Financial assurance

The obligation outlives the company that owes it

Almost every other risk in a project has a counterparty who is still around when the risk lands. Decommissioning does not. The obligation matures at the moment the asset stops earning, which is also the moment the owner has the least money and the least reason to perform it, and the owner is a special purpose vehicle whose entire design is to hold one asset with no recourse to its sponsor.

Ring-fencing is what makes non-recourse debt possible. It is also what leaves a landowner or a county holding an unsecured claim against a company whose only asset is the plant it now costs money to remove. Financial assurance is the answer to that: put the money, or a third party's promise of the money, outside the project company before it is needed.

The timing is the whole design problem. The cost lands in year twenty or later. The entity that can comfortably pay for it is solvent in the middle years. Whatever instrument is chosen has to survive a sale of the project, a refinancing, a change of operator and the insolvency of the party that posted it, and it has to be callable by a beneficiary who is not a bank and has no project-finance team.

Where the requirement comes from, and it is not one place

There is no federal decommissioning-assurance requirement for grid-scale storage on private land in the US, and no EU-wide equivalent either. In US practice the duty usually arrives from one of three directions: a county or municipal zoning ordinance that makes assurance a condition of the conditional-use permit, a state statute where one has been enacted for energy facilities, or the land lease itself, where the landowner negotiated security for removal at option signing rather than waiting to see what the county would ask for.

On federal land the right-of-way grant carries its own bonding regime administered by the land agency. In Europe the pattern is national and municipal — planning conditions and lease terms — with the EU Battery Regulation running alongside as producer responsibility for the batteries themselves. That is a different mechanism entirely, and it does not pay to take out foundations or restore a field.

The consequence for a developer is that this is a site-specific diligence item, not a policy you carry between projects. Two counties in the same state can require different instruments, different sizing bases and different re-estimation cycles. Find the requirement before the layout is frozen, because the shape of the obligation — day-one funded against phased, gross against net of salvage — moves real money in the model and stops being negotiable once it is a condition of approval.

Four instruments, and the questions that separate them

A surety bond is a three-party arrangement: the project company is the principal, a surety underwrites it, and the party relying on it is the obligee. It is cheap in cash terms, an annual premium rather than tied-up capital, and the surety underwrites the principal's credit, usually taking indemnities or collateral of its own. What the beneficiary gets is not a demand instrument. A surety can investigate a claim and can generally raise the defences the principal itself would have, so payment follows a process rather than a presentation of documents.

A standby letter of credit is the opposite trade. A bank undertakes to pay against conforming documents, independent of any dispute under the underlying obligation — the autonomy principle that ISP98 and UCP 600 are built around — so the beneficiary draws first and argues afterwards. The price is that the facility consumes the project's bank lines or is cash-collateralised, which is real capital.

The catch is tenor: no bank writes a twenty-year letter of credit. Standbys are commonly issued for a year with evergreen auto-renewal, so what actually protects a beneficiary over a project life is the right to draw the full face amount on notice of non-renewal. Read that clause before anything else in the instrument.

Cash in escrow, or a sinking fund built up on a schedule, removes third-party credit risk entirely, which is why jurisdictions like it and sponsors do not — it is the most expensive form to the project and the only one where the money is genuinely gone. How it is held decides how good it is: an escrow with a third-party agent and the jurisdiction named as beneficiary is a different animal from a reserve account inside the project company that a liquidator can reach.

A parent-company guarantee is cheapest and weakest, converting the question into the parent's credit twenty years out when the parent is frequently a fund with a defined life, and many jurisdictions will not accept one on its own. Four questions separate all of them for whoever has to rely on it. Does it pay on demand or after a process? Does it expire, and what happens if it is not renewed? Whose credit stands behind it, and will that entity exist at removal? And is the money inside or outside the project company's estate if the project company fails?

Sizing is where the disputes are

The number in the instrument is a cost estimate for work nobody has done yet, and the first argument is the basis. Gross removal cost is what it costs to take the plant out and restore the site. Net of salvage subtracts the expected scrap value of copper, steel, aluminium and whatever the batteries recover. The developer wants net, because it is a smaller number that ties up less capital.

The beneficiary wants gross, because a salvage credit is a twenty-year commodity forecast written into a security instrument as a deduction, and if it is wrong the shortfall appears exactly when there is nobody left to make it up. Jurisdictions split on this. Some allow the credit, some allow it subject to a floor, and some refuse it outright. Establish which applies before the financial model assumes one.

Who produces the estimate is the second argument, and the usual answer is an independent estimate stamped by a licensed engineer rather than a developer's spreadsheet, with the scope tied line by line to the decommissioning plan: module removal and dangerous-goods packaging, transport, treatment or disposal fees, transformer fluid handling, foundation removal to the stated depth, restoration, and the gen-tie easement.

The estimate then has to be re-run, because a single figure fixed at financial close is stale within a few years. A workable structure names a re-estimation cycle, names who performs it, and provides for the instrument to be stepped up when the number moves — and, less commonly, stepped down.

Escalation is the third argument. The estimate is in today's money and the work happens two decades out, so an instrument that is not indexed shrinks in real terms every year it sits. The fixes are an index applied to the face amount, a step-up schedule agreed at signing, or a sinking fund whose contributions escalate on the same basis.

Phasing raises the mirror-image problem. Staged funding is far cheaper than posting the whole amount on day one and jurisdictions often accept it, but the failure the instrument protects against is a project abandoned early, before a back-loaded schedule has built up anything to draw on. A funding curve concentrated in the last five years covers the case that worries nobody.

The beneficiary is not your lender

Every other security instrument in a BESS project runs to the lenders: the share pledge, the charges over accounts and contracts, the direct agreements. This one runs the other way. Its beneficiary is the landowner or the local jurisdiction, which makes it the single piece of security in the stack that the lenders cannot draw on, do not control, and generally treat as a cash-flow obligation to be covenanted rather than as an asset.

Where funding it sits in the payment waterfall is therefore a live negotiation. Above debt service and it keeps getting funded in a bad year; below it and it is the first thing to stop being funded when the project is in trouble, which is the same year the beneficiary starts to care.

It also has to be usable by someone with no project-finance capability. A county holding a bond for twenty years needs a written trigger it can act on — what counts as abandonment, typically a stated period with no operation, plus a notice-and-cure period for the owner — and a draw procedure that does not require litigation to start.

It needs the instrument to name it correctly and to survive a change of ownership, so assignment and successor language matters more here than in documents that get amended at every refinancing. And it needs someone to watch the calendar: an instrument that quietly lapses because nobody tracked a renewal date is worth nothing, which is why the non-renewal draw right in a standby letter of credit is the clause doing the actual work.

The last thing to write down is how it ends. Release should be tied to defined completion evidence out of the decommissioning plan — disposal and recycling certificates for the batteries, a survey confirming foundations came out to depth, a soil test where the restoration standard calls for one — and it can be staged, with the face amount reducing as work is verified in phases. An assurance with no defined release mechanism outlives the site it was posted for: the field is restored, and the instrument is still running because nobody wrote down what finished looks like.