Proof You Can Borrow Against. How Supply Chain Data Could Change Who Gets Credit

Proof You Can Borrow Against. How Supply Chain Data Could Change Who Gets Credit

The farmer has a real crop, a real buyer and a real track record. A bank can’t see any of it.

So in a lot of the farming communities I’ve worked in, the money for the next season arrives before the harvest does. Cash up front, paid back in crop. A cooperative or an aggregator hands over the money. The farmer buys seed, fertilizer and sometimes equipment. At harvest, the crop goes back to the buyer. And the terms are usually set by whoever has the money.

I’ve been thinking about this for a few years. I’ve always believed that traceability data could help lenders and insurers understand (and price) these farmers’ risk. Most of what they produce and get paid is recorded on paper, if it is recorded at all. Some payments move through mobile money or USSD phone transfers. But in many areas people are still paid in cash, so none of it builds a credit history a bank, lender or insurer can see.

Last time I wrote that traceability is where the work starts, and that finance was one of the places the data could go next. This one goes there. It is what I’m seeing and hearing, plus a bit of my own history with lending. So jump in and correct me where it doesn’t match what you’ve seen.

Why Is Credit So Hard to Get?

Let’s start with what happens without that history. People either don’t get credit at all, or they get it on poor terms. A bank can’t see income that never touched an account, so the farmer looks too risky. Microlenders step in without asking for collateral, which helps, but they charge higher rates to cover what they can’t see. For many farmers, the most reliable source of money is the buyer down the road.

That is where the forward-buying arrangements come in, and where people get stuck. Predatory isn’t the right word every time, but sometimes they can be quite extractive. The farmer needs money at planting, so they take the price and terms they’re given. At harvest the crop repays the advance, often at a price agreed months earlier. Then the next season comes, they need inputs again, and the cycle starts over. It is very hard to climb out of.

To be fair to the buyers, they carry real risk too. Some farmers sell to someone else at harvest, and some crops fail. That risk gets priced into the terms, and every farmer pays for it, good risk or not. The demand is clearly there. In its 2025 report, the advisory firm ISF Advisors put the annual finance gap for smallholder farmers at more than USD 200 billion.

What Can a Warehouse Receipt Do?

One of the big revelations for me came from a conversation with my advisor Lamon Rutten many years ago when we first started working together. Lamon has run commodity exchanges as a Managing Director in India and Indonesia, and before that spent many years at the UN working on commodity finance. As he tells it, some of the biggest social impact from those exchanges came from something quite unglamorous. The licensed warehouses set up around them.

The way he described it… a farmer or producer drops off their crop at a licensed warehouse and gets a warehouse receipt for what they dropped off. In some countries that receipt is bankable, so a bank will lend against it at much better rates than cash up front. The farmer doesn’t have to sell the day the harvest comes in. They can sell down the stored crop when they feel the price is right, and pay back the loan as they go.

India shows how far this can go. Its warehousing regulator has required receipts to be electronic since 2019, and banks lend against them. In late 2024 the government added a credit guarantee for these loans worth about USD 120 million, so farmers don’t have to sell in a hurry and banks feel safer lending.

In countries without this kind of system, people end up back in the first scenario. The crop is real and it is sitting there, but there is nothing a bank can lend against, because nobody independent can vouch for it.

Haven’t Lenders Done This Before?

Some of my thinking goes back to my time as a management consultant. Around 2021, I worked on a bank project in Australia that was starting to model credit risk using open banking data. Open banking began in the UK in 2018, when the competition regulator required the nine largest banks to let customers share their account data with other providers. Australia followed from 2020 under its Consumer Data Right.

The use-case we worked on was for credit cards. Normally you’d fill in a form saying what you earn, what you own and roughly what you spend. The bank takes those answers as true, adds a buffer, checks a credit bureau, and a risk engine sets your limit. Honestly, it was quite an archaic system. And most people underestimated how much they spent and overestimated how much they earned, which is exactly why the buffers were there.

Open banking let a customer choose to share their accounts at other banks. The system could then build a web of transactions across all of them, tracking money in, money out and the transfers in between, and turn it into curves and projections. Instead of guessing, the bank could see how income actually arrived and where it went. Better information meant the buffers didn’t have to do so much of the work.

Around the same time I spoke with the Yanir Yakutiel 🎗️ and Anna Hawter at Lumi, an Australian small-business lending fintech. They pulled a business’s financial records and checked the credit bureaus like anyone else. But they also had an advanced transaction-level risk engine that sorted and categorised how money actually moved through a business, and scored the risk from there. Looking back, they were well ahead of the curve leveraging AI and Machine Learning.

What stuck with me even more was who they lend to. Instead of competing head on with the big banks, they lent to a broader base of businesses. They didn’t pretend the risk was lower than it was. They built it into how they assessed and priced each loan, in a way that still worked for those businesses and still made sense for Lumi.

