Close
Climate Finance

Savings groups and biological inputs need sequenced systems

(17 min read)

Two things stand between an African smallholder and a profitable season: the money to buy what the crop needs, and a safe, legal product to buy with it. They have more in common than they seem. Three pieces you shared, one from CARE on digitising savings groups, one on designing credit for informal economies and one from SIANI on biological crop inputs in Africa and Asia, all make the same point in different words. Order matters, and a system that skips a step fails at scale. This article joins the finance and the inputs, using African numbers to ask how smallholder credit and crop protection can be built so that both hold up.

Read this first. We read all three pieces in full. The CARE blog gives programme results from its own pilots, and the credit piece is one practitioner’s view from a fintech product role, so we treat their figures as reported. For the SIANI piece we added Kenyan and East African reporting on biological inputs. Finance gap estimates differ a lot by source and year, and we show the spread. Research stops at 13 November 2026. In this expanded version we also set the argument against published studies: randomised trials, a Science panel study and on farm evaluations, each flagged by design and strength.

Seven findings

  • Farming is under lent. Agriculture gives about 30% of African GDP and over 60% of jobs but gets roughly 5% to 6% of commercial bank lending.
  • Order matters in digital savings groups. CARE finds that norms and skills must come before digitisation, credit and scale.
  • Records, not just data, unlock credit. Six to twelve months of transaction history fed a bureau score in Uganda, with portfolio at risk below 1%.
  • Credit is a system, not a feed. Data, identity, trust, risk and incentives must improve together.
  • Biologicals are blocked by rules and evidence. Eight East African countries have just agreed common registration rules, but implementation is the test.
  • The wider evidence is encouraging but uneven. Randomised trials back savings groups for access, and bundled credit, insurance and inputs for farm results, yet some trials find no effect.
  • Field proof is the shared gap. Both lenders and regulators need credible evidence of what happens on real farms.

Big in the economy, small in the loan book

The African Development Bank’s president told a Nairobi conference in 2025 that agriculture is 30% of the continent’s GDP but only 6% of commercial bank lending. The Star’s coverage of the April 2026 launch of a financing summit says the sector employs more than 60% of workers and receives under 5% of commercial lending, and that smallholders, women and youth are most excluded. The size of the annual gap is disputed: Aceli Africa estimated $65 billion for agricultural SMEs in the missing middle across Sub Saharan Africa, the Milken Institute cites $75 to $200 billion, and the Star reports over $100 billion.

Big in the economy, small in the loan book

Per cent. GDP share: African Development Bank. Workforce: The Star, 2026 (“more than 60%”). Bank lending: African Development Bank at 6%; other sources say under 5%

0%10%20%30%40%50%60%70%Share of|GDP: 30%30%Share ofGDPShare of|workforce (over): 60%60%Share ofworkforce (over)Share of commercial|bank lending: 6%6%Share of commercialbank lending
How big is the gap? Depends who you ask

Annual financing gap estimates, billions of US dollars. Aceli Africa 2020; Milken Institute; The Star on the FINAS 2026 launch; Africa.com. Definitions and years differ, so read as a spread, not a ranking

Agri-SME “missing middle”, Aceli 2020$65bnAnnual gap, Milken Institute$75 to $200bnAnnual gap, FINAS 2026 launch (over)$100bnAnnual gap, Africa.com feature$200bn

Why does the money stay away? Aceli’s analysis of 9,104 loans from 31 lenders found risk in agri SME lending at least twice that of other sectors, returns 4% to 5% lower in East Africa and high costs of serving small borrowers. A speaker at the 2025 conference compared a 14% yield on Kenyan government bonds with a 3% average return on agricultural SME lending.

Why banks prefer bonds

Per cent. Figures cited at the 2025 smallholder finance conference in Nairobi (AfDB account); the 3% is an average for agricultural SME lending

0%4%8%12%16%Kenyan government|bond yield: 14%14%Kenyan governmentbond yieldAverage return on|agri-SME lending: 3%3%Average return onagri-SME lending

The sequence that cannot be skipped

CARE estimates 500 million people in Africa and Asia belong to savings groups, about 80% of them women, and has run 10 digital pilots across eight countries. Its conclusion is that the steps go in order: norms, skills, digitisation, connection to formal finance, scale. On norms, GSMA’s 2025 data shows women are 36% less likely than men to own a mobile money account, worse than in 2021, and over 60% of Ugandan women CARE surveyed had heard of violence linked to phone use. A radio and community campaign in northern Uganda cut phone restrictions on women by 57%, and 81% of listeners said they took action.

