A downside case is a specific, evidence-backed story about how an asset could perform worse than expected under a given capital structure, mapped line by line into the model. A stress test is a tougher, lower-probability path built to find the first point where the structure stops working: liquidity runs out, covenants break, refinancing fails, or control shifts.
Finance teams often treat these as percentages, like “take 10% off revenue, 200 bps off margin,” and call it a day. That’s quick, and it’s also how people talk themselves into leverage they don’t really understand. A real downside case has a causal chain you can underwrite. A real stress test forces an uncomfortable question: when the math turns, who pays first?
Why downside cases and stress tests matter to decision-makers
Downside and stress cases exist to answer three questions that every disciplined investment committee asks, whether they say it out loud or not. The payoff is practical: you learn which assumptions matter, which documents matter more, and which actions still exist once performance turns.
First, you need to know what can go wrong that is specific to this business and this set of documents. Second, you need to know how fast it shows up, and who sees it first: management, lenders, or the cash balance. Third, you need to know what actions preserve option value instead of turning equity into a passenger.
If your “downside” still keeps plenty of liquidity and never approaches a binding covenant, it’s usually a valuation sensitivity dressed up as risk work. Conversely, if your “stress” breaks everything in month three with no operational link, it’s a crash test with no steering wheel.
Start with failure modes (not margin clips) to find the real fragilities
The reliable way to build a downside is to start with the ways the deal can fail, then back-solve the operating and financial assumptions. In leveraged deals, the failure points cluster in a few places, and naming them early keeps the scenario honest.
Cash shortfall is the classic one: working capital expands, margins compress, or capex runs over plan, and liquidity drains faster than anyone modeled. Contractual triggers come next: maintenance covenants, minimum liquidity tests, borrowing base reductions, step-ups, or change-of-control clauses get tripped. Then comes the refinancing wall: a maturity arrives when the market is shut or the credit profile has weakened.
Two others are often under-modeled. Structural leakage is when cash is trapped or value shifts through baskets, incremental facilities, or priming claims. Exit impairment is when buyers disappear, multiples reset, or a public window closes right when you need it.
Pick the top three fragilities that are specific to the asset and structure. Then build cases so at least one failure bucket becomes relevant, and show the path from operating drivers to the trigger. If no trigger is even approached, you have learned little about the structure. If every trigger fires immediately, you may have skipped the business reality that makes the path plausible.
Calibrate severity with evidence and “time-to-pain”
Severity should come from data, not rules of thumb. When you tie assumptions to history and comparables, you also make the scenario easier to defend in writing and easier to debate line by line.
Start with company history. If the business has already faced a shock, that pattern is your first candidate downside. Adjust for mix, pricing power, and cost structure changes since then, and be explicit about what changed and why it matters.
Then use peer drawdowns. Find peers with similar end markets and cost behavior. Use their worst rolling 12- or 24-month revenue and margin moves as guardrails, and document the differences that will affect outcomes: contract length, variable cost share, geography, and customer concentration.
Finally, map macro to micro. Don’t pipe a survey into a revenue haircut. The Federal Reserve’s Senior Loan Officer Opinion Survey has shown periods of tighter credit standards for commercial and industrial loans and commercial real estate. For a business selling into discretionary capex cycles, that usually means slower bookings, longer sales cycles, and higher churn risk among leveraged customers. That has timing and cash consequences, which is what you care about.
The most important design variable is “time-to-pain.” Businesses absorb shocks through backlog, renewal notice periods, inventory buffers, and cost rigidity. A model that assumes the full-year impact arrives cleanly in January usually flatters liquidity and mis-times covenant pressure.
A better model shows lags explicitly: revenue lag from backlog burn and pipeline conversion; margin lag from hedge roll-offs and pricing cycles; cost lag from severance, leases, and minimum purchase commitments; and working capital lag from receivables aging, inventory mis-forecasting, and vendor term tightening. Timing is not trivia because a two-quarter lag can move the liquidity pinch into a refinancing window, which turns a manageable downturn into a control event.
Build scenarios as causal chains you can actually underwrite
A coherent downside is a chain: an external shock or company event leads to commercial impact, then operating response, then cash flow and balance sheet impact, then covenant and financing consequences, and finally a set of strategic options.
Volume-led downturn: model elasticity, lags, and working capital stress
A volume-led downturn is common in cyclical industrials, distribution, and discretionary services. The modeling work is in volume elasticity, the speed of variable cost reduction, and how sticky fixed costs really are.
