Chapter Two

The Closed Loop

3 minute read · 12 paradoxes

In early 2026, the Government of India sent Bhutan Nu 2.5 billion in stimulus aid.

The aid was meant to help Bhutan through a fuel-price shock. Petrol had pushed past Nu 170 a litre. The state was subsidising the difference at the pump to protect households from the spike. The aid was the Indian government’s way of saying: we know this is hurting you. Here is some cash. Deploy it however you need to.

The aid arrived as flexible fiscal support. It could be used for chiwog road blacktopping. It could be used for the agricultural price guarantee. It could be used for youth employment, tourism development, creative industries, small-business credit. It could be used to soften the fuel shock for citizens at the pump.

By 21 May 2026, sixty-one percent of it was gone.

Guess where it went.

Nu 1.531 billion had been spent paying Indian Oil Marketing Companies for diesel and petrol. The fuel had been refined in Assam by Indian state-owned firms and sold to Bhutan at international prices. The Bhutanese state had used Indian aid to subsidise Indian fuel sold to Bhutanese citizens.

Chiwog roads: cut. Agricultural price guarantee: cut. Youth employment programmes: cut. Tourism development fund: cut. Creative industries: cut. Concessional small-business credit: cut. All of it cancelled to keep the pump price below Nu 100 a litre while the wholesale price climbed past Nu 200.

The aid arrived as aid. It left as fuel. It went back to the country that sent it.

This is not an accusation. It is arithmetic. The Bhutanese state was the conduit because there was nowhere else for the money to go. The fuel bill was real. The pump price was political. The aid was flexible. The development line items were not.

But the deeper question — why does the country that exports the cleanest electricity in South Asia end up borrowing from its electricity buyer to subsidise the fuel it imports from the same buyer — has a longer answer. It begins forty years before the aid envelope arrived.

Wangchuk runs a small pig farm in Tsirang, about thirty kilometres north of the Indian border. He raises mostly Yorkshire-Duroc crosses on a plot his father cleared in the 1980s. In a normal year he sells about a hundred and twenty pigs through the wholesale market in Phuentsholing — roughly twelve metric tonnes of pork at three hundred ngultrum a kilo, less feed costs and labour.

In March 2026, just as Wangchuk was preparing his second-quarter sales, a quiet announcement reached his cooperative office. The Agricultural Price Guarantee Scheme — a programme the Ministry of Agriculture had been preparing to expand in 2026 to support smallholder livestock producers — was being put on hold. The scheme would have set a floor price below which farmers would receive a state top-up. For Wangchuk’s pigs, the floor was at two hundred and twenty ngultrum a kilo. The market price in 2026 had dropped to one hundred and ninety.

Without the scheme, Wangchuk is selling at a loss. He has roughly 192 metric tonnes of unsold pork in cold storage at the regional cooperative — partly his own, partly held in common with three other farmers. His loan repayment to the Bhutan Development Bank, due in June, is going to be late.

The money that would have funded his price guarantee was real. It existed. It had been allocated under the Economic Stimulus Programme — the same envelope that was redirected, weeks earlier, to pay Indian fuel companies.

Wangchuk does not know this. He has not been told. The cooperative officer who informed him about the scheme being put on hold did not explain why. The why involves a chain of decisions made in Thimphu, in Delhi, in Indian Oil Marketing Company accounting offices, none of which routes through Tsirang.

But the chain runs through his farm anyway. Wangchuk’s unsold pork is connected, by a thread of fiscal arithmetic, to the Thimphu motorist filling up his Hilux at Nu 98 a litre — a price made possible by the Nu 101.35-per-litre state subsidy paid from the ESP envelope that would have funded Wangchuk’s price guarantee.

The Tsirang pig farmer is, in effect, paying for the Thimphu motorist’s fuel subsidy. And both are paying the Indian Oil Marketing Company.

This is the closed loop.

Tshering joined the Punatsangchhu-I dam construction team in 2008. She was twenty-four. She had just graduated from the College of Science and Technology in Phuentsholing. The dam was budgeted at Nu 35 billion and scheduled to commission within a decade.

Eighteen years later, in 2026, the dam is still not commissioned. The cost is now estimated at three to four times the original budget. Tshering is forty-two. Her hair has greyed at the temples. Her daughter is in college. She has spent the working years between twenty-four and forty-two on a single project that, when finally commissioned, will be the smaller and more expensive sibling of the next generation of mega-projects Bhutan has already committed to.

The 1,125-megawatt Dorjilung. The 740-megawatt Sankosh. The 540-megawatt Chamkharchhu I. Each will be financed on a similar model. Each will take a generation to build. Each will lock in tariff structures denominated in rupees, debt service first, retained rent last.

