Download OMB Circular A-129

---

Like the prior post about A-11, this one takes a ‘legalistic’ view of what’s required with respect to WIFIA loan optionality for federal oversight compliance, this time by OMB Circular A-129.

There is a lot in A-129 about fundamental credit program policy — what the program is for, what it accomplishes, whether there are alternatives, etc. Much of it is problematic for EPA WIFIA, if actual outcomes were ever honestly reviewed. But that’s a big topic for another post. Here’ll we just look at what A-129 appears to require for correct FCRA budgeting, a narrower topic with more objective standards to evaluate whether the Program is in compliance.

We’ll parse the brief language above. Obviously, the Program needs to be in strict compliance with FCRA law, but the top statement goes a bit further. ‘In accordance with FCRA’ implies compliance with the spirit and principles, as well as the letter, of the law. Or so it can be read — in any case, there aren’t any modifiers or qualifiers to the statement, so a more expansive or natural interpretation seems to be justified. I’m sure the authors of the 1967 Budget Commission and the drafters of the lapidary 1990 law would have wanted it to be seen that way.

The next phrase clarifies that compliance is pro-active: “…agencies must analyze and control the risk and cost of their programs.” A surface view, like “all our borrowers are highly rated, completely non-federal eligible entities, so everything is OK” is not at all sufficient. The risks and costs of financial portfolios are always complex, and all the more so for big, long-term infrastructure loans. Credit risk is one factor, and perhaps the primary one at most federal credit programs, but it’s not the only one, and especially not at EPA WIFIA. Maybe the folks at the Program expected the loans to be all about clean water and improving the environment and that sort of thing, and they didn’t sign up for abstract financial analysis. Too bad — if you commit $25b of taxpayers’ money, you’d better learn how to properly analyze the actual risks and costs of the portfolio or hire people that can.

Now comes an interesting sentence: “Agencies must develop statistical models predictive of defaults and other deviations from the loan contract.” As discussed in a number of prior posts, some statistical analysis and modelling will be required to correctly estimate and apportion for the cost of WIFIA loan optionality, taking into account the probabilities of various factors. Such models are of course already in standard use for credit risk but note that the sentence doesn’t say “utilize” statistical models. It says that the Program “must develop” them. In other words, consistent with the requirement to ‘analyze’, the Program can’t just rely on off-the-shelf models that work for most other federal credit programs and call it a day. If there’s evidence that Program loans pose some unusual or unique risks and costs, as there is clearly at WIFIA, a customized statistical and ‘predictive’ model must be developed. Complicated? Boring? A PITA? Again, too bad — if you can see a risk or cost, you must develop the appropriate models.

But we should be precise. The sentence talks about “other deviations from the loan contract”. Obviously, that includes all the other things a borrower might do, short of payment default, that are in violation of the loan contract and pose a risk or cost to the government. Failure to perform on a covenant, for example. But the exercise of WIFIA loan options is not a deviation from the loan contract in that sense. The optionality is doubtless clearly stated within the document, and the borrowers’ lawyers would make very sure about that in connection with their clients’ underlying motivations. Strictly and literally construed, the language would seem to limit model development to factors arising from loan contract default or near-default, with the costly consequences of loan option exercise given a free pass.

However, there is another, more natural, way to interpret what OMB was trying to convey with the word ‘deviation’. Yes, of course, all defaults are included. But it would make more sense if ‘deviations from original expectations about the loan’ were also included. When a WIFIA loan is executed, it is expected (per official EPA narrative, anyway) that it is to be a source of financing for a project. Cancellation or delayed disbursement is possible, but they would be a ‘deviation’ from the original intent and expectations, the precise description of which forms the bulk of the loan contract. It might be considered unusual (in fed credit land) for a loan cancellation or delayed draw to pose risks and costs, but if there’s evidence these ‘deviations’ do, then I think they come under this sentence.

More generally, whatever OMB had in mind, FCRA law and the Anti-Deficiency Act would seem to require that any identified risk or cost to the Program be analyzed and modelled with appropriate methodology, e.g. statistical for WIFIA optionality. There is no free pass for certain kinds of unexpected costs — if there were, this would be exploited as a budget-gaming loophole, which was exactly what FCRA law was intended to prevent. As noted above, the Program must act “In accordance with FCRA”.

