Skip to main content

What Australia's NAB taught me about emerging-market credit underwriting

Ahmed Arafat Joyadh · 2026-02-02

Three years inside National Australia Bank's investment banking division — and the credit principles I'm now applying to the most underbanked layer of South Asia's economy.

I spent 12 years in Australian financial services and 3 of those at NAB as an investment banker. The single most important thing I took with me — and what JOY's Joy Score credit infrastructure is built on — is something Australian banks understand and emerging-market lenders almost universally do not:

Credit risk is not a function of the borrower. It's a function of the borrower's environment.

Why Bangladesh workshops can't get capital today

Banks in BD ask the same three questions every emerging-market lender asks: collateral, salary slip, formal financials. Workshops have none of those. So the country's most economically active layer ends up running on personal trust networks, supplier credit, and informal moneylenders charging 30%+.

It's not that workshops can't pay. The default rates we observe in supplier credit data are lower than most consumer loan books. The problem is that legacy lenders have no way to see the signals.

30%+ Annualised cost of informal credit for BD workshops today

What an Australian bank does differently

At NAB, every credit decision was a multi-input vector — cash flows, transactional patterns, industry positioning, sector outlook, asset turnover. Not a single number. A risk surface.

Joy Score is the same idea, applied to data sources that didn't exist in Bangladesh five years ago:

  • Procurement velocity — how often a workshop reorders, what categories, seasonal patterns.
  • Supplier reliability — the supplier-side data we collect on every transaction.
  • Inventory turnover — the operational rhythm of a healthy workshop.
  • Cross-supplier behaviour — patterns visible only because JOY sits across multiple suppliers.
  • Vehicle-base concentration — fleet contracts, recurring service customers.

Why this only works as embedded credit

If you separate credit from commerce, you lose the data. You become another ZestMoney — a thin wrapper around a low-information underwriting model. Joy Score works because it's embedded. Every transaction enriches the model. Every workshop's behaviour informs the next workshop's underwriting.

Australia took 50 years to build the data infrastructure that lets a bank underwrite a small business in 90 seconds. Bangladesh doesn't have to wait — it can leapfrog, just like it did on mobile payments. JOY is the platform doing the leapfrogging.