Policy Brief on Sustainable Climate-Smart Agriculture for Food and Livelihood Security in Ethiopia (SCALE) Project: Baseline Evidence and Policy Directions

Policy Brief

Sustainable Climate-Smart Agriculture for Food and Livelihood Security in Ethiopia (SCALE) Project: Baseline Evidence and Policy Directions

Kebede,S.G., Amare G. H., Kahsay, G.A., Kahsay H. B.; Smith-Hall,C., Zemo, K. H.  

July 2026

1. Background and Rationale

Maize, teff and soybean sit at the core of Ethiopia’s food and agricultural economy. Maize provides roughly one‑third of cereal production and up to one‑fifth of national caloric intake, underpinning food security for low‑income rural households. Teff is the country’s highest valued crop by planted area and a key cash earner, while soybean combines growing market demand with critical agroecological functions through biological nitrogen fixation in nutrient‑depleted highland soils. Yetthese value chains are increasingly stressed by climate variability and systemic post‑harvest inefficiencies. Farmers in Jimma (Oromia) and Wolaita(South Ethiopia) face erratic rainfall, late onset and early cessation of rains, temperature shifts, droughts and associated price volatility. These shocks drive asset depletion and poverty traps, particularly where liquidity constraints and inadequate risk‑management options force distress sales at harvest.

At the same time, traditional storage systems, poor drying and handling practices cause substantial quantitative and qualitative losses, particularly for maize. Quality deterioration – mold, bad smell, insect damage and aflatoxin risk – reduces the value of harvests and limits access to higher‑value markets. Farmer organizations (FOs) and cooperatives could, in principle, mitigate many of these constraints by aggregating supply, improving access to inputs and services, and linking farmers to reliable off‑takers. However, the performance of Ethiopian cooperatives is highly heterogeneous, with governance weaknesses, opportunistic side‑selling and limited capital hindering their potential as resilience multipliers.

Against this backdrop, the SCALE project (2025–2028) – implemented by Agriterra, ETG, PSI, UCPH and Mary’s Meals International – aims to strengthen climate and financial resilience for 10,000 smallholder households in Jimma and Wolaita. The project combines climate‑smart agronomy, cooperative professionalization, improved post‑harvest management (including storage and quality control), biochar innovation, and market linkages anchored in ETG’s sourcing and Mary’s Meals’ nutrition agenda. This policy brief distills insights from the project’s baseline survey to inform policy and program design at national and regional levels.

2. Study Design and Analytical Approach

The baseline covers 2,006 households in 25 farmer organizations, of which 10 are treatment cooperatives and 15 control. Assignment is quasi‑experimental at cooperative level, reflecting operational plans, while an embedded household‑level randomized information experiment provides a pure RCT on climate information effects. A discrete choice experiment (DCE) on storage technologies (hermetic bags, modified Gombisa, metal silos) elicits preferences under different cost–loss combinations.

The survey instrument, built on LSMS‑ISA standards and implemented via Kobo Toolbox, covers:

1. Household demographics, wealth, and livestock (TLU)

2.  Food security (HFIAS)and food consumption patterns (FCS)

3. Shocks, climate change perceptions, and adaptation practices

4.  Production, land, labor, inputs, and crop losses

5. Income, expenditure, and infrastructure access

6.  Behavioral traits (risk, patience, altruism) and trust

7.  Biochar awareness and readiness

8. Cooperative governance, norms, services, and market relationships

Variables are constructed using transparent Stata code (PCA‑based wealth indices, food security indicators, standardized harvests, shock severity, risk indices, etc.), ensuring reproducibility and alignment with a pre‑analysis plan for future impact evaluation.

Baseline balance tests show broadly comparable treatment and control groups in demographics (age, education, household size), with modest but systematic differences in assets, livestock, and income. Income, food security status, biochar familiarity, reported crop losses, and climate adaptation measures also show geographic variation between the two zones that warrant spatial and covariate adjustment in subsequent econometric analyses.

3. Key Baseline Findings

3.1. Climate Resilience and Food Security

High exposure to climate and economic shocks: households in both treatment and control cooperatives report frequent exposure to drought, erratic rainfall, late onset and early cessation of rains, and temperature changes, often with moderate to high severity on a 1–5 scale. Economic shocks – sharp increases in input prices and drops in output prices – further compound vulnerability.

