 0.028** [0.393] [0.294] [0.346] 5. lack of financing 445 0.063 435 0.057 880 0.060 0.005 0.730 [0.235] [0.230] [0.231] 6. competition of imported products 445 0.011 435 0.016 880 0.014 -0.005 0.520 [0.097] [0.124] [0.111] 7. social conflicts 445 0.002 435 0.005 880 0.003 -0.002 0.553 [0.048] [0.068] [0.058] 8. lack of water for agriculture 445 0.061 435 0.067 880 0.064 -0.006 0.745 [0.252] [0.298] [0.279] 9. problems in transporting products-high distance 445 0.169 435 0.145 880 0.157 0.024 0.429 [0.457] [0.418] [0.433] 10. products damages 445 0.065 435 0.067 880 0.066 -0.001 0.934 [0.281] [0.246] [0.259] 11. low selling price 445 0.128 435 0.113 880 0.120 0.015 0.534 [0.400] [0.332] [0.366] 999 Other: limited/infertile land 445 0.020 435 0.025 880 0.023 -0.005 0.624 [0.157] [0.150] [0.153] covid-19 445 0.038 435 0.046 880 0.042 -0.008 0.565 [0.188] [0.213] [0.202] little children to care 445 0.000 435 0.002 880 0.001 -0.002 0.319 [0.000] [0.048] [0.034] local competition 445 0.004 435 0.007 880 0.006 -0.002 0.641 [0.067] [0.084] [0.075] unstable prices 445 0.004 435 0.002 880 0.003 0.002 0.571 [0.066] [0.048] [0.058] 99 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) lack of training/orientation about harvesting 445 0.004 435 0.009 880 0.007 -0.005 0.396 [0.066] [0.095] [0.082] low production 445 0.011 435 0.009 880 0.010 0.002 0.776 [0.119] [0.093] [0.107] other (lack of time/sickness/no answer) 445 0.027 435 0.067 880 0.047 -0.040*** 0.002*** [0.162] [0.211] [0.191] 1000. none 445 0.234 435 0.241 880 0.237 -0.008 0.811     [0.518]   [0.423]   [0.470]     The value displayed for t-tests are p-values. Standard deviations are robust. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 100 DIGITAGRO 2022 Table 5. Agricultural production     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) In the past year the household produced: 1. chard 445 0.065 435 0.131 880 0.098 -0.066*** 0.007*** [0.290] [0.421] [0.375] 2. avocado 445 0.040 435 0.039 880 0.040 0.001 0.918 [0.201] [0.188] [0.192] 3. garlic 445 0.002 435 0.002 880 0.002 -0.000 0.987 [0.048] [0.048] [0.048] 4. celery 445 0.034 435 0.051 880 0.042 -0.017 0.274 [0.200] [0.265] [0.241] 5. rice 445 0.000 435 0.007 880 0.003 -0.007* 0.067* [0.000] [0.078] [0.057] 6. broccoli 445 0.083 435 0.124 880 0.103 -0.041 0.134 [0.370] [0.432] [0.399] 7. sweet potato 445 0.004 435 0.005 880 0.005 -0.000 0.985 [0.065] [0.096] [0.082] 8. carambola 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 9. onion 445 0.094 435 0.113 880 0.103 -0.018 0.531 [0.467] [0.398] [0.432] 10. chile guaque 445 0.002 435 0.009 880 0.006 -0.007 0.163 [0.045] [0.094] [0.073] 11. chile pasa 445 0.009 435 0.005 880 0.007 0.004 0.409 [0.090] [0.067] [0.080] 12. chile pimiento 445 0.054 435 0.062 880 0.058 -0.008 0.698 [0.293] [0.327] [0.308] 101 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 13. cilantro 445 0.119 435 0.202 880 0.160 -0.083** 0.016** [0.441] [0.585] [0.535] 14. string beans 445 0.022 435 0.016 880 0.019 0.006 0.605 [0.225] [0.129] [0.184] 15. spinach 445 0.040 435 0.064 880 0.052 -0.024 0.192 [0.235] [0.302] [0.271] 16. black beans 445 0.247 435 0.237 880 0.242 0.010 0.774 [0.523] [0.532] [0.518] 17. guisquil 445 0.045 435 0.039 880 0.042 0.006 0.707 [0.237] [0.219] [0.226] 18. cashew 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 19. laurel 445 0.002 435 0.000 880 0.001 0.002 0.321 [0.048] [0.000] [0.034] 20. lemons 445 0.065 435 0.074 880 0.069 -0.008 0.678 [0.335] [0.278] [0.314] 21. corn 445 0.393 435 0.471 880 0.432 -0.078* 0.079* [0.598] [0.707] [0.661] 22. malanga 445 0.007 435 0.011 880 0.009 -0.005 0.460 [0.083] [0.106] [0.095] 23. mango 445 0.036 435 0.034 880 0.035 0.001 0.925 [0.222] [0.242] [0.231] 24. apples 445 0.011 435 0.009 880 0.010 0.002 0.765 [0.105] [0.096] [0.100] 25. hoja de maxan 445 0.016 435 0.021 880 0.018 -0.005 0.581 [0.128] [0.138] [0.132] 102 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 26. melon 445 0.002 435 0.000 880 0.001 0.002 0.317 [0.047] [0.000] [0.034] 27. miltomate 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 28. blackberry 445 0.009 435 0.005 880 0.007 0.004 0.426 [0.095] [0.067] [0.082] 29. naranjada 445 0.009 435 0.014 880 0.011 -0.005 0.490 [0.090] [0.114] [0.102] 30. oregano 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 31. potatoes 445 0.196 435 0.211 880 0.203 -0.016 0.732 [0.724] [0.666] [0.698] 32. papaya 445 0.031 435 0.018 880 0.025 0.013 0.321 [0.243] [0.132] [0.198] 33. pineapple 445 0.009 435 0.007 880 0.008 0.002 0.725 [0.095] [0.081] [0.087] 34. platano 445 0.085 435 0.092 880 0.089 -0.007 0.803 [0.345] [0.436] [0.396] 35. cabbage 445 0.083 435 0.110 880 0.097 -0.027 0.262 [0.348] [0.375] [0.367] 36. watermelon 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 37. tomatoes 445 0.139 435 0.099 880 0.119 0.040 0.193 [0.545] [0.353] [0.469] 38. thyme 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 103 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 39. yuca 445 0.022 435 0.021 880 0.022 0.002 0.855 [0.162] [0.180] [0.194] 40. carrot 445 0.153 435 0.177 880 0.165 -0.024 0.514 [0.561] [0.545] [0.558] 41. zapotes 445 0.000 435 0.002 880 0.001 -0.002 0.318 [0.000] [0.048] [0.034] 997. Otros 1 445 0.753 435 0.766 880 0.759 -0.013 0.712 [0.509] [0.520] [0.516] 998. Otros 2 445 0.429 435 0.464 880 0.447 -0.035 0.338 [0.532] [0.559] [0.547] 999. Otros 3 445 0.164 435 0.223 880 0.193 -0.059** 0.040** [0.387] [0.449] [0.413] Any agricultural product 445 1.000 435 1.000 880 1.000 N/A N/A [0.000] [0.000] [0.000] Any agricultural PAE product 445 0.919 435 0.929 880 0.924 -0.010 0.617     [0.300]   [0.273]   [0.288]     The value displayed for t-tests are p-values. Standard deviations are robust. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 104 DIGITAGRO 2022 Table 6. Agricultural sales     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) In the past year the household sold: 1. chard 445 0.025 435 0.037 880 0.031 -0.012 0.326 [0.168] [0.195] [0.183] 2. avocado 445 0.013 435 0.025 880 0.019 -0.012 0.222 [0.111] [0.166] [0.140] 3. garlic 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 4. celery 445 0.016 435 0.011 880 0.014 0.004 0.591 [0.141] [0.107] [0.133] 5. rice 445 0.000 435 0.002 880 0.001 -0.002 0.310 [0.000] [0.047] [0.033] 6. broccoli 445 0.040 435 0.069 880 0.055 -0.029 0.125 [0.243] [0.299] [0.269] 7. sweet potato 445 0.002 435 0.002 880 0.002 -0.000 0.987 [0.046] [0.048] [0.047] 8. carambola 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 9. onion 445 0.049 435 0.071 880 0.060 -0.022 0.295 [0.319] [0.306] [0.315] 10. chile guaque 445 0.002 435 0.005 880 0.003 -0.002 0.540 [0.045] [0.066] [0.056] 11. chile pasa 445 0.004 435 0.000 880 0.002 0.004 0.136 [0.064] [0.000] [0.046] 12. chile pimiento 445 0.034 435 0.028 880 0.031 0.006 0.674 [0.231] [0.199] [0.215] 105 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 13. cilantro 445 0.063 435 0.099 880 0.081 -0.036* 0.085* [0.253] [0.369] [0.329] 14. string beans 445 0.011 435 0.007 880 0.009 0.004 0.542 [0.124] [0.084] [0.106] 15. spinach 445 0.013 435 0.023 880 0.018 -0.010 0.325 [0.130] [0.155] [0.143] 16. black beans 445 0.110 435 0.110 880 0.110 -0.000 0.993 [0.396] [0.384] [0.388] 17. guisquil 445 0.029 435 0.023 880 0.026 0.006 0.642 [0.214] [0.177] [0.197] 18. cashew 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 19. laurel 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 20. lemons 445 0.038 435 0.037 880 0.037 0.001 0.916 [0.204] [0.196] [0.200] 21. corn 445 0.142 435 0.177 880 0.159 -0.035 0.261 [0.403] [0.530] [0.477] 22. malanga 445 0.002 435 0.005 880 0.003 -0.002 0.552 [0.048] [0.068] [0.058] 23. mango 445 0.011 435 0.011 880 0.011 -0.000 0.975 [0.119] [0.121] [0.120] 24. apples 445 0.007 435 0.000 880 0.003 0.007* 0.082* [0.082] [0.000] [0.058] 25. hoja de maxan 445 0.004 435 0.007 880 0.006 -0.002 0.618 [0.063] [0.079] [0.071] 106 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 26. melon 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 27. miltomate 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 28. blackberry 445 0.007 435 0.002 880 0.005 0.004 0.328 [0.082] [0.048] [0.068] 29. naranjada 445 0.002 435 0.002 880 0.002 -0.000 0.987 [0.048] [0.048] [0.048] 30. oregano 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 31. potatoes 445 0.162 435 0.163 880 0.163 -0.001 0.974 [0.661] [0.627] [0.649] 32. papaya 445 0.009 435 0.002 880 0.006 0.007 0.169 [0.090] [0.047] [0.073] 33. pineapple 445 0.009 435 0.000 880 0.005 0.009** 0.046** [0.095] [0.000] [0.067] 34. platano 445 0.049 435 0.062 880 0.056 -0.013 0.558 [0.264] [0.366] [0.316] 35. cabbage 445 0.036 435 0.044 880 0.040 -0.008 0.598 [0.208] [0.224] [0.216] 36. watermelon 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 37. tomatoes 445 0.090 435 0.057 880 0.074 0.032 0.227 [0.487] [0.279] [0.405] 38. thyme 445 0.000 435 0.000 880 0.000 N/A N/A [0.000] [0.000] [0.000] 107 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 39. yuca 445 0.013 435 0.007 880 0.010 0.007 0.267 [0.132] [0.083] [0.129] 40. carrot 445 0.101 435 0.110 880 0.106 -0.009 0.748 [0.437] [0.417] [0.429] 41. zapotes 445 0.000 435 0.002 880 0.001 -0.002 0.318 [0.000] [0.048] [0.034] 997. Otros 1 445 0.483 435 0.457 880 0.470 0.026 0.524 [0.582] [0.634] [0.619] 998. Otros 2 445 0.200 435 0.230 880 0.215 -0.030 0.294 [0.420] [0.430] [0.428] 999. Otros 3 445 0.072 435 0.106 880 0.089 -0.034* 0.078* [0.270] [0.290] [0.283] Any agricultural product sold 445 1.000 435 1.000 880 1.000 N/A N/A [0.000] [0.000] [0.000] Any agricultural PAE product sold 445 0.701 435 0.729 880 0.715 -0.028 0.533 [0.698] [0.624] [0.665] In the past year, the household sold to: 1. Coyote/Intermediary 445 0.063 435 0.067 880 0.065 -0.004 0.846 [0.265] [0.308] [0.288] 2. Association/Cooperative 445 0.025 435 0.011 880 0.018 0.013 0.285 [0.229] [0.124] [0.185] 3. Someone who resells to schools (PAE provider) 445 0.025 435 0.048 880 0.036 -0.024 0.106 [0.155] [0.262] [0.213] 4. School 445 0.031 435 0.044 880 0.037 -0.012 0.445 [0.195] [0.271] [0.236] 5. Merchant 445 0.088 435 0.078 880 0.083 0.009 0.612 [0.264] [0.290] [0.277] 108 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 6. Directly to the person 445 0.346 435 0.326 880 0.336 0.020 0.606 [0.561] [0.575] [0.572] 7. In the square/market 445 0.344 435 0.303 880 0.324 0.040 0.287 [0.588] [0.572] [0.602] 999. Other 445 0.256 435 0.297 880 0.276 -0.040 0.240 [0.528] [0.494] [0.519] Neighbors 445 0.110 435 0.108 880 0.109 0.002 0.938 [0.020] [0.017] [0.013] In the past year, the household sold mainly to: 1. Coyote/Intermediary 445 0.061 435 0.057 880 0.059 0.003 0.864 [0.263] [0.293] [0.278] 2. Association/Cooperative 445 0.020 435 0.009 880 0.015 0.011 0.334 [0.211] [0.115] [0.171] 3. Someone who resells to schools (PAE provider) 445 0.018 435 0.032 880 0.025 -0.014 0.213 [0.134] [0.197] [0.167] 4. School 445 0.022 435 0.041 880 0.032 -0.019 0.176 [0.165] [0.241] [0.208] 5. Merchant 445 0.079 435 0.060 880 0.069 0.019 0.282 [0.244] [0.275] [0.260] 6. Directly to the person 445 0.294 435 0.274 880 0.284 0.021 0.567 [0.512] [0.574] [0.549] 7. In the square/market 445 0.267 435 0.264 880 0.266 0.003 0.932 [0.532] [0.567] [0.567] 999. Other 445 0.238 435 0.262 880 0.250 -0.024 0.482 [0.521] [0.496] [0.515] Fraction of households that will have products to sell in the next 445 0.555 435 0.549 880 0.552 0.006 0.875 months [0.527] [0.540] [0.533] 109 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) Fraction of households with already a buyer for the next months 247 0.328 239 0.364 486 0.346 -0.036 0.377 [0.476] [0.427] [0.454] In the past year, selling decisions in the household were taken by: 1. Interviewed woman 445 0.649 435 0.637 880 0.643 0.013 0.715 [0.578] [0.457] [0.530] 2. Husband 445 0.492 435 0.490 880 0.491 0.002 0.946 [0.518] [0.568] [0.546] 3. Other member of the household 445 0.216 435 0.230 880 0.223 -0.014 0.625 [0.428] [0.439] [0.437] 4. Other NOT member of the household 445 0.020 435 0.034 880 0.027 -0.014 0.273 [0.165] [0.216] [0.191] -8. Not appliable 445 0.004 435 0.002 880 0.003 0.002 0.573 [0.067] [0.047] [0.058] Fraction of women that negotiated with buyers in the past year 445 0.288 435 0.301 880 0.294 -0.014 0.707     [0.536]   [0.531]   [0.534]     The value displayed for t-tests are p-values. Standard deviations are robust. