Ground Lease Rent Resets: CPI, FMV, and Hybrid Escalation Structures

Key Takeaways

Why Rent Resets Matter

A ground lease rent reset is the mechanism that adjusts the annual ground rent paid by a leasehold tenant to the fee simple landowner over the term of the lease. In a 99-year ground lease, the initial rent set at execution may bear no relationship to economic reality 30, 50, or 80 years later. Rent resets exist to bridge that gap, and the structure chosen determines whether the landlord or tenant captures the appreciation of the underlying land.

For acquisitions analysts, the rent reset structure is one of the largest variables in a ground lease cash flow model. Take an illustrative property with $4.2M of NOI before ground rent and a $600K annual ground rent, bought for $42.35M (an 8.5% unlevered yield after ground rent). Hold NOI flat for 20 years to isolate the rent effect. If the rent escalates at 2% a year, year-20 rent is $891,568 and the yield after ground rent is 7.8%. If the rent instead resets at year 20 to 6% of land value, uncapped, and the land has grown 4% a year from $10M, the new rent is $1,314,674 and the yield drops to 6.8%. At 6% land growth the reset is $1,924,281 and the yield is 5.4%.

Three reset structures recur in long-term ground leases: CPI-based escalation, fair market value (FMV) resets, and hybrid structures that combine elements of both. Each carries distinct risk profiles for tenants, landlords, and lenders. Understanding the mechanics, the modeling implications, and the litigation history of each structure is foundational to underwriting any leasehold interest.

CPI-Based Escalation

CPI-based escalation ties ground rent increases to the Consumer Price Index, typically the CPI-U (All Urban Consumers) series published monthly by the U.S. Bureau of Labor Statistics. The adjustment may occur annually, every five years, or at some other fixed interval. The formula is straightforward: the prior rent is multiplied by the change in CPI over the relevant period.

CPI escalation clauses often include caps and floors to bound the adjustment. Illustrative examples of how a ground lease might specify them:

• Annual CPI adjustment with a 2% floor and 4% cap
• Five-year compounded CPI adjustment with a cumulative cap of 20% per period (equivalent to ~3.7% annually)
• Annual CPI adjustment with a 6% annual cap but no cumulative cap

The cap-and-floor structure creates a corridor of predictability. On a $500,000 initial ground rent with a 2% floor and 4% cap applied annually, the rent at year 25 falls between $820,303 (at the 2% floor) and $1,332,919 (at the 4% cap). That's a $512,616 range: significant, but bounded. An uncapped FMV reset has no contractual ceiling; the year-25 rent depends on what the land is worth on the reset date.

The limitation of CPI-based structures is that CPI tracks consumer price inflation, not land values. Where land values grow faster than the CPI cap, a CPI-capped escalation transfers value from the landowner to the tenant. That is a feature for the tenant and a cost for the landowner, which is why landowners may push for FMV resets or hybrid structures.

FMV Resets and Arbitration Risk

A fair market value reset adjusts ground rent to a specified percentage of the then-current fair market value of the land. The percentage and the interval are set lease by lease. The Penn 1 lease discussed below uses 6% of land value, with resets at each of three 25-year renewal periods. The rent between reset dates may be fixed or may escalate via CPI.

The FMV percentage and reset interval determine the magnitude of the adjustment. Consider an illustrative ground lease with an initial land value of $10M, a 6% rent factor (so a $600,000 starting rent), and resets every 20 years:

Table 1: Illustrative FMV rent resets at 20-year intervals assuming 4% annual land appreciation and a 6% rent factor. By year 60, the ground rent is more than 10x the initial amount.

The theoretical elegance of FMV resets masks their practical problem: determining fair market value is inherently contentious. A lease commonly provides for mutual agreement first, failing which the dispute goes to appraisal or arbitration. Many leases use a three-appraiser panel, with each side picking one appraiser and the third chosen by the two party appraisers or by a body the lease names. The Penn 1 lease, for example, calls for a third appraiser chosen by the Real Estate Board of New York or the court (The Real Deal, February 2026). Read the clause in each lease rather than assuming a standard.

