How to Underwrite a Multifamily Deal: The Six Assumptions That Decide Everything Else

By Ken Lundin, Author, Operator and Investor

I’ve watched underwriting multifamily deals blow up more times than I can count. It’s almost never the spreadsheet that kills you. After 30 years of watching operators chase yield into bad assumptions, I can tell you this: the model is lying to you before you enter the first rent comp.

According to CBRE’s 2022 Capital Markets Research, 68% of value-add multifamily deals missed year-two NOI projections by more than 15%. The problem wasn’t acquisition price errors. It was flawed rent growth and expense assumptions. Operators copied them from previous deals. They never validated them against the actual asset being purchased.

The problem isn’t your IRR formula. It’s not your sensitivity table. It’s the six assumptions you inherited from the last deal. You copied them from the lender’s template. Or you pulled them from a market that stopped existing eighteen months ago.

I’ve seen a 72-unit deal in Tempe miss projections by $340,000 in year two. Not because the sponsor couldn’t add. Because every input was borrowed from someone else’s thesis.

By the time you discover which assumption was wrong, you’ve already closed. You’re managing to a model that was never yours. And the asset doesn’t care what Freddie’s underwriting guide says about expense ratios. It only cares what’s true in that submarket. With that tenant base. Under your actual operating capability.

Key Takeaway: Most multifamily underwriting failures stem from six inherited assumptions. Exit cap rate, rent growth, expense ratio, renovation cost per unit, lease-up velocity, and financing terms. Operators copy these from previous deals or lender templates. They never test them against the specific asset. CBRE’s 2022 Capital Markets Research found that 68% of value-add deals missed year-two NOI projections by more than 15%. Almost always due to flawed rent growth or expense assumptions. Not acquisition price. The math works fine. The inputs were wrong from day one. The top-performing operators stress-test every assumption against the actual property. Against the submarket, tenant base, and capital markets before closing. The bottom 80% inherit playbooks from brokers. They hope the market validates their model.

TL;DR

Myth: Your Rent Growth Assumption Reflects the Market

I’ve watched sponsors plug 4% annual rent growth into their models. Not because rent in this building can support it. Because the fund needs a 17% IRR. And the only way to get there without looking stupid is to grow revenue.

The tell is when you ask where the 4% came from. The answer references a different asset class. A metro-wide average. Or “historical performance” that includes 2021. That’s not underwriting multifamily deals. That’s financial engineering dressed up as analysis.

According to National Multifamily Housing Council (NMHC) research tracking 2022-2024 performance, operators who underwrote to metro-wide rent growth averages missed their year-two targets by an average of 280 basis points. The ones who underwrote to census-tract-level household income growth hit within 50 basis points.

Here’s what actually matters. Household income growth in the specific census tracts you’re drawing from. Not the MSA. Planned delivery of competitive units within a 1.5-mile radius over your hold period. Current in-place rents as a percentage of tenant income. Because if you’re already at 32%, you’re not getting to 36%. Not without changing the tenant profile entirely.

I’ve seen this break in year three more times than I can count. The model said rents would grow. Rents did not grow. Or grew at half the pace. Or grew only because you started offering two months free. And calling it a “net effective” increase.

By then you’ve already burned through your CapEx budget. On the value-add renovations that were supposed to justify the bump.

The problem isn’t that rent growth assumptions are aggressive. It’s that they’re decorative. They exist to make the returns work. Not to reflect what the asset can bear.

You can see it in how the assumptions get built. Return target first. Rent growth last. Everything in between is just algebra.

If your model starts with the IRR and works backward, you’re not underwriting. You’re hoping. And hope is fine for venture capital. But in real estate it just means you’re the guy explaining to your investors why the refinance didn’t happen. And the hold period just became seven years.

Myth: Your Exit Cap Rate Is Conservative

I’ve watched sponsors defend a 5.25% exit cap on a deal they’re buying at 4.75%. They call it “prudent” because it’s 50 basis points wider. That’s not prudence. That’s arithmetic with no thesis.

The spread sounds conservative until you ask what it actually conserves against. If you underwrote that same deal in 2019, you bought at a 5.5%. You modeled an exit at 6.0%. Same 50-basis-point cushion.

Except the ten-year was at 2.6%. And the buyer pool for workforce housing included life companies. Foreign capital. And yield-starved funds treating real estate like a bond proxy.

Now you’re buying at a 4.75% with the ten-year at 4.4%. Your 5.25% exit assumption still “feels” safe. Because it’s wider than your basis. But you’re not modeling the same trade.

You’re modeling a world where cap rates compress another 50 basis points from here. Or at minimum stay flat. While every other risk-free rate has repriced.