[Mmm…”exploited as a budget-gaming loophole” like, oh I don’t know, intentional reliance on PIA and mandatory appropriations to make WIFIA look magically cost-effective?]

Bottom line for this one? Whether they like it or not, I think EPA WIFIA was and is obligated to evaluate WIFIA optionality per the requirements of A-129. I have seen no evidence that they have done so or are planning to do so. It is possible that the Program simply didn’t disclose such evidence, but then why did OMB FY25 and FYTD26 loan subsidy apportionments apparently only include credit risk? Well, perhaps the picture is more complicated. Eventually, however, it’ll be clear whether or not EPA WIFIA is in compliance.

Download OMB 2025 Circular A-11 Federal Credit

---

I’ve written a lot about WIFIA loans’ optionality in recent posts. In this one, I’ll look at the specific connection between that optionality and what is required, per OMB’s Circular A-11, to be included in a loan’s subsidy cost estimate at commitment. I’m not a lawyer (though I’ve dealt with plenty of financial contracts), but I’ll take a legalistic approach here, as if A-11 was a ‘contractual obligation to evaluate loan cost’ and I wanted to see if the cost of WIFIA loan optionality was covered by it.

First, let’s review WIFIA loan optionality. There are two options that are statutorily permitted and undoubtedly written into the loan contract, though the word ‘option’ is probably not used:

• The option to cancel an undrawn loan commitment at any time, without penalty.

• The option to delay disbursement of the loan for any period of time after construction expenditures up to one year after project completion.

These options are probably present in some form in a lot of federal credit agreements, but they likely don’t have any predictable impact on loan subsidy cost. An individual might decide, after getting a student loan commitment, not to attend college or to defer it. Or a small agricultural project with a USDA loan commitment might be cancelled or delayed due to local funding issues. In these cases, the disbursement of the loan commitment is cancelled or delayed for highly idiosyncratic reasons that won’t be correlated to other factors included in the loan’s subsidy cost estimate. Hence, in most if not nearly all such cases, these options don’t need to be considered in the loan’s cost estimate at commitment, and the cost of final outcomes is correctly evaluated and budgeted in the standard FCRA re-estimate process.

However, things are different for WIFIA subsidy estimates, as actual operations FY18-FY25 make abundantly clear. The vast majority of WIFIA borrowers are highly rated public water systems with excellent access to the tax-exempt bond market and short-term financing. Their decisions with respect to cancellation or delay will be highly correlated with a major factor of loan cost — the level of US Treasury rates at the time of loan disbursement. Rational, sophisticated and well-resourced WIFIA borrowers can be expected to exercise their loan contract options under the following conditions:

• Cancellation of the undrawn loan commitment (if not reset) whenever UST rates have fallen below the loan commitment rate, subject to conditions in the tax-exempt bond market (typically highly correlated at about 80% of UST yields) and their perception of the value of other WIFIA loan features (likely to be relatively minor under current law).

• Disbursement will be delayed whenever the borrower’s short-term financing rates are below the WIFIA loan commitment rate. Since the UST yield curve is typically positively sloped, the delay may be lengthy, possibly right up to the maximum allowed, one year after project completion.

The exercise, or non-exercise, of these contractual options is predictable under circumstances that are not idiosyncratic but related to major observable macroeconomic metrics that the financial staff of large public water systems can be expected to use.

Importantly, the impact of option exercise on average WIFIA loan cost is also predictable:

• Exercising the cancellation option of an ‘out-of-the-money’ loan commitment will obviously have no cost impact, other than the return of previously apportioned subsidy to the Program. But the non-cancellation of an ‘in-the-money’ loan commitment and subsequent disbursement may result in material funding losses (i.e., the present value of loan debt service is lower than the funded principal amount at then-current UST discount rates). The option causes the loan to have an asymmetrical outcome with respect to funding cost. If UST rates rise after loan commitment, disbursement will incur funding losses. But if they fall, funding gains will not be realized because the commitment will be cancelled (or reset). For example, if within a certain period (say, three years), UST rises and falls are equally likely with a range (say, one standard deviation from the long-term average), it should be assumed that losses at disbursement will be incurred 50% of the time, with no commensurate gains. Even though UST interest rates are unpredictable, the asymmetrical outcomes of option exercise (or non-exercise) will have a predictable and quantifiable cost on average, similar to the way that loan credit losses can be predicted and quantified.

