A brief plea to the US Banking sector to make a pile of low-risk revenue while solving one of the energy transition’s dumbest, most stubborn inefficiencies – by David Riester
Background
The world doesn’t always function as it should. I don’t like the way airplanes are boarded. I think health insurers’ billing departments are designed to make “mistakes”. I bet we could do without notaries. But in this mad world drenched in dysfunction and entropy, absolutely nothing makes less sense to me than how American banks use (rather, don’t use) letters of credit in the renewable energy sector.
A couple foundational truths:
- There is a completely rational, logical, prudent, responsible way for banks to issue letters of credit at far greater volume, without it feeling like wasted ammunition from a “Basel III” perspective.
- The biggest market opportunity chasm (in one of the most aggressively pursued, crowded markets in the land, mind you) is best filled with “LCs”.
This is going to be a very simplistic proposal. Most readers will be saying or thinking “I’ve been saying this for years”. Indeed, same here. I’ve been telling any banker that’ll listen that there’s a ~$1b+ revenue opportunity just sitting there, waiting for even just one bank to evolve their thinking within the regulatory realities. For reasons I half-understand – yet still can’t believe continue to prevail – ships keep passing in the night, and this extremely important, high-growth sector remains inhibited by the failure of the banking sector to navigate institutional red-tape, regulatory hurdles, and deeply engrained behavioral patterns. Those phenomena explain the underutilization of LCs, but they don’t justify it. We need to figure this out together. Letter of credit use cases can be expanded in a manner that is “worth it” from a Basel III (2017 update) perspective, while satisfying bank risk management departments. The prize is a ~$1b revenue boost, with excellent risk adjusted returns on Tier 1 capital.
Current Forms of Financial Security
There are usually either 2 or 3 ways that a financial security can be satisfied. This can be found in most governing documents as “acceptable forms of financial security.” Go into OASIS and ctrl + F through a few of the relevant docs. You’ll find: i) cash is always an option, ii) LCs are always an option, and iii) surety bonds are sometimes an option. To set the stage, it’s important to clarify that an LC is a form of financial security (or “deposit”), but the opposite is not true, and the words are not interchangeable. As I (rather contemptuously) demonstrate in the Appendix, folks constantly conflate the two, which threatens to be the thing that finally nudges me over the edge into the abyss of complete madness. When someone who knows what they’re doing hears that conflation, it’s a dead giveaway that you are unsophisticated on this matter. With that in mind, lets hit the basics of letters of credit and surety bonds:
Letter of Credit ("LC")
A “Performance Letter of Credit” (or “Standby Letter of Credit”; same thing) is a financial instrument between i) an issuing bank (not a private equity fund, not an IPP, not a company; a bank) ii) an account party (whose obligation is being addressed with the LC), and iii) a beneficiary. Let’s define these:
- “Issuing Bank” = I’m going to say it again: Not a private equity fund, not an IPP, not a company… a bank. If a non-bank is saying they will write you a letter of credit, that should be an orangish-red flag. Someone is communicating disingenuously with you, presumably because they don’t think you understand the topic at hand. The game is afoot, and you’re the mark.
- “Account Party” (a.k.a. “applicant”) = the party whose obligation is being backstopped, usually the developer who has a financial security (or “deposit” if you prefer, but don’t say “LC”) to post.
- “Beneficiary” = the party requiring financial security or credit support of some fashion. In our world this is usually an ISO, a utility, an offtaker, or a swap[1] counterparty.
It’s a piece of paper – a very simple one – prepared by the issuing bank, describing what/whose obligation is being backstopped, and who will be paid cash if the obligation is not met. An LC usually costs around 1% of “face value” (the max amount that can be called by the beneficiary) per year. That’s very cheap, and LCs are pretty easy to put in place. If LCs are an option, you’d be pretty nuts to satisfy financial security requirements any other way.
That conditional “if” is the problem. Very few developers have the credit profile required to get an investment grade letter of credit. You essentially need to be of investment grade credit yourself, and even then you’ll probably be asked to post some collateral. My understanding is that the driver of this high bar is regulatory ratio requirements arising from the 2017 update to “Basel III”. In recent years, uncollateralized LCs burn up a lot of exposure runway afforded by a bank’s “Tier 1” capital, whereas collateralized LCs burn up 1/5 as much. So, it’s less about risk, and more about the opportunity cost of eating into the bank’s lending runway. This piece will go a little deeper on Basel III later, so we’re going to leave it here for now. But it’s important to acknowledge there are regulatory constraints in play that make this less straightforward than it may seem on the surface.