Reduce bookings and shipments based on customer budget cycles and end-market indicators. Keep fixed costs sticky and let variable costs adjust with a lag so operating deleveraging shows up where it does in real life. Stress working capital as well: inventory can build before production cuts, then reverse later. Extend DSO and add bad debt expense if customers feel stress. Only model procurement savings if contracts support it.
Price-cost squeeze: make timing mismatch the center of the story
A price-cost squeeze is common in consumer, food, chemicals, and any business with pass-through friction. Here, timing mismatch is the villain, so the model has to respect how contracts and competition delay recovery.
Move input indices first, apply hedge coverage and roll-off schedules, and then apply price increases with real contract timing and competitive limits. If price increases exceed category tolerance, model volume loss. Let EBITDA compression drive covenant pressure, and don’t hide it in a blended margin assumption.
Working capital and liquidity shock: show how “good EBITDA” can still fail
Working capital and liquidity shocks often blindside teams, especially in rapid-growth software, milestone billing services, and inventory-heavy models. EBITDA can look healthy while cash collapses, which is why the scenario analysis needs cash mechanics rather than top-line overrides.
Slow collections and reduce prepayments or upfront billings. Increase inventory days when forecasts miss and suppliers enforce minimums. Tighten vendor terms when suppliers sense risk. Reduce revolver availability if the borrowing base depends on eligible receivables or inventory. Add the cost of liquidity tools like factoring fees or letters of credit. The impact tag here is simple: cash breaks first, and the lender sees it before the sponsor sees it in the IRR.
Event risk: build a timeline, costs, and recovery friction
Event risk, whether regulatory, litigation, or operational disruption, needs a timeline, not a shrug. A credible case specifies detection, remediation, and normalization timing, then translates that into cash timing.
Model one-time costs, lost revenue, and ongoing compliance opex. Test liquidity under delayed insurance recovery and coverage disputes. Tie the case to reps and warranties, indemnities, and insurance terms, and then haircut recovery for timing, caps, and counterparty credit.
Make sure the model can carry the story (or the scenario won’t teach you anything)
Downside work fails when the model can’t represent the narrative cleanly. Before building cases, make sure the architecture can handle causal drivers and legal mechanics.
Revenue needs a driver bridge: price, volume, mix, churn, new logos, and cross-sell, so you can tell the story without blunt overrides. Costs need behavior splits: fixed, semi-variable, and variable, so you can show operating leverage and lags. Working capital needs schedules by component: receivables, inventory, payables, and deferred revenue, so you can see why cash moves when EBITDA does not.
Debt schedules must reflect the paper: amortization, cash interest, PIK features, floors, caps, call protection, and fees. Covenants and baskets need an engine: maintenance tests, incurrence tests, restricted payments, incremental debt capacity, and EBITDA add-backs. And you need a liquidity waterfall with minimum liquidity, borrowing base mechanics, cash traps, and distribution blockers visible.
If the downside “works” only by forcing the cash flow statement to balance, you haven’t learned how the deal behaves. Liquidity trouble usually comes from the interaction of working capital, capex, and financing fees. Model those mechanically, or you’ll miss the very thing the stress test is supposed to find.
Define each case by what binds so the committee knows what changes
Skip labels like “mild,” “moderate,” and “severe.” Label cases by the first binding constraint, because that’s where decisions and control change.
- Covenant bind: Identify the first maintenance breach and the cure or amendment path.
- Liquidity bind: Hit a minimum cash-plus-availability threshold and show how quickly it arrives.
- Refinancing bind: Assume the debt cannot be refinanced at maturity under plausible market terms.
- Collateral bind: Capture borrowing base shrinkage, reserves, margin calls, or other collateral-driven constraints.
- Exit bind: Show target returns fail under plausible exit multiples and holding periods while deleveraging is limited.
A stress case should usually force at least one bind. If nothing binds, you have a sensitivity, not a stress.
Where stress shows up first: flow of funds and control rights
Most deal models focus on EBITDA and leverage. Stress shows up first in cash controls and priority of payments, so you should map cash through the actual structure.
Cash dominion and lockbox terms change the speed and visibility of deterioration. Revolver mechanics matter as well: borrowing base formulas, eligibility, concentration limits, and reserves can cut availability when you need it most. Model reserves and dilution haircuts, not just a revolver cap.
Restricted payments and leakage matter because baskets become the economic battlefield under pressure. Intercreditor priority matters because first lien, second lien, super senior revolvers, and priming liens determine who can force a restructuring and what recoveries look like. Hedging can create collateral posting requirements when positions move out of the money, which drains liquidity precisely when margins compress.
