Title XVII

Constellation Energy has announced it will not be continuing its effort to build a third nuclear reactor at its Calvert Cliffs site in Maryland. I looked at the structure of this project, and the associated subsidies, in detail last year in this paper for the Nonproliferation Policy Education Center.

The firm has blamed the US Department of Energy for unreasonably high credit-subsidy payments associated with a $7.5 billion conditional loan guarantee committment to build the new facility.

According to Platt's:

In a letter addressed to DOE Deputy Secretary Daniel Poneman dated  Friday, Constellation Vice Chairman Michael Wallace said the US government's $880 million fee -- 11.6% -- for a loan guarantee to support the $10 billion project was too high.

"Such a sum would clearly destroy the project's economics (or the economics of any nuclear project for that matter), and was dramatically out of line with both our own and independent assessments of what the figure should reasonably be," Wallace wrote.

DOE agreed to cut the credit subsidy amount to 5 percent, in return for $300 million in additional equity from Constellation and power purchase guarantees.  Constellation said no, and thought a credit subsidy on the order of only 1-2% was appropriate.   Christine Tezak, an analyst at Robert W. Baird & Co. in Milwaukee said DOE's terms were no better than those on offer from private funders.  Her claim does not seem credible.   Private funders willing to do 80% debt financing at the risk-free rate in return for an up-front 11.6% payment?  Absent names and details, a good deal of skepticism about these other financing options seems warranted.

Color me skeptical as well on Constellation's claim that the high credit subsidy payment was the main reason for pulling out.  Constellation's own cost models (via presentations done by Constellation Senior VP Joe Turnage in 2008) estimated that the loan guarantees would reduce their cost of power production by 3.7 c/kWh levelized.  Again, using their own operating assumptions for the new reactor, this would have resulted in financing savings of nearly $500 million per reactor per year.  The payback on the "dramatically out of line" loan guarantee fee?  Less than two years.  And that was before rising reactor costs and the credit meltdown made the value of risk-free borrowing even larger than before.

There are two other explanations for Constellation's change of heart that are far more likely:

• Poor and worsening economics for nuclear new-build.  There is much wider agreement than a few years ago that the economics of nuclear power are just really.  The combination of worsening credit conditions, falling power prices due to the economic recession, the surge in low cost natural gas from fracking, inaction on carbon pricing, and a rather less than stellar delivery schedule on Areva reactors in Europe all suggest caution is greatly needed for an investment of this size.  The Baltimore Sun has a good summary of many of these issues. 

• Bargaining strategy.  Constellation may be playing project subsidy chicken with partners on both sides of the Atlantic -- with DOE and OMB to get the credit subsidies down to one or two percent; and with the government of France (through their export credit agency Coface) to come forward with the sovereign guarantees that were supposedly pending at earlier stages in the project. 

Hopefully the Obama administration will not intervene here to once again give out the dessert before making sure the spinach has been eaten.  The Administration ought to be indifferent to which low carbon energy options win in the marketplace, and the marketplace seems to have spoken on this project.  However, Platt's indicates DOE is still trying to woo the firm to take the federal billions -- not a great sign for the taxpayers who will ultimately bear the burden of any default.  In contrast, shareholders seemed happy with the change:  Constellation's stock price rose following the announcement and is expected to rise further as a result of the cancellation.

Westinghouse, on the other hand, is still signaling full speed ahead. Time will tell.  It is always so much easier to stay in the game when you are betting other people's money.

Update, 18 October 2010:  It looks like the poor economics may be the leading rationale for Constellation's recent actions, rather than a bargaining strategy to get more favorable terms on loan guarantees.  The firm offered to sell its stake in the nuclear project for $1 plus reimbursement of $117 million in incurred project development costs.

It is hard to imagine the executives of a public company extending an $8.3 billion line of credit to a risky new venture, and then be unwilling to provide any detail to shareholders on how they evaluated the benefits and risks of the deal, and what they were expecting to get paid for taking on that risk.

Yet, this is effectively what the US Department of Energy has done for an $8.3 billion loan guarantee package to support the construction of two new nuclear reactors in Georgia.  We know that DOE is required by statute to have gotten an independent assessment of the project risks without guarantees, yet no details on who did that analysis (to gauge their reputation and potential conflicts of interest) or its findings have been released.  

A key element of the loan guarantees relates to the size of the expected "credit subsidy" that the borrower is expected to prepay to the government.  (This metric is supposed to assess expected losses relative to the "risk-free" government cost of borrowing; even with no default, the utilities will recieve very large subsidies relative to the private borrowing rate on the proposed projects). DOE, with a strong goal of pushing loan guarantees out the door, has an incentive to down play the risks of initial guarantees.  Since actual risks will become evident only years from now if the projects start to veer off course (as has occurred with new reactors in both Finland and France), understating risks now increases the benefits to borrowers, expands the number of deals DOE can do with its pool of capital, and delays the day of reckoning. 