The mobile money gap

Index, men = 100. GSMA 2025 as cited by CARE: women are 36% less likely to own a mobile money account. The index of 64 is our arithmetic

04080120Men|(index): 100100Men(index)Women|(index): 6464Women(index)

Skills follow, spread across all members rather than held by one literate leader. Malawi’s programme runs on USSD through any phone, and Bangladesh’s platform works offline. Only then does digitising records add value, because paper ledgers already work and digital tools earn their place by cutting errors, keeping histories and producing records a lender accepts. In Uganda, Ensibuuko’s Chomoka tracked about $61,000 of internal loans across 500 groups with portfolio at risk below 1%, which on our arithmetic is about $122 a group and shows how small these loans are. Six to twelve months of that data feeds a credit score that lets a lender extend USSD loans without a branch visit or collateral, and FINCA reported portfolio at risk below 1% on loans made this way.

The pathway

Based on CARE’s pilots as described in its FinDev blog. Results are CARE’s own and from specific programmes in Uganda, Malawi and Bangladesh

NormsPhone limits forwomen down 57%(northern Uganda)SkillsWhole group learns;USSD or offlinetoolsDigitise$61,000 trackedin 500 groups;at risk under 1%Connect6 to 12 monthsof records feeda credit scoreScaleLink to markets,buyers and publicprogrammes

Credit is a system, not a data feed

The NextBillion piece argues that more data alone does not fix informal economy lending. Its author, who worked on a credit decisioning product used by lenders in Nigeria and elsewhere in Africa, describes five layers: data, identity, trust, risk and incentives. Identity is fragmented across phones, wallets and informal names. Trust is built through relationships rather than documents. And income swings that a bank reads as risk are normal in farming. The piece says the system must be an adaptive loop in which each layer improves the others.

Five layers that have to improve together

Based on the NextBillion article by Ayokunmi Sodamola (June 2026); the layout is ours

DataIdentityTrustRiskIncentivesAn adaptive loop,not a pipeline

For agriculture the implication is direct. A smallholder’s income depends on rain and harvest timing, so a lender needs to see seasonality and consistency rather than a flat monthly salary. Savings group records show behaviour over time, and plot level records could show production, which is why the CARE and NextBillion arguments point the same way. We have found no evidence yet of farm level data feeding savings group credit scores in Africa, so we treat that as an open opportunity.

The inputs side: rules, evidence and extension

The SIANI summary of discussions across Africa and Asia describes biological products, from biopesticides to biofertility products and biocontrol agents, as tools that work best inside integrated pest management and can cut residues for export standards. Africa’s market is small, registration is slow because it copies frameworks built for chemicals, and field performance is variable across temperature, humidity and timing. Farmers who expect chemical style speed need demonstrations and extension.

Kenya is moving. Its regulator, the Pest Control Products Board, set up a dedicated biopesticides division and in March 2026 consulted industry on registration data, with proposals including extending the renewal period from three to five years. A review in Frontiers in Sustainable Food Systems calls Kenya’s mechanisms comparatively advanced in East Africa but notes high costs, long timelines and low farmer awareness. On 19 October a regional workshop heard that eight countries, Kenya, Uganda, Tanzania, Rwanda, Burundi, Congo, South Sudan and Somalia, have adopted a common framework that lets evidence from one country support registration in others and allows faster emergency authorisation, for example for fall armyworm. The framework’s promoters say it must still be written into national law and funded.

2026 milestones for biological inputs

Dates from the Dutch embassy report on the PCPB consultation, the SEI roadmap, the Independent (Kampala) report on the Cotonou workshop and SIANI

30 MarKenya regulator andindustry review data rules25 AugKenya bio-inputroadmap published19 OctEight-country frameworkpresented in Cotonou11 NovSIANI shares lessonsfrom Africa and Asia

The Aflasafe case shows why regulation matters. The workshop heard that the biological product can cut aflatoxin by 80% to 90% and is registered in six African countries, and that Kenya spent KSh400 million on it in 2019, about a fifth of its fertiliser subsidy by that account. Procurement turned registration into real use.

When registration becomes procurement

Kenya’s 2019 Aflasafe purchase as presented at the October 2026 workshop, reported by the Independent (Kampala). The 80% remainder is our arithmetic

A: 20%F: 80%20%of the subsidyKSh400 million for Aflasafe: about a fifthRest of the fertiliser subsidy (our arithmetic)

What the wider evidence says

The CARE sequence rests on pilots, so it helps to set it beside studies built to test cause and effect. The picture is encouraging but uneven, and that is the point of reading across sources.

