Technical skills in private credit mean the ability to turn a credit agreement into numbers you can trust. Underwriting, in plain English, is the work of translating contractual cash flows, covenant terms, collateral rights, and messy operating reality into downside outcomes and decision thresholds.
Private credit underwriting is a modelling problem constrained by legal documentation and imperfect data. The core task is to build models that stay faithful to the documents, resist optimism drift, and rerun quickly under stress without breaking. If the model cannot be re-run in a live investment committee discussion, it is not a tool. It is a liability.
This article explains what “good” private credit modeling looks like in practice, so you can move faster in diligence, identify the real failure mode early, and communicate risk in decision-ready terms.
What modeling means in private credit (and what it does not)
A private credit model is a cash-flow and covenant compliance engine, not a valuation engine. It should answer four questions with minimal discretionary judgment.
First, can the borrower pay scheduled debt service in the base case and in a downside case using operating cash flow and available liquidity? Second, if it cannot, what fails first: a covenant test, a borrowing base deficiency, a liquidity shortfall, or a maturity wall? Third, when something fails, what can the lender do, given cure rights, grace periods, blocker thresholds, and the practical limits of enforcing security? Finally, if you end up in a work-out, what does recovery look like after seniority, guarantees, intercreditor terms, and enforcement friction take their share?
In practice, the “model” is a set of linked artifacts: a standardized input pack, projections, a contractual debt schedule, covenant and basket calculators tied to defined terms, and a scenario engine that can shock revenue, margins, working capital, capex, and rates without rewiring formulas. The scorecard is auditability and speed of reruns. Elegance does not pay the coupons.
A useful rule of thumb for investment committee
A model is decision-grade when you can explain every major output by pointing to either (1) a source document, (2) a defined term, or (3) a small set of explicit downside assumptions. If an output depends on “hand judgment” that you cannot re-create under pressure, it will not survive a hard committee.
Credit analysis sets the boundary conditions
Credit analysis constrains the model. It forces you to separate cash flows that are repeatable from cash flows that depend on a sponsor’s mood, friendly vendors, or accommodating markets. The analyst’s job is to identify operating leverage, balance sheet optionality, and lender control points.
The product may change the emphasis, but the technical spine stays the same.
Senior secured unitranche work leans on lien priority, covenant protections, and refinancing risk at maturity. Second lien and mezzanine work turns on standstill periods, payment blockages, and how long liquidity lasts before the junior gets a seat at the table. Asset-based lending lives and dies on eligibility, dilution, reserves, and borrowing base volatility. NAV lending hinges on concentration, valuation governance, leverage caps, and sweep mechanics. Real estate and infrastructure underwriting is about cash yield, DSCR, contract tenor, and sponsor support covenants that have real teeth.
Here is the simple boundary condition. If the downside case requires discretionary sponsor support, unsecured enterprise value, or easy capital markets access, the deal is not structurally senior no matter what the slide deck calls it. Labels do not rank claims. Documents do.
Data hygiene and the underwriting file
Private credit runs on information asymmetry. Management reporting is often unaudited, charts of accounts wander over time, and addbacks get negotiated like price. Technical skill starts with normalizing inputs and documenting each transformation so you can defend the result later, especially in an amendment or dispute.
A decision-useful underwriting file typically includes audited financials with notes, monthly management accounts, a bridge from audited results to run-rate, a detailed debt and lease schedule, customer and revenue concentration, backlog and churn where relevant, inventory and receivables aging for asset-based situations, capex split between maintenance and growth, and tax payment history. In sponsor-backed deals, you want the acquisition model outputs and purchase agreement economics, but you should not let them dictate the base case. Sponsors are paid to be optimistic.
Tie the model to a “source of truth” tab that records origin, period, and adjustments for each key figure. That discipline reduces time in diligence, reduces argument later, and improves close certainty. When definitions and knowledge get tested, you will want a clean record of what you relied on.
Fresh angle: build a “definition ledger” to prevent optimism drift
One practical upgrade that pays off immediately is a definition ledger: a short table that lists the exact defined terms you rely on (EBITDA, Consolidated Net Income, Permitted Indebtedness, Restricted Payments, Unrestricted Subsidiaries, Available Amount) and the model cell ranges that implement them. This is not busywork. It is how you keep addbacks, baskets, and pro forma adjustments from quietly expanding across drafts, versions, and late-night markups.
Build the operating model for stress, not precision
In private credit, a model that looks precise can still be wrong. Build for stress. Let the base case use management guidance if you must, but make the downside case mechanical and defensible.
Forecast revenue using drivers you can shock. For subscription businesses, model customers, churn, net retention, and price. For distribution and manufacturing, separate volume from price so you can model mix and input costs without hand-waving. For project and services revenue, spell out backlog burn and conversion rather than assuming a smooth growth rate.
