Scaling Up Radio and ICTs for Enhanced Extension Delivery and Development Impact Quantitative Baseline Report Catherine Ragasa, Lucia Carrillo, Kelvin Balakasi STRATEGY SUPPORT PROGRAM | REPORT MAY 2022 CONTENTS Executive Summary ............................................................................................................................. 4 Introduction .......................................................................................................................................... 7 SRIEED II and its theory of change .................................................................................................... 8 Evaluation design ................................................................................................................................ 9 Baseline data collection .................................................................................................................... 11 Baseline data ..................................................................................................................................... 14 Socioeconomic characteristics ....................................................................................................... 14 Access to extension services ......................................................................................................... 16 Technology awareness and adoption ............................................................................................. 21 Outcome indicators ........................................................................................................................ 32 Food security............................................................................................................................ 32 Agricultural income ................................................................................................................... 36 Resilience capacity .................................................................................................................. 39 Key observations and next steps ..................................................................................................... 43 About the Authors ............................................................................................................................. 45 References ......................................................................................................................................... 45 TABLES Table 1: Baseline outcome indicators. ................................................................................................... 5 Table 2: Total households interviewed in the baseline for the impact evaluation of the SRIEED II project ............................................................................................................................................................ 11 Table 3: Total households interviewed in the baseline survey for the impact evaluation of ICT hubs ... 12 Table 4: Characteristics of the household head ................................................................................... 15 Table 5: Main occupation of head and decisionmakers in the household (% of rural households) ....... 16 Table 6: Technology awareness and adoption (% of rural households) ............................................... 25 Table 7: Awareness and adoption of agricultural technologies, marketing, and nutrition and health related practices .................................................................................................................................. 27 Table 8: Number of agricultural and nutrition and health related technologies known and adopted by rural households .................................................................................................................................. 28 Table 9: Reasons for not adopting and disadopting promoted agricultural technologies ...................... 29 Table 10: Food security and dietary diversity indicators (% rural households except for the scores) ... 34 Table 11: Women's dietary diversity indicators (% of rural women respondents, except the scores) ... 36 Table 12: Crop productivity (kg/ha) during 2020-21 cropping season .................................................. 37 Table 13: Percentage of crop harvest sold .......................................................................................... 38 Table 14: Crop production value and sales revenue during 2020-21 cropping season ........................ 39 Table 15: Crop acreage and diversification during 2020-21 cropping season ...................................... 40 Table 16: Percentage of rural households engaged in different livelihoods ......................................... 41 Table 17: Percentage of rural households using coping strategies during the COVID-19 crisis ........... 42 Table 18: Percentage of rural households adopting climate smart technologies .................................. 43 FIGURES Figure 1: Theory of change adopted from the SRIEED II project ........................................................... 8 Figure 2: Distribution of sample households in treatment and control groups ...................................... 13 Figure 3: Access to extension services in the last two years, by topic and project district (% of rural households) ......................................................................................................................................... 17 Figure 4: Access to extension services in the last 12 months, by topic and project district (% of rural households) ......................................................................................................................................... 18 Figure 5: Access to extension services in the last 12 months, by gender of respondents and project district .................................................................................................................................................. 19 Figure 6: Access to extension services in the last two years, by source and project districts .............. 20 Figure 7: Access to extension services in the last 12 months, by main source and project district (% of rural households) ................................................................................................................................. 20 Figure 8: Potential media for information access ................................................................................. 21 Figure 9: Agricultural technology awareness ....................................................................................... 22 Figure 10: Agricultural technology adoption ........................................................................................ 23 Figure 11: Awareness and adoption of livestock, marketing and agroprocessing related practices ..... 23 4 EXECUTIVE SUMMARY This report summarizes the baseline data that describe the rural population of five districts in Malawi targeted in the Scaling up Radio and ICTs for Enhanced Extension Delivery (SRIEED) II project that started in 2020 and ends in 2024. It also provides the impact evaluation strategy for the overall project as well as a causal impact evaluation of a major component of the project (impact ICT hubs). Intervention: The SRIEED II project, led by Farm Radio Trust-Malawi and the Department of Agricultural Extension Services, aims to promote and scale agricultural innovations through low- cost information and communication technologies (ICT), including radio programming and digital platforms, in order to improve incomes, food security, and resilience of 1 million farmers in five districts (Kasungu, Lilongwe (East and West), Mzimba (North and South), Nkhota-kota, and Mangochi) by 2025. These outcomes will be achieved through (1) increased awareness, knowledge, and skills among smallholder farmers; (2) increased application and adoption of ag- ricultural innovations; (3) a conducive policy environment for digital extension; (4) improved ap- plication and use of digital tools; and (5) improved use of market linkage platforms. Working with existing rural producer groups (with 10–40 members) to strengthen them to become impact ICT hubs is central to the technology promotion approach. These hubs will be targeted as centers for providing demand-driven extension and models for early adopters of agricultural innovations. Demographics: A total of 34% of sample households are female-headed households and 29% are youth-headed households. Beyond headship and looking at both spouses and other mem- bers who are decision-makers or farmers within the household, 74% of the sample are dual- adult households (both women and men within the household), 23% are households with women only, and 3% are households with men only. Most households have one female and one male decision-maker or farmer (husband and wife). A total of 43% of households have at least one youth decision-maker or farmer; and 57% of the households have no youth decision- maker or farmer. Roughly a fourth of the households have one youth, and another fourth have two youths. Most heads and other decision-makers or farmers in the household have no formal schooling or have some years in primary or elementary (77%) (Table 1). A total of 13% have no formal schooling. A total of 21% have some years in high school, and only 2% completed high school or went to college or university. Baseline outcome indictors: Table 1 summarizes the outcome indicators at baseline. 5 Table 1: Baseline outcome indicators. Indicators Measurement Total HH By gender of head By age group of head By household type Female- headed Male- headed Youth- headed Non- youth- headed Dual- headed Women- only HH Men- only HH Indicator 1: % of house- holds experiencing hunger in the last month (July) % of rural households with "little or no hunger" based on Household Hunger Scale (HHS) 91 88 92 88 92 92 88 82 Indicator 2: % households in moderate or severe fam- ily food insecurity in tar- geted regions in the last month (July) % of rural households with "moderate or severe food insecurity" based on Household Food Insecurity Ac- cess Score (HFIAS) 60 60 60 60 60 60 60 60 Indicator 3: % of house- holds meeting 6 food groups minimum dietary di- versification requirements in the past 7-days Household Dietary Diversity Score (HDDS) (0-10) 7.4 7.2 7.6 7.4 7.5 7.5 7.3 7.2 Food Consumption Score (FCS) (0-126) 50.4 47.9 52.0 49.7 51.0 50.9 48.8 51.7 % with acceptable FCS (>35) 76 70 80 74 78 79 70 68 % of household achieving all required 6 food groups promoted in Malawi 36 32 39 35 37 36 36 39 Women's Dietary Diversity Score (WDDS) (0-10) 4.2 4.3 4.2 4.2 4.2 4.2 4.4 4.3 % of rural women achieving <5 of 10 food groups adapted from WDDS 38 38 38 37 39 37 40 45 % of rural women achieving all required 6 food groups promoted in Malawi 12 13 11 12 11 10 15 11 Indicator 4: % of house- holds with higher improved income from agriculture en- terprise in past 12 months Value of crop production (MWK/year) (median) 153,000 118,500 183,000 126,250 172,000 178,200 105,000 118,000 % of rural households selling crops 60 59 61 59 61 61 57 51 % of maize harvest sold (average) 4 4 5 4 4 6 5 4 % of total crop harvest sold (average) 24 22 24 25 21 14 23 24 Value of crop sales (MWK/year) (median) 13,500 10,000 15,000 12,500 15,000 17,500 8,750 2,000 Number of livestock units (average) 10 7 11 8 11 11 7 10 6 % of rural households with income from other agricul- tural activities 65 64 66 66 64 64 65 76 Indicator 5: % of house- holds resilient to climate change and weather-related shocks in past 12 months Simpson index for crop diversification (average) 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 % of acreage planted with nonmaize crops (average) 37 35 38 37 37 37 36 33 Total land owned or cultivated (average acre) 2.6 2.2 2.8 2.3 2.8 2.8 2.0 2.0 Livelihood diversification: count of livelihood activities rural households were engaged in the last 12 months 3.2 3.1 3.3 3.2 3.2 3.2 3.1 3.2 Income diversification: count of sales or income gen- erating activities rural households were engaged in the last 12 months 2.3 2.2 2.3 2.4 2.2 2.3 2.2 2.5 Number of climate-smart technologies adopted (aver- age) 8.3 7.7 8.6 8.0 8.5 8.5 7.4 8.1 % of rural households receiving extension services in the last 12 months 68 61 72 66 70 70 64 60 % of rural households receiving extension services on environment or climate change in the last 12 months 31 25 32 28 32 32 25 32 % of rural household resorting to negative coping strategy (e.g., reduced food or nonfood expenditures or selling assets) as a result of shock/crisis (e.g., COVID-19) 6 4 8 9 5 7 5 8 Indicator 6: % of house- holds aware of the exist- ence of the disseminated technologies Number of promoted technologies rural households were aware of (0-47) 28 27 29 27 29 29 27 28 Indicator 7: % of house- holds adopting dissemi- nated technologies Number of promoted technologies the rural house- holds were adopting in 2020/2021 (0-44) 17 17 18 16 18 17 16 18 Indicator 8: % of house- holds accessing digitally enabled, market-oriented advice in past 12 months % of rural households receiving extension services through radio and digital platforms in the last 12 months 52 47 55 55 51 55 44 60 % of rural households receiving extension services on marketing or agroprocessing in the last 12 months 26 33 35 34 34 36 28 23 Source: IFPRI/Wadonda household survey (July 2021). HH=household. USD 1=MWK 813 as of August 30, 2021. 