Warehouse Worked Example

All amounts in this article are in US dollars (USD). Every figure is invented for teaching. None of it is a market benchmark, a quoted price, or a real transaction. The point is to show how the pieces of a warehouse interact, so that when a term sheet lands on your desk you can rebuild the same arithmetic with the actual numbers in it.

Three terms recur throughout, so fix them first. Par means you pay the face value of the receivables, with no discount or premium. Eligible receivables are the subset of the pool that satisfies the facility's stated criteria — obligor concentration, seasoning, documentation, delinquency status, and whatever else the schedule lists. Assets you own that fail those tests are still yours; they simply do not count toward borrowing capacity. The advance rate is the percentage of eligible collateral the lender will lend against.

Day one: who put in what

You buy USD 10m of receivables at par. Your capital provider reviews the pool against the eligibility criteria and agrees that USD 9m qualifies. The advance rate is 80%. The initial borrowing limit is the lesser of the USD 12m commitment and 80% of USD 9m eligible collateral, so USD 7.2m, and you draw all of it.

That leaves USD 2.8m of the purchase price for you to fund. An 80% advance rate on eligible collateral is not 80% of what you spent. It is 80% of the slice the facility agreed to count, and the gap between purchased and eligible collateral is equity you supply.

On timing, funding and purchase can settle simultaneously. Your equity and the lender's advance can move into the same closing settlement rather than sequentially, so you should not assume you must own the entire pool before any lender money is available. What matters is that your contribution is real money committed at settlement, not a later top-up. The mechanics of that settlement are covered in how draws actually work.

Two costs land at closing on top of that. An arrangement fee of 1% on the USD 12m commitment is USD 120,000, and legal costs are USD 60,000. Assume you pay both in cash at closing rather than netting them from the first draw. Your opening cash contribution is therefore USD 2.98m.

One more thing about the USD 12m. It is a commitment, not a bank balance you can spend. In this example, the advance must be supported by eligible collateral acquired before or as part of the agreed settlement, so the unused USD 4.8m is capacity waiting for future assets. A separate USD 3m accordion sits alongside it, optional and uncommitted, and it is excluded from both the USD 12m and the unused-fee base unless and until someone actually commits it.

Year one: the full cost stack

Hold the book flat for a year. Assume average collateral of USD 10m and average drawn debt of USD 7.2m, with replenishment replacing runoff so both averages stay constant through the period. Nothing here models principal amortisation, taxes, hedges, additional reserves, or operating-company overhead. It is a simplified constant-average model, not a cash-flow forecast.

Income: the collateral throws off a gross cash yield of 12%, so USD 1.2m.

Costs for the year:

• Borrowing interest at 8% on USD 7.2m of average drawn debt: USD 576,000
• Servicing at 1% of collateral: USD 100,000
• Admin, a flat annual amount: USD 30,000
• Unused fee at 0.5% on the undrawn USD 4.8m: USD 24,000
• Arrangement and legal, charged once at closing and expensed in full in this teaching year rather than amortised across periods: USD 180,000

Total first-year costs: USD 910,000.

Credit losses sit separately from facility costs, because they are not a facility charge and are never added into the cost stack here. At an annual net credit loss of 2% of collateral, that is USD 200,000 charged against annual cash income.

Income of USD 1.2m, less USD 910,000 of costs, less USD 200,000 of losses, leaves a first-year surplus of USD 90,000.

What the result actually tells you

Three readings, each with a different use.

Cost per dollar of debt. First-year costs of USD 910,000 against USD 7.2m of average debt is 12.6389%, before losses. That is over four points above the 8% headline interest rate, and the scope includes servicing and admin alongside the pure financing charges. When you compare two term sheets, the interest spread is the part everyone negotiates, and it is not the whole picture. The full fee inventory walks through which charges to look for and how to avoid double-counting synonymous ones.

Year two, if nothing changes. The USD 180,000 of arrangement and legal cost is a one-time closing charge, taken entirely in year one and not spread forward. Strip it out and recurring costs fall to USD 730,000. Same income, same losses, and the surplus rises to USD 270,000. Year one is not representative of the steady state, and a structure that looks marginal over twelve months can look different in its second period. That argues for judging a facility over its full expected life rather than its first year.

