Close
Impact

Solar water and trusted advice raise farm incomes on cleared land

(18 min read)

Three newsletters from March 2026 describe one problem from three sides. WRI Africa shows farmers in Ethiopia’s Berbere district turning idle river water into several harvests a year with solar pumps. The Partnership for Economic Inclusion at the World Bank explains how coaching, now helped by AI, can reach more households without losing the trust that makes it work. And the Global Forest Watch review of the Congo Basin shows what happens to forests when small farms and cooking fuel have no better option. This article asks a practical question for cooperatives, agribusinesses and governments: what raises farm income on land that is already cleared, and how will anyone prove that it did?

Read this first. We read WRI’s Berbere report, WRI’s October 2026 diesel analysis, the Global Forest Watch Congo Basin review, the PEI paper on e coaching, and most of the World Bank paper on electrifying Africa. For PEI’s Graduation landscape and the World Bank paper on generative AI we relied on published summaries, so their figures are as reported. The Berbere results come from WRI’s farmer interviews in one pilot of 48 households, not from a controlled trial. Research stops at 15 November 2026.

From a pump to proof

The links this article follows. Figures are from the sources named in each section

Energy$19,200 for twosolar pumps inBerbereWaterThree to fourharvests a year,not oneCrops$709 from onionsin one seasonon averageAdviceAbout 50 to 90households percoach with AI helpProofPlot records forbanks, buyersand funders

Six findings

  • Energy, not water, kept Berbere’s farmers poor. Four rivers flow all year, yet petrol for irrigating one hectare cost $91 to $203 a year and many farmers went back to subsistence.
  • Crop choice moved incomes. Farmers reported averages of $709 from onions in one season, against about $207 a year from maize and teff sales before the pumps.
  • Diesel is a growing risk. African retail diesel rose 25% on average across 42 countries between mid January and mid June 2026, and 86% in Nigeria.
  • Power pays when it is used. World Bank economists find that electricity alone does not guarantee productive use, and that farmers value power for pumping water.
  • Forest loss in the Congo Basin is mostly about food and fuel. The Basin lost 780,000 hectares of primary forest in 2024, mainly to small scale farming, with charcoal and fire adding pressure.
  • Coaching scales when AI supports people. In some programmes one coach now supports closer to 90 households instead of about 50, and adoption depends on a trusted person introducing the tool.

Water was never the problem

Berbere sits at the foot of the Bale Mountains in Oromia and is surrounded by four rivers that flow all year. Farmers there grew mainly maize and teff, and the average household earned about $207 a year from crop sales, according to WRI’s interviews. A local NGO had already given 13 farmer groups petrol pumps, training and starter seeds. That worked until fuel became the constraint: petrol was costly to bring from nearby towns, supply was patchy and repairs were frequent, so many farmers returned to subsistence.

In 2024 WRI and DanChurchAid studied the district. With IKEA Foundation funding they then installed two 2.75 kilowatt solar pumps in Gebe Keku and Gora Hido for $19,200. The pumps irrigate 14 hectares and serve 48 households. Farmers can now harvest three to four times a year instead of once, and grow onion, tomato, chili and cabbage alongside staples. In the first season, April to June 2025, they produced over 5,100 kg of onions, 8,000 kg of peppers and 3,300 kg of cabbage.

What a pump changed on Berbere farms

US dollars. Crop bars are average earnings reported by farmers per production season (April to June 2025). The maize and teff bar is the average household’s yearly crop sales before the pumps. WRI interviews, one pilot

$0$200$400$600$800Maize and teff, per year before: $207$207Maize and teff,per year beforeOnions, per season: $709$709Onions,per seasonPeppers, per season: $274$274Peppers,per seasonCabbage, per season: $77$77Cabbage,per season$207 a year before the pumps

WRI describes these season earnings as 4 to 10 times what many farmers made in a year from teff and maize. On our arithmetic, the onion average alone is about 3.4 times the household’s earlier yearly crop income. Farmers put the money into cooking oil, flour, soap, seed, livestock and school supplies. These are first season reports from one pilot, and the figures are promising rather than proven.

