When do project loans actually default?
Early, and then much less. Moody's marginal annual default rate for project finance bank loans peaks in the first year after financial close and falls every year thereafter: on the 1983 to 2015 data it ran 1.47% in year one, 1.41% in year two and 1.18% in year three, decaying to 0.32% by year seven and 0.09% by year ten (Moody's Investors Service, Default and Recovery Rates for Project Finance Bank Loans, 1983-2015, published 6 March 2017). The 1983 to 2018 update carries the same shape at a lower level, from 1.08% in year one to 0.07% in year ten (Moody's Investors Service, Default and Recovery Rates for Project Finance Bank Loans, 1983-2018, published 17 August 2020).
Moody's is explicit about what the early peak reflects, attributing the elevated first and second year rates to higher default risk during the construction phase or start-up challenges during the ramp up of operations, and observing in the later study that marginal default rates tend to decline to levels consistent with those of single A-rated corporates by year seven. A project loan is not one credit for twenty years. It is a speculative credit for roughly two years and an investment grade one after that.
One correction is needed before a sponsor draws the wrong conclusion, because the folklore overshoots the data. Defaults are not mostly a construction phenomenon by count. Moody's phase breakdown on the 1983 to 2015 sample records 74 defaults during construction against 385 during operations, so roughly five sixths of defaults happen after the asset is running. That is arithmetic, not contradiction: there are far more operating years than construction years in any portfolio. The right statement is about hazard, not about totals. Any given year of construction is far more dangerous than any given year of operation, and there are not many of them.
| Year after financial close | 1983 to 2015 study | 1983 to 2018 study |
|---|---|---|
| Year 1 | 1.47% | 1.08% |
| Year 2 | 1.41% | 1.03% |
| Year 3 | 1.18% | 0.85% |
| Year 5 | 0.72% | 0.52% |
| Year 7 | 0.32% | 0.24% |
| Year 10 | 0.09% | 0.07% |
Does it matter which side of the line a default happens on?
Considerably, and this is where the phase distinction earns its keep. On the same Moody's 1983 to 2015 data, construction phase defaults recovered 70.2% on average against 81.2% for defaults during operations. The construction defaults also arrived faster, at an average of 2.5 years from financial close against 4.3 years for operating defaults.
The reason is intuitive once stated. A project that defaults in operations is a working asset with a revenue stream, a market value and a buyer somewhere; a project that defaults during construction is a partially built structure that produces nothing, may require the original contractor to finish, and may not be worth completing at all. Lenders are not merely more likely to lose in construction. They are likely to lose more.
Set against corporate credit, both numbers are strong. Moody's ultimate recovery rates on project finance bank loans averaged 79.5% on the Basel definition and 77.3% on Moody's own over the 1983 to 2015 period, and the 1983 to 2018 study put them at 77.9% and 75.8%. That is the structural security package, the direct agreements and the step-in rights doing the job they were drafted to do. It is also why lenders will accept a coverage ratio on a project that they would refuse on a company.
“Everything in a project financing is designed around one transition. Sponsors spend the negotiation on the twenty operating years and the lenders spend it on the two before them, and the lenders are reading the right data.”
How do rating agencies treat completion risk?
As a ceiling, not an adjustment. Fitch states in its completion risk criteria that ratings in project financings with completion risk are unlikely to exceed the 'A' category due to risks associated with construction and timely completion, and that for projects exposed to both completion risk and operating risk the overall rating will be constrained by the lower of the two risk assessments (Fitch Ratings, Completion Risk Rating Criteria, published 10 July 2025).
The same criteria carry a threshold worth knowing: a project with greater than 90% advancement in construction and no material outstanding design issues will not be subject to a detailed completion risk analysis. The line is treated as effectively crossed near the end of building rather than at the formal completion certificate, which is a practical acknowledgement that the risk collapses before the paperwork does.
The performance of rated infrastructure and project finance debt reflects the constraint working. Fitch reported no defaults at all across its rated global infrastructure and project finance portfolio in 2024, against a long-run average annual default rate since 2005 of 0.3%, and a ten-year average cumulative default rate over 2005 to 2024 of 3.80% across the portfolio, 2.02% for investment grade and 15.90% for speculative grade (Fitch Ratings, Global Infrastructure and Project Finance 2024 Transition and Default Study, published 5 May 2025). The portfolio's average and median rating was BBB plus across 468 ratings.
What actually protects a lender across the line?
Four things, and a sponsor should be able to name all of them before signing. The first is the construction contract itself: a fixed price, date certain, turnkey arrangement that transfers completion risk to a contractor with the balance sheet to carry it, backed by liquidated damages sized to service the debt through delay rather than to compensate for inconvenience. The Fitch criteria make the corollary explicit, capping counterparty dependency at the contractor's own creditworthiness where the contractor is irreplaceable.
The second is sponsor support that falls away rather than persisting: a completion guarantee or equivalent undertaking that keeps the sponsor on the hook until defined tests are met, which is the reason the financing is not truly limited recourse until that date. The third is the reserve and contingency package, contingency inside the construction budget for the cost overrun, and a debt service reserve for the delay, sized against a schedule that assumes things go wrong.
The fourth is the completion test itself, and it is the one sponsors read least carefully. Completion is a defined event with performance criteria, and until the asset demonstrates them the financing has not crossed anything, regardless of whether it is generating revenue. A plant running at 90% of its rated performance is producing cash and has not completed, and the difference between those two facts is where a surprising number of disputes live.
What should a sponsor negotiate before construction starts?
The tests, the schedule and the exit, in that order. The tests first: what exactly must the asset demonstrate, measured how, over what period, verified by whom, and what happens if it demonstrates most of it. A sensible package has a partial or delayed completion path with a defined economic consequence rather than a cliff, because near-misses are the common case and a structure with no answer for them converts a technical shortfall into a default.
The schedule second. Long stop dates, extension mechanics, force majeure relief and the interaction between the construction contract's schedule and the financing's are frequently drafted by different teams and frequently do not line up. A relief event that extends the contractor's obligations but not the financing's long stop date has quietly moved a risk onto the sponsor that everyone believed the contractor held.
And the exit third, because the construction facility is not the permanent one. What the loan becomes at completion, whether the margin steps down, whether the covenant package resets to operating tests, and whether the facility must be refinanced or merely may be, are decided at close and determine whether the sponsor arrives at the operating phase with a negotiation or a deadline. The data says the credit improves dramatically once the line is crossed. Whether the sponsor captures that improvement is a drafting question asked years earlier.
As of September 2026
Sources: marginal annual default rates by year after financial close, average ultimate recovery rates, the construction and operations phase default counts, the average years to default by phase, and the phase recovery rates are from Moody's Investors Service, Default and Recovery Rates for Project Finance Bank Loans, 1983-2015, published 6 March 2017, and Default and Recovery Rates for Project Finance Bank Loans, 1983-2018, published 17 August 2020, on the Basel definition of default unless stated. The rating ceiling on completion risk, the constraint to the lower of completion and operating risk assessments, the 90% construction advancement threshold and the treatment of an irreplaceable contractor are from Fitch Ratings, Completion Risk Rating Criteria, published 10 July 2025. The 2024 default experience, the long-run average annual default rate since 2005, the ten-year average cumulative default rates and the portfolio rating profile are from Fitch Ratings, Global Infrastructure and Project Finance 2024 Transition and Default Study, published 5 May 2025. Nothing here describes a transaction or engagement.
This position sits within our project finance practice.