Another relevant experience I had was during my first ever internship at Zip, one of the early buy now, pay later (BNPL or Pay-in-4) companies and a main competitor to Afterpay. Credit cards assessed a person once and gave them a limit. Zip, Afterpay and the other BNPL’s went to the transaction instead, one purchase at a time, and transformed how a lot of people spend and borrow.

So What Has Been Missing?

What all three taught me is simple (in theory…).

When lenders can see new data, they find a rate where the risk makes sense. Then they build a business around people they used to turn away.

So I’ve long believed something similar could work in the early stages of agricultural supply chains. The big gap was the data, a verifiable history that someone produced something, sold it and got paid for it.

From what I’m seeing, lending already reaches part of these supply-chains. Banks are often comfortable lending to exporters and processors, with their accounts, contracts and volumes. Much less reaches the cooperatives and aggregators, and very little reaches the farmers behind them. That looks a lot like the gap Lumi and other fintechs went after. A lender that could see verifiable records further down could price that risk fairly, without fighting the banks at the top.

And honestly, after several years of thinking about this, not that much has happened. I may be a bit biased, because I haven’t seen it in many of the markets we work in. It probably happens in places I haven’t looked yet, and I’ll be doing some research on that.

Some of it clearly does. In Kenya and Zambia, Apollo Agriculture lends seed and fertiliser to smallholders without asking for collateral. It decides using farm data its field officers collect, satellite images and credit bureau records where they exist. Most of the repayment falls due after harvest. Its investor Triodos Investment Management said in May 2025 that it had served more than 400,000 farmers across the two countries.

What feels different now is that the foundation I wrote about last time is being laid, often because rules demand it. Europe’s deforestation rules, its digital product passports and the US food traceability rule all ask for records of where things came from and who handled them. Once those records exist, they can tell a lender something too. Production histories, yields, who delivered what and when, and who got paid.

Where Does It Get Hard?

I see two ways to lend into this, and both take time. The first is unsecured, which is what most microloans already do. No collateral, so the lender leans on what it can learn and charges for what it can’t. Verifiable records could make that risk much easier to assess. The big question is whether banks and insurers can assess it today.

I don’t think they’ll hand that judgement to someone else just because the data exists. Lending is their core skill, and these are new pools of clients, many of them high risk. So there is a bit of a social curve to climb before lenders get comfortable. In theory it makes a lot of sense. In practice it will take a bit of time.

The second is secured lending against the crop or product itself, the warehouse receipt route. I think that one is harder. You can’t just lend money into private warehouses and hope for the best. Someone independent has to control what goes in and out, and in many countries the licensing and rules behind a bankable receipt don’t exist yet. That will take time as well.

And through all of it, the data belongs to the people who produced it. If a farmer’s records are going to be used to judge whether they get a loan, they need to understand that and agree to it first. Otherwise it is just one more thing being decided about them, without them.

How Could This Play Out?

Here is how I see it going if it works. A cooperative has three seasons of deliveries, payments and yields on record, checked by someone outside it. A lender uses that history to price an input loan at planting, and the farmer buys fertilizer without promising the crop at a price fixed months ahead. At harvest they sell to whoever pays best, repay the loan, and next season’s record is a little stronger than the last.

And here is how it goes if it doesn’t. The records get collected because a rule says so, and they sit in a compliance file. Lenders never build the skills to use them, so nothing changes in what they offer. Farmers keep taking cash up front on the buyer’s terms, and the data they gave ends up helping everyone except them. Most likely it lands somewhere in between, and differently in each country.

Who Moves First?

I’m optimistic this will come together, I’m just not sure when.

The capital is out there, and forward buying shows people clearly want money up front. It already happens today, just on terms that rarely favour the farmer. We’re working on something along these lines right now. Part of it is in metals and minerals, and part is in agri-markets with the 5am.earth Foundation, and financial institutions are one of the groups they are building for. From what I’m hearing, it could be a very healthy market to lend into once the mix is right.

The bigger questions are who the new players will be, what capabilities they need, and what new ways of thinking about risk have to take hold. I think a lot of the answers sit in markets that are a bit more mature and took this risk early. We have quite a few of those in Australia, where I grew up. I’m confident these ideas will keep spreading into the markets I’ve been lucky to work in. and that many of you reading this have worked in too.

It matters because these are people who are often left behind. And as the world keeps changing, getting the right capital at the right time matters more and more. For many producers…it is how you keep going, put your family in a good spot, and help the community around you grow.

If you lend to, insure or buy from producers in these supply chains, I’d love to hear from you. What would you need to see in a producer’s records before you’d offer better terms? And if you’ve seen this work somewhere, I’m always happy to compare notes.