The Underwriting Challenge

You are reviewing a Confidential Information Memorandum for a $10M EBITDA specialty manufacturing company seeking a $45M unitranche facility. The sponsor’s model shows leverage of 4.5x on adjusted EBITDA. Flip to the trailing twelve-month financials, however, and reported EBITDA is $7.4M — implying true leverage of 6.1x. The gap between those two numbers is a $2.6M stack of add-backs: restructuring charges, above-market owner compensation, one-time legal fees, and a COVID-era government credit. Each line item carries a story. The underwriter’s job is to determine how much of that story is defensible.

Adjusted EBITDA is the single most consequential number in a leveraged credit transaction. It sets leverage multiples, governs covenant compliance, drives valuation, and determines debt capacity. It also happens to be the number most susceptible to manipulation. IOSCO’s 2024 review of leveraged lending markets found that add-backs can account for a material share of borrower EBITDA projections in M&A and LBO transactions, and that there is clear evidence headline Debt/EBITDA may systematically understate true leverage when adjustments are applied aggressively. For a credit underwriter, building a disciplined normalization methodology is not optional — it is the foundation of the entire analysis.

The Analytical Framework: Three Layers of the EBITDA Bridge

Quality of Earnings (QoE) practice structures the EBITDA bridge into three distinct layers, and the rigor applied to each layer should scale with its uncertainty. Conflating them is one of the most common underwriting errors in practice.

Layer 1: Definitional Adjustments

These are the GAAP-to-EBITDA conversions: adding back interest expense, income taxes, depreciation, and amortization. They are non-controversial because they are mathematically defined. Every underwriter starts here, and no serious lender challenges them. The only work required at this layer is confirming the inputs tie to audited or reviewed financial statements.

Layer 2: Quality-of-Earnings Adjustments (One-Time Items)

These are the items claimed to be non-recurring: settled litigation, M&A advisory fees, redundant facility closure costs, non-cash impairments, and severance from a specific restructuring event. Each adjustment must survive a single, binary test: Will this item genuinely not recur? This is where the bulk of diligence effort concentrates and where the highest proportion of disputes arise. Lenders must demand source documentation — invoices, settlement agreements, board minutes — for every line item.

Layer 3: Normalization Adjustments (Structural Corrections)

These cover recurring items that are mispriced on the books: above-market owner compensation, below-market related-party rent, family members on payroll in non-substantive roles, or discretionary personal expenses run through the business. Unlike Layer 2, these items are recurring in the current ownership structure — the argument is only that a new ownership or management structure would eliminate them. Each normalization must therefore be benchmarked to a verifiable market-rate equivalent, not simply asserted.

“A defensible add-back needs source documentation, a narrative explaining why the item is genuinely non-recurring or owner-specific, a benchmark supporting any market-rate claim, and categorization consistent with how comparable transactions have treated the same type of item.”

— Private Credit Review, Underwriting EBITDA Adjustments in Sponsor-Backed Deals, 2026

The Lender’s Independent Credit EBITDA

Best-practice credit memos present two EBITDA figures side by side: the borrower-presented adjusted EBITDA, and the lender-independently-derived credit EBITDA, with an explicit bridge between them showing which adjustments were accepted in full, which were haircut, and which were rejected. This structural transparency prevents the underwriter from inadvertently inheriting the borrower’s optimism.

Step-by-Step: Normalizing a $10M EBITDA Manufacturing Borrower

Consider a representative lower-middle-market manufacturing company with $28M in revenue. The borrower presents adjusted EBITDA of $10.0M to support a $45M unitranche (4.5x leverage). Below is the underwriter’s annotated review of each adjustment category:

The disciplined underwrite reduces adjusted EBITDA by $2.02M — a 20% haircut — and reveals true leverage of 5.6x, materially above the sponsor’s headline. At 5.6x on a lower-middle-market manufacturing credit, the debt service coverage ratio tightens materially. Applying a 7.5% all-in rate on $45M of debt yields annual interest of $3.375M and a DSCR of approximately 1.24x — squarely in the warning zone, where middle market maintenance covenants typically begin at 1.20x to 1.35x.

Stress Testing the Normalized Figure

A single-scenario underwrite is insufficient. Prudent lenders model at minimum three scenarios against lender credit EBITDA, not borrower EBITDA:

• Base Case: $7.98M EBITDA, stable margins — DSCR 1.24x, leverage 5.6x.
• Adverse Case (-10% EBITDA): $7.18M EBITDA — DSCR 1.12x, leverage 6.3x. Covenant breach territory.
• Severe Case (-20% EBITDA): $6.38M EBITDA — DSCR 0.99x, leverage 7.1x. Distress scenario requiring waiver or amendment within 2 quarters.