Evidence scorecard

Our reading of the studies cited in this section, one row per study. Outcomes are those the authors report; “not measured or not reported here” does not mean no effect

Access to financeFarm orbusiness outcomeWellbeing orempowermentSavings groups, 3-country randomised trialPositive effect reportedPositive effect reportedPositive effect reportedSavings groups, Ghana randomised trialPositive effect reportedNo effect foundNo effect foundMobile money, Kenya household panel––Positive effect reportedInsurance + credit + inputs bundle, Ethiopia trial–Positive effect reported–Input loans + insurance + training, One Acre Fund trial–Positive effect reported–Index insurance bundled with loans, Senegal trial–No effect found–Push-pull, on-farm comparisons (not randomised)––Positive effect reportedNo effect foundPositive, not randomised– Not measured or not reported here

Savings groups. A randomised evaluation by Karlan, Thuysbaert, Udry and colleagues with CARE and 13 other organisations tracked over 12,000 households in 561 communities in Ghana, Uganda and Malawi for two to three years. It found more businesses (up 6%), better financial inclusion and more say for women in household decisions; 68% of members took at least one loan, with a median value of $19.70, and members used funds for farm investments, food and education. The Ghana only evaluation, covering 175 villages, found higher savings and loan access but no evidence of effects on average consumption, business outcomes or women’s empowerment, and its authors suggest savings were too small to matter. Taken together, savings groups reliably build savings and access, while larger welfare effects depend on context and on what comes next.

Who takes part and how far

Share of members or households. Three country figures: Innovations for Poverty Action summary of the PNAS study (2017). Ghana figures: J-PAL evaluation summary. Different samples; read each bar separately

0%20%40%60%80%Took at least|one loan|(3 countries): 68%68%Took at leastone loan(3 countries)Completed a full|cycle (3 countries): 61%61%Completed a fullcycle (3 countries)Completed a|cycle (Ghana): 69%69%Completed acycle (Ghana)Households with a|member (Ghana): 36%36%Households with amember (Ghana)

Mobile money. Suri and Jack’s five round household panel in Kenya (2008 to 2014) estimated that access to M-PESA lifted 194,000 households, 2% of Kenya’s total, out of poverty, driven by saving, financial resilience and a shift by women from farming into business. It is a long run, panel based estimate for Kenya, not a randomised trial, and effects elsewhere may differ.

What a payments system did

Share of Kenyan households lifted out of poverty by mobile money access. Suri and Jack, Science, 2016. The 98% remainder is our arithmetic

M: 2%O: 98%2%of households194,000 households lifted out of poverty: 2%All other Kenyan households (our arithmetic)

Credit, insurance and inputs together. This is where the evidence is most useful for farms and most mixed. A randomised experiment in Ethiopia testing four index insurance designs found that only when farmers took a package of insurance, credit and inputs did they significantly raise investment in modern technology and productivity. A preliminary randomised trial of One Acre Fund’s programme, which combines input loans, insurance and training, found yields up 32% and profits up 23%, with evidence that relaxing information constraints alone was not enough; that version was labelled preliminary, so treat its figures with caution. A 3ie summary of an IFPRI trial in Kenya found risk contingent credit raised fertiliser adoption more than traditional credit. But another 3ie supported evaluation in Senegal found no evidence that index insurance bundled with loans raised investment, practices or incomes overall, so bundling is not a guaranteed fix.

When finance, inputs and advice arrive together

Change in per farmer outcomes in a randomised trial of One Acre Fund’s programme. Preliminary working paper by Deutschmann, Duru, Siegal and Tjernström (2019); estimates may have changed

+0%+10%+20%+30%+40%Yield: +32%+32%YieldProfit: +23%+23%Profit

Biological and ecological pest control. The best documented African case is push pull, where desmodium and a border grass repel and trap stemborer moths and suppress Striga. In on farm evaluations across Kenya, Uganda and Tanzania, farmers using the climate adapted version lost about 5% of maize to stemborers while neighbours lost about 40%, and harvested around 2.5 times more, according to the International Centre of Insect Physiology and Ecology. A study of 250 adopters found yields 2.7 times higher against fall armyworm. These are comparisons on working farms, not randomised trials, so adopters may differ from neighbours. A survival analysis of adoption also shows many farmers exit within five years, which is why extension and support matter as much as efficacy.

How much maize pests destroyed

Share of maize destroyed by stemborers in fields of 400 farmers in Kenya, Uganda and Tanzania, as reported by icipe (Midega and colleagues). Neighbouring figure is approximate

0%10%20%30%40%50%Maize lost to|stemborers,|neighbouring fields: about 40%about 40%Maize lost tostemborers,neighbouring fieldsMaize lost to|stemborers,|push-pull fields: 5%5%Maize lost tostemborers,push-pull fields

Read together, the studies support the sequence argument. Savings groups work as a base, mobile money widens reach, and finance changes farm results when it comes with inputs, insurance and advice. In biological control, evidence of effect exists but adoption needs support. None of it replaces the local evidence a cooperative needs for its own crops.

Where the two stories meet: proof at plot level

Lenders who finance inputs and regulators who approve them need the same thing: reliable evidence of what happened on real farms. SIANI lists evidence on efficacy, application and farmer outcomes as a priority, and the credit literature asks for records that a lender trusts. Plot level data could serve both. 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 for biological input trials or with savings group credit scoring, so that role is our proposal.