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 110 DIGITAGRO 2022 Table 7. Perceptions about the SFP     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) Fraction of women knowing the SFP 445 0.766 435 0.766 880 0.766 0.001 0.979 [0.436] [0.456] [0.445] Fraction of women knowing the SFP thanks to: 1. MAGA 341 0.223 333 0.216 674 0.220 0.007 0.850 [0.361] [0.538] [0.458] 2. MINEDUC campaign 341 0.023 333 0.024 674 0.024 -0.001 0.964 [0.153] [0.169] [0.160] 3. other farmer/proveedor 341 0.044 333 0.030 674 0.037 0.014 0.312 [0.197] [0.160] [0.179] 4. school/teacher/OPF 341 0.557 333 0.556 674 0.556 0.002 0.968 [0.508] [0.548] [0.524] 999 Other: COCODE 341 0.015 333 0.006 674 0.010 0.009 0.222 [0.107] [0.075] [0.094] TV/news/internet/social media 341 0.012 333 0.024 674 0.018 -0.012 0.218 [0.105] [0.149] [0.129] relative/neighbor/friend 341 0.114 333 0.120 674 0.117 -0.006 0.822 [0.360] [0.299] [0.329] government/training/SFP person 341 0.038 333 0.030 674 0.034 0.008 0.553 [0.174] [0.192] [0.188] mayor 341 0.006 333 0.006 674 0.006 -0.000 0.984 [0.108] [0.075] [0.093] other 341 0.053 333 0.045 674 0.049 0.008 0.638 [0.228] [0.200] [0.215] Fraction of women that during the last year talked to (or a 341 0.296 333 0.318 674 0.307 -0.022 0.585 household member talked to) a SFP provider [0.524] [0.528] [0.524] 111 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) Fraction of women that during the last year did NOT talk to a SFP 341 0.698 333 0.661 674 0.680 0.037 0.364 provider (and no household member did) [0.522] [0.542] [0.529] Fraction of women not knowing if her or some household 341 0.006 333 0.021 674 0.013 -0.015* 0.080* member talked to a SFP provider in the last year [0.077] [0.137] [0.112] Fraction of women registered or with a household member 341 0.150 333 0.168 674 0.159 -0.019 0.602 registered to be a SFP provider [0.411] [0.506] [0.458] Fraction of women NOT registered and with no household 341 0.842 333 0.823 674 0.832 0.019 0.603 member registered to be a SFP provider [0.410] [0.513] [0.458] Fraction of women not knowing if a household member is 341 0.009 333 0.009 674 0.009 -0.000 0.980 registered to be a SFP provider [0.092] [0.119] [0.106] Fraction of women interested in registering to sell to SFP 287 0.463 274 0.427 561 0.446 0.036 0.380 [0.429] [0.539] [0.485] Fraction of women NOT interested in registering to sell to SFP 287 0.348 274 0.401 561 0.374 -0.053 0.213 [0.469] [0.519] [0.489] Fraction of women not knowing if they would be interested in 287 0.188 274 0.172 561 0.180 0.017 0.617 registering to sell to SFP [0.422] [0.360] [0.390] Fraction of women not willing to register because: 1. does not know how to register 100 0.070 110 0.100 210 0.086 -0.030 0.436 [0.260] [0.298] [0.283] 2. low selling prices 100 0.010 110 0.000 210 0.005 0.010 0.293 [0.095] [0.000] [0.067] 3. the process is difficult 100 0.090 110 0.055 210 0.071 0.035 0.322 [0.302] [0.200] [0.252] 4. SFP does not buy their products 100 0.070 110 0.018 210 0.043 0.052* 0.074* [0.258] [0.134] [0.212] 999 Other:lack of time 100 0.100 110 0.118 210 0.110 -0.018 0.704 [0.349] [0.343] [0.345] 112 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) not enough products/land or only seasonal products 100 0.370 110 0.427 210 0.400 -0.057 0.385 [0.436] [0.525] [0.493] required too many products 100 0.000 110 0.009 210 0.005 -0.009 0.324 [0.000] [0.096] [0.069] do not produce required products 100 0.030 110 0.027 210 0.029 0.003 0.906 [0.172] [0.163] [0.167] required high quality products 100 0.010 110 0.009 210 0.010 0.001 0.946 [0.099] [0.095] [0.097] too much responsibility/commitment 100 0.060 110 0.064 210 0.062 -0.004 0.917 [0.227] [0.278] [0.254] already sell to other buyers 100 0.040 110 0.045 210 0.043 -0.005 0.838 [0.195] [0.191] [0.193] lack of money 100 0.080 110 0.036 210 0.057 0.044 0.142 [0.237] [0.185] [0.217] lack of water 100 0.010 110 0.045 210 0.029 -0.035 0.115 [0.100] [0.210] [0.167] cannot issue invoices 100 0.000 110 0.009 210 0.005 -0.009 0.324 [0.000] [0.096] [0.069] need to ask to husband/father 100 0.040 110 0.018 210 0.029 0.022 0.320 [0.184] [0.124] [0.155] other reasons 100 0.130 110 0.127 210 0.129 0.003 0.951 [0.326] [0.310] [0.317] Fraction of women willing to register because: 1. know how to register 133 0.000 117 0.009 250 0.004 -0.009 0.322 [0.000] [0.093] [0.064] 2. high selling prices 133 0.075 117 0.094 250 0.084 -0.019 0.587 [0.247] [0.291] [0.267] 113 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 3. registering is easy 133 0.000 117 0.000 250 0.000 N/A N/A [0.000] [0.000] [0.000] 4. SFP buys their products 133 0.481 117 0.470 250 0.476 0.011 0.856 [0.476] [0.488] [0.479] 999 Other:to sell and earn more money 133 0.233 117 0.222 250 0.228 0.011 0.828 [0.384] [0.406] [0.396] help the husband 133 0.000 117 0.009 250 0.004 -0.009 0.322 [0.000] [0.093] [0.064] to be able to work/like to work in agriculture 133 0.038 117 0.026 250 0.032 0.012 0.576 [0.179] [0.159] [0.171] to help the community and schools, and give quality products to 133 0.150 117 0.154 250 0.152 -0.003 0.937 students [0.348] [0.362] [0.360] to sell locally, without transportation problems 133 0.015 117 0.009 250 0.012 0.006 0.634 [0.122] [0.092] [0.109] to have a fixed buyer 133 0.015 117 0.026 250 0.020 -0.011 0.613 [0.172] [0.159] [0.166] to know more about feeding 133 0.000 117 0.026 250 0.012 -0.026* 0.070* [0.000] [0.152] [0.107] other reasons 133 0.173 117 0.137 250 0.156 0.036 0.466 [0.430] [0.355] [0.394] Fraction of women currently trusting the MAGA: 1. a lot 445 0.445 435 0.478 880 0.461 -0.033 0.394 [0.575] [0.577] [0.569] 2. quite a bit 445 0.151 435 0.131 880 0.141 0.020 0.469 [0.420] [0.378] [0.397] 3. a little 445 0.227 435 0.209 880 0.218 0.018 0.502 [0.392] [0.401] [0.399] 114 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) 4. not at all 445 0.070 435 0.069 880 0.069 0.001 0.968 [0.256] [0.262] [0.257] 5. no answer/don’t know 445 0.108 435 0.113 880 0.110 -0.005 0.823 [0.321] [0.313] [0.318] Fraction of women currently trusting the MINEDUC: 1. a lot 445 0.458 435 0.552 880 0.505 -0.093** 0.012** [0.531] [0.536] [0.523] 2. quite a bit 445 0.155 435 0.110 880 0.133 0.045* 0.084* [0.416] [0.335] [0.376] 3. a little 445 0.209 435 0.191 880 0.200 0.018 0.509 [0.384] [0.433] [0.409] 4. not at all 445 0.085 435 0.057 880 0.072 0.028* 0.085* [0.258] [0.220] [0.240] 5. no answer/don’t know 445 0.092 435 0.090 880 0.091 0.002 0.893 [0.281] [0.274] [0.282] Fraction of women currently trusting the SAT: 1. a lot 445 0.211 435 0.239 880 0.225 -0.028 0.378 [0.492] [0.441] [0.466] 2. quite a bit 445 0.108 435 0.085 880 0.097 0.023 0.283 [0.359] [0.262] [0.314] 3. a little 445 0.189 435 0.175 880 0.182 0.014 0.602 [0.394] [0.408] [0.402] 4. not at all 445 0.135 435 0.133 880 0.134 0.001 0.951 [0.349] [0.375] [0.358] 5. no answer/don’t know 445 0.357 435 0.368 880 0.362 -0.011 0.762 [0.571] [0.466] [0.527] 115 DIGITAGRO 2022     (1)   (2)   (3)   t-test N Treatment N Control N Overall Difference p-value Variable Treatment Mean/SD Control Mean/SD Overall Mean/SD (1)-(2) (1)-(2) Fraction of women thinking they can register to sell to SFP 341 0.534 333 0.502 674 0.518 0.032 0.440 [0.542] [0.554] [0.551] Fraction of women thinking they canNOT register to sell to SFP 341 0.308 333 0.351 674 0.329 -0.043 0.252 [0.504] [0.477] [0.485] Fraction of women not knowing if they can register to sell to SFP 341 0.158 333 0.147 674 0.153 0.011 0.670 [0.349] [0.343] [0.350] Fraction of women NOT agreeing on surely being able to do a 445 0.067 435 0.087 880 0.077 -0.020 0.259 difficult work [0.233] [0.288] [0.264] Fraction of women not agreeing nor disagreeing on surely being 445 0.067 435 0.041 880 0.055 0.026 0.111 able to do a difficult work [0.267] [0.208] [0.237] Fraction of women agreeing on surely being able to do a difficult 445 0.789 435 0.818 880 0.803 -0.030 0.258 work [0.414] [0.353] [0.380] Fraction of women not knowing if they are surely able to do a 445 0.076 435 0.053 880 0.065 0.024 0.162 difficult work [0.278] [0.221] [0.253] Fraction of women NOT agreeing on being able to obtain 445 0.040 435 0.032 880 0.036 0.008 0.531 everything they aim at [0.221] [0.169] [0.199] Fraction of women not agreeing nor disagreeing on being able to 445 0.038 435 0.016 880 0.027 0.022* 0.062* obtain everything they aim at [0.215] [0.121] [0.175] Fraction of women agreeing on being able to obtain everything 445 0.892 435 0.920 880 0.906 -0.027 0.197 they aim to [0.336] [0.290] [0.317] Fraction of women not knowing if they are able to obtain 445 0.029 435 0.032 880 0.031 -0.003 0.835 everything they aim at   [0.208]   [0.216]   [0.212]       The value displayed for t-tests are p-values. Standard deviations are robust. ***, **, and * indicate significance at the 1, 5, and 10 percent critical level. 116 DIGITAGRO 2022 Annex 2: Take-up and response rate tables Table 8. Baseline response rate, by treatment group Individual Cluster Level/status Respondent cluster size N Participated Response rate N Participated Response rate Mean Sd Control 1585 436 27.5% 201 142 70.6% 7.9 0.6 Treatment 1457 445 30.5% 208 132 63.5% 7.0 0.5 Total 3042 881 29.0% 409 274 67.0% 7.4 0.4 Table 9. Endline response rate, by treatment group Individual Cluster Level/status Cluster size N Participated Response rate N Participated Response rate Mean sd Control 142 128 90.1% 142 128 90.1% 3.0 0.2 Treatment 131 117 89.3% 131 117 89.3% 3.4 0.3 Total 273 245 89.7% 273 245 89.7% 3.2 0.2 Table 10. Verified and assisted farmers, by treatment groups Variable Treatment Control Overall p-value Coefficient/SE N Mean/SE N Mean/SE N Mean/SE (T)-(C) 445 0.924 436 0.952 881 0.938 0.083* -0.028 Verified [0.012] [0.010] [0.008] [0.016] 445 0.013 436 0.023 881 0.018 0.294 0.009 Assisted [0.005] [0.007] [0.004] [0.009] 117 DIGITAGRO 2022 Table 11. Take-up questions for the treatment group Indicator Treatment Mean/[Sd] 1. Do you remember that a video was sent to you by WhatsApp about the School Feeding Program? Yes 0.9621 [0.1911] 2. Which message(s) do you remember the most from the video? (Open-ended question. Enumerator, select one) 1. What is the SFP 0.0557 [0.2298] 2. How to sell crops to the PAE 0.2689 [0.4441] 3. Products purchased by PAE schools 0.2557 [0.4370] 4. Recommendations on product quality 0.3279 [0.4702] 999. Other (Specify) 0.3148 [0.4652] 1000. None 0.1607 [0.3678] 3. Do you remember receiving text messages about the School Feeding Program? Yes 0.7855 [0.4111] 4. What do you remember most about the text messages? Open-ended question. Enumerator, select one) 1.How to sell crops to PAE 0.1968 [0.3984] 2. Products purchased by PAE schools 0.1285 [0.3353] 3. Contact information for MAGA specialists 0.0763 [0.2660] 4. Contact information