In March 2011, Construction Law International, the journal of the IBA's International Construction Projects Committee, published "Project financing of alliance-based projects" (Vol. 6, Issue 1, pp. 18-26). I wrote it with Troy Edwards, Michael Jordan and David Wilkie, then colleagues at Allen & Overy in London, Sydney and Amsterdam. The PDF is below, with their kind credit.
The question we asked was simple. Alliancing had by then a solid track record in Australia and a first Dutch application in the A2 Hooggelegen alliance. Owners liked it. Contractors liked it. Yet no project finance lender would touch it. Why not?
The lenders' answer was consistent: an alliance offers no fixed price, no fixed completion date and no real recourse against the contractor. A financier lends against certainty, and an alliance offers a target cost, a pain share/gain share mechanism and a no blame culture. From a lending desk, that looks like the borrower has given away the three things that protect the debt.
Our reply was that the certainty of a fixed price, fixed time contract is largely an appearance. The risks that derail a complex project, ground conditions, incomplete design, late permits, interfaces between packages, do not disappear because the contract says they belong to the contractor. They come back as claims, as delay, as a contractor in distress, or as a termination that leaves the lender with a half-built asset and a dispute. The real question for a financier is not which model offers a fixed price on paper, but which model best manages the risks that will occur anyway. On that question, we argued, an alliance with proper governance, open books and an integrated team has a fair claim to being the safer bet. We closed with a call to educate the financing community.
Rereading the piece fifteen years later, three things strike me.
First, what we left out. We said nothing about step-in rights, which is the instrument a lender reaches for first when a project goes wrong. In an alliance, with its integrated team and collective decision-making, it is not obvious whom a lender would step in against, or what it would step into. That gap in our argument was real, and it is where any serious attempt to bank an alliance would have to start.
Second, what our evidence was worth. The alliances we cited as proof, the Victorian projects and A2 Hooggelegen, were publicly funded. They had never been tested against a project finance lender. We were arguing that lenders should accept a model that no lender had yet been asked to accept.
Third, and most important, the question is still open. In 2026, Gowling WLG is publishing a series on precisely this problem, in almost the same terms we used in 2011. The lenders' objections have not moved. That tells me our diagnosis was too mild. We treated the problem as a knowledge gap, to be closed with education. It is not. The obstacle is institutional: the way capital is treated on a bank's balance sheet, the methodologies rating agencies apply to construction risk, and the absence of a precedent that a credit committee can point to. None of those is changed by a good article. They change when somebody banks the first one, and the first one is precisely what the institutions are built to avoid.
That is not a counsel of despair. It is a more accurate description of the work. It also changes who the audience is: not the lenders' lawyers, who understood the argument in 2011, but the people who set capital rules, write rating methodologies and decide which project gets to be the precedent. Public owners, who do not depend on project finance, may have a role to play here that they have not yet recognised: every publicly funded alliance that performs is a data point the institutions currently do not have.
I raise this now because it sits directly on top of two things I am working on this autumn. On 9 October in Dubrovnik I moderate a panel at the ICC Croatia regional conference on whether the contract model determines the dispute, with a joint background paper by five practitioners from Finland, Brazil, the Gulf, Croatia and the Netherlands. The panel's question is a close cousin of the 2011 article's: if the risks occur regardless of the model, what does the model actually decide? The financiers' fifteen-year no is a useful test case. It shows that a model can be better at managing risk and still not be chosen, because the choice is made by institutions that measure something else.
The article, in full, with credit to my co-authors, is attached below.
Read the full article (PDF, Construction Law International, March 2011)
Originally published in Construction Law International, Vol. 6, Issue 1, March 2011, pp. 18-26. Co-authors: Troy Edwards, Michael Jordan and David Wilkie, then at Allen & Overy, London and Sydney. Reproduced with the kind permission of the International Bar Association.