ASEAN Series Progress

Part 1 of 5 · 8 min read

90-second brief: Thailand's NCB reported 10.92% SME NPLs in June. The cooperatives holding those loans already knew in March. The gap isn't data quality — it's temporal architecture. By the time the bureau counts a default, the portfolio has already been in a CPL state for months.

For: Cooperative credit managers, SME association secretaries, portfolio officers, microfinance institutions, and anyone who reviews a quarterly report and already knows what it will say.

ASEAN Series · Signal Architecture

The bank found out in June. The shop knew in March.

MetriqOne ASEAN Series · Part I of V September 2026

A portfolio does not need a quarterly report to know something has shifted.

March. A cooperative credit officer in Chiang Mai sits with her member list. Three businesses she knows personally — a hardware supplier, a small contractor, a family-run restaurant — have already stretched their restocking cycles. Terms that used to run 30 days now need 45. The float that used to cover the gap doesn't stretch as far. Nobody writes any of this down in a form the head office would recognise.

Restocking takes an extra week to fund. A supplier who used to extend 45 days now wants 30. None of these is a crisis on its own. Together, they are a portfolio quietly renegotiating its own survival, one week at a time, with no one keeping score — and no signal reaching the officer who could restructure the terms before the default happens.

A cooperative credit officer reviewing member files, mid-arithmetic, before any bureau sees a signal

June. Thailand's National Credit Bureau publishes its quarterly numbers. Non-performing loans among SMEs and juristic persons:

10.92% of SME credit already non-performing — confirmed in June, felt in March

Another 4.95% flagged for "special attention" — not yet defaulted, but close enough that the bureau is watching. Combined, nearly 16% of the ฿2.54 trillion in business credit the NCB tracks is now either bad or turning. Total credit growth for the same businesses: 0.5% for the year. The system isn't lending its way out of this. It's watching it happen.

The NCB report is real, useful, and honest. It is also, structurally, an autopsy.

See how a Stacked Node Architecture catches this in March →

By the time the 10.92% figure exists, it has already happened to every business inside it. The cooperative member from March isn't in that number because she's about to default — she's in it, if she's in it at all, because she already did, months before anyone counted her.

Aged invoices and a bureau notice — evidence discovered after the fact

What the number is actually measuring

Read carefully, the NCB release is measuring two different things and naming them with one word. It measures non-performing loans — debt that has already failed. And it measures debt "requiring special attention" — the 4.95% still performing but showing the pattern that precedes failure. That second category is the closest the current system gets to a leading signal, and even it only exists at the loan-servicing level: a bank's internal classification, visible to the bank, invisible to the cooperative itself until the terms tighten.

Preventive restructuring — debt renegotiated before default — is up sharply this year: ฿330 billion against ฿250 billion a year earlier, and the accounts involved rose from 64,700 to 85,900. That's the system finding, after the fact, businesses that were already drifting and hadn't been given a way to say so earlier. The restructuring conversation happens at the moment the bank's own lagging model finally catches the pattern — not at the moment the cooperative officer first felt it.

The gap this leaves

A Stacked Node Architecture reads differently. The cooperative itself — not the lender, not the bureau — is the source of Context: what it's actually facing, logged as it happens, not reconstructed from repayment records a quarter later. A Performance Indicator that tracks days-to-restock, or the gap between invoice terms extended and invoice terms received, doesn't require a credit event to register. It moves before the loan does.

Crossing its CPL — the Counter Productive Level, the pre-signal threshold set below the point where the organisation's activity no longer serves the organisation, triggering Early Warning and Control before the process ever produces a special-cause signal — is what the cooperative's own evidence would have shown in March: a threshold crossed, five-plus months of lead time still available, long before any bureau's aggregate would have caught it in June.

None of this makes the NCB's data wrong. It makes it late by design — a property of measuring the loan rather than the business behind it. The 10.92% is the correct answer to the question a credit bureau is built to ask. It was never going to be the correct answer to the question the cooperative officer needed answered in March.

Why Thailand, first

Thailand publishes this kind of data cleanly, regularly, and in enough detail to see the shape of the problem — which is exactly why it's the right place to start naming it. The pattern here is not a Thailand problem. It's what a trailing-indicator system produces anywhere it's the only system running.

A row of shopfronts at dusk, watching the street

How many portfolios across the region are living through their own March right now, with no architecture built to hear them until the number arrives?

Part II looks at a country that just changed its own credit classification at the national level — and appears to be borrowing its way into the same blind spot. The Philippines, next. Read Part II →

Start your first deployment → See how it works

Which of these sounds like your organisation?

We find problems in quarterly reports, never in real time. That's the Thailand lag.

One number drives a decision that affects thousands. That's the Philippines single-metric trap.

Our headline looks calm but we know the churn is hidden. That's the Vietnam calm.

We look stable, but we're not watching for drift. That's the Malaysia false calm.

None of the above — but I'm not sure we'd know. That's why the architecture matters.

그게 중요할까요? — 측정해 보세요!

Signal Architecture, ASEAN, Credit Stress, Thailand, MetriqOne, Counter Productive Level

ASEAN Series Part II — The Philippines Part III — The Region Part IV — The Protection Gap Part V — The Build

Forward this case

One line: Thailand's NCB found out in June. The cooperative knew in March. Here's why that gap matters for your portfolio.

Two paragraphs: Thailand's National Credit Bureau reported 10.92% SME NPLs in June — a correct number that arrived five months after the businesses in that number already knew they were drifting. The problem isn't the bureau. It's the architecture: a system built to confirm what already happened, with no layer underneath to catch what's happening now. A Stacked Node Architecture, with a Counter Productive Level set below the point of failure, would have given the cooperative officer five months of lead time to restructure before the default ever registered.

MetriqOne · 잡음 없는 증거 · metriq.one