7 1. INTRODUCTION Despite significant progress achieved in the last few decades, poverty and food insecurity re- main major development challenges in Malawi. Based on the data from the Third Integrated Household Survey (IHS3) 2010–11 and Fourth Integrated Household Survey (IHS4) 2016–17, the national poverty rate remained steady at 51% in 2010–11 and 52% in 2016–17. Rural pov- erty, however, appears to have risen over the same period: from 57% in 2010–11 to 60% 2016– 17. As of 2019, Malawi ranked 172 out of 189 countries in the United Nations Development Pro- gramme’s Human Development Index and 149 out of 162 countries in the Gender Inequality In- dex (UNDP 2019). Malawi faces limited market and economic opportunities and high levels of food and nutrition in- security, compounded by issues of extreme weather events, increasing youth population and unemployment, and gender disparities. Malawi has low agricultural productivity, despite major investments in farm input subsidies for decades (Dorward and Chirwa 2011; Holden and Lun- duka 2010; Ragasa and Mazunda 2018; Ricker-Gilbert 2011). Improved technologies and pro- motion of these technologies offer a potential solution to some of these agricultural and develop- ment challenges. Information and communication technology (ICT) platforms offer a cost-effec- tive way of bringing knowledge and technologies to rural producers; however, there is current debate on whether these ICT tools alone can contribute to increased incomes and other devel- opment outcomes. The Scaling up Radio and ICTs for Enhanced Extension Delivery (SRIEED) II project, led by Farm Radio Trust-Malawi (FRT) and the Department of Agricultural Extension Services, aims to promote and scale agricultural innovations through ICTs, including radio programming and digi- tal platforms, in order to improve incomes, food security, and resilience of 1 million farmers in six districts (Kasungu, Lilongwe, Mzimba North, Mzimba South, Nkhota-kota, and Mangochi) by 2025. These outcomes will be achieved through (1) increased awareness, knowledge, and skills among smallholder farmers; (2) increased application and adoption of technologies; (3) an ena- bled conducive policy environment for digital extension; and (4) higher use of market linkage platforms. SRIEED II focuses on innovative and inclusive low-cost and high-impact digital solu- tions, including interactive radio and mobile platforms such as a farmer call center, apps and In- teractive Voice Response, and mediated farmer videos integrated with emerging agronomic weather and remote sensing technologies through a sustainable public-private partnership model. Central to the technology promotion approach is working with existing rural producer groups (with 5–40 members) to strengthen them to become radio listening clubs or impact ICT hubs. These hubs will provide demand-driven extension and be models for early adopters of im- proved agricultural technologies. Most of 2021 has focused on agricultural technology review and prioritization, listing and characterization of the potential ICT hubs, designing the targeting, monitoring and evaluation strategies, and setting up the baseline survey and interviews. Imple- mentation of interventions will start to intensify around September and October 2021 in time for the rainy season cropping. FRT engaged the International Food Policy Research Institute (IFPRI) to support the design of the impact evaluation, collect baseline data, and implement the 8 follow-up surveys for the project. This document provides background information to and reports on the results of the quantitative baseline survey with the rural population within the six focus districts of the SRIEED II project and comparison districts. 2. SRIEED II AND ITS THEORY OF CHANGE The SRIEED II project is grounded on the premise that there is a lack of information on the available improved agricultural technologies and a lack of skills and knowledge to grow produc- tive and profitable agricultural enterprises. To address this, the SRIEED II project focuses in technology transfer, extension services, and skills development through a combination of ap- proaches including low-cost ICT tools, farm demonstrations, face-to-face engagement with farmers, and marketing support as demanded by the farmers. These activities complement the investments made through the new Affordable Inputs Programme (AIP), which replaced the Farm Input Subsidy Programme (FISP). The activities start with a technology audit to compile the latest available improved technologies with evidence of impact. This is done through an extensive consultation with CGIAR centers, technical departments of the Ministry of Agriculture, Irrigation, and Water Development (MoAIWD), Department of Agricultural Research Services (DARS), District Agricultural Develop- ment Officers (DADOs), and farmers’ organizations. From the long list of available technologies, district-level prioritization is done with key stakeholders to focus on a shorter list of technologies for promotion and technology transfer. This will then be followed by extensive promotion and awareness campaigns through low-cost ICT tools (radio, short messaging service (SMS) push, interactive voice response (IVR), mobile platforms, and videos), farm demonstration, and tar- geted intensive engagement and facilitation of the ICT hubs. Through these activities, the project expects to see improvements in the use of ICT tools for ag- ricultural, market, and nutrition related information and the awareness and adoption of improved technologies being promoted. The approaches and activities under the project and the develop- ment impacts that the project aims to achieve are shown in Figure 1 Figure 1: Theory of change adopted from the SRIEED II project Source: Authors’ illustration, with inputs from the SRIEED II project team. Note: ICT=information and communication technologies; VAC=village agricultural committees; ASP=area stakeholder panels; DSP=district stakeholder panels; SMS=short messaging service; W=women, M=men, Y=youth 9 3. EVALUATION DESIGN The project evaluates the impact of the overall project and the impact of the ICT hub component at the household and individual levels. These evaluations will be complemented by project and process monitoring activities, cohort studies, and qualitative research. Impact evaluation of overall project: Changes in household and individual-level outcome indi- cators in the six project districts and in comparable districts are to be measured and quantified. Randomization of households or communities to receive the project interventions would have been the most rigorous method of evaluating project impacts. However, it was not possible within this project for various reasons. First, radio programming is a major component of the project but is difficult to randomize at the community level. Second, various activities will target most, if not all, communities in the six project districts and spillover of knowledge and services from direct beneficiaries to nonbeneficiaries can be a major issue. As such, the project districts will be compared with comparable districts and the project impacts will be quantified using differ- ence-in-differences (DID) method. This approach will not only provide information on how many beneficiaries experienced improvements in the outcome indicators but also quantify the average effect of the project on information access, technology awareness and adoption, productivity, food security, farm income, and other indicators, in relation to comparison districts. The compar- ison districts are based on their similarities with the project districts in terms of region, agricul- tural development division (ADD), agroecological zones, farming systems, and in consultation with project partners. The comparison districts are the rest of districts in the northern region; the rest of the districts in the central region, except Mchinji; and Machinga as the comparison district for Mangochi in the southern region. Data is also collected from both female and male respond- ents within the household and will therefore also cover gender-disaggregated data on infor- mation access and technology awareness and adoption, productivity indicators by the gender of the plot decision-maker, women’s dietary diversity in addition to household dietary diversity, and the project’s impact on women’s empowerment and gender equality. The evaluation will be based on baseline and endline household surveys, complemented by qualitative interviews. Impact evaluation of the ICT hub component: Strengthening ICT hubs is a major activity in SRIEED II, and a rigorous randomized controlled trial (RCT) is being implemented to evaluate the impact of these ICT hubs. The ICT hubs work with existing groups of producers and support them with ICT tools, agricultural and marketing extension services, market linkages, and other demand-driven support that will help these groups improve their productivity and incomes. The ICT hubs aim to combine concepts of collective action, information campaigns, and marketing support to enable smallholder farmers to improve their wellbeing. A total of 118 existing farmers’ groups in the six project districts were identified by FRT to have the potential to be developed as impact ICT hubs. Based on data collected, these groups are mostly informal, consist of about 10–40 members, and include mixed-gender groups and female-only groups. Most of these groups are formed to serve multiple functions including activities related to savings and loans. We anticipate strong information spillover within the groups but minimal information spillover across groups and villages. 