Principal, then interest, then penalty

On 19 March 2023 Thailand published an amendment to its student loan law. It came into force the next day.

Most of what it did was reported. Interest capped at 1 per cent a year. The penalty for late payment cut to 0.5 per cent. Guarantors abolished.

The clause nobody reported is section 19, which inserted a new section 44/1, and its third paragraph is eleven words long in Thai. When a borrower pays, the money is applied to principal then due, then to interest, then to the penalty.

That sentence is the article. Everything else follows from it.

Why the order decides everything

Think about what it replaced.

Before, a payment went to the penalty first. And the penalty, on the Fund’s own account, used to run at 12 to 18 per cent a year, later 7.5 per cent.

Now follow a borrower who fell behind. They owe principal. On the overdue part a penalty accrues at up to 18 per cent. They pay what they can. The payment is swallowed by the penalty. Next month the same principal is still outstanding, a new penalty accrues on it, and they pay again — into the penalty.

They can pay for years and never reduce what they borrowed. The balance does not fall because nothing they pay ever reaches it. This is not a failure of discipline. It is the order of appropriation, working exactly as written.

Reverse the order and put principal first, and every payment bites. Cap the interest at 1 per cent with compounding expressly prohibited, drop the penalty to 0.5 per cent, and the debt becomes something a person can actually extinguish.

That is what the 2023 amendment did. And then it did something much stranger.

It went backwards

Statutes normally take effect going forwards. This one was written to reach back.

Section 27 applies the new repayment provisions to borrowers and guarantors who had already borrowed or guaranteed before it came into force.

Section 29 goes further, and is the provision worth reading twice. Any rule the Board makes under the Act that is favourable to borrowers or guarantors must be made to apply to people who borrowed or guaranteed before the Act — “even where there is a final judgment of the court, or execution is under way.”

Section 44 carries the same instruction into the mechanics. The Fund may reduce, restructure, novate or suspend a debt during litigation, after final judgment, or during enforcement. And where a debt is novated after final judgment or during execution, the judgment debt is extinguished, and any seizure or attachment of property must be lifted.

Read plainly: Parliament instructed a state creditor to go back to cases it had already won and unwind them.

And section 29’s last line releases the guarantor — where the Fund grants forbearance under section 44 after the Act, the guarantor is discharged from the debt.

Article 176 Taking turns found Thai teachers guaranteeing each other’s loans into ruin, with the guarantor’s liability outliving the borrower’s capacity to pay. This statute did the opposite thing deliberately: it prohibited the Fund from requiring a guarantor in any case at all, and then went back and let the existing ones out.

Section 22 completes it. The debt is extinguished on death; on bankruptcy, for whatever the estate does not cover, unless the bankruptcy was fraudulent; on disability preventing work; and on serious illness preventing work. Section 28 applies that retrospectively too — though with a limit that tells you something about how these things are drafted: anyone who already paid gets nothing back.

Then somebody had to do the arithmetic

Here is where a generous law meets an accounting system.

Changing the order of appropriation retrospectively means you cannot simply adjust a balance. You have to take every payment a borrower has made since their first due date, discard how it was applied at the time, and apply it again under the new order, with the new interest rate and the new penalty rate, through however many years have passed.

For every account.

The Fund began showing recalculated balances in its app in August 2025, two and a half years after the Act took effect. It had finished 556,000 accounts. A week later, another 430,000. A week after that, 770,000 more, for a running total of 1.75 million.

It reached roughly 3.9 million accounts by June 2026 — more than three years after the law came into force.

And about 200,000 of them were wrong. Borrowers who had closed their accounts found balances had reappeared.

The disclosure the Fund made about itself

This is the part that deserves attention, and the source for it is the Fund’s own published explanation.

Some borrowers opened the app and found their debt had gone up.

The Fund explains why. From when the law took effect, and while it was still building the system to comply with it, every payment received was applied 100 per cent to principal. Not principal then interest then penalty — all of it to principal, with interest and penalty simply not deducted at all.