• The exercise of the disbursement delay option will increase the time between loan commitment and drawdown, increasing the potential difference between the UST curve at loan commitment (which sets the loan’s rate) and at disbursement (which determines the funding cost). Combined with the cancellation option, the delay option will increase the average funding cost of the loan if or when disbursed. This increased cost can also be quantified, using historical interest rate patterns (e.g., frequency and slope of yield curve) and past borrower behavior (e.g., typical disbursement delay).

Now, let’s parse through the relevant language in Circular A-11 ‘contract’ above and see if it’s applicable to the fact set outlined above:

• The ‘estimated cash flows’ used for calculating the NPV of the loan, both at loan commitment and at disbursement are obviously forward looking and include predictable contingencies. Estimates for credit losses are not based on the loan’s non-default at commitment, but on what is expected to happen on average in the future. Likewise, the fact that the loan’s interest rate at commitment will not (by definition) cause an NPV shortfall doesn’t mean that an average shortfall won’t occur or cannot be predicted at disbursement, even if interest rates themselves are unpredictable.

• The requirement that the estimates ‘include the effects’ of a list of contingencies means that anything that could affect the NPV of a loan must be considered, excepting only administrative costs. There are no other explicit qualifiers or limitations, though materiality can be assumed. I think this simply says, “if can you foresee and quantify the effect, you need to include it” in the estimate. That would be consistent with basic FCRA principles.

• One effect is specifically mentioned: “…expected actions by…the borrower within the terms of the loan contract, such as an exercise by the borrower of an option included in the loan contract.” Note that the word ‘option’ is not capitalized or defined within A-11 and is therefore applicable to the substantive optionality in a WIFIA loan contract, regardless of whether the word ‘option’ is used or not. The ‘expected actions’ by a typical highly rated WIFIA borrower would include (1) the exercise of the cancellation option on out-of-the-money commitments, but the non-cancellation of in-the-money commitments, and (2) the exercise of the delay option. Both of these actions are, on average, predictable and will have material effects on loan cost, as described above.

Bottom line? I think if A-11 was a legal contract, OMB would be obligated to estimate (or cause the Program itself to estimate) the cost of WIFIA loan optionality at loan commitment execution. With the estimate, OMB would then be obligated to apportion for the cost against Program budget authority under FCRA law and the Anti-Deficiency Act. It should be noted that there’s no evidence that OMB has ever estimated this cost in their apportionments FY18-FYTD26.

Of course, A-11 is not a legal contract with the WIFIA Program or anyone else. It’s OMB’s own guidance. Guidance for what? I assume primarily to manage all the numbers arising from FCRA law for budgeting purposes. But perhaps also to demonstrate how their processes are in compliance with the ADA? if so, there might be some legal aspects to this after all.

---

Imagine a Federal Program in which an Eligible Participant can bet on a coin flip. The Program pays $1 if it shows heads, but the Participant pays $1 to the Program if tails. [1] The Program is thought to be essentially costless for taxpayers because while individual coin flips are unpredictable, it is assumed that the fair coin has the usual 50/50 probability, and in time payouts and receipts will net to zero. [2] Of course, the Program will need Permanent Indefinite Authority to incur Mandatory Appropriations to pay Participants if there’s a run of heads in the short term — that’s ok, they’ll be balanced by gains from tails eventually. [3] There’ll also be a careful apportionment of a very small amount of Discretionary Appropriation for costs to cover Program admin and in the unlikely event that a Participant fails to pay up after a run of tails. Everything is done strictly the by book. [4]

Eligible Participants, who are large creditworthy entities capable of undertaking big projects, aren’t initially interested. They don’t gamble. But then their sophisticated Coin Flip Advisors analyze the legislative fine print and realize that under most circumstances, the flip can be cancelled if the coin shows tails whereas payout is certain if it shows heads. [5] Moreover, they think that such a cancellation would be a bit embarrassing for the Program, and that political pressure might be brought to bear on the Program to re-execute the flip (a ‘reflip’) whenever it comes up tails. [6]

Soon Eligible Participants are lining up at the door. For obvious reasons, they don’t exactly advertise their precise motivation, and nobody asks — they’re eligible, after all. The Program looks like a smashing success.

