Google-Gemini-‘Discussion-on-possibility-of-splitting-EPA-WIFIA-InRecap-011026-1

Download PDF

---

Okay — this is a radical idea. But the more I thought about it, I began to see that something needed to be done, a decisive action, a bold but discrete first step. So, I ‘discussed’ the idea with Google Gemini. Of course.

In olden times, before 2024, your ancestral self would sketch out a new idea using scrawled written notes and a lot of Google searches about various aspects and potentially relevant facts. Now that we have advanced beyond such primitive behavior, we simply ask Google Gemini to run around the web and summarize the whole thing in neat, bullet-pointed, grammatically perfect and (apparently) logical text.

Very handy. But a new skill is absolutely required — the acute and immediate detection of slop and hallucinations. That’s not so easy in highly technical matters — you have to know what you’re doing. But to be fair, the essence of that skill was required for pre-2024 idea development by scrawling and searching as well, albeit in a different sequence. Then, you crossed out a lot of just-written nonsense (“what was I thinking?”) and dismissed dead-end search results after a quick review. With Gemini output, it’s all dumped upfront and you refine the idea by removing the dross.

There’s a fair amount of slop and hallucination in the above PDF transcript (e.g., the split and move to Interior ‘has been discussed’ — WTF?) but a lot of the output appears useful and benign, at least as a guide for early thinking. I’ll leave it to the reader to decide — I left the transcript unedited.

Back to the idea itself, here’s a rationale for a split that I asked Gemini to ‘consider’ (page 3 of the PDF transcript):

I didn’t see any showstoppers in the silicon’s answers. So, perhaps more carbon-based thinking about this radical solution to WIFIA reform is warranted.

---

On the surface, the above article essentially reprises, in more formal and explanatory language, this recent post — The Zeldin Letters.

On the next level, the article is a continuation of efforts to cut through the WIFIA ‘narrative’ and present some facts about what might really be happening at a bellwether federal infrastructure loan program. Not so much to ‘report the truth’ (I’m not a reporter, much less an idealist) but because truth matters for effective policy design. Like a consumer product, you first need cold science and detailed engineering to make the thing work as intended. Only then can you let the marketing people loose with more-or-less truth-adjacent narratives in order to sell it.

As described in the prior two WFM articles, Explaining the Decline in WIFIA Loan Volume: Part 1 and Part 2, the WIFIA loan ‘product’ appears to be failing, certainly in terms of its promising start, but also more importantly in terms of its potential effectiveness in helping federal policy address US infrastructure renewal. Fixing this faulty ‘product’ is a group effort (federal finance is a national public good, after all), so I think it’s useful to ‘disassemble’ it in public — not to criticize, but as the basis for improvement.

There is a third level. For various reasons, I think WIFIA’s mandatory spending issue should start to be surfaced in public, but that’s not so easy to do if the topic is described in a standalone, abstract way. The wretched FCRA Criteria had this problem — as I think its proponents craftily intended — of being impossible to criticize because nobody really knows what it is. Floating above criticism or even widespread understanding, a distortion of federal budgeting law or loan program policy intent (harmful to most but useful to some) can be sustained indefinitely.

The ‘reset risk’ trope helps bring WIFIA’s mandatory spending issue back to earth and at least be understood, if not dealt with:

• The ‘risk’ topic itself is practically clickbait for those WIFIA borrowers who might be relying on the reset for their interest rate management strategies. Looking at you, last five WIFIA loan commitments executed at about 4.8%.

• A discussion of how the reset works introduces the cost of FCRA interest rate re-estimates in a comprehensible way. Treasury borrowing at 5% to fund a WIFIA loan asset earning 2% is obviously a loser for taxpayers, no?

• The actual intentions of Aa3/AA- borrowers getting a WIFIA loan commitment in the first place can be highlighted. ‘Sophisticated’ borrowers using an ‘interest rate option’ to hit a ‘low point in an interest rate cycle’ sounds more like a description of Wall Street traders than simple Main Street folks with few financing alternatives using a ‘low-cost’ federal loan to build a ‘critical’ project for thirsty local kids, as the standard narrative would have it.

• And speaking of ‘narratives’ — could it be that that “simplistic narratives about this issue” are “ugly” because they are both straightforward and true? Do you see it now? Do you? Okay, then let me spell it out: Narrative games can work both ways.

