The Underwriting Problem: Whose Number Are You Lending Against?

Every credit memo starts with the same question: what does this business actually earn? The answer is rarely the number on the cover of the CIM. Borrowers and their advisors routinely present an adjusted EBITDA figure that has been scrubbed of expenses they characterize as non-recurring, non-operating, or discretionary. Some of those adjustments are legitimate. Many are not.

The gap between borrower-presented EBITDA and lender-adjusted EBITDA is not a rounding error — it is a structural risk embedded in the deal from day one. Every turn of leverage, every debt service coverage ratio, every covenant threshold is calibrated against that number. If the base is inflated, the loan is mispriced, the covenants are set too loosely, and the first credit stress will arrive faster and hit harder than the underwriting model predicted.

The discipline of EBITDA normalization is one of the most consequential analytical skills a credit professional exercises. It requires financial forensics, sector judgment, and the institutional confidence to push back on numbers presented by well-compensated sell-side advisors. This article walks through the methodology, the mechanics, and the red flags — with a worked example and a full underwriting checklist.

The Analytical Framework: Lender-Adjusted EBITDA

The starting point for any normalization exercise is a clear conceptual distinction. Reported EBITDA is what the income statement shows. Adjusted EBITDA is what the borrower claims after removing items they consider extraordinary. Normalized EBITDA is what a disciplined lender independently concludes represents sustainable, forward-looking cash earnings — and these three numbers are almost never the same.

A practical formula, drawn from institutional practice, reads as follows: Normalized EBITDA = Reported EBITDA + Accepted Add-Backs − Non-Recurring Income ± Owner-Compensation Normalization ± Accounting & Classification Adjustments + Run-Rate Effects Already Implemented and Supportable. The sign and category of each item matter as much as the dollar amount.

Critically, the cleanest credit memo shows both borrower-adjusted EBITDA and lender-adjusted EBITDA, with a clear, line-by-line bridge between them. This discipline forces the underwriter to own every decision and ensures that downstream ratios — leverage, DSCR, interest coverage — are computed off a single, consistently applied earnings base. The four broadly accepted add-back categories institutional lenders work from are: (1) excess owner compensation above documented market-replacement cost; (2) verified one-time expenses with no probability of recurrence; (3) non-cash charges such as stock-based compensation and D&A; and (4) related-party transactions normalized to arm’s-length market rates.

Example: Analyzing a $10M EBITDA Manufacturing Business

Consider a mid-market industrial manufacturer presenting a trailing-twelve-month adjusted EBITDA of $10.0 million in its CIM. The sponsor’s add-back schedule is aggressive. The lender’s independent review produces a materially different picture:

The Analysis: How Inflated EBITDA Destroys Ratio Integrity

The downstream consequences of unchallenged add-backs are not merely academic. When a lender accepts the borrower’s $10.3M EBITDA on a $52.5M debt package, the apparent leverage of 5.1x sits comfortably within most middle-market credit policy guidelines. The debt service coverage ratio at a 7.5% blended cost and five-year amortization appears adequate. Covenants are set with appropriate headroom.

But when normalized to $8.2M, leverage jumps to 6.4x — a figure that crosses the FDIC’s higher-risk commercial loan threshold of 4.0x total debt-to-EBITDA and the 3.0x senior-debt benchmark used in bank regulatory assessments. DSCR compresses from a comfortable 1.6x to a marginal 1.27x, leaving almost no buffer against an operating shortfall before cash flow is insufficient to service debt. At that level, a 10–15% revenue decline — well within the range of a mild economic slowdown — puts the borrower in technical distress.

Stress testing must be applied to the lender-adjusted number, never the borrower-presented one. A basic downside scenario — revenues down 10%, margins contracting 150bps — should be modeled before final credit approval. If DSCR falls below 1.10x in a moderate downside, the capital structure deserves a hard look at sizing, pricing, or both.

The CARVE Test: A Practical Add-Back Decision Filter

Experienced underwriters apply a structured decision filter to each proposed add-back. The core question for every item: Is this cost required to generate revenue, serve customers, meet compliance requirements, or maintain the asset base? If the answer to any of those conditions is yes, the cost belongs in operating EBITDA. An add-back that cannot clearly pass this test should be haircut or excluded — regardless of how the borrower categorizes it. Pro-forma revenue adjustments deserve particular skepticism: a contract not yet started, a pipeline opportunity, or a management forecast is not the same as a documented expense that will not recur.

Industry Standards: What the Data Says About EBITDA Risk and Credit Outcomes

The underwriting stakes are clarified by credit performance data. The leverage environment of 2023–2025 saw typical middle-market deal leverage normalize to 4.5x–5.5x as rate hikes and weaker exit conditions forced lenders to reassert discipline — a meaningful reset from the 5x–6x structures common in prior cycles. That structural tightening was overdue: years of aggressive add-backs and borrower-friendly covenant definitions had systematically overstated earnings and understated leverage across the market.

