NovFind the assumption doing most of the work
A simple rental example shows how one changed input affects the remainder, before considering costs and adverse conditions together.
This article reflects the named expert’s practical perspective. See NovAsia’s editorial policy for how material is prepared and reviewed.
Which input would you be reluctant to change? That is often where I begin testing a property illustration. A result may look comfortable because one assumption is quietly doing most of the work.
Use an invented case: ten paid months at USD 900 produce USD 9,000. Deduct specified annual costs of USD 3,000 and USD 6,000 remains before tax, finance and anything not included. With eight paid months and the other assumptions held constant, receipts are USD 7,200 and the remainder is USD 4,200.
Two fewer paid months reduce that remainder by USD 1,800, or 30%. This is not a forecast for Cambodian occupancy. It is an explanation of the arithmetic behind the result.
First isolate, then combine
Changing one input makes the effect easier to understand. It is not the whole downside exercise. Rent, vacancy and expenses can change together, so a separate combined scenario should follow, using defensible assumptions rather than frightening numbers chosen for effect.
Costs need attention before that exercise. Some continue while the property is empty; some depend on use; others arise when a tenancy changes. Keeping every cost fixed may be useful for an initial demonstration but should not be mistaken for a finished operating forecast.
If the calculation becomes acceptable only after one convenient number is restored, that number's basis deserves scrutiny. Does evidence support it for this property and period? Or are we preserving it because the offer no longer looks attractive without it?
The useful outcome is a precise research question. The buyer knows which assumption needs stronger support. That is more actionable than labelling the whole presentation optimistic or conservative, and it does not require me to make the investment decision for them.
After the first single-variable test, I look for the input to which the outcome is most sensitive, not necessarily the largest number in the model. A high monthly rent may attract attention, yet a small change in it might matter less than losing two paid months. Sensitivity analysis helps direct due diligence toward the assumptions capable of changing the decision.
Sensitivity and likelihood are different ideas. A model can be highly sensitive to an event that has weak evidence of occurring, while a small cost increase may be quite plausible but have little effect on the result. I would not convert the sensitivity table into a risk ranking. It shows impact. The probability and evidence behind each change still have to be considered separately.
A useful extension is to identify the point at which the scenario no longer meets the buyer's purpose. That threshold might be a minimum cash remainder, a maximum period without receipts or a cap on additional funding. The question then becomes concrete: at what input does this property stop doing the job it was selected to do?
Combined scenarios need discipline too. Cutting the rent, increasing vacancy and inflating every expense at once can create a dramatic downside case that teaches very little. Inputs should change for a reason. Lower occupancy may reduce some variable costs; a tenant change may create a one-off expense; certain fixed commitments may continue regardless. A credible model reflects those relationships instead of simply making every line worse.
I also keep the original base case visible beside each test. If five inputs change simultaneously in every new version, the buyer cannot see what caused the difference. The value of scenario work is transparency. Each variation should tell us something about the mechanism, not merely produce another percentage.
The source of the baseline assumption matters as much as its sensitivity. A figure supported by an existing agreement is different from one taken from an asking price, and both are different from a number chosen because it produces the desired return. Testing a weak assumption does not strengthen it; the exercise tells us how much effort should go into verifying it.
Used this way, the model stops pretending to forecast the exact future. It becomes a map of dependence: which assumptions carry the result, how much room exists before the outcome changes materially, and what information is worth obtaining next. For a buyer, that is more useful than a single precise yield that conceals how easily it moves.