Cost and lifespanSolar pumpPetrol pump
Upfront costUS$19,200US$2,700
Lifespan20 years10 years
Maintenance a yearUnder US$10Over US$1,300
FuelNoneUS$91 to US$203 per hectare a year
Lifecycle costLowestRoughly 3 times solar

Solar costs more on day one, but WRI estimates petrol pumps cost roughly three times as much over their life, with a payback of six to eight years for solar. On our arithmetic, the $19,200 works out at about $400 a household, and running the same 14 hectares on petrol would cost $1,274 to $2,842 a year in fuel alone. Most farmers cannot pay upfront, so WRI’s convening of government, donors, banks and suppliers produced a cooperative model: farmers deal directly with financial institutions, negotiate with suppliers and receive after sale service, using local bank loans, pay as you go plans and donor subsidies. Nearly 78% of farmers interviewed in neighbouring villages want solar, and WRI’s models show 30 to 142 hectares per village and 50 to 260 farmers per site with more funding.

Why solar is no longer optional

Diesel runs pumps, mills, cold rooms and the trucks that carry produce, so every price shock reaches the farm gate. WRI’s October 2026 analysis, drawing on IFPRI, shows African retail diesel up 25% on average across 42 countries between mid January and mid June 2026, with increases of 86% in Nigeria and more than 50% in Tanzania and Ethiopia. Fertiliser rose 30% to 45% in several African markets. A women’s groundnut cooperative in Nigeria reported that raw groundnut and oil prices rose 33%.

Where diesel rose most in 2026

Per cent. Country and Africa bars: retail diesel, mid January to June 2026 (IFPRI, as shown by WRI). Grey bar: global average, second quarter 2026 against a year earlier, a different basis. “50%+” means more than 50%

0%25%50%75%100%NigeriaNigeria: 86%86%TanzaniaTanzania: 50%+50%+EthiopiaEthiopia: 50%+50%+Africa average, 42 countriesAfrica average, 42 countries: 25%25%World average, year on yearWorld average, year on year: 32.9%32.9%

Behind the pump sits a shadow power system. An Energy for Growth Hub study, cited by WRI, estimated 150 to 190 GW of distributed diesel generator capacity running across Africa in 2024, against 268 GW of grid capacity. Only 33% of rural people in sub Saharan Africa have electricity. The shift is already under way: WRI reports that 17 GW of solar will be installed in Africa in 2026, up 45% from 2025, with about three quarters of the 2023 to 2025 growth coming from distributed systems. In Chamwino, Tanzania, farmers reported that mango production and income roughly doubled within a season after a solar pump and drip system replaced diesel.

Africa’s shadow power system, 2024

Gigawatts. Energy for Growth Hub estimate as cited by WRI. Diesel generator capacity is a range, shown as a floating column

0 GW100 GW200 GW300 GWGrid-connected capacity: 268 GW268 GWGrid-connectedcapacityDistributed diesel generators: 150 to 190 GW150 to 190 GWDistributed dieselgenerators

The headroom is large because almost nothing is irrigated. The World Bank estimates that 95% of sub Saharan Africa’s 183 million hectares of cultivated land is rain fed. Other sources put irrigated cropland at 5% or 6%, which is the same story at a different decimal.

Africa’s cropland waits for water

Share of sub Saharan Africa’s 183 million hectares of cultivated land. World Bank estimate as reported by Mongabay; the water managed share is stated as “less than 5%”

Rain-fed: 95%Any water management: under 5%95%rain-fedRain-fed: 95%Any water management: under 5%

Studies of how much could take solar pumps differ in method but agree on scale. A 2023 study in Environmental Research Letters found that over a third of the unmet crop water needs of 19 major crops on African smallholder land could be met with standalone solar irrigation, and identified 10 million hectares with profit potential above $100 a hectare a year. A 2024 crop by crop analysis found about 9.34 million hectares of rain fed cropland both hydrologically and economically feasible. In Ethiopia, researchers found solar irrigation suited to 9% of irrigated land and 18% of rain fed land. WRI’s own analysis finds about 12.6 million Ethiopians living where farming activity and solar potential are high but grid access is poor.