This stress architecture directly mirrors market data: S&P Global analysis of middle market credits found that a mild stress scenario — a 10% EBITDA decline and a 50-basis-point rate increase — pushes more than 25% of firms below 1.0x interest coverage, signaling distress.

What Institutional Lenders Look For: Market Standards and Credit Performance Data

Leverage and Covenant Norms (2025–2026)

Market data provides clear reference points for disciplined underwriting. According to PGIM, middle market first-lien debt averaged 4.5x leverage (Debt/EBITDA) as of late 2024, compared to 5.8x in the broadly syndicated loan market — a structurally more conservative posture that partially reflects the absence of market price discovery in direct lending. For lower-middle-market transactions (sub-$10M EBITDA), total leverage generally clears between 2.5x and 3.25x, while upper mid-market buyouts range from 5.0x to 5.5x.

Leverage covenant calibration follows a consistent institutional methodology: the covenant is typically set with a 25–35% cushion above the borrower’s own base case model, as documented by Sidley Austin’s March 2026 review of private credit covenant practices. Cushions above this range risk creating “cov-loose” structures that strip lenders of early warning rights. The key structural risk is the gap between contractual covenant EBITDA (which includes permitted add-backs) and reported GAAP EBITDA. The Financial Stability Board has noted that, with add-backs stripped out, true leverage in private credit borrowers could be closer to 7x, against a headline-reported 5–6x.

IOSCO’s leveraged lending review confirmed that add-backs and synergy adjustments in current deals are subject to EBITDA adjustments of 15–30%, and that historical best practice called for caps between 5% and 25% of prior-twelve-month EBITDA — with Fitch identifying a trend toward eliminating caps from loan documentation entirely. Sophisticated lenders are now pushing back with explicit contractual add-back caps of 20–25% of reported EBITDA and time limits of 18–24 months for synergy realizations.

Common Red Flags in EBITDA Normalization

• Uncapped pro forma synergies: Add-backs for future cost savings with no contractual realization timeline or escrow mechanism. Reject or cap at 18 months with a burn-down schedule.
• Revenue add-backs presented as EBITDA add-backs: Borrowers occasionally present lost revenue from a divested segment as a positive adjustment. Challenge the revenue base before normalizing margins.
• Recurring “one-time” charges: Three consecutive years of “non-recurring” restructuring charges are, by definition, recurring. Apply a three-year average to stress-test the pattern.
• Marketing and headcount cuts pre-sale: Sellers who slash marketing spend to inflate short-term EBITDA before a transaction are inflating a number that will revert. Underwriters must normalize downward to reflect what the business actually requires to sustain revenue.
• Undisclosed non-recurring revenue: One-time government contracts, insurance proceeds, or litigation settlements that inflate TTM revenue without being flagged. Always analyze revenue detail at the customer or contract level.
• Borrowed EBITDA gap: The difference between covenant EBITDA and reported EBITDA can be 20–40% in aggressive structures. Monitor both figures independently post-close.

The EBITDA Normalization Underwriting Checklist

Use the following checklist at each stage of the credit process: initial screen, credit committee memo, and post-close monitoring.

Key Takeaways

• Adjusted EBITDA is a starting point, not a conclusion. The borrower’s number reflects negotiation strategy; the underwriter’s job is to derive an independent credit EBITDA grounded in documentation and conservative benchmarks.
• The taxonomy of adjustments determines the burden of proof. Definitional adjustments require no argument; quality-of-earnings items require non-recurrence evidence; normalization adjustments require market-rate benchmarks. Apply proportionate scrutiny at each layer.
• The leverage gap is the unpriced risk. The Financial Stability Board estimates that true leverage in private credit, with add-backs stripped out, may be 7x against a headline 5–6x. Every 1.0x of understated leverage reduces covenant cushion and accelerates the path to distress under adverse conditions.
• Recovery rates no longer provide the same margin of safety. Moody’s estimated 2024 first-lien ultimate recoveries at approximately 68%, versus a long-run historical average in the high 70s — driven by increased first-lien leverage, fewer subordinated cushions, and the prevalence of distressed exchanges that deliver weaker outcomes than traditional bankruptcy processes.
• Covenant calibration is only as good as the EBITDA definition underneath it. Set leverage covenants against lender credit EBITDA with a 25–35% cushion, impose contractual add-back caps of 20–25%, and require a quarterly EBITDA reconciliation certificate. The covenant is meaningless if the EBITDA definition permits unlimited adjustment.