Tshering’s career arc is not unusual. It is the model career arc of the Bhutanese hydropower sector. The careers are as long as the dams.

The loop did not start in 2026. It started forty years earlier, with a perfectly reasonable decision about hydropower.

In 1985, Bhutan had cheap rivers and no domestic market. The country had perhaps four hundred thousand citizens, almost all rural, almost all using firewood and kerosene for cooking and lighting. There was no electrical grid to speak of outside Thimphu, Paro, and Phuentsholing. The total domestic electricity demand could have been met by a single small powerhouse.

But the country had something else: gradient. The major rivers — the Wang Chhu, the Punatsang Chhu, the Mangde Chhu — drop from four thousand metres to two hundred metres over distances of two hundred kilometres or less. That kind of gradient, against that volume of water, can be turned into electricity at a fraction of the cost per megawatt that any thermal plant requires.

Bhutan had no money to build dams. India did. India also had a giant grid that was growing at fifteen percent a year and constantly short of generation. The deal almost wrote itself. The Indian Exim Bank would lend the capital. The Indian utility would commit to buying the power. The tariff for each project would be set, at signing, to amortise the loan within the contract period. The tariff would be denominated in rupees. The contract would typically run twenty-five to thirty years.

This was the model that built Chhukha in 1986. Then Kurichhu in 2002. Then Tala in 2007. Then Mangdechhu in 2019. Then Punatsangchhu-I and Punatsangchhu-II, both still under construction nearly two decades after groundbreaking.

In other words: the dams were built to repay India for building the dams.

This was correct in 1985. Domestic demand was negligible. The export model brought in foreign currency. It built Bhutanese engineering capability — civil engineers, electrical engineers, project managers — that the country had no other way of developing at scale. It tied Bhutan’s most important infrastructure to a single neighbour whose interests were predictable and whose grid demand was structurally growing.

Forty years later, here is the result.

Read the full paradox →

Tala began construction in 1997 and commissioned in 2007. Its power purchase agreement with India was negotiated in stages between 1996 and 2005, with the operative tariff structure locked in 2002 — five years before the first kilowatt flowed.

The tariff was set at INR 2.12 per unit, denominated in rupees, with no escalation clause indexed to currency, fuel prices, or comparable spot-market rates.

In 2002, INR 2.12 represented roughly USD 0.044 per unit. In 2026, with the rupee having depreciated meaningfully against the dollar, INR 2.12 represents roughly USD 0.025. In dollar terms, Bhutan now sells Tala’s electricity at about half what the same tariff represented when it was signed.

The contract runs until 2032. Renewal is on the calendar in 2026-27. The terms of that renewal will set the template for Chhukha, Kurichhu, Basochhu, and Mangdechhu renewals across the next decade — together about fifty-seven percent of the country’s export volume.

The decision-makers signing the renewal will be working from a 1990s template against 2026 economic conditions. Whether the template gets updated, or whether the new tariff is set to amortise new loans on a similar twenty-five-year basis, is the single most consequential electricity-policy question on the country’s calendar.

Bhutan now pays its own factories roughly Nu 1.60 per unit for the same electricity that India buys, after debt service, at a net rate to Bhutan of between Nu 0.60 and Nu 1.00 per unit.

That comparison, written out as a sentence, is what it sounds like. Bhutanese factories pay more for Bhutanese electricity than India does.

The reason is that the price India pays sounds higher than it is. The contract price under Tala is INR 2.12 per unit. India pays Bhutan that. But the Indian Exim Bank loan that built Tala requires roughly thirty-five percent of every rupee earned to flow back as debt service. After that flow, what Bhutan retains net is roughly INR 1.20 to 1.40 per unit. Bhutanese factories — the HV1 industrial category — pay a domestic tariff of Nu 1.60 per unit. They pay more than the net rate at which the country exports.

Under the 2025–2028 tariff application the Bhutan Power Corporation filed in December 2025, the unsubsidised cost of supply for Bhutanese households would be Nu 5.73 per unit — the actual consumer rate, set by the government subsidy still to be approved, would be lower. Even at the unsubsidised figure, that is roughly three times what India pays gross, and roughly six times what Bhutan retains net.

Read the full paradox →

Of every Indian rupee Bhutan earns from electricity exports, roughly thirty-five rupees out of a hundred flow back to India as debt service on the Exim Bank loans that financed the dams. The Tala loan was structured as a thirty-year amortisation with a fixed coupon. Debt service on Tala alone runs roughly USD 90 million a year against export earnings of about USD 250 million.

