Bhutan's Loan Book: The Year Tourism Turned

Buried in Bank of Bhutan’s audited accounts for 2025 is a table called Loans and NPL by Sectoral Classification. It is not discussed in the narrative sections. It runs to fourteen lines and it says something the commentary around Bhutanese banking has largely missed.

Here is the line that matters:

Service and tourism non-performing loans went from Nu 30.6 million to Nu 635.7 million in twelve months. A twentyfold increase. Nearly 2,000 percent.

To put that in proportion: Bank of Bhutan’s total non-performing loans rose by about Nu 1.08 billion across the year. Tourism alone accounts for 56% of it — more than every other sector in the book combined.

Everyone is watching the wrong sector

The conversation about Bhutanese credit risk has, reasonably enough, been about housing. Housing is the largest single category on the book — Nu 29.8 billion, roughly a third of all lending — and its bad debts did rise sharply, from Nu 389.6 million to Nu 686.3 million.

That is a 76% increase on a loan book that grew only 3.6%, which is a genuine deterioration and worth its own attention. But it is not the emergency. Housing’s default rate is 2.30%. Tourism’s rise is nearly twenty-six times larger in percentage terms and more than twice as large in ngultrum.

The full table, ranked by how fast the bad debts grew:

What happened to tourism: too many beds

The mechanism is documented, and it is not really a tourism story. It is a property story wearing a tourism label.

Bhutan built hotels faster than it acquired guests. In 2019 the country had 160 certified hotels. By September 2025 it had 397 — more than double — concentrated in Thimphu, Paro and Punakha, and comprising 8,625 rooms. Three-star properties account for the largest share at 202.

Arrivals went the other way. Bhutan received 315,599 visitors in 2019. The borders reopened only in September 2022 and the country saw 103,066 visitors in 2023, recovering to 145,065 in 2024 and 209,376 in 2025 — still a third below the 2019 peak, six years on. Between January and August 2025 it received 115,536.

The result is the number that explains the whole line item: average industry occupancy of 20.8 percent, with some three-star hotels in single digits and rooms selling at Nu 2,500 a night for two people including breakfast and dinner — pricing aimed, as Kuensel put it, less at profit than survival. Break-even across 397 hotels would require something like 6,000 tourists arriving every day.

The picture has not improved since. In August 2026 a Bumthang hotelier described the district’s occupancy as eight to ten percent — “we work for three months and spend the earnings over the next three months. So we are left with nothing.”

The Hotel and Restaurant Association of Bhutan puts roughly 70 percent of hotels nationwide in financial difficulty. At least five have ceased operating. Others are surviving by moving into other businesses or liquidating assets, including land and property. Its chairperson’s assessment: “Hotels below three stars are in intensive care units.”

That is a construction boom meeting a demand curve that never came back — the ordinary anatomy of an overbuilding bust, financed by bank credit and now surfacing as bad debt.

Why it surfaced in 2025 — and it was not the deferments ending

The natural assumption is that the numbers jumped because a pandemic-era moratorium lapsed and hidden defaults became visible. That is the tidy explanation, and the timeline does not support it.

Tourism was the sector most sheltered through the pandemic — at peak, roughly three-quarters of hotel and tourism lending sat deferred. But the repayment moratorium did not end in 2025. It was extended — to 30 September, then 30 December, then 31 January 2026 — and finally lifted only on 7 April 2026, with banks given until 30 June to restructure what remained.

So the December 2025 figure above was recorded while the deferment was still in force. And at that date, according to the RMA, 85 percent of hotel and tourism borrowers were still servicing their loans, with only 15 percent in deferment.

That changes the reading. The bad debts did not appear because a measurement holiday ended. They appeared among borrowers who were supposed to be paying and could not — at 20.8 percent occupancy, on loans taken to build hotels for guests who did not come. The deferment was not hiding the problem; it was holding back an even larger one.

And the trend continued. By 30 June 2026 — after the moratorium finally lifted — Bank of Bhutan’s hotel and tourism NPLs stood at Nu 1.149 billion, on a book that had grown to Nu 29.0 billion. The Nu 635.7 million in this article’s headline nearly doubled again in six months.

Across the whole system, tourism NPLs stood at Nu 549.21 million in May 2025. A single bank was above that by December, and at double it by mid-2026.