Whatever the reasons, the reality is there are maybe 20–30 established developers in the US who would meet the credit thresholds, and even those parties are often forced to post some collateral. Most of them are big name companies, usually vertically integrated or active in many sectors… often public. It is incredible how many companies in the energy transition are operating under the incorrect assumption that they will be able to secure uncollateralized LCs to satisfy deposits. Almost no one can.
Surety Bond
A surety bond is a binding contract entered into by the same three parties found in an LC, with different names. The “obligee” is the beneficiary – the party requiring the financial security. The “principal” is like the “account party” for an LC – the party needing to satisfy the financial security. The “surety” is the party backstopping the obligation and promising the obligee that, if the principal fails to deliver, the surety will make it right.
The main difference between an LC and a surety bond, is that an LC is thought to be easier to collect upon. A surety generally has a right to subrogation (an opportunity to get the result instead of sending a check). Between i) subrogation, ii) the payment terms (longer in surety), and iii) the burden to prove there was a breach by the principal, a beneficiary/obligee faces a higher probability that the party backstopping the obligation (surety) resists or delays payment. There’s also the reality that letters of credit escape traditional “contract law”, having their own set of legal principles – this is most relevant in the context of bankruptcy. All that said, I’ve never seen or heard of a surety failing to meet their obligations in the event of a legitimate draw on financial security satisfied by a surety bond. On a relative basis, it’s not quite as airtight as an LC… but on an absolute basis, it’s a very reliable form of financial security.
Sureties play a big role in the industry. It’s a red-tape laden, quirky world rife with irrational behavior and archaic rules dating back to early English common law. Nevertheless, many of them have an impressive grasp of this whole energy project deposits game and do a rather nice job of filling the LC gap as much as circumstances permit. They cost about 1.5–3.0%; so, they are quite affordable. They are more of a hassle to secure than LCs, but easier than closing a loan. Segue has a surety facility and we use it (for our investment partners’ benefit) whenever we can, to good effect. The rub with surety bonds is that many ISOs/utilities don’t take surety bonds as financial security.
Any fan of the energy transition should be trying to change that reality. The topic of why some utilities and ISOs don’t accept surety bonds is ripe for discussion, and should be included in any queue reform discussion. Some parties nominally accept surety bonds but set their requirements such that actually using them is wildly impractical (or outright impossible). SPP is a good example. They accept surety bonds, but they insist upon an impossibly demanding form of bond (and don’t budge from it), and then impose aggregation limits on different bond providers. NYISO is almost as bad, though a modicum more constructive. ERCOT doesn’t accept sureties, neither does most of MISO (except Ameren who is forward thinking on sureties). If a utility or ISO is serious about participating constructively in the energy transition, there’s no lower hanging fruit than accepting surety bonds as a form of financial security and working with the surety community to create a reasonable form of bond.
Financial Security Cost of Capital Spectrum
Deposit Loan
The final option for developers and project owners is posting cash raised from a third party. These days, that tends to cost around 12–18%. Did you just notice how much distance along the cost of capital spectrum we just leaped? That’s the market inefficiency underpinning this whole piece. Two years ago you could get this late-stage dev loan (for refundable deposits, or really late stage non-refundable) for, say, ~8–10%. But, then this part of the capital markets (private credit, specialty banks) tightened up due to economic conditions and the denominator effect. Now they have an “I don’t get out of bed for anything less than a 1.25x” attitude. The reasoning is simple: money is hard enough to come by right now, and there always seems to be someone asking for ~14% money for something that would garner much lower interest in a different market. Many of those opportunities are more compelling from a risk-adjusted return standpoint, so anything <15% just feels like it has too much opportunity cost. It’s not that the risks necessitate a better return, it’s that I can do better for the same amount of risk somewhere in the market, and, well… capitalism. In any event, even when this was 8%, that’s still a big gap there, not to mention quite a bit more work and legal costs to make this land, as compared to an LC or surety bond.
The rank order of attractiveness is clear and obvious. Here’s the developer “decision tree”, represented visually:
And, for my kindred, in Excel:
=if([LC AVAILABLE], LC, if(and([SURETY ACCEPTED],[SURETY AVAILABLE]), SURETY, if([DEBT AVAILABLE], DEBT, KILL PROJECT)))
How Should This All Be Working?
What fundamentally makes sense? That’s a subjective matter on which reasonable people might disagree, but I believe that i) refundable deposits, and ii) non-refundable deposits for which it’s very, very difficult to imagine scenarios where the security is forfeited… should be satisfied with a “cashless” solution (LCs or surety bonds), regardless of developer credit. Most non-refundable deposits, especially those posted for projects that still carry material attrition risk, should be met with cash raised from either high interest debt, or equity, as the risk profile of such deposits is materially the same as early/mid stage development expenses, which is to say… high.
