How Customer Float Funds Growth and What Happens When It Reverses

Some businesses are funded by their customers. Cash arrives before the cost is incurred, invested capital shrinks toward zero, and growth pays for itself. The same structure is a margin call the moment growth stalls: the float that funded the business becomes a claim against it.

The mathematics are the point. Return on invested capital is profit divided by capital employed. Negative working capital attacks the denominator. Penman calls this operating-liability leverage: operating liabilities as a share of net operating assets, which mechanically lifts the return on that smaller base. McKinsey's valuation consequence is the same idea from the other side. Invested capital is operating assets minus operating liabilities. When the next dollar of growth needs almost no capital, the reinvestment rate in the continuing-value formula approaches zero. Growth is self-funded.

Buffett's version is the cleanest. Insurance float is cost-free capital whenever underwriting breaks even, measured as the underwriting result divided by average float. There is no guarantee it stays cost-free, and shrinking premium volume starts a "depressing sequence" of forced asset sales. The size of the float is not the tell. The cost of it is.

The cash comes from four families, and they do not last the same way.

• Customer float — premiums, deferred revenue, memberships, upfront service fees. Insurers, SaaS, KONE's service contracts, Costco memberships.
• Supplier leverage — extended payables won through buyer power. Walmart, Home Depot.
• Operating-model float — a negative cash-conversion cycle designed into the business. Amazon's collect-before-pay marketplace. Dell's build-to-order model in its prime.
• Franchise float — fees collected ahead of parent-company costs on long agreements. Domino's.

A COVID stress test ranked the durability: insurance float held and grew, SaaS deferred revenue mostly held, franchise float was next, and supplier leverage was the first to be withdrawn. Mixed-source firms deserve the most conservative reading. Amazon has at least three of the four.

The float is only quality when something else holds it in place. Without pricing power, switching costs, or scale, negative working capital is fragility wearing quality's clothes. Dell's advantage evaporated once competitors copied the model. Greenwald's point is sharper still: "spontaneous liabilities" (payables, accrued wages, taxes) lower the reproduction cost a new entrant has to pay. The same structure that lifts earnings-power value lowers the floor underneath it. Mihaljevic called this the holy-grail paradox — high returns on capital and balance-sheet protection almost never coexist, because the thing that produces the returns removes the downside. If the moat fails, there is nothing underneath. Customer prepayments are senior claims, not free money.

That is also why the factor shows up in returns. Wang, in the Journal of Financial Economics (2019), found that a low-minus-high cash-conversion-cycle portfolio earned roughly 5–7% annual alpha on U.S. stocks from 1976 to 2015, surviving a Fama-French five-factor adjustment and lasting more than three years after formation. German equities replicated it. After the usual post-publication decay, a realistic forward read is closer to 3–5%. Nissim and Penman (2003) showed operating-liability leverage predicts future return on equity even after controlling for current profitability. None of that is a reason to buy a short cash cycle on its own. Bates, Kahle, and Stulz documented that corporate cash-to-assets doubled from 1980 to 2006 as the economy shifted toward intangibles, which mechanically inflates the population of low-cycle firms.

When the machine runs in reverse, it is fast. Suppliers withdraw credit when they sense distress. Customers stop prepaying when the product is in doubt. The float that funded growth becomes a claim that has to be repaid out of a business that is already weakening. Distress research puts working-capital deterioration only one to two years ahead of insolvency once it starts. High fixed costs plus a reversing float compound rather than add. That double pattern is the modern casebook: Bed Bath & Beyond, Tuesday Morning, WeWork, Peloton.

Gateway shifted to financing risky customers in 2000–01; net operating assets ballooned by about $1.1 billion, free cash flow swung to −$855 million, and $876 million was written off the next year. Lucent's receivables rose from 20% to 27% of sales as it lent to its own customers; operating cash flow went from $2.1 billion to −$276 million. Home Depot under Nardelli stretched payables from 22 to 34 days for a $3 billion one-time cash-flow boost that fully reversed within a year. Northern Rock's "extreme dependence on a fickle source of funding" produced the first UK bank run in a century.