for PAE suppliers 0.2129 [0.4101] 118 DIGITAGRO 2022 Indicator Treatment Mean/[Sd] 999. Other (Specify) 0.1606 [0.3679] 1000. None 0.4779 [0.5005] Annex 3: Sample balance at endline and quality of the empirical design The experimental design resulted in comparable treatment and control groups according to observable characteristics. At endline, survey participants appeared similar across the treatment and control group on a set of characteristics that had been measured at baseline. For instance, in Table 12, the two groups are not statistically different in terms of similar age, education level, and household size. Also, they are identical on a set of agricultural characteristics like land size, climatic zone, and chemical pesticides or fertilizer use. Furthermore, the groups are equally likely to have harvested SFP products in the last year and registered as SFP providers. The only statistically significant difference among the two groups is an imbalance in the probability of having sold SFP products over the previous year. Overall, the set of covariates in Table 12 fails to predict treatment status after conditioning on stratification variables, with a p-value equal to 0.119. Excluding the only unbalanced variable, the p-value equals 0.453 – so the regression analysis performed to obtain this report’s results controls for this variable at baseline. In addition, the treatment status does not predict attrition between the baseline and endline: a regression of an indicator variable on attrition against treatment status, conditional on stratification variables, yields a non-statistically significant coefficient (p-value equal to 0.747 with a standard error of 0.03). Table 12. Balance table at endline (1) (2) (3) (1)-(2) Treatment Control Mean/(SD) Total/Mean(SD) Mean/(SD) Coeffictient/(SE) p-value Interviewee age 38.9 38.23 38.26 0.01 0.995 (12.57) (12.00) (12.20) (1.04) Interviewee completed primary education 0.67 0.65 0.66 -0.01 0.824 (0.58) (0.59) (0.59) (0.04) Household head completed primary education 0.74 0.76 0.75 -0.02 0.426 (0.45) (0.47) (0.46) (0.03) Number of household members 6.21 6.25 6.23 -0.01 0.961 (3.26) (2.92) (3.09) (0.25) Household harvested and sold their products in the past 1.00 1.00 1.00 - - 12 months (0.00) (0.00) (0.00) - - 119 DIGITAGRO 2022 (1) (2) (3) (1)-(2) Treatment Control Mean/(SD) Total/Mean(SD) Mean/(SD) Coeffictient/(SE) p-value In the last year the household harvested traditional PAE 0.83 0.88 0.85 -0.02 0.448 crops (0.48) (0.42) (0.46) (0.03) In the last year, the household harvested sold traditional 0.68 0.77 0.72 -0.77 0.069* PAE crops (0.62) (0.57) (0.60) (0.04) Interviewee registered or with a household member 0.12 0.16 0.14 -0.03 0.387 registered to be PAE provider (0.35) (0.48) (0.42) (0.03) Interviewee knowing about the existence of PAE 0.79 0.77 0.78 0.02 0.567 (0.41) (0.42) (0.42) (0.03) Households having used chemical fertilizers in the last 0.71 0.67 0.69 0.07 0.152 year (0.51) (0.64) (0.58) (0.05) Households having used chemical pesticides/herbicide in 0.56 0.51 0.54 0.03 0.404 the last year (0.51) (0.57) (0.54) (0.04) Climatic zone: Valley 0.22 0.23 0.22 0.03 0.145 (0.84) (0.87) (0.85) (0.02) Climatic zone: Plateau 0.36 0.39 0.38 0 0.898 (1.05) (0.97) (1.02) (0.02) Climatic zone: Coast 0.42 0.39 0.40 -0.03 0.269 (1.11) (0.97 (1.05) (0.03) Observations 325 314 639 (Clusters) (111) (124) (228) The value displayed for t-test are the differences in the means across the groups. Standard deviations are clustered at variable village. Robust standard errors in parentheses***p<0.01,** p<0.05,* p>0.1 120 DIGITAGRO 2022 Annex 4: Detailed impact evaluation results To evaluate the impact of the DIGITAGRO information campaign, the econometric specification to estimate the treatment effects relied on OLS (ordinary least squares) regression analysis. Given an individual-level outcome of interest and a village-level treatment assignment indicator , the regression equation for a woman belonging to village and randomization stratum (defined by the intersection of municipality and an indicator for villages with high rates of WhatsApp adoption) is as follows: yivs=α+βTv+x’ ivsπ+δₛ+εivs, where δₛ is a stratum fixed effect, x’ is a row-vector of covariates (cf. list in Table 13), and ε is an error term clustered at the village level. The coefficient of interest is β, which allows to determine whether women belonging to the treatment group experience differential outcomes as a result of receiving the information campaign. The rest of this annex contains the results tables for the main regressions lying behind the findings in Chapter 5 of this report. Table 13. Covariates included in all regressions Control Type HH characteristics Number (N) of children in the HH Continuous N. of HH members Continuous N. of adult men in the HH Continuous N. of adult women in the HH Continuous N. of adults older than 65 in the HH Continuous HH head Self Discrete Partner Discrete Parent Discrete Other (omitted) Discrete Individual characteristics Age 18-29 Discrete 30-39 Discrete 40-49 Discrete 50 more (omitted) Discrete 121 DIGITAGRO 2022 Control Type Education level (attained) None Discrete Primary Discrete Secondary Discrete Tertiary (omitted) Discrete Marital status Partnered Discrete Single (omitted) Discrete Agricultural practices at baseline Fertilizers Discrete Pesticides/herbicides Discrete Improved seeds Discrete Technified irrigation system Discrete Machinery Discrete None (omitted) Discrete Land size at baseline 0 - 1 cuerdas Discrete >1 - 2 cuerdas Discrete >2 - 3 cuerdas Discrete >4 - 5 cuerdas (omitted) Discrete >6-10 Discrete >10 Discrete Land ownership Discrete Landowner   Climate Zone Plateau area Discrete Central area (omitted) Discrete Coastal area Discrete Producer of SFP products at baseline Discrete 122 DIGITAGRO 2022 Control Type Products’ climatic zone Cold Discrete Warm Discrete Both (omitted) Discrete Type of crop Permanent crops Discrete Temporary crops Discrete Undefined (omitted) Discrete Table 14. Treatment effect on information intake about the School Feeding Program Aware of the SFP Knows SFP buys from local Knows that can register Knows the steps to HH have a provider’s Knows which VARIABLES farmers as SFP provider register contact information products the school   buy (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Effect of the Treatment 0.0294 0.0400 0.116*** 0.114*** 0.0792* 0.0872* 0.0875** 0.0994*** 0.0722 0.0921* 0.0399 0.0682*   (0.0320) (0.0282) (0.0433) (0.0406) (0.0470) (0.0476) (0.0360) (0.0348) (0.0522) (0.0494) (0.0357) (0.0381)   0.360 0.157 0.00810 0.00551 0.0938 0.0683 0.0159 0.00477 0.168 0.0636 0.265 0.0744 Outcome mean in control 0.880 0.880 0.573 0.573 0.411 0.411 0.138 0.138 0.332 0.332 0.750 0.750 group Num clusters 229 229 194 194 194 194 194 194 207 207 229 229 Observations 625 625 511 511 511 511 511 511 469 469 625 625 R-squared 0.090 0.165 0.184 0.295 0.131 0.229 0.163 0.260 0.164 0.270 0.093 0.155 Stratum FE YES YES YES YES YES YES YES YES YES YES YES YES Controls NO YES NO YES NO YES NO YES NO YES NO YES Control for outcome at NO YES NO YES NO YES NO YES NO YES NO YES BL Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 123 DIGITAGRO 2022 Table 15. Treatment effect on SFP information intake by participation in traditional extension programs Has participated in agricultural training Has never participated in agricultural training VARIABLES Aware of the Knows SFP can Knows that can Knows the steps Aware of the SFP Knows SFP can buy Knows that can Knows the steps SFPk buy products from register as SFP to register products from local register as SFP to register local markets markets (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) Effect of the -0.0324 -0.0442 -0.0351 -0.00896 0.0185 0.0399 0.108** 0.102* 0.114* 0.102* 0.252*** 0.268*** 0.150** 0.153** 0.0798 0.0984* Treatment (0.0315) (0.0308) (0.0638) (0.0570) (0.0666) (0.0799) (0.0534) (0.0591) (0.0620) (0.0550) (0.0618) (0.0675) (0.0742) (0.0765) (0.0512) (0.0572)   0.305 0.153 0.583 0.875 0.781 0.618 0.0453 0.0852 0.0682 0.0643 7.79e-05 0.000117 0.0456 0.0468 0.122 0.0880   Outcome mean in 0.924 0.924 0.695 0.695 0.445 0.445 0.172 0.172 0.824 0.824 0.441 0.441 0.373 0.373 0.102 0.102 control group Num clusters 174 174 138 138 138 138 138 138 151 151 132 132 132 132 132 132 Observations 340 340 263 263 263 263 263 263 285 285 248 248 248 248 248 248 R-squared 0.141 0.257 0.235 0.400 0.207 0.317 0.195 0.322 0.227 0.346 0.336 0.492 0.208 0.455 0.233 0.403 Stratum FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES Controls NO YES NO YES NO YES NO YES NO YES NO YES NO YES NO YES Control for NO YES NO YES NO YES NO YES NO YES YES NO YES NO NO YES outcome at BL Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 124 DIGITAGRO 2022 Table 16. Treatment effect on main information source about the SFP Aware of the SFP Knows through MAGA MAGA Videos&SMS School/Teachers/SPO Other sources VARIABLES or Videos&SMS   (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Effect of the Treatment 0.0294 0.0400 0.0972** 0.117*** 0.0561 0.0756** 0.0494*** 0.0520*** -0.116*** -0.117*** 0.00115 -0.00470   (0.0320) (0.0282) (0.0406) (0.0353) (0.0394) (0.0328) (0.0125) (0.0140) (0.0408) (0.0380) (0.0401) (0.0418)   0.360 0.157 0.0176 0.00112 0.156 0.0223 0.000111 0.000262 0.00487 0.00233 0.977 0.911 Outcome mean in control 0.880 0.880 0.199 0.199 0.199 0.199 0 0 0.576 0.576 0.354 0.354 group Num clusters 229 229 214 214 214 214 214 214 214 214 214 214 Observations 625 625 560 560 560 560 560 560 560 560 560 560 R-squared 0.090 0.165 0.154 0.372 0.142 0.392 0.089 0.146 0.121 0.306 0.082 0.222 Stratum FE YES YES YES YES YES YES YES YES YES YES YES YES Controls NO YES NO YES NO YES NO YES NO YES NO YES Control for outcome at BL NO YES NO YES NO YES NO NO NO YES NO YES Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 125 DIGITAGRO 2022 Table 17. Treatment effect on women’s sales Any agricultural or Traditional SFP Traditional SFP Traditional non-SFP COVID-19 SFP COVID-19 SFP COVID-19 non-SFP VARIABLES animal product agricultural product animal product animal product agricultural product animal product animal product   (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) Effect of the Treatment -0.0554 -0.0324 0.00197 0.0170 0.0518 0.0771* -0.00622 -0.0128 0.0398 0.0317 0.0423 0.0632 0.0293 0.0379   (0.0359) (0.0331) (0.0426) (0.0396) (0.0385) (0.0402) (0.0240) (0.0238) (0.0426) (0.0413) (0.0392) (0.0412) (0.0327) (0.0327)   0.125 0.329 0.963 0.668 0.180 0.0565 0.795 0.593 0.351 0.444 0.281 0.126 0.372 0.247 Outcome mean in 0.838 0.838 0.347 0.347 0.380 0.380 0.0844 0.0844 0.282 0.282 0.341 0.341 0.146 0.146 control group Num clusters 229 229 229 229 229 229 229 229 229 229 229 229 229 229 Observations 625 625 625 625 625 625 625 625 625 625 625 625 625 625 R-squared 0.157 0.308 0.112 0.283 0.145 0.225 0.159 0.236 0.116 0.268 0.145 0.217 0.109 0.186 Stratum FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES Controls NO NO NO YES NO YES NO YES NO YES NO YES NO YES Control for outcome at NO NO NO YES NO YES NO YES NO YES NO NO NO NO BL Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 126 DIGITAGRO 2022 Table 18. Treatment effect on women’s sales by agency level (animal products) Participating Not participating VARIABLES Sold any Eggs Chicken meat Cow Cheese Cow meat Sold any Eggs Chicken meat Cow Cheese Cow SFP animal SFP animal meat product product (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) Effect of the 0.0623 0.0794 0.0518 0.0778 0.0659** 