Since the final months of 2008, three gas-fired power deals have been hovering on the edge of the US bank market. The three, from an NRG/United Illuminating joint venture, a tie-up between EQT and Fortistar, and from an SNC/EIF/Suez-led grouping, all benefit from power purchase agreements. They all use proven technology, and they all feature well-known sponsors.

But all of them have been subject to various tweaks and alterations, most of them not pricing-related, to bring in lenders. And the normal jostling for titles between arrangers, intense during the best of times, has acquired a sharp edge. If the similarities with the independent power project market of the mid-1990s were pronounced early in 2008, they are even more stark now.

Midland Cogen comes out first

The furthest advanced deal is the one without any construction risk. A joint venture of the EQT Infrastructure Fund and Fortistar signed a definitive agreement on 19 March to buy the Midland Cogeneration plant. The two are thought to be paying roughly $1.1 billion for the 1,560MW cogeneration project, located in Midland, Michigan, near the headquarters of Dow Chemical. They are financing the acquisition with $515 million in debt, led by Union Bank of California and WestLB.

Midland Cogen is something of an anomaly, since it is one of the largest cogeneration plants in the US, selling power to Consumers Energy, steam and power to Dow Chemical and steam to Dow Corning, under long-term power purchase agreements. It was developed as a nuclear plant, reconfigured as a gas-fired plant after 80% of the work on the nuclear plant was complete.

Consumers was the developer of the plant, and the strain of the cost overruns was the main force behind its creation of the CMS Energy unregulated subsidiary, which was once a force in Latin American and Middle Eastern power. It refinanced the asset with a lease-bond issue in the early 1990s, and sold its 49% stake in the venture to GSO Capital Partners, now a Blackstone affiliate, and Rockland Capital Energy Investments for $60.5 million in 2006.

GSO and Rockland, along with Dow, are selling their stakes in the plant, which is now clear of the debt and the lease. The new owners have made such a healthy return by keeping faith with the Michigan regulatory process. CMS sold the plant, which accounts for close to 10% of the state's capacity, at a time of high gas prices. It had quarrelled with the state for much of the plant's history over how much of the plant's costs could be passed on to consumers.

The new owners, however, assembled an amended and restated power purchase agreement for 1,240MW of its capacity, between Midland and Consumers, which runs to 2025 and was signed in 1986. The price of getting regulatory certainty was a reduction in the agreement's tariff, a result that Michigan's Public Service Commission loudly trumpeted, and the offtaker's ability to shop for cheaper power elsewhere.

The July 2008 agreement was followed by an auction, which a joint venture of European infrastructure fund EQT and New York-based independent power producer and landfill gas operator Fortistar won in October 2008. The two benefited from EQT's judicious hiring of several former employees of ABB, which installed the turbines at Midland. EQT is putting up 65% of the equity, with Fortistar contributing the remainder.

The financing lacked for similar moments of serendipity, not least because the new owners have had to put up roughly 45% of the price as equity. The financing that the two sent to market consisted of a $525 million seven-year loan with minimal amortization, but after a consultation with the small number of big underwriters standing in the New York bank market was reconfigured to consist of $275 million in amortizing term debt, a $140 million bullet tranche and $100 million in working capital debt.

WestLB and UBOC are not providing the bullet tranche, whose refinancing risk is a source of particular horror at credit committees. Among the banks rounding out the senior lender group on the term debt are understood to be Calyon, CoBank, GE Energy Financial Services, Natixis and US Bank. GSO and Rockland are among the entities providing the bullet debt, mostly because they like the 350bp over Libor pricing and 300bp fees, according to sources close to the sponsors*.

Astoria II: Better script, how's the acting?

The most familiar credit in the market is the financing for the $1.1 billion Astoria II unit in New York City. The financing benefits from a longer and stronger power purchase agreement than Midland, since it has a 20-year power purchase agreement with the New York Power Authority, a AA-/Aa2-rated municipal power authority. The agreement calls for the plant to be online by 2011, since it will in part replace the Charles Poletti plant, which is located, like Astoria and much of the city's generating fleet, in Queens.

Astoria II is the 500MW follow-up to a 500MW first unit at the same site, which has the same set of sponsors. The first unit was financed in 2004 using $690 million in Credit Suisse-led B loan debt, and then refinanced in April 2005 using $725 million of Calyon-led high-yield debt. It has a 10-year power purchase agreement with the city's private utility, Consolidated Edison.

The original developer of the project SCS Energy, sold much of its stake to Suez in mid-2008, and had brought in EIF and La Caisse as equity providers before the first financing. Rounding out the group is the plant's engineering, procurement and construction (EPC) contractor SNC-Lavalin.

