The base case tells you what happens when the chosen assumptions behave. Stress testing asks what happens when they do not.
Stress testing means changing important assumptions to see how the model behaves.
Sensitivity analysis means showing the effect of one or two variables across a range.
Neither exercise predicts the future. They are hypothetical tests. A flight simulator does not promise an engine failure; it shows whether the crew knows what one would do.
Hit the controls that move the aircraft
Start with the variables that can change cash flow, debt survival, or sale value:
| Variable | Why it matters |
|---|---|
| Rent growth | Compounds through revenue |
| Vacancy and bad debt | Hits collections directly |
| Taxes and insurance | Can move fast and hard |
| Interest rate or debt service | Controls cash-flow pressure |
| Exit cap rate | Drives terminal value |
| Renovation cost and timing | Controls value-add execution |
Do not spend the first hour making office supplies 2% worse while terminal value is hauling the entire return uphill.
Use the rent roll and T-12 for revenue history, tax bills and insurance quotes for expenses, the lender quote for debt, contractor bids for renovations, and sales comps for the exit range. A sensitivity range without a reason is just a decorative slider.
Run the two-way simulator
Assume, hypothetically, that projected exit NOI is the key driver:
| 5.50% exit | 6.00% exit | 6.50% exit | |
|---|---|---|---|
| $900k NOI | $16.36M | $15.00M | $13.85M |
| $1.0M NOI | $18.18M | $16.67M | $15.38M |
| $1.1M NOI | $20.00M | $18.33M | $16.92M |
Read the rows first: that is the effect of the exit cap changing while NOI stays fixed. Read the columns next: that is the effect of NOI changing while the exit cap stays fixed.
If value swings hard between adjacent cells, you found a control surface with teeth. The output is not guaranteed at any point in the table; the table shows how much the conclusion depends on each assumed input.
Make the misses travel together
Single-variable tests isolate cause. Real properties prefer simultaneous inconveniences.
Build one hypothetical downside case:
| Assumption | Base | Downside |
|---|---|---|
| Rent growth | 4.0% | 2.0% |
| Bad debt | 1.5% | 3.5% |
| Insurance growth | 6.0% | 18.0% |
| Exit cap | 5.50% | 6.25% |
| Renovation cost | $9,000/unit | $10,500/unit |
Then make the model answer practical questions:
- Does cash flow cover debt service and required reserves?
- When does available cash get tight?
- Which renovation work gets delayed or reduced?
- Does the loan mature before the plan stabilizes?
- Who must supply cash if two misses overlap?
One warning light is analysis. Five warning lights are a meeting.
Rank the damage
Do not stop at “returns went down.” Name what caused the damage:
| Rank | Assumption | Damage if wrong |
|---|---|---|
| 1 | Exit cap | High |
| 2 | Stabilized NOI | High |
| 3 | Debt service | Medium/high |
| 4 | Renovation cost | Medium |
| 5 | Lease-up speed | Medium |
Your ranking may differ by deal. That is the point. If nobody can identify the top sensitivity and trace it to a source, nobody understands which part of the plan is flying closest to its limit.
Debrief the failure
Run the downside before the investment committee summary, investor memo, or family conversation. Save the case. Write down the first covenant, reserve, timeline, or return measure that fails and the assumption that caused it.
The pretty case will still be available. The simulator is where you learn whether the base case survives contact with its own limits.
PR Steinfurth Equity provides educational information only. Nothing on this website is an offer to sell or a solicitation of an offer to buy any security, nor investment, legal, or tax advice. Any securities offering is made only to qualified investors through official offering documents. Real estate investments involve risk, including possible loss of principal. Past performance is not indicative of future results.