Food insecurity remains widespread, with spatial disparities:

·      58.7%of control and 52.5% of treatment households are food insecure (HFIAS > 0),a statistically significant difference.

·      Mean HFIAS scores are 4.5 (control) vs. 3.5 (treatment), indicating somewhat milder food access constraints in treatment cooperatives but still substantial vulnerability overall.

·      In Jimma, food insecurity levels and HFIA categories are similar across groups; in Wolaita, severity is markedly higher, especially among control households (HFIAS 6.6 vs. 5.1 in treatment).

·      Food consumption score also shows marked difference between the two zones, with 27.2% and 9.8 % of households in Wolaita and Jimma respectively fall under the “poor” food security threshold.  

·      Own production of food remains the main source of food in both zones than market purchases.

Climate change is perceived as real and harmful: most households report perceiving changes in rainfall and temperature over the last decade. For late onset or early cessation of rainy seasons, 85–92% of respondents experiencing such changes report livelihood impacts, with severity ratings around 4/7. Temperature changes are also widely noticed, though somewhat fewer households directly link them to livelihoods. Treatment and control groups hold similar perceptions, implying no major baseline differences in climate awareness.

Adaptation practices are widespread but uneven: farmers already adopt a broad portfolio of 20 adaptation practices, including rainwater harvesting, irrigation, water and soil conservation, agroforestry, grazing management, conservation agriculture, drought‑tolerant and short‑maturing varieties, and crop/livestock diversification. However, adoptionlevels differ significantly by zone and treatment status:

·  Rainwater harvesting: 18.1% Jimma vs. 22.2% Wolaita

·  Irrigation: 21.4% vs. 71.4%

·  Agroforestry: 18.4% vs. 82.4%

·  Grazing management: 26.5%vs. 65.8%

·  Drought‑tolerant varieties: 35.7% vs. 77.1%

·  Short‑maturing varieties: 55.3% vs. 22.2%

·  Crop diversification: 69.5%vs.98.5%

·  Livestock diversification: 51.7% vs. 94.7%

The striking exception is crop insurance use, in which no farmer in both zones has used the scheme. Organic soil enhancement and soil and water conservation practices also show comparable differences between the zones, while use of chemical fertilizer and pesticides/herbicides is similar across the zones.

Overall farmers response to climate stressors is one of modest but they also face key constraints such as lack of information and training, labor shortages and high upfront costs.

Behavioral traits – slightly higher risk tolerance and patience among treatment households: on 0–10 scales, treatment households report significantly higher willingness to take risks in agriculture, rainfall, markets and financial decisions (differences of about 0.25–0.50points). They also exhibit marginally higher patience and altruism. These differences are modest but consistent with their slightly higher asset and livestock holdings.

Trust in local institutions is high; distant institutions lag: households express the greatest trust in neighbors, cooperative leaders and community actors, and lower trust in more distant institutions (higher‑level officials, anonymous traders, media). Treatment households report slightly higher trust in cooperatives and financial institutions, but the overall patterns are similar across groups.

3.2.         Livelihoods, Production and Market Access

Land, plot quality and production patterns: most cultivated plots are owned, titled and certified (approx. 86–88%). Treatment households allocate a larger share of land to annual crops (66.4% vs.50.6%) and grazing, while control households have a relatively higher share in homesteads and perennial crops. Treatment plots are more often sloping, with slightly higher incidence of mild erosion and more plots rated as low fertility. Soil and water conservation structures exist on a minority of plots and are frequently poorly maintained.

Agricultural mechanization is very low across both zones. Nearly 90 percent of farmers continue to rely on traditional cultivation methods as farm power (animal traction and hand).

Crop choice and seasonality: maize dominates Meher production in both groups (around 68–71% of plots), with teff playing a major complementary role. Soybean is present on about 10–11% of Meher plots, but in Belg it is significantly more prevalent on treatment plots (38.0% vs. 21.3% in control). Control households cultivate more cassava and Enset, while treatment households grow more pepper and some pulses, reflecting agroecological differences and market opportunities.

Productivity and returns for focus crops: treatment households obtain higher yields for focus crops

·   Maize (Meher): 747kg/plot vs. 682 kg in control

·   Maize (Belg): 678 kg vs.444 kg

·   Soybean (Belg): 392 kg vs. 229 kg

·  Teff (Meher): slightly higher yields and significantly higher value per plot

These yield differences translate into higher gross margins for treatment farmers, particularly in maize and soybean, though substantial within‑group heterogeneity remains, suggesting important scope for productivity gains.

