A direct lending model built from the borrower’s perspective will pass internal review. It will not pass a credit committee. This guide explains the structural differences — yield-on-cost, PIK toggle, covenant headroom, and lender-side IRR — and how to build each component correctly in Excel.
Private credit has moved from a niche asset class to a core component of institutional portfolios. Blackstone Credit, Apollo, Ares, and dozens of mid-market direct lenders in New York and across the United States now deploy billions annually — often in competition with, rather than alongside, traditional bank financing. Yet most financial models circulating in deal rooms are built from the borrower’s perspective: they optimize for leverage, coverage, and equity returns.
What they miss is the other side of the table.
A direct lender’s credit team models a deal completely differently. They start with yield, not EBITDA. They stress covenant headroom, not IRR. They toggle PIK versus cash-pay interest not as a feature — but as a risk signal. If you’re working on private credit transactions, sitting in a credit fund, or modeling a deal where direct lending is part of the capital structure, building the lender’s perspective into your model is not optional.
Here’s how to do it.
Why a Standard LBO Model Won’t Survive Credit Committee Scrutiny
Most analysts who’ve built LBO models are comfortable with three-statement integration, debt schedules, and returns analysis. But here’s what we see when deal teams bring direct lending models into our review sessions in New York: the model answers the wrong questions.
An LBO model is built to answer: “What equity return does the sponsor achieve given these assumptions?” A direct lending model must answer a different set of questions entirely:
- What is the all-in yield on this loan, and does it compensate for the credit risk?
- At what EBITDA level does the borrower breach its maintenance covenant — and what is the cushion today?
- If the borrower elects PIK for two consecutive periods, what happens to the lender’s effective yield and the debt quantum?
- Under a downside scenario, is there enough asset coverage for the lender to recover principal?
- What is the lender’s cash-on-cash return and gross IRR across the investment horizon?
These are structurally different outputs from a structurally different model. Building them requires adding layers to what you already know — not rebuilding from scratch.
What’s missing from most direct lending models Most models we review have a debt schedule, but no PIK toggle. They show interest expense, but not yield-on-cost. They include a covenant table as a static exhibit — not as a live, formula-driven output. These are not minor omissions. They are the components credit committees use to make decisions.
The Lender’s Model Architecture: Six Components That Must Be Built Correctly
1. Yield-on-Cost: The Starting Point, Not the Result
In a traditional LBO model, interest expense is a cost to the equity sponsor. In a direct lending model, it is the primary revenue driver. Yield-on-cost (YOC) represents the total return the lender earns on deployed capital — and it must be modeled explicitly, not backed into.
YOC = (Cash Interest + PIK Interest + Upfront Fee Amortization + OID Amortization) / Average Outstanding Balance
Each component needs its own row in the model. The upfront fee and OID (original issue discount) are not one-time cash items — they amortize over the loan life and materially affect YOC calculations. Missing them understates the lender’s return and overstates deal risk.
2. PIK Toggle: Model It as a Decision, Not a Default
Paid-in-kind (PIK) interest means the borrower does not pay cash interest — instead, the outstanding balance increases by the interest amount each period. For the lender, PIK is a double-edged structure: it preserves the borrower’s liquidity but compounds the debt burden and increases recovery risk.
The model needs a PIK toggle — a single input cell (typically 0 or 1, or a percentage split) that drives four separate calculations simultaneously:
- The cash interest line in the borrower’s income statement and cash flow
- The PIK interest accrual that increases the outstanding principal balance
- The lender’s effective yield recalculation based on the new outstanding balance
- The covenant headroom recalculation, since PIK increases leverage metrics each period
A PIK toggle should never be hardcoded. Structuring it as a dynamic driver means you can run a sensitivity table comparing cash-pay versus full PIK versus 50/50 split in seconds — which is exactly what a credit committee will ask for.
3. Covenant Headroom: Live Output, Not a Static Exhibit
Maintenance covenants in direct lending agreements are typically set around leverage (Net Debt / EBITDA) and coverage (EBITDA / Cash Interest). The covenant levels are negotiated at close, but headroom — the distance between actual metrics and the covenant threshold — needs to be a live model output across the entire forecast horizon.
What this means structurally: the covenant table must be formula-driven, pulling from the model’s live income statement and debt schedule. The common mistake is inserting a static screenshot of Day 1 metrics into an appendix. That tells the credit team nothing about covenant risk in Year 2 or Year 3 under a downside scenario.
A properly built covenant section in a direct lending model includes:
- Leverage covenant: trailing twelve-month (TTM) EBITDA vs. total net debt, with the threshold clearly labeled
- Coverage covenant: TTM EBITDA vs. cash interest (not total interest — PIK is excluded from coverage tests in most agreements)
- Headroom calculation: absolute amount and percentage cushion, both live formula outputs
- Covenant breach flag: a conditional cell that highlights red when headroom falls below a defined threshold (commonly 15–20%)
The covenant test most models get wrong EBITDA for covenant purposes is rarely the same as accounting EBITDA. Most credit agreements allow addbacks — restructuring costs, one-time items, run-rate synergies. Your covenant calculation must reflect the contractual definition, not the GAAP line. This is the difference between a model that survives legal review and one that doesn’t.
4. Debt Schedule: Structure Before Returns
The debt schedule in a direct lending model differs from an LBO debt schedule in one important way: it must account for the full economics of the instrument, not just the principal and coupon.