A test before scaling

CARE offers a simple test, which we adapt for agriculture. Programmes should be able to answer yes to each question before they scale.

QuestionIf the answer is no
Can women safely and independently use the technology?Invest in norms and community engagement
Can more than one group member operate it?Invest in shared skills
Is there a low tech fallback?Add USSD, voice or agent support
Were the tools co designed with users?Redesign with members
Do records show individual behaviour over time?Build the record first
Is credit introduced only after group maturity?Delay lending
Is there a real market or buyer beyond credit?Connect to value chains before scaling

What to do next

  1. Cooperatives and savings groups should keep individual, dated records of savings, loans and deliveries, on paper first if needed.
  2. Programme designers should apply the sequence: shift norms, build shared skills, then digitise, then link to lenders.
  3. Lenders should test scoring that treats seasonal income swings as normal, and keep loans small enough to serve cheaply.
  4. Regulators should write the East African framework into national rules and fund laboratories and evaluators.
  5. Input companies should run field trials on real farms and publish results farmers can see.
  6. Governments and funders should link registered biological products to procurement and extension, as the Aflasafe case suggests.

What we can and cannot claim

StatementEvidence status
500 million savings group members, 80% womenCARE’s estimate
Phone restrictions cut 57% in northern UgandaCARE programme result; not independently evaluated
Chomoka: $61,000, 500 groups, under 1% at riskCARE’s account of a partner platform
Agriculture gets 5% to 6% of bank lendingAfDB and press estimates; sources differ
Annual finance gap of $65 billion to $200 billionEstimates vary in scope and year
Eight countries adopted a biopesticide frameworkIndependent (Kampala) report of a workshop statement; not read in primary text
Aflasafe cuts aflatoxin by 80% to 90%Workshop presentation; not independently verified here
Savings groups raised business counts by 6% in three countriesPNAS randomised study via IPA summary; Ghana only trial found no business effect
M-PESA lifted 194,000 households out of povertyScience panel study; not randomised; press accounts differ on the number of women moving into business
One Acre Fund raised yields 32% and profits 23%Preliminary working paper; not the final published estimate
Push pull cuts stemborer losses to about 5% from about 40%icipe led on farm comparisons; not randomised
Plot data could support credit and input evidenceOur argument; not tested

The finance and the inputs fit one logic: build trust in the right order and the numbers follow. A savings group that has changed norms, shared skills and kept honest records can earn a credit score. A biological product that has been tested in real fields, registered in good time and bought by programmes can reach a farmer who trusts it. Both depend on evidence that people on the ground can see and lenders and regulators can check. Africa does not lack capital, ideas or farmers willing to try new things; it lacks enough carefully sequenced systems that connect them, and those are the systems worth scaling.

Sources & notes

SOURCES: Kaduru, Sriram and Liu (CARE), Digitizing Savings Groups: Five Steps That Cannot Be Skipped, FinDev Gateway (24 June 2026); Ayokunmi Sodamola, Designing Credit Systems for Informal Economies, NextBillion (30 June 2026); SIANI, Agrobiologicals for sustainable agriculture: lessons from Africa and Asia (11 November 2026); The Independent (Kampala) via allAfrica, East African states adopt common biopesticide rules; Dutch embassy, Kenya biopesticide registration consultation; FFTC summary of Alwora et al., Frontiers in Sustainable Food Systems, 2026; SEI, Kenya bio based inputs roadmap; Aceli Africa, Bridging the Financing Gap (2020); The Star, Africa faces US$100 billion annual agriculture financing gap; AfDB via allAfrica on smallholder financing; Milken Institute, Billion dollar opportunities; IPA, three country savings group study; J-PAL, Ghana savings groups evaluation; Suri and Jack, Science, 2016; Marr et al., interlinked insurance, credit and inputs, Ethiopia; Deutschmann et al., One Acre Fund trial (preliminary); 3ie Agricultural Insurance Evidence Program; icipe on climate smart push pull; Midega et al., push pull and fall armyworm; Tropentag 2025 poster on push pull adoption and dis adoption; AgriKonnekt.

Differences between sources: press accounts of M-PESA give different counts of women who moved out of farming into business, so we use only the Science paper’s findings; the One Acre Fund results come from a preliminary draft; the annual agricultural finance gap is quoted from $65 billion (agri SMEs only) to $200 billion; agriculture’s share of bank lending is 6% (AfDB) or under 5% (other estimates).

Our own arithmetic: the 98% remainder of Kenyan households; the mobile money index of 64; about $122 per group; the 80% Aflasafe remainder.

Not read: CARE’s underlying evaluations, GSMA’s 2025 report, the Cotonou workshop papers, Kenya’s draft registration data requirements. Programme results are as stated by their organisations.