A lot of what looks like float is theater. Reverse factoring roughly doubles apparent payment terms — Wuttke and co-authors measured +55.6 days on average — and by early 2023 about 80 S&P 500 companies had disclosed $64 billion of supplier-finance obligations. Carillion reported positive operating cash flow that was negative without the supplier-finance inflow, then collapsed. Abengoa defaulted when a facility was cancelled. Peregrine faked receivable sales to suppress days sales outstanding. Royal Ahold booked $700 million of fake vendor rebates. Tesco overstated supplier income by $400 million. Penman's manipulation tables show the quieter version: draining deferred revenue to flatter reported sales. Management language about "aggressively managing working capital" is itself a flag in Schilit's forensic canon.

Not all of the accounting signature is even usable. Funds held for marketplace sellers can be restricted. Government progress payments can become immediately repayable if a program is cancelled. Regulated utilities and hospitals with mandated collections earn capped returns, so their negative working capital is not a quality signal.

Watch the source, then watch it move. A cash-conversion cycle that is negative and stable for about five years is the signature — days sales outstanding plus days inventory, minus days payable. Compute payable days on purchases, not cost of goods, when the position is serious. Classify the family before judging durability: customer float fails through churn, supplier leverage through term renegotiation, operating-model float through competitive copy, franchise float through franchisee default. For subscription businesses, deferred revenue growing at or with revenue is healthy; several consecutive quarters of deferred revenue lagging revenue is the earliest reversal signal. Prakash and Sinha found a 1% change in the deferred-revenue component maps to a large change in year-ahead margins, so judge the move against the firm's own trailing year, not a universal cutoff.

When invested capital is near or below zero, abandon ROIC. It is undefined at zero and inverted below it. McKinsey's TradeCo example had ROIC jumping from 316% to 632% while value declined. Switch to economic profit: operating profit after tax minus a capital charge. If the add-back from negative capital dwarfs actual operating profit, the business is being flattered by its balance sheet.

One legitimate exception: a company in the first two or three years of a deliberate license-to-subscription transition will show exactly the decay signals as a feature. Tzuo called it the fish. Adobe, Microsoft, Autodesk, PTC, and Bentley all troughed for roughly that long before cash-flow growth returned. Transitions that drag well past that range without recovering should be re-read as decline.

The usable posture is to hold both books at once. From a going-concern compounding view, customer float is zero-cost funding. From a credit and liquidation view, it is a stack of senior claims. Value the compounding on the first book. Size the downside on the second. The float is a lock-in only while the moat that created it is still there.

Sources

Penman, Financial Statement Analysis and Security Valuation, 5th ed. (operating-liability leverage; fragility when suppliers withdraw credit); McKinsey (Koller, Goedhart, Wessels), Valuation, 7th ed. (invested capital as operating assets minus operating liabilities; TradeCo); Buffett, The Essays of Warren Buffett (Cunningham ed.); Schilit & Perler, Financial Shenanigans, 4th ed.; Mihaljevic, The Manual of Ideas; Greenwald, Competition Demystified and Value Investing; Cunningham, Eide & Hargreaves, Quality Investing; Tzuo, Subscribed; Dorsey, The Five Rules for Successful Stock Investing; Damodaran, Investment Valuation, 3rd ed.; Chancellor, Capital Returns; Klarman, Margin of Safety; Graham & Dodd, Security Analysis; Wang, "The Cash Conversion Cycle Spread," Journal of Financial Economics, 2019; Nissim & Penman, 2003; Prakash & Sinha, Contemporary Accounting Research, 2013; Wuttke et al., Journal of Operations Management, 2025; Bates, Kahle & Stulz; FASB ASU 2022-04. Named cases from those sources and public filings: Amazon, Berkshire/GEICO, KONE, Costco, Dell, Adobe, Gateway, Lucent, Home Depot, Northern Rock, Bed Bath & Beyond, WeWork, Peloton, Carillion, Ahold, Tesco.

Not AI-generated signals.