IMAGE VIA RONB/FACEBOOK
IN THE mid-1990s Nepal’s public capital stock stood at about three-quarters of its GDP. By 2019 it had sunk to 54%. A nation needing more bridges, roads and irrigation canals had somehow ended up with fewer of them. The figures tell a story of systemic failure, punctuated by brief spurts of progress.
The earthquake of 2015 forced capital spending upwards, peaking at 11.4% of GDP in 2021. Yet the discipline did not stick. By 2024 public investment had slumped back to 7.9% of GDP. This is inadequate for a country the World Bank reckons must spend 10-15% of national income annually for a decade just to plug existing infrastructure gaps.
Worse, even the money allocated is often left on the table. Federal capital budget execution, which averaged 74% in the years before the pandemic, has fallen to around 60%. In 2022 foreign-funded projects spent less than two-fifths of what they had been given.
These administrative failures amount to losses equal to 5% of GDP a year over the past four years. For a country where forbidding geography makes connectivity the difference between subsistence and prosperity, such waste is existential.
The gap between planning and execution runs through every stage of a project. Take the National Project Bank, set up in 2020 to bring order to deciding what to build. In theory projects must pass a proper appraisal. In practice, many still find their way into the budget through back doors.
When the National Planning Commission updated the guidelines in March 2025, it made sensible improvements. But the rules still do not specify how feasibility studies should be done. The same requirements apply whether it is a small irrigation scheme or a trans-Himalayan highway.
If upstream choices are flawed, downstream execution is paralysed. Two bottlenecks—tree cutting and land acquisition—can halt projects for years. Securing permission to fell trees typically consumes 22 to 24 months. The process requires a manual census of every tree, conducted by officials counting trunks and estimating volumes by hand.
That happens after the Environmental Impact Assessment has already made its own survey. When the two counts diverge, a supplementary EIA is ordered and the clock resets. Digital tools like LiDAR surveys, which could map forests in hours, go unused.
Land acquisition takes two to three years on average. The core problem is valuation. Nepal maintains at least three separate land valuations: one for transfer tax; one for property tax; and one for government compensation. None reliably reflects market prices. The minimum valuation used for compensation understates true values, inviting litigation.
Once projects finally break ground, new torments await. Cash management practices concentrate spending in the final quarter. Despite rules allowing full-year commitments, the Treasury releases funds quarterly and rations cash when revenues fall short.
For foreign-funded projects, reimbursements for late-year spending arrive only after the government has spent its own money in the first half. If domestic co-financing falls short, the pipeline runs dry.
Budget reallocations, known as virements, offer a theoretical escape valve. But moving money between projects requires dealing with regulations that seem designed to discourage the attempt. A routine Rs1bn transfer from the transport ministry’s contingency to road maintenance took 59 days, 25 steps, five memoranda and approvals from eight divisions. Line ministries have formal discretion but lack the trained staff to exercise it. In practice everything goes through Finance and everything slows down.
Procurement provides the final act of self-sabotage. World Bank data show Nepal taking the longest in South Asia to award contracts: 231 business days on average, compared with a regional average of 192. Section 59 of the Public Procurement Act, which makes defaulting contractors liable for the full outstanding contract amount regardless of work completed, deters serious firms from bidding and guarantees litigation.
The government has a blueprint. The March guidelines, if enforced, could impose discipline. The tree clearance process could be streamlined by merging the two surveys. Digital tools could replace manual tree-counting.
The Land Acquisition Act of 1977 could be amended to mandate market valuations. Cash releases for foreign-funded projects could be decoupled from the reimbursement cycle. Procurement thresholds could be raised; abnormally low bids rejected; contract termination clauses restored to their original form.
Whether Nepal’s government can follow them is another question. The blueprint exists. The building does not. ■
