Going-In vs Exit Cap Rate: The Spread, the Reversion, and the 2026 Underwriting Discipline

Key Takeaways

Why the Exit Cap Decides Returns

In a 5–10 year hold underwrite, the exit cap rate is the input that moves levered IRR the most per basis point of change. Rent growth compounds across the hold and matters; expense growth runs against it and matters; debt cost determines the leverage uplift. None of them moves IRR as much per basis point as the exit cap. In the model used in this article, the reversion is about 65–77% of undiscounted unlevered cash inflows over a 5–10 year hold, and the terminal value is roughlyYear 11 forward NOI ÷ exit cap. The denominator is a single number. Move it 25 bps, and you move about 4% of unlevered exit value, which, after 60% LTV leverage, translates to roughly 100–170 bps of levered IRR on a 5–7 year hold.

This article models the going-in to exit cap spread as a function of three drivers (hold period length, cyclical positioning, and structural fundamentals at the projected exit). It shows what the 25–50 bps convention is doing, when it's right, and when it's wrong; it works through what 25 and 50 bps of exit cap expansion cost at 3, 5, 7, and 10 year holds; it spells out the Year 11 forward NOI convention; and it gives the IC defense template that turns “going-in plus 50” from a default into a deliberate underwriting assumption.

The article sits inside the valuation cluster.The pillar covers what a cap rate is and how it's computed.The decomposition article covers how to build a cap rate from risk-free rate, risk premium, and growth components. This article covers what happens to that cap rate across the hold and how to underwrite the exit. Theterminal value sibling goes deeper on the hold-period choice itself; this article focuses on the spread.

The Going-In Cap

The going-in cap rate is the cap rate at acquisition: Year 1 forward NOI divided by purchase price. In broker pitchbooks it's sometimes called the entry cap rate, and some brokers quote it on trailing NOI; this article uses Year 1 forward NOI so that the going-in and exit caps are defined the same way. The going-in cap is the observable price of the deal: what the market is asking you to pay for current cash flow.

Three sources of truth converge on the going-in cap in institutional practice. The first is the broker quote: the whisper number in the BOV. The second is the comp set: recent transactions of comparable assets in the same submarket, published by data providers such as MSCI Real Capital Analytics and tracked by the broker's capital markets desk. The third is the build-up: the institutional decomposition into risk-free rate plus risk premium less long-run growth, walked through in thecap rate decomposition article. In a normal market the three should land close together. When they disagree, the underwriter has to attribute the disagreement to a specific component: growth view, sponsor premium, structural premium, or a stale comp set.

The going-in cap ranges used as illustrative assumptions in this article's examples (a May 2026 calibration, not survey data):

• Multifamily core (Sun Belt, garden, stabilized): 5.50–6.25%
• Industrial core (Class A logistics): 4.75–5.50%
• Anchored retail (grocery-anchored, top-quartile): 6.25–7.00%
• CBD office (Class A trophy): 6.50–7.50%
• Suburban office (commodity): 8.50–10.50%
• Self-storage (institutional core): 5.50–6.25%
• Full-service hospitality (urban): 7.50–9.00%

These are working assumptions for the worked examples, so check them against a current comp set before using them on a live deal. The macro backdrop they sit against is documented: the 10-year Treasury averaged 1.45% in 2021, 4.21% in 2024, and stood at 4.56% on May 22, 2026 (FRED DGS10). The institutional reset is real: cap rates have not returned to their 2021 lows, and the underwriting discipline in the rest of this article is built around that gap.

The Exit Cap

The exit cap rate is the cap rate assumed in the underwriting at sale, the denominator in the terminal value calculation. Appraisers call it the terminal cap rate (also written reversion cap rate); broker pitchbooks call it the exit cap. All of these terms refer to the same input: the cap rate the underwriter assumes the asset will trade at when sold.

