The suggestion that the NSW government had done nothing wrong would normally seem unlikely, irrespective of context. Even more so when the context is a $2.6bn capital investment project that’s at risk of collapsing, requiring a massive government bailout, or both.
However, the funding shortfall threatening the public-private partnership (PPP) to build 78 new Waratah suburban trains for Sydney CityRail services is an exception. The NSW government did a good job in managing risks for this deal, and it’s at risk of having to stump up extra taxpayer’s cash for reasons nobody can blame it for not foreseeing.
On the plus side, even if the government does have to step in, it’s unlikely the NSW taxpayer will lose much. The biggest loser is likely to be Downer, the train’s builder, which is exactly where the blame should lie. Unfortunately for the NSW government, the deal is arcane enough that the press and the opposition can easily claim otherwise.
Why is PPP a good idea?
PPPs have become popular worldwide as a way of funding new rail rolling stock programmes because of the transfer of risk they involve. Under traditional contracts, a government agency buys some trains from a private contractor, who builds them and hands them over to the agency when they’re built, with the agency paying to maintain them until they’re no longer economic to repair. Under a PPP, the agency agrees with a private contractor that the contractor will provide it with working trains for a fixed time period, with repair and maintenance work carried out by the contractor.
The advantage of a traditional deal is that it’s straightforward. The disadvantage is that it sets up a terrible incentive structure. The manufacturer’s best option is to promise the lowest possible upfront cost, then deliver as many trains as possible as soon as possible, even if they won’t reliably last for their design life, so they can win the contract and get paid. The government’s best option is to take the lowest bid and accept as many trains as soon as possible, even if they won’t reliably last for their design life. So both sides have a good motive to agree to underspecified trains that will cost more in the long term. Not just in actual maintenance money but also in disruption, which is the absolute killer: a half-hour delay to a morning commuter train carrying 800 passengers costs the NSW economy over $10,000 in lost work hours.
The last trains procured in NSW under a traditional contract were the Tangara sets in the late 1980s and early 1990s. They’ve suffered from reliability problems throughout their working life, partly because the manufacturer built them down to the minimum spec, and partly because RailCorp specified cheaper components in the design than the manufacturer recommended. The first trains procured under a PPP, on the other hand, are the Millennium trains. Now, these also arrived late and suffered from initial reliability problems – but crucially, the manufacturer paid for the delays, and now that the trains are in reliable service, the manufacturer will pay for all future maintenance to keep them running.
The PPP approach for train procurement has also worked in Melbourne, as well as in countless successful projects in Europe, North America and Asia.
So what’s gone wrong on the Waratah project, then?
Well, first of all, as with almost every new rolling stock project in the world ever, the engineering side of the project is running late. Integrating multinational suppliers’ work has taken longer than expected, and the trains have encountered unexpected problems when they’re actually run on RailCorp’s rails. This is the consortium’s fault for not managing the project well enough, and for not pricing risks and delays into the original contract.
So Reliance Rail – the legal entity that’s providing the trains to RailCorp – is liable for a sizeable compensation bill that will wipe out its profit (Reliance’s main shareholder, Downer EDI, has already written off its entire equity investment in the consortium). This is exactly the point of doing a PPP – the private sector partner is paying for its failures, and the NSW taxpayer isn’t. Hurrah! And this would have been the end of the story, but for some complicated financial engineering…
Reliance is a company created solely to buy Waratahs from Downer EDI, and provide them to RailCorp with Waratahs. When the project was launched in 2006, Reliance raised a total of $274m from its shareholders (Downer, AMP, and, erm, RBS and Babcock & Brown), and $1.9bn in AAA-rated bonds. It also got its bankers (NAB, Westpac, Sumitomo and Mizuho) to pledge another $357m in debt financing that would be drawn after 2012 if required – effectively, an overdraft facility to cover the possibility of overruns.
