Close
Climate Finance

Adaptation money follows carbon while farmers struggle to prove results

(11 min read)

Between October and December 2025 the newsletters that trade think tanks, impact investors and development banks send out kept returning to one point from different directions: money, market access and policy attention go to what can be measured, trusted and verified. This article reads about twenty of those stories beside outside data from UNEP, the World Bank, Kenya’s statistics office and Our World in Data. It asks what they mean for the work Kilimora does: Digital MRV (measurement, reporting and verification) and technical assistance that link smallholder cooperatives to climate finance and fairer markets.

Four numbers that frame the article

Sources: UNEP Adaptation Gap Report 2025; World Bank Global Findex 2025; Kenya Economic Survey 2026.

12–14×adaptation needs indeveloping countries versuscurrent flows (UNEP)US$26bninternational publicadaptation finance, 2023(down from US$28bn)40%of adults in Sub-SaharanAfrica hold a mobilemoney account (2024)23.2%of Kenya’s GDP comes fromagriculture, which grew onlyabout 3% in 2025

Six findings

  1. Climate money follows carbon. Roughly 15% of global climate finance goes to adaptation, and under 3% reaches local communities, according to Kuza One in NextBillion. UNEP’s 2025 data show the same gap from the other side.
  2. Farmers are exposed and the numbers show it. Kenya’s tea fell 8% and sugarcane deliveries fell 24.7% in 2025 after erratic rains, while maize and potatoes held up.
  3. Digital finance has arrived, but credit has not. Mobile money reaches 40% of adults in the region, yet only 12% borrow formally, against 24% in comparable economies.
  4. Trade rules now move in coalitions. The MC14 conference let the e commerce moratorium lapse, yet groups of willing members kept working.
  5. Supply chains are being redrawn. Tariffs sent carmakers regionalising, and Ethiopia’s imports from China rose to about a third of the total.
  6. Lenders are cutting their own transaction costs. The World Bank and ADB now let one lead lender run a project for both.

The first leak: adaptation money that never lands

Sheena Raikundalia of Kuza One argues that the mitigation versus adaptation framing shapes how funding flows and how success is measured. Mitigation attracts the bulk of funding because it produces carbon outcomes that can be sold, while adaptation’s local benefits are harder to quantify and trade. Her warning is that African landscapes risk becoming carbon farms for the Global North. A companion NextBillion article from Sustaina Harvest makes the point from the farm: many stakeholders still prefer short term fixes and crisis spending to investing in technologies, ecosystems and people that turn climate risk into opportunity.

UNEP’s Adaptation Gap Report 2025, released ahead of COP30, quantifies the shortfall. Developing countries need an estimated US$310 billion a year by 2035 on modelled costs, and US$365 billion on needs stated in national plans. International public adaptation finance was US$26 billion in 2023, and 59% of recent flows were debt, which raises affordability concerns. UNEP sees private finance rising from about US$5 billion to perhaps US$50 billion a year if policy and blended finance de risk it.

The adaptation gap, in billions of US dollars a year

UNEP Adaptation Gap Report 2025. Bar length is proportional to value. The need is the lower, modelled estimate for 2035; flows are 2023 and private figures are current and potential

Estimated need by 2035310+310+Public adaptation flows, 20232626Private flows today55Private potential with policy5050

Kilimora’s Theory of Change starts from the same gap: small farming cooperatives already do work that helps the climate, but funders want evidence they can trust, and producing it costs too much. AgriKonnekt is built for that step. It checks each farm plot with sensors and satellites and trains cooperatives to use the results. The Kuza argument adds a point we would press: practices such as mulching help crops survive drought and also build soil carbon, so verification should record both effects and not force a farmer to pick a label.

Why the stakes are high in Kenya

Kenya’s Economic Survey 2026 shows what a thin adaptation budget costs. Agriculture provided 23.2% of GDP in 2025, up from 22.4%, and crops made up Sh2.8 trillion of the sector’s Sh4.1 trillion. Yet sector growth slowed to about 3% (reported as 2.8% to 3.1% by different outlets) because rainfall was above average in the long rains and below average in the short rains. Smallholders, who rely on rain, bore the swings: droughts also cut the collateral value of farm assets, which restricts credit.