A complete direct lending debt schedule includes:
- Opening principal balance
- PIK accrual additions (if applicable)
- Scheduled amortization (direct lending loans are typically bullet or minimal-amortization)
- Optional prepayment modeling with prepayment premium (call protection is material in credit economics)
- Closing balance — which feeds both the lender’s outstanding balance and the borrower’s leverage calculation
Call protection schedules — where early repayment triggers a fee, commonly structured as 2%/1%/0% over the first three years — must be modeled explicitly if the instrument includes them. They affect the lender’s yield in early-exit scenarios and are a material input to the IRR calculation.
5. Lender-Side IRR: Not the Same as Sponsor IRR
This is where most models conflate the borrower and lender perspectives entirely.
Sponsor IRR models equity returns: the delta between equity invested at entry and equity value at exit. Lender IRR models the return on debt capital: cash flows from origination to repayment, including fees, interest (cash and PIK), any call premium, and principal recovery.
The lender-side IRR calculation:
- Origination: negative cash flow equal to (principal deployed minus OID minus upfront fee retained)
- Each period: positive cash flow from cash interest received
- Each PIK period: no cash flow (PIK accrues but is not received)
- Exit: positive cash flow from principal repayment plus any call premium
- Output: XIRR across all dated cash flows
The distinction matters because lender IRR is what the credit fund reports to its LPs. It is the metric that drives mandate compliance, benchmark comparisons, and ultimately, capital allocation decisions. Building it correctly — including the OID and fee treatment — is non-negotiable in any professional direct lending model.
6. Downside Scenario: Recovery Analysis, Not Just EBITDA Sensitization
Standard LBO downside scenarios test equity returns under pressure. Direct lending downside scenarios ask a different question: if the borrower defaults, what does the lender recover?
A credit-focused downside scenario should include:
- Distressed EBITDA assumption driving the leverage and coverage metrics to breach
- Enterprise value in distress — typically a haircut to the base case exit multiple, often 4–5x versus 7–8x in the base
- Recovery waterfall: senior secured lenders are paid first, but the model must account for super-senior revolvers, transaction costs, and any structural subordination
- Recovery rate as a percentage of outstanding principal: the output credit committees use to set position sizing and covenants
How to Structure This in Excel
The model architecture should follow a clear separation between input, calculation, and output — the same principle that applies to any bank-grade financial model.
Recommended tab structure for a direct lending model:
- Assumptions — all deal inputs: rate, spread, OID, fees, amortization schedule, PIK toggle, covenant thresholds
- Operating Model — borrower’s three-statement (P&L, balance sheet, cash flow)
- Debt Schedule — principal balance, PIK accrual, amortization, call protection waterfall
- Credit Metrics — live covenant headroom, leverage, coverage — formula-driven from operating model and debt schedule
- Returns — lender IRR, YOC, cash-on-cash — base and downside
- Sensitivity Tables — YOC and IRR across EBITDA growth and spread scenarios; covenant headroom across leverage scenarios
Each tab should have a clear color-coding convention: blue for hard-coded inputs, black for formulas, green for output cells. Any analyst picking up the model should be able to identify at a glance where assumptions live and where outputs flow.
Where This Connects to What You Already Know
If you’ve worked through our LBO model tutorial or our credit metrics dashboard guide, the core mechanics here are extensions of existing frameworks — not replacements. The three-statement integration, debt schedule logic, and sensitivity table structure are identical. What changes is the perspective: you’re now building outputs that answer a lender’s questions, not a sponsor’s.
The PIK toggle and lender IRR are the highest-delta additions. Both are buildable in a few hours once the underlying debt schedule is clean. The covenant section is the component most often done incorrectly — not because it’s complex, but because analysts treat it as a reporting appendix rather than a live analytical tool.
If you’re preparing for a role at a direct lending fund, a business development company (BDC), or a credit-focused PE firm — understanding how to build and defend this model is what separates a candidate who knows LBOs from one who’s actually built credit analysis from the lender’s seat.
Working on a direct lending transaction or preparing for a credit-focused role? If you’re building or reviewing a direct lending model and need a framework that holds up in a credit committee — we work through exactly these mechanics in our training sessions, using live deal-based case studies. Let’s talk through what that looks like for your situation.
Frequently Asked Questions
What’s the difference between lender IRR and borrower IRR in a direct lending model? Lender IRR measures the return on debt capital deployed — including fees, OID, cash interest, PIK accrual, and principal recovery. Borrower IRR (or sponsor IRR) measures equity returns. The two are calculated on different cash flow streams and answer fundamentally different investment questions.
How does a PIK toggle affect covenant headroom? PIK interest increases the outstanding principal balance each period instead of being paid in cash. This directly increases the leverage metric (Net Debt / EBITDA) — which means PIK reduces covenant headroom automatically. A model without a live PIK-to-covenant linkage will understate credit risk in stressed scenarios.
What EBITDA definition should I use for covenant calculations? The covenant agreement defines it — not GAAP. Most direct lending agreements permit addbacks for one-time items, restructuring costs, and run-rate synergies. Your model must reflect the contractual EBITDA definition, not the accounting line, or the covenant calculation will be structurally incorrect.
Is direct lending modeling covered in your financial modeling courses? Credit and debt structure mechanics — including debt schedules, covenant modeling, and returns analysis — are covered in our advanced training tracks. If direct lending modeling is a specific objective, we recommend discussing a tailored session that covers the lender-perspective components described here.
For a structured breakdown of how direct lending mechanics connect to LBO and credit model architecture, the resources below are the right starting point.