Nepal’s new average-bid procurement system could undermine genuine competition, encourage collusion and turn public contracting into a lottery, particularly given the construction sector’s existing weaknesses. The government should reconsider the model and adopt the better-tested approach of screening abnormally low bids while continuing to award contracts to the lowest responsive bidder.
What actually changed
On July 16, 2026, the Public Procurement Monitoring Office confirmed that the Second Amendment to the Public Procurement Act had been published in the Nepal Gazette and taken legal effect. It is the most consequential rewrite of Nepal's procurement law since the original 2007 Act, and it followed an unusual path: the government issued it as a presidential ordinance in April, using its fast-track constitutional authority to bypass the slower two-house legislative process, before Parliament formally converted it into an Act in July.
The single change likely to matter most for how public money is spent is this: Nepal has abandoned the lowest-evaluated-responsive-bidder rule that governed public works contracts for nineteen years. In its place is an average-bid formula. The contracting agency now totals the remaining submitted bids, discards any bid more than 30 percent below the engineer's cost estimate before doing the sum, calculates the average of what remains, and awards the contract to whichever bidder is numerically closest to that average — provided the winning bid still sits at or below the official cost estimate.
The Federation of Contractors' Associations of Nepal (FCAN) welcomed the change publicly, framing it as a fix for years of quality shortcuts and stalled projects caused by reckless underbidding. That framing is not wrong about the underlying problem. It is wrong about the fix.
The arithmetic problem, worked through
The mechanism that sounds neutral — “average” — has a reassuring, moderate ring to it. Run the numbers and the neutrality disappears.
Take a contract estimated at NPR 100 million. Five firms bid. One of them, Firm A, genuinely has lower overheads, a shorter haul distance, owned rather than rented machinery, and bids honestly at NPR 72 million — a legitimate 28 percent saving for the state. The other four firms, who have bid against each other in the same district, quote NPR 96, 97, 98 and 99 million: close to the estimate, comfortably profitable, and mutually predictable.
Under the old rule, Firm A wins at NPR 72 million, and the government pockets the difference — subject to standard checks that it can actually deliver at that price. Under the new rule, the average of all five bids is NPR 92.4 million. The bid closest to that average is NPR 96 million. Firm A, the efficient bidder, loses. The government pays NPR 24 million more than it needed to, for the same scope of work, awarded to a firm that did nothing to earn it beyond guessing where the pack would cluster.
This is not a contrived edge case. It is the textbook description of how average-bid auctions are expected to behave once bidders understand the rule, published in a 2024 study of bid-rigging patterns in Crime, Law and Social Change. When bidders anticipate the format, the theoretical outcome converges toward everyone bidding at or near the reserve price, the government pays close to the maximum, and the eventual winner is picked essentially at random rather than on merit. Genuine cost efficiency stops being an advantage. It becomes a liability, because it pulls a firm away from the number that wins.
A method with a research paper trail, not a blank slate
NRB issues Rs 25 billion in bonds, invites bids from BFIs
Nepal is not the first country to try this, and that is precisely the problem. The results elsewhere are already published, and they are not encouraging.
Italy ran local public works procurement on an average-bid-style formula for decades, generating one of the largest empirical datasets in procurement economics. Bocconi University economist Francesco Decarolis, working through Bank of Italy and structural-econometric studies of thousands of Italian contracts, found three consistent effects: bidding behaviour converges toward uniformity rather than genuine competition; the resulting allocation of contracts behaves statistically more like a lottery than an auction; and average-bid formats are typically less efficient for the procuring authority than a straightforward lowest-price auction paired with proper screening.
The collusion channel is the sharper concern. Because the winning number is a function of what everyone else bids, a cartel that can insert even a handful of coordinated “filler” bids can drag the calculated average toward whatever price its preferred member wants to win at. A study of road-works tenders in Turin — run under a formula close to Nepal's, including a trimmed average — found firms doing exactly this, expanding the pool of bidders artificially to shift the number. Italy's competition authority later named the average-bid criterion specifically as structurally favourable to collusion, a finding echoed in a 2025 review of European competition tools in public procurement. Nepal's own engineering literature has flagged the identical vulnerability: firms registering associated shell companies purely to bid alongside their real one and tilt the mean, a tactic documented internationally as far back as 1993.
Academic surveys place average-bid auctions in the procurement law of several other countries — Chile, China, Colombia, Japan, Peru, Malaysia, Switzerland and Taiwan among them, alongside Italy. That list is sometimes cited by supporters as evidence the method is mainstream and safe. It is more accurately read the other way: it is a well-studied method precisely because so many jurisdictions that tried it later found reason to constrain it. Italy's own 2016 and 2023 procurement code reforms progressively restricted the circumstances in which average-price and automatic-anomaly-exclusion criteria could be used, in direct response to the collusion patterns researchers had documented.
Why Nepal is a worse market to try this in than the ones it was tried in before
Every vulnerability documented abroad already exists — and in some respects is sharper — in Nepal’s construction sector.
Nepal's major road, building, and irrigation contracts are won by a comparatively small, repeat cast of firms who bid against each other. In such a familiar-face market, tacit coordination is easiest to sustain and hardest to prove. Italy's antitrust authority, which first named the average-bid criterion as pro-collusive, is a mature, well-resourced regulator with decades of case history. Nepal's Competition Promotion and Market Protection Board and the Commission for the Investigation of Abuse of Authority already lag behind on existing bid-rigging and blacklisting complaints under the old, simpler rule. There is no evidence either body has been resourced to detect the more subtle padding-of-the-average behaviour this formula invites.
Nepal’s e-GP portal centralizes bid submission, enhancing transparency. However, centralization alone does not prevent collusion among firms. On the contrary, it can make it easier for cartels to monitor competing bidders and adjust their numbers accordingly.