(Note: often a financial security deposit is partially refundable, partially not; the same logic would apply such that “mixed” deposits are partially covered with LC/surety, partially with cash. We do this all the time, and it’s fine.)
What Is Actually Happening?
That’s how it should work. But in reality: i) beneficiaries too often insist on either cash or LCs, because they are not forced to accept surety bonds, and ii) banks that can issue LCs underappreciate the significance of a deposit being refundable, and generally take a completely binary approach to deciding if an LC will be issued.
Two wildly divergent capital costs may – and often do – apply to the exact same fundamental project circumstances. The same deposit might be satisfied with a 1% LC by one party, and 18% debt by another, where the only difference is that the first party has a A- rating, while the second merely has a $400m balance sheet and 10 years of operating history. Now, I’m not saying principal credit should be ignored – it’s important, no doubt – but a 17% gap is not the picture of an efficient, properly functioning market. Perhaps I’m barking up the wrong tree and should take that complaint to the rating agencies, yet… we have a huge market of lenders underwriting unrated, non-recourse project financing loans, using the “spread” to account for differences in underlying risk. This is not so different.
Nothing even vaguely resembling “pricing in risk” occurs in today’s LC market; almost all the “questions” in that decision tree are binary. You are either deemed creditworthy enough to get an LC, or not; if you are, the price is always the same. A surety bond is either an acceptable form of financial security, or it’s not. You are either deemed creditworthy enough to be the primary indemnitor behind a surety bond, or not; if you are, the price is always the same. Etc. There is no nuance… there are no shades of grey.
Here’s the Bank “decision tree”, represented visually:
That is why that chasm exists in the cost of capital/solution spectrum – a severe rigidity of approach that leads to the wrong strategic conclusions within the regulation-constrained environment.
Let's take a minute to talk about those regulatory realities, as I think doing so uncovers why the LC market is so exclusive. This space is mostly affected by lending ratio constraints originating from what’s called “Basel III”. Basel III is an international regulatory accord that came together after the global financial crisis. All the central banks got together to try and improve the regulation, supervision, and management of banks on a global basis. The main topic was “leverage ratios” (lending ratios) and reserve capital requirements. They agreed on some high-level principles and the participants in the summit have been implementing them ever since. There was a follow up session in 2017 that was meant to settle a few especially stubborn items. That session covered a lot of turf relevant to LCs, including some bewilderingly punitive elements.
Each nation is responsible for codifying their own specific regulations. In the US, the FDIC has been in charge. They released updated proposed rules in mid-September of this year, and here are the most germane morsels:
The main mechanism is maximum allowable leverage ratios after applying prescribed risk-weightings to different categories of instrument/risk, all measured against their Tier 1 (essentially cash and cash equivalents) capital. The risk factor (“Credit Conversion Factor”, or “CCF”) is really low for things like highly rated covered bonds (1:330) and much lower for positions that are i) less liquid, ii) longer duration, and iii) higher risk of default. As an example, at the other end of the spectrum is equity (1:13). Each category’s CCF determines how much of the bank’s Tier 1 capital base is used up by holding a position in that category. Letters of credit took a hit here, with a CCF of 100% for uncollateralized LCs, which means they are subject to a 1:33 ratio – toward the bad end of the spectrum.
It's best to think about this through the lens of opportunity cost. If you assume a bank’s Tier 1 capital is fixed – which it is in the immediate term – then every investment/loan/guaranty on the bank’s books burns a portion of a fixed pool defined by Basel III leverage ratios. So, banks must be very strategic when deciding what to put on their books.
If a letter of credit earns you 1% per annum while burning through quite a bit of your leverage ratio runway, it’s not an especially attractive instrument to lean into. Therein likes the problem… and the opportunity.
How To Build A Hugely Profitable Business Unit, In Two Simple Steps!
Thematically, the solution involves trading complete, uncompromising rigidity, for a touch of pliability. Plastic instead of steel.
What if the LC cost wasn’t 1%? What if it was meaningfully higher, such that every LC you write actually extends your leverage ratio runway, because it brings in more Tier 1 capital than the Tier 1 capital you burn by writing the LC ([LC Amount] ÷ 33)? The arithmetic suggests anything above a 3% fee would be accretive to the Tier 1 leverage ratio on a net basis.
Why not think about it in terms of pricing in risk by calibrating a “spread” based on assessment of the fundamental, underlying risk – just as lenders do when determining the spread for a project loan? The considerations are i) principal/guarantor credit, ii) refundability, and iii) asset (project) value and risk. Yes, this introduces operational friction and costs for a bank, but if you price the instrument properly, it can absolutely be worth a bank’s resources to apply a bit more thought and process to the underwriting.
