How the tool estimates the accordion as debt-incurrence optionality.

Four distinct questions

Debt permission, lien permission, ranking and financing economics are not interchangeable. This Lab models selected free-and-clear and ratio debt routes. The priority overlay is an assumption, not proof that the debt may be incurred, secured or ranked as shown.

The economic output is a borrower financing-benefit teaching estimate: the assumed spread saving on modeled borrowing, less the borrower's modeled MFN cost on existing debt. It is not existing lenders' credit loss, covenant compensation or a market price. Quantifying lender harm would require a separate counterfactual credit model, including use of proceeds, default and recovery.

Current selected-route capacity estimate: define the routes before adding

The fixed and grower prongs are greater-of alternatives, not additive baskets. Prior use is deducted once from this basket, and previously incurred debt must already be included in opening leverage. No repayment-credit prong is inferred.

The stacking switch matters. When enabled, same-date free-and-clear borrowing is excluded while testing the ratio borrowing. When disabled, this Lab assumes the free tranche is used first and fully counted. That conservative convention does not assert that every agreement without explicit simultaneous-incurrence wording prohibits ratio-first sequencing; the actual allocation and sequencing rules control.

Opening leverage and the ceiling must use the same gross/net basis and debt perimeter (first-lien, secured or total). The reference face amount used for PV01 is separate from D. The model never nets newly borrowed proceeds as cash, never credits acquisition EBITDA automatically, and does not infer secured permission from debt capacity.

Capacity is not borrowing

Selected-route capacity is an unweighted snapshot. Expected borrowing applies one draw probability to the selected free and ratio amounts. The default 55% is a teaching input, not an empirical forecast or evidence of funding commitments.

When future uncertainty is disabled, current EBITDA and covenant debt are frozen until the draw date. When enabled, the model changes the distribution of EBITDA, not its mean. In both cases, existing covenant debt and the current free basket remain fixed, and one borrowing is modeled. Repeated draws, repayments, reclassification and optimal exercise are not simulated.

Primary authority and private study context

The agreement, not a generic formula, determines contractual availability. The public explanations below are independently authored teaching summaries supported by primary filings:

• Simultaneous-incurrence treatment. A filed commitment document expressly describes ratio-first allocation and exclusion of same-date fixed-prong borrowing when testing the ratio prong. SEC-filed example, Incremental Amount, B-6.
• Different ratio perimeters and no proceeds netting. Another filed agreement uses distinct tests for first-lien, junior-lien and unsecured borrowing, excludes new cash proceeds from netting, and deems incremental revolving commitments fully drawn. SEC-filed incremental facility example.
• MFN instrument eligibility and yield definition. A credit agreement conditions MFN on specific instruments, ranking, incurrence date and carve-outs. It also specifies treatment of OID, lender fees and floors when comparing pricing. These are document-specific filters, not a universal MFN formula. SEC-filed agreement, section 2.20(F).

The LevFin Book supplies the teaching framework. The calculator neither republishes private research nor imports workbook outputs as market prices. Parameters, the scenario distribution and payment schedules below are explicit teaching assumptions.

Inputs and interpretation

On smaller screens, scroll horizontally to see the complete table.

A coherent single-date sensitivity

With uncertainty enabled, EBITDA at draw date t follows the mean-preserving lognormal assumption below. Z is standard normal; Φ is its cumulative distribution. This is a scenario distribution, not an asserted risk-neutral measure.

The formula is the positive-part expectation derived from a truncated lognormal variable. It already contains state probabilities: do not multiply it by Φ(d₂) again. At zero volatility or zero time, it reduces exactly to current ratio capacity. The future-state increment is a separate teaching sensitivity, never added to capacity today. It varies continuously through the zero-headroom boundary.

A negative mean EBITDA trend, cash generation, contractual EBITDA add-backs and debt repayment would require a richer state model. The mean-preserving assumption intentionally does not promise future deleveraging. The lognormal distribution also rules out non-positive EBITDA states; that is a material limitation, not an empirical assertion.