Include the fee and leakage stack that persists even as performance deteriorates: recurring sponsor and management fees where contractual, transaction fees, and OID amortization where they affect cash and covenants. Small items become big when liquidity is thin.
Match the “paper” to the model so the scenario is executable
Stress tests often assume flexibility the documents do not allow. Build a term-to-model checklist and reconcile it against executed or near-final drafts.
Credit agreements and fee letters govern covenants, baskets, pricing, floors, amortization, and reporting. Security documents govern collateral scope and enforcement. Intercreditor agreements govern standstills, payment blockages, lien priorities, and enforcement rights. Purchase agreements govern working capital true-ups, indemnities, earnouts, and closing conditions that change starting leverage and liquidity.
If a downside relies on indemnity recovery or escrow, test timing, caps, baskets, and indemnitor credit. On many sponsor-backed deals, practical recovery is limited, and slow.
Fees, accounting, tax, and regulation: the quiet drivers you still need to model
Model cash, not headline pricing. Include upfront fees and OID cash effects, commitment and unused line fees, letter of credit fees, and realistic amendment and waiver fees. If your covenant bind case assumes relief, model the economics: pricing step-ups, tighter covenants, extra reporting, additional collateral, and tighter restricted payments.
Accounting definitions matter because covenant EBITDA is negotiated, and lender tolerance for add-backs shrinks under pressure. Revenue recognition can shift revenue and cash differently under churn and downgrades. If there are JVs, SPVs, or receivables programs, check consolidation because that’s where off-balance-sheet liquidity can vanish when you need it.
Tax can change cash and upstreaming capacity, including withholding on intercompany payments, interest limitation rules like US 163(j), and leakage in restructurings. Keep regulatory edge cases short and concrete because timing is often the whole ballgame.
Model the real playbook: actions, constraints, and second-order effects
A downside case that says “management cuts costs” is not underwritten. Put actions into the model as an operating plan with constraints.
- Owner and approvals: Name the decision owner and approvals required, such as board, lender consent, unions, or regulators.
- Timing and ramp: State earliest implementation dates, ramp speed, and one-time costs.
- EBITDA vs cash: Separate accounting benefits from cash effects, since severance can consume cash before savings arrive.
- Second-order effects: Show how cuts can raise churn, reduce uptime, or shrink the buyer universe at exit.
Include a delayed-response variant because teams often react late, and a two-quarter delay can turn a waiver into a restructuring. Include an overreaction variant too, since harsh cuts can preserve liquidity and impair the franchise, shrinking exit options.
A fresh angle: treat scenario design like “information security” for your model
Scenario work improves when you treat it as a control function, not just analysis. In practice, that means you should ask where incentives, reporting cadence, and model governance could allow risk to hide until it is too late.
Start by aligning scenarios with the management reporting cycle and lender reporting requirements. If management sees KPIs monthly but lenders see a borrowing base weekly, the first “alarm bell” might be the revolver, not the P&L. Next, tie scenarios to early-warning indicators that can be operationalized, such as backlog coverage, customer credit downgrades, vendor term changes, or churn cohort behavior. Finally, record assumptions and overrides with the same rigor you would apply to access rights, because the fastest way to lose credibility is an untraceable “plug” that drives the break.
Change the structure when the downside is unacceptable
If the downside reveals a bind you can’t live with, the decision is often “change the deal,” not “walk.” Reduce leverage, change amortization, add committed liquidity, extend maturities, tighten cash controls and reporting, or use seller financing or earnouts to reduce day-one cash outlay while recognizing enforcement and subordination issues.
The point is to make structure serve the business reality, not to make the model serve the purchase price. If you need a refresher on how scenario frameworks differ from sensitivities, see key modeling terms and essential reading for analysts and investors.
Closeout pattern for scenario records
Archive the scenario book and supporting files with an index, version history, Q&A log, user list, and full audit logs. Hash the archived package so later disputes have a clean integrity check. Apply a retention schedule, then require vendor deletion with a destruction certificate when retention ends. Keep legal holds above deletion, every time.
Conclusion
A strong downside case explains how performance deteriorates and why, while a real stress test finds the first binding constraint that changes control. When you build scenarios from failure modes, calibrate them with evidence and time-to-pain, and match them to the legal documents, you stop debating abstract “haircuts” and start underwriting the structure that actually governs outcomes.