10 This research is set up as a cluster-randomized controlled trial (cRCT). The units of randomiza- tion are farmers’ groups, which will be targeted as ICT hubs. We used a computer program (Stata) to randomly assign groups into treatment and control. Half of these groups will be sup- ported with a comprehensive package of support including Short Message Service (SMS) push, IVR, video-based extension, and market linkage support, and the other half will be the control group and not receive the comprehensive support. All groups will have access to radios, radio programming, and a call center, which are national or district-wide in coverage and difficult to control or assign to different groups. This cRCT and impact evaluation will not only provide infor- mation on how many beneficiaries experienced improvements in the indicators but also quantify the average effect of the ICT hub on technology adoption, productivity, food security, farm in- come, and other indicators. Gender- and age-disaggregated data will also be analyzed. The evaluation will be based on baseline and endline household surveys, completed by qualitative interviews. Given random assignment to the treatment, intent-to-treat (ITT) effects are estimated by ordinary least squares (OLS), where the variable of interest is the indicator variable equal to 1 if the ICT hub was assigned to a treatment group. We will compare results from the midline and endline surveys using single difference (SD) and analysis of covariance (ANCOVA) to estimate the causal impacts of the different treatments. The outcome using SD (equation 1) and ANCOVA (equation 2) can then be written as (1) 𝑌1𝑖𝑗 = 𝛼0 + 𝛽1,𝑆𝐷𝑇𝑣 + 𝛾𝑥𝑋0𝑖𝑗 + 𝜀1𝑖𝑗 (2) 𝑌1𝑖𝑗 = 𝛼0 + 𝛽1,𝐴𝑇𝑣 + 𝛽𝑦𝑌0𝑖𝑗 + 𝛾𝑥𝑋0𝑖𝑗 + 𝜀1𝑖𝑗, where Y is the outcome indicator at time 0 (baseline) or time 1 (endline); i is the individual or household; j is the group; a is the intercept; 𝛽1 measures the average effects of the treatment T; X is a vector of control variables; and 𝜀 is the error term, which is clustered at the group level. We test the null hypothesis 𝛽1= 0. If rejected, we conclude that the treatment or intervention package has significant effect to the magnitude of 𝛽1. Using the baseline data already collected, we tested for balance in the baseline characteristics. Annex Table 1 provides the results of the test for bal- ance. Almost all baseline characteristics are similar between treatment and control households. We conclude that there is a good baseline balance between treatment and control. We also plan to conduct a detailed cost-effectiveness analysis of the interventions. Further, we propose a qualitative study to gather in-depth insights on the treatment effects and identify mech- anisms and pathways of change. We also aim to identify components that work better than others via qualitative insights and feedback from farmers. The qualitative study will consist of focus group discussions as well as interviews with household members. The qualitative component is critical for determining why the treatment options did or did not have the intended effect and to help inform the design of the most effective programs. 11 4. BASELINE DATA COLLECTION The baseline household survey was completed on July 31, 2021 and is linked with the third round household survey conducted by the International Food Policy Research Institute (IFPRI) that monitors national access to extension services and technology adoption1. The survey co- vers not only the six project districts but also comparison districts. The sampling is proportional to the size of the rural population in the district, so that each rural household has equal chance of being selected for the survey. To account for oversampling in some of the districts (for exam- ple, some districts of interest were oversampled to allow for analysis at the district level), sam- pling weights were applied so that the sample remains representative to the total rural popula- tion in the districts of interest. The baseline data for SRIEED II cover 1,767 randomly selected households in the six focus dis- tricts of the SRIEED II project and 853 households in the comparison districts (Table 2). Table 2: Total households interviewed in the baseline for the impact evaluation of the SRIEED II project 1 See details at https://www.ifpri.org/project/pluralistic-extension-system-malawi Districts Households interviewed Project districts Comparison districts Total Kasungu 246 246 Lilongwe East 244 244 Lilongwe West 312 312 Mangochi 353 353 Mzimba North 189 189 Mzimba South 246 246 Nkhota-Kota 177 177 Chitipa 27 27 Dedza 223 223 Dowa 101 101 Karonga 30 30 Machinga 232 232 Nkhata- Bay 28 28 Ntcheu 137 137 Ntchisi 20 20 Rumphi 16 16 Salima 39 39 Total 1,767 853 2,620 Source: IFPRI/Wadonda household survey (2021). 12 In addition, there are 118 rural producer groups (with 5–40 members each) in the survey, and 5–10 members of each group were randomly selected and interviewed for the baseline survey evaluating the impact of the ICT hubs. In total, 1,012 household members in these sample groups were included in the baseline survey (Table 3). The distribution of the sample house- holds in the treatment and control ICT hubs are shown in Figure 2. Table 3: Total households interviewed in the baseline survey for the impact evalu- ation of ICT hubs Districts Total Treatment Control Kasungu 169 82 87 Lilongwe 192 96 97 Mangochi 176 88 88 Mzimba North 152 82 70 Mzimba South 179 90 89 Nkhota-Kota 143 73 70 Total 1,012 511 501 Source: IFPRI/Wadonda household survey (2021). 13 Figure 2: Distribution of sample households in treatment and control groups Source of raw data: IFPRI/Wadonda household survey (2021). See Annex Figure 1 for more detailed map per cluster. We use the value of agricultural production and sales revenue as the main outcome indicators in power calculations. Based on the IFPRI survey of extension services and technology adop- tion in 2016 and 2018, the average value of crop production was MWK (Malawian kwacha) 200,900 (USD 255), and the standard deviation was MWK 244,000 (USD 310). The project tar- gets a 20% increase in the average value of production, and the minimum sample size needed to statistically detect this is 367. The average sales revenue was MWK 31,000 (USD 39), and the standard deviation was MWK 68,880 (USD 89). About half of the sample farming house- holds did not engage in crop sales. For those who did engage in crop sales, the average sales amount was MWK 62,000 (USD 79). The project targets a 30% increase in average sales reve- nue among households that were selling during baseline (increase of an average sales revenue of MWK 18,600 [USD 24]). For those not selling during baseline, the project targets a similar 30% increase over the average sales revenue among the baseline sellers (from zero to an aver- age revenue of MWK 18,600 [USD 24]). The minimum sample size needed to statistically de- 14 tect this effect size is 267. The sample sizes in the baseline survey (for both project and com- parison districts) are much more than the minimum samples required. To adjust the standard error due to clustering at the group level and to account for potential attrition, the target sample size per treatment arm is about 500 (9 members per ICT hub). In total, we interviewed 511 households in the treatment group and 501 households in the control group, for a total of 1,012 households interviewed for the ICT hub impact evaluation. 5. BASELINE DATA This section summarizes the baseline indicator data collected. Annex Table 1 shows the list of indicators used and their measurements. 5.1 Socioeconomic characteristics Gender and age group: Two classifications of gendered household types are used in this re- port: (1) household based on the gender of the designated head (female- versus male-headed households); (2) household based on the gender composition of all decisionmakers or adult members in the household: (1) dual-headed or dual-adult households; (2) households with fe- male adults only; and (3) households with male adults only. The latter is an increasing used gendered household type classification (see the widely used Women’s Empowerment Index in Agriculture (WEAI) methodology and sampling; Ragasa et al. 2019). A total of 34% of sample households are female-headed households. Looking at decision-makers within the household, 74% are households with both women and men, 23% are households with women only, and 3% are households with men only We follow similar household classifications for youth and nonyouth. Twenty-nine percent household are headed by youth. A total of 43% of households have at least one youth decision-maker or farmer. Of which, roughly a fourth of the households have one youth decision-maker or farmer, and another fourth have two youth decision-makers or farmers. Education level. Most heads and other decision-makers or farmers in the household have no formal schooling, have some years in primary or elementary, or have completed primary or ele- mentary school (77%) (Table 4). A total of 13% have no formal schooling. A total of 23% have some years in high school, completed high school, or achieve higher education. Of which, 21% have some years in high school, and only 2% completed high school or went to college or uni- versity. Household size is 5.4 in project districts and 5.1 in comparison districts. The average age of the household head is 45, 79% are married, and 72% can read or write in Chichewa, but only 35% can read or write in English. A total of 12% have been lead farmers in the past, but only 7% are currently active lead farmers. These figures are similar between project and comparison dis- tricts. 15 Table 4: Characteristics of the household head Indicators Project district Comparison districts % female 34 33 Age (mean) 45 44 % youth 29 28 % married 79 79 % who has ever been a lead farmer 12 12 % who is an active farmer 7 7 % who reads/writes Chichewa 72 77 % who reads/writes English 35 36 Education level (% of household heads) No formal schooling 12 11 Some years in elementary school 50 47 Elementary school graduate 15 18 At least some years in high school 23 24 Household size 5.4 5.1 N 1,738 826 Source: IFPRI/Wadonda household survey (2021). Farming is the main occupation by 81% of the heads and 87% of decisionmakers within the household (Table 5). Roughly 16% of heads are employed or working on family business. These figures are similar between project and comparison districts. On average, roughly 20% of decisionmakers within the household are employed or own business as their main occupation. 16 Table 5: Main occupation of head and decisionmakers in the household (% of ru- ral households) Main occupation Head Decisionmakers Project districts Comparison districts Project districts Comparison districts Farmer 80.6 71.1 87.0 81.1 Employee 4.6 4.1 5.9 5.0 Family business worker 5.0 7.9 6.8 12.3 Self-employed 6.2 8.9 8.0 11.1 Employer 0.1 0.2 0.2 0.3 Unemployed-worked before or seeking work 1.2 0.8 2.1 1.1 Non-worker, not seeking for work 0.2 0.3 1.2 1.7 Homemaker 2.4 5.8 9.8 17.5 Student 0.2 0.2 1.3 1.4 Other 0.1 1.3 0.3 2.1 N 1744 832 1766 853 Source: IFPRI/Wadonda household survey (2021). 5.2 Access to extension services The survey asks whether a male or female respondent in the household has received infor- mation or advice on agricultural production, marketing or agroprocessing, environment or cli- mate change adaptation, livestock, aquaculture, fisheries, nutrition, and nonfarm livelihoods in the last two years and in the last 12 months. In the project districts, 73% have received agricul- tural production advice, 54% have received advice on marketing or agroprocessing, 56% have received advice on environment or climate change adaptation, and 76% have received advice on nutrition or health in the last two years (Figure 3). A slightly lower proportion of households in project districts have received advice agricultural topics than in comparison districts. 