Three things distinguish the exit cap from the going-in cap. First, the exit cap is not observable: nobody quotes it; the underwriter chooses it. Second, the exit cap should be built up fromexit-date components: an estimate of the 10-year Treasury at the exit date, the asset-class risk premium at exit (which can differ from today's if structural fundamentals are changing), and the growth view embedded in the comp set at exit. Third, the exit cap carries structural conservatism: underwriters add a buffer above what a pure build-up would produce, on the principle that forecasting components 5–10 years out is more uncertain than observing them today.

The exit cap is the place in the model where the institutional discipline shows. A 5.88% going-in cap with a 6.38% exit cap (going-in + 50) is a conservative underwrite: you're assuming the asset trades 50 bps wider at exit than it does today, which compensates for forecast uncertainty and the natural structural deterioration of any building as it ages out of its current vintage. A 5.88% going-in with a 5.88% exit (flat spread) is a no-change assumption: the asset trades at exit at the same risk-adjusted price as today. A 5.88% going-in with a 5.63% exit (negative 25 spread) is a market call: the underwriter is betting cap rates compress, and the return arithmetic depends on that compression happening.

The Spread: The 25-50 bps Convention

The convention used in this article adds 25–50 bps to the going-in for value-add deals,flat-to-25 bps for core, and 50–100 bps for opportunistic. The convention is a starting point, not an answer. The convention encodes three things:

• Structural deterioration of any building. A property that is 7 years older at exit than today has more deferred capex, less efficient mechanical systems, and a higher capex reserve requirement. Even with no change in the macro picture, the exit cap should compensate for vintage drift. This article uses an illustrative 5 bps per year of hold for the structural aging of the building.
• Forecast uncertainty. The 10-year Treasury at exit is unknown. The risk premium at exit is unknown. The growth view at exit is unknown. Adding a buffer above the forward-component build-up compensates the buyer for that uncertainty.
• Institutional discipline. A positive spread embeds the principle that the deal should clear its return target on a conservative reversion, not on a market call. If the underwriting clears IRR with a flat or tightening spread, the deal is making the market call, and the IC should know.

Asset class and strategy refine the convention. The illustrative defaults used in this article look like this:

Table 1: Illustrative going-in to exit cap spreads by asset class and strategy. Both columns are working assumptions for this article's examples, not survey data. The “convention” column is the baseline spread; the “2026 calibration” column reflects current forecast uncertainty and asset-class-specific structural views. Suburban office is the outlier: structural repricing puts it materially wider than convention. Industrial is the other direction: secular tailwind keeps the spread tight.

The three drivers framework

The spread is not a constant; it's a function of three drivers. The convention is a population average across all three. Underwriting a specific deal requires decomposing the spread into the three drivers and sizing each one against the deal's specific facts.

Driver 1: Hold period length. A 3-year hold concentrates the cap-rate exposure on a single near-term exit. A 10-year hold dilutes it: the same percentage hit to exit value is annualized over more years, and interim cash flow makes up a larger share of total proceeds. NOI growth does not absorb the cap-rate move, because a given cap-rate move changes exit value by the same percentage whatever the hold or the growth rate. The convention should therefore not scale spread in proportion to hold length (a 10-year hold doesn't deserve 5x the spread of a 2-year hold), because the IRR cost of any given spread falls as the hold lengthens.

Driver 2: Cyclical positioning. Underwriting in 2021 (caps tight, the 10-year averaging 1.45%) is fundamentally different from underwriting in 2024 (caps wide, the 10-year averaging 4.21%; FRED DGS10 annual averages). When current cap rates are tight versus the long-run mean, mean-reversion expectations argue for a wider exit cap: the spread should reflect the gravitational pull back toward the long-run risk premium. When current cap rates are wide, the opposite applies. As a working view for May 2026, with the 10-year at 4.56% on May 22 and caps assumed to be stabilizing after the 2022–2024 widening, the cyclical contribution to the spread is small: cap rates are close to their long-run risk-adjusted mean, so mean-reversion isn't pulling them materially in either direction.