This tangled history and diverse sponsor group, not to mention the logistical difficulties associated with building pants in the city, complicates the financing process. At one time several banks' names have been attached to the project, including Calyon, WestLB and Natixis, all of which have laid claim to lead arranger title. Most recent indications are that Calyon, EDC, Natixis and WestLB have credit approval, for underwriting commitments of roughly $100 million each.

Still to be determined, and the financing structure is still in flux, is the total debt component, which will depend upon responses from other potential lenders. The presence of the deep-pocketed EIF, La Caisse Suez and SNC-Lavalin, the last of which put in an all-equity option on its bid for the Montreal Symphony Hall PPP, will be reassuring to those banks that are already committed. But the sponsors, which have so far found it difficult to present a coherent message to the market, will need to pull together smartly to bring the deal home.

GenConn looks to peak

The last of the three is a portfolio financing for GenConn, a 50/50 joint venture between NRG Energy and United Illuminating. The biggest complicating factor in this financing has been the bid by Exelon Corporation for NRG, which has expended considerable management resources resisting the bid. Nevertheless the plant again benefits from a long-term power purchase agreement, this time a regulated 30-year contract for differences with Northeast Utilities' subsidiary Connecticut Light & Power (rated BBB/Baa1, S&P/Moody's).

The plants – Devon and Middletown – each use four GE LM6000 turbines, have a capacity of roughly 190MW, and are to be built without a fixed-price EPC contract, with NRG managing the process.

The contract provides for the developer to recover its capital costs and a reasonable return n equity, with a floor of 9.755, based on a 50/50 debt/equity ratio. This recovery is netted from whatever the project receives selling power into the merchant New England ISO. It has a 5% construction cost contingency, with reviews if overruns increase to between 5% and 10%, and annual review of the plant's operating costs.

With a power purchase agreement this generous, the sponsors assumed that a long-dated construction financing would be possible. In October they mulled publicly the possibility of a private placement to fund the project. However, the latest version of the financing contemplates a construction-plus-five-years mini-perm financing and three-year working capital debt at the holding company level totalling roughly $291 million, and $243 million in 2.25-year equity bridge debt, split equally between each sponsor. This could then be refinanced in the capital markets.

Two lead arrangers, Royal Bank of Scotland and Union Bank of California, have launched syndication on the $534 million in financing, which is priced at roughly 350bp over Libor for the operating company debt. They hope to wrap up syndication by the middle of April, on the back of the project's low gearing, long contracts and solid sponsor group.

What's the right way?

Some immediate lessons are clear from the recent burst of financing activity. Communication with potential participants is vital, even if there are too many of them. Bankers familiar with all three credits give best marks on this count to GenConn and the fewest to Astoria.

The pricing floor for a US power project, indeed even a well-regarded toll road with no traffic risk, is 300bp over Libor, and more likely 350bp. Or, as one syndications banker put it "300 is the new 100", referring to the floor that sponsors tried to break through during good credit market conditions. "It's getting harder," he added "to say we're in the business of pricing risk carefully, when we apply such a simple number to all new deals."

One explanation might be that only very similar high-quality deals are coming to market. A good long-term power purchase agreement with a top-rated counterparty, far from being a unique selling point, is now a prerequisite. To date, few sponsors have dared to come to market with a power hedge from a financial institution. The last deal to do so was Topaz Power, which reluctantly acceded to a hedge with Morgan Stanley Capital Group for its $740 million debt financing in May 2008.

But Topaz has since indicated that the sharp fall in power prices in Texas, while accompanied by a fall in gas prices, has made the hedge a much more attractive proposition than it first feared. Several hedge providers, including Credit Suisse, RBS Sempra, Goldman Sachs' J Aron and Morgan Stanley, are still understood to be open to new deals, and one developer, say bankers familiar with the market, is trying to interest banks in a project with such a hedge.

Projects have needed to shave additional years from debt maturities. The US definition of a miniperm loan has shrunk from construction-plus-five years to five years including construction. Lenders will ask for the maximum amortization possible, and an aggressive deployment of cash sweeps.

This means that only the fastest construction prospects, most of them gas-fired and using proven technology, can hope for financing. Coal projects, even if they could obtain permits, do not fit the bill. LS Power, probably the most successful coal plant developer, recently abandoned development of a 1,600MW coal project, White Pine, and shifted its attention to a nearby transmission prospect.

Meanwhile, project finance lenders say that they will shortly be looking at a new raft of wind projects. These, thanks to the Obama administration's stimulus package, may not need to find tax equity, where liquidity has been limited, and could instead take an upfront cash grant from the US Department of Energy. As has been the case in recent PPP financings, this cash cushion is designed to make lenders more comfortable.

But the projects would still receive tax benefits, primarily accelerated depreciation, that their owners could not exploit. "For that," explains one banker, apparently entirely seriously, "we're looking at doing a leveraged lease of some sort." The spirit of the 90s lives on.

*This sentence corrected to clarify that the two commercial banks are not providing the bullet tranche