17 Figure 3: Access to extension services in the last two years, by topic and project district (% of rural households) Source: IFPRI/Wadonda household survey (2021). Note: “Agriculture” is the aggregate for any advice received related to agricultural production, marketing or agroprocessing, environ- ment or climate change adaptation, livestock, aquaculture, or fisheries, and excludes advice related to nutrition and nonfarm liveli- hoods. In the project districts, 68% of rural households have received agricultural advice, 34% have re- ceived marketing or agroprocessing advice, 30% have received information on environment and climate change adaptation; and 59% have received nutrition–related advice in the last 12 months (Figure 4). A slightly lower proportion of households in project districts has received ad- vice on agricultural production and nutrition than in comparison districts, but more rural house- holds in project districts have received advice on marketing/agroprocessing, livestock, and non- farm livelihoods than in comparison districts. 82 73 54 56 50 11 6 76 27 88 75 45 60 47 8 6 82 22 0 10 20 30 40 50 60 70 80 90 100 Project districts Comparison districts 18 Figure 4: Access to extension services in the last 12 months, by topic and project district (% of rural households) Source: IFPRI/Wadonda household survey (2021). Note: “Agriculture” is the aggregate for any advice received related to agricultural production, marketing or agroprocessing, environ- ment or climate change adaptation, livestock, aquaculture, or fisheries, and excludes advice related to nutrition and nonfarm liveli- hoods. More men (67%) were accessing agricultural extension services than women (55%) in 2020/21 (Figure 5). 68 56 34 30 28 2 1 59 8 72 57 26 31 22 2 3 64 6 0 10 20 30 40 50 60 70 80 Project districts Comparison districts 19 Figure 5: Access to extension services in the last 12 months, by gender of re- spondents and project district Source: IFPRI/Wadonda household survey (2021). Note: “Agriculture” is the aggregate for any advice received related to agricultural production, marketing or agroprocessing, environ- ment or climate change adaptation, livestock, aquaculture, or fisheries, and excludes advice related to nutrition and nonfarm liveli- hoods. In the project districts, 58% of rural households have received advice through ICT tools, mostly from radio programming (57%) (Figure 6). Only 5% of households have received advice or infor- mation through SMS and 2% through TV programs. A slightly greater proportion of rural house- holds in project districts has accessed extension than in comparison districts from most sources, but the opposite is the case when information was accessed via other farmers and radio pro- gramming. 55 45 24 19 18 1 52 7 57 44 18 18 13 1 58 5 67 54 33 31 27 3 52 7 68 52 20 30 19 5 50 3 0 10 20 30 40 50 60 70 80 A gr ic u lt u re A g P ro d u ct io n M ar ke ti n g/ A gr o p ro ce ss in g E n vr io n m en t/ C lim at e C h an ge L iv es to ck A q u ac u lt u re /F is h er ie s N u tr it io n N o n -F ar m L iv e lih o o d s A gr ic u lt u re A g P ro d u ct io n M ar ke ti n g/ A gr o p ro ce ss in g E n vr io n m en t/ C lim at e C h an ge L iv es to ck A q u ac u lt u re /F is h er ie s N u tr it io n N o n -F ar m L iv e lih o o d s Project districts Comparison districts % o f sa m p le r u ra l w o m en a n d m en Female Male 20 Figure 6: Access to extension services in the last two years, by source and pro- ject districts Source: IFPRI/Wadonda household survey (2021). Note: NGO=nongovernmental organization; FBO=farmer-based organization; FFS=farmer field school; ICT=information and com- munication technology; TV=television; SMS=Short Message Service. ICT Tools consist of radio, television, internet, SMS, video, social media, mobile app and call center. In the project districts, 53% of rural households have received advice from government exten- sion agents in the last 12 months (Figure 7). A total of 53% have received agricultural advice through ICT tools, almost all of it from radio programming (52%). Only 3% of households have received advice from SMS, and 1% have received it from television programs. A greater propor- tion of rural households in project districts has accessed extension from the government, non- governmental organizations (NGOs), farmer-based organizations (FBOs), and lead farmers than in comparison districts. Similar proportions of rural households in project and comparison dis- tricts have received advice from radio programming and other ICT tools. Figure 7: Access to extension services in the last 12 months, by main source and project district (% of rural households) Source: IFPRI/Wadonda household survey (2021). 57 6 20 12 25 33 5 58 57 1.9 0.3 5.3 0.3 0.1 0.5 0.3 50 3 16 7 19 39 2 64 63 1.2 0.2 5.2 0.2 0.1 0.6 0 0 10 20 30 40 50 60 70 (% o f ru ra l h o u se h o ld s Project districts Comparison districts 53 3 17 7 21 24 3 53 52 1.1 0.4 3.4 0.3 0.3 0.1 0.2 43 2 11 3 15 25 2 52 51 0.9 0.1 3.7 0.2 0.2 0.3 0.0 0 10 20 30 40 50 60 Project district Comparison districts 21 Note: NGO=nongovernmental organization; FBO=farmer-based organization; FFS=farmer field school; ICT=information and com- munication technology; TV=television; SMS=Short Messaging Service. ICT Tools consist of radio, television, internet, SMS, video, social media, mobile app and call center. Most rural women and men have access to radio and phone and go to the nearest market at least once a week (Figure 8). These are potential sources of information and media for greater technology promotion. The project should continue the radio programming as greater than 50 percent of respondents are getting information them it. The project can expand the usage of phone/mobile platforms for extension; it is currently being utilizing for agriculture advice by <5 of respondents, but 58 percent and 65 percent of women and men respondents use phone every day and 72 percent and 85 percent of women and men respondents use phone at least once a week. Seventy-three percent of women and 85% of men go to nearest market at least once a week; the project can explore extension services and technology promotion in market place or around market days or through traders. Figure 8: Potential media for information access Source: IFPRI/Wadonda household survey (2021). 5.3 Technology awareness and adoption The questions in the baseline survey are the following:  Are you aware or have you heard of [technology 1]?  Did you ever practice or adopt it?  Why did you not practice or adopt it?  Are you still adopting it?  Why did you stop adopting it? For some agricultural technologies, the responses above are triangulated with data from the plot-level production practices in baseline survey module C. 77 15 85 24 85 60 8 72 12 73 0 10 20 30 40 50 60 70 80 90 Listens to radio at least once a week Watches television at least once a week Uses phone at least once a week Goes to nearest town at least once a week Goes to nearest market at least once a week % o f sa m p le r u ra l w o m en a n d m en Male Female 22 The survey included a total of 47 agricultural technologies, including 35 crop and livestock production technologies, 5 marketing practices, and 7 nutrition and health–related practices (Table 6). There is generally a high level of awareness of most of the technologies promoted though adoption of many of them is low (Table 7). Eight technologies have a high awareness- adoption gap (at least 40% of rural households are aware but have not adopted), and 18 technologies have a medium awareness-adoption gap (20–39% of rural housheolds are aware but have not adopted) (Table 7). The pattern is similar in the project and comparison districts. Of 47 technologies, rural households are aware of 28 technologies on average, and have tried or adopted 18 technologies on average (Table 8). On average, a rural household has disadopted one technology over time. These patterns are similar for project and comparison districts. In the project districts, female-headed or women-only households are aware of and have adopted fewer technologies than male-headed households or those with men (a difference of two technologies on average). Youth-headed households are also aware of and have adopted fewer technologies than non-youth-headed households (a difference of two technologies on average). Across individual respondents, we see some differences in technology awareness and adoption beween women and men (Figure 9-11). A significantly greater proportion of men are aware and adopt most technologies than women (Figure 9 and Figure 10). A greater proportion of men than women are aware and adopt livestock, marketing, and agroprocessing related technologies (Figure 10). Figure 9: Agricultural technology awareness Source: IFPRI/Wadonda household survey (2021). 0 10 20 30 40 50 60 70 80 90 100 % o f re sp o n d en ts Male Female 23 Figure 10: Agricultural technology adoption Source: IFPRI/Wadonda household survey (2021). Figure 11: Awareness and adoption of livestock, marketing and agroprocessing related practices 0 10 20 30 40 50 60 70 80 90 % o f re sp o n d en ts Male Female 92 69 29 14 90 62 21 9 0 20 40 60 80 100 % o f re sp o n d en ts a. Livestock technology awareness Male Female 0 20 40 60 80 100 % o f re sp o n d en ts b. Livestock technology adoption Male Female 24 Source: IFPRI/Wadonda household survey (2021). Many of the agricultural practices have not been adopted mainly because farmers lack sufficient knowledge of and exposure to the technology in order to adopt (Table 9). These agricultural practices include zero or minimum tillage, pit planting, and water harvesting. Farmers also report that these practices are time consuming or labor intensive, which discourages adoption and encourages disadoption. “[Zero or minimum tillage] requires more labour, so me being a woman I can’t manage,” says a woman respondent. For some technologies, such as planting vetivar grass, double-up legume intercropping, fodder trees in crop plots, and double-row soybean planting, the reason given for nonadoption/disadoption is lack of seeds. For less common technologies, such as double-up legume intercropping and double-row soybean planting, lack of knowledge/exposure to the technology is also cited as a common reason. Nonadoption of other practices, like fallow, occurs mainly because insufficient resources and land mean that households have to cultivate every year. For expensive inputs such as herbicides, inoculants, and hermetic bags for storage, the main reason for nonadoption is lack of funds. Agricultural technologies that require more inputs, such as fertilizer, are also not well adopted or are frequently disadopted. For example, one-one maize planting (sasakawa), despite being well-known, is not well adopted and has a high level of disadoption because it requires more fertilizer than households’ old planting practices do. For organic fertilizer, the main reason for nonadoption is the lack of material/crop/manure. For some newer practices, such as pelletized tobacco waste/manure and composting toilets, the reason is a lack of knowledge or exposure. Additionally, in the case of composting toilets, the technology is also considered time consuming and unsafe with the potenital to cause the spread of disease. Those techonologies focused on dietary diversity and nutrition, implementaiton and disadoption are mainly because of a lack of funds to buy the required food. 70 57 45 36 23 26 64 43 34 32 14 18 0 10 20 30 40 50 60 70 80 Grading or sorting out produce Use of hermetic bags for storage Collective marketing Use of Mandela cock drying for aflatoxin management Warehouse receipt system Commodity aggregation % o f fe m al e an d m al e re sp o n d en ts c. Awareness of marketing or agroprocessing practices Male Female 25 Table 6: Technology awareness and adoption (% of rural households) Technology Project districts Comparison districts Aware Ever adopted Still adopt- ing Did not adopt Dis- adopted Aware Ever adopted Still adopt- ing Did not