Driver 3: Structural fundamentals at the projected exit. What is the asset class doing in secular trajectory? These are working views, not forecasts. Industrial: positive (e-commerce, supply-constrained land, modest construction pipeline), so a flat or even tightening exit cap is defensible. Multifamily: neutral-to-positive (housing shortage, rent growth normalizing), so the convention spread or modestly tighter. Suburban office: negative (work-from-home structural reset, oversupply, capex needed for habitability); convention says +50 to +100 bps, structural reality says +100 to +200 bps. The structural fundamentals can override the convention by a factor of two.

When the Spread Inverts

A negative spread (exit cap tighter than going-in) is the rare case. The underwriter is explicitly assuming cap rates compress during the hold. The honesty test: when is that assumption defensible?

Case 1: Cyclical mean reversion from a clearly wide entry. Take an illustrative industrial deal acquired at a 6.50% going-in cap, well above the 4.75–5.50% range assumed above for Class A logistics, at a point when the cycle sits at peak risk premium. That is acquiring at a level the long-run mean cannot defend. If the long-run risk premium normalizes and the 10-year settles lower, suppose the implied exit cap is 5.75–6.00%. That is a spread of −50 to −75 bps, and it is the underwriter taking a cyclical mean-reversion view explicitly. It can defend at IC if the cyclical case is documented.

Case 2: Structural improvement of the specific asset. A value-add deal that takes a Class B multifamily property and renovates it to Class A pricing should compress its cap rate on the basis of the asset's own re-rating, not on macro. The going-in cap reflects Class B pricing; the exit cap should reflect Class A pricing, which this article illustratively assumes is 50–75 bps tighter. The spread inverts not because the market compresses, but because the asset itself moves up the quality curve. This is the cleanest defensible negative spread.

Case 3: A sponsor or operator re-rating. Same idea, sponsor-level. A new sponsor with a long track record bidding for an asset where the seller is a forced seller (refi failure, defaulted partnership) can underwrite a tighter exit cap on the basis that the new sponsor's reputation will re-price the asset closer to institutional-median sponsor premium. This is defensible, sometimes. The IC should require explicit attribution.

Outside these three cases, a tightening exit cap is a market call. The article-level rule of thumb:

The Sensitivity Math: What 25 bps Costs

The leverage of the exit cap assumption is the central message of this article. The spread is often described as a rule of thumb (“add 25 bps”) without showing what that 25 bps costs. The math decomposes into two pieces: the impact on unlevered exit value, and the compounded impact on levered IRR.

Unlevered: the exit value math

Exit value is Year-N+1 forward NOI divided by exit cap. Because exit value moves inversely with the cap, a change in the exit cap from c to c + Δc moves exit value by exactly:

ΔExit Value ÷ Exit Value = −Δc ÷ (c + Δc)

Worked: a 5.88% exit cap moving to 6.13% (+25 bps) reduces exit value by (0.25 / 6.13) = 4.08%. For a $20M exit value, that's an $816K loss of unleveraged terminal value. A 50-bps move (to 6.38%) costs 7.84%; a 100-bps move (to 6.88%) costs 14.5%. The relationship is convex: a 100-bps move costs about 1.9 times what a 50-bps move does, not twice, because the move is divided by a larger new cap each time, so each incremental basis point costs less than the one before.

Levered: the IRR math

The IRR impact is larger than the exit value impact because the equity check is a fraction of the deal: the lost value falls entirely on the equity. The grid below uses one model throughout: Year-1 NOI of $1.00M growing 3% a year, a purchase price of Year-1 NOI ÷ 5.88% ($17.01M), 60% LTV interest-only debt at 6.00%, annual cash flows, exit at Year N+1 forward NOI ÷ exit cap with the debt repaid at exit, and no closing costs, sale costs, capex reserves or fees. For a 60% LTV core deal, a 25-bps exit cap miss (flat to 6.13%) costs about 316 bps of levered IRR on a 3-year hold, 167 bps on a 5-year hold, 106 bps on a 7-year hold and 63 bps on a 10-year hold. Leverage raises these numbers modestly: at 65% LTV (same 6.00% debt) the 5-year and 7-year costs are about 186 and 117 bps, and at 70% LTV about 210 and 129 bps. Real deals carry costs and capex that this simplified model omits, so treat the grid as a sizing tool and re-run it in your own model.