How much land could take solar pumps

Million hectares. Three studies with different scopes, crops and cost assumptions, so read them as a spread, not a total. The Mali figure covers one country

0 Mha3 Mha6 Mha9 Mha12 MhaFeasible rainfed cropland (Wamalwa 2024)Feasible rainfed cropland (Wamalwa 2024): 9.349.34Profit above $100 a hectare (Falchetta 2023)Profit above $100 a hectare (Falchetta 2023): 1010Suited to solar, Mali only (ILSSI)Suited to solar, Mali only (ILSSI): 4.44.4

Power pays only when it is used

A World Bank working paper from March 2026 argues that electrifying Africa is a twofold task: help poor people use power to raise their incomes, and build trust in tariffs and taxes so the infrastructure can be paid for. An estimated 43% of people in sub Saharan Africa lack electricity. The authors read the evidence as mixed. Electricity raises incomes in some places, but small villages often need complements such as roads, irrigation, credit and markets before activity takes off.

Two findings speak straight to farms. Take up of grid power is higher in villages that already have productive uses such as grain mills, and farmers say they would invest in power to pump water rather than to light their homes. And in Ethiopia, villages that received electricity saw substantial gains in agricultural productivity and non farm business compared with those that did not. The paper also reports that 57% of unelectrified Africans live within an hour’s travel of an urban agglomeration. Our reading is that the market towns where produce is gathered and processed are natural anchors for pumps, mills and cold rooms. The paper’s main proposal, large grants to local governments for last mile connections and the complements that make power productive, is an idea to be tested, not a proven fix.

The forest next door

Raising income on cleared land matters most where the alternative is clearing forest. The Congo Basin is the clearest case. Global Forest Watch reports that it removed an average of 160 million tonnes of carbon dioxide equivalent a year between 2015 and 2024 and supports about 100 million people. In 2024 it lost 780,000 hectares of primary forest, 0.44% of its primary forest against 0.38% in 2023, a 14.2% rise. The DRC accounted for over 75% of the loss.

The forest that holds the carbon

Primary forest as a share of land area in Congo Basin countries, 2024. Global Forest Watch review published by WRI Africa. Cameroon and the Central African Republic are Basin countries, but the review text gives no share for them

TanzaniaW. SaharaDem. Rep. Congo: 42% primary forestSomaliaKenyaSudanChadSouth AfricaLesothoZimbabweBotswanaNamibiaSenegalMaliMauritaniaBeninNigerNigeriaCameroonTogoGhanaCôte d'IvoireGuineaGuinea-BissauLiberiaSierra LeoneBurkina FasoCentral African Rep.Congo: 60% primary forestGabon: 85% primary forestEq. Guinea: 80% primary forestZambiaMalawiMozambiqueeSwatiniAngolaBurundiMadagascarGambiaTunisiaAlgeriaEritreaMoroccoEgyptLibyaEthiopiaDjiboutiSomalilandUgandaRwandaS. SudanGabon85% primary forestEquatorial Guinea80% primary forestRepublic of Congo60% primary forestDemocratic Republic of the Congo42% primary forestCameroonShare not given in the sourceCentral African RepublicShare not given in the sourcePrimary forest, share of land area40%85%

Small scale shifting agriculture remains the dominant driver across most of the region. Permanent agriculture is rising: it caused 40% of Cameroon’s tree cover loss from 2001 to 2024, against 15% in the DRC, and DRC cocoa output more than quadrupled between 2015 and 2023. Fuel matters too, because about 95% of the DRC’s people rely on biomass for energy. Around Kisangani, demand for food and energy from the city drives most of the clearing, largely for food crops. In 2024 Gabon lost 0.07% of its primary forest and the DRC 0.56%.

Protected areas lose trees most slowly, and community forests, which hold the best chance of linking tenure to livelihoods, are losing more. DRC community forest concessions grew from under 1 million to more than 4.5 million hectares in a decade, yet loss inside them has risen steadily, because investment in sustainable practices has not kept pace with secured rights. WRI’s April 2026 update found that non fire primary forest loss in the DRC hit a record in 2025, linked to small scale farming, firewood and charcoal, conflict related displacement and mining.