The state’s response is a four percent interest subsidy for hotels of four stars and below, with roughly Nu 845 million budgeted for a year. Enrolment jumped from 25 hotels to 218 between July and August 2026 as the scheme was widened. Actual disbursement, though, was Nu 3.75 million in May and Nu 7.21 million in June — a little over one percent of the budget, six months in. The support exists on paper; whether it arrives before the loans go bad is a separate question.

Where the bad loans are

The RMA’s district figures locate the problem precisely, and they land where the hotels were built.

(RMA figures as at May 2025. The non-performing column is tourism-sector — the three districts account for 81% of the Nu 549.21M national tourism total. The portfolio column is reported as the district’s overall loan book, so the two columns may not share a base and no default rate is calculated from them here.)

Thimphu alone carries 58% of the country’s bad tourism debt. Thimphu and Paro are also the two towns where the hotel count more than doubled. The distress is where the building was.

One detail is worth holding for later. Hotels in difficulty are liquidating land and property to stay alive. In a country where good land almost never comes to market, distressed hotel sales may be among the few urban transactions actually occurring — which means the visible price data is disproportionately made of forced sales.

The quiet one: agriculture

Tourism is the loudest line. Agriculture is the worst.

Its default rate is 15.46% — the highest in the entire book, five times the bank’s average, and on a loan book that shrank slightly over the year. Trade and commerce runs at 12.61%, transport at 11.12%.

These are the sectors where lending has been chronically difficult and the numbers reflect it, year after year, without much comment. A tenth of the country works in agriculture and a sixth of the money lent to it is not being repaid.

And then the line nobody expected

The fastest-growing category on the entire book is education lending: up 47.9%, from Nu 6.6 billion to Nu 9.8 billion in a single year.

That figure needs context, because it is not the whole country following suit. System-wide education lending stood at Nu 13.4 billion at the end of 2025 — which means Bank of Bhutan alone now holds roughly 73 percent of every education loan in Bhutan. In 2021 the category was dominated by the pension fund; four years later one commercial bank has most of it.

Set that beside a fact from elsewhere. Australian student-visa grants to Bhutanese nationals fell from 15,552 in FY2022-23 to 9,787, then to 6,186 in FY2024-25 — a 60% collapse from the peak — and in January 2026 Australia moved Bhutan to its strictest visa-scrutiny tier. The RMA’s own annual report for 2023/24 recorded education credit declining 5.2%, citing “shifts in education policies abroad.” It rebounded 5.2% the following year.

So one bank expanded education lending by nearly half, and took three-quarters of the national market, across the period in which the principal destination for Bhutanese students closed by 60%.

There are innocent readings, and they should be given. Loans are drawn before departure and the lending cycle lags the visa cycle; students already abroad borrow for later years; other destinations have absorbed some redirected demand. The loans themselves are structured cautiously — secured against a land thram up to Nu 5 million, or unsecured only with a family guarantor and a household income cap — and the RMA limits any bank’s exposure to 10 percent of its portfolio. The default rate on this book is 0.75 percent at BoB and 1.0 percent system-wide, so nothing has gone wrong yet.

But the timing is worth stating plainly, because education loans are underwritten against an expectation — that the borrower will study abroad, earn in a hard currency, and repay from it. That expectation was being written into new loans at an accelerating rate, and concentrated into a single institution, during the two years the pathway narrowed. If it holds, the growth is prudent. If it does not, this line is where it will show — and the current default rate tells you nothing about that, because these loans have not yet reached repayment.

What the table is actually for

Two things are worth taking from it.

First, that sectoral disclosure is the most useful thing in a Bhutanese bank’s annual report and almost nobody reads it. The headline numbers — profit, total NPL ratio, capital adequacy — compress everything into figures that can only move slowly. The sectoral table is where a 2,000% change sits in plain sight, published, audited, and unremarked.

Second, that the aggregate is misleading in both directions. Bank of Bhutan’s overall default rate is 2.93%, which sounds unremarkable and internationally is. Inside it are sectors at 15.46% and 12.61%, and sectors at 0.32%. A single national NPL ratio tells you almost nothing about who is failing to pay, and the answer this year is: hotels, farms, and traders — not, mainly, households with mortgages.

The bank remains profitable and comfortably capitalised, with a capital adequacy ratio of 13.85% against a 12.5% floor. Bhutan’s banking system posted record profits in FY2025. None of that is contradicted by the table.

But profits are a report on the year that finished, and non-performing loans are a report on the years that have not. In 2025, with the deferments still in force, the books began to show that Bhutan’s tourism sector had been carrying something for five years that nobody had yet been required to count.