How to Detect Debt That the Snapshot Misses
For the nonfinancial corporations in the Federal Reserve study, the hard part is not naming a missing liability; it is deciding what evidence can reveal it without turning an unusual number into an accusation. This chapter explains how the study converts operating-lease commitments and interest expense into comparable signals of hidden leverage. The reader's task is to understand what each measure estimates, why its assumptions matter, and how to keep a provisional diagnosis separate from proof of intent.
Turning lease commitments into debt-like exposure
A schedule of future lease payments is not yet a leverage ratio. It describes cash the corporation expects to pay at different points in the future, while reported debt is usually presented as an amount outstanding at the reporting date. To compare the two, the study translates future operating-lease payments into a present value: an estimate of what that future payment stream is worth in today's terms under a chosen discount rate.
The logic is familiar from any comparison of cash received now with cash paid later. A dollar due next year is not financially identical to a dollar due ten years from now because the later payment can be discounted to reflect the time value of money. A lease schedule that simply adds every future payment would overstate the amount comparable to a current debt balance. Discounting creates a common unit, while still leaving an important judgment: the selected rate changes the size of the estimated liability.
The paper builds this estimate from the payment information that firms reported in Compustat. Before the accounting change, firms disclosed payments for each of the next five years, a total for the following five years, and payments beyond five years. When the annual breakdown was unavailable, the study assumed the aggregate amount was amortized across the first five years. It distributed payments beyond five years across years six through ten. Those assumptions make the data usable across firms, but they are not observations of the exact payment calendar for every company.
For the discount rate, the study uses the within-year average yield on Moody's Baa corporate bonds. The author describes this as an intermediate choice. A lessee's selected implicit rate can be relatively high, which produces a lower present value and therefore a smaller reported-equivalent liability. A lessor looking at the asset may use a different valuation perspective. The Baa yield is intended to reflect a corporate financing rate while recognizing that a lease is relatively safe or collateralized compared with unsecured borrowing. The result is a research estimate, not a claim that every firm would calculate the same liability under every accounting or valuation convention.
Once the estimate is constructed, the study adds the present value of off-balance-sheet operating leases to reported debt to produce a broader measure of total debt. It can then distinguish the leverage visible in the old balance sheet from hidden leverage represented by the lease estimate, and from non-lease leverage after operating-lease liabilities are recognized. That separation is crucial. If the objective is to measure whether the corporation borrowed more after disclosure, simply adding a lease liability to the denominator would confuse a presentation change with a new debt decision. The non-lease measure attempts to isolate borrowing other than the lease itself.
The study also needs a rule for deciding when lease exposure is substantial rather than merely present. It flags an operating-lease position when adding the estimated lease liability would raise leverage by more than the firm's actual leverage, by more than one standard deviation relative to firms in the same period, or by more than two standard deviations relative to the firm's own history. The three comparisons answer different questions. The first asks whether the lease is large compared with what the firm already reports. The second asks whether it is unusual among peers. The third asks whether it is unusual for that corporation.
These thresholds are useful for identifying firms whose reported profile would change materially under recognition. They are not a legal definition of hidden debt and do not identify motive. A firm can have a large lease because its business naturally depends on rented property or equipment. A firm can also have a small lease that is economically meaningful for its cash flow but does not cross the study's threshold. The flag is therefore a way to define exposure for a comparison, not a verdict about the corporation.
The measurement also clarifies why the notes mattered before ASC 842. The lease schedule contained the raw material for a debt-like reconstruction. A reader who discounted the payments could reach a different leverage assessment from one who used only the recognized debt line. Recognition later reduced that gap for the common balance sheet, but the research still has to estimate the pre-recognition exposure in a consistent way to compare firms across time.