At first, there’s a brief run of tails. But these are reflipped into heads. Then there’s a longer run of heads. The payouts at this point are large — over ten times the Program’s Discretionary Appropriations for the period. However, the cost is covered by off-budget Mandatory Appropriations buried in arcane accounts. If questions are asked, the huge payouts can simply be explained as an ‘unpredictable run of bad luck in the short run that will be balanced by commensurate gains in the long run’. [7]

The Program comes up for reauthorization. The Eligible Participants’ lobbyists do their job. They use the Program’s Official Budget Numbers, and everything looks great. The reauthorizing legislation passes without any discussion or debate — why should there be any? [8]

After ten years, a Serious Audit is done in connection with an impending federal bankruptcy. The Audit finds that the Program consistently made transfer payments from taxpayers to Eligible Participants in annual amounts approximately five times the Program’s Discretionary Appropriations on average, all through Mandatory Appropriations. No material gains were recorded. It was also unclear what, if any, real-world policy objectives were achieved by the coin-flipping Program.

The Auditors ask the Program’s Budget Oversight Officials some awkward questions:

• If the asymmetrical payout built into Program legislation was understood by the Coin-Flip Advisors, why was this not also understood by Budget Oversight Officials? It was clearly possible, per Program statutes. Why did you assume the Program was essentially costless?

• The asymmetrical payout was obviously the primary motivation for Eligible Participants to use the Program. Why else would they use the Program? Why did you not ask the Program to investigate and report on Participant motivation as part of the annual reviews required under your Budget Circulars?

• As the Mandatory Appropriations built up to a very large amount relative to Discretionary Appropriations, why did you not re-examine your assumptions?

• Did you rely on the off-budget and automatic nature of PIA Mandatory Appropriations to obscure what was happening? If so, why? [At this point, the Auditors advise the Budget Officials to be careful in their answers, noting that violations of the Anti-Deficiency Act carry both civil and criminal penalties.]

It is not known how the Budget Officials responded. Shortly thereafter, the impending federal bankruptcy caused a major financial crisis and economic collapse. In the civil disturbances that followed, issues arising from the federal Coin Flip Program were considered unimportant.

______________________________________________________________________________________

Notes

[1] A WIFIA fixed-rate loan commitment is executed at a UST rate several years before it is drawn and funded at then-current UST rates. Interest rates are essentially unpredictable in that time frame — UST rates might rise (the commitment rate is an in-the-money ‘heads’ relative to the borrower’s alternatives, a funding loss for the Program) or fall (the commitment is an out-of-the-money ‘tails’, funding gain). A coin flip.

[2] UST rates are cyclical and revert to the mean, so over time rises and falls will roughly balance out. In theory, therefore, Program funding losses and gains will likewise balance out, if the ‘payout’ is equally certain for losses and gains.

[3] FCRA budgeting bases a loan’s cost on its net present value as discounted by the then-current UST curve — (1) the PV of the loan’s debt service (less expected credit losses) compared to (2) the PV of the Treasuries that would hypothetically be issued to fund the specific loan, the debt service on which would be exactly matched to the loan’s debt service. For loans that have a delayed drawdown, part of the challenge is to separate the ‘noise’ of random interest rate movements from the true ‘signal’ of the loan’s cost for policy purposes. FCRA technical re-estimates accomplish this with PIA to cover ‘random’ funding losses, which are expected to be balanced by equally ‘random’ funding gains over time. The same idea works for random runs of heads or tails in the Coin Flip Program.

[4] In WIFIA Program, credit loss reserve is a miniscule 0.71% of loan commitment amount — Program admin costs are about half that, 0.30% of commitments. Nearly free, right?

[5] At WIFIA, the legislative print isn’t actually very fine — explicit and often advertised by the Program. But I’m sure that sophisticated water system financial staff and their advisors understood the implications and did the numbers in the context of their tax-exempt bond alternatives. Since the bonds had near-UST rates highly correlated to the Treasury curve, these borrowers could cancel a loan commitment that was even slightly out-of-the-money (a ‘tails’) because the bond alternative was better.

[6] WIFIA did the first resets (essentially, a re-flip of the interest rate coin) in 2020 after some large borrowers with commitments executed in 2018 at higher rates ‘suggested’ them. I have no doubt that the threat of cancellation was implied, and perhaps some political pressure.

