A pro forma built on one cost number, one interest rate and one absorption pace is not a financial model — it is a single scenario dressed up to look like one. Sensitivity testing is the discipline of finding out, before capital is committed, which of those single numbers actually decides whether the deal works.
Key takeaways
A pro forma answers one question: does the deal work. Sensitivity testing answers a more useful one: how far does reality have to move from the assumptions before it stops working. The gap between those two questions is exactly the gap a lender's credit committee and an equity partner's investment committee are going to probe, so building the sensitivity table before they ask for it is cheaper than building it under pressure after they do.
It is tempting to sensitize whatever the spreadsheet makes easy to change — a soft-cost percentage, a marketing budget line — because the formula already references a single cell. That is usually not where the risk lives. The variables that actually decide most development pro formas are hard construction cost, the interest rate on the construction facility, the pace and price of absorption or lease-up, and the exit cap rate if the plan is to sell or refinance on completion. A sensitivity exercise that spends its time on soft-cost contingency while leaving hard cost and absorption pace untouched has tested the wrong thing well.
Statistics Canada's own release data makes the case for why cost and activity assumptions deserve the scrutiny. The June 2026 building permits release reported the national total rebounding by $2.3 billion, or 18.5%, to $14.9 billion, after two consecutive monthly declines in April and May. A market that moves that much month to month at the national level is not a market where a pro forma's single-point cost and timing assumptions can be treated as safe defaults — they are a snapshot of one moment in a series that has already shown it can swing double digits in either direction within a single quarter.
A rate shock is usually modelled as though the full loan amount is drawn and accruing interest from day one, which overstates the impact on a construction facility. A construction mortgage advances funds in staged draws as milestones are reached, and interest accrues only on amounts already advanced, not on the full committed facility. That means a rate increase in month 3, while only 15% of the facility has been drawn, costs far less in absolute dollars than the same rate increase applied against a fully-drawn balance in month 14. A sensitivity table that applies one flat rate shock across the whole construction period, rather than against the actual projected draw curve, will overstate early-stage interest risk and understate late-stage risk in roughly equal measure — which nets out to a number that happens to look plausible while being wrong on the shape.
The other financing variable worth isolating separately is the contingency draw itself. A cost overrun in month 4 that gets funded through an accelerated draw doesn't just cost the overrun amount — it also pulls forward the interest-accrual curve on everything drawn after it, compounding a cost problem into a financing-cost problem. Modelling the two as one line item hides which lever actually moved the return.
Cost escalation risk isn't only a materials question. Canada's construction sector employs more than 1.6 million people, and the industry's own forecasting has been flagging sustained hiring pressure and workforce renewal challenges as retirements continue through the coming decade. A pro forma that models a flat hard-cost escalation percentage without a labour-tightness variable behind it is really modelling one cause of cost overrun — materials pricing — while leaving out another that's shown up repeatedly in recent Canadian project histories: a trade shortage in a specific region pushing wage rates, and therefore the schedule, past what the base case assumed. A sensitivity table with a labour-tightness scenario, flexed independently of materials cost, catches a failure mode a materials-only stress test misses entirely.
A common shortcut is to apply a single haircut to projected sale prices or lease-up rents and call that the downside case. That treats price and pace as the same risk, when they usually aren't. A unit that sells for 5% less than projected but sells on schedule is a very different outcome from a unit that sells at full price but takes twice as long to move — the second scenario carries extra carrying cost, extra interest accrual, and in a phased project, a knock-on delay to the next phase's mobilization. A sensitivity table that varies price and pace independently, rather than bundling them into one "downside" column, shows which one actually drives the deal's return more.
Worked example — recomputing, not eyeballing, a stress case
Base case: $10.0M hard cost, average draw balance $6.0M over an 18-month construction period at a 6.0% construction rate, giving roughly $6.0M × 6.0% × 1.5 years = $540,000 in construction-period interest.
Stress case: hard cost rises 8% to $10.8M, and the rate rises 150 basis points to 7.5%. If the draw curve's shape stays the same in proportion, the average draw balance scales with cost to roughly $6.48M. Interest becomes $6.48M × 7.5% × 1.5 years = $729,000 — a $189,000 increase, which is 35% higher financing cost against an 8% cost increase and a 150-bps rate move, not a simple additive sum of the two shocks.
That non-linearity is the entire point of recomputing rather than eyeballing: a pro forma that just adds "8% cost overrun" and "1.5% rate increase" as two separate line-item deductions from the base case's profit will materially understate how the two compound through the draw balance.
Related reading: the inputs a feasibility study needs and the draw schedule a lender will fund.
Fewer, done properly, beats more done carelessly. A base case plus one downside and one upside on each of the two or three variables that actually move the return — typically hard cost, financing cost and absorption pace — covers most of what a lender or equity partner will ask about.
Two. A price cut with on-schedule sales and a full-price sale that takes twice as long carry different carrying-cost and financing consequences, and bundling them into one "downside" haircut hides which one is actually driving the return.
The drawn balance. Construction-period interest accrues only on funds actually advanced, so a rate stress test applied against the average projected draw balance, following the real draw curve, is materially different from one applied against the fully committed facility from day one.
Applying shocks as simple additive deductions from the base case rather than recomputing the model with the new inputs flowing through every downstream formula — cost and financing shocks compound rather than add, and an additive stress case understates the real exposure.
A 30-minute call is enough to tell you whether AI pays for itself here.