From the point of view of everyone involved at the time, this was a fine deal. Reliance had enough cash to pay Downer to actually build the trains, and a guarantee of getting its hands on extra cash from the banks if its funding obligations ran ahead of delivery-based payments from the NSW government. NSW could see that Reliance had easily enough funding to deliver the trains and compensate for overruns. The bondholders had super-safe AAA bonds. And the banks’ overdrafts would have been on the same investor-friendly terms as the bondholders.
Unfortunately, the deal also involved one of the more surreal bits of the pre-GFC financial system. Reliance paid two large insurance companies – XL (now Syncora) and FDIC – to insure its debt. In other words, if Reliance were to go bust – say, because it spent all its capital and was completely unable to deliver any trains, or because NSW underwent a Communist revolution and refused to honour any private sector contracts – then XL and FDIC would make sure Reliance’s bondholders and bankers were repaid. This guarantee is how Reliance managed to get its bonds rated as AAA (‘almost totally safe’) despite being exposed to some operational risk.
You might be wondering why it was cost-effective for insurance companies to take on financial risks by insuring loans, rather than simply getting banks and bondholders to accept slightly riskier terms in exchange for higher interest payments. At the time, it was universally accepted that the logic was simply based on different risk appetites for different types of investor, with enormously clever computer models backing this up.
As it turned out, the actual reason was that the insurance companies had massively cocked up – they hugely underestimated the chances of defaults on the private-sector loans they made (mostly in US real estate. Yes, subprime mortgages again). And when the real estate companies whose bonds they were guaranteeing went bust, the insurance companies took the hit. So Syncora and FDIC are now both in serious financial trouble, and may become insolvent.
Who loses, and what can we do about it?
The insurance crisis doesn’t matter from the point of view of Reliance’s bonds: they’ve been sold, and the insurers’ failure is the bondholders’ problem. The same is true with the shareholders’ money. Reliance has $2.2bn in real money, and it will cover the vast majority of the project’s cost.
But it does matter for the overdraft: the bankers say that without the AAA rating that the insurance would have provided, their agreement to provide the extra $357m is null and void. So there is a danger that in the final stages of the roll-out, before the company starts receiving regular payments from RailCorp, it will run out of cash. This is what’s driving front-page headlines about ‘rail crisis goes off track’.
It’s the job of the opposition and the press to talk this up into a major disaster, with “$2.6 billion project in jeopardy and NSW taxpayer on the hook” headlines. But in truth, it isn’t a major disaster at all.
The insurers aren’t bust yet, and may not actually go bust at all – in which case, the banks will have to provide the overdraft. Even if they don’t, the bondholders will lose far more money if the contract collapses than if they agree to let the provider of the completion funding take precedence over their bonds (in other words, ensuring that if the company did go bust, the $357m would be paid before any of the bondholders are paid). So there’s a strong incentive for everyone in the private sector to work things out.
But let’s assume the worst financial scenario happens, and the NSW government ends up deciding to guarantee the $357m itself rather than see the introduction of the trains delayed until everyone’s sued everyone else to death. Reliance would pay the government back with interest over the life of the contract. The only risk to the NSW taxpayer would be if the project went so badly wrong that even with the $357m overdraft, the company was unable to bring the trains into service. This doesn’t look likely, considering the level of delays and the reported engineering quality of the trains so far.
A government bailout defeats some of the point of a PPP, in that NSW now has a financial interest in ensuring Reliance can pay its debts. But on the other hand, the private sector has already lost $274m (the value of shareholder equity in Reliance, which has been wiped out), and would lose still more if the final project went so wrong that the company did collapse (the $1.8bn in bonds). Even if NSW does end up having to underwrite some of the debt, this will still be a better deal for taxpayers than a traditional purchasing model.
4 thoughts on “Waratah trains: why the NSW government isn’t at fault”
Have you read The Black Swan by Nassim Nicholas Taleb? If not I'd recommend it as an interesting read, if not what do you think of it's conclusions regarding predicting (pricing in) risk?
Taleb is interesting. Dsquared occasionally of these parts has a good not-exactly-countering-but-tempering view of his work.
Very well written.