A base-case pro forma is one possible future. The actual future will be different — sometimes a little, sometimes a lot. Sensitivity analysis is the discipline of figuring out HOW different, in WHICH variables, and what that means for the deal's returns.

We touched on sensitivity in Course 5, Lesson 6. This lesson goes deeper: not just how to build sensitivity tables, but how to interpret them and use them to make better decisions.

The core question

Sensitivity analysis answers a single question: how confident am I in the base case, and how much can I afford to be wrong?

A deal with a base-case 15% IRR isn't useful information by itself. The same 15% IRR can mean:

"Returns are 14-16% under almost any reasonable assumption" → robust, fund it
"Returns are 8-22% across the realistic range" → fragile, depends on getting lucky
"Returns swing from −5% to +35% across the realistic range" → speculative, treat accordingly

Same headline number. Three completely different deals.

Sensitivity analysis is what reveals the difference. Without it, you're just looking at one point estimate.

Single-variable sensitivity (the basic form)

The simplest sensitivity analysis sweeps one input across a range of values and shows how the output changes. We covered the Excel mechanics (Data Tables) in Course 5.

The key inputs to sensitize:

Exit cap rate — almost always the biggest single driver
Year-1 NOI — both rent and expenses
Rent growth rate
Vacancy rate
Operating expense growth
Interest rate (if rate sensitivity is meaningful)
Hold period

For each, sweep ±15-25% around your base case and see how IRR moves. Record the extremes.

A useful chart format

Variable swept    Low value    Base    High value    IRR range
Exit cap rate       5.00%      6.00%      7.00%       4.5%-12.8%   (8.3% range)
Y1 NOI            -10%        Base       +10%         9.5%-13.5%   (4.0% range)
Rent growth        1.5%       3.0%       4.5%         8.5%-13.0%   (4.5% range)
Vacancy            8.5%       6.5%       4.5%         9.0%-12.5%   (3.5% range)
OpEx growth        4.5%       3.0%       1.5%        10.0%-12.0%   (2.0% range)
Interest rate      8.0%       7.0%       6.0%         9.0%-12.5%   (3.5% range)

The "IRR range" column tells you which inputs matter most. In this example:

Exit cap rate drives an 8.3% point swing → biggest driver
Rent growth, vacancy, and interest rate are each 3-5% point drivers → secondary
Year-1 NOI matters but less than expected
OpEx growth matters least

This ranking is immediately useful: it tells you where to spend your due diligence energy. If exit cap is the dominant driver, you should be obsessed with the exit cap assumption — pull recent comps, talk to brokers about future trends, run multiple exit scenarios. Don't waste time arguing about whether OpEx growth is 2.8% or 3.2%.

Multi-variable sensitivity (the matrix)

A two-variable matrix captures interactions between inputs. The classic combinations:

Exit cap × rent growth

Most common matrix. Shows how returns change as the two biggest drivers move together.

                    Rent growth
Exit cap     1%      2%      3%      4%      5%
   5.00%   11.5%   13.0%   14.6%   16.2%   17.8%
   5.50%   10.0%   11.5%   13.1%   14.7%   16.3%
   6.00%    8.6%   10.1%   11.7%   13.3%   14.9%
   6.50%    7.2%    8.7%   10.3%   11.9%   13.5%
   7.00%    5.9%    7.4%    9.0%   10.6%   12.2%

Reading this: base case is 6% exit cap, 3% growth → 11.7% IRR. Realistic downside (6.5% exit, 2% growth) → 8.7%. Realistic upside (5.5% exit, 4% growth) → 14.7%. Range of 8.7% to 14.7% — that's the deal's actual return profile.

Color-code with conditional formatting (red for <5%, yellow for 5-10%, green for 10-15%, dark green for 15%+) and the matrix tells the story at a glance.

Other useful matrices

Interest rate × LTV — debt structure scenarios
Vacancy × rent growth — operational scenarios
Year-1 NOI × Exit cap — entry vs. exit pricing
Construction cost × lease-up time — for development deals
Hold period × exit cap — when to exit

Pick the two variables most relevant to the specific deal you're underwriting. Don't run all of them by default — you'll drown in data.

