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	Typical sizing
Debt service reserve (DSRA)	A temporary cash shortfall meeting debt service	6 months, sometimes 12, of principal and interest
Maintenance reserve (MRA)	Scheduled maintenance the operating budget cannot absorb	Forward-looking, typically 12–24 months of the maintenance plan
Major overhaul / augmentation reserve	Large periodic capital events	Accrued to the forecast cost, on a schedule
Decommissioning reserve	End-of-life obligations	Accrued over asset life to the estimated cost
Insurance deductible reserve	The uninsured first loss on a claim	The deductible, or a multiple of it
Tax reserve	A lumpy cash tax liability	The forecast liability

Not every structure carries all of them. The DSRA is near-universal; the maintenance reserve is standard on any asset with a serious maintenance plan; the others appear where the specific risk is material.

How Is Each One Sized?

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

Backward-looking: the DSRA. Six or twelve months of debt service is a duration, not a risk calculation. The question it answers is "how long can the project be down before it misses a payment," and six months is the market's judgement of how long a serious but recoverable outage takes to fix. It is not derived from the cash flow distribution and does not claim to be.

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 unambiguous and the practice is inconsistent.

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.

The breakeven LC fee — the rate at which the two are equivalent — is the spread itself, 8%. Any LC priced below eight hundred basis points beats cash funding on these assumptions, which is every letter of credit in any normal market by a very large margin.

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.

Neither refinement changes the direction of the answer in any realistic case. They change its size, which matters when the letter of credit facility is scarce and the question becomes which use of it earns the most.

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.

Where Does "Six Months" Come From?

Convention, and it is worth asking whether the convention fits the asset.

Six months of debt service is the standard DSRA across infrastructure, and it originates in a world of large rotating machinery — a gas turbine, a hydro unit, a process plant — where a serious mechanical failure means a long outage, a bespoke replacement part, and a repair measured in quarters.

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

Solar is modular. A utility-scale solar plant has thousands of near-identical components and no single point of failure whose loss stops the whole plant. The realistic bad outcome is a partial derate, not a six-month total outage — and the exceptions that would cause one, such as a transformer failure or a fire, are insurable and have their own recovery paths. On that analysis six months of full debt service is generous.

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 degrades rather than fails. A battery's bad outcome is usually capacity loss requiring augmentation, which is a capital event with a lead time rather than a sudden outage. That argues for a larger augmentation reserve and possibly a smaller DSRA.

The general point is that a DSRA is sized against a duration-of-outage assumption that is almost never stated. Stating it is worthwhile: "six months" is a conclusion, and the premise behind it should be a specific view about how this technology fails and how long recovery takes. On that basis a solar sponsor has a real argument for four months and a wind sponsor may deserve nine.

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 lenders simply require it. Particularly for the first period after completion, or where the sponsor is unrated. This is a credit judgement rather than an analysis, and it is negotiable more often than sponsors assume — a common compromise is cash for the first two or three years, converting to an LC once an operating record exists.

Inertia. The least good reason and a common one. Reserve funding is drafted from a precedent, nobody computes the carry, and the structure is agreed because it looks standard.

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. Full funding of all reserves is almost always a distribution condition. A drawn reserve therefore blocks distributions until restored, independently of whether the coverage test is met — which is why the lock-up post identified reserve restoration as one of the conditions worth negotiating.

With the DSCR. The reserve should not normally be included in the coverage calculation, because a ratio that passes because a reserve was drawn is measuring the wrong thing. It is commonly included in LLCR and PLCR, which measure available value rather than operating performance.

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. A reserve reduced or released once the project has demonstrated a track record — three years of coverage above a threshold, say. Less common, more valuable than a step-down, and usually available only where the sponsor asks.

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 segregated, charged account holding typically six or twelve months of principal and interest, available to meet debt service if operating cash flow falls short. It is the last line before a payment default.

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?

On the arithmetic, yes — the breakeven fee is the full spread, 8% on these assumptions, and no LC is priced anywhere near that. The constraints are facility capacity, the issuer's credit, and whether the lender will accept one.

Should the DSRA be included in the DSCR?

No. A coverage ratio that passes because a reserve was drawn is measuring the wrong thing. The reserve is commonly included in LLCR and PLCR, which measure available value rather than operating performance.

Is six months the right size for a renewable project?

It is a convention inherited from large rotating machinery, where a serious failure means a long outage. Solar is modular and arguably needs less; wind carries serial-defect risk and arguably needs more. The number should follow a stated view on how the technology fails and how long recovery takes.

Can a reserve balance be released early?

Sometimes. A DSRA maintained at six months of current debt service falls as the facility amortises, releasing cash steadily — which is materially better than one fixed at six months of original debt service. Ask for the first; precedents often provide the second.

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 series 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: Pivotal180 — What Is a DSRA? · Breaking Into Wall Street — Debt Service Reserve Account in Project Finance · Forvis Mazars — Features of a Cash Flow Waterfall in Project Finance · LexisNexis — Project Finance Financial Covenants