Where DRC’s trees are lost fastest

Tree cover loss, per cent a year, by land management category, DRC 2015 to 2024. Global Forest Watch review, all tree cover, not only primary forest

0.0%0.25%0.5%0.75%1.0%Protected areas: 0.26%0.26%ProtectedareasCommunity forest concessions: 0.61%0.61%Community forestconcessionsLogging concessions: 0.76%0.76%Loggingconcessions

Higher farm income does not by itself stop clearing. Whether it does depends on land rights and on whether forests are protected, and the sources we read do not test it. What they do show is that the pressure comes from food and fuel on small plots, which is where water, yields and cooking energy are the levers.

The human half of the harvest

Pumps do not run farms; people do. PEI’s 2025 survey of 173 Graduation programmes, which combine coaching, productive assets, consumption support, skills training and links to savings groups, found sub Saharan Africa the most active region, at 65%. About 2.3 million households were enrolled as of January 2025 and nearly 6.7 million had been reached to date. PEI warns that the survey may not have reached every organisation and that recent funding cuts may have kept some from taking part.

Who runs Graduation programmes

Share of 173 programmes in PEI’s 2025 survey. The remainder is our arithmetic and includes government led and other lead agencies

NGO-led: 78%Other lead agencies (our arithmetic): 22%173programmesNGO-led: 78%Other lead agencies (our arithmetic): 22%

The evidence for the approach is strong in Africa. Randomised trials in six countries, including Ethiopia and Ghana, found higher consumption and assets in every country except Honduras. In Tigray, Ethiopia, households that received the programme held 68% more in assets a year after it ended. Honduras carries an agricultural lesson: most households chose chickens as their asset, many chickens died of illness, and assets fell. The choice of livelihood matters as much as the transfer. IFPRI’s 2025 review of Ethiopia’s safety net adds that intensive NGO models raised assets and incomes but rarely let households exit the programme, while government led efforts struggled with overstretched systems, weak credit and limited market links. Those are the gaps a cooperative with finance and buyers can fill.

Coaching is the costly part, which is why PEI’s March paper on e coaching matters. Fundación Capital traces fifteen years from tablets to chatbots to AI assistants and concludes that technology works when it reinforces human coaches. In some programmes coach caseloads rose from about 50 households to closer to 90. Its AI tools answer from a curated, verified knowledge base, accept voice messages for people with low literacy, and are adopted more when a coach or trusted local figure introduces them. The authors also note that AI needs stable connectivity and digital comfort, so no single channel suits everyone.

More households per coach

Households supported by one coach in some programmes after e coaching tools were added. Fundación Capital, as described in the PEI paper; not uniform across programmes

0255075100Before e-coaching, about: About 50About 50Before e-coaching,aboutWith e-coaching, closer to: Closer to 90Closer to 90With e-coaching,closer to

The same pattern shows up in farm advice. Digital Green’s Farmer.Chat, used in Kenya, Ethiopia, Nigeria and India, engaged over 15,000 farmers and handled more than 300,000 queries in an early study, and Digital Green’s early estimate puts advice at $0.35 a farmer against $35 for conventional extension. A study of the tool found farmers trusted it more when familiar extension officers introduced it. A March 2026 World Bank and ILO working paper adds a caution: in developing economies, workers who could gain from generative AI face digital infrastructure gaps that may stop them capturing the gains. That paper studies jobs, not farms, but the caution carries over: advice by phone works only where phones, data and trust exist.

Making the gains visible

Banks, buyers and funders back what they can see. WRI Africa’s Restoration Launchpad guidebook, flagged in the March digest, sets out how to monitor, report and verify local land restoration at scale, combining practitioner knowledge with high resolution satellite imagery and AI. The same logic fits irrigated plots: a cooperative asking for a pump loan needs records of hectares irrigated, harvests a year and sales. AgriKonnekt is built to check farm plots with sensors and satellites so cooperatives can show results to funders, according to the AgriKonnekt page. We have not seen it used to verify irrigated yields or avoided clearing, so that use is our proposal.

