There are two ways to feed more Africans. Farmers can clear more land, or they can grow more on the land they have. Wildlife pays for the first, and farm finance can pay for the second. This article asks what the evidence says about each, what the World Bank Group and the International Fund for Agricultural Development promised at the October 2026 Annual Meetings in Bangkok, and what a lender should learn from a microfinance crisis in Cambodia before lending at scale. It follows our earlier pieces on credit, solar tools, markets, cocoa, horticulture and data.
Read this first. The yield and habitat evidence is peer reviewed but not new: the habitat study appeared in 2021 and Our World in Data’s Africa yields article in 2022, though your newsletter featured it again. They are projections to 2050, not measurements. The World Bank plenary figures come from a French language page of the speech, and programme results from the IFAD report are the programmes’ own claims. The Cambodia case is outside Africa and still contested. We read the NextBillion summary of it, plus Human Rights Watch’s account. Research stops at 18 November 2026.
Six findings
- Yields decide how much land farming takes. By 2050, business as usual needs 26% more cropland, and Sub Saharan Africa’s average habitat loss would exceed 12%, twice the global 6%. Closing yield gaps would cut it to about 1%.
- Africa has grown output mostly by adding land. From 1980 to 2018, Sub Saharan cereal production roughly tripled while cereal land rose from 48 to 112 million hectares.
- Wildlife is under pressure. The 2026 Living Planet Report puts the average fall in monitored African wildlife at 80% since 1970, against 73% worldwide, though the index is not a count of lost animals.
- The promise is large. AgriConnect aims to move 300 million smallholders from subsistence to surplus by 2030, and co financing between IFAD and the Bank’s concessional arm rose from $646 million in 2021 to $1.56 billion in 2025.
- Credit can hurt. In Cambodia, the ombudsman of the International Finance Corporation found the institution failed to protect poor and vulnerable borrowers, a finding its board rejected.
- Redress is weak. Only 4 of 15 digital public systems studied in five countries, including Benin, Ethiopia and South Africa, met every test of a working complaints channel.
Yields decide how much land farming takes
Our World in Data’s Hannah Ritchie summarises a 2021 study in Nature Sustainability by Williams and colleagues. By 2050 the world needs 26% more cropland, about 3.4 million square kilometres, if nothing changes. Of 19,859 land vertebrates assessed, 88% lose some habitat, 6% see no change and 6% gain, mostly farmland birds. The global mean loss is 6%, but in Sub Saharan Africa it is over 12%. Some 1,280 species lose at least a quarter of their habitat, and 93% of them live in Sub Saharan Africa or Latin America. Closing yield gaps to 80% of each place’s potential cuts Africa’s mean loss to about 1%.
The earlier Africa yields article shows how the gap arose. From 1980 to 2018, South Asia’s cereal production rose 133% with no change in land use, while Sub Saharan production roughly tripled as land went from 48 to 112 million hectares. Sub Saharan yields rose only about 30% and sit at half of India’s and a fifth of the United States’. More than half the labour force works in agriculture, and 44% of people live below $3 a day. Our reading is that yield is the lever a farm lender can pull, through seed, fertiliser, water and storage.
Williams et al. (2021) via Our World in Data. Sub Saharan Africa is “over 12%”, plotted as 12
Our World in Data, 2022. Land under cereals in Sub Saharan Africa; 2.3 times is our arithmetic
Williams et al. (2021) via Our World in Data. Share of 19,859 land vertebrates, business as usual
What the wildlife data say and do not say
The 2026 Living Planet Report, from the World Wide Fund for Nature and the Zoological Society of London, reports a 73% average fall in monitored wildlife populations from 1970 to 2022. Our World in Data warns that this is not 73% of animals lost or 73% of species in decline: it averages changes across nearly 36,000 populations, and about half of those studied have fallen while about half have risen or held steady. For Africa the report says monitored populations fell 80% on average, second to Latin America and the Caribbean at 95%, as reported by RFI on 8 October 2026. It cites farm expansion, logging, mining and roads in the Congo Basin, and notes that Africa’s steeper fall partly reflects richer wildlife when monitoring began, and that some southern African elephant populations are recovering. It gives no share of the fall caused by farming.
