Reserve Accounts: What They Protect Against, and What They Cost to Hold

A six-month debt service reserve on a $150m facility is about $7.98m of cash, sitting in an account, earning a deposit rate, for the life of the loan.

Equity does not value that money at a deposit rate. It values it at its own cost of capital, and the gap between the two is a carry cost of roughly $638,000 a year — $4.63m in present value over an eighteen-year facility.

The same reserve backed by a letter of credit at 150 basis points costs $120,000 a year, or $867,000 in present value.

The difference is $3.76m, and it is decided by a line in the term sheet that most sponsors treat as administrative.

This post covers what each reserve protects against, how each is sized, the cash-versus-letter-of-credit arithmetic, and why lenders sometimes insist on cash despite it.

ℹ️ Note: All figures are labelled assumptions. The carry cost depends entirely on the spread between the sponsor's cost of capital and the deposit rate, which varies by sponsor and by rate environment.

What Is a Reserve Account?

A segregated account holding cash against a specific future obligation, funded ahead of distributions and released only on defined conditions. It converts a potential future liquidity shortfall into a present cash requirement — which is precisely what a lender wants and precisely what it costs equity.

All reserves sit inside the security perimeter, which means the cash is charged as well as segregated. That is what distinguishes a reserve from a prudent cash balance.

The Reserve Stack

Reserve	Protects against	Sizing basis to extract
Debt service reserve (DSRA)	Temporary cash shortfall meeting debt service	contractual Required Balance; can be months of debt service or another formula
Maintenance reserve (MRA)	Scheduled maintenance / capital maintenance	approved maintenance plan and timing
Major overhaul / augmentation reserve	Large periodic capital events	forecast event cost and funding schedule
Decommissioning reserve	End-of-life obligations	legal / contractual obligation and escalation assumptions
Insurance deductible reserve	Uninsured first loss	policy deductibles and financing requirements
Tax reserve	Lumpy tax obligations	forecast tax liability and payment timing

Not every structure carries all of them, and even the DSRA can be funded in different forms. Vistra disclosed a 2024 project financing with a $29 million debt-service-reserve letter-of-credit facility, while other filed agreements require cash balances in a controlled DSRA.

How Is Each One Sized?

Three different logics, and conflating them produces reserves that are the wrong size in both directions.

Contractual formula: the DSRA. The required balance is a credit term, not a universal outage-duration rule. One filed project facility required the next three months of principal and interest; another requires ongoing deposits through the cash waterfall until a defined Required Balance is reached.

Forward-looking: the maintenance reserve. This one is derived from a forecast — the maintenance plan, with its scheduled interventions and their costs, and the reserve funded so the cash is there when each falls due. It is therefore only as good as the maintenance plan, and a plan that understates the frequency or cost of major interventions produces a reserve that is systematically short.

Accrual: decommissioning and overhauls. Funded on a schedule that accumulates to a known future amount by a known future date. The sizing question is the amount and the date; the funding question is the profile, and a straight-line accrual to a cost that is itself escalating will undershoot.

The augmentation case from the waterfall post is the hard one: certain in principle, uncertain in timing, driven by how hard the asset has been cycled. Tying the funding rate to measured throughput rather than to elapsed time is the structure that handles it, and it is more drafting work than a flat annual contribution.

Cash or Letter of Credit?

The arithmetic is straightforward once the assumptions are explicit, but the funding choice is not universally one-sided.

Assumptions, labelled as such:

Annual debt service                             $15,952,917
DSRA at 6 months                                 $7,976,458
Sponsor cost of capital                                 12%
Deposit rate on reserve balances                         4%
Letter of credit fee                                   1.5%
Facility tenor                                     18 years

Cash funding:

Annual carry cost   7,976,458 × (12% − 4%)       =    $638,117
PV over 18 years at 12%                          =  $4,626,135

Letter of credit:

Annual fee          7,976,458 × 1.5%             =    $119,647
PV over 18 years at 12%                          =    $867,400

Difference: $3,758,735 in favour of the letter of credit.

Under the worked assumptions, the simple breakeven LC fee equals the assumed 8% opportunity-cost spread. That is not a market statement. The actual comparison must also reflect LC reimbursement collateral, commitment / fronting fees, issuer replacement risk, sponsor LC capacity, deposit yield and whether reserve cash can earn permitted investment income.