After debt service flows back, Bhutan retains about Nu 0.60 to Nu 1.00 per unit exported — less than what its own factories pay domestically. The same structure applies to Chhukha (now mostly amortised), Kurichhu (still amortising), Mangdechhu (early amortisation), and both Punatsangchhu projects when they eventually commission.

The net retention rate per unit, averaged across the active fleet, is around forty percent of gross. The other sixty percent flows out as debt service to Indian lenders, transmission and operational costs, and grid balancing fees.

This is what makes the export-as-rent argument complicated. The exports are real. The rent is small. The difference disappears into a flow that crosses the border in the other direction.

The Prime Minister, in March 2026, sent the Bhutan Power Corporation (BPC) proposal back to the regulatory authority for re-examination. The pressure had been to recover system costs accumulated over decades — grid expansion, transmission losses, the rural-electrification subsidy, the cost of integrating Punatsangchhu-II (PHPA-II) into the grid. The recovery had to land somewhere. The proposal had landed it on 99.96 percent of customers.

Twenty-three industrial customers in Bhutan consume eighty-eight percent of the country’s electricity. They are the cement plants, the ferrosilicon producers, the steel mills, the calcium-carbide manufacturers — the energy-intensive industries that have grown up around the country’s cheap power as the country’s main industrial sector.

The remaining 99.96 percent of customers — every household, every shop, every clinic, every school, every government office, every restaurant, every farm with electric pumping — consume the remaining twelve percent.

When the BPC proposed doubling the low-voltage tariff, the twenty-three industrial customers were not asked to absorb a comparable percentage increase. They negotiate their own arrangements with the grid operator under separate contracts. The recovery burden, by structural default, was placed on the 99.96 percent who consume twelve percent.

A primary school teacher in Thimphu opens her electricity bill in February 2026. The number is Nu 600. Her rate is Nu 2.66 per unit. By the end of the year, if the proposed revision had passed unsubsidised, the rate would have been Nu 5.73. Her bill would be about Nu 1,290. Her salary, in the meantime, has not moved.

Two hundred kilometres south, in the Pasakha industrial estate, a cement factory consumes about fifty thousand units a month — roughly two hundred times the teacher’s household use. Its rate, under the same proposal, was not changing in the same way.

This was not a glitch. It was the structure.

Read the full paradox →

Bhutanese hydropower is overwhelmingly run-of-river. The country has, with limited exceptions, almost no reservoir storage. The implication is that electricity production tracks the monsoon.

In summer (June through September), when the rivers are running full, Bhutan generates roughly four times its domestic demand and exports the surplus to India. In winter (December through March), when the rivers are low, Bhutan generates less than its domestic demand and imports electricity from India.

The pricing of these flows is asymmetric. Bhutan sells its summer surplus at the contracted PPA tariff — INR 2.12 to 2.55 per unit depending on the project. Bhutan buys its winter shortfall on the Indian spot market — typically INR 6 to 8 per unit, occasionally higher during winter demand peaks.

So Bhutan exports cheap and imports expensive across the same border, in the same year, in the same currency. The seasonal arithmetic costs the country roughly USD 25 to 40 million a year in lost rent that would not have existed if the country had reservoir storage adequate to time-shift production into winter.

PHPA-II and Dorjilung include some reservoir capacity. The legacy fleet does not.

There is one more piece of the loop, and it operates not at the kilowatt level but at the ngultrum level.

In 1974, Bhutan introduced its first national currency, the ngultrum, and pegged it one-to-one to the Indian rupee. The decision was correct. Bhutan in 1974 had almost no trade outside India, no monetary policy capacity, and a tiny banking sector. A floating currency would have been ruinously volatile. The peg has held without interruption for fifty-two years — the longest fixed exchange rate in South Asia.

Its consequence today is that the Reserve Bank of India sets Bhutan’s interest rates. The Reserve Bank of India sets Bhutan’s inflation. The Reserve Bank of India sets the price of money for every Bhutanese borrower and every Bhutanese saver. There is no Bhutanese person in the room when the decision is made.

Most Bhutanese citizens do not know this. They were not taught it in school.

In a closed loop with a counterparty that holds your currency peg, that lent you the capital for your largest infrastructure projects, that buys your largest export at terms it sets, and that sells you your largest import at the prices it negotiates — the leverage runs one way. The loop is not symmetric. It has a direction.

Read the full paradox →

The ngultrum and the Indian rupee have traded at par since 1974. The peg has never been broken. Indian rupees circulate freely as currency inside Bhutan. Cross-border currency movement is unregulated for personal use up to certain thresholds and largely unmonitored above them.