Living Planet Report 2026 via RFI and Our World in Data. An index of monitored populations, not lost animals
The money on the table
In his plenary of 16 October, according to the French language page of the speech, World Bank Group President Ajay Banga said the Group wants to help 300 million smallholders move from subsistence to surplus by 2030, with 200 million as its own target and partners making up the rest. AgriConnect has set up activities to support 82 million farmers in its first full year and is preparing the next 118 million. He said agro industry is one of five sectors that received about 55% of financing and mobilised private capital, and that the Lobito corridor, built for minerals, increasingly carries farmers’ harvests. Under Mission 300 with the African Development Bank, 59 million Africans have gained electricity access: 12 million in 2024, 18 million in 2025 and 29 million last year. He also said 1.2 billion young people will enter the labour market in 10 to 15 years, against about 400 million jobs, a gap of about 800 million.
World Bank Group plenary remarks, 16 October 2026. Plans and set ups, not yields
World Bank Group plenary remarks. Fiscal years; the total is our check (12 plus 18 plus 29)
The International Fund for Agricultural Development, reported on 14 October 2026, signed a new co financing framework with the World Bank at the Annual Meetings. It runs five years and renews once. Co financing with the Bank’s International Development Association rose from about $646 million in 2021 to $1.56 billion in 2025, about 2.4 times (our arithmetic). The only African example given is Lesotho, where a $70.5 million programme reports nearly 1,500 jobs, nutrition education for 217,000 people, over 106,000 people adopting climate smart practices and weather services reaching more than 900,000. Those are the programme’s own figures, with no yield or income result. We found no African yield outcomes for AgriConnect itself.
What Cambodia teaches a farm lender
Cambodia is not Africa, but its microfinance crisis is a warning about credit pushed faster than borrowers can bear. NextBillion’s summary of consultant Daniel Rozas’s argument says the International Finance Corporation invested in the country’s largest and fastest growing lenders despite warnings that the sector was overheating, and that its board dismissed its own ombudsman’s findings. Human Rights Watch reports that the ombudsman published a 142 page investigation on 24 June 2026, finding the Corporation had not met its sustainability obligations to protect poor and vulnerable people, had not ensured lenders had adequate grievance mechanisms, and had accepted land in Indigenous territories as collateral without community consent. The Corporation’s board rejected the finding of policy noncompliance and approved help for complainants to use existing complaint procedures, with no compensation, debt relief or land restitution. The Corporation has invested more than $400 million in Cambodian microfinance in 10 years, and the complaint, filed in 2022, covered 18 borrowers. Human Rights Watch is an advocacy group and the Corporation disputes the finding, so we treat this as a contested record, not a verdict.
Human Rights Watch, 26 June 2026
The lesson for African farm credit is about design, not geography. Loans should rest on cash flow and the harvest calendar, not on a family’s land or home. A borrower needs a complaints channel that works. Here the evidence is thin: researchers for a Gates Foundation project, writing in NextBillion on 12 October 2026, studied 15 digital public systems for payments, identity and data exchange in Brazil, India, South Africa, Benin and Ethiopia. Only 4 met every test of a live grievance mechanism, 8 were substantially underway and the status of the remaining 3 is not stated (our arithmetic). They found standardisation ends when something goes wrong, and recommend shared categories, a common case record and public scorecards. Farm credit that runs on these systems will inherit their gaps.
Porteous et al., NextBillion, 12 October 2026. The 3 is our arithmetic (15 less 4 and 8)
Our proposal
What to do next
- Make yield the lending target: finance seed, fertiliser, water and storage that raise output on the land farmers already hold.
- Publish AgriConnect’s results as yields and incomes per farmer, not only as farmers reached.
- Do not lean on land as collateral for small farm loans, especially where tenure is communal.
