IRR Calculation and Interpretation in CRE Deals

IRR is the discount rate that makes the net present value of a deal's entire cash flow stream, every dollar in and every dollar out, equal to zero. That single number is why it dominates commercial real estate underwriting: it's the only common metric that accounts for the acquisition outlay, the hold-period cash flows, and the exit proceeds all at once, weighted for when each dollar actually showed up. Cap rate gives a snapshot of the market in one year. Cash-on-cash gives an income yield in one year. ROI tells you total profit but not when it landed, and CAGR smooths two endpoints together while ignoring everything in between. IRR is built to punish idle capital and reward money that arrives early, and that's why institutional investors, family offices, and private equity funds treat it as the baseline for judging a deal: not because it's the best number, but because it's the only one doing that particular job.

How IRR is actually calculated: the mechanics sponsors and brokers need to know

The formula looks clean on paper: NPV equals CF₀ plus CF₁ divided by (1+r), plus CF₂ divided by (1+r) squared, and so on through CFₙ, set equal to zero and solved for r. There's no clean algebraic way to isolate r once a deal has more than a couple of cash flow periods, so IRR gets solved through guesswork dressed up as math. Excel's =IRR() function, along with most financial calculators, uses an iterative numerical method: guess a rate, check how close NPV lands to zero, adjust, repeat.

Setting up the cash flows correctly matters more than the math itself. Year 0 is the equity investment, booked as a negative outflow. Years 1 through N are net cash flow after debt service. The final year folds that year's operating cash flow together with net sale proceeds. Take a $1.5 million flex building bought with $500,000 of equity, throwing off $30,000 a year after debt service, sold in Year 5 for net proceeds of $800,000. Run that through the formula and it lands at 14.9%. That number then gets stacked up against the investor's hurdle rate, the floor below which the deal simply doesn't clear.

One distinction trips up more models than it should. Standard IRR assumes cash flows land neatly at year-end, but real deals rarely cooperate that way. XIRR assigns actual calendar dates to each cash flow instead of forcing everything into annual buckets, and deals where timing is uneven across the calendar benefit from that precision rather than the standard annual-period assumption. Run the standard version on a deal like that and the output is technically calculated but doesn't reflect how the money actually moved.

Then there's levered versus unlevered, and this is where sponsor-investor conversations go sideways more often than they should. Levered IRR uses cash flows after debt service, and it's what shows up in nearly every syndication offering memorandum, because leverage boosts returns when a property's yield beats the cost of debt. Unlevered IRR strips financing out entirely and measures the asset on its own. In a healthy deal, levered IRR runs higher. But when debt costs more than the property yields, a condition sometimes labeled negative leverage, the relationship can flip, and levered IRR may come in lower than unlevered. Anyone reading a pro forma should confirm which version they're looking at before comparing it to anything else. Loan-to-value, rent growth assumptions, operating expenses, taxes, insurance, reserves, capex, management fees, exit cap rate, the debt service schedule: every one of those feeds the final number, and moving any single input moves the output.

What counts as a good IRR: benchmarks by strategy and asset class

There's no single number that works across strategies, and treating an 18% IRR as automatically better than a 9% one, without asking what sits underneath either figure, is the most common mistake investors make when comparing deals. A rough starting frame puts core or stabilized deals at 8% to 10%, value-add at 12% to 18%, and opportunistic or ground-up development north of 20%.

A leverage-adjusted frame tells a slightly different story. Unlevered IRR on a five-to-ten year hold typically runs 6% to 11%, with lower-risk, longer-hold deals sitting near the bottom of that band. Levered IRR over the same window spans roughly 7% to 20%, and where a deal lands in that range comes down to its risk profile, not to how skilled the sponsor claims to be.

Stabilized multifamily tends to underwrite in the 8% to 12% range, assuming standard leverage, a five-to-ten year hold, and an exit cap that isn't wishful thinking. Preferred return structures in institutional-quality syndications have settled, per Primior Group's analysis of the 2025-2026 market, at 7% to 9% for stabilized assets and 8% to 10% for ground-up development.

Every one of these ranges assumes something: leverage in the neighborhood of 25% to 30% down, a five-to-ten year hold, an exit cap rate that isn't fantasy. Change any of those inputs and the benchmark moves with it, which is the whole point of treating IRR as a range rather than a target.