Three readings off the grid matter institutionally.

First, the cost of a 25-bps miss is roughly 100–170 bps of IRR for a 5–7 year hold(about 167 bps at 5 years and 106 bps at 7 years for flat to 6.13%; about 162 and 102 bps for 6.38% to 6.63%). Compare it with other inputs in the same model at the 6.38% base case and a 7-year hold: NOI growth of 2.5% instead of 3.0% costs about 110 bps of IRR, growth of 2.0% costs about 223 bps, and debt at 7.00% instead of 6.00% costs about 133 bps. Per basis point of change, the exit cap is the largest lever (about 4.1 bps of IRR per bp of exit cap, against about 2.2 per bp of annual NOI growth and 1.3 per bp of debt cost). But the inputs are not equally forecastable, and a 50-bps-per-year shortfall in NOI growth costs about as much as a 25-bps cap miss. The exit cap earns its attention because it is a single number applied to the largest cash flow, not because every other input is small.

Second, the hold period is a hedging instrument. A 3-year hold concentrates cap-rate exposure on the exit; on a 10-year hold, the same percentage hit to exit value is annualized over ten years and offset by more interim cash flow. The same 25-bps miss costs about 316 bps of IRR on a 3-year hold and 63 bps on a 10-year hold. Underwriters who can flex the hold should treat hold length as a cap-rate risk management lever, not just a return-target lever.

Third, the grid is close to symmetric. A 25-bps tighter exit (5.88% to 5.63%) adds about 324, 171, 109 and 64 bps of IRR at 3, 5, 7 and 10 years, against costs of 316, 167, 106 and 63 bps for the same-sized widening. The discipline of underwriting the wider cap is not about payoff asymmetry. It is about not relying on compression to clear the return target: the upside is real, but it is a market call, and the base case should not need it.

How to Model It: The Year 11 Forward NOI Convention

Once the exit cap is chosen, the reversion calculation has one more convention that is easy to miss. The terminal value formula is:

Reversion Value = NOI_{year N+1, forward} ÷ Exit Cap

The numerator is forward NOI (the next year's projected NOI), not the trailing Year N NOI in the year of exit. For a 10-year hold, the reversion uses Year 11 forward NOI. For a 7-year hold, Year 8 forward. For a 5-year hold, Year 6 forward.

Why forward? Because the cap rate is itself defined as NOI ÷ Value, and the NOI that the buyer at exit is pricing is the next year's cash flow. The going-in cap on day one is computed on the buyer's expected Year 1 NOI; the exit cap by the same convention is computed on the next buyer's expected Year 1 NOI, which from the seller's perspective is Year N+1 forward.

Using Year N trailing instead of Year N+1 forward understates terminal value by one year of NOI growth. At 3% NOI growth, that's a 2.9% valuation error (1 − 1/1.03) on the largest single cash flow in the model. At 60% LTV and a 6.38% exit cap, it costs about 124 bps of levered IRR on a 5-year hold, 78 bps on a 7-year hold and 46 bps on a 10-year hold, roughly three-quarters of the cost of a 25-bps exit cap miss at the same hold. It is hidden inside a convention that is easy to miss. The pillararticle on the cap rate calculator and formulaintroduces the formula; this article spells out the convention.