Two refinements are worth making because they narrow the gap without closing it.

The deposit rate is not always 4%. In a high short-rate environment the spread compresses and cash funding becomes less punitive; in a zero-rate environment it widens to nearly the full cost of capital. The comparison is therefore rate-environment dependent, and a structure agreed in one environment may be the wrong one in another — which is an argument for the flexibility to switch, negotiated at the outset, rather than for picking correctly once.

The cost of capital is the right discount rate only if the capital has an alternative use. A sponsor with a pipeline of projects competing for equity genuinely forgoes 12% by locking $8m in an account. One holding surplus cash with nowhere to deploy it forgoes considerably less, and for that sponsor the comparison is much closer.

Those refinements can change both the size and, in some cases, the direction of the result. The worked example is a capital-allocation framework, not a conclusion that an LC always dominates cash.

And the sizing matters on the same terms: moving from a six-month to a twelve-month DSRA costs a further $4.63m in present value if funded in cash, and about $867,000 if backed by a letter of credit. A lender asking for twelve months rather than six is asking for something whose cost depends entirely on how it is provided.

What Does the Required Balance Actually Protect?

Whatever shortfall period and credit risk the lenders chose to protect through the contractual Required Balance.

Do not reverse-engineer a universal technology story from a six-month number. Public financings show three-month balances, six-month assumptions, dynamic formulas and LC-backed facilities. The useful diligence question is why this required balance is appropriate for this revenue profile, outage risk and payment frequency.

Renewable assets fail differently, and the difference cuts both ways.

Solar failure modes vary. Module- or inverter-level failures can create partial derates, while transformer, substation, interconnection or casualty events can create larger outages. Reserve sizing should therefore use the project's actual single-point-of-failure analysis, spares strategy, insurance waiting periods and repair lead times rather than assuming solar always justifies a shorter DSRA.

Wind has serial-defect risk. The failure mode that genuinely threatens a wind project is not one turbine but a design or manufacturing defect affecting the whole fleet — a gearbox, a blade, a main bearing — requiring a campaign across every unit. That is exactly the scenario a six-month reserve was designed for, and arguably is not long enough.

Storage has both degradation and outage risks. Capacity fade can create planned augmentation needs, while PCS, transformer, fire, control-system or grid events can create abrupt revenue interruptions. Model the augmentation reserve and debt-service liquidity separately.

The general point is that reserve size should be connected to the cash-shortfall risk it is intended to absorb. Technology failure mode is one input, but revenue seasonality, insurance waiting periods, payment dates, maintenance exposure and lender risk appetite can be equally important.

What If the Reserve Is Never Drawn?

Which is the usual outcome, and it is the right way to frame the cost.

A DSRA exists for an event that, for a well-built contracted asset, mostly does not happen. The cost, however, is certain: $4.63m of carry over the facility, paid whether or not the reserve is ever touched.

That makes it an insurance premium, and it should be assessed like one.

Certain cost of a cash-funded DSRA                = $4,626,135
Loss it protects against                          =  a payment default
Probability over 18 years (assumption)            =          3%

At a 3% probability, the project is paying $4.63m to avoid a 3% chance of an event. Whether that is good value depends entirely on the consequence — and the consequence of a payment default in a non-recourse financing is not a missed payment, it is acceleration, enforcement and the loss of the whole equity position.

So the correct comparison is not $4.63m against the missed payment; it is $4.63m against 3% of the entire equity value. On the numbers from earlier in this series, 3% of $115m is $3.5m — which makes the reserve roughly fair value in cash and comfortably good value as a letter of credit.

That framing does two useful things. It explains why lenders are unyielding on the DSRA while being flexible on much larger items: they are pricing a tail, not a expected loss. And it confirms that the sensible sponsor response is not to fight the reserve but to fight the funding method, because the protection is worth having and the carry is not.

So Why Does Anyone Fund in Cash?

Four reasons, and two of them are good.