Labor and input intensity: treatment plots are more labor‑ and input‑intensive,

·  Land preparation labor(Meher): 16.9 vs. 14.1 person‑days per plot

·  Fertilizer application labor (Meher): 4.4 vs. 3.8 person‑days

·  Greater reliance on purchased and exchanged seed in some seasons

·   Slightly higher application of inorganic fertilizers in several cases

These patterns suggest that higher productivity among treatment households is partly driven by higher factor use, not just inherent agro‑ecological advantages.

Income, expenditure and diets: cereal crop sales are the leading source of income for both groups in both zones closely followed by livestock sales. Treatment households in both zones earn higher share of their income from sale of crops than control households:

·  On average nearly 43%vs.27% and 40% vs. 29% of income for treatment vs. control households in Jimma and Wolaita respectively comes from  cereal sales.

·  Commercial crops like coffee and khat in Jimma and Cassava in Wolaita also contribute significantly to overall household income.

·  Income from non farm activities like petty trading is the fourth largest source to households in Wolaita.

Total annual income is modest but slightly higher among treatment households, mirroring the asset and livestock patterns. Non‑food expenditures (health, education, clothing, transport) are similar across groups, and Food Consumption Scores (FCS) point to slightly more diversified diets in treatment households, though both groups remain close to the borderline between acceptable and vulnerable consumption.

Infrastructure and service access: treatment farmers face higher transaction costs. Despite being somewhat better off, treatment households are typically farther from towns, markets and cooperative offices and incur systematically higher transport costs to reach key infrastructure (nearest town, regular market, kebele office, main roads). For example, one‑way public transport from the village center to the nearest town costs ETB 194.5for treatment versus ETB 140.5 for control households.

This combination – slightly higher income but higher transaction costs – implies that improved aggregation, storage and cooperative marketing could generate substantial net welfare gains by lowering per‑unit marketing costs.

Post‑harvest losses and quality degradation are pervasive: maize shows the highest loss incidence, with 72.3% of control and 65.1% of treatment households reporting losses, more pronounced in Jimma than Wolaita. Losses are also notable for teff, soybean, pepper and coffee. Losses occur at both harvest and post‑harvest stages, but storage is clearly the dominant point of highest loss, accounting for over 40% of reported cases in both groups.

Traditional storage (Gombisa, polypropylene or jute bags) dominates; improved storage technologies are rare. A large majority of households have experienced mold, bad smell or insect damage in stored grain, and quality deterioration is frequently linked to lower sale prices per 50 kg bag. Most farmers believe improved protection is very orextremely important, and many express concern about invisible health risks (mold toxins, chemical residues).

3.3.    Cooperatives, Governance and Inclusive Growth

Heterogeneous cooperative structures and capital base: treatment cooperatives are older and larger on average, while control cooperatives often have higher recorded asset values. All charge registration fees, though only about a quarter levy annual membership fee. Governance structures (elected committees, monitoring bodies) are widely present, but their functionality and enforcement capacity vary.

Norms, rules and enforcement gaps: households across groups perceive early selling and defection from cooperative agreements as likely and recurrent. Roughly two‑thirds report that their cooperative has formal rules to address such behavior, but the mix of sanctions (fines, suspension of benefits, expulsion) is uneven and seldom backed by consistent enforcement. Expectations of opportunistic behavior are high, reflecting structural incentives to side‑sell in the presence of liquidity constraints and thin local markets.

Service provision: farmer organizations provide core services – input supply, collective marketing, storage, basic training – but quality ratings are modest (typically 2–3 on a 5‑point scale). Government extension and FOs are the primary sources of information and advice; training is far less common than basic information provision. Quality of advisory services from financial institutions and NGOs, where accessed, is often rated higher, but reach is limited.

Inclusion: membership criteria rarely explicitly target women or youth, though both groups are numerically present in most cooperatives. Leadership positions remain overwhelmingly male‑dominated, with limited youth representation. Barriers include social norms, perceptions of leadership as male domains, and lack of confidence or experience among women and young members.