0.0703** 0.00595 0.00712 -0.00476 -0.00387 0.0850 0.139* 0.0557 0.0999 0.0148 0.0209 0.0286 0.0329 0 0 Treatment (0.0537) (0.0563) (0.0548) (0.0571) (0.0308) (0.0354) (0.0166) (0.0182) (0.00529) (0.00458) (0.0672) (0.0779) (0.0663) (0.0789) (0.0502) (0.0570) (0.0317) (0.0286) (0) (0)   0.247 0.160 0.346 0.175 0.0336 0.0486 0.721 0.696 0.369 0.399 0.208 0.0776 0.402 0.208 0.770 0.715 0.369 0.252       Outcome mean in 0.378 0.378 0.353 0.353 0.0547 0.0547 0.0249 0.0249 0.00498 0.00498 0.383 0.383 0.318 0.318 0.0935 0.0935 0.0467 0.0467 0 0 control group Num clusters 195 195 195 195 195 195 195 195 195 195 124 124 124 124 124 124 124 124 124 124 Observations 403 403 403 403 403 403 403 403 403 403 222 222 222 222 222 222 222 222 222 222 R-squared 0.177 0.282 0.154 0.254 0.106 0.179 0.136 0.231 0.043 0.104 0.227 0.404 0.242 0.404 0.150 0.290 0.215 0.406     Stratum FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES Controls NO YES NO YES NO YES NO YES NO YES NO YES NO YES NO YES NO YES NO YES Control for outcome NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO at BL Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 127 DIGITAGRO 2022 Table 19. Treatment effect on sales of SFP animal products, by marital status Partnered Single VARIABLES Sold any SFP Eggs Chicken meat Cow Cheese Cow meat Product Eggs Chicken meat Cow cheese Cow animal product meat (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) Effect of the 0.131*** 0.175*** 0.114** 0.154*** 0.0753** 0.0974*** 0.00754 0.0100 -0.00510 -0.00481 -0.0875 -0.0820 -0.0746 -0.0230 0.0408 -0.00836 0.0120 0.00506 0 0 Treatment (0.0443) (0.0461) (0.0466) (0.0499) (0.0308) (0.0368) (0.0190) (0.0191) (0.00554) (0.00538) (0.0829) (0.108) (0.0760) (0.0972) (0.0458) (0.0676) (0.0315) (0.0384) (0) (0)   0.00359 0.000187 0.0158 0.00237 0.0155 0.00884 0.691 0.599 0.359 0.373 0.294 0.448 0.328 0.813 0.375 0.902 0.705 0.895 Outcome mean in 0.367 0.367 0.332 0.332 0.0664 0.0664 0.0310 0.0310 0.00442 0.00442 0.415 0.415 0.366 0.366 0.0732 0.0732 0.0366 0.0366 0 0 control group Num clusters 196 196 196 196 196 196 196 196 196 196 109 109 109 109 109 109 109 109 109 109 Observations 454 454 454 454 454 454 454 454 454 454 171 171 171 171 171 171 171 171 171 171 R-squared 0.195 0.274 0.198 0.270 0.113 0.169 0.114 0.203 0.039 0.089 0.299 0.503 0.326 0.522 0.230 0.417 0.521 0.601     Stratum FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES Controls NO YES NO YES NO YES NO YES NO YES NO YES NO YES NO YES NO YES NO YES Control for outcome NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO at BL Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 128 DIGITAGRO 2022 Table 20. Treatment effect on women’s decision-making around sales of agricultural and animal products, by marital status Total Partnered women VARIABLES Selling decision for Agricultural Selling decision for Animal Selling decision for Agricultural Selling decision for Animal products done by interviewed product done by interviewed products done by interviewed product done by interviewed women women women women (1) (2) (3) (4) (5) (6) (7) (8) Effect of the Treatment 0.0389 -0.0361 0.0269 0.0361 0.0314 -0.00268 0.0719 0.0988   (0.0566) (0.0598) (0.0427) (0.0384) (0.0739) (0.0753) (0.0476) (0.0597)   0.493 0.547 0.529 0.349 0.672 0.972 0.133 0.1000 Outcome mean in control group 0.654 0.654 0.800 0.800 0.662 0.662 0.833 0.833 Num clusters 161 161 176 176 142 142 148 148 Observations 348 348 337 337 264 264 253 253 R-squared 0.109 0.332 0.131 0.379 0.173 0.365 0.149 0.326 Stratum FE YES YES YES YES YES YES YES YES Controls NO YES NO YES NO YES NO YES Control for outcome at BL NO YES NO YES NO YES NO YES Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 129 DIGITAGRO 2022 Table 21. Treatment effect on willingness to participate in the SFP HH registered to sell crops to the Interested in registering Sold crops to a registered SFP Plans to speak to a registered SFP VARIABLES SFP provider provider   (13) (14) (27) (28) (29) (30) (31) (32) Effect of the Treatment -0.00728 0.0108 -0.0212 -0.0104 -0.0146 0.0120 -0.0599 -0.0256   (0.0357) (0.0261) (0.0453) (0.0480) (0.0426) (0.0420) (0.0363) (0.0355)   0.839 0.679 0.640 0.829 0.732 0.776 0.101 0.471 Outcome mean in control group 0.201 0.201 0.326 0.326 0.193 0.193 0.202 0.202 Num clusters 229 229 198 198 207 207 207 207 Observations 625 625 501 501 475 475 471 471 R-squared 0.201 0.503 0.133 0.242 0.201 0.322 0.245 0.408 Stratum FE YES YES YES YES YES YES YES YES Controls NO YES NO YES NO YES NO YES Control for outcome at BL NO YES NO YES NO YES NO YES Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 130 DIGITAGRO 2022 Table 22. Treatment effect on selling barriers Could not find where to sell your products Problems transporting the harvest Low sale price VARIABLES   (1) (2) (3) (4) (5) (6) Effect of the Treatment 0.0296 0.0380 -0.0239 -0.0190 0.0262 0.0200   (0.0262) (0.0264) (0.0351) (0.0349) (0.0333) (0.0316)   0.260 0.151 0.497 0.585 0.432 0.527 Outcome mean in control group 0.104 0.104 0.185 0.185 0.192 0.192 Num clusters 229 229 229 229 229 229 Observations 625 625 625 625 625 625 R-squared 0.115 0.167 0.105 0.216 0.155 0.214 Stratum FE YES YES YES YES YES YES Controls NO YES NO YES NO YES Control for outcome at BL NO YES NO YES NO YES Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 131 DIGITAGRO 2022 Table 23. Adoption of agricultural practices, by SFP registration status   (1) (2) (3) (1)-(2)     Registered Unregistered Total Mean/(SD) Mean/(SD) Mean/(SD) Difference p-value Natural fertilizers 0.92 0.89 0.89 0.03 0.318   (0.30) (0.36) (0.35) (0.03)   Chemical fertilizers 0.76 0.68 0.69 0.09 0.062*   (0.42) (0.60) (0.58) (0.05)   Pesticides/herbicides 0.71 0.5 0.54 0.21 0.000***   (0.43) (0.54) (0.54) (0.05)   Improved Seed 0.6 0.48 0.5 0.12 0.016**   (0.49) (0.49) (0.50) (0.05)   Technified irrigation system 0.42 0.26 0.29 0.16 0.001***   (0.51) (0.53) (0.56) (0.05)   Machinery 0.1 0.03 0.05 0.06 0.025**   (0.28) (0.20) (0.21) (0.03)   None 0.02 0.02 0.02 0 0.980   (0.13) (0.12) (0.12) (0.01)   Observations 114 503 617     (Clusters) (84) (192) (228)     The value displayed for t-tests are the differences in the means across the groups. Standard deviations are clustered at variable village. Fixed effects using variable stratum are included in all estimation regressions. *p<.1; **p<0.05; ***p<0.01 132 DIGITAGRO 2022 Table 24. Treatment effect on registration process awareness and perceptions Aware of the Knows SFP can IKnows that can Knows the steps to Easy to register as Interested in Sold crops to a Plans to speak SFP buy products register as SFP register provider registering registered SFP to a registered from local provider provider SFP provider VARIABLES markets   (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) Effect of the Treatment 0.0294 0.0400 0.116*** 0.114*** 0.0792* 0.0872* 0.0875** 0.0994*** 0.0729** 0.0695** -0.0212 -0.0104 -0.0146 0.0120 -0.0599 -0.0256 (0.0320) (0.0282) (0.0433) (0.0406) (0.0470) (0.0476) (0.0360) (0.0348) (0.0294) (0.0278) (0.0453) (0.0480) (0.0426) (0.0420) (0.0363) (0.0355)   0.360 0.157 0.00810 0.00551 0.0938 0.0683 0.0159 0.00477 0.0139 0.0133 0.640 0.829 0.732 0.776 0.101 0.471   Outcome mean in 0.880 0.880 0.573 0.573 0.411 0.411 0.138 0.138 0.0569 0.0569 0.326 0.326 0.193 0.193 0.202 0.202 control group Num clusters 229 229 194 194 194 194 194 194 194 194 198 198 207 207 207 207 Observations 625 625 511 511 511 511 511 511 511 511 501 501 475 475 471 471 R-squared 0.090 0.165 0.184 0.295 0.131 0.229 0.163 0.260 0.212 0.304 0.133 0.242 0.201 0.322 0.245 0.408 Stratum FE YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES YES Controls NO YES NO YES NO YES NO YES NO YES NO YES NO YES NO YES Control for outcome NO YES NO YES NO YES NO YES NO YES NO YES NO YES NO YES at BL Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 133 DIGITAGRO 2022 Table 25. Treatment effect on institutional perception (MAGA) Trust in Maga very Trust in Maga some Trust in Maga little Trust in Maga none Trust in Maga don’t Trust Maga index VARIABLES much what know/no reply   (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Effect of the Treatment 0.0538 0.0540 -0.0533 -0.0603* -0.0177 -0.0111 -0.0258 -0.0209 0.0430 0.0385 0.169** 0.149*   (0.0387) (0.0366) (0.0339) (0.0354) (0.0333) (0.0360) (0.0202) (0.0189) (0.0272) (0.0251) (0.0825) (0.0844)   0.165 0.141 0.117 0.0893 0.596 0.757 0.204 0.269 0.116 0.125 0.0416 0.0784 Outcome mean in control group 0.351 0.351 0.201 0.201 0.273 0.273 0.0682 0.0682 0.107 0.107 2.935 2.935 Num clusters 229 229 229 229 229 229 229 229 229 229 214 214 Observations 625 625 625 625 625 625 625 625 625 625 547 547 R-squared 0.124 0.299 0.087 0.185 0.128 0.206 0.124 0.204 0.074 0.189 0.175 0.335 Stratum FE YES YES YES YES YES YES YES YES YES YES YES YES Controls NO YES NO YES NO YES NO YES NO YES NO YES Control for outcome at BL NO YES NO YES NO YES NO YES NO YES NO YES Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 134 DIGITAGRO 2022 Table 26. Treatment effect on institutional perception (MINEDUC) Trust in Mineduc Trust in Mineduc Trust in Mineduc Trust in Mineduc Trust in Minedu don’t Trust Mineduc VARIABLES very much some what little none know/no reply index   (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Effect of the Treatment -0.0248 -0.0322 -0.0179 -0.0126 -0.00763 0.00856 -0.0201 -0.0258 0.0704*** 0.0712*** 0.0276 0.00260   (0.0325) (0.0361) (0.0329) (0.0354) (0.0338) (0.0365) (0.0224) (0.0222) (0.0227) (0.0244) (0.0827) (0.0866)   0.447 0.373 0.586 0.722 0.822 0.815 0.371 0.247 0.00214 0.00393 0.739 0.976 Outcome mean in control group 0.338 0.338 0.201 0.201 0.305 0.305 0.0812 0.0812 0.0747 0.0747 2.860 2.860 Num clusters 229 229 229 229 229 229 229 229 229 229 218 218 Observations 625 625 625 625 625 625 625 625 625 625 558 558 R-squared 0.089 0.216 0.121 0.178 0.107 0.217 0.099 0.173 0.104 0.197 0.121 0.241 Stratum FE YES YES YES YES YES YES YES YES YES YES YES YES Controls NO YES NO YES NO YES NO YES NO YES NO YES Control for outcome at BL NO YES NO YES NO YES NO YES NO YES NO YES Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 135 DIGITAGRO 2022 Table 27. Treatment effect on institutional perception (SAT) Trust in SAT Trust in SAT some Trust in SAT a little Trust in SAT none Trust in SAT don’t Trust SAT index VARIABLES what know/no reply   (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Effect of the Treatment 0.0123 0.0142 0.0155 0.0164 -0.00778 0.00538 -0.0523* -0.0515* 0.0323 0.0128 0.221* 0.215*   (0.0291) (0.0297) (0.0294) (0.0295) (0.0335) (0.0340) (0.0272) (0.0268) (0.0436) (0.0451) (0.118) (0.122)   0.674 0.632 0.599 0.580 0.816 0.874 0.0562 0.0559 0.460 0.776 0.0631 0.0804 Outcome mean in control group 0.127 0.127 0.130 0.130 0.208 0.208 0.159 0.159 0.377 0.377 2.359 2.359 Num clusters 229 229 229 229 229 229 229 229 229 229 189 189 Observations 625 625 625 625 625 625 625 625 625 625 384 384 R-squared 0.152 0.227 0.071 0.165 0.149 0.236 0.082 0.158 0.088 0.216 0.173 0.302 Stratum FE YES YES YES YES YES YES YES YES YES YES YES YES Controls NO YES NO YES NO YES NO YES NO YES NO YES Control for outcome at BL NO YES NO YES NO YES NO YES NO YES NO YES Standard errors clustered by village shown in parenthesis, p-values in italics Covariate variables include are presented in Treatment effects on entry and willingness to participate in the SFP *p<.1; **p<0.05; ***p<0.01 136 DIGITAGRO 2022 Annex 5. Robustness checks using the Inverse Probability Weighting Estimator (IPWE) Figure 24. Treatment effects on entry and