Model margins by separating fixed and variable components. One EBITDA margin line hides the risk you are paid to take. In a contraction, gross margin compresses, SG&A stays sticky, and working capital often reverses. Give yourself levers: a gross margin shock, an SG&A elasticity assumption, and a restructuring case with one-time costs and a clear path to run-rate savings.
Working capital is the most common liquidity surprise. A single “change in working capital” plug is fine only when the deal has wide liquidity and low leverage, which is not what most private credit looks like. When the structure is tight, model DSO, DIO, and DPO so the cash conversion cycle moves with the business. If receivables drive revolver availability, include dilution, returns, and concentration because eligibility governs the cash you can actually draw.
Capex must be split between maintenance and growth. EBITDA can ignore capex. Cash cannot. The underwriting question is whether maintenance capex can be deferred without impairing uptime, compliance, or revenue, and how quickly the bill shows up.
The cash bridge that drives the credit decision
Private credit is governed by cash, not accrual. The centerpiece of the model should be a cash bridge from EBITDA to cash available for debt service.
Start with EBITDA. Subtract cash taxes, maintenance capex, and cash interest. Adjust for working capital. Subtract mandatory amortization and material lease payments. Add or subtract one-time cash items with explicit timing. The output is free cash flow after debt service and the resulting cash build or burn.
Keep two views side by side: accounting cash flow from operations and “credit cash flow” used in covenant definitions. A borrower can be covenant-compliant while burning cash if addbacks inflate EBITDA and working capital deteriorates. Your job is to make that visible early, because timing risk shows up first in cash, not in ratios.
Debt schedule mechanics: model the contract, not the idea
A private credit debt schedule must replicate the credit agreement. That means rate definitions, floors, margins, day counts, payment frequency, amortization, optional prepayments, call protection, and default interest.
Floating-rate loans need explicit base rates and reset timing. Term SOFR in the US and SONIA in the UK dominate many structures. Floors matter, and so do the months when the rate resets and the cash interest actually leaves the account. For hedged borrowers, model swaps or caps as separate cash flows and confirm the hedge profile matches the debt profile. If the hedge rolls off before maturity, model the cliff. That is where risk migrates.
PIK toggles should be modeled as principal accretion that raises future interest and changes covenant ratios. Revolvers and delayed draws need availability logic. If availability depends on a borrowing base or springing covenant, do not assume the line is there just because the commitment exists.
For multi-tranche stacks, follow the waterfall. Many models quietly pay down the highest-cost tranche first. Agreements often require pro rata sharing or restrict junior payments under an intercreditor. A small error in the waterfall can flip your recovery conclusion. Small errors are expensive.
Covenant modeling: definitions first, math second
Covenants are defined terms plus testing mechanics plus cure rights. Ratios come last. The covenant model must mirror the agreement, not management’s adjusted EBITDA slide.
Maintenance covenants often include total leverage, first lien leverage, FCCR, interest coverage, and minimum liquidity. Incurrence tests live in restricted payments, investments, debt and liens, and asset sales. Even deals marketed as covenant-light may still have springing revolver tests or liquidity triggers.
Technical requirements are straightforward, but they take discipline. Build an EBITDA addback schedule with caps, baskets, and time limits. Separate “run-rate cost savings” from extraordinary items and apply the contract’s limits, not the sponsor’s enthusiasm. Model net debt the way the agreement defines it, including restricted cash, trapped cash, cash outside the group, and letters of credit if included. Track pro forma mechanics for acquisitions and dispositions only to the extent the agreement permits them. If there is an equity cure, model capacity, timing, and effect exactly as drafted; cures often have annual limits and may change ratios without adding cash.
Report headroom as ratio cushion and as implied EBITDA decline or debt paydown that would trigger a breach. Decision makers think in operational terms: “How much can the business miss before we lose control?” Give them the answer.
- Defined terms: Copy definitions into the model or reference them explicitly so the math cannot drift from the document.
- Addback discipline: Apply caps and sunset periods, and track cumulative usage, not just current-period adjustments.
- Cure realism: Treat equity cures as a governance and timing question, not a magical ratio fix.
Liquidity runway and default timing: the output that matters
If you remember one thing, remember this: liquidity runway is central. A borrower can look solvent and still default because cash timing fails.
When liquidity is tight, use a month-by-month cash forecast even if you keep annual projections for the long view. Incorporate cash on hand, revolver availability net of borrowing base and reserves, seasonal working capital, interest months, tax months, earnouts, and deferred consideration. Layer in covenant test dates, notice periods, and grace periods, because remedies depend on when a default is recognized and can be declared.
Identify the first failure mode in each scenario. If the first failure is a covenant trip while liquidity remains adequate, the lender has negotiating leverage and time. If the first failure is cash, options narrow quickly and recovery risk rises. Timing changes outcomes, and outcomes drive returns.
Collateral and security: turn legal rights into model inputs
Security packages get summarized, then forgotten. That habit costs money. Collateral affects recovery timing, enforcement cost, and whether someone can prime you.