The appraisal process creates large modeling uncertainty. In the Vornado/Penn 1 dispute (discussed below), the landowner's appraiser valued the land at roughly $1.9B, while the panel's award implied a value near $250M. And as Penn 1 shows, the award itself can be challenged in court, which stretches the period of uncertainty by years.

Hybrid Structures

Hybrid rent reset structures address the limitations of pure CPI and pure FMV mechanisms. One hybrid pattern: CPI-based annual escalation between periodic FMV resets, with a cumulative cap on the reset rent.

An illustrative hybrid clause, written for this article and consistent with the $10M land and $600,000 starting rent used above, might read:

Ground rent shall increase annually by the greater of (a) 2.0% or (b) the percentage increase in CPI-U, provided that the annual increase shall not exceed 4.0%. At the beginning of each 20-year period following the commencement date, the ground rent shall be reset to the greater of (i) the ground rent then in effect and (ii) the lesser of (A) 6.0% of the then-fair market value of the land and (B) the rent that would have resulted from compounding the initial rent at 3.5% per annum from the commencement date.

The cumulative cap is the critical provision. In the example above, the 3.5% compounded cap means that regardless of how much the land appreciates, the FMV reset cannot lift year-20 rent above $600,000 × (1.035)^20 = $1,193,873. Even at 4% land appreciation, the uncapped reset would be $1,314,674 (6% of $21.9M), so the cap binds by a narrow margin. A 6% land appreciation rate would produce an uncapped year-20 reset of $1,924,281 (6% of $32.1M), 61% above the cap. With 8% land appreciation, the uncapped reset would be $2,796,574 (6% of $46.6M), but the 3.5% cap still holds it to $1,193,873.

The hybrid structure provides three things simultaneously:

• Predictability for the tenant and lender. The cumulative cap creates a maximum rent trajectory that can be underwritten with certainty.
• Partial upside for the landlord. The 2% floor guarantees minimum growth, and the FMV reset lets the landlord capture land appreciation up to the cap. Above 3.5% a year compounded, the cap hands the excess to the tenant, so the landlord's participation is bounded, not full.
• Reduced arbitration risk. The cap limits the stakes of any FMV dispute. When the cap clearly binds (in the example above, at every reset once land grows 4% a year or more), the exact FMV determination stops mattering.

Not every institutional ground lease uses an FMV reset. Safehold's 2024 Form 10-K describes contractual base rent increases "either at a specified percentage or consumer price index ('CPI') based, or both," with many leases including a periodic increase based on prior years' cumulative CPI growth. Those CPI lookbacks are generally capped between 3.0% and 3.5% and generally start between years 11 and 21 of the lease term. That is a capped CPI structure, not an FMV reset. Beyond the escalators, the 10-K points to residual rights to take ownership of the buildings and other improvements at lease end, rather than to reappraising the land mid-term.

Modeling Rent Over 99 Years

A 99-year ground lease cash flow model must accommodate multiple reset mechanisms across a time horizon that exceeds most practitioners' experience. The modeling challenge is not just mathematical; it requires assumptions about economic variables (inflation, land appreciation, discount rates) that compound dramatically over multi-generational time frames.

The key modeling inputs for a ground lease rent projection:

• Initial ground rent: the contractual rent at lease execution
• Escalation type: CPI, fixed percentage, FMV, or hybrid
• Escalation frequency: annual, every 5 years, or at each reset interval the lease sets
• CPI assumption: long-run CPI-U forecast (this article's low, base and high scenarios use 2.0%, 2.5% and 3.5%; Figure 2 also shows 3% CPI and 2% fixed escalation)
• Land appreciation rate: for FMV resets (this article's low, base and high scenarios use 3%, 4% and 6%; Figure 2 adds 5% and a 7% stress case, and the cap example tests 8%)
• FMV rent factor: percentage of land value applied at reset, taken from the lease (6% in this article's examples)
• Caps and floors: annual caps, cumulative caps, minimum increases
• Discount rate: for present-valuing each rent stream, set by its risk (see Common Mistakes below)

The leasehold value equals the fee simple value less the leased fee value, where the leased fee is the present value of the ground rent plus the present value of the reversion at lease expiry. Alternatively, value the leasehold directly as the present value of NOI after ground rent over the remaining term. The leasehold vs fee simple article in this series works through both methods. A higher ground rent trajectory means lower leasehold value, and at some point the ground rent consumes enough of the NOI that the leasehold becomes unfinanceable.