From an interest mismatch to a provisional diagnosis
Intra-period borrowing requires a different construction because the principal may have been repaid before the reporting date. The paper begins with interest expense, which accumulates over the period, and compares it with debt reported at the prior quarter-end. In simplified form, the quarterly implied interest rate is interest expense divided by prior-quarter end debt, annualized by multiplying by four and expressed as a percentage. The annual version uses interest expense divided by the average of beginning- and end-of-year reported debt.
The formula is not intended to recover the exact rate paid on a particular loan. It combines at least two things: ordinary interest on debt that was visible and interest paid on borrowing that may no longer be visible. The paper therefore does not compound the quarterly annualization. Compounding would imply that every part of the measured expense came from a continuously outstanding balance, which is precisely what the diagnostic cannot assume.
A simple example shows the direction of the signal. Suppose a corporation reports a modest debt balance at the end of a quarter but had borrowed and repaid additional cash during that quarter. The interest expense includes the cost of the temporary borrowing, while the ending balance does not include its principal. Relative to the reported denominator, the implied rate rises. That rise is consistent with intra-period borrowing, but it could also arise if the corporation paid an unusually high rate on its visible debt or incurred another form of financing cost.
The paper addresses that differential by estimating a counterfactual interest expense. It uses a proxy for the risk-free rate, Moody's Baa spread as a price-of-risk measure, the firm's Ohlson default-risk measure, and an interaction between the spread and default risk. The purpose is to ask how much interest would be expected from the debt the firm reports, given the general interest-rate environment and its observable risk. A high raw implied rate caused by a distressed borrower should not automatically be classified in the same way as a high residual after risk and default have been considered.
The suspicious-interest flag is based on several comparisons. The study flags interest paid at more than twice the predicted amount, a log gap unusually large relative to other firms in the period, or a gap unusually large relative to the firm's own history. It also flags cases in which positive, nontrivial interest expense is reported while debt is zero. These rules capture different manifestations of the same symptom: financing cost that is difficult to reconcile with the debt visible at the relevant point in time.
The word “symptom” matters. A doctor would not diagnose a disease from one abnormal reading without considering other causes; the same discipline applies here. An unusually high interest-to-debt relationship is a starting point for investigating the balance-sheet denominator, not evidence that the corporation intentionally hid a loan. The study's counterfactual removes some explanations, but it cannot guarantee that every remaining gap is intra-period borrowing.
Accounts payable is one alternative. A corporation that delays payment to suppliers may incur an implicit financing cost that appears in financial results without looking like a conventional loan. The paper finds that reliance on accounts payable is positively related to suspiciously high implied interest. That is a relevant warning, not a discarded nuisance. The author reports that accounts payable has limited explanatory power, that the results survive restrictions to firms with low accounts payable or controls for it, and that payable financing alone does not fit the timing of the changes around disclosure rules. The conclusion is not that accounts payable never matters; it is that it does not fully explain the observed pattern.
Financial distress is another alternative. A corporation close to default may genuinely pay high interest, so a high implied rate could reflect the price of risk rather than a missing principal balance. That is why the study adjusts for default risk and examines the excess over a risk-adjusted prediction. Even then, the measure remains imperfect. Default-risk models are themselves estimates, and a firm's actual borrowing terms can differ from the variables used to predict them.
The lease measure has its own uncertainty. Future payment schedules may be incomplete or aggregated, and distributing payments across years requires assumptions. The discount rate can materially change present value. A higher rate produces a smaller estimated liability; a lower rate produces a larger one. The paper also notes that lessees may choose rates at or above 10 percent, while the author uses the Baa yield to account for the collateralized nature of leasing and the changing interest-rate environment. That choice supports comparability across the sample, but it should not be mistaken for a universally correct valuation.
The two measurement paths should not be treated as interchangeable simply because both end in a flag. The lease path starts with future contractual payments and estimates a stock of debt-like exposure. The interest path starts with a period flow and asks whether the reported debt balance is too small to explain it. One looks forward through commitments; the other looks backward through financing cost. Their different denominators and time horizons are the reason the study keeps them distinct.