adopt Dis- adopted Crop production technolo- gies One-one maize planting 95 68 57 27 11 98 63 39 35 24 Crop residue incorpora- tion 92 82 80 10 3 97 92 88 5 4 Crop rotation 90 71 69 19 3 85 63 59 22 4 Mixed cropping 88 73 68 15 5 96 83 79 13 4 Mulching 87 73 72 14 1 92 75 72 17 3 Planting Vetiver grass 86 53 52 32 2 89 47 44 42 3 Agroforestry 84 63 62 20 1 86 71 70 14 1 Cereal-legume 82 69 67 14 1 94 86 85 8 1 Contour bunds 76 60 60 16 1 90 80 78 10 2 Compost manure 75 54 50 21 5 84 63 50 21 14 Box ridges 75 57 56 18 1 84 65 63 18 3 General manure from do- mestic rubbish pits 73 56 53 18 2 70 48 43 22 5 Fallow 72 26 24 46 2 77 31 27 45 4 Pit planting 70 26 17 43 9 71 21 12 50 9 Zero or minimum tillage 69 28 19 41 9 80 25 15 54 10 Herbicide 67 16 11 51 5 76 15 11 61 3 Mechanical control 63 55 55 7 0 74 70 69 4 1 Proper ridge spacing 61 51 51 10 0 65 56 55 10 0 Double row soybean plant- ing 55 30 26 26 4 51 22 18 29 4 Double up legume inter- cropping 53 30 28 23 2 61 32 29 29 3 Mbeya manure 43 22 19 21 3 59 32 25 27 7 Composting toilets 41 4 3 37 1 47 10 6 38 4 Pelletized tobacco waste/manure 38 11 8 27 3 37 9 5 28 4 Water harvesting in pits or swales 34 9 8 25 1 44 11 9 33 1 Biological control 31 19 19 12 0 30 22 22 7 1 26 Inoculant 30 9 7 21 1 26 5 4 20 1 Reduced use of pesticide 29 17 16 13 0 28 15 14 13 1 Soil cover 26 18 17 8 1 39 27 27 11 1 Soil testing 18 4 3 14 1 22 5 3 18 1 Consulted a plant clinic or plant doctor 14 8 8 6 0 11 4 4 7 0 Livestock-related prac- tices Livestock/animal manure 94 74 67 20 7 96 81 70 15 11 Improved livestock hous- ing 68 28 25 40 3 65 24 17 42 7 Fodder trees in crop plots 31 8 8 23 0 36 11 11 25 0 Hay/silage making 15 3 2 12 1 21 5 5 16 1 Marketing and agropro- cessing practices Grading or sorting out produce 68 65 65 3 0 67 62 60 6 1 Use of hermetic bags for storage 50 15 13 35 2 61 15 10 45 5 Collective marketing 39 nd nd 39 nd 49 nd nd 49 nd Mandela cock drying for aflatoxin mgt 37 nd nd 37 nd 32 nd nd 32 nd Warehouse receipt system 22 nd nd 22 nd 13 nd nd 13 nd Commodity aggregation 20 nd nd 20 nd 22 nd nd 22 nd Nutrition and health–re- lated practices Washing hands before preparing and consum- ing food 99 99 99 0 0 98 98 98 0 0 Dietary diversity 93 68 66 24 2 92 63 54 29 9 Using iodized salt in food preparation 91 88 88 3 0 94 94 93 1 0 Backyard gardening 82 56 51 26 5 86 59 47 27 12 Consuming iron-rich foods 62 50 49 12 1 61 50 47 11 3 Food budgeting/food cal- endar 65 46 46 19 0 54 35 34 19 1 Orange-fleshed sweet po- tato 70 59 58 11 1 70 54 48 16 5 Source: IFPRI/Wadonda household survey (2021). Note: Nd = no data 27 Table 7: Awareness and adoption of agricultural technologies, marketing, and nu- trition and health related practices Awareness Adoption Awareness-adoption gap High awareness (>=70% aware) High adoption (at least 70% adopted) High gap (40–60%) Washing hands before preparing and consuming food Washing hands before preparing and consuming food Herbicide One-one maize planting Using iodized salt in food preparation Pit planting Livestock/animal manure Crop residue incorporation Zero or minimum tillage Dietary diversity Mulching Fallow Crop residue incorporation Medium adoption (50–69% adopted) Improved livestock housing Using iodized salt in food preparation Crop rotation Composting toilets Crop rotation Mixed cropping One-one maize planting Mixed cropping Livestock/animal manure Use of hermetic bags for storage Mulching Cereal-legume intercropping Medium gap (20–39%) Planting Vetivar grass Dietary diversity Planting Vetivar grass Agroforestry Grading or sorting out produce Backyard gardening Backyard gardening Agroforestry Pelletized tobacco waste/manure Cereal-legume intercropping Contour bunds Double row soybean planting Contour bunds Orange-fleshed sweet potato Dietary diversity Compost manure One-one maize planting Livestock/animal manure Box ridges Box ridges Compost manure Manure from domestic rubbish pits Mechanical control Water harvesting in pits or swales Fallow Manure from domestic rubbish pits Double up legume intercropping Pit planting Planting Vetivar grass Mbeya manure Orange-fleshed sweet potato Backyard gardening Fodder trees in crop plots Medium awareness (50–69% aware) Proper ridge spacing Inoculant Zero or minimum tillage Compost manure Agroforestry Improved livestock housing Some adoption (25–49% adopted) Crop rotation Grading or sorting out produce Consuming iron-rich foods Mixed cropping Herbicide Food budgeting/food calendar Box ridges Food budgeting/food calendar Double up legume intercropping Manure from domestic rubbish pits Mechanical control Double row soybean planting Food budgeting/food calendar Consuming iron-rich foods Improved livestock housing Proper ridge spacing Fallow Double row soybean planting Low adoption (<25% adopted) Double up legume intercropping Biological control Use of hermetic bags for storage Zero or minimum tillage Some awareness (25–49% aware) Mbeya manure Mbeya manure Soil cover Composting toilets Pit planting Collective marketing Reduced use of pesticide Pelletized tobacco waste/manure Use of hermetic bags for storage Mandela cock drying for aflatoxin mgt Herbicide Water harvesting in pits or swales Water harvesting in pits or swales Biological control for pests Consulted a plant clinic or plant doctor Fodder trees in crop plots Fodder trees in crop plots Inoculant Pelletized tobacco waste/manure Reduced use of pesticide Inoculant Soil cover Soil testing Low awareness (<25% aware) Composting toilets Warehouse receipt system Hay/silage making Commodity aggregation Collective marketing Soil testing Mandela cock drying for aflatoxin mgt Hay/silage making Warehouse receipt system Consulted a plant clinic or plant doctor Commodity aggregation Source: IFPRI/Wadonda household survey (2021). Note: Regular fonts indicate crop production-related technologies; in italics are marketing or agroprocessing related practices; in bold are livestock production related technologies; and in underline are food or nutrition related practices. 28 Table 8: Number of agricultural and nutrition and health related technologies known and adopted by rural households Known Ever adopted Still adopted Project districts 28 18 17 By gender of head Female-headed 27 17 17 Male-headed 29 19 18 By household type Dual-headed HH 29 19 17 HH with women-only 27 17 16 HH with men-only 28 19 18 By age group of head Youth-headed 27 17 16 Non-youth-headed 29 19 18 Comparison districts 28 18 16 By gender of head Female-headed 25 16 14 Male-headed 29 19 17 By household type Dual-headed HH 29 18 17 HH with women-only 25 16 14 HH with men-only 25 16 14 By age group of head Youth-headed 28 18 16 Non-youth-headed 28 18 16 Source: IFPRI/Wadonda household survey (2021). 29 Table 9: Reasons for not adopting and disadopting promoted agricultural technologies Reasons Zero or minimum till- age Pit planting Water harvesting in pits or swales Planting Vetivar grass to control for soil erosion Double up legume intercropping Did not adopt Disadopt Did not adopt Disadopt Did not adopt Disadopt Did not adopt Disadopt Did not adopt Disadopt We do not know much about the technology 73 0 71 0 79 0 21 5 55 0 Lack of materials/seeds/soil cover/mulch/manure 12 11 9 3 4 0 60 39 33 59 Lack of funds or resources to implement 6 8 8 8 10 0 7 13 6 6 We are not interested (e.g., have rented land, believe that it will destroy soil fertility in the case of herbicide, consider composting toilet disgusting) 2 10 3 4 0 0 2 4 1 13 We are lazy 1 11 3 9 2 11 1 0 0 0 Time-consuming to implement 0 9 2 25 0 29 0 5 0 0 Not applicable or appropriate (e.g., waterlogging area, land not for legumes, area is free from erosion, not growing soya, not raising livestock, etc.) 0 3 0 3 2 24 5 17 1 0 We do not see benefit (e.g., more pests/termites, crops do not grow well, livestock destroy garden, etc.) 1 9 0 4 0 0 1 13 4 10 We are used to/prefer old/other practice (e.g., traditional ridging, animal or compost pit manure, or inorganic fertilizer, dimba gardens) 3 11 0 4 1 0 1 0 0 0 Lack of water 0 0 0 0 0 0 0 0 0 0 Theft 0 0 0 0 0 0 0 0 0 0 Not enough land 0 5 0 3 0 0 0 4 0 6 It requires more fertilizer or other inputs 2 17 4 34 3 24 2 0 1 6 Inability/cannot manage/sick/old age 1 5 1 3 0 12 0 0 0 0 Source: IFPRI/Wadonda household survey (2021). 30 Table 9: Reasons for not adopting and disadopting agricultural technologies being promoted. (continued) Reasons Double row soybean planting Fodder trees in crop plots Backyard gardening One-one planting (Sasakawa) Fallow Did not adopt Disadopt Did not adopt Disadopt Did not adopt Disadopt Did not adopt Disadopt We do not know much about the technology 58 0 52 0 14 0 26 2 23 0 Lack of materials/seeds/soil cover/mulch/manure 32 36 37 45 36 12 31 9 23 6 Lack of funds or resources to implement 5 9 8 55 16 7 35 10 31 6 We are not interested (e.g., have rented land, believe that it will destroy soil fertility in the case of herbicide, con- sider composting toilet disgusting) 0 0 0 0 2 8 1 4 0 0 We are lazy 0 5 1 0 6 7 1 0 0 0 Time-consuming to implement 0 0 0 0 1 3 0 3 0 0 Not applicable or appropriate (e.g., waterlogging area, land not for legumes, area is free from erosion, not growing soya, not raising livestock, etc.) 2 11 1 0 1 3 1 0 1 0 We do not see benefit (e.g., more pests/termites, crops do not grow well, livestock destroy garden, etc.) 0 2 0 0 8 10 1 1 0 0 We are used to/prefer old/other practice (e.g., traditional ridging, animal or compost pit manure, or inorganic fer- tilizer, dimba gardens) 0 3 0 0 5 12 1 4 1 0 Lack of water 0 0 0 0 7 23 0 0 0 0 Theft 0 0 0 0 0 0 0 0 0 0 Not enough land 0 15 0 0 0 3 0 0 0 88 It requires more fertilizer or other inputs 1 18 0 0 4 6 4 65 21 0 Inability/cannot manage/sick/old age 0 0 0 0 0 4 0 3 0 0 Source: IFPRI/Wadonda household survey (2021). 31 Table 9: Reasons for not adopting and disadopting agricultural technologies being promoted. (continued) Reasons Herbicide Inoculant Hermetic bags for storage Dietary diversity Food budgeting or food calendar Did not adopt Disadopt Did not adopt Disadopt Did not adopt Disadopt Did not adopt Disadopt We do not know much about the technology 17 0 56 0 19 0 7 0 41 0 Lack of materials/seeds/soil cover/mulch/manure 11 9 13 12 5 6 7 11 6 36 Lack of funds or resources to implement 70 78 30 83 76 77 86 77 53 64 We are not interested (e.g., have rented land, believe that it will destroy soil fertility in the case of herbicide, con- sider composting toilet disgusting) 1 2 0 0 0 6 0 0 0 0 We are lazy 0 2 0 0 0 6 0 0 0 0 Time-consuming to implement 0 0 0 0 0 6 0 0 0 0 Not applicable or appropriate (e.g., waterlogging area, land not for legumes, area is free from erosion, not growing soya, not raising livestock, etc.) 1 4 1 5 0 0 0 0 0 0 We do not see benefit (e.g., more pests/termites, crops do not grow well, livestock destroy garden, etc.) 1 0 0 0 0 0 0 0 0 0 We are used to/prefer old/other practice (e.g., traditional ridging, animal or compost pit manure, or inorganic fer- tilizer, dimba gardens) 0 2 0 0 0 0 0 0 0 0 Lack of water 0 0 0 0 0 0 0 0 0 0 Theft 0 0 0 0 0 0 0 0 0 0 Not enough land 0 4 0 0 0 0 0 0 0 0 It requires more fertilizer or other inputs 0 0 0 0 0 0 0 0 0 0 Inability/cannot manage/sick/old age 0 0 0 0 0 0 0 12 0 0 Source: IFPRI/Wadonda household survey (2021). 