- Size repayments to the harvest and to cash flow, and watch arrears as an early warning.
- Require every farm lender to run a complaints channel that works, tested by mystery shopping.
- Use the Lobito corridor and Mission 300 power to open storage and markets, so higher yields can be sold.
- Track habitat and land use around financed farms, so the case for yields is measured and not assumed.
What we can and cannot claim
| Statement | Evidence status |
|---|---|
| Closing yield gaps would cut African habitat loss to about 1% | Peer reviewed model, 2021; a projection to 2050 |
| African cereal production tripled mainly by adding land | Our World in Data, 1980 to 2018 |
| African monitored wildlife fell 80% since 1970 | Living Planet Report via RFI; an index, not a count |
| Farming caused that fall in Africa | Not shown; no share given |
| AgriConnect supports 82 million farmers | Activities set up, per the speech; no outcome data |
| Co financing grew from $646 million to $1.56 billion | IFAD report, 2021 to 2025 |
| Lesotho programme raised farmers’ incomes | Not shown; jobs and reach only |
| IFC harmed Cambodian borrowers | Contested; ombudsman and advocacy findings, board rejects |
| Only 4 of 15 digital systems have a working complaints channel | Research summary on one site; five countries |
The choice between more land and more yield is already being made, one loan at a time. The evidence says Africa grew its harvests mostly by clearing more, that closing yield gaps would almost end habitat loss in the region by 2050, and that the money to try is on the table. What is unproven is whether that money will raise yields on real farms, and whether the borrowers will be protected if it does not. Lend against cash flow and the harvest, keep land out of small loans, give every borrower a complaints channel that works and count the yield, and credit can feed more people without taking more habitat.
Sources & notes
SOURCES: Our World in Data, To protect the world’s wildlife, we must improve crop yields, especially across Africa (Ritchie; Williams et al., Nature Sustainability, 2021); Our World in Data, Increasing agricultural productivity across Sub Saharan Africa is one of the most important problems this century (Ritchie, April 2022); RFI via allAfrica, Africa’s wildlife population plunges by 80% in five decades, WWF warns (8 October 2026); Devdiscourse, IFAD and World Bank forge new financing pact (14 October 2026); World Bank Group, Remarks by President Ajay Banga at the 2026 Annual Meetings plenary (16 October 2026, French page); Human Rights Watch, Cambodia: World Bank’s IFC rejects microfinance harm findings (26 June 2026); NextBillion, Digital public infrastructure is supposed to be interoperable (Porteous et al., 12 October 2026); NextBillion newsletter: Is the World Bank to blame for the Cambodian microfinance crisis? (Rozas); Our World in Data, The OWID Brief (October 2026); World Bank Civil Society, President Banga’s plenary address and Missed Bangkok? newsletters; World Bank Civil Society, CSPF schedule for 16 October 2026.
Differences between sources: the speech page and your newsletter give the jobs gap as 1.2 billion young people against about 400 million jobs, while the news report in our data article gave 420 million, so we use 400 million here; the speech counts 59 million people with power access under Mission 300, consistent with the 50 million plus we cited in our first article; the 73% decline is an average across populations, not a share of animals; the habitat study and the Africa yields article are older than the newsletter that re featured them; the 300 million farmer ambition includes partners and the Bank’s own target is 200 million; Human Rights Watch and the ombudsman say the Corporation breached its policy, while its board says there was no noncompliance; the ombudsman’s report date follows Human Rights Watch, though other file names suggest earlier stages; we read only the NextBillion summary of the Cambodia piece.
Our own arithmetic: 2.3 times more cereal land (112 divided by 48); 100 million partners’ share (300 less 200); 82 plus 118 equals 200; 59 million (12 plus 18 plus 29); 2.4 times co financing (1.56 billion over 646 million); the 3 unstated systems (15 less 4 and 8) and the pie shares of 26.7%, 53.3% and 20.0%; and the 12% of species with no change or gain (6% plus 6%).