The dummy-company tactic is not hypothetical here either. Nepali engineering researchers examining bid-evaluation practices have separately documented unbalanced bidding and bid-rigging behaviour in road and bridge contracts as an existing, live problem. The underlying industry conditions the new formula depends on being clean are the same conditions the pre-existing literature says are already compromised.
Nepal's own numbers show the diagnosis is right and the prescription is still wrong
To be fair to the amendment's supporters, the underlying problem is real and well documented in Nepal's own data, not invented as a pretext.
An Asian Development Bank assessment found that only 15 percent of tasks in Nepal's public construction sector were completed within their expected timeframe, with an average delay of 37 months. Separate academic studies of rural-municipality projects recorded time overruns in more than half of surveyed contracts, ranging from 24 percent to over 500 percent of the planned schedule. One district-level study of road and bridge contracts found bids averaging 37.5 percent below the tender price were directly associated with cost overruns, schedule slippage, and resource shortfalls on site.
That is a genuine case for intervening in how bids are screened. It is not, on its own, a case for average-bid specifically. The same body of Nepali delay research consistently ranks other causes above underbidding alone: delayed site handover, delayed payments from the client side, poor contract administration, and weather and geological conditions all feature as high or higher-ranked causes of time and cost overrun than the winning bid price itself. One study of World Bank–financed rural projects in Nepal noted explicitly that donor-financed contracts, evaluated under a different methodology that screens for abnormally low bids without abandoning the lowest-price principle, do not show the same overrun pattern driven by bid price alone. That is a strong hint about where the actual fix lies — not in changing who wins the auction, but in how badly low bids are screened before the award, and in fixing the client-side failures (site readiness, payment discipline) that the same literature says matter just as much.
A reform bundled in the industry's favour
Context matters here, and it is worth stating plainly. FCAN has been formally asking the government to adopt exactly this average-bid formula since at least 2012 — a decade-long lobbying position, not a sudden technical insight. The same amendment that adopted their preferred pricing formula also softened what happens when a contractor defaults: previously, a contractor whose contract was terminated for non-performance was liable for the full cost of finishing the remaining work; under the new Act, the government instead seizes the performance security and recovers only the gap between the original contract value and the updated cost estimate for the unfinished portion — a liability limit construction companies had also long opposed and have now had reduced.
Taken together — the pricing formula the industry asked for, the liability cap the industry asked for, delivered via the fast-track ordinance route specifically chosen to avoid the slower scrutiny a full bill would have faced — this reads less like a neutral technical correction to a real problem and more like a package negotiated substantially on the regulated industry's own terms.
The fault line donor-financed Nepal now has to manage
There is a structural complication that has had almost no public attention: Nepal's major infrastructure portfolio is not procured under one rulebook. World Bank IDA credits and ADB loans all require, under their own procurement regulations, evaluation on the basis of the lowest evaluated substantially responsive bid — not an average-bid formula. Where those lenders worry about unrealistically low bids, their standard tool is statistical: calculate the mean and standard deviation of submitted bids, flag anything falling roughly one standard deviation or more below the mean as an Abnormally Low Bid, and subject it to enhanced due diligence on the bidder's capacity to deliver at that price, while still awarding to the cheapest bidder that passes the check.
That is a materially different philosophy from Nepal's new domestic rule, and it means the country now runs two procurement systems in parallel: GoN-financed contracts evaluated on proximity to an average, and MDB-financed contracts evaluated on lowest responsive price with statistical screening. Every project office managing blended or co-financed portfolios — and Nepal has many — now has to apply two different award philosophies depending on the funding source of a given package, with two different risks of dispute, two different training needs for evaluation committees, and two different sets of assumptions bidders have to hold in their heads when deciding how to price a tender. That is added transactional friction the amendment's drafters do not appear to have weighed against the benefit it is meant to deliver.
What would actually make this defensible
None of this means the goal behind the reform — stopping unrealistic underbidding — is wrong to pursue. It means the specific mechanism chosen is the wrong tool, and better-tested alternatives already exist.
Adopt statistical screening for abnormally low bids instead of, or alongside, the average-bid award rule. The World Bank’s guidance — mean minus one standard deviation as the flag line, followed by a documented capability review rather than automatic disqualification — achieves the stated goal without abandoning the lowest-price principle that rewards genuine efficiency.
Trim the formula if average-bid is kept. Discard the highest and lowest 10 percent of bids before calculating the average, as some Italian municipalities eventually did, to blunt the effect of a single padded or shell bid.
Publish bid sets and calculations for every contract above a modest threshold. Transparency after the fact is the cheapest available check against coordinated bidding.
Screen bidder ownership before award, not after disputes. Cross-checking beneficial ownership records can catch shell entrants tied to an existing bidder.
Mandate review clauses in implementing regulations. Require the PPMO to publish contract-cost, dispute, and delay data after two years, with a statutory obligation for Parliament to revisit the formula if outcomes have not measurably improved.
Fix client-side failures such as late site handover and slow payment release. A pricing formula change that leaves these untouched will only partially improve outcomes.
The bottom line
Nepal has swapped a well-known, badly managed problem of reckless underbidding for a different, arguably worse-documented one — imported wholesale from a jurisdiction that spent two decades trying to engineer the collusion risk back out of the same formula. It did so through a process that skipped full legislative scrutiny, at the request of the industry that stood to benefit twice over in the same package. The fix is not to shrug and hope Nepal is the exception. It is to import the safeguards along with the mechanism — or better, to adopt the screening approach the country’s donor-financed projects already use, and that a growing body of research suggests actually works.
The author is affiliated with DT Global International Development USA, Ltd.
