By Lakshman Pandey

Banking News – Nepal’s banks have been required to screen loans for flood and disaster risk since 2022. The flood in the month of Bhadau on the Bhote Koshi shows how far a completed checklist is from a managed risk.

On the morning of Bhadau 10, a flood came down the Bhote Koshi from the Tibetan side of the border, carrying boulders, mud, and debris. By the time it had passed through Rasuwa and Nuwakot, it had hit 13 hydropower projects with 759 MW of capacity, seven of them operating and six under construction, according to the government’s rapid damage assessment. About 430 MW of generation dropped off the national grid. The assessment puts physical damage to hydropower at about Rs 120 billion and the cost of rebuilding the energy sector at Rs 390 billion.

The human toll comes first. By Ashoj 7, official records counted 1,451 bodies recovered and more than 6,000 people still missing, figures the government treats as provisional. More than 800 of the missing were workers at hydropower sites.

For bank boards, the flood also raises a narrower question that will not go away once the river settles. The first count for lenders was almost reassuring. The Nepal Bankers’ Association’s preliminary tally, reported on Bhadau 16, found four bank-financed hydropower projects damaged. Loans at risk were put at Rs 6.5 billion in one account and Rs 7.5 billion in another. Commercial banks have lent more than Rs 500 billion to the energy sector, about 9.5 percent of all their loans, so the damaged exposure is between one and two percent of that book. The heavier losses fell on projects financed by the government, the Nepal Electricity Authority, state pension funds, and foreign lenders.

That arithmetic should comfort no one. The association counts eight or nine bank-financed projects in the affected districts, depending on the report, and four of them were hit in a single morning. A fifth, carrying Rs 9.53 billion in loans, was not damaged but cannot sell its power because the flood destroyed its transmission line. Banks have a further Rs 2.5 billion lent to households and businesses in the flood zone.

Nor did the risk wait for another monsoon. On Ashoj 25, less than seven weeks later, a flash flood on the Budhi Gandaki in Gorkha damaged four hydropower projects under construction, with a combined capacity of 412 MW. It swept away the main camp of the 341 MW Budhi Gandaki project, along with bridges, access roads, and equipment at three smaller ones. Early analysis pointed to a landslide that blocked the river and then gave way, not to rainfall. The same river produced a similar flash flood in Baisakh 2081. How much bank money sits in those four projects has not been reported. Two rivers in seven weeks make this a recurring risk, and a live one. The next river or incident may hold a larger share of the domestic loan book. So the question every credit committee now faces is a simple one. Should the original loan assessment have asked whether these sites were exposed to exactly this kind of event?

The Rule Was Already There

The answer is yes, and the rulebook already said so. In Falgun 2078, Nepal Rastra Bank revised its Environmental and Social Risk Management guidelines. The revision told banks to assess climate risks such as floods and droughts before lending, added a climate checklist to be completed before a loan is issued, and required banks to check whether a project site lies in a disaster-prone area.

Nor was the hazard obscure. A World Bank-funded study published after the 2072 earthquake warned that a large glacial lake outburst could sweep away every plant built in cascade along trans-Himalayan rivers such as the Rasuwa Bhote Koshi. Floods and landslides have damaged hydropower plants in every significant monsoon since 2078, including 26 projects that year and 30 in 2080.

So the failure is not a missing rule. A bank could complete the checklist, file it, pass its inspection, and still have no idea what an outburst flood would do to its loan book. Compliance and exposure sat side by side.

Where the Assessment Falls Short

There is no method. ICIMOD’s review after the flood found that Nepal’s rules set requirements without a way of meeting them. They offer no guidance on climate models, no standard way to quantify outburst flood risk, and no benchmark for adequate screening. A checklist can be ticked without any hazard modeling behind it.

The design basis looks backward. Plants are typically built to withstand flows several times their design estimate, calculated from historical records. Banks then rely on the developer’s own feasibility study. Neither captures an outburst flood, and neither allows for a climate that is changing the record.

Risk is judged one loan at a time. Most banks address climate at the level of the single transaction, and portfolio-wide assessment is rare. Yet the Trishuli corridor holds about a dozen plants on one river system. One event reaches many borrowers at once, and a file-by-file review cannot see that.

Guarantees stand in for analysis. Projects are financed mainly on personal and corporate guarantees. A guarantee may help recovery, but it does not price the risk or shape the design, the insurance, or the covenants.

The mandate pulls the other way. Banks must place 10 percent of their loans in energy. One banker told Onlinekhabar that banks lend to the sector despite the risks because of that requirement. The regulator is asking for volume and caution in the same breath.

The old book was never revisited. Most projects in the corridor were built or started over the past 20 to 25 years, long before the 2078 checklist. Assessment happens once, at sanction, while the conditions upstream keep changing.

Rasuwagadhi shows the cost. The 111 MW plant was damaged by a flood in Asar 2082 and repaired at the same site. Kumari Bank lent Rs 2 billion and Nabil Bank Rs 498 million for the reconstruction. Fourteen months later the plant was destroyed. Lending into a rebuild is a second credit decision, and it is the moment to ask what has changed on the river.

The hazard starts across the border. This flood entered from Tibet, where Nepali banks and developers have little data and less warning.

Eight Questions for the Risk Committee

A board does not need a hydrologist to find out where it stands. It needs to ask eight questions and insist that the answers come from the loan files, not from the policy manual.

1. Governance. Does our risk appetite statement name physical climate risk, and does this committee see it reported?

2. Site screening. Does every hydropower file name the hazards at the site and upstream of it, using data that did not come from the borrower?

3. Method. Do we have a written way of assessing those hazards that goes beyond the regulator’s checklist?

4. Decision. Has a climate assessment ever changed a price, a covenant, or a design, or led us to decline a loan?

5. Concentration. Do we know our total exposure on each river corridor, including consortium shares and loans against hydropower stocks?

6. Insurance. Do our borrowers’ policies cover outburst floods at replacement cost, and could the insurers pay if several plants claimed at once?

7. Monitoring. Have we reassessed the loans we approved before 2078, and do we review them when conditions upstream change?

8. Capital. Have we tested what the loss of a whole corridor would do to our provisions and our capital?

A bank that can answer only with a completed checklist is compliant. A bank whose answers show up in pricing, limits, and declined proposals is managing the risk. Most of the sector, on the evidence available, is closer to the first.

Before the Next Monsoon

Three steps need no new regulation and little money. Map the hydropower book by river corridor. Read every insurance policy for what it covers and who gets paid. Flag every financed site that sits downstream of a known dangerous glacial lake. Any bank can do all three in months.

The harder work follows. It means a written assessment method, corridor limits that bind new lending, a review of the pre-2078 book, and a stress test built around the loss of a cascade.

The regulator has its own part. A checklist without a method invites box-ticking, and a lending quota without a risk standard invites worse. ICIMOD and the Water and Energy Commission Secretariat are developing Nepal’s first climate-resilient hydropower guideline, which could give banks the common reference they lack.

None of this would have predicted Bhadau 10. The Asian Development Bank’s own climate assessment for Upper Trishuli-1, one of the damaged projects, had rated the risk low. If a multilateral lender with far deeper resources could misjudge the river, no Nepali bank should expect to forecast it. Forecasting is not the test. The test is whether a bank knows where its exposure sits, has priced and insured it, and has limited how much of it rests on one river.

The Bhote Koshi answered that question for 13 projects in a single morning. The Budhi Gandaki asked it again in Ashoj, without waiting for rain. The next monsoon is about seven months away.