The peg is what makes the hydropower export model possible. Without a stable rupee-ngultrum rate, the rupee-denominated PPA tariffs would expose Bhutan to currency risk that the country has no monetary policy capacity to hedge.

The peg is also what makes the import bill possible. Bhutan’s fuel, food, and machinery imports from India are priced in rupees and paid for in ngultrums at par.

The peg works, in 2026, because both sides accept it. The Reserve Bank of India holds Bhutanese rupee reserves as part of its standard reserve management. Bhutan holds Indian rupees as part of its foreign-exchange reserves.

The asymmetry is that India sets the monetary policy that determines what its rupee is worth, and Bhutan inherits that determination without participating in it. The peg is the deepest piece of the loop. It is also the piece that is hardest to change, because changing it requires Bhutan to develop a monetary-policy capacity it has not historically needed.

The peg deserves a closer look. It is older than every other currency arrangement in the region, and the assumption that it must continue forever has not been examined.

Nepal’s rupee was pegged to the Indian rupee in 1993 — thirty-three years of peg, twenty fewer than Bhutan’s. Sri Lanka has run a managed float since 1977. Bangladesh moved to a managed-float regime in 2003. The Maldivian rufiyaa has been pegged to the US dollar, not to a regional currency, since 2001 — twenty-five years. Malaysia, Thailand, Indonesia, the Philippines, and Singapore have all run floating or basket-managed currencies for decades. Bhutan’s fifty-two-year rupee peg is the oldest fixed-exchange-rate arrangement in South Asia, and one of the oldest in Asia overall. Only the Hong Kong dollar’s peg to the US dollar — since 1983, forty-three years — comes close.

The peg’s longevity is sometimes cited as a feature. It is also worth examining what the peg has cost.

The Indian rupee has depreciated steadily against the US dollar across the entire peg period. In 1974, when the peg was set, one US dollar bought about INR 8. In 1995, about INR 33. In 2010, about INR 45. In 2020, about INR 75. In 2026, about INR 84. Across fifty-two years the rupee has lost approximately ninety percent of its dollar purchasing power.

The ngultrum has lost the same proportion across the same period. Not because Bhutan managed its own economy poorly. Because India did, and the peg passed the cost through.

The peg is, in this frame, a one-way ratchet. It buys Bhutan stability against India — useful for sixty to seventy percent of the country’s external trade. It also buys Bhutan, automatically, a steady erosion against everything that is not India. The erosion is invisible to most Bhutanese because most Bhutanese do not transact in non-rupee currencies often enough to notice. The erosion is very visible in the country’s reserves, in its sovereign-debt position, and in the cost of any international procurement.

Could the peg be something other than the rupee?

The Maldives, twenty-five years ago, made the institutional choice to peg to the US dollar instead of to a regional currency. The arrangement has held through Asian financial crises, COVID, and various regional shocks. The Maldivian economy — tourism-dominant, dollar-denominated revenues — fits the peg, and the peg has been a stability anchor rather than a transmission vulnerability.

Bhutan in 2026 is not the Maldives. The hydropower exports are rupee-denominated. The Indian fuel-import bill is rupee-denominated. The cross-border labour mobility runs through the rupee. Switching the peg to the US dollar tomorrow would impose immediate adjustment costs on every cross-border transaction. The export contracts would have to be renegotiated. The import pricing would need to be reset. The cross-border labour markets would face new friction. The Indian counterparty would, predictably, be unhappy.

But the switch does not have to happen tomorrow. It could be prepared for over a decade and executed at a moment chosen in advance.

Here is the thought experiment, stated plainly enough to debate.

Suppose the Royal Monetary Authority began, today, a fifteen-year transition programme. Dollar reserves accumulated at a faster pace than rupee reserves. New sovereign-debt instruments denominated in dollars where the country has the choice. GMC’s financial-services district operated in dollar terms — a dollar-zone within a rupee-zone country, the way Hong Kong’s dollar has coexisted with the renminbi for decades.

Suppose the trigger condition is announced now, with the date left open. On the day the Indian rupee crosses INR 100 to USD 1 — which the rupee’s secular trajectory suggests will arrive sometime in the early 2030s — the peg breaks. The ngultrum, at that moment, is fixed to the US dollar at 100 ngultrum to one dollar. From that day forward, the ngultrum trades against the rupee at whatever the dollar-rupee rate produces. But the ngultrum holds against the dollar at 100. Indefinitely.

What does that do?

