EBITDA Normalization in Credit Underwriting: How Lenders Build a Defensible Earnings Base
When a borrower hands you an adjusted EBITDA schedule, the real underwriting work begins. Here is how institutional lenders decompose, verify, and stress-test every line item — before a single covenant is drafted or a dollar is deployed.
The Underwriting Challenge
Imagine you are underwriting a $10 million EBITDA manufacturing company. The borrower’s investment banker presents a neat CIM showing $12.7 million in “adjusted EBITDA” — a 27% lift from reported results. The add-back schedule includes excess owner compensation, one-time legal fees, a closed plant’s severance costs, and $800,000 in “run-rate cost savings” from a reorganization completed six weeks ago. The ask: 5.0x leverage, or $63.5 million in senior debt.
This scenario plays out thousands of times a year in middle-market direct lending. The central question is not whether adjustments exist — every operating business has them — but whether each adjustment is sustainable, transferable, verifiable, and non-recurring. The difference between a rigorously underwritten EBITDA and a management-presented one can shift implied leverage by a full turn or more.
According to the Financial Stability Board’s May 2026 report on private credit vulnerabilities, private credit borrowers typically report leverage of 5x–6x debt to EBITDA, but once aggressive add-backs are stripped out, true leverage may be closer to 7x. That is not a rounding error — it is a structural risk embedded in the EBITDA definition itself. For any lender operating in this market, normalization discipline is not optional. It is the foundation on which every other underwriting judgment rests.
The Analytical Framework: Four Categories of EBITDA Adjustments
Institutional lenders organize add-back analysis into four tiers, each carrying a different burden of proof and level of lender scrutiny.
Tier 1 — Clearly Defensible
These adjustments represent costs that genuinely do not recur and can be documented to source records. They include: one-time litigation settlements tied to a specific resolved matter; transaction advisory fees directly related to the sale process; and non-cash charges such as stock-based compensation (where the lender independently assesses ongoing retention costs).
Lender stance: Accept with documentation. Require invoices, bank payment records, or board minutes. A legal fee add-back is supportable when matched to a settlement invoice and confirmed bank payment — not when it appears only in a management Excel schedule.
Tier 2 — Grey Zone (Requires Haircut)
These adjustments are conceptually valid but frequently overstated. The most common are: (a) excess owner compensation — the add-back is only the delta between actual W-2 and a market-rate replacement cost, not the full salary; (b) related-party rent normalization — the add-back is limited to the difference between actual lease payments and a comparable market rate, verified with third-party comps; and (c) restructuring charges — only if the restructuring is complete, with no further costs expected.
Lender stance: Haircut aggressively. Require payroll records for compensation, CoStar comps for rent, and a signed lease termination notice for restructuring. Lenders should independently calculate replacement cost rather than accepting management’s estimate.
Tier 3 — Run-Rate and Pro Forma (High Skepticism)
These are the most dangerous category in a credit file. Run-rate adjustments annualize cost savings or revenue from recent changes — a new customer contract signed three months ago, a headcount reduction implemented last quarter. Pro forma adjustments project future benefits from an acquisition or restructuring not yet reflected in historical financials.
Lender stance: A run-rate argument is only credible if supported by at least two full quarters of actual results. Annualizing a cost reduction implemented six weeks before measurement is a projection embedded in historical EBITDA — a category error that should be rejected or significantly haircut.
Tier 4 — Red Flags (Reject)
Certain adjustments should be rejected regardless of how they are framed: normalized revenue from a non-recurring customer windfall (e.g., hurricane emergency orders); “synergy” savings that depend entirely on post-close actions the new owner has not yet taken; deferred maintenance that has been capitalized rather than expensed; and EBITDA adjustments that conflict with working capital trends or cash conversion.
Lender stance: Reject and document the reason in the credit file. If the borrower pushes back, view the resistance as a due diligence signal about management’s credibility.
The Analysis: Walking Through a Real Adjustment Schedule
Below is a representative normalization walk for a $10M EBITDA manufacturing business — illustrating how reported EBITDA, seller-adjusted EBITDA, and lender-underwritten EBITDA diverge at each step.
