The Big Read: the diaspora is rebuilding its own credit graph

In a Dakar courtyard last November, twelve women passed a paper notebook around a low table. Each entry was a name, a date, a number in CFA francs. The notebook clears more credit in their neighbourhood than the nearest licensed bank branch.

They call it a tontine. In Lagos it is an esusu. In Beirut a jamia. The form is older than central banking. What is new — and what we set out to document over six months in six corridors — is that the notebook is becoming a database.

I. The notebook is becoming a database

The Senegalese diaspora moves roughly EUR 2 billion home each year through licensed channels. We estimate, conservatively, that another EUR 1.4 to 1.8 billion moves through unlicensed corridors — agent-to-agent, WhatsApp-cleared, in cash on arrival. The licensed figure is published by BCEAO every quarter. The unlicensed one appears in no public document.

What has changed since 2023 is not the volume — informal flows have been roughly constant in real terms for two decades — but the persistence layer. A tontine that used to live in a notebook now lives in a WhatsApp group, a Google Sheet, or, increasingly, a small mobile app written by someone's nephew in Lagos. The data is structured. It can be queried. It can, in principle, become collateral.

II. Six corridors, one pattern

We selected six corridors that together account for roughly 38% of LMIC remittance volumes: UK→Nigeria, France→Senegal, Canada→Morocco, UAE→India, USA→Mexico, and Australia→Philippines. In each corridor we interviewed at least one informal operator, one licensed MTO branch manager, one regulator, and one diaspora-finance founder.

The pattern repeats with unsettling consistency. The licensed rail is faster and safer than it was five years ago — Wise, Remitly and the local mobile-money networks have closed most of the speed gap. But it is also more expensive at the bottom of the market, where transfers under USD 100 still carry effective costs above 8%, and where KYC frictions disenfranchise undocumented senders entirely.

III. The credit graph nobody is building

Every informal operator we interviewed maintains, in some form, a reputation ledger: who pays back, who delays, who defaults. These ledgers are not anonymised, are not transferable, and are not bankable. They are also remarkably accurate. One Casablanca operator, who has been clearing MAD-EUR for fifteen years, estimates his loss rate at 'between 0.4 and 0.6 percent.' That figure, if true, would compare favourably with most retail consumer-lending books in the Eurozone.

What incumbents have missed — and what a handful of corridor-native startups are now beginning to understand — is that the question is not how to extract the informal sector into the formal one. It is how to give the informal sector instruments that respect its existing trust structures while making its repayment data legible to outside capital.

IV. What this means for cross-border capital

For the cross-border lender, the implication is uncomfortable. The non-resident mortgage market — the market Nova RealFi operates in — is downstream of exactly this informal credit infrastructure. The young Senegalese engineer in Paris who wants to buy a building in Dakar has, almost always, a parallel financial life her French bank cannot see. Her family's tontine record is, in practice, a better predictor of her capacity to service a euro-denominated loan than her three most recent payslips.

None of this is published in a regulator's annual report. We spent six months trying to write it down.

Methodology

This essay draws on 47 interviews across six corridors between October 2025 and April 2026, supplemented by transaction-level data shared on background by three diaspora-finance operators (anonymised at their request), and on the public datasets cited in our /method page. Estimates of unlicensed flows use the Freund-Spatafora residual methodology applied to BCEAO, CBN, Bank Al-Maghrib, RBI, Banco de México and BSP data; full workings on request.