Marketing relations: most households sell to a variety of buyers (consumers, brokers, traders, processors, cooperatives).Linking farmers to markets is the single most important service of the market operators in both zones.  Trust in main buyers is generally high, but formal contracts are extremely rare. Price negotiations are limited, and many farmers simply accept buyer‑posted prices. Cooperative marketing plays a relatively small role in actual sales volumes at baseline.

3.4.         Biochar Awareness and Readiness

Biochar is a central innovation in SCALE’s agronomic strategy, but the baseline reveals an almost complete knowledge vacuum:

·  Only 1.5–2.5% of households are familiar with biochar

·   Nearly 80 percent of those reported familiar are from Jimma.

·   Actual use is virtually non‑existent

·  Among the few informed households, radio and research institutions are prominent information sources ;radio reaches a higher share of informed treatment households

·   Most farmers perceive trying new soil amendments as risky, and many prefer to adopt only after observing successful use by others.

This implies that biochar interventions will be starting from near zero awareness in a risk‑averseenvironment – placing a premium on demonstrations, farmer‑to‑farmer learningand careful risk communication.

4.  Policy Implications

·   Resilience must betackled on multiple fronts simultaneously: high climate exposure, widespreadfood insecurity, and pervasive post‑harvest losses suggest thatsingle‑dimensional interventions (e.g. only promoting new seed varieties) willhave limited impact unless complemented by improved risk management, storage, andmarket access.

·  Treatment–control differences reflect light advantages, not structural imbalance: treatment households are somewhat wealthier, more risk‑tolerant, and more input‑intensive, yet they face higher marketing costs and cultivate more erosion‑prone land. Policies should therefore view treatment areas not as fundamentally different but as places where there is already some readiness to leverage more advanced technologies and institutional innovations.

·   Post‑harvest storage isthe single most powerful lever for short‑term gains: given the magnitude of storagelosses, the ubiquity of quality problems and strong expressed demand for betterstorage, investments in hermetic storage, improved Gombisa, and cooperativestorage services promise immediate returns in food availability, pricestabilization and health outcomes.

·  Cooperatives are indispensable but require governance upgrading: cooperative structures exist and are trusted locally, yet norms of early selling and side‑selling remain strong, enforcement is weak and leadership is not inclusive. Turning cooperatives into reliable vehicles for inclusive growth will require business‑oriented governance reforms, performance‑based incentives and explicit gender–youth strategies.

·  Biochar and advanced climate‑smart practices need dedicated technology introduction strategies: extremely low awareness and high perceived risk suggest that biochar cannot simply be added to existing extension messages. A more intensive, phased strategy is required, starting with demonstration plots, participatory trials, and possibly risk‑sharing mechanisms (subsidized pilots, input credit).

·  Spatial heterogeneity – especially Jimma vs. Wolaita – must inform targeting: the substantially higher food insecurity in Wolaita, coupled with differences in climate change responses, biochar, cropping systems and market access, implies that zone‑specific packages may be more effective than one‑size‑fits‑all interventions.

5.  Implications for Monitoring and Impact Evaluation

The baseline provides arobust platform for Difference‑in‑Differences and RCT‑based impact evaluation.Going forward, it will be crucial to:

·  Track changes in foodsecurity (HFIAS, FCS), adaptation practices, yield and income indicators, andcooperative performance;

·  Explicitly control forbaseline differences in wealth, TLU, risk preferences and market access ineconometric models;

·  Disaggregate impacts byzone (Jimma vs. Wolaita), gender, youth status and wealth quintiles, to assessdistributional outcomes and inclusiveness.

6.   Concluding Remarks

The SCALE baseline depicts a farming landscape under significant stress but also rich in existing adaptive practices and institutional infrastructure. Households are highly exposed to climatic and economic shocks; food insecurity remains prevalent especially in Wolaita and post‑harvest losses and quality degradation drain value from already modest production. At the same time, farmers are actively experimenting with water, soil and diversification strategies; cooperatives are present and partially trusted; and there is clear latent demand for better storage and more reliable market linkages.

The challenge and opportunity for policymakers and development partners is to connect these pieces: to couple climate‑smart production with robust post‑harvest systems, professionalized cooperatives and inclusive market arrangements that share risks and returns more fairly. If SCALE and aligned public programs succeed in turning cooperatives into effective hubs for technology, information and fair marketing, Ethiopia’s maize, teff and soybean systems can become not only more productive, but also more resilient, equitable and nutrition‑sensitive.

‍