Sector_In-Depth-Water-and-Wastewater-Utilities-08Jan2026-PBC_1467169-Inrecap-comments-010926

Download PDF with InRecap Comments

---

This report makes a big deal about the Trump 2.0 delays, but the arguments that these caused, or could cause if repeated, actual project construction inception delays or higher costs for ratepayers are pretty…slight. The report is more a declarative ‘narrative’ (the pitch is summarized in the title) than the kind of detailed and fact-based analysis you’d expect from Moody’s.

It struck me later that a WIFIA report with an emphasis on Trump 2.0 delays was perhaps requested or suggested by the credit agency’s water authority clients as part of the ‘pressure campaign’ described in The Zeldin Letters post a few days ago. If that’s the case, I think the analysts did the best they could with scant evidence of any past or future material impact of the delays.

Still, a few interesting points:

• The report acknowledges, though it doesn’t highlight, the facts that WIFIA volume has consistently declined since 2022 and that new annual appropriations for the Program don’t get fully utilized. The chart showing WIFIA’s rise and decline 2018-2024 speaks for itself.

• For once — finally! — there is a direct comparison to tax-exempt bonds as the baseline against which WIFIA’s constantly touted ‘low cost’ should be measured. As usual, undiscounted numbers and some high-side approximations are used in the quick comparison. Even so, the numbers, described narrative-style as ‘significant cost savings’, aren’t exactly compelling. At all. Which kind of undercuts the whole ‘costly Trump 2.0 delay’ narrative, no? I’m sure the analysts knew this — but business is business and the client is always right. We’ve all been there.

• Much better was the discussion about how a WIFIA loan’s non-rate features (slower amortization, deferral, sculpting, etc.) help issuers to manage more affordable rate increases — including details about a great real-world example in the recently closed City of Joliet loan. I got the distinct impression that the analysts would have preferred to spend more time on this area if the ‘costly Trump 2.0 delays’ narrative wasn’t the report’s apparent priority.

• Overall, what comes out is this: The ‘costly delay narrative’ is unpersuasive and easily forgotten, but the discussion of WIFIA’s non-rate features hints at much more substantive things about how well-designed federal finance can facilitate infrastructure affordability. That’s something that Moody’s with its huge base of experience and data in the public sector could expand on very constructively, including considering WIFIA amendments for the 55Y loan term and small project combinations. Wouldn’t the analysts have more fun with that?

Lot of numbers in the chart above, but for now just focus on the discounted present value of debt service difference (Diff DS PV) for the 55Y and 35Y term cases. That’s how sophisticated issuers would compare long-term financing alternatives (e.g., precisely how muni bond issuers evaluate an advance refunding using DBC software packages).

Current Law: 35Y Post-Construction Term — Good Enough for Reset Games

Tax-exempt bond rates are currently pretty good compared to UST yields (e.g., 30Y UST of 4.85% vs. MBIS Index of 3.63%, for a 74% ratio), so it’s no surprise that under current law, a 49% WIFIA Loan and 51% Bond combination is underwater compared to a similar straight bond. The estimated amount is about 1.6% of project cost — negative, but not by much.

Still, the overall case for a WIFIA/Bond combo might be positive when the value of non-rate features is included. But most of all — by far — is the fact that the WIFIA loan commitment rate can be reset at some point, perhaps years, in the future.

If the reset remains available, the current rates and NPV comparisons don’t matter that much compared to what might become available in the next year or so. Well, let’s see — election year monetary stimulus? Geopolitical crisis liquidity flood? Deflationary recession after the AI bubble bursts? Trump wildcard of some sort — oh I don’t know, something crazy like launching his own quantitative easing through the home mortgage GSEs?

In all these cases, UST yields are likely to fall faster than bond rates — at which point, the reset can capture the value. If things go the other way, the worst case is you’ve got the WIFIA commitment in hand, effectively capping a minor downside.

I think the reset features explains much of why there was rush by WIFIA applicants to close ASAP at the end of 2025 — i.e., when the prospect of an especially ‘interesting’ 2026 was becoming clearer — despite relatively unfavorable numbers for a WIFIA loan.

Of course, this strategy depends on the reset remaining available — another wildcard, as recently discussed in this post, The Zeldin Letters, and (with a good deal more diplomacy) in this WFM article, Is WIFIA’s Interest Rate Reset Feature at Risk?.