willingness to participate in the SFP 0.5 Proportion of women 0.4 0.3 0.2 0.309 0.296 0.301 0.1 0.200 0.179 0.197 0.191 0.189 0.163 0.0 Interested in registering Sold crops to a registered SFP provider Plans to speak to a registered SFP provider Control Treatment Treatment (IPW) Annex 6. Results of focus group discussions and of complementary surveys in San Marcos A6.1 Focus groups with farmers and OPFs at project design stage In-person focus groups with OPFs and farmers were carried out before the struck of the COVID-19 pandemic. In total, 10 farmers and 10 OPFs participated in dedicated face-to-face activities. Main SFP-related challenges mentioned by school parents’ organizations (OPFs) » For the most part, members of the OPFs do not have sufficient preparation to manage the records of food purchases from suppliers. Some claim that, at the time of entering the OPF, they did not have enough information. For this reason, it is common for school personnel such as teachers and principals to manage purchase records from vendors. » Both OPFs and school staff may have favoritism towards a provider despite considerations of quality and price. » Frequently, the mothers who should act as cooks do not show up to cook the day they are assigned. Since it is a volunteer job, there are no penalties associated with its failure. » The OPFs are elected in an assembly of parents and the elected members feel obliged to accept the position, due to high social pressure. Their main incentive for joining is ensuring better nutrition for their children. » In any case, OPFs spend a lot of time working on the SFP and this time is unpaid. Apart from the administrative work of the OPF, the members also take turns supervising the cooks’ food preparation each day. In addition, many members themselves also serve as cooks at the school. This means that, in addition to their home and family care duties, they have to work at school early in the day. Some mothers declare that they start their work in the school kitchen at 3 in the morning to prepare the menus.  » The lack of infrastructure in schools prevents the proper preparation, storage, and provision of food. » For example, a mother member of an OPF states that, once they receive the vegetables from the providers, they have to take them home to be able to cut them and store them for the following days. » In addition, there is a lack of refrigeration systems to preserve the food that is purchased from the producers. » Also, in some schools there is no dining room, so the children eat sitting on the floor or on their desks. 137 DIGITAGRO 2022 » Some wells are running dry and schools are having trouble getting water. Sometimes the mothers go to ask the neighbors for water. » Many times, the pots and other cooking utensils belong to the families that lend them to the school. Sometimes when there is money left over for food, the school buys these utensils, but always recording the purchase as if they were food. » In many schools there is no wood for the cooking fire: the children carry it on their shoulders when they go to school in the morning. » Sometimes the menus do not match what is produced in the area. This happens mainly with the national menus, but also with the departmental ones if there is a lot of geographic/climate variability (eg highlands vs. coast in San Marcos). » Especially in these cases, the food has to pass through several intermediaries, which makes it costlier than it could be. » For example, one school mentioned that the provider had to fetch the required products in the local market. » In a few cases, food quantity seems to be a problem: » There are no standard measurements (bowls or plates) for offering food to students. So cooks have to approximately calculate how much to give to each: this means that some students receive more than others. » Sometimes it seems that prices are so high that the OPF cannot buy all the ingredients needed for the menu. So they resort to giving smaller portions (e.g., half an orange per student instead of one). Main SFP-related challenges mentioned by providers and producers » The main challenge for farmers is to register with the SAT, which implies that very few family farming producers are willing to be SFP providers. Some reasons for fearing the SAT are as follows: » Not knowing the process to register, there is no clarity about the required documentation and bureaucratic language; and the SAT personnel do not help and are not friendly with the farmers. » The accounting mechanism is difficult, which means that many have to hire an accountant and register their sales, the cost of which is also high: for example, in some places along the coast an expert accountant is paid 1,000 Quetzales to register with the SAT, plus 300 Quetzales per month. The cost seems to be less expensive in the highlands, 25 Quetzales per month. » Pay the tax associated with sales. » Perception that SAT investigates other types of assets of registered providers. » Example: in February 2020, 11 providers were catering to 43 schools in one municipality on the coast. Originally in 2018, when MAGA brought together family farmers to guide them to sell to schools, approximately 35 producers from the community attended the training, but when they were told they needed to sign up for the SAT, only 5 remained. » If a provider makes invoices for more than 150,000 Quetzales a year, he begins to pay a tax rate of 18 percent, while the tax rate for invoices of less than 150,000 Quetzales a year is only 5 percent. This is an important incentive against their growth. In certain municipalities, the providers in the network divide the schools among themselves so as to stay below that threshold. 138 DIGITAGRO 2022 » There are challenges with production and storage of their products. The storage of agricultural production is a particularly important challenge. Although they receive training and follow-up from MAGA and other actors (e.g. FAO), farmers have problems dealing with: » Food loss and waste » Pests and diseases » Climate change and pollution. » The transportation of products is challenging and roads are in bad state (especially in the highlands): in the coastal area, transport challenges are the main reason to sell to the coyotes. In many cases, there are schools that are neglected by the SFP because there are no providers in their area, and it is not convenient for the providers that exist to go that far: some make an effort for the well-being of the children, but this does not always happen. » Producers on the coast do not seem to have access to alternative lucrative markets. In the highlands, on the other hand, producers can market their products at the municipal level and in the most important markets in the departments of San Marcos and Quetzaltenango: the role of the coyotes is therefore more limited. This also implies that coyotes on the coast pay lower prices because they have to take the product to a distant market, with high transportation costs. » Not all provider networks work as they should, because many were recently established and still lack organization and social capital. The level of organization of agricultural producers is higher in the highlands than in the coastal zone, which makes a difference in terms of their participation as providers within the SFP. » On the coast, providers seem to be in better educational and economic conditions than the producers. In the highlands it seems to be the other way around: the suppliers are smaller farmers, and they buy from producers who are larger and more commercialized. » Sometimes, the ingredients of the menus are found only in one area of the department and not in another: for example, the watermelon is found in the highlands, and the pineapple on the coast. For the supplier then it is difficult to find these products within the local family agriculture and ends up buying them in the markets. Other times, producers travel to other areas of the department to sell their products and buy those they do not have, but this creates high transportation costs and also waste if they do not manage to sell all the produce they bought. » Recently, the creation of networks of inter-municipal suppliers is being promoted to carry out exchanges between the hot zone and the cold zone of the department. » Also, the creation of family farming markets is being considered, so that the production that is not sold among suppliers is not wasted and can be sold to the public. The role of women in agriculture and in the SFP framework » Women consistently highlight not having the support of their husband/partner to carry out their business activities as an important factor limiting their commercial participation in the SFP. But they also mention that, when husbands see the household income go up, they change their minds. » Currently, their main work is focused on the home, with activities such as preparing food for family consumption, caring for children and other domestic chores. » In agriculture, they participate in harvest work and distribution of production. » In general, they manage the financial resources obtained from production: some can decide what to do with the money they get from their sales, in other cases it is decided jointly with the partner. 139 DIGITAGRO 2022 » Many women are dedicated to the commercialization of their husbands’ production: in these cases, they themselves are the SFP provider and not the husband. » In general, the presence of women producers is considered uncommon on the coast, but it is not totally absent: » Providers state that they mainly buy eggs and cheese from women producers: these products are easier to produce while doing housework. » Many women also buy chickens from the market, raise them for a while, and then sell them to the provider/school. » Several women are producers of herbs and leafy vegetables. » Although the work is more frequent among men due to the physical workload, there are also women who produce fruits such as rambutan, banana, and pineapple. » Some women also join a cooperative to sell in bulk (e.g. eggs) or to process some foods, such as pineapple jam. However, many times they lack the skills and training to formalize and to make these businesses profitable. Before starting the business, many were afraid that this activity might be a scam and they did not think they could have this entrepreneurial role, but now they are growing. » Many mention commercial agriculture as a salvation since their husbands emigrated and in many cases abandoned them when emigrating. Now they can earn a living autonomously. » There are opportunities to promote the participation of women in the supply of products such as vegetables, nightshade, chipilín, eggs, chicken, milk and dairy (cheese), and other processed products. A6.2 Virtual focus group with OPFs at endline The main criteria for the composition of the focus groups were the area (rural or urban) and size (total number of students). Eventually, 15 OPFs participated, divided into four specific virtual venues: the first gathered five OPFs of urban schools, the second five small schools in rural areas, the third three big rural schools, and the fourth two medium-size rural schools. Relationship with providers (duration, how many, how did they choose the provider, main opportunities and problems, quality of the products delivered, any changes over time). How many suppliers does the school have? Almost all schools have two types of providers: groceries and agricultural providers. The former offer schools processed