Kenya’s 2025 harvest: winners and losers from erratic rain

Change in output, 2025 against 2024, from Kenya’s Economic Survey 2026 as reported. Beans, coffee and maize are our calculation from the reported volumes

Sugarcane deliveries−24.7%−24.7%Wheat−18.2%−18.2%Beans−11.9%−11.9%Tea−8%−8%Coffee+3.8%+3.8%Maize+2.2%+2.2%Potatoes+13.6%+13.6%Millet+14.3%+14.3%

The pattern matters for verification. Cooperatives that can show which practices held yields steady, and where, are better placed to win adaptation finance, insurance and offtake terms than those that can only report a bad year.

Digital finance: wide, but shallow on credit

The World Bank’s Global Findex 2025 gives the backdrop to two NextBillion pieces. Payal Dalal of the Mastercard Center for Inclusive Growth notes that 79% of adults worldwide now hold an account, against 62% ten years earlier, but that inclusion is still delivered as separate loans, savings and insurance products that customers must stitch into a safety net themselves. In Sub Saharan Africa the share with mobile money accounts rose from 11% in 2014 to 40% in 2024, overtaking bank accounts at 38%. Kenya is the pioneer: 32% of its adults borrowed from mobile money providers.

Mobile money has overtaken bank accounts in Sub Saharan Africa

Share of adults with an account, World Bank Global Findex. Bar length is proportional to the share

Mobile money, 201411%11%Mobile money, 202127%27%Mobile money, 202440%40%Bank-type account, 202438%38%

The gap sits in credit. Only 12% of adults in the region borrowed formally in 2024, and mobile money providers supplied nearly 60% of those loans. For farmers that means short, small loans, not the working capital or equipment finance a cooperative needs. Nathaniel Adeniyi Akande’s warning about AI powered lending applies directly: bias, fraud and weak data governance can exclude the people a scoring model is meant to reach. The Findex adds a caution of its own, since only about half of mobile money owners in the region protect their phones with a password, and women lag men in phone ownership and account access. Julia Kho and Marianne Vidal Marin’s ten year data on first time SME finance vehicles, from the Dutch Good Growth Fund and I&P, point to catalytic capital as the ingredient that makes small funds work. Koheun Lee’s CARE experiment shows that AI tools trained without women’s input can get women’s lives wrong, so any model scoring a cooperative needs inclusive data.

Maintenance beats construction, and so does follow through

Cecilia Gamba of LeFil Consulting reports that billions spent on rural water in Kenya have left much of the country underserved because systems are not maintained, and she describes a pilot that tests a market based approach to maintenance. The lesson carries over to agriculture: sensors, dashboards and solar pumps that nobody services stop delivering evidence. Related NextBillion work on PAYGo solar (‘PAYGo 2.0’) and four last mile distribution models from Hystra says the same, that reaching rural customers is a matter of repeat service and repayment, not a single sale. Kilimora’s training and technical assistance exist so that Digital MRV keeps working after the project launch.

Trade: coalitions are the working unit

The Hinrich Foundation’s coverage reads differently with hindsight. Former US ambassador Maria Pagan argued before MC14 that the e commerce moratorium and the Investment Facilitation for Development Agreement should come first, and that the “single undertaking” no longer works. Bernard Hoekman made the related case for variable geometry. Dawn Shackleford argued in December that most favoured nation treatment is not dead yet but needs rescuing. Dmitry Grozoubinski and Matthew Wilson showed why small states struggle: a junior attaché in Geneva can cost upward of US$200,000 a year, or about 1% of a developing country’s ministry funding.

MC14 in Yaoundé, Cameroon, in March 2026 bore much of this out. Reports said the moratorium on duties on electronic transmissions lapsed after roughly 25 years of extensions, and that the investment facilitation deal, backed by 129 of 166 members, was blocked by India’s opposition, while analysts noted a widely supported plurilateral e commerce agreement. For cooperatives selling into export chains, rules on digital records and investment will come from groups of willing countries, so records need to satisfy the strictest likely buyer.

Supply chains: who sells to Africa, and who buys

Our World in Data’s December Data Insight shows how far African trade has shifted. In the early 1990s over 40% of Ethiopia’s imports came from Europe and very little from China; today China supplies about one third. Trade statistics from WITS and TrendEconomy trace the climb in between.

