The loan officer in Besisahar showed me a ledger of forty-one outstanding agricultural loans, of which he expected to recover perhaps thirty-five. On the remaining six, the security was land. He had never taken any of it, and when I asked whether he would, he laughed in a way that made the question seem naive.
He explained: the parcel is a third of a hectare on a slope, it is registered to a man who has been in Malaysia for four years, the family living on it are his brother's, and it has no road access. If the cooperative seized it, they would own an unsellable field and would have destroyed their relationship with every borrower in three villages. The security is not security. It is a form.
I have spent two years looking at rural credit in the middle hills and this is the central fact I keep returning to, and it is not the one the literature emphasises. The literature is mostly about interest rates and financial inclusion. The binding constraint is that collateral does not function, and every institution has quietly built around that fact in a way that shapes who gets credit far more than any rate does.
Here is how the building-around works, in practice.
Since land cannot be realised, the cooperative lends on reputation. That means the loan officer's knowledge of the family — who repaid in 2019, whose brother sends money, which household had a bad year for reasons everybody understands and which had a bad year because the son gambles. This is genuinely good information, better than any credit score would be, and it is why cooperative recovery rates are as high as they are.
But reputation is only legible to somebody embedded in the place, and it is not portable. A man who has an excellent standing in Besisahar has nothing at all forty kilometres away. He cannot take his creditworthiness with him, which means he cannot borrow where the opportunity is, which means capital does not follow returns; it follows familiarity. Every village has its own credit market with its own ceiling, and the ceiling is set by how much the cooperative has, not by how much the borrowers could productively use.
The second consequence is the one I find harder to write about. Reputation lending is systematically unavailable to anyone whose standing is not their own. A daughter-in-law running a poultry operation borrows on her husband's family's name or not at all. A household that arrived from another district in the last decade is outside the information network no matter how solvent it is. A man whose brother defaulted carries it. None of this is written in a policy and all of it is enforced more reliably than any policy could be.
Now the complication that I think most outside analysts get wrong, and that I got wrong for the first year.
The obvious remedy is to make land collateral work: clean up the land registry, formalise titles, enable foreclosure, and the standard property-rights argument follows — secure title unlocks credit, credit unlocks investment. It is an elegant argument and there is a serious literature behind it.
It would be a disaster here, and the reason is the same reason the loan officer laughed. In a landscape where a third of working-age men are abroad, titles are held by people who are not present, occupied by relatives with informal claims that everyone recognises and no document records. Make foreclosure genuinely enforceable and you have not created security; you have created a mechanism by which a family living on and farming a plot can be removed by a creditor holding paper signed by an absent brother in Qatar. The village knows this. It is why the informal claim is respected and the document is not.
So the honest position is that the collateral problem cannot be fixed by fixing the collateral, at least not before the migration pattern that hollowed out the ownership changes. That returns me to the same place my other arguments about this country return to, which is tiresome but I think correct: this is downstream of the absence of work.
What can be done in the meantime is smaller. Portable reputation is the piece that could actually be built. If the cooperatives in a district shared a repayment history — not a score, just a record of who repaid what and when, contributed by the institutions that already hold it — a borrower's standing would travel with them as far as the district. That is one database, a governance agreement, and no change to property law at all.
I have raised this in two meetings. The objection, both times, was that cooperatives would not want to share their good borrowers with each other. Which is honest, and is a competition problem rather than a technical one, and is at least the right size of obstacle to be arguing about.
The officer let me photograph the ledger with the names covered. Six loans he does not expect to recover, secured on land he will never take, extended anyway, because the alternative was lending nothing to six families he has known his whole life. That is either a scandal or the only functioning part of the system, and after two years I still cannot decide which.