Start with collateral scope and perfection. Confirm first-priority liens on substantially all assets and list excluded assets, foreign subsidiary equity, real estate, IP, or key contracts. Confirm deposit account and securities account control. If cash can move without your control, the lien can look strong on paper and weak in practice. Control agreements raise close certainty and improve recovery optics in stress.
Guarantee coverage matters as much as collateral. Map the guarantor group and its limitations: fraudulent conveyance, financial assistance, and local law constraints. In cross-border structures, assess upstream guarantee enforceability and whether withholding taxes, exchange controls, or trapped cash impede cash movement.
In ABL structures, model eligibility and reserves. Borrowing base availability is the real liquidity, not the stated commitment.
Documentation literacy is a technical skill
Private credit analysts must read documents like technicians. Build a document-to-model map that ties each major mechanic to the relevant sections of the credit agreement, intercreditor agreement, and security documents.
Pull from the credit agreement: interest mechanics, amortization, optional prepayments, covenants, reporting, events of default, mandatory prepayments, and definitions. Pull from the fee letter: OID, ticking fees, unused fees, and when they are payable. Pull from the intercreditor: standstill, payment blockage, lien priority, turnover, and enforcement control. Pull from the security agreement and mortgages: collateral scope, perfection steps, remedies. Pull from the guaranty: scope, limitations, release mechanics. If there is hedging, include cash settlement and termination events.
Execution order matters. If perfection steps are deferred post-close, list them with deadlines and consequences. A missing control agreement is not a checklist item. It changes the recovery thesis.
Fee stack and economics: measure what reduces lender IRR
“Spread plus OID” is shorthand. It is not a return calculation. Compute lender IRR and money multiple net of fees and expected draw timing, and show who pays what and when.
Model OID, upfront fees, unused commitment fees, ticking fees between signing and funding, amendment fees, and prepayment premiums. Even if the borrower pays agency and servicing costs, those costs still affect credit risk through cash leakage. In cross-border structures, model withholding and gross-up clauses and make a sober call on treaty relief based on lender domicile and documentation realities.
PIK features and exit fees matter because they compound and they raise the refinancing burden at maturity. Higher effective yield is welcome, but it can also increase default probability. Your model should make that trade-off plain.
Scenario design and recovery: the sober view
Stress testing should follow real failure modes, not arbitrary percentage cuts. At minimum, run revenue decline with margin compression, working capital outflow, capex rigidity, customer loss or contract non-renewal, input cost spikes, and delayed price pass-through. For cyclicals, model a two-year trough with a slow recovery. For sponsor-backed deals, include a refinancing stress where the borrower must refinance under higher rates and tighter leverage.
Rates matter because floating-rate loans transmit policy changes into interest expense fast. Shock base rates and test hedge expiry and break costs. The output should include minimum liquidity, covenant headroom over time, a default date if any, and a break-even table that shows what EBITDA decline or working capital outflow triggers failure.
Recovery work should be haircut-driven and net of friction. Use multiple frames, going-concern value under a downside multiple, liquidation value of hard assets net of costs, and stabilized cash generation value, then reconcile them. Subtract priming claims such as ABL, any permitted priming liens, taxes that prime in certain jurisdictions, and administrative costs. Model time to recovery because time drives IRR. A high nominal recovery in two years can still disappoint.
Jurisdiction affects bargaining and timing. US Chapter 11 differs from UK restructuring plans and many European processes. Even if you do not simulate the process, reflect whether venue increases timing risk or priming risk.
Monitoring and the closeout discipline
A good underwriting model becomes a portfolio tool. Build it so it can be refreshed with monthly reporting inputs, covenant calculators, and variance analysis. Require the reporting package in the documents: monthly financials when leverage is high or liquidity is thin; borrowing base certificates for ABL; budgets and variance explanations; KPIs tied to drivers like churn, backlog, utilization, and pricing.
Use operational triggers, not only covenant triggers. Covenants can trip late. Margin erosion, DSO creep, inventory build, and capex deferral can show stress earlier and preserve options.
When the work is done, treat the underwriting record like an asset. Archive the index, versions, Q&A, users, and full audit logs. Hash the final package. Apply retention rules. If a vendor hosts any part of the data, require deletion and a destruction certificate. And remember the simple hierarchy: legal holds override deletion.
Key Takeaway
Private credit modeling is not about a perfect forecast. It is about a fast, auditable engine that turns documents into cash, covenants, and control, so you can see the first failure mode early and price, structure, and negotiate accordingly.
Related Reading
- Financial covenants
- security packages
- intercreditor agreements
- asset-based lending
- unitranche
- Debt scheduling in financial modeling
- Stress testing financial models
Live Source Verification
I selected the sources below because they are stable, public references that define or explain core concepts used in private credit underwriting (EBITDA, covenants, borrowing bases, SOFR, and restructuring/recovery mechanics). Each link points to a well-known publisher page that is typically accessible without paywalls and suitable for reader verification.