The LC facility is finite. This is the real constraint and it is a portfolio problem. The interconnection post described queue deposits consuming letter of credit capacity; the PPA credit support post described post-COD security doing the same; and now reserve accounts want it too. A developer running several projects can exhaust its LC facility long before it exhausts its equity, and at that point cash funding is not a choice.

LC provider credit is a real exposure. A letter of credit is only as good as the bank issuing it, and lenders impose rating requirements on the issuer with a replacement obligation if it is downgraded. That obligation is a contingent call on the sponsor at exactly the wrong moment in a credit cycle.

Some financings require cash rather than substitute credit support. Others permit an LC or a switch between forms subject to issuer and replacement conditions. Treat the permitted form as a document term, not a convention.

Precedent can obscure cost. A reserve copied from an earlier deal may be defensible, but the sponsor should still quantify its carry cost, funding form and release profile before accepting the precedent.

How Do Reserves Interact With Everything Else?

Three connections that matter and are frequently modelled separately.

With the waterfall. Reserve funding sits below senior debt service, which as the waterfall post established means it is below the CFADS line and does not affect the coverage ratio or the sized debt. It reduces distributions, not leverage.

With the lock-up. Restricted-payment conditions may require specified reserve accounts to be at their Required Balance before distributions resume. Check the actual test rather than assuming every reserve deficiency blocks equity cash.

With the coverage ratios. Whether reserve cash enters DSCR, LLCR or PLCR is a definition question. Reproduce the finance-document formula exactly and show reserve draws separately so operating weakness is not hidden.

The sequence in a deteriorating project runs through all three: coverage weakens, distributions stop, the maintenance reserve is drawn, the DSRA is drawn, and only then is a payment missed. Each reserve is a step on that ladder, and the total length of the ladder is what a reserve package actually buys.

When Can a Reserve Be Released?

Less often than sponsors expect, and the release conditions are worth as much attention as the sizing.

On final repayment. Universal and uncontroversial: once the debt is repaid the DSRA has nothing to protect and is released in full. That is eighteen years away and contributes almost nothing in present value.

On a step-down. Some facilities reduce the required balance as the debt amortises — maintaining six months of current debt service, which falls as the balance falls. This is the most valuable release mechanism available and is worth asking for explicitly. On a sculpted facility with declining debt service, a reserve maintained at six months of current service releases cash steadily throughout the term rather than in a single payment at the end.

On a performance trigger. Some facilities permit a step-down or release after specified performance conditions; if so, model the precise test date and amount released.

Never, in practice, for the maintenance reserve. A forward-looking maintenance reserve is topped up as it is drawn and as the plan rolls forward, so it is a permanent balance rather than a temporary one. The release comes only at the end.

Two drafting points repay attention. A DSRA fixed at six months of the original debt service, rather than current service, is a materially worse deal on a sculpted or amortising facility — the reserve stays large while the obligation it protects shrinks. And release conditions that require no default subsisting plus all other reserves fully funded can prevent a release that the sizing rule permits, which is the same interaction the lock-up post described.

How Do You Model Reserves in Excel?

As balances with funding and release rules, and with the carry cost reported as its own line.

The reserve mechanics

Required(t)     = per the sizing rule for that reserve
Funding(t)      = MAX(0, Required(t) − Opening(t))      [from the waterfall]
Drawing(t)      = MAX(0, Obligation(t) − Available_Cash(t))
Closing(t)      = Opening(t) + Funding(t) − Drawing(t) + Interest(t)

Distribution_Test = AND(all reserves fully funded, coverage met, no default)

The carry cost, which almost no model reports

PF_ReserveCarryCost(t) = Cash_Balance(t) × (Cost_of_Capital − Deposit_Rate)

PV over the facility, cash-funded                = $4,626,135
PV over the facility, LC-backed                  =   $867,400
PF_LCSaving                                      = $3,758,735

The breakeven that settles the argument

Breakeven_LC_Fee = Cost_of_Capital − Deposit_Rate    = 8.0%

Any letter of credit priced below that beats cash. Stating the breakeven is more useful than comparing two specific prices, because it survives a change in either.