foods (e.g., sugar, flour, milk powder), while the latter offer semi-perishable crops. So, most only have two providers that fulfil these roles. However, for a few schools, their one provider does both. This last occurrence does not seem to be determined by the number of students but rather the providers’ capacity. For example, one urban school has around 120 students and only one provider. Since when do you work with these suppliers? Before the law came into force, most did not have this distinction between providers and were only working with one. Except for the biggest of the urban pre-primary schools with up to seven grocery providers, almost all schools have one provider for each type. Some OPF presidents recall that they have been working with groceries before the law came into force. Consequently, the grocery providers were maintained, and agricultural providers were engaged later. One medium and one large rural school had changed their providers based on performance evaluation. How did you find them? Did you choose your provider? What did you consider when choosing them? It is more accessible to find a grocery provider since most can provide invoices and are rather ubiquitous. But, 140 DIGITAGRO 2022 on the other hand, agriculture providers are introduced by the MAGA. Most OPF choose a provider from a pool based on who can accommodate its budget. The bigger schools communicate the food quantities they need and select the provider with the better offer, while the other select based on price. Other criteria for picking a provider are their offer of the products contained in the menus and the quality of the crops. Have you faced problems in the relationship with SFP-registered providers? If so, what were the main issues? If applicable, why have you continued to purchase from your current providers? None of the participants pointed to a problem in the relationship, but rather most expressed discomfort with the prices offered. The price is perceived as too high compared to the local market. Generally, are you satisfied with the quality of the food that your providers sell you? In cases not, how much food = have you rejected? One issue the president of one of the bigger rural schools (around 600 students) brought up was that they could not check the quality of all crops because of the large volume. So, in a few cases, this lack of control resulted in giving food bags with some discomposed crops to the families of the kids. But the other schools did not have any issue with quality. And, even though they evaluate changing providers every year to search for better quality, only one of the interviewees has done it. Further, they point to certain benefits of working with local providers. First, transportation is cheaper because of proximity. For example, a president of an OPF of a medium-size primary rural school reported that the providers pay for the transportation of the crops to the school; in return, the school staff arrange the logistics and provide the manual labor needed. Second, working with someone from the community creates a higher degree of accountability and trust. Then, if there is any problem with the crops delivered, they can sort it out quickly and more efficiently. Finally, there is a perception that people of the community where the school is located will care more about the food quality. What can be improved on the side of the suppliers? Criticisms are mostly directed to how the program operates rather than the providers themselves. Prices What do you think of the prices set by providers in comparison to the local market? Do you think that the price of the purchased food is justified by the quality? Prices are perceived as too high compared with the local markets; however, a few factors justify the cost. If given a chance, most OPFs would buy the crops in the local market at a more affordable price. For example, some OPF presidents mentioned that prices are set too high and do not reflect the local reality, which in turn results in buying less food for the kids. On the other hand, most also recognized that buying from agricultural providers offers two advantages that partially justify the price. First, providers provide invoices that are necessary to coordinate funds with higher authorities. Second, it is easier to guarantee the quality and freshness of the crops when there is a long-term agreement between parties. Under this contractual relationship, OPFs perceived a better chance to obtain quality products as efforts are aligned with the provider instead of going to the local market and picking what is available. Are the prices offered by providers variable? Fixed? Are you able to negotiate prices with suppliers? The authorities fix prices, and OPFs are not able to negotiate. The only mechanism to look for better prices seems to be to change providers eventually. A president of a medium-size primary rural school commented that prices are fixed for almost a year. Since the authorities fix prices for each procurement cycle, this perception might stem from the fact that prices vary marginally across the school year and also from the difficulty of changing providers. Demanded products and frequency Normally, do you buy more typical products from the coast, the highlands, or the valley? What products do you purchase most frequently? Are you able to obtain these products for your schools? 141 DIGITAGRO 2022 All the schools comply with choosing one of the menus of the default options established by the authorities. Then, the bags contain the crops and groceries specified in these menus, but some products can be replaced if needed. The authorities also publish a list of replacements for the crops where the OPF can choose from, based on local preferences and the supply of the provider. However, a concern is that replacement products sometimes are hard to adapt. For example, a crop like yucca, even if abundant in their region, is not part of the usual diet of the kids, so it is changed to potato. But, sometimes, with other crops, a suitable replacement might not be available. Furthermore, there is a need to adapt quality according to what can obtained locally (e.g., size of the product), but this is not usually considered, and it is reflected in the prices paid. In the ambit of COVID-19, the fact that food bags are delivered per kid and not per household may cause substantial food waste. For example, if there is a prescribed quantity of 12 units of lemons per bag per kid, a household can end up with 60 lemons if it has five kids in school – and these families might even be producing these crops on their own for subsistence. However, the OPFs cannot adapt the quantities or change the bag’s content to avoid these waste of resources. Are the products that are purchased with most frequency different from those that were purchased frequently before the pandemic? Yes, only a few semi-perishable crops from agriculture remain. Since most schools now deliver foods in bags once a month, the product had to change to avoid food waste instead of cooking daily for the kids. Current method of bookkeeping and demand. Possible platform adoption? Knowledge and use of computer/smartphone/internet. Could you describe the method used for bookkeeping and how you keep track of the food you buy from SFP providers? Under current processes, do you think it is easy to fulfill the SFP’s administrative requirements? The individual in charge of bookkeeping varies across schools. Some members of the OPF have the position of accountant (with a significant learning curve for parents), while others rely on the school personnel like teachers or directors. One of the medium-size rural schools’ presidents commented that the MAGA supported them doing the bookkeeping and that their primary function was to supervise the logistics. So, the common perception is that these procedures must be supported by more knowledgeable staff. OPFs are generally more involved in the logistics; for example, going to banks or tax authorities in the municipality, other red tape, arranging transportation. All these procedures take approximately around three to four days per procurement cycle. The cost of food and transport to get to the municipality is covered by themselves. Another pressing issue relates to transparency between the school and the kids’ parents, as parents do not understand that the price is higher to comply with the tax authorities (providers issuing invoices) and ensure food quality, and therefore question the OPF’s integrity. The OPF usually publish all the accounting in the schools’ blackboard or otherwise share their spending reports, but there are still complaints about how the money is spent. How much experience do you have in the use of computers and the internet? And smartphones? Would you be interested in being able to view and purchase the products offered by different vendors in San Marcos through a smartphone application? OPFs in urban areas said they would be interested in testing an app like the one described. However, they are concerned that it would cause delays from the provider because of a lack of skills in using ICT tools. For example, one of the presidents said that the provider had problems not correctly issuing electronic invoices. And this technology barrier caused the provider to be frustrated and refuse to give the invoice properly. The OPFs of rural schools pointed some difficulties for adoption. Nevertheless, they seem excited about technology reducing the time and costs spent going to the municipal head to do the program paperwork. Furthermore, they are keen to have a more efficient connection with authorities. They commented on three issues: internet access, 142 DIGITAGRO 2022 smartphone ownership, the learning curve of the app. One of the presidents suggested that it would be good to put the teachers in charge because they are more likely to have a smartphone, whereas not many people in the community have them because of the cost. Also, they suggested distributing devices to them with the app pre- installed. They pointed out that this last suggestion would also solve the internet access problem. Challenges during the COVID pandemic. During COVID, how has the relationship and the purchasing process between OPF and suppliers changed? The common feeling is that the pandemic has eased the relationship between them and the provider. With the schools now delivering food bags, coordination with the providers is less frequent outside the one or two weeks when the purchases occur. So, they only have a couple of days of intense work. Contrary to when the food was delivered daily, OPFs feel that this frequency is more sustainable for both: from the OPF side, the burden of juggling cooking the meals and coordinating with providers has been lifted; at the same time, the providers can better plan their production and buy food from the markets or farmers. Do you think that suppliers have been able to adapt effectively to changes caused by the pandemic? Have some products that you previously purchased from providers been in short supply due to the pandemic? A pressing issue relates more to how the authorities schedule the calendar of the purchases: in the week of the school purchase, the prices offered by farmers and in the local markets spike because of high demand. Since most of the providers in the municipality are demanding a high volume of crops simultaneously, the price goes up to the level where it generates problems with the school budget. Furthermore, the fact that there is not enough stock