On the default side, Moody’s analysis indicates that the private credit default rate in 2025 likely ranged between 1.6% and 4.7%, depending on whether distressed exchanges are included in the count. Notably, distressed restructurings — workouts, indenture modifications, debt-for-equity swaps — accounted for roughly 65% of all corporate defaults in 2025. This is a critical underwriting insight: the deal does not have to go formally bankrupt to inflict losses on lenders. Maturity extensions and term modifications under financial duress are credit events in all but name.

Recovery rate data reinforces the urgency of getting the earnings base right at origination. The 2024 first-lien term loan ultimate recovery rate fell to 49.2%, versus a long-run average of 76.4% per Moody’s Annual Default Study — the most severe first-lien recovery deterioration since the 2008–2009 financial crisis cohort. Second-lien recovery in 2024 came in at just 14.5%, against a long-run average of 46.5%. For direct lenders operating in the middle market, Blue Owl has cited recovery rates in direct lending at roughly 75% and above, versus approximately 40–60% for broadly syndicated loans — a function of tighter covenant structures and closer lender-borrower relationships.

Common Red Flags Experienced Lenders Watch For

• Add-back stack exceeding 20–25% of reported EBITDA: When total adjustments materially exceed one-quarter of stated earnings, the normalized figure is highly sensitive to judgment calls and requires extraordinary documentation scrutiny.
• Recurring “one-time” charges: Legal fees, restructuring costs, or consulting expenses that appear in multiple trailing periods are, by definition, part of the run-rate cost structure and should not be added back.
• Pro-forma revenue without signed, collectible contracts: Management projections and pipeline opportunities do not belong in TTM EBITDA. Revenue-side add-backs require near-term, documented evidence of collectibility.
• Owner compensation below market rate: A seller deliberately underpaying themselves inflates EBITDA. The lender must normalize downward for a market-rate replacement — an adjustment that cuts earnings, not adds to them.
• Add-backs that contradict the working capital analysis: If a cost is removed from EBITDA as non-recurring but the working capital bridge includes it as an ongoing liability, the presentations are internally inconsistent and signal broader diligence failure.

Underwriting Checklist: EBITDA Normalization Due Diligence

The following step-by-step checklist reflects institutional lender practice for credit submissions involving adjusted EBITDA. It should be completed independently — not handed to the borrower’s accountants.

1. Anchor to audited or reviewed financials. Start with the most recent three-year audited P&L and independently tie TTM figures to bank statements, not management accounts alone.
2. Build your own EBITDA bridge. Construct the normalized EBITDA schedule line by line, independent of the borrower’s add-back schedule. Compare the two and document every variance.
3. Classify every adjustment. Label each item: verified one-time, excess owner comp, related-party normalization, non-cash, or rejected. The rationale must be written into the credit memo.
4. Apply the CARVE test to each add-back. Is this cost required to generate revenue, serve customers, meet compliance, maintain assets, or retain key staff? If yes, it belongs in operating EBITDA.
5. Independently benchmark owner compensation. Use current industry compensation surveys to document market-rate replacement cost. Normalize in both directions — excess comp is added back; below-market comp is subtracted.
6. Verify related-party transactions at arm’s length. Obtain third-party market comparables for any lease, service agreement, or intercompany transaction. Document the source (e.g., CoStar, broker opinion, trade data).
7. Interrogate non-recurring revenue. Identify any top-line item that is unlikely to recur — large government contracts, windfall customer orders, litigation settlements received. Remove from the TTM base and document the source.
8. Reconcile EBITDA to covenants. Confirm that DSCR, leverage, and all financial covenant definitions in the draft credit agreement use the same lender-adjusted EBITDA base — not the borrower’s adjusted figure.
9. Stress test the normalized number. Run a base, downside, and severe downside scenario. If DSCR falls below 1.10x under moderate stress, revisit deal sizing, structure, or pricing before approval.
10. Document the complete bridge in the credit memo. The credit file must include both the borrower-presented EBITDA and the lender-adjusted EBITDA, with a clear reconciliation that any examiner or credit committee member can follow without additional explanation.

Documentation Requirements

• Three years of audited (or reviewed) financial statements
• TTM income statement tied to bank-level cash receipts and disbursements
• Owner compensation W-2s or K-1s for all periods under review
• Third-party lease comps or broker opinion for related-party real estate
• Invoices or court documentation for claimed one-time legal or restructuring costs
• Signed customer contracts supporting any run-rate revenue assumptions
• Management rep letter affirming completeness of the add-back disclosure