It locks in Bhutan’s dollar purchasing power at the moment of the switch. As the rupee continues to depreciate in the years that follow — INR 110, INR 120, INR 130 to one dollar — the ngultrum does not follow. A Bhutanese citizen’s hundred-ngultrum note always buys one US dollar of internationally traded goods. The country’s reserves, denominated in dollars, do not erode against the global price of imported machinery, technology, education, healthcare equipment, or capital instruments.

The trade-off lands on the rupee side. The ngultrum, once decoupled, strengthens against the rupee. Bhutanese imports from India become cheaper in ngultrum terms. Bhutanese exports to India become, in rupee-denominated terms, more expensive. The hydropower PPAs, still rupee-denominated, need to be renegotiated against the changing relative-purchasing-power picture. The cross-border labour markets face new pricing dynamics. The Indian counterparty has every reason to push back.

And the switch signals, to international capital, that Bhutan has committed to a stable currency anchor that is not regional but global. GMC’s positioning as an international financial centre benefits from this. Foreign direct investment, currently priced through ngultrum-rupee currency risk that is effectively rupee-dollar risk, has a cleaner anchor to price against.

The thought experiment is not free. It is also not technically impossible. The Maldives did it with much less institutional capacity than Bhutan has. Hong Kong’s dollar peg has held for forty-three years through every regional crisis. The institutional question is whether Bhutan should begin building, now, the capacity to make the choice when the trigger arrives.

The country has not had this conversation publicly. The Royal Monetary Authority is, by mandate, a transmission agent for an inherited regime. The Ministry of Finance has not framed the peg as a policy variable. The political discussion has not yet acknowledged that the peg is a choice rather than a constraint.

The conversation is overdue. The rupee will cross INR 100 to USD 1 sometime in the next decade. When it does, Bhutan can either continue the peg — and continue the decline — or break it. The decision to break it cannot be made on the day it needs to be made. It has to be prepared for, with reserves accumulated and instruments restructured and institutional capacity built, for ten to fifteen years in advance.

The window for that preparation, in 2026, is open. Whether the country uses it is the question.

Read the full paradox →

Every Power Purchase Agreement Bhutan has ever signed with India is denominated in Indian rupees. None of them — Chhukha 1986, Tala 2007, Mangdechhu 2019, Punatsangchhu-II 2026 — contains a single clause referencing the US dollar. The rupee has lost roughly eighty-five percent of its US dollar value since the Chhukha plant was commissioned. The contracts continued to deliver rupees at the agreed rate. The rupees, measured against what a Bhutanese could once buy with them overseas, kept shrinking.

Add up the gap across every operating plant, year by year, project by project, and the cumulative loss reaches roughly USD 1.85 billion. That is roughly half of what the country produces in a year. It does not appear in any official report, because the report is written in ngultrum, and the ngultrum is still pegged at one-to-one with the rupee. The accounting is honest. The purchasing power is gone.

Read the full paradox →

Bhutan’s hydropower plants are mostly run-of-river. They generate what the rivers give in the hour they give it. In December, the rivers give roughly one-third of what they gave in August. Domestic demand does not fall to match. The gap is bridged by imports from the Indian Energy Exchange, purchased on the day-ahead auction at whatever the All-India market clears at — typically four to six rupees a unit during the dry months.

The country’s largest domestic industrial customers — the cement plants, the ferro-silicon furnaces, the cryptocurrency-mining operations — pay the Bhutan Power Corporation a tariff that has been held at one rupee sixty paise per unit since 2022. The same electricity. The same wires. The same hour. Bought at five rupees, sold at one rupee sixty. The Bhutan Power Corporation absorbs the difference. The difference, summed across roughly eleven hundred gigawatt-hours of imports in 2025, comes to between thirty-three and forty-four million US dollars a year. The number does not appear as a line item in any official report. It sits inside the utility’s compressed margin, the dividend to the Druk Holding and Investments group that is slightly smaller than it could have been, the tariff increase that the residential consumer is being asked to absorb on a household electricity bill. The Indian electrons get paid for. The Bhutanese factory pays a third of what they cost. The remainder is paid silently.

Pull back from Wangchuk and the schoolteacher and the cement factory. What does the closed loop look like at the level of the country’s accounts?

In 2023, Bhutan earned roughly USD 208 million from electricity exports. In 2025, it spent roughly USD 442 million on fuel imports. Net energy position: minus USD 234 million.

The 2025 figure is partly distorted — Punatsangchhu-II commissioned in late 2024, and 2025 electricity exports were higher than 2023’s. A like-for-like comparison after PHPA-II ramps up will look better. But the structural pattern is unmistakable. The cleanest electricity exporter in South Asia is, on net, a fossil-fuel-deficit country with its largest neighbour.