in the local markets to supply all the schools in a single week causes delays in the schedule. However, schools cannot negotiate the dates when they should deliver the food bags to the household, as the local authorities define this schedule. One of the schools mentions that the sellers from the local markets hide the crops from the providers to artificially raise the prices, and providers are forced to take some products at a loss to comply with the schedule. These phenomena also prevent new providers from entering the program. They also say that it would be beneficial if the purchases could be on different dates for different schools depending on the zone (e.g., coast, highland, central area). A6.3 Virtual focus groups with MAGA extensionists at endline The virtual focus group gathered 20 MAGA extensionists, divided into two groups of 13 and seven participants respectively (based on officials’ availability). Challenges of attracting new suppliers to the SFP. What assistance do they specifically provide to support a producer who wishes to register for the SFP? How could this service be improved? What information is provided to producers that are interested in registering for the SFP? How? When contacted by aspiring providers, officials give general information about the law and the program. After that, they explain more detailed information related to the process of registration, including information of typification of the producer by the MAGA, commitments of providers, and taxation. Commitments include active engagement with program activities, which include attending provider network meetings, as well as receiving technical assistance and training. These extension efforts are designed to support providers to comply with SFP standards. Providers must also participate in meetings with MINEDUC officials and the CTIDAE. In addition, officials report that they try to clarify to farmers that being a provider is a beneficial long- term commitment. The information about the program is mainly given in face-to-face meetings. For example, officials take advantage 143 DIGITAGRO 2022 of existing spaces like farmers’ schools (CADER), municipal councils (COMUDE), and associations or cooperatives meetings. Another vital space for advertising the program is within the schools themselves, with the help of school staff, which allows engaging farmers close to the school or those whose children study there. Outside of face-to-face meetings, how do producers communicate with you to resolve their doubts about the SFP? The main pool of aspiring providers is drawn from these activities. On the other hand, some officials report that individuals outside these activities contact them, although it is mostly grocery store owners with monopolistic ambitions. Officials rarely rely on the phone as a mean to disseminate information about the program, as they mention it is hard to communicate about the program through phone. Do you think that all the questions that producers have about the SFP are completely resolved? What were/ are the challenges? Most of the difficulties to understand the registration process are related to taxation. Aspiring providers fear the tax authority since they have not had any contact with it before. The main fear is that taxation will rip out the benefits from selling to schools. Another difficulty is the lack of management skills of farmers, which hinders the understanding of the supply chain and the role of the provider. In certain communities, a further problem is that some established providers’ networks attempt to misinform aspiring ones, to discourage the entry of new providers and secure the market position of incumbents. On the other hand, in other communities, the opposite happens, and incumbent providers help spreading information about the program. What do you think are the benefits for producers that register with the SFP? What difficulties do you think producers face? The main benefits for providers are the generation of stable household business, as well as the possibility to sell more products and generate more income through this new market. Another benefit that providers receive is training and technical assistance by the MAGA, NGOs, and multilateral organizations (especially FAO): as part of the joint efforts to strengthen the market, providers’ receive agricultural extension services to help them meet schools’ demand and comply with food safety standards. On the other hand, a substantial difficulty faced by providers pertains to prices, which result from the negotiation between the MAGA and the MINEDUC – the former representing the network of providers and the latter the schools. Most MAGA officials claim that the other side has little awareness of the additional costs for farmers to supply to the program (e.g. transportation, food safety standards, and taxation), and that the school authorities use the local farmers’ market prices as a reference when setting the prices. This reduces the providers’ profit margins because these markets do not factor in extra compliance costs with the program, lowering the profitability of selling to schools. Officials from the highlands also mention that SFP prices are standardized at the municipality level without considering the price difference between local crop varieties. For example, the corn in the highlands (maíz criollo) has a higher cost of production than the one produced on the coast, and has thus a higher price: even though the program’s guidelines allow to switch the crops demanded by schools with local varieties, this does not consider the differences in prices. The same happens with the local variety of beans in the highlands. Thus, providers are discouraged from selling local produce because low prices reduce profits. Another difficulty for providers is that the delivery date to schools is currently announced in a squeezed time frame because of the pandemic. This predicament presents two challenges for providers. First, transportation becomes harder: officials report that getting a vehicle to deliver the products is already a problem for some producers, but this worsens when combined with tight delivery schedules. Second, prices offered by the farmers’ support network fluctuate with the local market prices, forcing some providers to take higher prices at a loss. 144 DIGITAGRO 2022 Providers’ networks strengthen the market and the resilience of registered providers in the face of these challenges. For example, new providers can cooperate with more established providers for transportation. But most importantly, they can look at them as role models and anchor their goals and expected profits by looking at their success in the program. Mode and frequency of agricultural extension provided in COVID context. Compared to pre-COVID times, how has the format of training producers changed? Training is returning slowly to normality, but most officials mention that they provide face-to-face training to producers less frequently, at a smaller scale, and inconsistently. During the early stages of the recovery of the pandemic, officials tried remote extension services. Unfortunately, these attempts yielded mixed results. In the highlands, there is a consensus among officials that remote methods do not work, because of challenges related to culture, education, internet access, smartphone ownership. On the other hand, in other municipalities, the officials say they have a WhatsApp group where farmers can consult – and if more detailed technical assistance is needed, they can send videos and pictures so the official can evaluate and advise accordingly. In places where farmers receive remittances from overseas, remote extension is more doable, because relatives in other countries usually gift smartphones and phone credit. Were they trained on the School Feeding Program? In most municipalities, training and assistance to registered providers were slowed down because of the pandemic context. Also, in one case, the local activities carried out to search for new producers stopped. For non-providers, training must be shorter in time and the information given is more concise because of COVID restrictions. Combined with less frequent trainings, this reduces the amount of exposure to farmers to the program. But the training about the School Feeding Program is still in the agenda of the CADER. During this pandemic, how did you and participants meet to train farmers’ schools (CADER) leaders? How frequent were these meetings? Previously, each municipality had different face-to-face training methodologies decided by the official, but now possibilities are reduced. For example, one municipality uses demonstration plots as a means of training in the CADER, but, because of COVID, it has been hard to coordinate a visit to these plots with the farmers. The result is lower attention rates, and prolonged periods between visits. Also, there is always the possibility that visits must be suddenly postponed or canceled as soon as a COVID case is found in a community. The same applies to all training methodologies: for instance, in another place face-to-face group trainings to CADER leaders are being held but less frequently than before. Officials report that the rise in the prices of transportation are a further obstacle to farmers to reach the training facilities or meeting areas. And the meetings of CADER members in the communities, how have they functioned during the pandemic? The perception for the meetings of the CADER members is the same as with the leaders. And in some communities, officials report that they have halted activities. Compared with male producers, did you notice any change in women’s attendance at meetings? In one municipality, the official reported that CADER are a predominant female space: out of ten participants, seven are women and three are men. The most notable change is not the change in gender composition, but rather the decrease in overall attendance: before it was 15-20 attendees, now it is ten. Validate the usefulness of the extension videos and whether/how they would use it in their work (including what, in their opinion, would be the main barriers). 