Alas, real world experience suggests it’s not that simple. Either there’s a false premise in there, or something else is going on. I know this because I’ve made that case to about 20 bankers in the last 8 years, and it never goes anywhere. Here’s one such story:
Together with one of our developer partners, we pared down a pipeline’s universe of financial security deposits to the least risky bits – only fully refundable deposits held in escrow, where the lender could “hold the joystick” for calling that money back from the utility until their loan was repaid. This should probably cost something very close to the risk-free rate, maybe SOFR + 100. Our partner – the would-be borrower - proposed a 9% interest rate (way above SOFR + 100). A lender said they’d “take it back”. Then, as we pre-planned, the next day our partner preemptively shot the banker a note saying, “You know, if you prefer to get the same 9% by posting an LC (with the same duration) for us, that would work too.” In other words: an infinite IRR on a cashless investment, same risks. The banker said “Really?! Oh that would be really interesting for us. Let me pull in the LC group.” A couple days passed, and then the banker sheepishly asked us for our Moody’s credit rating (an absurd question to ask a brand-new development JV). Upon being rebuffed, he said the LC group couldn’t do anything for sub-A-rated credits.
So what happened there? One branch of a banking institution (the commercial lending or project financing group) was operating within a different framework than the other (the trade finance group). I believe the trade finance group i) bumped into institutional constraints designed to protect against the opportunity cost imposed by Basel III regulations, and/or ii) was not incentivized to do anything other than check their box (or not), and then write an LC (or not). In the trade finance groups of today, that seems to be the job.
But that setup – including the incentive structure – is built upon a false premise: that all letters of credit must cost ~1%. Had the bank agreed to write an uncollateralized letter of credit for a refundable deposit and charged, say, 7%, everyone would have been better off. They simply couldn’t. We’ve been operating with stale assumptions that haven’t been challenged for years, and a regulatory environment that stifles one’s imagination (“could I really charge an amount for an LC that changes the opportunity cost calculus of burning so much Tier 1 capital???”).
With that in mind, we need some thought leaders in the banking sector to rethink the role that letters of credit play in their portfolio, dropping long held assumptions. Additionally, banks will probably need to staff the trade finance group a bit differently – pulling in folks like those in project finance or commercial lending – who can think in shades of grey, apply creativity, and incorporate a business development mindset. The bank needs to empower them to price in risk, using basic logic and deduction in a system that can certainly remain rule based. That space needs to be created within a well-constructed risk management regime – lines that can’t be crossed, exposure limits to different beneficiaries, etc. You can set a pricing system that appropriately accounts for the realities of Basel III, maybe even expanding the pool of Tier 1 capital instead of burning through the existing Tier 1 capital runway. Bear in mind that LC fees are paid up front.
I appreciate this plan may contain a whiff of cowboyishness, but we are talking about very low risk scenarios. 5%–10% might normally be associated with hairier loans, but remember the only reason those numbers hunt is because the alternatives are poor – not because the underlying risk requires a higher handle. We shouldn’t be using this product for situations with lingering binary risk. Like before, the LCs would be written for very low risk scenarios. Same product (letter of credit), different price. Same risk, but for different reasons (refundability instead of guarantor credit).
What if the price of the LC tracked the credit of the ultimate indemnitor/guarantor, much like bonds of size $X’s get interest rates of Y% track the strength of the Company Z’s credit? Same thing, really: a year of $X of risk/liability exposure to credit Z makes you Y * X in revenue. For credits that are off the spectrum, and essentially worthless, an LC provider is underwriting the circumstances and security as opposed to the guarantor/indemnitor. In my mind, provided the LC is for genuinely low-risk applications like refundable deposits, a bank should still issue an LC for that, just at a high price point that reflects a belt-but-no-suspenders underwrite.
What price, you ask? Well… look at the giant gap in that cost of capital spectrum and pick your number. “But we can’t charge 8% for a letter of credit, that’s just not how it’s done!” you say? Why not? Try it. See what happens. I’m literally begging you.
Think About Credit In a Non-Binary Way
Ok fine, you don’t like the idea of just charging more for a weak or non-existent indemnitor credit, regardless of the fundamental risk. Let me try a different angle. Maybe you can introduce a whiff of nuance to the way you think about credit. Two examples stick out for me:
The first is cash collateralization. As it stands now, a creditworthy entity gets an LC with little hard cash collateral requirements, and a non-creditworthy entity gets an LC only if they post 100% cash as collateral. How about a partially cash-collateralized LC, with LC fees sticking a bit closer to the traditional price range? For a developer, this is not the dream, but it’s still far better than satisfying security deposits with cash raised at their corporate cost of capital (usually high-priced equity).