And the trajectory is unfavourable. As domestic vehicle ownership grows — Bhutan has approximately 9.6 vehicles per kilometre of road, denser than India’s national average — the fuel bill grows. As the cement, ferrosilicon, and steel industries expand, domestic electricity demand grows, and the export volume net of domestic use either holds or declines. As Punatsangchhu-I finally commissions, eventually, two decades after groundbreaking, the debt service on its rupee-denominated loan begins to absorb its export earnings before the earnings reach the Bhutanese exchequer.

The country’s hydropower economy is not a market in the conventional sense. It is a closed loop with a single counterparty that captures most of the rent before it leaves the border.

Read the full paradox →

In May 2026, the World Bank signed a loan to Bhutan for the Dorjilung hydropower project — 1,125 megawatts, USD 515 million. The per-installed-kilowatt cost works out to approximately USD 458 per kilowatt.

Global hydropower capex benchmarks for large projects in mountainous terrain run USD 900 to USD 1,500 per kilowatt. Dorjilung at USD 458 is roughly half the global average.

Either Bhutanese geology is extraordinarily favourable, the international competitive bidding produced unusual cost efficiency, or the project will experience cost overruns later. The third possibility is the most likely.

Punatsangchhu-I started at Nu 35 billion and is now budgeted at roughly Nu 100 billion. Punatsangchhu-II started at Nu 71 billion and is now budgeted at over Nu 110 billion. If Dorjilung’s true cost ends up at global benchmark levels — USD 900 million to USD 1.2 billion — the additional USD 400 to 700 million will come from somewhere. More bilateral debt. Supplementary multilateral lending. Sovereign guarantees. Equity infusion. Whichever source, the country’s debt stock grows beyond what was committed at signing.

Cleanest export. Lowest net price. Single buyer.

Dirtiest import. Highest pump price. Same counterparty.

Aid to bridge the gap. Same counterparty.

The dams were the right decision. The loop is the consequence.

Read the full paradox →

The simplest way to see the PPA economics is this. Bhutan sells the same kilowatt to India at INR 2.12 (Tala) to INR 2.55 (Mangdechhu) per unit. Bhutan buys the same kilowatt back from India at INR 6 to 8 per unit during winter shortfalls. Bhutan’s own domestic industrial customers pay INR 1.60 per unit. Bhutan’s own domestic households pay INR 2.66 per unit, headed toward an unsubsidised Nu 5.73 if the 2025–28 proposal had passed.

Four prices for the same commodity. Four counterparties. Four directions of flow.

Multiplied across the seven thousand gigawatt-hours of annual HV1 consumption, the implicit subsidy from foregone Punatsangchhu-II reference-tariff revenue to those twenty-three industrial customers runs to about USD 281 million a year.

Read the full paradox →

The chapter has named four separate structural leaks in the closed loop with India. Each was authored by a different decision, in a different decade, by a different ministry. Each is defensible on its own terms.

The first: every hydropower contract Bhutan has ever signed is denominated in Indian rupees with no clause referencing the US dollar. The rupee has lost roughly eighty-five percent of its dollar value since the first dam came online. The cumulative loss, measured in dollars Bhutanese reserves can no longer buy, is about USD 1.85 billion.

The second: India pays its smaller currency partners in the Common Monetary Area a yearly seigniorage transfer for the rand circulating in their economies. Bhutan has shared a currency with India for fifty-two years and has never asked for the equivalent. The unclaimed flow is about USD 10 million a year.

The third: Bhutan’s largest domestic industrial customers pay a per-unit electricity tariff lower than the cheapest tariff at which Bhutan sells the same electricity to India. The implicit subsidy is about USD 281 million a year.

The fourth: every winter, when the rivers run low, the Bhutan Power Corporation imports electricity from the Indian spot market and resells it to those same domestic customers at a third of the import price. The annual cash loss is about USD 33 to 44 million.

Four leaks. Four ministries. Four decisions, each one taken in its own decade, each one for its own reason. Together they describe a single direction. The ngultrum’s one-to-one peg with the rupee makes the direction invisible in the country’s own accounting, because the ngultrums always balance. The dollars do not.

The country is not a market. It is a logistics hub for a flow that mostly runs through it.

Read the full paradox →

A taxi driver pulls into a Babesa fuel station on a May 2026 evening. His Mahindra Bolero is on near-empty after a twelve-hour shift. He hands over six thousand and fifty-three ngultrum in mixed notes — the cost of sixty litres at the consumer pump price of one hundred and zero point eight nine ngultrum per litre. The transaction at the counter is simple. The receipt shows the volume, the price, the total. The driver drives away.