Taxpayers have exactly the opposite interest, since the point of greatest control over our ultimate bailout tab is in the terms of the initial guarantees, and whether or not to extend the credit at all.

In March, a group of environmental groups issued this letter as a followup to a number of freedom of information act (FOIA) requests that DOE had effectively ignored.  Here is one of the rejections to an NRDC request as an example.  The goal of the FOIA's was to obtain the necessary detail on the loan agreements with which to assess the quality of DOE's due diligence, and their assumptions regarding the risk of the loan. 

DOE's Office of the Loan Guarantee, under the Executive Directorship of Jonathan Silver, has continued to release nothing.  As a result, the Southern Alliance for Clean Energy has moved from FOIA to litigation in an effort to ensure transparency and accountability in this massive lending program.   It is surprising that these actions have not yet engaged the imaginations and lawyers from groups focused solely on good governance and fiscal controls.  There are clearly important principles at play that will affect government accountability well beyond the energy area.  This is well illustrated by the conflict between DOE and the Office of Management and Budget on the proper way to calculate credit subsidies, though I've seen little in the public sphere since this article late last year. 

With total energy loan guarantee commitments reaching $111 billion, and much more on the way if the administration is successful, the scale of this program certaintly warrants attention even by those concerned only about the fiscal rules by which our government operates.

First, a disclaimer of sorts:  as I've stated many times, I think the large scale credit support to private energy enterprises is a bad idea and not likely to end well.  That said, I also think that the dynamics of how credit support is provided is an important issue for evaluation for many areas of policy; a start of that discussion is below.

While much of the discussion on the US Department of Energy's credit support to new nuclear reactors and other energy technologies refers to "loan guarantees," this is only part of the story.  The statute authorizing this support also requires that where the federally-guaranteed portion is 100% of the project debt, the Federal Financing Bank must be the lender.  That is, if the federal government is on the hook for all of the debt, the loan guarantee actually becomes a direct loan.

I've gotten a number of questions on whether this is a good thing.  While my answer (that it depends) may seem unsatisfying, thinking about the incentive structure of the direct loans versus the guarantees is quite useful.  The same dynamics at play with respect to subsidized credit for nuclear reactors are also involved with other government lending activities -- for example, student loans, a financial activity that has only recently been effectively nationalized. 

The direct lending versus guarantee decision involves a tension between direct costs (in which the direct loan approach can be better) and risk vetting (where the decentralized guarantees distributed by other lenders can be better).  There is also a trade-off between static efficiency (how you behave when you already know the market and what you need to do) and dynamic efficiency (how you adjust operations to deal with changing market and policy conditions).  

For any of these programs, if you know the policy, have trained your staff, and bear all of the default risk anyway, running the lending as a direct loan probably can save money -- both in administrative and markup costs; and potentially in terms of reduced conflicts of interest.  The FFB has less incentive to pump up the loan disbursement volume as high as possible relative to a Goldman Sachs, for example.

However, these types of conditions do not apply to all situations.  Where policy conditions change; staff needs to be hired, fired, and trained in order to do their job efficiently and well; and there is alignment of interest between the lender and the long-term performance of the loan they are authorizing, direct loans are probably less efficient.  While there may be some additional markups on the lending that does go out, the responsiveness of the decentralized organization in terms of finding and vetting opportunities, and monitoring them over time, is likely to more than compensate.  The information that the decentralized lenders see, process, and act on can be immensely valuable in making better decisions with better long-term outcomes -- but only if the lenders share a material stake in that positive long-term outcome. 

In truth, this issue applies far more widely than whether the federal government should make direct loans instead of guarantees.  Franchising is an interesting example of these same tensions entirely within the private sector (credit for this example belongs to Mike Jensen, one of my grad school professors, not to me).  The direct financial costs of franchising are much higher than for a parent corporation to simply own the store and hire a manager.  But the agency costs -- the losses from the local staff and manager acting on their own behalf rather than that of the firm -- are quite high.  Jensen noted that as the number of stores in a particular region grew, the parent corporation at some point would be able to build in sufficient local oversight to overcome the agency costs, and would start to operate company-owned stores rather than franchises.  Jensen's papers on this topic are in academic journals and (to my knowledge) are not available for free on the internet.  But perusing this book on franchise economics will give you a feel for the issues.  