What to do next

  1. Cooperatives should price pumps on lifecycle cost, not upfront price, and organise as groups so they can reach banks, bargain with suppliers and secure after sale service.
  2. Every pump needs a market plan: which crops, which buyers, which weeks. Record hectares irrigated, harvests a year and sales from the first season.
  3. Energy planners should use open geospatial tools to find places where farming, sunshine and weak grid overlap, and start with market towns.
  4. Governments and donors should pair community forest tenure with investment in farming and clean cooking, so that forests earn rather than only cost.
  5. Programme designers should let AI handle routine questions, keep human coaches for decisions, introduce tools through trusted people and offer voice.
  6. Funders should ask for verified results on plots and fund proper evaluations of pilots such as Berbere before scaling on first season averages.

African farms are being asked to feed more people, cool a warming climate and keep forests standing, all on land that is already cleared. The evidence assembled here points to a modest set of levers that work together: reliable water powered by the sun, advice delivered by people who are trusted and helped by technology, finance that reaches groups, and records that prove what happened on the ground. None of them is proven at scale, and the forest link in particular remains a hope rather than a finding. But where a river runs past a field and a farmer has the means to lift it, the pump pays, the market responds and the pressure on forest eases. The countries and cooperatives that treat energy, advice and proof as one system will be the ones that raise incomes without clearing another hectare.

Sources & notes

SOURCES

WRI, Solar Brings Long Awaited Water to Ethiopian Farmers (24 February 2026); WRI, As Diesel Costs Surge for Farmers and Rural Businesses, Solar Offers a Lifeline (30 October 2026); Global Forest Watch and WRI Africa, Why Forest Loss in the Congo Basin Requires Immediate Action (25 February 2026); Mongabay, Tropical forest loss falls in 2025 but world still off track (April 2026); WRI, Tropical Rainforest Loss Drops 36% in 2025 (29 April 2026); World Bank, Policy Research Working Paper 11339, Electrifying Africa’s Economic Transformation (March 2026); World Bank, Policy Research Working Paper 11328, Disruption without Dividend? (March 2026); PEI, Scaling Human Support: From In Person Coaching to AI Enabled E Coaching (March 2026); PEI, Landscape of Graduation Programmes, summary on socialprotection.org (2025); PEI, Landscape of Graduation Programs: Approach and Summary of Key Findings (October 2025); Banerjee and others, A multifaceted program causes lasting progress for the very poor, Science (2015); J-PAL, Graduating the Ultra Poor in Ethiopia; J-PAL, Graduating Low Income Households in Honduras; IFPRI, Graduating from Ethiopia’s Productive Safety Net Programme: What have we learned? (October 2025); Farmer.Chat: Scaling AI Powered Agricultural Services for Smallholder Farmers (arXiv); Devdiscourse, AI in agriculture: Can digital tools support learning, not just yield?; Digital Green, Farmer.Chat; Falchetta and others, Solar irrigation in sub Saharan Africa: economic feasibility and development potential, Environmental Research Letters (2023); IIASA record for the Falchetta study; Wamalwa and others, Solar irrigation potential in Sub Saharan Africa: a crop specific techno economic analysis (2024); Feed the Future Innovation Lab for Small Scale Irrigation, Demand for solar powered irrigation is growing across sub Saharan Africa (2023); Mongabay, Smallholder agriculture blossoming with the use of renewables in Africa (March 2025); IWMI, Revolutionizing smallholder irrigation in Africa; AgriKonnekt; WRI Africa, Africa Digest (March 2026); Partnership for Economic Inclusion newsletter (March 2026); World Bank Policy Research Working Papers list (March 2026)

Differences between sources: the share of African cropland that is irrigated is 5% (ILSSI), 6% (IWMI) and under 5% (World Bank, via Mongabay); estimates of land suited to solar irrigation (4.4, 9.34 and 10 million hectares) differ in scope, crops and cost assumptions; the African diesel rise (25%, January to June 2026) and the global rise (32.9%, second quarter against a year earlier) use different bases; WRI gives the solar pump payback as six to eight years in one passage and just over eight years in another.

Our own arithmetic: onion earnings of $709 divided by the earlier $207 a year is about 3.4; $19,200 divided by 48 households is about $400; 14 hectares at $91 to $203 of petrol a hectare is $1,274 to $2,842 a year; the 22% of programmes that are not NGO led is 100% minus 78%.