The Underwriting Challenge

Every credit officer has seen it: a borrower’s Confidential Information Memorandum arrives with a management-adjusted EBITDA that sits materially above what the audited financials show. The gap is explained by a neat schedule of “one-time” items — legal settlements, integration costs, executive severance, a COVID-era revenue dip, and projected cost savings not yet realized. The sponsor’s deal team presents the adjusted figure with confidence. The leverage ratio looks acceptable. The debt service coverage ratio clears the hurdle. The deal looks approvable.

The problem is that normalized EBITDA is the single most consequential input in middle market credit underwriting, and it is also the input most susceptible to inflation. It is the denominator in every leverage ratio, the numerator’s context in every coverage test, and the basis against which covenant headroom is measured. A $1 million overstatement in EBITDA on a deal priced at 5.0x leverage can conceal $5 million of true economic debt capacity — before the first dollar is ever drawn.

The stakes are measurable at the portfolio level. According to S&P Global Ratings, average leverage across its U.S. middle market credit estimate population reached 7.0x in 2024, with 20% of companies operating below 1.0x EBITDA interest coverage — a cohort where even modest EBITDA overstatement can accelerate distress. Getting the earnings figure right is not a clerical exercise. It is the foundation of every credit decision.

The Analytical Framework: From Reported to Lender-Underwritten EBITDA

Experienced underwriters recognize that EBITDA normalization moves through at least four distinct stages, each with a different evidentiary standard:

1. Reported EBITDA — the accounting starting point drawn directly from audited or reviewed financial statements.
2. Management-Adjusted EBITDA — the seller or sponsor’s proposed bridge, disclosed in an add-back schedule attached to the CIM.
3. Normalized EBITDA — the more defensible estimate of sustainable, run-rate earnings after independent scrutiny of each proposed adjustment.
4. Lender-Underwritten (Buyer-Accepted) EBITDA — the conservative figure the credit desk actually uses to size debt, set covenants, and stress test coverage.

The gap between Stage 2 and Stage 4 is where credit risk lives. Practitioners consistently note that the most common source of disagreement in underwriting is not the definition of EBITDA itself — it is whether a proposed adjustment deserves full credit, partial credit, or none at all.

The CARVE Test

A practical screening tool used by institutional credit officers is the CARVE test, which requires each proposed add-back to satisfy five criteria before acceptance:

• C — Cash: Was the expense actually paid in cash, or is it a non-cash accounting entry?
• A — Auditable: Can the adjustment be verified through supporting invoices, board minutes, legal filings, or third-party confirmations?
• R — Recurring: Has this expense appeared in prior periods? If so, it may be structural, not one-time.
• V — Verifiable: Can an independent accountant or QoE firm reproduce the figure independently?
• E — Excluded-from-operations: Is the cost genuinely outside the normal course of business, or is it required to generate revenue, serve customers, or maintain the asset base?

Credit analysts applying this framework note that any add-back passing only two or three of five CARVE checks should be haircut or excluded entirely — regardless of how the borrower categorizes it. The cleanest credit memo presents both the borrower-adjusted EBITDA and the lender-adjusted EBITDA, with a transparent bridge between them.

Categories of Commonly Accepted Add-Backs

The add-backs most consistently accepted by institutional lenders fall into four categories: (1) excess owner compensation above documented market-rate replacement cost, (2) genuinely non-recurring costs supported by invoices — such as a single acquisition-related legal fee or a plant closure charge that will not repeat, (3) non-cash charges such as stock-based compensation, impairments, and amortization of intangibles, and (4) related-party transactions repriced to arm’s-length market terms. Even within these categories, the amount of the add-back matters as much as its classification — a lender should never accept the borrower’s number without independently testing what must remain in the business to sustain operations.

The Analysis: A $10M EBITDA Manufacturing Company

Consider a hypothetical sponsor-backed acquisition of a specialty manufacturer with the following management-presented financials:

At $25M of senior debt, the sponsor presents a leverage ratio of 2.5x. Attractive. But let’s apply the CARVE test:

• Owner Compensation ($900K): Supported by a market compensation study showing $800K replacement cost. The lender gives credit for $700K after haircut. Accepted: $700K.
• Legal Settlement ($600K): Single occurrence, fully documented. Passes all five CARVE criteria. Accepted: $600K.
• Restructuring ($500K): A company with “one-time” restructuring charges in multiple consecutive years does not have one-time charges — it has a recurring cost structure. Review of prior three years shows restructuring costs in two of three periods. Rejected: $0.
• Projected Synergies ($600K): Future benefits not yet implemented. Synergies belong to the acquirer, not the target — they do not belong in standalone normalized EBITDA. Rejected: $0.

At $25M of senior debt, the true leverage is now 2.87x — not 2.5x. More importantly, watch what happens to the DSCR. Assuming annual debt service of $3.0M:

• Management DSCR: $10,000 / $3,000 = 3.33x
• Lender-Underwritten DSCR: $8,700 / $3,000 = 2.90x

Both may clear a 1.25x coverage hurdle, but the margin of safety is materially different. Credit analysts note that a $200K swing in accepted add-backs on a $1M debt service base can shift DSCR from 1.15x to 1.35x — enough to change a decline into an approval. The evidence standard must match the economic impact.