Maybe a bank creates three internal risk categories, with 25%, 50%, and 75% cash collateralization requirements. Let's say the 25% category is for decent size companies that have non-trivial amount of asset value and liquidity, yet falls short of true investment grade thresholds. 50% is for companies with asset values that are at least, say, 5x the LC face value. 75% is for companies that don’t meet that bar. Many ERCOT trading desks use this approach, and it seems to work just fine for them.
Illustrative Approach to Sliding Cash Collateral Requirement
That’s just a made-up example I conjured in 30 seconds. There are no “right” or “wrong” thresholds or prices. What does seem objectively wrong, however, is that the bail bond market offers alleged felons more flexibility and creativity than nine-figure trade finance opportunities.
The other potential approach to introducing a little bit of flex to the way a bank thinks about guarantor/indemnitor credit is credit enhancement products like a revolving line of credit. The savvier insurance company lenders have long used this to unlock LC access for their borrowers. A bank might adopt the strategy of offering LCs to entities that have a line of credit of at least X% of the LC amount, provided that the call conditions and mechanics of the revolver are well calibrated. From the developer’s perspective, the combined cost of i) that revolver, and ii) the LC, is still going to be less than the ~15% loan they’d otherwise fall back on. Lending institutions that like revolving lines of credit might take notice here. Perhaps partner up with a bank that issues LCs and work together to create a product like this. I’m only aware of one such product/partnership, and it exists primarily to satisfy operating phase security requirements where the bank offering the revolver is also the perm lender. More tailored two-party hybrid products should exist.
The Prize
Here’s some very basic market size analysis:
Addressable Market: I’m going to focus on sectors I’m certain have demand for these solutions – solar, storage, and wind – which will inherently underestimate the market size (which includes equipment, derivatives, other technologies, etc.). U.S. Dept. of Energy projections for 2023 capacity:
Solar = 32 GW
Wind = 107 GW
BESS = 10 GW
Total = 149 GW
Financial Security Assumptions:
- Average security deposits required = $0.05/W. We did some analysis across our project portfolio, including deposit needs related to IX, PPAs, purchase orders, and swaps. The range is $0.00/W – $0.37/W, the average is close to $0.05/W. It’s worth noting that with i) the nation’s electrical grid becoming ever more “picked-over”, and ii) interconnection reform (FERC and elsewhere), this number is more likely to increase than decrease.
- Average duration of deposit need = 3 years. In our portfolio the range is 0 – 7 years, with the average a little under 3 years.
So, in total an estimated financial security need of ~$22b, annually. Now, we need to multiple that by an assumed average LC cost/fee to get to the bank’s revenue. Here’s a spectrum of solar/wind/storage revenue across a variety of average LC costs:
My takeaway here is that, if a bank were to ascribe to my proposed “price in the risk” approach, the likely market size is $1b+, just for wind, solar, and storage. I don’t consider this an especially tough $1b to achieve, given it’s a relatively low friction, opex-light business unit. Yes, if done properly it absolutely uses up bank lending capacity, but, if we rethink pricing, these instruments can be net additive to leverage ratio bandwidth, revenue, and profits.
Conclusion
It’s time. Our sector has been fumbling around, failing to address an unnecessary bottleneck for far too long. No doubt, the realities of the market/regulations/governance/human-nature are such that things are not quite as simple as I’ve laid out. I appreciate that, and genuinely welcome feedback as to where my understanding of the banking sector is off-base. There are many reasons why, to date, LCs have not been a viable solution for most of the market. As laid out, I think many of them are silly and addressable, but some are more rational and/or stubborn. We should be talking about all these considerations. My goal in this piece is to “get it all on the table” to create a solid foundation from which to carve out solutions and make letters of credit more widely available for appropriate energy transition use cases, while also calling attention to the missed opportunities to leverage the surety bond market. People and institutions famous for their rigidity and risk aversion will need to adopt a slightly different approach/mindset. That’s hard to do, but banks can evolve… people can grow. Frankly, over the years I’ve been delightfully surprised at how the US banking sector has adapted to the conditions of the clean energy market. I see no reason why that can’t happen in the financial security, letter of credit space too.
Appendix
[1] Dear Developer: Don’t say “hedge,” it’s a dead giveaway you’ve never closed a fixed-for-floating electricity swap, because nowhere in that journey is the word “hedge” used. Big trading organizations with power desks and big books of positions can call it a hedge; you call it “a swap”, or an “ISDA” if you want to overcorrect the other way.
