Borrowing to eat

Thai household debt reached ฿16.44 trillion in the fourth quarter of 2025 — 86.7% of GDP.

It peaked around 95.5% in early 2021 and has stayed high since. By regional standards it is among the heaviest household debt burdens in Asia.

That number gets reported every quarter and then nothing much is said about it. Two things about it are more interesting than the number itself, and neither appears in the quarterly coverage.

One: what the borrowing is for has changed

The recent growth is not mortgages. It is not cars.

It is personal consumption loans for day-to-day expenses.

Households are shifting toward state-owned financial institutions, savings cooperatives and pawnshops, as commercial bank lending tightens. Bank lending is ฿6.09 trillion of the total; specialised financial institutions are at ฿4.55 trillion and savings cooperatives ฿2.53 trillion, and both are rising.

And the distress sits exactly where you’d expect. Credit cards carry the highest NPL ratio at 4.61%, and as of September 2024, 64% of non-performing loan accounts were credit card and personal loans.

Borrowing to buy a house is one thing. Borrowing to buy food is another. Thailand’s household debt is increasingly the second, and a debt ratio composed of consumption loans behaves nothing like one composed of mortgages — there is no asset at the end of it, and no point at which it stops being a monthly problem.

Two: the number can’t see the rest of it

Here is the measurement point, and it matters more than the level.

Research on Thai credit segmentation puts formal debt at 91.4% of household debt to GDP, and informal debt at 25.2% gross, or 12.3% net.

The headline ratio measures what the financial system can see.

Which means that when banks tighten — as they have — the borrowing does not stop. It moves. To the cooperative, then to the pawnshop, then to the lender on the soi with no paperwork and no ceiling on the rate.

The debt doesn’t leave the economy. It leaves the statistics.

So a falling household debt-to-GDP ratio in Thailand can mean households are deleveraging, or it can mean they have moved to lenders nobody counts. You cannot tell which from the number, and I have not seen anyone acknowledge that when quoting it.

The people doing the moving are identifiable: research finds the unemployed, retirees, business owners and freelancers relying disproportionately on informal channels. Which is to say — everyone this site has spent a hundred articles describing.

What ninety-nine articles were actually about

I did not plan for the hundredth article to be this one, and then it became obvious.

Article 92: 27% of Thai parents reported borrowing — from relatives, pawnshops or loan sharks — to keep a child in school, against a ฿400 uniform subsidy and a ฿1,500 uniform.

Article 80: the gold shop as the household credit line, because it pawns without a credit check.

Article 57: savings circles, funeral societies and family remittances — an entire informal financial architecture built where the formal one doesn’t reach.

Article 74: a minimum wage around ฿10,400 a month, against a street vendor’s ฿20,000 and a win rider’s ฿24,500.

Article 38 and 67: a ฿600 state pension, topped up by adult children at a median ฿22,250 a year, under an obligation whose arithmetic is failing.

Article 97: cheap imported goods as the most direct real-terms income gain available to poor households — which tells you how little else was on offer.

Article 89: protection running inversely to exposure across nine separate systems.

฿16.44 trillion is what all of that looks like added together.

It is not a credit story. It is an income story wearing a credit story’s clothes. People do not borrow at pawnshop rates for daily expenses because credit is too available. They do it because the gap between what they earn and what they need is structural, and borrowing is the only instrument they have been given to close it.

Article 09 asked what happened to the country that grew at 7%. This is a large part of the answer. A household spending a significant share of income servicing consumption debt is a household not spending, not saving and not investing — and seventy-one million of them is a demand problem that no interest rate fixes.

The relief schemes, fairly

Something is being done, and it deserves a fair hearing rather than the reflexive moral-hazard response.

“Clear Debt, Move Forward” — a joint Finance Ministry, Bank of Thailand and Thai Bankers Association programme — targets small individual debtors with non-performing loans under ฿100,000, with around 3.4 million people in scope and debt transfers beginning 1 January 2026. A related plan addresses 2.3 million NPL accounts.

Election proposals have gone further, including clearing unsecured NPLs up to ฿200,000 on payment of 10%, and a budget allocation in the billions for informal debt.

TDRI and others have warned that repeated relief weakens financial discipline. That warning is legitimate and I’m not going to dismiss it.

But both things are true at once, and the argument usually only holds one.

A ฿90,000 personal loan taken to cover living costs is not imprudence. It is triage, and treating its forgiveness as a moral hazard problem misdescribes what happened.

And repeated, election-adjacent forgiveness does erode repayment norms in a way that makes future credit more expensive for exactly the same people.

The resolution isn’t in the debt policy at all. It’s upstream. A relief scheme addresses a stock. The flow is what put it there, and the flow is wages, the ฿600 pension, the ฿400 uniform subsidy, and every other place this site has found a number set below the cost of the thing it was for.

What I’d want measured

Publish an informal debt estimate alongside the official ratio, every quarter. Not instead — alongside. Until both move in the same report, nobody can tell whether an improvement is real, and the entire policy debate is conducted on a number that stops at the boundary of the formal system.

Break the ratio down by purpose. Mortgage debt and consumption debt are different economic objects and reporting them as one obscures the only thing that matters.

And track the pawnshops. Article 80 established that gold pawning is regulated and partly municipal. Pawn volumes are a real-time indicator of household stress that requires no survey, and I have not found anyone publishing them as one.

Common misconceptions

“86.7% is manageable.” The level matters less than the composition, and the composition has shifted toward consumption lending with no asset behind it.

“Household debt is falling, so things are improving.” It may be moving to lenders outside the statistics. The official figure measures the formal system only.

“People are borrowing irresponsibly.” The growth is in personal consumption loans for day-to-day expenses, among groups with the least access to formal credit.

“Debt relief is just vote-buying.” Some of it addresses genuine triage cases; repeated relief also does weaken repayment norms. Both are true.

“This is a banking problem.” Banks are tightening. The borrowing is moving to cooperatives, pawnshops and informal lenders, which is a different and less visible problem.