So principal fell faster than it legally should have. The Fund’s own diagram labels the two columns “temporary appropriation” and “appropriation according to law.”

When it finally ran the legal calculation, the unpaid interest and penalty had to come out of payments that had been credited to principal — and the principal balance rose.

Nobody was cheated and the Fund published the explanation itself. But consider the position of a borrower who paid every month for three years, watched the balance fall, and then watched it go back up because the institution had been using an interim method it knew was not the statutory one. A repayment system’s most valuable property is that a person can predict what their payment will do, and for those years it did not have that property.

And after all of it, two million people are in arrears

Reporting in April 2026 put borrowers in default at over two million, owing more than ฿100 billion. Around 100,000 who had been in arrears for four years or more were sent notices terminating their contracts, with legal proceedings to follow.

So: interest at 1 per cent, no compounding, a penalty of 0.5 per cent, no guarantor, a two-year grace period after graduation, up to fifteen years to repay, monthly or quarterly or annual instalments at the borrower’s choice, the debt written off on death or incapacity, and judgments reopened. By the standard of Thai consumer debt this is an extraordinarily soft instrument. Article 149’s illegal lending apps and article 180’s hire purchase are a different universe.

And default runs to two million people.

Which points at the mechanism, not the terms

Section 23 of the amendment rewrote how the money is actually collected. An employer who pays assessable income under section 40(1) of the Revenue Code — the employment category — must deduct the amount the Fund specifies from a borrower-employee’s pay and remit it to the Revenue Department on the withholding-tax timetable.

It is an elegant piece of design. No collection agency, no court, no letters. The money never reaches the borrower.

It also has one requirement: an employer, paying employment income, running payroll.

Article 57 established that Thailand’s parallel economy is not a fringe and article 89 The wrong way round found the pattern this is an instance of: a Thai protection or obligation attaches to a category, and the category is drawn around people the administration can already see. Payroll deduction reaches the graduate in a salaried job at a registered employer. It does not reach the graduate selling online, riding for a platform, working in a family business, farming, or moving between informal jobs — which, on this archive’s own evidence, is a great many of them.

So the collection mechanism is most effective on exactly the borrowers least likely to default, and least effective on the ones most likely to. The two million are not, on the whole, people dodging a payroll deduction. They are people there is no payroll to deduct from.

The Act shows some awareness of this. A new section 38/1 requires the Fund to collect and publish, at least once a year, statistics on whether its borrowers found work and what kind of work it was, together with forecasts of future demand — so that a student can see what a given course actually leads to before borrowing against it.

That is the right instinct and it is worth saying so. It also concedes the point: if the Fund has to tell students which degrees lead to employment, it knows that repayment depends on landing in the part of the economy its collection mechanism can reach.

What the article will not say

It will not say the 2023 reform failed. It cut the penalty by a factor of between twenty-four and thirty-six, stopped payments disappearing into charges, released guarantors, and reopened judgments. For a borrower who is employed, it is transformative, and for a guarantor it may have ended something that was going to follow them for life.

What it says is narrower. The reform fixed the arithmetic of the debt. It did not change where the debt is collected from — and in a country where the formal payroll is not where most people are, that is the constraint that decides the outcome.

---

Common misconceptions

“The interest rate was the problem.” Interest is capped at 1 per cent and compounding is prohibited. The penalty, formerly as much as 18 per cent a year, combined with an order of appropriation that paid it first, is what stopped balances falling.

“The new rules only apply to new loans.” They were written to apply retrospectively to existing borrowers and guarantors, expressly including cases with a final judgment or under enforcement.

“If my balance went up, there’s been a mistake.” Not necessarily. The Fund states that during the transition it applied payments entirely to principal, and that correcting this can raise the principal balance. Separately, about 200,000 accounts were wrong.

“Guarantors are still liable.” The Fund is prohibited from requiring a guarantor in any case, and existing guarantors are discharged where forbearance is granted under section 44.

“The debt follows you to the grave.” It is extinguished on death, and on disability or serious illness preventing work.