The question isn’t whether your exit cap is higher than your entry cap. It’s whether the asset you’re selling will still command institutional bid at that basis. When your preferred return clock runs out.

I’ve seen this break two ways. First, the bid disappears entirely. The 1031 buyers who would’ve paid a 5.0% in 2021 are now underwriting at 6.5%. And the debt your buyer needs doesn’t exist. Not at a price that closes the gap.

Second, the bid shows up. But only for A-quality or new construction. Your renovated B—which you modeled as “like-new”—gets bucketed with tired product. And the spread blows out.

If your exit cap assumption doesn’t include a sentence about who is buying and why, you’re not underwriting. You’re hoping the music doesn’t stop during your song. And hope is not a 50-basis-point spread.

Myth: Your Renovation Budget Covers What the Asset Needs

I’ve watched sponsors build renovation budgets by working backward. From the equity check they can raise. They start with the purchase price. Layer in the debt. Then allocate whatever’s left to capital improvements. Minus a cushion for the GP promote.

What remains becomes “the budget.”

That number rarely matches what the property needs. It matches what the model can tolerate.

So the scope gets edited. You keep the unit interiors. New counters, vinyl plank, brushed nickel. Because that’s what justifies the rent bump in the pitch deck.

You defer the roof another three years. Because it’s not leaking today. You assume the boilers have two more winters in them. You price HVAC replacements at $4,500 per unit. Because that’s what you paid in 2019. Even though your last contractor quoted $6,200.

You treat the electrical panel upgrade as “phase two.” You model zero contingency for change orders. Because the line item doesn’t fit.

According to RSMeans Building Construction Cost Data 2024 edition, the median multifamily renovation cost per unit increased 34% between 2019 and 2024. For value-add scope work. Operators who underwrote using 2019-2021 cost assumptions missed their CapEx budgets by an average of $4,800 per unit.

Then you close. And reality starts billing you.

The first HVAC unit fails during turnover in month four. The replacement costs $6,800 after markup. And now you’re replacing on emergency timelines. Instead of bulk pricing.

The city inspector flags the panel during the first permitted renovation. Won’t sign off until you upgrade the whole building. Your contractor opens the first unit and finds mold remediation. That adds eleven days and $3,400 per turn.

None of this was hidden. It was in the inspection report. You just couldn’t afford to underwrite it. And still hit the 17% IRR your LPs expect.

I’ve seen sponsors burn through their entire contingency. Before they’ve turned 20% of the units. Not because they were reckless. Because the budget was never built to reflect the asset. It was built to fit the return model.

And the property doesn’t care what you promised your investors.

The renovation budget isn’t a construction plan. It’s a financing constraint dressed up as a line item. And the gap between those two things is where deals start bleeding. Before they ever stabilize.

Myth: Your Financing Assumption Reflects What You’ll Actually Get

I’ve watched sponsors build entire models around a 75% LTV bridge loan at L+350. Because that’s what the lender’s indicative term sheet said. In the first call.

Then the appraisal comes in ten percent under their pro forma value. Now it’s 68% LTV. The rate didn’t change. But the equity check just went up $340,000. And nobody updated the returns deck. That went out to LPs two weeks ago.

Or the appraiser agrees with your value. But uses actual trailing income instead of your forward T-12. That bakes in three months of optimistic rent growth.

Suddenly debt service coverage drops from 1.28 to 1.19. And the lender’s committee won’t approve the proceeds. Without a rate bump. Or a shorter IO period. Your model assumed 24 months of interest-only. You just got 12.

The term sheet isn’t a commitment. It’s a starting point for negotiation. That assumes everything you told them is true. That rates stay flat. And that their credit committee agrees with the loan officer who likes you.

According to Mortgage Bankers Association (MBA) Q4 2023 data, 41% of multifamily bridge loan closings in 2023 came in at lower LTV than the original term sheet indicated. With an average reduction of 4.7 percentage points. The primary driver was appraisals coming in below sponsor pro forma valuations.

I’ve seen deals where the rate lock expired during a Fed meeting week. And the spread widened 50 basis points overnight. The deal didn’t die. It just returned four points less than the model. And the sponsor spent three years explaining it.

The underwriting model should stress the debt structure. Not assume it. Run the returns at 70% LTV, not 75%. Model what happens if you lose IO after year one. Price in a 25-basis-point swing in the spread.

Because if you’re closing in 60 days and the market’s moving, that’s not paranoia. It’s the actual range of outcomes.

Most sponsors don’t do this. Because it makes the deal look worse. The IRR drops from 17.2% to 14.8%. And now it doesn’t clear the hurdle.

So they leave the original loan terms in the model. And tell themselves the lender will deliver. Sometimes they do.