The sizing sensitivity

DSRA 6 months,  cash                             = $4,626,135 PV cost
DSRA 12 months, cash                             = $9,252,270 PV cost
DSRA 6 months,  LC                               =   $867,400 PV cost
DSRA 12 months, LC                               = $1,734,800 PV cost

ℹ️ Note: Reserve balances earn interest, and that interest belongs to the project. A model that holds a reserve balance and ignores the interest income understates CFADS slightly every year — small, but it is free and it compounds over eighteen years.

To build the reserve stack with funding rules, carry costs and the LC comparison, prompt Dezzmond with your debt service profile and maintenance plan.

What Do Sponsors and Lenders Actually Check?

• Is the DSRA cash-funded or LC-backed, and has the carry cost been computed?
• What is the breakeven LC fee, and is the available pricing below it?
• What else needs the LC facility — interconnection deposits, PPA credit support, other projects?
• Is the maintenance reserve derived from the maintenance plan, and is that plan credible?
• Does the augmentation reserve fund against throughput or against elapsed time?
• Is reserve restoration a condition to distributions, and how quickly can it be achieved?
• Does the model credit interest earned on reserve balances?

Frequently Asked Questions

What is a debt service reserve account?

A lender-controlled reserve or permitted substitute credit support maintained at the Required Balance defined by the finance documents and available for specified debt-service shortfalls. Filed financings use different durations and formulas.

What does a cash-funded reserve cost?

The spread between the sponsor's cost of capital and the deposit rate, for the life of the facility. On a $7.98m reserve at a 12% cost of capital and a 4% deposit rate that is $638,000 a year, or $4.63m in present value over eighteen years.

Is a letter of credit always cheaper?

No. In the worked example an LC is cheaper because its assumed annual fee is below the assumed opportunity-cost spread on trapped cash. The actual comparison must include LC fees, collateral / reimbursement terms, issuer requirements, deposit yield and sponsor capital constraints.

Should the DSRA be included in the DSCR?

Use the contractual ratio definition. For internal analysis, also report operating coverage before reserve draws so a liquidity support mechanism does not mask deterioration in the underlying project cash flow.

Is six months the right size for a renewable project?

Not by default. The finance documents set the Required Balance. Filed project facilities include shorter and differently defined reserves, so the appropriate size should be tested against the project's actual cash-shortfall risks and payment profile.

Can a reserve balance be released early?

Sometimes. If the Required Balance is linked to future debt service, it can decline as the debt amortises; other facilities use a fixed or differently defined amount. The release profile should be modelled from the actual formula.

How is a maintenance reserve sized?

Forward-looking, from the maintenance plan — the scheduled interventions and their costs, funded so the money is there when each falls due. It is only as reliable as that plan, and an optimistic plan produces a systematically short reserve.

Closing: A Cost That Never Appears as a Cost

Reserve accounts are the quietest expensive thing in a project financing. No payment is made, no charge appears in the profit and loss, and the money is still there — it is simply not available.

That makes the cost genuinely easy to miss. A model that funds a reserve and holds the balance shows no expense at all; the carry appears only in the equity IRR, as a return on capital that was committed and could not be distributed. On an eighteen-year facility it is four and a half million dollars, on a reserve most people would describe as a $7.98m item.

The comparison with a letter of credit settles it arithmetically and does not settle it practically, because letter of credit capacity is a scarce resource with three or four competing demands in any active developer — queue deposits, PPA security, reserves, and whatever the next project needs. That is a portfolio capital allocation question and it is the right frame: the question is not whether an LC beats cash on this reserve, which it does by a wide margin, but which of the competing uses of a finite LC facility earns the most.

Answering that requires computing the carry cost of every cash alternative, which almost nobody does. The number to start with is the breakeven fee — the spread between the cost of capital and the deposit rate — because it converts a comparison of two prices into a single figure that survives both of them changing.

the next post assembles everything this material has covered into the document it all lands in: sources and uses, and why the two sides balancing is the weakest check in project finance.

Sources: Vistra 2024 filing — $29m debt-service-reserve letter-of-credit facility · SEC-filed facility — three-month DSRA requirement · SEC-filed facility — ongoing DSRA funding through the cash waterfall to Required Balance · 2026 Project Swift loan-guarantee agreement — replacement / release of cash in DSRA and acceptable LC mechanics · Dezzmond analysis; worked-example carry-cost assumptions are illustrative.