A projected IRR north of 25% on a five-year value-add deal earns a harder look before it earns any trust, and it should only survive that look if the sponsor's track record shows completed deals that actually delivered near those numbers, not just modeled them. Most of the time, anything above 25% is the product of aggressive underwriting or an unusually short hold, and that's a reason to interrogate every assumption line by line rather than walk away outright. On the other end, a levered IRR below 8% sits at the low end of even stabilized multifamily benchmarks. At that point the deal has to justify tying up capital in an illiquid asset some other way: tax benefits, depreciation, or a specific portfolio fit, because the return alone isn't doing the work.

The three limitations that make IRR misleading when used alone

IRR has three structural blind spots, and all three get exploited, sometimes deliberately, in deal presentations.

First, IRR says nothing about scale. A 30% IRR over a two-year hold that returns $1.69 per dollar invested is a smaller wealth event than a 20% IRR over six years returning $2.60 per dollar. IRR measures speed, not size, and treating a high IRR as automatically better ignores how much capital actually grew.

Second, IRR rewards money that comes back early and punishes money that comes back late, even when both deals are equally profitable in total dollar terms. That bias makes IRR a weak standalone tool for long-hold, income-focused strategies, where most of the value shows up as appreciation at exit rather than early distributions.

Third, and more technical: when a project's cash flows flip sign more than once, say a large renovation outflow lands in Year 3 of an otherwise cash-flowing property, the IRR equation can produce more than one mathematically valid answer. Excel will still hand back a number, and that's the trap: the software doesn't flag that the answer isn't unique, so the number looks just as trustworthy as any other IRR on the page when it isn't. MIRR was built as a partial fix, swapping the implicit reinvestment assumption baked into standard IRR for an explicit finance rate and reinvestment rate. But MIRR carries its own trap: it can spit out a comfortable-looking figure, often landing near the cost of capital, built entirely from assumed rates that don't correspond to anything in the deal's actual capital stack. A constructed number is not a market-derived one, and treating it that way is the mistake.

The market is already adjusting for this. Per McKinsey's 2026 Global Private Markets Report, five-year rolling distributions as a share of AUM hit their lowest recorded level in 2025, roughly 6% in the first half of the year against a ten-year average near 14%. That slowdown in actual cash returned has pushed limited partners toward MOIC (multiple on invested capital) as the primary metric, with IRR now used to check the multiple rather than lead the conversation.

Equity multiple as IRR's necessary companion metric

Equity multiple is total cash returned divided by total equity invested. IRR measures speed. Equity multiple measures size. Reading only one is how sponsors get away with dressing up a mediocre deal, and no investor should sign off on a deal memo that shows just one of the two.

A 25% IRR sitting on a 1.3x equity multiple describes a quick return on a fairly small gain. It looks sharp on the cover page of an offering memorandum, but the actual wealth created is modest. A 15% IRR on a 2.0x multiple describes a longer path to returns, with meaningful absolute wealth creation over the hold. Two deals can even post the identical equity multiple, one over three years and one over ten, and land on wildly different IRRs, because the multiple tells you what you made while IRR tells you what that took in time.

Reading the two side by side surfaces problems that neither reveals alone. A strong going-in cap rate paired with a weak projected IRR usually means the model leans on aggressive rent growth or is counting on cap rate compression at exit to make the numbers work. A high IRR paired with a low equity multiple often signals a short hold where most of the return comes from debt paydown or a fast flip, not real wealth building. A strong equity multiple paired with a below-target IRR usually means capital sat in the deal longer than it should have, relative to the fund's reinvestment cycle.

Every offering memorandum and every investment committee presentation should show both numbers side by side. IRR alone is an incomplete picture, and equity multiple alone leaves out the cost of time. Cash-on-cash return has its place too, mainly for annual income planning and judging early-year performance, but it belongs as a supplementary panel, not a stand-in for either of the other two over a full hold period.

Why the exit cap rate is the assumption that controls IRR more than any other

In most CRE pro formas, somewhere between 60% and 80% of total return traces back to the assumed exit price. That makes the exit cap rate the single most consequential assumption in the whole model, more influential than rent growth, more influential than expense ratios, more influential than nearly anything else an underwriter types into a spreadsheet.

The sensitivity isn't subtle. A 25-basis-point bump in the exit cap alone, holding everything else constant, can knock 100 to 200 basis points off projected IRR, depending on leverage and hold length. Market convention usually sets the exit cap 25 to 75 basis points above the going-in cap. A property bought at a 5.5% going-in cap would typically carry an exit assumption in the 5.75% to 6.25% range, on the theory that cap rates drift upward as an asset ages through the hold.