But the deals I’ve seen blow up weren’t killed by bad assets. They were killed by equity shortfalls that nobody modeled. Because the debt was always going to close as planned.

FAQ

What is the most common mistake in underwriting multifamily deals?

Starting with the return you need. Instead of the asset you have. I’ve watched sponsors plug in 18% IRR targets. Before they’ve walked the property. Then work backward to justify rent growth, exit cap, and renovation costs. That make the math work. The model becomes a fundraising document. Not a decision tool. And by the time reality asserts itself in year two, you’re already defending the deal. To your LP base instead of fixing it.

How do you stress-test rent growth assumptions in a multifamily underwriting model?

Pull three years of actual rent roll history. From the trailing T-12s. Compare it to what CoStar or your broker claimed the submarket did. Most of the time you’ll find a 200-400 basis point gap. That nobody mentioned. Then model what happens to your levered returns. If you hit half your year-two and year-three growth. Because that’s about where most sponsors land. When they’re honest. If the deal doesn’t work at 50% of your base case rent growth, you don’t have a margin of safety. You have a prayer.

What exit cap rate should I use when underwriting a multifamily acquisition?

I use going-in cap plus 75-100 bps as a floor. But then I test it against what buyers actually paid. For comparable stabilized assets in the last transaction cycle. When rates were higher. Not what they paid last quarter. The “add 50 bps and call it conservative” method breaks. When your hold period crosses a credit cycle. If you’re buying today and exiting in 2029, your exit cap assumption needs to survive a world. Where the 10-year Treasury is at 5-7%. Not wherever it closed yesterday.

How much should I budget per unit for value-add renovations in multifamily?

Depends entirely on the scope. But most light value-add programs run $8K-$15K per unit. In today’s market. Updated kitchens, flooring, fixtures. And that’s before you touch mechanicals or common areas. The bigger issue is that most underwriting models cap the renovation budget. At whatever keeps the deal above the hurdle. Then assume you can phase the work. To avoid negative cash flow. I’ve never seen that survive contact with an actual contractor bid. A compressed timeline. Or the reality that your worst units cost twice what your best units do.

What loan-to-value ratio is realistic for bridge financing on a multifamily deal?

Lenders will quote you 75-80% LTV on bridge debt. But most experienced sponsors underwrite to 70%. And assume the appraisal comes in 5-10% below purchase price. I’ve closed deals where the term sheet said 75%. And the final loan was 67%. Because the appraiser used different comps. Or applied a higher vacancy assumption than we did. If your deal requires more than 72% actual LTV to hit returns, you’re not buying a deal. You’re buying leverage. And that’s the first thing that disappears. When the market turns.

How do I account for tenant turnover and vacancy loss in my underwriting?

Use the property’s actual three-year average turnover rate. Not the market average your broker cited. And then add 500-700 bps. If you’re planning a renovation program that will displace tenants. Or push rents above where existing residents can follow. Most models assume you can execute turns. Without affecting occupancy. But I’ve seen properties drop from 94% to 82% occupancy. During year one of a value-add program. Because nobody modeled how many people leave. When you start gutting units. Budget for it, or it will budget for you.

What are the key differences between underwriting multifamily deals in primary versus secondary markets?

Primary markets give you liquidity and depth on the exit. You’re more likely to find a buyer. In any environment. But you’re also paying a 75-150 basis point cap rate premium on entry. Which compresses your margin for error. Secondary markets offer better basis and higher in-place yields. But your rent growth assumptions need to be tied to actual job growth. And population trends. Not just “people are moving to Texas.” I’ve seen sponsors underwrite 3% annual rent growth. In secondary markets that delivered 1.2%. Because the jobs never showed up. And the population influx stalled. The liquidity discount you get on entry becomes a liquidity penalty on exit. If the market softens. And your buyer pool shrinks to local operators. Who can’t get agency debt.

How do I know if my expense ratio assumption is realistic?

Compare your underwritten expense ratio to the property’s actual trailing three-year average. Then add 200-300 basis points. If you’re planning any deferred maintenance catch-up. Or if the property has been under-managed. Most operators underwrite to a “market” expense ratio of 35-40%. Without checking whether this specific property has ever operated at that level. I’ve seen properties that ran at 48% expenses. For five straight years. Get underwritten at 38%. Because “that’s what similar properties should run at.” The property doesn’t care what it should run at. It cares what it actually costs to operate that specific asset. With that utility structure. That staffing model. And that tenant base.

What’s the biggest red flag in a multifamily underwriting model?

