A Credit Model That Starts With Cash
Why a credit model should be built from cash collected rather than reported earnings, walked through line by line down to what the debt can actually bear.
Earnings are an opinion. A careful one, reached under a thick set of accounting rules, but an opinion all the same. Someone decided when a sale counted as earned, how many years a machine would last, which costs belonged to this period and which could wait for the next. None of that makes the number wrong. It makes it a view about the business, and a lender is never repaid with a view. A lender is repaid with money that has actually arrived, after the business has paid what it needs to keep running. So when I build a credit model, I start at the cash and work down one line at a time, and I do not let the income statement choose the order.
Line one is cash collected from customers
The first line is not revenue. It is what customers paid during the period, which is revenue adjusted for the change in what they still owe. In a steady business the two sit close together and nobody notices the difference. In a business that is stretching payment terms to win volume, or booking sales that will be collected slowly, the gap opens quietly and then all at once. I want that gap visible on line one, because every line beneath it inherits the error if it is hidden. When collections trail reported sales for several periods running, the income statement is describing a company the bank account has not met.
Line two is cash paid to run the business
Next come the operating outflows actually paid: wages, materials, rent, freight, insurance, the utility bill. Again the point is payment, not expense. A company can report a tidy margin while paying suppliers later each quarter, which flatters cash for a while and then reverses when the suppliers push back. I separate recurring costs from items management describes as one time, and I keep the one time items in the model anyway. A charge that appears every year under a new name is a cost of doing business. The line has to show what leaves the account, whatever the footnote calls it.
Line three is working capital, the quiet borrower
Inventory, receivables and payables together behave like a revolving loan the business makes to itself. When sales grow, that loan grows too, and it consumes cash before any profit shows up. When sales shrink, the loan unwinds and releases cash, which can make a declining business look briefly flush. I model working capital as its own line with its own drivers, days of inventory and days to collect, rather than a plug at the bottom. A lender who ignores this line tends to discover it during the one year it matters most, when growth is financed from the very cash that was supposed to pay interest.
Line four is the spending that keeps the business whole
Some capital spending is optional and some is not. The part that is not, replacing worn equipment, maintaining the plant, keeping the fleet running and the software supported, belongs above the debt in the model, because skipping it only borrows from the future. Management will often present one figure for all capital spending and then describe most of it as growth. I try to estimate the maintenance share independently, from the age of the assets and the depreciation history, and I treat that share as a fixed cost. Growth spending I leave for later, as a use of cash the lender can argue about. A useful check is to compare depreciation with what the company has actually spent on upkeep over several periods. When upkeep runs persistently under depreciation, the assets are aging faster than they are being renewed, and the bill is waiting somewhere ahead.
Line five is everything that must be paid before the lender
Between operating cash and the debt sit obligations that behave like debt without being called debt. Long leases are one. Pension contributions, earnouts owed to former owners, deferred payments on an acquisition and contractual minimum purchases are others. Each has a claim on cash that arrives on a schedule and rarely waits. I list them individually rather than netting them, because their timing matters as much as their size. A company can look comfortably covered on an annual basis and still run short in the quarter when a large deferred payment falls due alongside a seasonal build in inventory.
Line six is cash available to service debt
What remains after the first five lines is the number the whole model exists to produce. It is the money the company could hand to its lenders in a period without starving itself. I compare it with the reported earnings figure the deal was marketed on, and the reconciliation between the two is often the most useful page in the file. Where they agree, I relax a little. Where they diverge, the divergence usually has a name, such as slow collections, deferred maintenance or a lease that was left out, and naming it is most of the analysis.
Line seven is what the debt actually asks for
Only now does the debt enter. Interest is the obvious claim, but scheduled amortization is just as real, and so is any required cash sweep. I set the full contractual demand against line six period by period rather than as a single annual ratio. Interest coverage on reported earnings can look generous while principal payments quietly exceed what the business throws off. The comparison that matters is cash available against cash demanded, and the gap between them, in each period, is the margin of safety. If that gap is thin in the base case, no covenant package will make it thick. Covenants can warn a lender early and give it a voice early, but they do not create cash. They only change who holds the pen once the cash runs short.
Line eight is the same model under stress
The stress case uses the same lines in the same order, which is the discipline of building it this way. I do not cut a single top line percentage and let everything else follow. I slow collections on line one, hold costs sticky on line two, let working capital absorb cash on line three and keep maintenance spending fixed on line four. Businesses under pressure rarely shrink neatly. They collect late, keep paying staff they hope to need again and still have to replace the forklift. Stressing each line separately shows which one breaks first, and that is the line I ask management about.
What the order protects against
Building from cash down takes longer than starting at earnings and adjusting. The extra time is the point. Every assumption has to be stated where it belongs, and nothing can be buried in a single adjustment row labeled normalization. When someone asks why the lender is comfortable, the answer can be traced line by line from a customer’s payment to a coupon, and every step can be argued with on its own terms. A model built this way is also easier to update, because each new quarter lands on a line with a clear meaning rather than on a figure that already absorbed a dozen quiet decisions.
A lender gets paid from the bank account, so the model starts where the bank account starts.
That is why the top of every credit model I build reads the same way. Not revenue, not earnings, not an adjusted figure someone else prepared. Line one is cash collected from customers, and every other line is there to explain how much of it can reach the lender.