32 Table 9: Reasons for not adopting and disadopting agricultural technologies being promoted. (continued) Reasons Composting toilets Pelletized tobacco waste/manure Mbeya manure Livestock/animal manure Improved livestock housing Did not adopt Disadopt Did not adopt Disadopt Did not adopt Disadopt Did not adopt Disadopt We do not know much about the technology 74 0 45 0 60 0 14 0 15 0 Lack of materials/seeds/soil cover/mulch/manure 11 13 35 26 20 34 38 30 28 9 Lack of funds or resources to implement 9 11 19 39 18 12 36 27 47 8 We are not interested (e.g., have rented land, believe that it will destroy soil fertility in the case of herbicide, con- sider composting toilet disgusting) 2 30 0 3 0 7 1 3 0 0 We are lazy 1 0 0 3 0 7 2 2 1 5 Time-consuming to implement 0 12 0 0 0 11 0 3 0 0 Not applicable or appropriate (e.g., waterlogging area, land not for legumes, area is free from erosion, not growing soya, not raising livestock, etc.) 0 0 1 21 0 2 2 16 6 43 We do not see benefit (e.g., more pests/termites, crops do not grow well, livestock destroy garden, etc.) 0 16 0 3 0 0 0 1 0 8 We are used to/prefer old/other practice (e.g., traditional ridging, animal or compost pit manure, or inorganic fer- tilizer, dimba gardens) 1 0 0 6 1 15 3 14 0 8 Lack of water 0 0 0 0 0 0 0 0 0 0 Theft 0 0 0 0 0 0 0 0 2 14 Not enough land 0 0 0 0 0 0 0 1 0 0 It requires more fertilizer or other inputs 1 17 0 0 1 10 1 1 0 5 Inability/cannot manage/sick/old age 0 0 0 0 0 2 0 1 0 0 Source: IFPRI/Wadonda household survey (2021). 32 5.4 Outcome indicators 5.4.1 Food security Several indicators of household and individual-level food security are used. Household Hunger Scale (HHS)2 measures the extent of hunger or severe food insecurity (Ballard et al. 2008; Deitchler et al. 2010). The HHS is most appropriate for use in areas of substantial food inse- curity. The HHS covers a recall period of 30 days and consists of two types of questions: nine “occur- rence” and nine “frequency-of-occurrence” questions. The respondent is first asked if a given condition was experienced (yes or no) and, if it was, then with what frequency (rarely, sometimes, or often). The three questions asked are: Q1. In the past 30 days, was there ever no food to eat of any kind in your house because of lack of resources to get food? Q2. In the past 30 days, did you or any household member go to sleep at night hungry because there was not enough food? Q3. In the past 30 days, did you or any household member go a whole day and night without eating anything at all because there was not enough food? Household Food Insecurity Access Score (HFIAS)3 is an expanded version of HHS and covers nine conditions. Similar to HHS, HFIAS covers a recall period of 30 days, and consists of two types of questions: nine “occurrence” and nine “frequency-of-occurrence” questions. The respondent is first asked if a given condition was experienced (yes or no) and, if it was, then with what frequency (rarely, sometimes, or often). The resulting responses can be transformed into either a continuous or a categor- ical indicator of food (in)security. When calculating the HFIAS as a continuous indicator, each of the nine questions is scored from 0 to 3, with 3 being the highest frequency of occurrence; and the score for each is added together. The total HFIAS can range from 0 to 27, indicating the degree of insecure food access. As a categorical variable, households are categorized as food secure, mildly food inse- cure, moderately food insecure, or severely food insecure (for more details see Table 4 in Coates et al. 2007). Households that respond affirmatively to the more severe behaviors (or experience them more frequently) are classified as more severely food insecure. Household Dietary Diversity Score (HDDS)4 is a population-level indicator of household food access. Household dietary diversity can be described as the number of food groups consumed by a household over a given reference period, and it is an important indicator of food security for many reasons. A more diversified household diet is correlated with caloric and protein adequacy, percentage of protein from animal sources, and household income (Swindale and Bilinsky 2006). The HDDS indicator provides a glimpse of a household’s ability to access food as well as its socioeconomic status, based on consump- tion over the previous 24 hours (Kennedy et al. 2011). An option is to establish a target using the aver- age dietary diversity of 33% of households with the highest diversity. The 10 food groups included in HDDS are as follows: 1. Cereals 2 https://inddex.nutrition.tufts.edu/data4diets/indicator/household-hunger-scale-hhs 3 https://inddex.nutrition.tufts.edu/data4diets/indicator/household-food-insecurity-access-scale-hfias; http://www.fao.org/fileadmin/user_up- load/eufao-fsi4dm/doc-training/hfias.pdf 4 https://inddex.nutrition.tufts.edu/data4diets/indicator/household-dietary-diversity-score-hdds http://www.fao.org/fileadmin/user_upload/eufao-fsi4dm/doc-training/hfias.pdf http://www.fao.org/fileadmin/user_upload/eufao-fsi4dm/doc-training/hfias.pdf http://www.fantaproject.org/sites/default/files/resources/HDDS_v2_Sep06_0.pdf http://www.fao.org/fileadmin/user_upload/wa_workshop/docs/FAO-guidelines-dietary-diversity2011.pdf https://inddex.nutrition.tufts.edu/data4diets/indicator/household-hunger-scale-hhs https://inddex.nutrition.tufts.edu/data4diets/indicator/household-food-insecurity-access-scale-hfias http://www.fao.org/fileadmin/user_upload/eufao-fsi4dm/doc-training/hfias.pdf http://www.fao.org/fileadmin/user_upload/eufao-fsi4dm/doc-training/hfias.pdf https://inddex.nutrition.tufts.edu/data4diets/indicator/household-dietary-diversity-score-hdds 33 2. Roots, tubers, and plantains 3. Legumes and nuts 4. Vegetables 5. Meat, fish, and animal products 6. Dairy 7. Fruits 8. Vegetables 9. Sugar, sugar products, and honey 10. Fats and oil 11. Spices and condiments Food Consumption Score (FCS)5 is a proxy indicator of household caloric availability. The FCS ag- gregates household-level data on the diversity and frequency of food groups consumed over the previ- ous seven days, and then weights that result according to the relative nutritional value of the consumed food groups. For instance, food groups containing nutritionally dense foods, such as animal products, are given greater weight than those containing less nutritionally dense foods, such as tubers. The score ranges from 0 to 126. Based on this score, a household’s food consumption can be further classified into one of three categories: 0–21 is Poor, 21.5–35.0 is Borderline, and >35 is Acceptable. This indica- tor is useful for categorizing and tracking households’ food security across time, specifically as a proxy for the quantity dimension (i.e. caloric sufficiency) for which this indicator has been validated. In addition to these standard food security and dietary diversity indicators, we also include the percent of households consuming all six required food groups being promoted in Malawi―cereals and roots, legumes and nuts, meat and fish, vegetables, fruits, and fats/oil. The HHS is low in both project and comparison districts. Almost all rural households in the project and comparison districts have experienced “little or no hunger” (Table 10). Female-headed households, youth-headed households, and women-only and men-only households are more hungry than other household groups. Using the HFIAS we find that 30% of households in project districts are food secure, 10% are mildly food insecure, 40% are moderately food insecure, and 20% are severely food insecure. The FCS is 50.4 out of 126, and the HDDS is 7.4 out of 10 food groups. Most households (76%) have an acceptable FCS, but only 36% of rural households have consumed all six required food groups be- ing promoted in Malawi. 5 https://inddex.nutrition.tufts.edu/data4diets/indicator/food-consumption-score-fcs?back=/data4diets/indicators https://inddex.nutrition.tufts.edu/data4diets/indicator/food-consumption-score-fcs?back=/data4diets/indicators 34 Table 10: Food security and dietary diversity indicators (% rural households except for the scores) Indicators Project districts Compa-ri- son dis- tricts FHH MHH YHH NYHH DHH Women- only Men- only Household hunger score (0–6) Mean 0.3 0.5 0.4 0.3 0.4 0.3 0.3 0.4 0.5 Median 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 SD 0.9 1.0 1.0 0.8 1.0 0.8 0.9 1.0 1.0 HH categories based on HHS Little or no hunger 91 86 88 92 88 92 92 88 82 Moderate hunger 8 12 10 7 10 6 7 10 18 Severe hunger 2 2 2 1 2 1 2 1 0 HH categories based on Household food insecurity score (HFIAS) (0-27) Food secure 30 30 30 30 30 30 30 30 40 Mildly food insecure 10 10 10 10 10 10 10 10 10 Moderately food insecure 40 40 40 40 40 40 40 40 30 Severely food insecure 20 30 20 20 20 20 20 20 30 Household dietary diversity (HDDS) (0–10) 7.4 7.0 7.2 7.6 7.4 7.5 7.5 7.3 7.2 Cereals 100 100 100 100 100 100 100 100 100 Roots, tubers and plantain 77 78 74 78 75 78 78 73 63 Legumes and nuts 82 78 81 83 82 82 83 80 70 Vegetables 96 97 95 97 96 96 97 94 93 Meat, fish, and animal products 57 39 53 59 58 56 56 57 72 Fruits 57 50 52 60 57 57 58 55 59 Milk and milk products 25 19 23 27 26 25 26 24 25 Fats/oil 87 81 87 87 87 87 87 86 90 Sugar/sugar products/honey 70 62 66 73 70 70 72 67 57 Spices/condiments 94 94 94 94 94 95 94 94 96 HH categories based on Food consump- tion score (FCS) (0–126) 50.4 44.4 47.9 52.0 49.7 51.0 50.9 48.8 51.7 Poor FCS 4 5 4 3 4 4 3 4 9 Borderline FCS 20 27 26 17 22 19 18 26 23 Acceptable FCS 76 68 70 80 74 78 79 70 68 Consumes all 6 main food groups 36 26 32 39 35 37 36 36 39 N 1734 844 559 1159 727 1006 1317 369 47 Source: IFPRI/Wadonda household survey (2021). Note: FHH=female-headed household; MHH=male-headed household; YHH=youth-headed household; NYHH=non-youth-headed household; DHH=dual-headed household; HH=household; SD=standard deviation. 35 We also include measures at the individual level, particularly for women in rural households. Evidence suggests that women are more likely to be food insecure within the household and that the nutrition of mothers is highly correlated with the nutrition of children. We used the standard measure of women’s dietary diversity score, and expanded it to provide a proxy measure of the proportion of the sample women who are eating the minimum nutritious diet. The Minimum Dietary Diversity for Women (MDD- W) is an internationally validated proxy indicator for the probability of micronutrient adequacy. Women aged 15 to 49 years are more likely to have achieved micronutrient adequacy if they consume on aver- age at least 5 out of 10 healthy food groups in a 24-hour period (FAO and FHI360 2016; Martin-Prével et al. 2015). There is no validated cut-off for other age groups. In this study, we consider the primary woman decision-maker in the household, including those above 49 years, and adopt the MDD-W threshold of 5 food groups as an indication of improved probability of adequate dietary diversity and mi- cronutrient adequacy. In the baseline survey, 77% of women are aged 15–49 in both project and com- parison districts. Interviews followed FAO and FHI360 (2016) recommendations on food groupings for the 24-hour recall period and interview processes. The women’s dietary diversity outcomes monitored are (1) women dietary diversity score (WDDS) measured by the number of healthy food groups con- sumed (0–10), (2) inadequate dietary diversity (<5 food groups), and (3) consumption of each food group. The 10 food groups used to estimate WDDS, based on MDDW food groupings, are the follow- ing: 1. Cereals 2. Pulses 3. Nuts and seeds 4. Dairy 5. Meat and fish 6. Eggs 7. Dark green leafy vegetables 8. Vitamin A–rich fruits and vegetables 9. Other vegetables 10. Other fruits We also include an indicator aligned with the six promoted food groups in Malawi: staples, legumes and nuts, meat and fish, vegetables, fruits, and fats and oils. We added an indicator on the percent of women respondents consuming all six food groups. In the project districts, the average WDDS was 4.2 out of 10 food groups (Table 11).6 Only 38% of women have a WDDS of at least 5 out of 10 food groups. Only 12% of women in project districts con- sume all 6 required food groups being promoted in Malawi. Similar patterns are observed in project and comparison districts and across household types. 