The Tourism Industry on Holiday from Its Debt

The full numbers

Bhutan’s tourism sector held about Nu 27-29 billion in credit at December 2023. Of that, 75% was under payment deferment — meaning borrowers were not required to make interest or principal payments. Reported gross NPL ratio in tourism was 0.54%. Most peer Asian economies began their COVID forbearance unwinds in 2022 and had substantially completed by Dec 2023, with deferred-to-NPL conversion rates of 15-30%.

Imagine this

A hotel owner in Paro took a Nu 50 million loan in 2018 to build a 30-room boutique hotel. The hotel opened in November 2019. By March 2020, COVID had shut the borders. Through 2020-2022, occupancy was 0%. The bank deferred her loan payments. In 2023, occupancy returned to 30% — far below break-even. Her loan stayed deferred. Her loan stayed deferred until 7 April 2026, when the moratorium was finally lifted — six years after it began. She is technically a “performing” borrower because her loan is “deferred” and not “delinquent.” On the bank’s books, her Nu 50 million is a healthy asset, not a problem. The bank reports it as part of its low-NPL portfolio. In reality, she has paid almost nothing in seven years and may never catch up to the original schedule. What happens when the bank eventually has to mark her loan to current performance? It moves from “performing” to “non-performing.” Her loan, and thousands like it, suddenly appears on bank balance sheets as bad debt. The country’s banking sector NPL ratio jumps from headline low to peer-realistic — say, from 5% to 15%. That’s the unwind that hasn’t happened.

Where this came from

COVID-era loan forbearance was a globally coordinated response by central banks to prevent mass bankruptcy during the pandemic shutdowns. Every Asian central bank did some version of it. The intent was to bridge borrowers through the shutdown until normal activity resumed.

Peer countries began unwinding in 2022 as their tourism sectors recovered. Bhutan’s tourism sector recovered more slowly (210K arrivals in 2025 vs 315K pre-COVID = ~67% recovery), so the policy logic for continued forbearance has been “let them recover more before reclassifying.”

Why this matters now

The longer forbearance continues, the bigger the eventual reclassification when it ends. Banks that have been reporting low NPLs for five years will suddenly report 3-4x higher NPLs in one quarter when the policy ends. Capital adequacy ratios will be affected.

Lending capacity will tighten just as the country is trying to fund the 13th FYP’s Economic Transformation Programme. A delayed, larger NPL shock in 2027-2028 is meaningfully harder to absorb than a smaller, earlier one in 2023.

What it should be

By 2025-26, COVID-era forbearance should be largely unwound. Loans should have either returned to performing schedules, been restructured formally with new terms, or been classified honestly as NPLs. The peer-country experience is that unwinds cause short-term NPL spikes (15-30% of deferred loans surface as NPL) but restore the banking sector to a sustainable footing. How others did the unwind

• Indonesia — deferred-to-NPL conversion ~25% as forbearance unwound 2022-23. Banking sector NPL temporarily spiked from ~2.5% to ~3.5%, then recovered as restructured loans began performing.
• Philippines — 20-30% conversion 2022-23. NPL touched 4.2% before declining.
• Vietnam — 15-25% conversion 2022-23. The country accepted a temporary NPL spike as the price of cleaning up.
• Thailand — completed its tourism forbearance unwind by mid-2023 despite delayed tourism recovery.
• Bhutan — unwound last, and only in 2026. The moratorium was extended to 30 September, then 30 December, then 31 January 2026, and lifted on 7 April 2026, with banks given until 30 June to restructure what remained.

The question we should be sitting with

Was the unwind managed — or merely postponed until it could no longer be? The deferments ended on 7 April 2026, and the answer is now partly visible: at Bank of Bhutan alone, hotel and tourism non-performing loans reached Nu 1.149 billion by 30 June 2026, roughly double the Nu 635.7 million recorded six months earlier. Whose balance sheet absorbs the rest?