[7] WIFIA reset the ‘tails’ outcomes from 2018-2020 and thereafter UST rates more-or-less steadily rose to current levels, a run of ‘heads’ and funding losses for the Program. Losses totaled $2.1b for the whole period FY22-FY25, about ten times the discretionary appropriations. No one seems concerned, at least not publicly.

[8] This is where we are now — I expect that ‘rubber stamp’ reauthorization is most likely outcome. But…maybe not.

From written testimony of witness appearing on behalf of AMWA at the House hearing yesterday, 2/24/26:

---

You can’t blame the witness. EPA WIFIA itself puts forth the standard narrative about ‘low-cost’ loans and job creation, but that’s boilerplate exaggeration for federal programs and likely to be understood by Congress in that light.

More harmful is the soundbite, also part of the standard narrative, that a ‘modest federal investment can support an exponential amount of direct credit assistance’. What Congressperson can resist this siren call? A trivial amount of budget authority somehow produces a large number of large-scale, high credit quality federal loans that deliver ‘savings’ and ‘jobs’ and ‘clean water’ and all the rest — like magic!

But it’s not magic. EPA WIFIA is in fact writing unhedged call options embedded in loan commitments. The cost is trivial — until it’s not:

---

Again, you can’t blame the witness or even the water lobbyist she’s representing — they’re all just doing their jobs. Or Congress. You could see at the hearing that a lot is going on in the US water infrastructure sector and the many federal programs and policies involved with it. No one has time to track down what might be actually happening at a relatively small loan program. Instead, they all rely — with complete justification — on official federal budget numbers for advocacy and policy decisions. The oft touted ‘100-1’ WIFIA leverage is indisputably based on official federal budget numbers for executed loan commitments and apportionments.

Of course, a few lobbyists, a few Congresspeople and probably more than a few finance staff at big water systems understand how WIFIA loans actually work. But it is not their job to question the official numbers and the narrative soundbites that accurately summarize those numbers for advocacy and policy purposes.

Well, whose job is it? Are they doing their job? Or is $2.1b in funding losses an indication that they might not be?

Water lobbyists’ 2018 letter opposing sub-UST rates for SRFS

---

As discussed in this LR WFM article, the sub-UST rates proposed in the 2018 SRF-WIN bill might have been effective in encouraging smaller SRFs to start leveraging their portfolios of small project loans. Prudent leverage can expand their lending capacity and reduce reliance on annual federal grant capital for new loans.

But the proposal was shot down by the water lobbyists representing large municipal systems. A central argument — maybe the key one — was that sub-UST rates are ‘costly’ in terms of WIFIA’s credit subsidy loan ratios. Compared to what? Possible positive outcomes for both small SRFs and federal taxpayers as local resources are utilized more efficiently? No, nothing so banal. The lobbyists compared the cost only to the ‘magical’ ratios that lending to their membership appeared to offer. 100-1! Nearly free!

Except the ‘magic’ was based on bad budgeting. The value of WIFIA loans to Aa3/AA- water systems with excellent tax-exempt financing alternatives was primarily in the loans’ embedded interest rate call options, the uncapped cost of which would eventually surface when the loans were funded. Since this near-certain cost was not considered by EPA WIFIA or OMB in the apportionment of the loans’ credit subsidy, (perhaps, though not necessarily, in the conscious expectation that FCRA authority to incur off-budget mandatory appropriations would cover it) these very attractive options looked ‘free’.

Actual WIFIA outcomes FY18-FY25 show that the embedded options are far from ‘free’ for federal taxpayers. The $2.1b in mandatory appropriations over the period were due to rising UST rates and the consequent ‘exercise’ of the call options through loan drawdown. Since it is extremely unlikely that WIFIA will realize $2.1b in gains when UST rates are falling because highly rated borrowers will not draw ‘out-of-the-money’ loan commitments, the $2.1b of cost can be considered permanent for the FY18-FY25 portfolio.

‘Permanent’ as in a $2.1b transfer payment from federal taxpayers to the water ratepayers of those communities lucky enough to have a large, highly rated water system with sophisticated financial staff and smart advisors. Like AAA-rated Silicon Valley Clean Water, which has secured multiple WIFIA loans. Perhaps these ratepayers didn’t exactly need a federal subsidy, but I’m sure they appreciated it.

One wonders what $2.1b transferred to small SRFs through sub-UST SWIFIA loans to encourage and facilitate more efficient utilization of their capital might have accomplished by now?