Stress Testing the Normalized Base

S&P Global’s mild stress scenario — a 10% EBITDA decline and a 50-basis-point rate increase — pushes over 25% of middle market firms below 1.0x coverage. Applied to our example: a 10% haircut on lender-underwritten EBITDA ($8,700K × 0.90 = $7,830K) with a $200K increase in debt service produces a stressed DSCR of 2.47x — still manageable. But if the underwriter had used management’s $10M figure and the company underperforms by 20%, stressed DSCR collapses to 2.67x on management’s numbers but only 2.32x on underwritten numbers. The difference is the lender’s margin of error at origination.

Industry Standards: What Institutional Lenders and the Data Show

The Covenant Linkage Problem

EBITDA normalization does not end at underwriting — it flows directly into covenant compliance calculations for the life of the loan. Normalized EBITDA is the denominator in leverage ratios used for covenant compliance, and overly aggressive add-backs can make a highly levered company appear to have more headroom than it actually has, obscuring credit deterioration until it is too late.

IOSCO’s leveraged finance review found evidence that headline debt-to-EBITDA ratios may be systematically understated due to add-backs for synergies and operational improvements that are uncertain in both magnitude and timing. As competition between broadly syndicated and direct lenders intensifies, Moody’s has noted that covenant flexibility and other borrower-friendly characteristics are increasingly migrating into private credit structures.

The Early-Warning Value of Maintenance Covenants

Moody’s Default Research Database and S&P Global Ratings data both point to the same directional finding: lenders armed with a maintenance covenant consistently recover more than lenders who discover the problem only at payment default — averaging 10 percentage points across multiple default cycles. Covenant-heavy loans show senior secured recovery rates of 75% or higher, while covenant-lite loans show recoveries in the 65–67% range.

Credit Performance Data Box

The Most Common Red Flags

• Recurring “Non-Recurring” Items: Restructuring, litigation, or integration charges appearing in multiple trailing periods are structural costs, not adjustments.
• Unimplemented Synergies: Cost saves and revenue improvements not yet reflected in actual results belong to the acquirer’s upside case, not the lender’s underwriting base.
• Revenue-Side Adjustments: Add-backs that effectively gross up revenue for a lost contract or underperforming product line deserve far higher scrutiny than cost-side items.
• Add-Back Schedules Without Documentation: Any adjustment lacking a corresponding invoice, board resolution, or third-party verification should be excluded by default.
• Borrower-Only EBITDA Bridge: Covenant definitions may permit adjustments that a credit officer should still haircut for underwriting conservatism. A single EBITDA figure without a lender-side bridge is a red flag.

EBITDA Normalization Underwriting Checklist

The following diligence framework should be applied to every credit that presents a management-adjusted EBITDA materially above reported EBITDA.

Step 1 — Build the Bridge Independently

• ✓ Obtain the borrower’s full add-back schedule with line-item detail
• ✓ Reconstruct reported EBITDA from audited statements independently; do not accept the borrower’s starting point without verification
• ✓ Map each add-back to a specific year and dollar amount across at least three trailing years

Step 2 — Apply the CARVE Test to Each Line Item

• ✓ Score each add-back across all five CARVE criteria (Cash, Auditable, Recurring, Verifiable, Excluded-from-operations)
• ✓ Require supporting documentation: invoices, legal settlements, severance agreements, compensation benchmarks
• ✓ Apply a haircut to any item scoring below 4 of 5; reject any item scoring below 3 of 5

Step 3 — Separate Historical, Run-Rate, and Prospective Adjustments

• ✓ Capitalize only supportable historical adjustments; give partial credit to implemented run-rate changes; reserve prospective improvements for the borrower’s upside case only
• ✓ Flag any synergy add-backs and confirm they are excluded from the lender-underwritten EBITDA base

Step 4 — Commission or Review a Quality of Earnings Report

• ✓ For transactions above your firm’s materiality threshold, require a third-party QoE from an independent accounting firm
• ✓ Adjustments disclosed by the seller should be independently verified by the buyer’s accountant in a quality of earnings review
• ✓ Confirm that QoE-rejected add-backs are removed from your underwriting base — not merely footnoted

Step 5 — Translate to Covenant Structure

• ✓ Set the financial maintenance covenant leverage ceiling at lender-underwritten EBITDA, not management-adjusted EBITDA
• ✓ Build in at least 15–20% headroom between underwritten leverage and the covenant ceiling at origination
• ✓ Define EBITDA in the credit agreement with specific, enumerated add-back permissions — avoid open-ended “other non-recurring” language
• ✓ Require quarterly compliance certificates with supporting EBITDA reconciliation schedules

Key Documentation Requirements

• ✓ Three years of audited or reviewed financial statements
• ✓ Trailing twelve months (LTM) management accounts with month-by-month detail
• ✓ Itemized add-back schedule with per-item supporting documentation
• ✓ Independent QoE report (for transactions above materiality threshold)
• ✓ Management compensation benchmarking study if owner comp add-back exceeds $250K
• ✓ Board resolutions or legal counsel letters for settlement or restructuring add-backs