| # | Adjustment Item | Seller Add-Back | Lender Underwriting | Rationale |
|---|---|---|---|---|
| — | Reported EBITDA | $10,000,000 | $10,000,000 | Per audited financials, TTM |
| 1 | Owner compensation excess | +$500,000 | +$300,000 | Actual W-2 $540K; market-rate CEO replacement estimated at $240K. Lender haircuts to the market differential only. |
| 2 | Related-party rent normalization | +$180,000 | +$120,000 | Actual rent $480K to owner LLC; market comp from third-party broker indicates $360K. Lender accepts verified delta only. |
| 3 | One-time legal settlement | +$165,000 | +$165,000 | Wrongful termination suit, settled with prejudice. Bank payment confirmed. No ongoing exposure. |
| 4 | Restructuring / severance (plant closure) | +$85,000 | +$50,000 | Closure confirmed. Lender haircuts 40% for potential lingering integration costs not yet invoiced. |
| 5 | Run-rate savings (new ERP system) | +$400,000 | $0 | ERP went live 6 weeks ago. No full quarter of demonstrated savings. Rejected as projection in historical EBITDA. |
| 6 | One-time revenue windfall (emergency order) | $0 (not removed) | −$175,000 | Lender identifies non-recurring customer order in revenue. Subtracts from EBITDA — adjustments must go both directions. |
| 7 | Discretionary owner perks | +$95,000 | +$75,000 | Club memberships and personal vehicle accepted; spouse payroll add-back haircut pending job description verification. |
| Adjusted / Lender-Underwritten EBITDA | $11,425,000 | $10,535,000 | ~$890K gap (8%) between seller and lender view | |
At a 5.0x leverage multiple, the seller’s adjusted EBITDA supports $57.1M in debt. The lender-underwritten base supports $52.7M — a difference of $4.4 million. Importantly, the cleanest credit memo shows both figures in parallel, with an explicit bridge between them, so the credit committee can evaluate where the underwriter exercised judgment and why.
“The cleanest credit memo shows both borrower-adjusted EBITDA and lender-adjusted EBITDA, with a clear bridge between them.”
— Crediflow Credit Analysis, August 2026
Industry Standards: What Institutional Lenders Require
Leverage Benchmarks in the Middle Market
Understanding where your underwritten EBITDA lands in the capital structure is the direct output of this normalization work. Current market benchmarks provide important guardrails. Mid-market leveraged buyouts typically carry total debt of 4.0x–5.0x EBITDA as market standard, with lower-middle-market transactions (sub-$25M EBITDA) generally ranging between 3.5x–4.5x. Anything exceeding 5.5x in the middle market requires exceptional credit strength or structuring justification.
Covenant Design Must Follow the Earnings Base
The underwritten EBITDA number is not just a valuation input — it is the foundation for every maintenance covenant in the credit agreement. Leverage maintenance tests are typically set with a cushion of 25%–35% above the borrower’s own base-case projections, then applied quarterly. That cushion is only meaningful if the underlying EBITDA definition is rigorously controlled. When covenants are written against an inflated EBITDA definition, leverage tests can allow earnings to fall 30%–40% before triggering a breach — eliminating the early-warning function that covenants exist to provide.
The structural risk here is compounding: in the upper-middle market, roughly 30% of recent deals are now covenant-lite, up from just 5% a decade ago, reducing lender visibility into deteriorating credit performance. Without maintenance covenants, lenders typically gain control only at payment default — when leverage may already be at 8x–10x and recovery values have eroded significantly.
📊 Credit Performance Data Box
- Private credit default rate (2025, Moody’s): Estimated at 1.6%–4.7% depending on whether distressed exchanges are included. Distressed restructurings accounted for approximately 65% of all credit events in 2025.
- First-lien senior secured recovery rate (Moody’s, 2024 expected): 57.6% across North America and Europe — below the long-term historical average of 65.0%. Structural deterioration, flexible documentation, and reduced junior debt cushions are the primary drivers.
- Covenant-lite risk penalty (Moody’s): Expected recovery on cov-lite first-lien structures is approximately 50%, versus 65% for structures that retain financial maintenance covenants — a 15-point recovery differential tied directly to covenant design.
- Distressed restructuring recidivism (Moody’s Default & Recovery Database, 1979–2026): More than 70% of eventual hard defaults following a distressed restructuring occur within the first two years of the soft credit event. Borrowers that restructured in 2023–2024 are now in their most vulnerable window.
- Selective defaults (S&P, 2024): Distressed exchanges classified as selective defaults outpaced conventional defaults by a 5:1 ratio in 2024, driven by flexible covenants and payment-in-kind conversions — highlighting how covenant erosion translates directly into disguised credit deterioration.
- Middle-market leverage benchmarks (Bain & Company / PGIM, 2024): Debt-to-EBITDA ratios for U.S. LBOs averaged approximately 4.9x in 2024, below pre-pandemic peaks. At least one financial maintenance covenant is standard for private credit deals under $50M EBITDA.
Common Red Flags in EBITDA Schedules
- ⚠ Recurring “one-time” items: If a cost appears in the add-back schedule across three of the last five years, it is maintenance expense — not a normalization. Reclassify it.