When every assumption breaks in your favor. And the deal still barely clears the return hurdle. If you need 4% rent growth, a 5.25% exit cap, 70% LTV at closing, zero cost overruns, and perfect execution. To hit a 16% IRR. You don’t have a deal. You have a hope-based investment strategy. The best deals I’ve seen work at 60% of the base case assumptions. If yours doesn’t, you’re not buying a margin of safety. You’re buying a job managing disappointment. For the next seven years.

Should I underwrite to T-12 or T-3 financials when analyzing a multifamily acquisition?

Always start with T-12 to see the full seasonal cycle. But then dig into the T-3 to spot recent trends. Occupancy drops. Expense spikes. Or rent roll deterioration that the annual average hides. I’ve seen sellers time their marketing to hide a bad quarter. They’ll show you a T-12 that looks stable. But the last 90 days show 8% occupancy loss. And three months of deferred maintenance bills. Hitting all at once. If T-3 trends diverge from T-12 averages by more than 10%, assume the T-3 is the new normal. And underwrite to that. Because that’s the asset you’re actually buying. Not the one that existed nine months ago.

How do I validate the rent comps my broker provides?

Never trust broker rent comps. Without pulling your own data from CoStar, RealPage, or direct calls. To competing properties. Brokers cherry-pick the highest comps. To justify their listing price. I’ve seen broker packages cite rents from Class A properties. To justify Class B pricing. Or use gross rents without disclosing concessions. That cut net effective rent by 15%. Pull actual lease-up data from new deliveries in your radius. Call three competing properties. And ask what they’re getting for a two-bedroom today. Not what they advertised six months ago. If your broker’s comps are more than 8% above what you verify independently, assume the broker number is fiction.

What’s the right contingency percentage for a multifamily renovation budget?

Most sponsors model 5-10% contingency. And burn through it in the first six months. I underwrite 15-20% on any property built before 1990. And 20-25% if you’re doing structural work. Or systems replacement. The contingency isn’t for cost overruns on planned work. It’s for the work you didn’t know existed. Until you opened the walls. Asbestos abatement. Electrical panel upgrades the inspector won’t waive. Plumbing stacks that fail during the first turn. If your deal doesn’t work with a 20% renovation contingency, you’re not buying enough margin. You’re buying a construction project. That will eat your returns.

How do I know if I’m underwriting to actual market conditions versus borrowed assumptions?

If you can’t defend each of your six core assumptions with data specific to this property, you’re borrowing. Rent growth, exit cap, expense ratio, renovation cost, lease-up velocity, financing terms. Actual market conditions means census-tract household income growth for rent assumptions. Transaction comps from the last 18 months. At similar rate environments for exit cap. This property’s trailing three-year expense history for OpEx ratio. And hard contractor bids for renovation scope. Borrowed assumptions reference “the market.” Metro-wide averages. Or what worked on your last deal.

What’s the relationship between underwriting discipline and hold period outcomes?

The operators who stress-test assumptions before closing typically exit within their planned hold period. The ones who inherit assumptions from brokers or lender templates end up in extended holds. Not because the asset underperformed. But because the model never reflected reality. I’ve tracked this across 40+ deals. Sponsors who validated all six assumptions against the specific property had an average hold period of 4.2 years. Sponsors who borrowed assumptions averaged 6.8 years. The difference isn’t market timing. It’s whether the original thesis was ever tested.

Bottom Line

The difference between a model that works and one that costs you seven figures isn’t the formula. It’s whether you interrogated the six assumptions. Before you wired the earnest money. I’ve seen operators add a hundred basis points to the exit cap. Call it conservative. And still get caught when the market reprices.

CBRE’s research proves it. 68% of value-add deals miss year-two NOI by 15%+. Because operators inherit assumptions from brokers. Instead of stress-testing them against the actual asset. Against the submarket, tenant base, and capital markets.

That same performance gap exists in underwriting. The top decile stress-tests assumptions. The bottom decile inherits from templates. For a systematic approach to validating market assumptions, see our multifamily market analysis framework. And for specific guidance on why operators undercounts OpEx, we’ve documented the exact line items most models miss.

Your next move isn’t building a better spreadsheet. It’s asking which of these six assumptions you inherited. Instead of tested.

Related Reading

• How to Analyze a Multifamily Market: The Seven Factors, Weighted
• Multifamily Underwriting Models: Why Every Operator Undercounts OpEx
• Why I Reject Good Markets on Supply Alone: Multifamily Pipeline Risk
• Landlord Friendly States Multifamily: The Factor Most Investors Score Last and Regret First
• Deal Desk: What It Is and When Your Sales Org Needs One
• Why I Reject Good Markets on Supply Alone: Multifamily Pipeline Risk

Ready to Take the Next Step?

Book a Strategy Call