145 DIGITAGRO 2022 Do you think the information provided in the video was accessible for producers interested in registering with the SFP? The overall sentiment towards the videos is positive. All officials agreed that the information was accessible to producers, and the pace was good for learning. But there were several suggestions to improve the quality of the video. First is that it must contain filmographic material of the agricultural practices in situ. For example, it is advisable to display local crop varieties, while the one depicted in the video is an export product—not usually produced by local farmers. Another improvement is showing schools which lack infrastructure and are in poor areas, to ground the videos to the local reality and make them more appealing. Do you think the video content complements your efforts to encourage and assist producers that are interested in registering with the SFP? All officials agree that the videos complement their extension efforts. However, one crucial point is that the videos could undermine their current work if there is no clear distinction between registering as a provider and joining as a support farmer. He reported that their efforts focus on incorporating new providers instead of the latter. If most farmers join as support farmers, providers will become mostly another intermediary. This, in turn, will defeat the purpose of the program and their work. An aspiring provider should only become a support farmer if they want to engage in the SFP but do not have the means to comply with the registration process. What do you think are the benefits in the dissemination of videos as tools for agricultural extension? What about the challenges? Some of the benefits mentioned were that the videos can help them to provide technical advice when they are not available to attend farmers. Also, the videos can be seen repeatedly by the farmers and by them, so they can improve their knowledge about the program and agricultural practices. An important challenge, on the other hand, is the lack of access to smartphones and internet. However, officials pointed out that it will increase with time, and that they already seeing this trend. The second challenge has more to do with the SFP challenges than the videos themselves. A shared feeling among officials is that the SFP is not profitable for smaller farmers. High barriers of entry due to costs combined with small profitability discourage small farmers from joining. These conditions result in only the medium to large producers getting profit from the program. What do you think is needed to make the videos as useful as possible for producers? Officials advice to utilize the power of social media and local tv networks for higher engagement. Some officials pointed out that Facebook videos have a broad audience among farmers, possibly broader than WhatsApp. But the most crucial piece needed to make the videos most helpful for the producers is to generate synergy with the efforts of the MAGA. Officials state that any information campaign must be accompanied by field practice: “La palabra enseña pero el ejemplo arrastra” [The word teaches but example creates change]. Disseminating information is key to awareness about the SFP, but adoption requires accompaniment in the field for farmers to become providers. A6.4 Survey of MAGA extensionists on contacts with treated group participants at endline 18 MAGA officials received a self-administered online survey through Google Surveys to check whether they had experienced an increase in contacts from farmers interested to know more about the SFP. Almost all the officials answered that, between June and July 2021, they had received a call from a farmer asking about the School Feeding Program, with an average of three farmers per extensionist. Half of them reported that at least one person that called told them they had watched a video about the program. 146 DIGITAGRO 2022 70 percent of all officials reported they were asked by at least one farmer about the contact information of existing providers: overall, 38 percent of the requests for information received were about the providers’ contact information. Furthermore, 70 percent of officials declared that more than half of the farmers asked about the process for registering as SFP provider. Almost all extensionists reported that they continued providing extension services to farmers during COVID-19: most of them offered these services at least once a month, and 41 percent gave support to farmers at least once a week. In terms of methods for assisting farmers, 41 percent provided technical assistance face-to-face, 41 percent provided it through telephone, and the rest combined both techniques. A6.5 Survey of existing SFP providers The sample consists of 173 providers already formally registered with the SFP. Most of the interviewees (84 percent) are registered as individuals, while the remaining are associations or cooperatives centered around the sale of key crops. Among the total of cooperatives, 36 percent are onion producers, 40 percent potato producers, 43 percent tomato producers, and 36 percent carrot producers. Demographic characteristics Women’s participation as SFP providers is not small. Of the total sample, 39 percent of the interviewees are women, with no significant differences compared in demographic and household characteristics with respect to their male counterparts. The average years of education are 8 and almost half of providers have completed secondary education. Of the remaining half, most have at least some primary education. The average provider household size has 6 members comprised of one to two children, and two women and two men of working age. The head is frequently male (64 percent of the total). Household agriculture Input usage is widespread and there are no noticeable differences in input usage between male and female providers. Most households used natural and chemical fertilizers in the last year (93 and 83 percent, respectively), and 67 percent used improved seeds. Pesticides or herbicides usage is also high at 78 percent. On average, eight out of ten providers hire labor for agricultural activities. Machinery such as tractors, planters, and scrapers are not used commonly. Only 8 percent of the farmers have used machinery in the last year. On the other hand, half of providers use technical irrigation systems such as spray, exuding, or drip irrigation on farms. The households of women providers have less agricultural land than their counterparts. On average providers’ households have 14 cuerdas for agricultural use, but women providers have a surface of 10 while males have 16 cuerdas. Sales Providers’ household income depends heavily on agriculture and selling to the schools is an important source. Eight out ten declared most of the household income comes from agriculture. While one-third of the agricultural income of the household comes from selling to the SFP market. Most of the providers own grocery stores. Despite selling crops in their stores (79 percent), however, almost all consider themselves farmers: nearly all declared harvesting the household land or a shared land during the year of the survey. Further, they declare they have sold or will sell their harvested production. In the months prior to the interview, registered SFP providers were selling to the SFP market, and almost all of them (90 percent) did it directly or indirectly. Of the total sample, 86 percent declared they had sold to schools, while 39 percent reported selling to other providers. Of the few that did not sell to the schools, there is no clear trend on why, but the two most frequent answers are that (i) the SFP market demands a higher quality standard of food; and (ii) that there was too little produce to sell. 147 DIGITAGRO 2022 Intermediaries are still important buyers, but there are other more frequent buyer alternatives for providers. 26 percent of providers sell to intermediaries, but the most common market besides the SFP is the square, local market, or terminal. 69 percent also sell directly to family members or neighbors. Participation in the SFP On average, providers had been in the program for three years at the time of the interview, and half had signed up recently. 71 percent of providers first heard of the program through the MAGA, 10 percent by the word of acquaintances. 65 percent found the registration process easy. There are no significant differences between males and females in time spent as providers or perception on ease of registration. Registered producers participated actively in the deliveries of 2021. 70 percent participated in all the deliveries to the schools. In these deliveries, each provider commonly serves tree to five schools inside a single municipality. Half of the sample declare being the sole providers of the schools they serve Most providers rely on other producers to deliver to schools. 82% percent declared to have bought food from other producers to sell to schools at some point in time: on average they buy from 5 producers. This is consistent with the efforts of the MAGA to strengthen the providers’ support networks. Providers tend to organize themselves in networks that exhibit oligopolistic behaviors. 68 percent of the sample are currently part of a network of providers organized by the MAGA. The network usually decides which schools will be served by which provider to (62 percent mention this happened in the last sale). It is a common practice to discuss the prices with the ministry officials, but the network usually agrees upon a set of prices to propose to the officials beforehand (61 percent of the total). Economic reasons propel providers to participate in the SFP, especially for females. When asked about the benefits from participating in the program, the most frequent answer is having a secure buyer (38 percent) and generating more income (37 percent). The third one is support to the community and children (29 percent). Female providers do not cite this last option as often as their male counterparts (22 percent for women and 33 percent for men). On the other hand, the main difficulties perceived in participating in the program are mainly low prices, followed by the demand of high-quality food (14 percent) and transportation (13 percent). Female providers find it substantially less troublesome than men to comply with high quality food standards (9 percent for women and 17 percent for men). The main channel of communication with schools is a phone call (70 percent). This is followed by face-to-face visits (35 percent) and WhatsApp (17 percent). Coordination with support farmers 61 percent of interviewed providers received calls from new producers that tried to sell them products for schools. For the last delivery, these providers declared to have received an average of 4 calls from new producers (i.e. producers they had not bought products from in previous deliveries). A third of them feel that calls from other producers are higher compared to prior months. 67 percent of providers communicated with farmers to coordinate a buy or to be offered crops to sell to the program. The main barrier for new support producers to sell to providers is low demand from providers. Of interviewed providers who sold to schools in the last delivery before the interview, only 38 percent bought any products from new producers. The rest bought to already established producers from previous sales (42 percent) or did not buy from any producer at all (20 percent). The most cited reasons for not buying from new producers is having availability of enough production (34 percent) or having established relationships with other producers (25 percent). 148 DIGITAGRO 2022 Most producers are however open to the idea of buying products from other producers, but they do not have the contact numbers. 72 percent of providers that sold to schools mention that they plan to buy products from other producers later. Nonetheless, only half of them (47 percent) have the phone numbers of any producers for this. And only 4 out of 10 of the providers that were called by new producers have their contacts. SFP app Most of the SFP providers have smartphones, but there is a gap between male and female providers: while 79 percent of providers have smartphones, 83 percent of males do vis-à-vis 72 percent of females. This gap disappears when asked about internet access (85 percent for male and 88 percent for female providers). All providers are interested in the idea of having a smartphone app to organize their sales to schools, with no difference between genders. The most frequently mentioned difficulties in this sense are not understanding how to use their phone correctly (35 percent) and not having a stable internet connection (14 percent). 149 DIGITAGRO 2022 DIGITAGRO Investing in digital technology to increase market access for women agri-preneurs in Guatemala Report No: ACS34060 150