Building the exit cap input cell

The exit cap is built up from forward components, then compared to the convention default. Six steps:

• 1. Forward 10-year Treasury. Use the current 10-year spot as the base case; reference a published forward-rate consensus for a forward path; sensitize ±100 bps as separate scenarios. Don't bake a forecast into the base case.
• 2. Forward risk premium. For most asset classes, assume the current risk premium holds at exit. For sectors with structural change (suburban office, certain retail), the risk premium at exit should widen from current; size the widening against the asset-class view.
• 3. Forward growth view. The long-run growth view tends to be optimistic at any given cycle point, so a modest trim is one way to widen the exit cap (each 25 bps trimmed from growth adds 25 bps to the cap). Size any trim together with steps 4 and 5.
• 4. Vintage drift. Add an illustrative 5 bps per year of hold for the structural aging of the asset.
• 5. Structural conservatism buffer. Add a buffer as institutional discipline, sized so that steps 3 through 5 together land inside the spread range for the asset class.
• 6. Compare to convention. If the build-up exit cap is materially different from going-in + 25/50, attribute the difference to one of the five steps above and document the rationale.

Steps 3, 4 and 5 are overlapping ways to widen the exit cap above the going-in, so they are not additive on top of the convention. On a 7-year hold, 5 bps a year of drift alone is 35 bps; adding a further 25–50 bps buffer would put the total at 60–85 bps, above the 25–50 bps range for core multifamily in Table 1. The output is the exit cap input cell. The discipline is that every basis point in the cell is attributable to a specific component, not a number pulled from convention because that's what the model template showed.

The IC Defense Template

The exit cap is the input that the IC asks about most often. The defense should be one sentence:

“Exit cap of X.XX% reflects (a) a spot 10-year of Y.YY%, (b) a risk premium of Z.ZZ% less a long-run growth assumption of G.GG%, consistent with [asset class / market / cycle] at projected exit, (c) V bps of vintage drift, and (d) B bps of institutional conservatism buffer above the build-up, which sum to X.XX%.”

Worked: a 6.38% exit cap on a 7-year multifamily hold defends as “4.56% spot 10-year (FRED DGS10, May 22, 2026) + 4.32% risk premium (held at the level implied by the 5.88% going-in) − 3.00% long-run NOI growth (held flat) + 0.35% vintage drift (7 years × 5 bps) + 0.15% conservatism buffer = 6.38%, which is going-in + 50 bps.” The components sum to the stated exit cap: 4.56 + 4.32 − 3.00 = 5.88, then 5.88 + 0.35 + 0.15 = 6.38. The risk premium and buffer are illustrative inputs; in a live deal each is sourced and attributed.

What the IC won't accept: “We used going-in + 50 bps per house convention.” That's not a defense; that's a template. The convention is a starting point. The deal-specific defense decomposes the number into its drivers and tests each one. The build-up exists for exactly this conversation.

The 2021 to 2024 stress case

The recent cycle is instructive. A 2021 underwrite that used a flat-to-25 bps exit cap spread was implicitly assuming that the 2021 rate environment was the long-run mean. FRED's annual averages for the 10-year Treasury show how far that environment moved: 1.45% in 2021 and 4.21% in 2024, a rise of 276 bps. The grid above shows how fast IRR falls as the exit cap moves away from flat: on the same inputs, a 7-year hold goes from a 12.5% levered IRR at a flat exit cap to 9.4% at +75 bps.

The discipline-level lesson is that flat spread is the assumption to question, not the assumption to default to. An underwriter who used a flat exit cap in 2021 was not making an explicit market call; they were using convention. The convention was exposed because the cycle position was unusual. The working approach in this article is to use the convention (25–50 bps wider) at minimum, and to widen further when current cap rates look tight against the long-run mean. In this article's working view, the 2026 environment is closer to the long-run mean than 2021 was, but not at the long-run mean, and modest mean-reversion is still part of a defensible underwrite.