Scenario analysis (named cases)

Sensitivity is mechanical — sweep inputs across ranges. Scenario analysis is narrative — define a few specific stories and run the model for each.

The standard three scenarios:

Base case

Your best guess. Most-likely values for every input. This is what you'd present to a partner if asked "what do you expect?"

Upside case (P25 — better than expected)

What happens if things go better than you expect, but not unrealistically so:

Rent growth at the higher end of your range
Vacancy lower
Exit cap rate compresses (favorable)
OpEx growth modest

This isn't "what if everything goes perfectly?" It's "what if the market cooperates and execution is solid?"

Downside case (P75 — worse than expected)

What happens if things go against you, but not catastrophically:

Rent growth at the lower end of your range
Vacancy higher
Exit cap rate expands (unfavorable)
OpEx growth elevated (insurance, taxes)

This isn't "what if the world ends?" It's "what if the market is mediocre and we hit normal headwinds?"

Each scenario produces a complete pro forma and outputs (IRR, equity multiple, NPV, DSCR). Present them together:

Metric               Downside    Base    Upside
Levered IRR             6.2%   12.5%    18.0%
Equity multiple         1.30x  1.80x    2.30x
NPV at 10%            -$85K   $145K    $410K
Year-1 DSCR             1.05    1.20     1.35
Probability (subjective) 30%    50%      20%

The probability column is subjective but useful — it forces you to think about likelihood, not just possibility.

Probability-weighted IRR

Multiply each scenario's IRR by its probability and sum:

0.30 × 6.2% + 0.50 × 12.5% + 0.20 × 18.0%
= 1.86% + 6.25% + 3.60%
= 11.71%

The probability-weighted IRR is closer to your true expected return than the single base case. In this example, the base case (12.5%) overstates the probability-weighted return (11.7%) by ~1 point. Multiply that across many deals over a career and the difference is enormous.

A fourth scenario: stress / disaster

Beyond the standard three, run a fourth scenario for serious deals: the stress case. This isn't "downside" — it's "what if multiple things go wrong simultaneously?"

Example stress assumptions:

Rent growth zero or negative
Vacancy 200 bps higher than base
Exit cap rate 100 bps higher than base
OpEx growth at insurance-driven 6%
Interest rate at refinance is 100 bps higher
6 months of additional re-leasing at sale

If the deal still survives the stress case (positive equity, no DSCR breach, IRR above zero), it's a robust deal. If the stress case wipes out equity, you have a fragile deal that depends on multiple things going right.

The stress case isn't your base case. It's a "catastrophe insurance" check — making sure you can survive if you're wrong on multiple dimensions at once.

Sensitivity reveals your assumptions

A useful side effect of sensitivity analysis: it forces you to be honest about your assumptions.

Try this exercise. For your base-case pro forma, ask yourself for each input:

What number am I using?
Why? (What's the source — appraisal? broker? guess?)
What's the range of plausible values?
Where in that range did I land — toward optimistic, neutral, or conservative?

Most beginners realize when they do this honestly that 60-80% of their inputs are toward the optimistic end of plausible. That's not maliciousness — it's the natural human bias toward outcomes you want to be true.

Sensitivity analysis catches this bias by showing what happens if your "neutral" inputs turn out to be slightly less than you guessed.

When sensitivity is misleading

A few caveats:

1. Correlated variables

Sensitivity treats each variable independently. In reality, many inputs are correlated:

High vacancy and slow rent growth tend to happen together
Rising interest rates and rising cap rates tend to happen together
Insurance increases and rising property taxes can correlate

If you run a multi-variable sensitivity that treats these as independent, you'll understate the true risk. Build in some correlation when constructing scenarios.

Sensitivity analysis answers: how confident am I in the base case?
Sweep each key input ±15-25% around base case to find the biggest drivers
Multi-variable matrices show interactions between two variables
Scenario analysis defines named cases (base, upside, downside) with full pro formas
Probability-weighted IRR is closer to true expected return than the single base case
Always run a stress case for downside protection
Watch for correlated inputs and tail events that escape simple sensitivity

Next lesson: building the three-scenario pro forma — turning sensitivity discipline into a structured comparison you can present to a partner or lender.

Sensitivity and Scenario Analysis