- ⚠ Run-rate adjustments without track record: Any run-rate savings argument must be supported by at least two full quarters of actual post-change results. Projections embedded in historical EBITDA are a category error.
- ⚠ Aggressive pro forma M&A add-backs: Anticipated cost savings from acquisitions should be rejected if they depend on post-closing integration actions not yet completed or demonstrated.
- ⚠ EBITDA-to-cash conversion disconnect: If normalized EBITDA is rising but operating cash flow and working capital are deteriorating, the adjustments are masking a real economic problem. Always reconcile EBITDA to cash flow from operations.
- ⚠ Deferred maintenance treated as non-recurring: Skipping fleet replacements or delaying facility repairs to boost near-term EBITDA is a liability, not an add-back. Sophisticated lenders will impute normalized maintenance capex.
- ⚠ EBITDA definition drift between covenant and underwriting: The contractual EBITDA in the credit agreement may permit adjustments that a credit officer would haircut for underwriting purposes. The gap between defined EBITDA and underwritten EBITDA is often the single largest unpriced risk in a credit file.
The EBITDA Normalization Underwriting Checklist
Use this framework to structure the EBITDA normalization process on every new credit. Each step should be documented in the credit file with supporting evidence.
- Step 1 — Start with historical financials, not projections. Pull three years of audited financials plus the most recent trailing twelve months (TTM). Spread each year independently before comparing.
- Step 2 — Classify every proposed add-back. Apply the four-tier framework: Clearly Defensible / Grey Zone / Run-Rate / Red Flag. Document the classification rationale in the credit memo.
- Step 3 — Verify to source documents. Each adjustment must tie to financial statements, tax returns, bank statements, payroll records, invoices, board minutes, or general ledger detail. An add-back that exists only in a management-prepared Excel schedule should be rejected or held pending documentation.
- Step 4 — Apply independent market benchmarks. For owner compensation, use published compensation guides (e.g., Robert Half, Radford) for the specific role and geography. For related-party rent, require third-party market comps (CoStar or equivalent) for comparable industrial or commercial space.
- Step 5 — Check for replacement costs and working capital effects. Before accepting an add-back, ask: if this cost goes away, does the business still function? Does removing it create a hidden future cost? Does it alter working capital or cash conversion?
- Step 6 — Build the lender’s own EBITDA bridge. Construct a parallel schedule showing reported EBITDA, borrower-adjusted EBITDA, and lender-underwritten EBITDA with explicit haircuts and rejections at each line item.
- Step 7 — Stress-test the underwritten EBITDA. Apply a 15%–25% downside scenario to the lender-adjusted number. At that stressed level, does the coverage ratio remain above 1.25x? Does leverage stay inside the covenant cushion? If not, the structure requires additional protection.
- Step 8 — Reconcile EBITDA to cash flow. Compare normalized EBITDA to operating cash flow and free cash flow over the same periods. A persistent and growing gap is a signal that adjustments are masking a structural cash problem, not a reporting artifact.
- Step 9 — Lock the covenant EBITDA definition. Ensure the credit agreement’s defined EBITDA is consistent with the underwritten number — not materially more permissive. The covenant should have teeth proportional to the risk. For sub-$50M EBITDA borrowers, insist on at least one financial maintenance covenant tested quarterly.
- Step 10 — Document everything. The credit file should contain the full adjustment schedule, tier classification, source documents reviewed, items haircut or rejected with reasons, and the reconciliation to cash flow. This documentation is the audit trail for the credit decision — and the portfolio monitoring baseline.
Key Takeaways
- EBITDA normalization is not about inflating a deal into viability — it is about establishing a credible, evidence-backed earnings base that every downstream underwriting decision depends on.
- The gap between seller-adjusted EBITDA and lender-underwritten EBITDA directly determines effective leverage. At 5.0x, a $900K difference in the earnings base translates to $4.5M in structural overcapitalization — a first-lien exposure that may not be recoverable in a downside scenario.
- Recovery risk is real and structural. Moody’s estimates that first-lien recovery rates on cov-lite structures fall to approximately 50% — a 15-point discount to structures with financial maintenance covenants. The EBITDA definition in the credit agreement directly influences this outcome.
- Run-rate and pro forma add-backs are the highest-risk category in any credit file. Require at least two full quarters of demonstrated actual results before crediting any run-rate savings. Projections embedded in historical EBITDA are a fundamental category error.
- Build the lender’s EBITDA bridge independently, in parallel with the borrower’s schedule. The credit memo must show both numbers with an explicit reconciliation. This is not bureaucratic process — it is the primary defense against a credit that looks good on entry and deteriorates within 18 months of close.