Simple return on your cash. USD 90,000 of first-year surplus against USD 2.98m of opening cash contribution is about 3.02%. Treat this as a single-period ratio and nothing more. It is not an IRR, not a money multiple, and not an investor return forecast. It ignores timing, assumes a flat replenished book held at constant averages, and says nothing about exit or renewal.

Where the structure breaks

Hold every other input fixed and ask how much credit loss the facility absorbs before the surplus disappears. Income of USD 1.2m less costs of USD 910,000 leaves USD 290,000, which is 2.9% of collateral. Your assumption was 2%. In this fictional case the headroom is 90 basis points of collateral in year one. It widens once the one-time upfront costs drop out.

Two separate stresses are worth testing. Do not combine them unless you rebuild the whole calculation explicitly.

Downside A: losses double. Net credit losses run at 4% of collateral instead of 2%, so USD 400,000, with everything else unchanged. Income USD 1.2m, costs USD 910,000, losses USD 400,000, and the year closes at a USD 110,000 deficit. Loss performance is one reason capital providers spend serious diligence time on loss curves. Getting that data defensible is the core of what you need before you approach anyone.

Downside B: the book comes in half-size. Origination disappoints and collateral averages USD 5m rather than USD 10m. Eligible collateral is USD 4.5m, debt at 80% is USD 3.6m, and your equity before fees is USD 1.4m, with opening cash of USD 1.58m. The commitment stays at USD 12m and the upfront and admin amounts do not shrink.

Income falls to USD 600,000. Interest falls to USD 288,000. Servicing falls to USD 50,000. But the unused fee rises to USD 42,000, because the fee is charged on commitment less drawn debt and there is now USD 8.4m undrawn. That fee moves inversely with utilisation rather than sitting fixed. Admin stays at USD 30,000 and the one-time upfront stays at USD 180,000. Costs total USD 590,000, losses are USD 100,000, and the year closes at a USD 90,000 deficit.

Cost per dollar of debt jumps to 16.3889%. Nothing went wrong with credit. The facility was sized for a book you did not build, and the fixed costs had a smaller base to sit on. Oversizing a commitment is not free optionality, and that tradeoff belongs in the conversation covered in negotiating terms.

Push it further. If you draw nothing at all, you still owe the arrangement fee, the legal costs, the admin amount, and unused fee on the entire commitment under these fictional terms. Cost per dollar of debt is undefined, because the denominator is zero. Which of these charges appear, and on what base, is a matter of the specific document — read it rather than assuming any particular fee applies or is waived.

The collateral test is a separate question

One snapshot, deliberately kept apart from the annual model. Suppose eligible collateral falls from USD 9m to USD 8m while drawn debt is still USD 7.2m. The borrowing base now supports USD 6.4m, and you are USD 800,000 over the limit.

Eligibility deterioration is not the same thing as realised credit loss. Assets can fall out of eligibility for concentration breaches, missing documentation, delinquency triggers, or age. Some of that becomes loss and some does not, so this USD 800,000 shortfall does not get added to the 2% annual loss assumption. What matters is that the shortfall must be fixed, and whether you do that with cash, with substitute collateral, over what period, and with what consequences if you cannot, is written into the documents.

This example is framed around a US-style structure; terms, remedies and legal treatment vary by contract and jurisdiction, and your counsel decides what your specific cure rights actually are.

What to carry into a term-sheet comparison

The lesson is that a warehouse layers largely fixed costs onto a variable-income book. Interest scales with what you draw. In this model, arrangement fees, legal costs and admin remain fixed, while the unused fee rises as drawings fall. At the planned balance, the model leaves USD 90,000 after the stated costs and losses, before omitted expenses. Miss the plan, or take more loss than you underwrote, and the same structure turns negative without anything dramatic happening.

Rebuild this arithmetic with real numbers before you sign, and ask:

• What is my cost per dollar of debt, including every fee rather than just the spread?
• What does the second year look like once one-time upfront costs fall away?
• At what loss rate does this stop working, and how far is that from my base case?
• What does this cost if I only reach half my origination plan?
• What is the fee position if I draw nothing for a quarter?
• What happens if eligible collateral falls while my debt stays put, and what specifically are my cure rights?

If you can answer all six from the document rather than from a conversation, you understand the facility. The sequencing from term sheet through to first cash is covered in getting to close.