The 2021 and 2022 vintage of deals is the hard lesson in what happens when that convention gets applied too narrowly. Sponsors underwriting during that stretch modeled exit cap sensitivity in a band of plus or minus 25 to 50 basis points, which looked conservative at the time. Then cap rates expanded 150 to 200 basis points through 2023 into 2025, while floating-rate debt modeled as static sat exposed as SOFR moved roughly 500 basis points. The models weren't wrong given the ranges they tested. The ranges themselves were too narrow for what actually happened, and the deals that got hurt worst were the ones that treated a quarter-point stress test as good enough.

The fix that's become standard is a cap-rate sensitivity table: IRR and equity multiple recalculated at exit caps ranging from 50 basis points below the base case to 100 basis points above it, in 25-basis-point steps. Alongside that table, sharper underwriters calculate a breakeven exit cap, the rate at which the deal returns exactly 100% of invested equity. If that breakeven sits comfortably below the underwriting assumption, the deal carries a margin of safety. If it sits above, the deal needs cap rate compression just to avoid losing money, and that's a much riskier bet than the headline IRR lets on.

How IRR appears in offering memoranda and what sponsors must disclose alongside it

IRR is the headline number in nearly every private CRE offering memorandum, and for good reason: it's the one figure that claims to capture the whole arc of a deal, from the first check to the last distribution. But that headline figure is a projection, not a result. It's built on assumptions about rent growth, exit cap rate, hold period, and debt structure, and every one of those assumptions is a place the number can be nudged in a favorable direction.

A sponsor projecting a 25% IRR on a ground-up development deal is making a claim that only holds up if that sponsor's prior completed projects actually delivered returns near that level, on a verified basis, not just on paper. The right response to a headline IRR isn't to accept it. It's to ask what's underneath it: what rent growth is baked in, what exit cap is assumed, and critically, whether the number on the page is levered or unlevered. Most sponsor decks show the levered figure, because it's the bigger one, and that's not automatically a problem as long as it's labeled clearly.

A separate, often underappreciated risk sits in the source documents themselves. Offering memoranda, rent rolls, T-12 statements, and operating statements frequently disagree with each other on the exact figures that feed the IRR model: vacancy rates, in-place rents, expense line items. When a discrepancy between two sources gets quietly resolved in favor of one document, and the conflicting number from the other gets tossed without anyone flagging it, that error flows straight into the return assumptions. The safer workflow surfaces those conflicts for an analyst to judge, rather than resolving them silently behind the scenes.

A well-built OM discloses more than the headline IRR: the equity multiple, the hold-period assumption, the exit cap rate assumption, a clear levered-versus-unlevered label, and a sensitivity table showing how the return moves under a stressed exit. Where a preferred return is part of the structure, that figure, running 7% to 9% for stabilized assets and 8% to 10% for ground-up development in the current market per Primior Group's analysis, should sit clearly subordinate to, and consistent with, the projected deal-level IRR. If the preferred return and the headline IRR don't line up logically, that's worth asking about directly.

Where manual IRR modeling breaks down at deal volume and how automation addresses it

CRE underwriting rarely falls apart because the people running the numbers don't know what they're doing. It falls apart because that expertise gets buried under low-value handling work: pulling figures out of T-12s, rent rolls, operating statements, appraisals, offering memoranda, and legacy loan files, then retyping them line by line into an internal model.

Every one of those handoffs is a place a number can get transcribed wrong, a discrepancy between two source documents can get missed, or a stale assumption can survive into a live model without anyone catching it. That risk doesn't scale in a straight line with deal volume. A shop underwriting five deals a month can catch errors through sheer attention to each one. A shop underwriting fifty can't rely on the same method, because review capacity per deal shrinks even as the cost of a bad IRR assumption stays exactly the same.

Automated data extraction and modeling tools built for this specific bottleneck don't replace the analyst's judgment about what makes a good deal. What they change is where the analyst's time goes. Instead of manually keying numbers from a rent roll into a template, the analyst reviews flagged discrepancies between source documents and makes the calls that actually require judgment. Solving for IRR, running sensitivity tables across exit cap scenarios, toggling between levered and unlevered views: all of that can run systematically once the underlying data is captured cleanly. That frees up the analyst for the part of the job no machine can do: deciding whether the sponsor's exit cap assumption is credible, whether the rent growth projection matches what's actually happening in the submarket, and whether the whole package holds together as a real investment thesis rather than a spreadsheet reverse-engineered to hit a target number.