6 The score is much lower than HDDS because different food groups are used. The WDDS uses 24-hour recall versus seven-day recall and it adapts a different process of enumeration in which respondents enumerate all the foods that they ate the day before the survey, and the enu- merators record the food group versus yes-no question in the HDDS above. 36 Table 11: Women's dietary diversity indicators (% of rural women respondents, except the scores) Indicators Project dis- tricts Comparison districts FHH MHH YHH NYHH DHH Women- only Women's dietary diversity score (0–10) 4.2 3.9 4.3 4.2 4.2 4.2 4.2 4.4 Cereals 99 100 100 99 99 100 99 100 Pulses 32 35 32 31 31 32 31 33 Nuts and seeds 39 30 42 37 36 42 37 44 Dairy 8 4 9 7 10 6 8 9 Meat and fish 46 40 47 46 48 45 46 45 Eggs 11 6 14 9 13 9 9 15 Dark green leafy vegetables 58 58 56 59 56 59 56 61 Vitamin A–rich fruits and vegetables 37 33 37 37 36 38 36 40 Other vegetables 79 75 75 81 77 80 79 77 Other fruits 16 13 17 16 18 15 16 19 % of women with score >=5 38 31 38 38 37 39 37 40 % who consumes all 6 food groups 12 6 13 11 12 11 10 15 N 1444 712 490 940 604 839 1091 329 Source: IFPRI/Wadonda household survey (2021). Note: FHH=female-headed household; MHH=male-headed household; YHH=youth-headed household; NYHH=non-youth-headed household; DHH=dual-headed household; HH=household. 5.4.2 Agricultural income Several indicators are used to measure agricultural income: (1) productivity (kg/ha); (2) value of crop production (MWK); (3) percent of crops sold; (4) value of crop sales revenue (MWK); and (5) percent of households with other (non-crop) agricultural incomes. Maize productivity was 2,275 kg/ha on average in the project districts (Table 12). In the 2020–21 crop- ping season, about 10% of the plots were not harvested during the survey time, and respondents were asked to estimate their likely harvest given how the crop had been growing and compared to last sea- son’s experience. Including those responses did not change the crop productivity estimates. 37 Table 12: Crop productivity (kg/ha) during 2020-21 cropping season Project districts Comparison districts Crop Mean SD Median N Mean SD Median N Maize (Chimanga) 2275 1727 1853 1737 2123 1824 1544 831 Rice (Mpunga) 1961 1544 1641 117 1878 2169 1025 89 Other cereals (wheat, millet, sorghum) 789 803 494 52 1288 1286 855 73 Tubers/rootcrops 3508 3011 2471 338 3553 4051 1883 148 Beans (Nyemba) 911 834 741 867 915 1147 544 520 Groundnut (Mtedza) 1391 1284 1030 767 1797 2014 1149 254 Vegetables 3018 3344 1400 315 2145 3028 988 202 Oilseeds 747 680 692 24 2705 4040 890 4 Fibers 6670 3701 8031 18 2192 2547 1147 10 Tobacco (Fodya) 1690 1453 1369 124 1156 1119 988 36 Other (specify) 1343 939 1236 82 1581 1604 912 75 Total 1895 1823 1378 4441 1924 2291 1206 2246 Source: IFPRI/Wadonda household survey (2021). Note: SD=standard deviation. N=number of plots. In project districts, 4% of maize harvested was sold by survey time (July) on average, which is similar to the 2018 figure (Table 13). Comparison districts had a similar figure. On average, 24% of total crop har- vested was sold by survey time, which is slightly greater than in 2018 and comparison groups. 38 Table 13: Percentage of crop harvest sold 2021 2018 Mean Median SD N Mean Median SD N Project districts Maize 4 0 10 1530 6 0 14 793 Rice 35 30 34 110 33 10 36 18 Other cereals 5 0 17 39 6 0 12 29 Tubers/Rootcrops 53 70 41 124 31 10 35 132 Beans 56 75 39 687 40 37 37 422 Groundnut 25 0 32 716 23 0 29 381 Vegetables 6 0 22 262 5 0 18 316 Oilseeds 86 100 31 16 64 100 46 18 Fibers 54 50 44 6 69 100 48 9 Tobacco 75 100 39 118 88 100 29 106 Weighted % of harvest sold 24 0 35 21 0 33 Comparison districts Maize 5 0 12 808 6 0 14 843 Rice 24 0 32 90 27 20 30 64 Other cereals 2 0 7 66 3 0 15 95 Tubers/Rootcrops 54 50 38 47 44 50 35 103 Beans 43 43 40 396 32 5 37 473 Groundnut 21 0 31 249 20 0 30 350 Vegetables 9 0 25 178 4 0 19 184 Oilseeds 90 100 22 3 43 0 54 6 Fibers 86 100 36 5 89 100 30 17 Tobacco 64 100 48 34 96 100 18 71 Weighted % of harvest sold 19 0 32 20 0 33 Source: IFPRI/Wadonda household survey (2021). Note: SD=standard deviation; N=number of observations. In the project districts, the average crop production value was MWK 328,518 (median=MWK 153,000) and the crop sales revenue was MWK 121,861 (median=MWK 13,500) and many households had zero sales (Table 14). These figures are higher in project districts than in comparison districts. Figures are lower for female-headed households, youth-headed households, and households with women only and men only compared to other households. Only 60% of farmers were selling harvested crops in both pro- ject and comparison districts. In addition to crop sales, 65% of rural households have other (non-crop) agricultural incomes, mainly from livestock and by-products sales, in both project and comparison dis- tricts. 39 Table 14: Crop production value and sales revenue during 2020-21 cropping season Indicators Project districts Comparison districts FHH MHH YHH NYHH DHH Women- only Men- only Total crop produc- tion value (MWK) Mean 328,518 281,864 258,703 366,092 285,797 360,473 369,468 213,933 182,191 Median 153,000 116,200 118,500 183,000 126,250 172,000 178,200 105,000 118,000 SD 637,042 735,643 650,721 630,042 623,739 645,278 704,244 371,933 205,758 Total crop sales revenue (MWK) (as of July 2021) Mean 121,861 79,945 88,580 139,601 117,180 125,363 147,642 49,411 32,070 Median 13,500 6,000 10,000 15,000 12,500 15,000 17,500 8,750 2,000 SD 506,514 523,560 558,315 479,354 550,801 470,930 579,875 142,199 100,369 N 1746 849 553 1177 729 1017 1332 365 49 Source: IFPRI/Wadonda household survey (2021). Note: FHH=female-headed household; MHH=male-headed household; YHH=youth-headed household; NYHH=non-youth-headed household; DHH=dual-headed household; HH=household; SD=standard deviation; N=number of observations. USD 1=MWK 813 as of August 30, 2021. 5.4.3 Resilience capacity Climate shocks have devastating effects on the well-being and livelihoods of the poor, including higher risk of mortality (Kahn 2005), reductions in productive assets (Carter et al. 2007; Carter and Barrett 2006; Hoddinott 2006), and declining human capital (Alderman et al. 2006; Dercon et al. 2005). Given the extent of these effects, substantial attention has been paid to measure resilience, or “the capacity to avoid and escape from [poverty] over time and in the face of myriad stressors and shocks” (Barrett and Constas 2014) and understand the factors that affect and promote it. In Malawi, persistent drought has required humanitarian assistance while development programs seek to protect households from the risk of future shocks (Ellis and Manda 2012). Aside from weather-related shocks, the COVID-19 crisis has also caused income loss, sickness or death in the family, reduced food access and consumption, and deepened poverty and hunger for many households. To measure resilience capacity, we use various indicators including the degree of crop diversification, livelihood and income diversification, assets (including land and livestock), adoption of climate-smart agricultural technologies, and coping strategies. Access to extension services, especially on environ- ment and climate change, is also added as an indicator for resilience capacity. In the project districts, 68% of rural households have received agricultural advice and 30% have received advice and infor- mation specifically on environment and climate change in the last 12 months. These percentages are similar in comparison districts and across household types. Crop diversification. In the project districts, the average land cultivated in 2020–21 was 2.6 acres (or roughly 1 hectare) (median=2.0 acres) (Table 15). On average, 63% of acreage was cultivated with maize while the other 37% was cultivated with non-maize crops, mostly legumes. The Simpson index for crop diversification (SID) is defined at the household level as 1 minus the summation of the square 40 of share of land allocated to each crop. The SID is 0 if all land is allocated to one crop. Some house- holds have a SID close to 1, determined by both the number of crops grown and how equally land is allocated across crops. The SID for rural households in the project districts is 0.4 on average (me- dian=0.5). Table 15: Crop acreage and diversification during 2020-21 cropping season Total acres cultivated % of land with maize SID (0–1) N Mean Median SD Mean Median SD Mean Median SD Project districts 2.6 2.0 2.0 63 60 25 0.4 0.5 0.2 1747 By gender of head Female-headed 2.2 1.7 1.8 65 62 24 0.4 0.5 0.2 553 Male-headed 2.8 2.3 2.1 62 60 26 0.4 0.5 0.2 1177 By household type Dual-headed HH 2.8 2.3 2.2 63 60 26 0.4 0.5 0.2 1332 HH with women-only 2.0 1.5 1.5 64 62 25 0.4 0.5 0.2 365 HH with men-only 2.0 1.5 1.4 67 67 24 0.4 0.4 0.2 49 By age group of head Youth-headed 2.3 2.0 1.9 63 60 26 0.4 0.5 0.2 729 Non-youth-headed 2.8 2.3 2.1 63 60 25 0.4 0.5 0.2 1017 Comparison dis- tricts 2.5 2.0 1.9 64 67 26 0.4 0.4 0.2 849 By gender of head Female-headed 1.9 1.5 1.5 67 67 27 0.4 0.4 0.2 212 Male-headed 2.6 2.0 1.9 64 67 26 0.4 0.4 0.2 619 By household type Dual-headed HH 2.6 2.0 2.0 64 67 26 0.4 0.4 0.2 705 HH with women-only 1.9 1.5 1.4 68 70 28 0.3 0.4 0.3 123 HH with men-only 2.1 1.7 1.7 63 67 27 0.4 0.4 0.2 21 By age group of head Youth-headed 2.3 2.0 1.7 64 67 27 0.4 0.4 0.2 451 Non-youth-headed 2.7 2.0 2.2 64 67 26 0.4 0.4 0.2 398 Source: IFPRI/Wadonda household survey (2021). Note: SID=Simpson index for crop diversification; HH=household; SD=standard deviation; N=number of observations. Livelihood diversification. More-diversified livelihoods or income sources reduce vulnerability and in- crease the likelihood of smoothing consumption. We considered 11 livelihood activities―crop farming, tree crop production or harvesting, livestock raising, aquaculture or fisheries, trading, processing, other nonfarm business, farm employment, nonfarm employment, receiving remittances, and receiving pen- 41 sion. A rural household in the project districts is engaged in 3–4 of these activities on average and re- ceived income or sales revenue from 2–3 of these activities (Table 16). While almost all rural house- holds in the sample are engaged in crop farming, only 60% sold any of the crop harvest. Approximately 75% of rural households raise livestock, and 65% sold livestock or by-products in the last 12 months. On average, a rural household owns and raises about 10 livestock units. A total of 43% of households in the project districts have tree crops, but less than 1% sold tree crops. Aquaculture or fisheries are not common in the sample households, with less than 1% involved. Slightly more than 20% of households are engaged in trading, other nonfarm businesses, or farm employment. Households in project and comparison districts engage in similar livelihood strategies with the exception of tree crop production and harvesting and remittances where households in project districts tend to engage more. Table 16: Percentage of rural households engaged in different livelihoods Activity Project districts Comparison districts % of rural households engaged in … 1 Crop farming 98.6 99.7 2 Crop sales 59.9 54.2 3 Tree crop production or harvesting 43.3 28.7 4 Tree crops sales 0.4 1.4 5 Livestock raising 74.5 77.2 6 Livestock sales 64.7 65.2 7 Aquaculture or capture fisheries 0.6 1.3 8 Fish sales 0.6 1.3 9 Trading 21.3 22.6 10 Processing (buying raw materials and processing) 6.2 5.4 11 Other own enterprises or businesses 22.4 22.9 12 Farm employment 28.3 34.2 13 Nonfarm salary/wage employment 8.3 11.3 14 Remittances/gifts from relatives or friends 14.4 8.1 15 Pension 0.7 0.3 Livelihood Diversification Index (count: 1,3,5,7, 9-15) 3.2 3.1 Income Diversification Index (count: 