What does not appear at the counter: the Royal Government of Bhutan has, in parallel, paid the same fuel station one thousand three hundred and eighty ngultrum on his behalf. Twenty-three ngultrum per litre, the subsidy bridge between the cushioned pump price he saw and the actual landed cost the Indian Oil Marketing Company billed the Bhutanese distributor — one hundred and twenty-three point eight nine ngultrum per litre. There is no receipt line for the subsidy. No notification to the driver’s phone. No annual statement. No budget item on any household-facing document. The driver has no way of knowing how much Bhutan paid the Indian distributor to put diesel in his tank. He thinks he paid one hundred a litre. He did. And so did the country, on his behalf, again.

On the seventeenth of April 2026, the gap was at its most extreme. The actual landed cost of diesel that day was one hundred and ninety-nine point six six ngultrum per litre. The consumer pump price was held at ninety-eight point three one. The government wrote a cheque for one hundred and one point three five per litre to bridge the gap. The government paid more per litre than the consumer that day. By May, global pressure had eased and the bridge had narrowed to twenty-three ngultrum a litre, but the structural mechanism had not changed. It will not change for as long as the country’s fuel benchmark is the speculative Arab Gulf Gasoil index rather than crude oil, and for as long as the Indian Oil Marketing Companies decline, as they have for the four years since Bhutan first requested it, to provide an internal break-up of the prices they charge their largest peg-pegged customer.

The Prime Minister stated, in the May 2026 National Assembly session, that the government had spent one point four five billion ngultrum on fuel subsidies to date. The petrol subsidy was lifted on the sixteenth of May. The diesel subsidy continues at twenty-three ngultrum a litre. At the current consumption, the annualised burn-rate is approximately five to seven billion ngultrum per year. Twenty-five to thirty-five percent of the annual Government of India grant that anchors the country’s fiscal stability.

The structural feature of the current mechanism is that it is invisible by design. A pump-subsidy routes money in a way that no household experiences as a deposit. A cash transfer of equivalent value — paid monthly into every household’s bank account through the existing mobile-money and National Payments Corporation of India rail that already moves billions of ngultrum in remittances across the country every month — would route the same total amount in a way that every household experiences as a benefit. The fiscal cost is identical. The political result inverts.

Indonesia under President Joko Widodo removed its gasoline subsidy in 2015 and routed the equivalent fiscal space into infrastructure and conditional cash transfers; the reform held. India in 2014 converted the LPG cooking-gas subsidy from an invisible pump subsidy to an Aadhaar-linked direct bank-account transfer; the conversion identified three and a half crore duplicate or ineligible beneficiaries and saves the Indian exchequer approximately fourteen thousand crore rupees a year. Saudi Arabia in 2018 launched the Citizen’s Account, a monthly cash deposit that offsets fuel and electricity price reform. Each of these reforms is studied as the textbook case of visible cash beating invisible pump on every dimension: fiscal cost, distributional fairness, price signal, political tractability. Bhutan has not yet made the move. The one point four five billion ngultrum spent to date could have funded the design and first-year payouts of the equivalent transfer scheme.

Read the full paradox →

The diesel subsidy described in the previous sidebar — twenty-three ngultrum per litre, one point four five billion ngultrum spent through May 2026 — has an annualised burn rate of approximately five to seven billion ngultrum per year if continued. That sum is twenty-five to thirty-five percent of the annual Government of India grant. It is structurally unsustainable.

The fiscal arithmetic forces an exit. The forward question is whether the exit is designed or chaotic.

Consider what the same five billion ngultrum per year buys, if redirected. An electric bus of the Yutong or BYD class — Chinese-import grade, suitable for the East-West highway and the mountain valleys — costs roughly thirty million ngultrum. The same five billion ngultrum that currently disappears into one year of diesel tank-fills could instead deliver one hundred and sixty-seven electric buses. Or two hundred and fifty fast-charging hubs across the country’s main highways. Or one hundred kilometres of full mountain-road reconstruction. Or one BRT corridor between Babesa and Norzin Lam, with money left over.

Across five years, twenty-five billion ngultrum of redirected subsidy buys: a working Thimphu-to-Phuentsholing BRT line, a fleet of approximately five hundred electric buses serving Thimphu and Paro and Phuentsholing and Gelephu, one thousand kilometres of mountain-road upgrade in the monument-dense valleys, and the charging-and-EV-incentive infrastructure that would accelerate private-vehicle electrification beyond the ninety-nine-EV-plus-forty-five-e-bus government-fleet procurement already announced.

