Sensitivity Analysis: The Stress Test
Your Project Report Must Survive
Every financial projection is wrong — the only question is by how much, and in which direction. Sensitivity analysis is how lenders find out whether your project survives that inevitable error. Here is the complete method, with interactive simulators to stress-test your own numbers.
Lenders Don't Finance Forecasts. They Finance Resilience.
Open any project report and you will find a beautiful base case: revenue climbing year on year, margins holding steady, every loan instalment paid on schedule. Credit officers see hundreds of these — and they trust none of them at face value. Not because applicants lie, but because the future does not cooperate with spreadsheets. Commodity prices move. A competitor discounts. The ramp-up takes six months longer than planned.
So before sanctioning a loan, every serious lender asks the same quiet question: "What happens to repayment if these assumptions bend?" Sensitivity analysis is the formal, numerical answer. It re-runs your financial model with deliberately worsened inputs — sales down 10%, input costs up 10%, interest up a couple of points — and checks whether the project can still service its debt.
A projection tells the lender what you expect. Sensitivity analysis tells them what happens when you're wrong — and that second answer is the one that gets financed.
What Sensitivity Analysis Actually Is
Sensitivity analysis (also called what-if analysis) measures how much a project's output metrics — DSCR, break-even point, IRR, NPV — change when one input assumption changes and everything else is held constant. In modelling literature this is the one-variable-at-a-time (OVAT) approach, and its isolation of variables is precisely what makes it powerful: it tells you not just that your project is fragile, but where.
If a 10% revenue drop cuts EBITDA by 22%, the sensitivity coefficient is 2.2. Any variable with a coefficient above 1 amplifies shocks — and any such variable with a wide plausible range is a critical variable.
Where It Sits in the Risk-Analysis Family
Sensitivity analysis is the first of three escalating techniques. A bank-ready project report needs the first, and the second strengthens it. The third belongs to large institutional deals:
Move a single assumption (price, cost, volume) while holding everything else constant. Reveals which variable your project is most exposed to — the "critical variable".
Required in every DPRChange several variables together into internally consistent futures — best case, base case, worst case. Tests plausible storylines rather than isolated shocks.
Expected by most lendersAssign a probability range to each input and run thousands of random draws. Outputs the probability of default or negative NPV. Used for large or institutional projects.
Institutional-gradeThe Variables Lenders Actually Stress
You cannot stress everything — a sensitivity section that varies forty inputs is noise, not analysis. Professional practice concentrates on the handful of variables that are both uncertain and high-impact. Six dominate real credit appraisals:
| Variable | Typical Stress Band | What It Tests | Where It Hits |
|---|---|---|---|
| Selling price | −5% to −15% | Pricing power & competitive pressure | Margin — every lost point is pure profit |
| Sales volume / utilisation | −10% to −20% | Demand assumptions & capacity ramp-up | Contribution margin, break-even point |
| Raw material / input cost | +5% to +15% | Supply-chain and commodity exposure | Variable costs, gross margin |
| Interest rate | +100 to +300 bps | Financing risk on floating-rate debt | Debt service, DSCR directly |
| Project cost overrun | +10% capex | Execution & construction risk | Funding gap, depreciation, loan size |
| Receivable days | +15 to +30 days | Collection discipline of the market | Working capital, liquidity buffer |
The first three — selling price, volume, and input cost — form the minimum stress set for any project report. Interest-rate stress matters wherever the debt is floating-rate; capex overrun matters most for greenfield and construction-heavy projects; receivable-day stress matters for B2B businesses selling on credit.
Stress-Test Your Own Numbers
This simulator does what a credit analyst does on day one: it takes your base case, applies a revenue stress, and watches your DSCR. Crucially, it models operating leverage — your fixed costs don't shrink when sales do, so EBITDA falls faster than revenue. Set the sliders to your project and find the stress level where you stop clearing 1.25x:
Two things to notice as you drag. First, the amplification factor: with a typical cost structure, a 10% revenue dip produces a 20–25% EBITDA dip. Applicants who simply scale profit down 10% in their "stressed case" are flagged instantly by experienced reviewers — the arithmetic doesn't survive contact with a fixed-cost schedule. Second, the interaction with debt sizing: the same stress that is survivable at a modest loan size becomes fatal at an aggressive one. Sensitivity analysis is ultimately a loan-sizing tool.
The Tornado Chart: One Picture, Every Exposure
The classic way to present OVAT results is the tornado chart: every variable gets a horizontal bar showing the output's swing between the adverse and favorable case, sorted widest-first. The result looks like a tornado — and the widest bar at the top is, by definition, your critical variable: the assumption your project lives or dies by.
Read the chart the way a credit officer would. At a ±10% swing, this base case survives everything except a price shock — a 10% price cut lands DSCR exactly at 1.00x, the edge of default. That single insight should reshape the whole business plan: price protection (contracts, differentiation, switching costs) becomes the risk-mitigation story your report needs to tell. That is the real purpose of sensitivity analysis — not decorating the report, but directing management attention to the one number that matters most.
Presenting Sensitivity Analysis in a DPR
In a Detailed Project Report, the sensitivity section belongs after the base-case financial projections and ratio analysis, before the final viability conclusion — it is the evidence on which the viability claim rests. The standard presentation is a compact scenario table tracking the three metrics every appraiser anchors on: DSCR (repayment capacity), break-even point (margin of safety), and IRR (return above the cost of capital):
| Scenario | Avg DSCR | Break-even (% of sales) | Project IRR | Reading |
|---|---|---|---|---|
| Base case | 1.67x | 44% | 21.8% | Comfortable headroom |
| Sales volume −10% | 1.37x | 49% | 17.2% | Clears the floor |
| Input costs +10% | 1.30x | 51% | 16.1% | Clears the floor |
| Selling price −10% | 1.00x | 57% | 11.4% | Breach — mitigation required |
| Combined: volume −10%, costs +10% | 1.04x | 56% | 10.9% | Borderline — lender will probe |
Illustrative figures from the tornado chart's base case ($1.0M revenue, 55% variable costs, $200k fixed, $150k debt service). Your report's numbers must reconcile to your own financial model.
The Three Rules of a Credible Sensitivity Table
- Every scenario must reconcile to the base model. The stressed DSCR figures must be re-computed from the full financial model — P&L, repayment schedule, working capital — not approximated by scaling the base numbers. Reviewers spot-check this, and a table that doesn't tie back destroys trust in the entire report.
- Include one combined scenario. Real downturns arrive in packs: volume falls and costs rise. A single compound scenario (the last row above) is the strongest honesty signal a report can send — and lenders increasingly ask for it explicitly.
- Attach a mitigation line to every breach. A stressed case that dips below 1.25x is not fatal if the report states the response: cost pass-through clauses, promoter funds on standby, a debt-service reserve account, insurance. An unexplained breach is a rejection; an explained one is a risk-management conversation.
Six Mistakes That Quietly Destroy Credibility
Stress It Yourself, Before the Bank Does
Every project report will be stress-tested. The only choice you have is who runs the test first. If the lender runs it, you find out about your critical variable in a rejection letter. If you run it, the sensitivity section becomes your strongest exhibit: proof that you understand your own risk, that the loan is sized to survive a bad year, and that management has already planned the response to each plausible shock.
Lenders don't finance best cases. They finance projects that survive bad ones — and reward the applicants who prove it first.