What the flags can and cannot tell us
The study's classifications are designed to make an empirical comparison possible. A researcher needs a repeatable way to identify corporations whose reported leverage would change materially if lease obligations were recognized, and corporations whose interest expense is unusually high relative to visible debt. Without such rules, “hidden leverage” would describe an intuition rather than a testable pattern.
But repeatability is not the same as completeness. A firm below the threshold may still have a meaningful operating lease. A corporation may borrow and repay within a period without producing an interest pattern large enough to cross the right-tail rule. A normal implied rate is not evidence that no temporary borrowing occurred. Conversely, a flag does not identify the precise instrument, the amount borrowed, or the decision-maker who chose the reporting pattern. The measures identify observations that merit an explanation.
The paper's use of peer and historical comparisons helps with this problem. A high interest ratio can be normal for an entire period of high rates, but unusual if it is high relative to firms facing the same environment. A lease can be normal for a capital-intensive industry, but unusual relative to comparable firms or to the corporation's own history. Comparing a firm with itself and with its peers reduces the chance that a general market condition is mistaken for a firm-specific anomaly, though it does not remove every source of selection or measurement error.
Reporting frequency also changes what can be seen. Quarterly data can reveal a debt balance that repeatedly falls at quarter-end, while annual data may smooth over some of those movements. Annual interest expense can still be compared with beginning- and end-of-year debt, but a borrower that draws and repays several times may look different from one that carries a balance for the full year. The study uses both frequencies for different purposes; a reader should not move a quarterly percentage or flag definition into an annual interpretation without checking the construction.
The same caution applies to the historical accounting transition. Before ASC 842, the study estimates lease liabilities from disclosed schedules. After recognition, operating leases enter the balance sheet and affect the reported variables directly. A change in the measured ratio can therefore reflect an accounting change even when the underlying contract and payment schedule did not change. To identify behavior rather than arithmetic, the study separately constructs non-lease leverage and tests outcomes around the implementation event. That distinction will matter when later chapters interpret post-revelation borrowing.
A useful diagnosis therefore has three layers. First comes the observed symptom: a low reported leverage ratio, a large note commitment, an unusual interest expense, or a reporting-date movement. Second comes the proposed mechanism: continuing lease exposure or borrowing that was repaid before the snapshot. Third comes verification through alternative explanations, peer and historical comparisons, and behavior after a change in disclosure. The study contributes evidence at all three layers, but the confidence differs by layer. The signal can be observed; the mechanism is inferred; intent remains a further claim.
This hierarchy is especially important for the phrase “false leverage profile.” The paper argues that the negative relationship between hidden and reported leverage, together with monitoring patterns, is consistent with some firms using off-balance-sheet debt to appear less leveraged. The measurement rules make that hypothesis testable. They do not show that every flagged corporation sought to mislead investors, nor that a lower reported ratio was the only reason to choose a lease or short-term loan.
The strongest reading of the measures is consequently conditional: if the payment reconstruction is reasonable, if the risk-adjusted interest expectation is well specified, and if alternatives such as distress and payables do not account for the gap, then the remaining pattern is evidence consistent with hidden leverage. New evidence could change that interpretation. A more complete borrowing ledger could show that the interest expense came from ordinary visible loans. Better lease terms could change the present value materially. A different disclosure rule could alter quarter-end behavior without any firm-level concealment motive. Or the relationship could disappear when firms are compared within a narrower industry and fiscal-year framework.
That conditional structure is not a weakness to hide. It is what makes the evidence useful. Financial statements rarely reveal every financing decision in one line, and a research measure that fills the gap necessarily combines observations with assumptions. The reader should ask not merely whether a corporation has been flagged, but which part of the diagnosis is directly recorded, which part is reconstructed, and which part remains uncertain.
With that boundary in place, the next synthesis can connect the two measures into one reading model. It can ask what it means when lease exposure, interest expense, and reporting-date debt point in the same direction, while preserving the difference between an incomplete snapshot and a proven explanation.