Amended Law: 55Y Post-Construction Term — Calmer Waters

Now consider the case if a 55Y post-construction loan term was available in the WIFIA loan. There’s a solid NPV benefit of 4.5% of project cost, all due to the fact that UST post-30-year ‘flat forward’ rates apply to more of the project loan balance, something that market-traded, primarily retail-owned tax-exempt bonds can never match.

For a long-lived project, the ability to capture some of the UST flat-forward curve would be a unique and sustainable capability of a 55Y WIFIA loan. The attitude towards seeking a WIFIA commitment could change — the value is economic and fundamental, ultimately based on intrinsic federal strengths as a lender.

Sure, WIFIA’s interest rate management features would still look good, but they wouldn’t be the central focus, and there’d be more applicants that can’t’ play the arbitrage game (e.g., large, one-off water management and flood control projects that are too idiosyncratic to get good rates or even access to the bond market, maybe someday small project combinations, and so on). WIFIA would have a substantive — and defendable — purpose and role in US water infrastructure renewal, with demonstrable additionality and measurable policy outcomes, not just false narrative. Isn’t that what WIFIA stakeholders should really want as we proceed into ever more turbulent times?

House-Senate-Conference-Approp-01052026-CWIFP-Section

Download PDF

Update 430pm: House approves minibus spending bills in bipartisan 397-28 vote

Text here as (I think) passed: HR 6938. No changes re the CWIFP and WIFIA aspects discussed in this post, so I’m guessing the ‘long march’ to WIFIA reform remains, well, long. But that’s okay — fundamental reform is never easy, and it wouldn’t be any fun if it were.

---

As you’d expect, there’s a lot going on in Congress’ spending bills. But for now, I’m narrowly focused on the fate of two federal loan programs, WIFIA and WIFIA’s Army Corps section, CWIFP.

CWIFP Non-FCRA Ineligibility

In a recent prior post, Senate S.3293’s Non-FCRA Provisos and HR. 6229’s FCRA Amendment — Good Faith Policy or End Run? I expressed some concern about provisos in the December Senate spending bill that effectively made CWIFP cost share loans to dam and levee projects with any federal ownership at all ineligible, regardless of the FCRA Criteria (which, of course, currently still apply).

That means that if even if some type of FCRA amendment passes and the wretched FCRA Criteria are revised or superseded, dam and levee projects with any amount of federal ownership are just flat-out ineligible for CWIFP cost share loans using FY26 appropriations.

Well, most of these ‘non-FCRA ineligibility’ provisos survived, as highlighted in the PDF above, abstracted from House Rules Committee spending package released on January 5, 2026. As discussed in the prior post, there may be a valid policy rationale for the restrictions, and it’s important to note that federal ‘ownership’ is clearer and effectively less restrictive than the FCRA Criteria’s nebulous ‘federal involvement’ or arbitrary disqualification based on past history and Congressional authorization. But still…it’s hard to understand.

Perhaps in all the non-stop federal tumult these days, no one was focused on these eligibility nuances for such a small federal loan program, and the Senate bill’s language just got added without much review or discussion? That’d be understandable — but I’d note that one highly technical proviso amending the definition of ‘project’ for CWIFP got dropped in conference, and someone took the time to cut $2.8m of couch change out of the $5m that the Senate bill originally proposed.

WIFIA Pointless Funding

Elsewhere in the Rules Committee’s I&E package, the EPA’s WIFIA got about $65m of new appropriations. Um…why? The program already has about $220m of ‘until expended’ carryover funding, enough for ten years of loan volume at the current pace. And no policy riders or amendments that would expand WIFIA’s capabilities to pick up that pace were included — just the money.

So — let me get this straight:

• You cut $2.8m of couch change from a dam & levee loan program that could probably use more funding if its original pre-FCRA Criteria statutory eligibility were restored but also ensured with repeated provisos that the original statutory eligibility is never restored for the current funding?

• You give $65m to a loan program which doesn’t need it, a program that’s primarily used by highly rated water agencies as a near-substitute for tax-exempt bonds in ‘occasionally useful’ interest rate management, and which has become less popular because tax-exempt rates are better, all while racking up huge mandatory spending for interest rate re-estimates — and you don’t change a thing?

Okay. I’m hoping this is a ‘work in progress’. But if it’s the final outcome for FY26 spending legislation, then there’s a whole lot more to do.