Common Mistakes

Six errors that consistently distort exit cap underwriting:

• Defaulting to the convention without a deal-specific defense. Going-in + 25 bps is a template, not a defense. An IC question about the exit cap should be answered with attribution to forward components: the 10-year at exit, risk premium at exit, growth view at exit, vintage drift, conservatism buffer. Not with “that's the convention.”

• Using trailing Year N NOI in the reversion instead of Year N+1 forward. The cap rate is defined on forward NOI. Using trailing understates terminal value by one year of NOI growth (about 3% at 3% growth, and proportionally more on a value-add ramp where NOI grows faster). At 60% LTV and a 6.38% exit cap that is about 46–124 bps of levered IRR, depending on whether the hold is 10 or 5 years.

• Assuming flat or tightening exit cap on a base case. Flat-to-tighter exit caps are upside cases. The base case clears IRR with going-in + 25 to +50 bps; the upside case explores what tighter spreads would mean. If the deal only clears IRR on the upside case, the deal isn't clearing. The underwriting is buying a market call.

• Holding the risk premium constant when fundamentals are changing. When structural risk-premium widening is under way (suburban office is the example), trailing comps lag it, and holding the risk premium at the trailing-comp level produces exit caps that are too tight. When the structural fundamentals are clearly changing, the risk premium at exit has to widen explicitly; the comp set is a lagging indicator.

• Ignoring the hold-period leverage. Short holds (3 years) concentrate cap-rate exposure on a single exit; long holds (10+ years) annualize the same exit-value hit over more years and offset it with more interim cash flow. Underwriters who choose hold length without sensitizing exit cap leverage are missing the dominant axis of risk in their model. The hold period should be a deliberate choice that reflects the underwriter's confidence in the exit cap forecast.

• Conflating exit cap with discount rate. The exit cap is the cap rate at sale; the discount rate is the cost of capital used in the DCF. They are related by the Gordon Growth identity (discount rate = cap rate + long-run growth) but they are not the same number. A DCF that uses the exit cap as the discount rate understates the discount by the growth term, the long-run NOI growth rate, which is 300 bps at the 3% growth used in this article (5.88% + 3.00% = 8.88%), and overvalues the asset. See theDCF vs. direct capitalization article for the resolution.

Related Articles

The going-in to exit cap spread spokes off the cap rate pillar. Related deep-dives in the valuation cluster:

• Cap Rate Calculator and Formula: the pillar. NOI ÷ Value, the formula's three rearrangements, asset-class benchmarks, the calculator widget.
• Cap Rate Decomposition: Risk-Free Rate, Risk Premium, and Growth: the build-up that gives you the going-in cap from components; same build-up applied to exit-date components gives the exit cap.
• Terminal Value, Reversion, and Exit Cap Across the Hold Period: the hold-period choice and the full reversion mechanics, with the Year-11-forward convention extended.
• DCF vs. Direct Capitalization: When to Use Which: the methodological choice; the exit cap is the parameter that distinguishes a direct-cap valuation from a hold-period DCF.
• NOI: Institutional vs. Broker Calculation: what counts as NOI in the cap rate formula; the definitional differences that swing the going-in and exit cap.

FAQ

Sources

External sources cited throughout this article:

• Federal Reserve Economic Data, 10-Year Treasury Constant Maturity Rate (DGS10): daily 10-year Treasury yield series; the source of the 4.56% (May 22, 2026) and the 2021 and 2024 annual-average yields cited above.
• Green Street Commercial Property Price Index (CPPI): monthly price index built on REIT-owned asset valuations.
• NCREIF Property Index (NPI): quarterly, unleveraged total returns for institutionally held private commercial real estate, weighted by market value.
• MSCI Real Capital Analytics: transaction pricing, volume and liquidity data used for comp-set cap rates.
• Gordon, M.J. and Shapiro, E. (1956), "Capital Equipment Analysis: The Required Rate of Profit," Management Science, 3(1): original publication of the Gordon Growth Model, from which the cap rate identity (cap = r − g) is derived.