How to Underwrite a Real Estate Deal: A Practical Developer's Guide
Underwriting is where deals live or die -- here's how institutional development teams stress-test a project before committing capital.
Underwriting is not about predicting the future with precision. It's about understanding where your assumptions are weak, what returns look like under stress, and whether the risk-adjusted outcome justifies the capital.
For institutional developers and their investment partners, underwriting is a structured process with distinct phases. Miss one, and you're not underwriting -- you're guessing.
What Underwriting Is (and Isn't)
Underwriting is financial feasibility analysis. It answers one question: at these assumptions, does this deal pencil?
It is not valuation -- that's the appraiser's job. It is not market research -- that's an input to the underwrite, not the output.
A development underwrite differs from an acquisition underwrite. Development underwrites model costs that haven't occurred yet -- land, hard costs, soft costs, carry -- against revenues that won't be realized for years. Acquisition underwrites work from existing income and a purchase price. Both require the same rigor. Development underwrites carry more uncertainty.
Step 1: Revenue Assumptions
Start with market rent. What are comparable stabilized buildings achieving per square foot or per unit? What's the vacancy rate in the submarket? How is absorption tracking against new supply coming online?
These inputs come from market research: broker surveys, lease comps, supply pipeline analysis. AI can accelerate this -- pulling rent comps from multiple data sources, modeling absorption curves against supply -- but the human judgment call on micro-location and product quality is still the deciding input.
For development, you're underwriting rents that won't be achieved for two to four years. Apply a conservative discount to current rents unless the market is structurally undersupplied. Assume some lease-up risk.
Step 2: Cost Structure
Development costs fall into four buckets:
Land -- acquisition price, closing costs, carry through the entitlement period
Hard costs -- construction, site work, infrastructure, contingency (typically 5-10% of hard costs for budget purposes)
Soft costs -- architecture, engineering, permits, legal, financing fees, developer fee
Carry costs -- interest during construction, operating costs during lease-up before stabilization
Hard costs are the largest variable and the most volatile. Ground-up industrial runs $80-$150 per square foot depending on specification and market. Data centers run $10-$20 million per megawatt for shell and core before tenant fit-out. Multifamily high-rise in a Tier 1 market can exceed $500 per square foot all-in.
Get a cost estimate from a real contractor, not a pro forma template. Underwriting with stale or generic cost numbers is one of the most common failure modes in development.
Step 3: Returns Analysis
The key metrics every institutional underwrite produces:
Yield on Cost. Stabilized net operating income divided by total development cost. Compare this to market cap rates for stabilized assets. If your yield on cost exceeds the prevailing stabilized cap rate by 150-200 basis points, you have a development spread worth pursuing. If the spread is tighter, you need to ask why you're taking development risk.
IRR (Internal Rate of Return). The discount rate at which net present value equals zero across the full investment period. Institutional return targets for development are typically 15-20% levered IRR, depending on asset class and risk profile.
Equity Multiple. Total equity returned divided by equity invested. A 2.0x equity multiple over a five-year hold means you doubled your money. Targets vary by vehicle and strategy; most institutional development mandates require 1.8x-2.5x.
Cash-on-Cash Return. Annual distributions divided by equity invested. More relevant for stabilized acquisitions than development projects, but worth modeling for the operating phase.
Step 4: Debt Sizing
Construction loans are typically 60-65% LTC (loan to cost), floating rate, with a term sized to cover construction and initial lease-up. Model the loan at current benchmark rates plus a spread -- and then stress it 100-200 basis points higher, because construction timelines run long.
Permanent financing -- a sale or refinance at stabilization -- is assumed at a market exit cap rate. That exit cap rate assumption is often where optimism lives and deals break.
Step 5: Sensitivity Analysis
No underwrite is complete without sensitivity testing. At minimum, stress:
Rents down 10%, down 20%
Hard costs up 10%, up 15%
Construction timeline extended 6 months, 12 months
Exit cap rates up 25 bps, up 50 bps
The purpose is not to find a scenario where the deal fails. It's to understand which variables are actually driving returns and where the deal breaks. A project where a 10% rent reduction kills the IRR is not a project you take to investment committee.
The most informative sensitivity table shows which assumptions matter most. If the deal only works with rents at the top of your comp range and costs at the bottom of your estimate, you don't have a deal -- you have an aspiration.
Where AI Changes the Workflow
Assembling the inputs for an underwrite has always been time-consuming: pulling rent comps, gathering cost benchmarks, sourcing cap rate data, confirming debt pricing. AI-native tools can compress that research phase from days to hours.
More powerful is AI's ability to run programmatic sensitivity analysis -- varying multiple inputs simultaneously and surfacing the combinations that produce unacceptable outcomes. That analysis used to require a full day in Excel and still got done manually. Now it takes minutes, and the output is more comprehensive.
The human judgment -- what rent assumption is defensible, what exit cap rate is credible for this submarket in three years, whether the contractor's cost estimate is realistic -- still sits with the developer.
The Discipline Behind the Numbers
An underwrite is only as good as the assumptions behind it. The most common failure mode is optimistic revenue assumptions combined with understated costs and compressed timelines. The second most common is a single-scenario model that never gets stress-tested.
The teams that consistently make money in development are not the ones with the most sophisticated models. They're the ones who know which assumptions matter most and are honest with themselves about where the uncertainty lives.
Get those right, and the model is just arithmetic.