Is your project finance model ready for submission?
Between an accepted file and a returned one, the difference rarely lies in the quality of the project itself. It lies in the financial model — its rigour, its traceability, its ability to withstand a lender’s questions. An evaluator — whether assessing a file for AFD, the AfDB or the World Bank — does not read a project finance model the way one reads a spreadsheet: they look for breaking points. Here is the checklist our team applies before every submission, ordered the way an evaluator actually inspects a model. It targets revenue-generating projects — infrastructure, energy, PPP — where debt coverage and cash-flow robustness sit at the heart of the analysis.
Structure and assumptions
A single, centralised assumptions sheet. Every assumption in one place, never buried in formulas. An evaluator must be able to change one input and watch the effect flow through — without hunting across twelve tabs. An assumption hard-coded inside a formula is an invisible assumption, and therefore a suspect one.
Every figure has a documented source. A rate, a price, a volume: where does it come from? Market study, contract, comparable, expert judgement. A figure without an origin is the first one the evaluator will challenge. Traceability isn’t a nicety — it’s what makes the model defensible.
Clear separation between assumptions, calculations and results. Three distinct zones. Never type data into a calculation cell. This discipline prevents silent errors and lets a third party understand the model without a manual.
Consistent currency unit and inflation treatment. Nominal or real — chosen, owned, consistent throughout. Inflation assumptions made explicit. A model that mixes nominal and real produces wrong ratios that go unnoticed until the one question too many.
Debt coverage ratios
DSCR calculated on CFADS, not EBITDA. This is what separates an expert model from an approximate one. Debt service is covered with cash actually available — CFADS, cash flow available for debt service — not with an accounting profit. Confusing the two artificially inflates coverage and is spotted immediately.
The minimum DSCR is met in every period. Not on average: in every period, across the whole loan life. Lenders set a floor in the term sheet — often in the region of 1.30x to 1.40x depending on sector and risk. A single dip below the threshold, even a brief one, is a stress signal the evaluator will catch.
LLCR and PLCR present, not just DSCR. The DSCR measures the short term, period by period. The LLCR (loan-life coverage) and PLCR (project-life coverage) measure long-term capacity. The three together tell the full debt story; DSCR alone only speaks to the present.
Debt sizing follows the cash-flow profile. Debt sizing and sculpting consistent with the project’s real flows — repayments shaped to period-by-period capacity, not a flat profile pasted onto irregular cash flows.
Profitability and sustainability
Positive financial NPV, discounted at the right rate. Financial NPV discounted at the weighted average cost of capital. A positive financial NPV means the project creates value beyond the cost of its financing. The discount rate used must be justified, not set arbitrarily.
Project IRR and equity IRR distinguished. Two different indicators, never conflated. The project IRR measures intrinsic profitability; the equity IRR, the return on equity after leverage. Presenting them separately shows command of the financing structure.
Fiscal sustainability for the public party. In a PPP, the evaluator also looks at what the project costs — and may cost — the public side. The logic of PFRAM (the PPP fiscal risk assessment model developed by the IMF and the World Bank) is the reference here: firm and contingent liabilities made visible.
Balancing subsidy quantified, if the project depends on one. If the project doesn’t stand without public support, the amount must be estimated and its assumptions explicit — not drowned in the model. An evaluator prefers an owned, justified subsidy to a dressed-up return.
Robustness: sensitivities and scenarios
Sensitivity analysis on the key variables. Prices, volumes, rates, construction delays: what happens to the model if each moves? A model that never breaks is a model that wasn’t pushed hard enough. The evaluator wants to see where the fragilities are, not a demonstration that all is well.
Scenarios, not just a central case. Base case, low case, stress case. A file presenting only an optimistic scenario disqualifies itself. Credibility comes from the ability to show the project under strain.
Contingencies provisioned. A reserve for uncertainties on investment costs — usually in the region of 5 to 15% depending on project maturity. Contingencies don’t replace good preparation, and they sit as a separate line, not a margin hidden in the costs.
Stress tests built in from the design stage. Designed into the model’s architecture, not bolted on at the end to tick a box. A model built to be stressed takes it cleanly; a model stressed afterwards cracks.
Auditability and format
An audit trail for every figure. Every result must trace back to its originating assumption. This is what lets an evaluator — or a lender, later — verify without rebuilding the model.
Zero formula errors, balance / flow consistency checked. Integrity checks in place and green. A single balance-sheet inconsistency calls the reliability of the whole into question. Cell-by-cell verification is not optional.
A format legible to a third party. The model will be read by someone who didn’t build it. Clear structure, named tabs, consistent colour conventions, obvious navigation. A model that is correct but unreadable loses half its value at submission.
A methodology note attached. A few pages explaining the logic, the choices, the limits. It turns a spreadsheet into a defensible file and saves the evaluator valuable time — for which they’ll be grateful.
In summary
These points are no guarantee of acceptance — no model is. But they are, in order, the ones a lender or an evaluator tests first. A model that clears them all won’t be rejected on technical grounds: it will be judged on the project, which is exactly what you want.
Unsure about one of these points in your own model? Our team carries out pre-submission model audits — an outside view, cell by cell, to the standards of the major funders (AFD, AfDB, World Bank). Request an audit →