2,4,6,8,9-15) 2.3 2.3 Number of Livestock units 9.7 11.7 N 1767 853 Source: IFPRI/Wadonda household survey (2021). Note: N=number of observations. Coping strategies. When facing crises or shocks, households often resort to coping strategies that negatively affect their well-being. We measure coping strategies using the Coping Strategies Index (CSI), a composite score that weights the frequency with which a household engages in a negative 42 coping strategy (over the past seven days) by the severity of that strategy (Maxwell et al. 2003; Knip- penberg et al. 2018), although we do not use weights. COVID-19 affected many households in Malawi in 2020, but 57% of rural households say they were un- affected (Table 17). Of those that were, 4% used up savings, less than 1% sold land or other assets, 4% reduced nonfood expenditure, and 5% reduced food expenditure. These patterns are generally sim- ilar in comparison districts. In project districts, more male-headed households, men-only households, and youth-headed households have resorted to coping strategies than their counterparts. Table 17: Percentage of rural households using coping strategies during the COVID-19 crisis Not af- fected Affected but did nothing Bor- rowed Used sav- ings Sold land/ as- sets Reduced nonfood exp. Reduced food exp. Negative coping indi- cator Project districts 57.3 10.7 5.1 4.1 0.3 4.0 5.5 6.4 By gender of head Female-headed 59.7 11.5 5.0 3.4 0.0 2.8 3.9 4.3 Male-headed 55.9 10.4 5.3 4.5 0.4 4.6 6.2 7.5 By household type Dual-headed HH 57.0 10.2 5.0 4.0 0.4 4.2 5.6 6.8 HH with women-only 57.9 12.5 5.4 5.0 0.0 3.5 4.6 4.8 HH with men-only 58.9 12.0 6.8 0.4 0.0 4.7 8.4 8.4 By age group of head Youth-headed 59.6 11.6 5.9 4.4 0.6 5.5 7.6 8.6 Non-youth-headed 55.6 10.1 4.6 3.9 0.0 2.9 3.8 4.7 Comparison districts 55.9 11.4 6.5 4.5 0.9 7.1 7.8 10.7 By gender of head Female-headed 61.5 10.0 4.4 1.9 0.2 5.0 7.4 8.1 Male-headed 54.4 11.2 7.3 5.3 1.1 7.8 8.1 11.8 By household type Dual-headed HH 55.7 11.1 7.1 5.2 1.1 8.0 8.4 11.7 HH with women-only 60.4 10.3 3.1 2.2 0.0 2.8 6.0 6.7 HH with men-only 34.5 26.3 11.4 0.0 0.0 5.0 2.2 5.0 By age group of head Youth-headed 58.9 11.6 7.3 4.3 1.0 6.9 7.0 10.3 Non-youth-headed 52.7 11.2 5.7 4.8 0.7 7.2 8.8 11.1 Source: IFPRI/Wadonda household survey (2021). Note: HH=household 43 Adoption of climate-smart or sustainable technologies: In the project districts, the average number of climate-smart technologies adopted by rural households is between 8 and 9 (Table 18). This figure is similar between project and control districts, but it is slightly lower for female-headed, youth-headed, and households with women-only or men-only households. Irrigation and water harvesting technologies are not commonly adopted. Table 18: Percentage of rural households adopting climate smart technologies Technology Project districts Comparison districts FHH MHH DHH Women- only HH Men- only HH YHH NYHH Soil cover 20 23 15 22 19 20 22 12 13 Zero or minimum tillage 32 29 29 34 28 35 33 25 33 Mulching 75 75 76 75 73 78 75 76 80 Pit planting 26 25 20 29 23 28 29 17 18 Crop rotation 75 68 69 78 72 78 78 66 68 Cereal-legume intercropping 66 77 65 66 67 65 65 67 80 Double up legume intercrop- ping 35 31 34 36 31 38 36 34 31 Agroforestry 70 68 65 72 68 71 71 64 78 Mixed cropping 69 79 67 70 70 69 70 66 81 Box ridges 61 65 56 63 58 62 62 55 57 Contour bunds 63 74 56 67 62 65 67 54 52 Planting Vetivar grass 58 51 52 60 55 60 61 49 44 Water harvesting in pits or swales 11 10 10 11 11 10 11 8 12 Backyard gardening 62 60 57 65 63 62 64 54 65 Erosion Control 65 75 62 66 62 66 65 64 56 Irrigation 2 2 1 2 2 1 2 0 4 Count of components 8.3 8.6 7.7 8.6 8.0 8.5 8.5 7.4 8.1 Source: IFPRI/Wadonda household survey (2021) 6. KEY OBSERVATIONS AND NEXT STEPS Based on the baseline survey results, five key observations stand out. First, the project districts have very low incidence of severe hunger and households had adequate access to energy and calorie re- quirements. However, project districts had low dietary quality measured in terms of minimum food groups consumed to achieve dietary diversity and micronutrient adequacy. There is a huge opportunity for the SRIEED II project to help improve this. Second, there is also a huge opportunity for greater awareness and promotion of marketing practices, which is currently low. Third, access to extension ser- vices that share advice on livestock and environmental practices remains low and is an area needing greater promotion by the project. Fourth, lack of knowledge and information is the primary reason that farmers do not adopt many technologies, despite being aware of those technologies. This indicates the 44 need for more intensive and frequent promotion of various technologies, including the use of farm demonstrations to improve farmers’ understanding of their potential and application. Lastly, while non- maize crops are being planted, the level of commercialization remains low, with rural households selling only 20% of their crop produce on average. Improving maize productivity, venturing into high-value cash crops, and supporting farmers with market linkages are some strategies that the project can fur- ther invest in. Baseline data also shows that project and comparison districts are similar in almost all of the demo- graphic and outcome indicators measured giving us confidence that the comparison districts can be used as the control group against which the outcome indicators can be compared. Once households are disaggregated by the gendered household types, some of the indicators begin to show differences between various groups. Nonetheless, we will continue to monitor indicators along these categories. Within the household, women are equally involved in agriculture as men and most plots are jointly man- aged by both women and men. While the gender gap has been reducing over time, some differences in access to extension services and technology awareness and adoption between female and male re- spondents remain. This is another opportunity for the project to further improve gender equity in these indicators. We will expand our analysis of the baseline data to determine gendered differences in many of the indi- cators, as well as differences across youth and non-youth. We will also analyze and measure dimen- sions of women’s empowerment to determine to what extent the interventions can effectively close the gender gap. We also collected data and information on the activities, membership, and functions of pro- ducers’ groups as potential impact ICT hubs. In the future we will be analyzing these data to further un- derstand how the SRIEED II project can strengthen these groups. We will also start discussions with Farm Radio Trust on the implementation of the interventions with these impact ICT hubs. 45 ABOUT THE AUTHORS Catherine Ragasa is a Senior Research Fellow in the Development Strategy and Governance Division of the International Food Policy Research Institute (IFPRI), Washington, DC. Lucia Carrillo is a Research Analyst in the Development Strategy and Governance Division of the In- ternational Food Policy Research Institute (IFPRI), Washington, DC. Kelvin Balakasi is a Data Manager at Wadonda Consult, Zomba, Malawi. REFERENCES Alderman, H., J. Hoddinott, and B. Kinsey. 2006. “Long Term Consequences of Early Childhood Malnutrition.” Oxford Economic Papers 58 (3): 450–474. https://doi.org/10.1093/oep/gpl008 Anderson, M. L. 2008. “Multiple Inference and Gender Differences in the Effects of Early Intervention: A Reevaluation of the Abecedarian, Perry Preschool, and Early Training Projects.” Journal of the American Statistical Asso- ciation 103 (484): 1481–1495. https://doi.org/10.1198/016214508000000841 Ballard, Terri; Coates, Jennifer; Swindale, Anne; and Deitchler, Megan. Household Hunger Scale: Indicator Definition and Measurement Guide. Washington, DC: Food and Nutrition Technical Assistance II Project, FHI 360. Barrett, C. B., and M. A. Constas. 2014. “Toward a Theory of Resilience for International Development Applications.” Proceedings of the Na- tional Academy of Sciences 111 (40): 14625–14630. https://doi.org/10.1073/pnas.1320880111 Benson, T., A. Erman, and B. Baulch. 2019. Change and rigidity in youth employment patterns in Malawi. In Youth and jobs in rural Africa: Beyond stylized facts. Mueller, Valerie; and Thurlow, James (Eds.). Chapter 6, Pp. 137-171. New York, NY: International Food Policy Research Institute (IFPRI) and Oxford University Press. Carter, M. R., and C. B. Barrett. 2006. “The Economics of Poverty Traps and Persistent Poverty: An Asset-based Approach.” Journal of Devel- opment Studies 42 (2): 178–199. https://doi.org/10.1080/00220380500405261 Carter, M. R., P. D. Little, T. Mogues, and W. Negatu. 2007. “Poverty Traps and Natural Disasters in Ethiopia and Honduras.” World Develop- ment 35 (5): 835–856. https://doi.org/10.1016/j.worlddev.2006.09.010. Coates, Jennifer, Anne Swindale and Paula Bilinsky. 2007. Household Food Insecurity Access Scale (HFIAS) for Measurement of Household Food Access: Indicator Guide (v. 3). Washington, D.C.:FHI 360/FANTA. http://www.fantaproject.org/sites/default/files/re- sources/HFIAS_ENG_v3_Aug07.pdf Deitchler, Megan, Terri Ballard, Anne Swindale and Jennifer Coates. Validation of a Measure of Household Hunger for Cross-Cultural Use. Washington, DC: Food and Nutrition Technical Assistance II Project (FANTA-2), FHI 360, 2010.Dorward, A., and E. Chirwa. 2011. The Malawi agricultural input subsidy program: 2005–6 to 2008–9. International Journal of Agricultural Sustainability, 9(1), 232–247. Dercon, S., J. Hoddinott, and T. Woldehanna. 2005. “Shocks and Consumption in 15 Ethiopian Villages, 1999–2004.” Journal of African Econ- omies 14 (4): 559–585. https://doi.org/10.1093/jae/eji022 Ellis, F., and E. Manda. 2012. “Seasonal Food Crises and Policy Responses: A Narrative Account of Three Food Security Crises in Malawi.” World Development 40 (7): 1407–1417. https://doi.org/10.1016/j.worlddev.2012.03.005 Hoddinott, J. 2006. “Shocks and Their Consequences Across and Within Households in Rural Zimbabwe.” Journal of Development Studies 42 (2): 301–321. https://doi.org/10.1080/00220380500405501 Holden, S. T., and R. Lunduka, R. 2010. Too poor to be efficient? Impacts of the targeted fertilizer subsidy program in Malawi on farm plot level input use, crop choice and land productivity. Noragric Report 55, Department of Economics and Resource Management, Norwegian University of Life Sciences. Kahn, M. E. 2005. “The Death Toll from Natural Disasters: The Role of Income, Geography, and Institutions.” Review of Economics and Statis- tics 87 (2): 271–284. https://doi.org/10.1162/0034653053970339 Kennedy, G., Terri Ballard and Marie Claude Dop. 2011 Guidelines for Measuring Household and Individual Dietary Diversity. FAO, Rome. Knippenberg E., N. Jensen, and M. Constas. 2018. Resilience, Shocks, and the Dynamics of Food Insecurity: Evidence from Malawi. Working Paper. Ithaca, NY: Cornell University. Maxwell, D., B. Watkins, R. Wheeler, G. Collins. 2003. “The Coping Strategies Index: A tool for rapidly measuring food security and the impact of food aid programs in emergencies.” Paper presented at the FAO International Workshop on “Food Security in Complex Emergencies: building policy frameworks to add/ ress longer-term programming challenges.” Tivoli, September 23–25. Ragasa, C., and J. Mazunda. 2018. The Impact of Agricultural Extension Services in the Context of a Heavily Subsidized Input System: The Case of Malawi, World Development 105: 25-47. Ricker-Gilbert, J., and T. S. Jayne. 2011. What are the enduring effects of fertiliser subsidy programs on recipient farm households? Evidence from Malawi. Department of Agricultural, Food and Resource Economics Staff Paper 2011-09. Michigan St