This incentive issue may be part of the logic on the distinction between a 100% guarantee of debt versus lower levels that is made in the Title XVII program:  so long as the originator is also at risk, they will execute due diligence on the borrower at the outset, and do a better job monitoring the loan over the course of its life.  However, the logic has a few holes even for cases where the feds don't guarantee all of the debt.  First, since the DOE chooses the borrower, there is no benefit from outsourcing in terms of deal vetting and selection.  Second, while less than 100% federal guarantees may mean the private lender bears risk on the debt issuance (in theory providing an incentive for due diligence), that is not the only possible outcome.  Some other party (a foreign government, perhaps) may be picking up the rest of the risk; or the lender may immediately securitize and sell the residual risk onto capital markets.  The resulting zero or very small stake in long-term performance of the loan would create a very weak incentive for oversight and management even where the federal guarantees were for less than 100% of debt.

Perhaps the reputational hit to a bank, or internal requirements, or their basic "nature" as bankers would lead them to vet deals anyway?  This was an argument a number of banks made themselves in pushing for 100% debt guarantees in 2007.  Consider the comments submitted by H. John Gilbertson Jr. (Managing Director) and Alejandro Hernadez (Associate) at Goldman Sachs: 

"In addition, DOE’s concern that the 10% non-guaranteed portion is needed to make sure lenders have 'skin in the game' and  perform adequate due diligence is inconsistent with the realities of the capital markets. Even when debt instruments are backed by government guarantees and other forms of high-quality credit support, lenders and their counsel do not throw caution to the wind and abandon their diligence efforts. Lenders are by nature 'downside animals' and will perform appropriate diligence notwithstanding the existence of a guarantee, and often their internal policies require them to do so."

I discounted this argument then, and I think my view has been borne out.  There is not much evidence that investment banks focused on collecting fees for high deal flow in the area of mortgage-backed securities slowed the production line much to vet the deals or stop them because they were unsound.  Where project risks are high, and oversight important, it is not at all clear that maximizing liquidity, or minimizing loan processing or borrowing costs, should trump proper incentive alignment for long-term venture success.

So what makes sense for energy lending?  First, a logical incentive structure -- though this issue seems to be mostly ignored by those in DOE and beneficiary industries so focused on making taxpayers a major investor in high risk energy assets. Second, FFB involvement might make sense not just based on percentage of debt guaranteed, but for any deal above a certain size (say $500 million) in which the originator does not retain a minimum percentage of the debt risk over the long term.  That approach could avoid paying the private sector high fees for basic issuance of debt instruments over which they are exercising no due diligence anyway; and force greater visibility on the cost of these programs at the federal level.  If the FFB plans to sell the debt using private banks, however, care is needed to be sure the banks don't just simply recapture the economic rents at that point. 

In addition, the FFB has historically been a mechanism for centralizing financing functions to reduce costs, rather than playing a due diligence role.  For such large support to single enterprises as Title XVII and CEDA will entail (far higher than in other government programs), an independent due diligence function -- working on behalf of the Treasury and the taxpayer rather than the policy objectives of the White House or Secretary of Energy -- might make a great deal of sense.

I've just posted the slides from a presentation I gave at the New America Foundation in December.  The discussion provides some additional detail on why I am so critical of the Obama administration's push for massive loan guarantees to energy facilities. 

There are three main issues here.  The first, as illustrated in the chart reproduced below, the scale of lending is far larger than other similar programs the federal government has undertaken in the past.  Title XVII of the Energy Policy Act of 2005, in combination with stimulus spending, is now well over $100 billion in authorized credit support.  The Clean Energy Deployment Administration (CEDA) -- thankfully still only a proposed law -- could be even larger.  The idea that experience from existing export credit agencies or subsidized loans to rural energy facilities can serve as a sound model for running these much larger programs is, in my view, misguided.  Steps to boost review, staff skills, and incentive alignment appropriate for a program of this scale do not seem to have been taken.  In fact, the information I've received from people more closely linked than I to the agencies involved with assessing the risks of the loan guarantees indicates that political pressure to minimize the estimates of financial risk (and hence any required credit subsidy prepayments) continue unabated.

The second important issue is that the lending programs are focused primarily on energy technlogy development, with the mistaken assumption that if you build it, they will come.  Yet, experts such as Innosight, LLC (Clayton Christensen's firm) that focus on disruptive transformation of industries, identify the technology as only one of four key attributes needed for success (see slide 4).  In fact, when a political process drives capital allocation, the most heavily funded groups tend to be the most politically powerful, not the sectors most able to deliver carbon reductions at low cost (slide 5). 

Finally, the presentation identifies a number of key program attributes that are likely to contribute to a higher or a lower chance of success.  When the CEDA program is evaluated against these criteria (slides 6-8), the deficits in program structure and control become all too evident.