Discount the actual modeled cash-flow legs

Define A(t,n) as the time-zero PV of one unit per year, paid annually after t with a final stub at t+n. Every payment is discounted at exp(−r × payment date). No survival haircut is applied.

The four-year shortcut is shown only as a comparison. It is not added to the PV01 result, and no evidence upgrade follows from either conversion. All euro values, including PV01, use EUR millions; PV01 therefore means EUR millions per basis point.

MFN: trigger window is not uplift duration

The selected borrowing can trigger MFN only if the instrument is eligible, the draw occurs before the sunset, and protected debt is still outstanding. Once triggered, the modeled uplift continues from the draw date until protected maturity, not merely until the sunset.

B(t, maturity) is the remaining protected-loan spread annuity: payments stay on the existing annual coupon dates, and the first affected coupon accrues only from the draw date. This avoids shifting the existing loan payment calendar. Zero borrowing produces zero MFN cash flow. Increasing the sunset past an already eligible draw date does not extend the uplift's modeled life. A negative net benefit is retained: it says the specified MFN cost exceeds the assumed funding advantage, not that an optimal borrower would choose that transaction. No optimally exercised option or lender-loss claim is implied.

Worked default case

Teaching inputs: EBITDA €1,687m; covenant leverage 4.5x; ceiling 6.0x; €175m free amount; no grower or prior use; simultaneous-incurrence exclusion enabled; draw at year 0.5; 55% draw probability; 8% annual EBITDA volatility; 100 bps funding advantage for five years; 4% continuous discounting. The reference face is €6,180m with maturity at year 5 and 100% protection. MFN has a 50 bps cushion, 100 bps gap, 30% conditional trigger probability and a 12-month sunset.

Turning off the simultaneous-incurrence exclusion reduces current ratio capacity by €175m in this case; the free-first total becomes €2,530.5m. Turning off MFN leaves the benefit unchanged and removes its cost. Setting draw probability to zero leaves capacity unchanged but sets financing benefit, MFN and expected borrowing to zero.

How the exhibits reconcile

• Priority split: a mutually exclusive allocation of current selected-route capacity; not extra capacity and not automatic MFN eligibility.
• Capacity bridge: current selected-route capacity estimate + expected future ratio increment − draw-probability haircut = expected borrowing. The ratio-state expectation is not probability-weighted a second time.
• Bps bridge: positive financing benefits less negative borrower MFN cost = signed net benefit. Every row uses the same reference PV01. MFN is simultaneously an uplift to existing lenders, not a credit-loss offset calculated by this model.

Why this remains a teaching estimate

A positive-part payoff does not by itself establish an option price. A calibrated result would require a documented pricing measure, market inputs and calibration, an encoded agreement, default/survival/recovery consistency and independently reviewed exercise rules. None is supplied here.

The closed-form terminal expectation is appropriate for the explicitly simplified one-date scenario. Reclassification, repeat exercise, market windows and path-dependent contractual tests require a different model. No probability displayed here is represented as a real-world forecast.

What is deliberately outside the model

• Other debt baskets, voluntary repayment credits, acquired debt, acquisition EBITDA, automatic reclassification, no-worse tests, cross-covenant blockers and limited-condition testing.
• Actual lender commitments, borrower optimisation, issue-size thresholds and MFN amount, purpose, currency and maturity carve-outs. The eligibility switch must be set after external document review.
• OID conversion, fees, base-rate floors and an all-in-yield solver. The entered gap must already reflect the comparison required by the clause.
• Default, recovery, amortisation, refinancing, prepayment, state-dependent access, correlations and transaction costs. The reference instrument is a bullet with annual spread cash flows and explicit 100% survival.
• Negative EBITDA states, expected EBITDA growth, cash accumulation and debt reduction. The free basket is frozen at its current computed amount even with future uncertainty enabled.
• Legal conclusions from assumed payment/lien buckets. Structural subordination, entity perimeters, collateral and guarantees require separate analysis.

Where this sits in the Book

Debt and liens govern how leverage enters the structure; restricted payments and investments govern how value may leave. Read this Lab beside the Builder Basket Lab and the LevFin Book. Keep capacity, actual use, borrower benefit and lender credit risk separate throughout.

Common questions