That same government-fleet announcement, made the same May 2026 week as the one point four five billion ngultrum subsidy disclosure, costs roughly four to five billion ngultrum total. The single year of subsidy that the country has already committed could have funded the entire announced electrification programme — with money to spare for road maintenance.

The per-vehicle math is the regressive transfer made explicit. Distribute the five billion ngultrum across the roughly fifty-five thousand active diesel vehicles in the country and the implicit subsidy is about ninety thousand ngultrum per diesel vehicle per year. The household with a Bolero or a diesel pickup is receiving the cash-equivalent of a Nu 7,500 monthly salary supplement — invisibly, through the pump. The household without a vehicle, which is roughly ninety-three percent of Bhutanese households, receives zero. The subsidy is collected from all — through foregone tax revenue, through diverted Economic Stimulus Plan funds, through future fiscal-space pressure on the next budget — and benefits roughly seven percent of the population. It is the most regressive fiscal flow currently active in the country.

The asset-versus-consumption arithmetic is more striking still. A one-thousand-three-hundred-and-eighty ngultrum subsidy poured into a sixty-litre Bolero tank burns once and is gone. The same money contributed to an electric bus that runs for ten to twelve years carries that money through approximately four hundred and eighty thousand passenger-trips per bus over its useful life. The per-rupee social return on infrastructure investment is roughly twenty to forty times the per-rupee return on subsidised fuel-combustion. Same fiscal cost. Different physics.

Singapore made this choice in the 1970s. The Mass Rapid Transit system was built precisely because the government refused to underwrite private-car ownership; today Singapore has the highest per-capita transit ridership in Southeast Asia and one of the most expensive private-car-ownership markets in the world. Norway made this choice in the 2010s. EV transition was funded by oil-revenue redirection plus aggressive private-car disincentives — tolls, congestion fees, parking restrictions; today EVs are roughly ninety percent of new-car sales. Bogotá made this choice in the 2000s. TransMilenio BRT was built without national fuel subsidy, financed through bond issues and a dedicated transport-fuel-tax surcharge; today it carries 2.4 million passengers a day. Indonesia made this choice in 2014. The gasoline subsidy was removed; the freed fiscal space went into rural infrastructure and conditional cash transfers; the modal shift to public transport was visible by year three.

Bhutan has not yet made the choice. The current policy is to maintain the fuel subsidy at five to seven billion ngultrum per year while the transport-infrastructure pipeline is funded at a fraction of that figure, while vehicle stock grows five to eight percent per year, and while public transport remains effectively non-existent outside the largest cities.

The structural sequencing — for the choice to be politically tractable rather than fiscally forced — would be a twelve-month phased subsidy removal in four quarterly steps of five ngultrum per litre, a Transport Transformation Fund ring-fenced in budget law and published monthly, a Year-1-to-2 spending priority of road maintenance and EV purchase incentives, a Year-3-to-5 priority of the five-hundred-bus build-out, and a Year-4-to-6 priority of the BRT corridor. Critically, every household would receive a monthly Nu 1,500 transport-allowance deposit for the first twenty-four months of the subsidy removal — a visible deposit hitting bank accounts before the higher pump price hits petrol stations, addressing the political-asymmetry problem the previous sidebar described.

This is the prescription that closes the chapter’s argument. The country is currently choosing to spend its single largest household-facing fiscal commitment on keeping today’s diesel Boleros running cheaply. The same money, redirected, would build the bus network that the ninety-three percent of Bhutanese households who do not own vehicles have been waiting for. The choice is between burning the money in fuel tanks and building the network that would make the fuel tanks less necessary.

Wangchuk drives his pickup back to his farm in Tsirang in late May 2026. The road from Phuentsholing to Tsirang is one of the better-paved ones in the country, partly because it serves the southern industrial belt. He passes three cement plants, two ferrosilicon factories, and a fuel station where the pump price is Nu 98 a litre — a price he can read on the sign without having to slow down.

He thinks about the pump price. He thinks about his pigs in cold storage. He does not connect the two.

Nobody has told him they are connected. The only person who could have explained the connection — the cooperative officer who informed him about the price-guarantee scheme being put on hold — did not explain it because the cooperative officer does not know either.

The loop runs through Wangchuk’s farm. The loop runs through the schoolteacher’s apartment. The loop runs through the cement factory across the road. None of them sees the others. All of them are part of the same arithmetic.

The same forty-year arc that built the closed loop also built something else: a labour market, a school system, an English-language pipeline. By 2026, that pipeline has produced something the 1985 decision did not anticipate.

It has produced about forty thousand Bhutanese in Australia.

Read the full paradox →