SOMETIME IN THE mid-2000s, a Nepali migrant worker in Qatar wired his first month’s wages home to his family in Sindhuli, a poor hill district. His wife deposited the money in the nearest branch of a commercial bank, where it earns, today, 2.93% per year. Every month the money sits in that account, it buys a little less rice thanks to inflation. The bank meanwhile lends it out at 7.12% and pockets the spread. This is the business model of an entire sector.
Nepal’s banking system holds deposits equivalent to 126% of GDP—a ratio that would be respectable in Germany and extraordinary in a country whose GDP per capita sits below $1,500. That mountain of savings has not been built through thrift or industrial dynamism.
It has been assembled almost entirely from remittances: the roughly $10bn or so sent annually by the 3m or so Nepalis working in the Gulf states, Malaysia, Japan and South Korea. Their labour, performed in construction sites and factories far from the Himalayas, has created one of the most deposit-rich banking sectors, relative to the size of the underlying economy, in Asia.
What Nepal’s 54 licensed banks and financial institutions have done with that windfall is, by any fair accounting, close to nothing. Total credit to the private sector stands at 95% of GDP, leaving a gap of 31 percentage points between what has been deposited and what has been lent.
The credit-to-deposit ratio at commercial banks sits at 74%, meaning that for every 100 rupees placed with a bank, only 74 rupees find their way back into the economy as loans. The other 26 rupees sit in government securities and liquid instruments, earning safe and undemanding returns for shareholders without troubling anyone with the messy business of credit underwriting.
The interest rate spread, the gap between what banks pay depositors and charge borrowers, stood at 3.55 percentage points in January, according to Nepal Rastra Bank data. That figure has barely shifted in five years. Between April 2020 and January 2026, it fell from roughly 4.6%, to 3.55%: less than one percentage point of compression across a full rate cycle that saw deposit rates peak above 10% and then collapse back below 4%.
In a real competitive market, a rate cycle of that magnitude should have shaken the spread considerably. It did not, because the 20 commercial banks that dominate the financial sector did not compete. They moved in lockstep.
The pattern is visible in the data and remarked upon by economists in Kathmandu. When one bank adjusts its fixed deposit rate, others follow within days: not in response to market signals, but as if someone had sent a circular. The weighted average deposit rate across all commercial banks converged to 3.56% in January, with variance across institutions so low it strains coincidence. Twenty banks. One price. The word for this, in other contexts, is cartel.
The depositors who subsidise this arrangement have no meaningful exit. Nepal has no accessible retail bond market. Its stock exchange, NEPSE, is volatile enough to have burned a generation of retail investors. Property in Kathmandu, the one alternative store of value that middle-class families might have contemplated, is now priced beyond most of them.
So the remittance money arrives, goes into the nearest savings account and earns its negative real return most years while the bank decides what to do with it.
What the banks have decided to do, largely, is lend to property developers in Kathmandu. During the years 2019 to 2022, commercial banks extended enormous volumes of credit to apartment projects, commercial real estate and construction firms clustered in the valley. The logic was legible: urban property, however richly valued, offered collateral. Agricultural lending in the hill districts offered goats. The banks chose apartments.
The consequences are now visible in the non-performing loan statistics. Bad debts across the system have risen from 1.3% of total loans in mid-2021 to 5.42% by January, a more than fourfold deterioration in under five years. For Class C finance companies, which serve smaller borrowers and operate with slimmer capital buffers, the NPL ratio has reached 11.86%. Construction and real-estate lending sits at the heart of this deterioration. Banks discovered condominiums during the boom, and are now discovering what happens when property prices correct and interest rates spike.
The response to soaring bad debt has been to tighten lending further. Banks that once poured credit into Kathmandu real estate have retrenched across the board, reducing the credit-deposit ratio and holding more assets in government paper. The businesses as well as farmers and entrepreneurs who had no access to credit at the peak of the lending cycle have even less now.
That is a classic second-order effect of bad banking: the recklessness is concentrated in one sector, but the punishment is distributed across all of them.
Agriculture makes the point painfully. Two-thirds of Nepal’s population depends on farming, directly or not. Agricultural and forestry lending has stagnated and is now in slight decline in nominal terms. Commercial banks participate in the sector primarily because the Nepal Rastra Bank requires them to: a deprived sector lending quota mandates that a minimum share of loans go to underserved borrowers. The aggregate compliance figure—5.29% for commercial banks—looks technically satisfactory. It is mostly fiction.
Banks meet the quota by wholesale on-lending to microfinance institutions, which then lend to actual rural borrowers at considerably higher rates. The bank books the compliance, avoids the risk and never meets the farmer.
Geography makes the exclusion concrete. Nepal’s 6,511 bank branches are distributed by profit potential rather than population need. The Kathmandu Valley, home to less than 15% of the country’s people, holds a wildly disproportionate share of branches, ATMs and loan officers. In Bhojpur district—in the eastern hills, reachable by road from Dharan but not easily—there is one bank branch for every 6,317 people. In wealthier parts of the valley the ratio is closer to one per 2,000. The Nepal Rastra Bank publishes this data. It has not acted on it.
The regulator’s posture throughout this story is that of an observer. The NRB mandates lending quotas and has pushed several rounds of bank consolidation over the past decade. It has done essentially none of the things that would fix the structural problems described above. It neither does public stress testing of the system’s most fundamental vulnerability.
That vulnerability is the remittance dependency itself. Nepal’s entire deposit base, and therefore its entire banking system, rests on the continued willingness of Gulf and Asian economies to employ Nepali workers in huge numbers.
Saudi Arabia’s Vision 2030 and the UAE’s Emiratisation programme are explicitly designed to replace low-skilled foreign labour with nationals. Automation is steadily reducing demand for semi-skilled manufacturing workers in Malaysia. South Korea’s employment permit scheme, a major source of relatively well-paid Nepali migration, is subject to bilateral renegotiation. None of this is secret. None of it features in the NRB’s published stress scenarios.
A 25% reduction in remittance flows, a plausible outcome over the next decade not a tail risk, would materially shrink the deposit base; tighten liquidity across the system; and bring sudden pressure on a spread that has been maintained through captive deposits rather than real banking skill.
Capital adequacy currently sits at 12.57% of risk-weighted assets, above the minimum. But capital fund growth slowed to near zero in recent months even as bad loans kept rising. The comfortable numbers are backward-looking. They describe a system that has not yet faced a real test.
Nepal’s banks have built not a financial system but a remittance collection service. They take deposits from the families of migrant workers at rates that erode purchasing power; hold a quarter of those deposits in safe government paper; lend the rest primarily to urban property and commercial borrowers; and distribute the resulting profits to a narrow shareholder base. The rural household that sent money to the bank in Sindhuli has, in a very precise sense, subsidised the Kathmandu condominium market.
The toll-booth analogy is inexact in one respect: toll booths at least maintain the road. ■