China’s share of Ethiopia’s imports

2012, 2018 and 2022 from WITS and TrendEconomy trade data; the latest figure and the early-1990s Europe share from Our World in Data (IMF data). The sources use different methods, so treat the comparison as indicative

Europe, early 1990sover 40%over 40%China, 201221%21%China, 201828%28%China, 202229%29%China, latest (OWID)about 33%about 33%

Three other stories show how rules and ownership reshape trade. The hfAI Lab, a new Hinrich product built with generative AI under human supervision, projected that US auto tariffs pushed carmakers towards regionalisation, China as a hedge, closer investor communication and operational adjustments. The Nexperia episode showed a US ownership rule prompting a Dutch intervention and a Chinese export response. The industrial policy series from Julia Tijaja and Arata Kuno adds that governments are returning to subsidies after disruption. For an African exporter the same logic holds: a rule changes upstream and the cost lands downstream. Better records of origin, inputs and emissions claims are the cheapest insurance.

What the development banks just showed

The World Bank and ADB’s Full Mutual Reliance Framework, announced on 4 December 2025 with projects in Fiji and Tonga, lets a country deal with one lead lender for design, preparation, supervision and evaluation. The banks promise faster delivery and lower transaction costs, with about 20 more projects in the pipeline, including agribusiness. For a smallholder cooperative the nearest equivalent is one trusted verification record accepted by several funders, so field data is not rebuilt for each lender. That is the logic behind Kilimora’s Cooperative Edge proposal, which won the second UNCTAD Supply Chain Innovation Challenge, and AgriKonnekt’s Gold Award at the 2026 UN FAO Global AgriInno Challenge. Corporate programmes point the same way: Mars has committed US$20 million to sustainable rice sourcing over 2020 to 2030, and the ClimaFii Alliance from Shell Foundation and Accion funds climate innovation for microenterprises across Sub Saharan Africa and South Asia. Both reward partners who can show what happened on the ground.

The verdict, and what we would ask for

The money in these stories leaks at the same few points: it follows what is easy to count, it favours large and well documented actors, and it falls short of the cost of proving results at small scale. Three signals would change this view: a measurable rise in the adaptation share that reaches communities, published accuracy and bias results for AI scoring tools, and a verification record accepted by more than one funder.

  • Fund adaptation outcomes in cooperatives, and count practices that deliver both adaptation and mitigation once.
  • Move adaptation finance towards grants and concessional terms, as UNEP urges, to avoid a debt trap.
  • Publish accuracy, bias and data governance results for any AI model that scores farmers or plots.
  • Pair mobile money with working capital credit and insurance for cooperatives, not only small loans.
  • Fund maintenance and training so sensors, pumps and dashboards keep producing evidence.
  • Agree a shared verification record so one set of field data serves several funders.
  • Prepare cooperative data for the strictest export and digital trade buyer, not the average.
Sources & notes

Sources: Hinrich Thought Leadership of 2, 9 and 16 December 2025 (Hinrich Foundation research page, where the essays by Hoekman, Tijaja and Kuno, Thompson, Shackleford, Grozoubinski and Wilson, and the hfAI Lab auto study are listed); NextBillion newsletters of December 2025 (guest articles index); Our World in Data Data Insight, December 2025. Maria L. Pagan, Want to rally the WTO for MC14?; Hinrich Foundation, Variable geometry in trade governance; NextBillion, Context instead of carbon, AI risk management in digital finance and The world is losing the fight against malaria; Our World in Data, Ethiopia imports much more from China than it used to; WITS, Ethiopia 2012 and 2018; TrendEconomy, Ethiopia imports 2022; UNEP, Adaptation Gap Report 2025: Running on Empty, and UNEP press summary; Stockholm Environment Institute summary (debt share of adaptation flows); World Bank Global Findex 2025 via TechCentral, FXC Intelligence and Ecofin Agency; The Standard, Unfavourable rains slow down agriculture output, and Financial Fortune on the Economic Survey 2026; FAO/AGNES, Situational analysis of Kenya’s agriculture NDC; Lawfare, The Nexperia crisis; Hogan Lovells, MC14; Bocconi IEP, What the failed WTO meeting revealed; Deccan Chronicle, MC14 ends without deal; World Bank, Full Mutual Reliance Framework event, and ADB, first projects announcement; Kilimora, kilimora.africa. Figures are as reported by the sources; chart bar lengths and the recommendations are our own reading, and the bean, coffee and maize changes were calculated from reported volumes.