The Vornado / Penn 1 Dispute

The Penn 1 (One Penn Plaza, Manhattan) rent reset illustrates the real-world stakes of FMV resets. Vornado Realty Trust controls the tower through a ground lease on land owned by the Korein family. The 1970 lease sets rent at 6% of the land's fair market value (Hoodline, February 2026). Vornado's filings describe fair market value resets at each of three 25-year renewal periods, the first of which began on June 17, 2023 (Vornado Form 10-Q for the quarter ended June 30, 2026). Before the reset, Vornado paid about $2.5M a year.

The parties could not agree, and the reset went to an appraisal panel. On April 22, 2025, the panel set annual rent at $15M, roughly six times the prior rent but below the $26.2M a year Vornado had been accruing, so Vornado reversed about $17.2M of previously booked rent expense (Vornado third quarter 2025 results). The landowner then moved to vacate the award, and on October 31, 2025, the New York County Supreme Court granted that motion. Vornado is appealing. A separate pending action concerns whether a sublease affects the value of the land. Vornado has disclosed that, if the vacatur is reversed on appeal and the landowner then prevails in that sublease litigation, the panel's decision sets annual rent for the 25-year term at $20.22M, retroactive to June 17, 2023. Meanwhile Vornado is paying based on $15M a year.

The positions were far apart. The Koreins' appraiser valued the land at about $1.9B, which at 6% implies roughly $114M a year in rent. Vornado's appraiser valued it at close to nothing, in part because of the sublease, which the Koreins say expired long ago; the panel's figures came to $250M if the land is encumbered by the sublease and $337M if it is not (The Real Deal, February 2026). At 6%, those are the $15M and $20.22M outcomes. In February 2026 the family sued to have a court adopt the $1.9B value. A Manhattan judge dismissed the suit in June 2026, writing that "The Court is simply not empowered to determine the fair market value in lieu of enforcing the contractual provision between the parties" (The Real Deal, June 2026). More than three years into the term, the rent is still not final.

This case highlights several structural risks inherent to FMV resets:

• Valuation methodology disputes. What exactly is being appraised? At Penn 1, the fight turned partly on whether an alleged sublease encumbers the land, which Vornado's appraiser treated as pushing land value toward zero. The answer dramatically changes the result, as the gap between a $250M panel figure and a $1.9B landowner appraisal shows.
• Arbitration is not the last word. The losing side can still move to vacate. Penn 1's reset has run through a panel award, a vacatur, an appeal, a sublease lawsuit, and a dismissed valuation suit.
• Market timing. A reset that occurs during a market peak (or trough) locks in that valuation until the next reset, 25 years later at Penn 1. Unless the lease provides a mechanism, the tenant cannot renegotiate if the market declines after the reset.
• Retroactive exposure. The reset date does not move while the dispute runs. Vornado paid about $22.4M of prior-period ground rent in 2025 based on the panel determination and now pays at the $15M rate, so its open exposure is the gap to whatever rent is finally fixed: at $20.22M, about $5.22M a year back to June 2023.

Common Mistakes

These errors appear repeatedly in ground lease underwriting, particularly from analysts who have modeled fee simple deals but are new to leasehold structures:

• Using a single rent escalation assumption. Ground lease rent is one of the highest-sensitivity variables in a leasehold model. A single-scenario projection (e.g., "2.5% annual CPI") creates false precision. Run the low, base and high cases used in this article (CPI at 2.0%, 2.5% and 3.5%; land appreciation at 3%, 4% and 6%), plus stress cases where the lease has FMV exposure. The range of outcomes is the output, not a single number.
• Ignoring the cumulative cap when it exists. Some hybrid structures include a cumulative compounded cap (e.g., 3.5% per annum from commencement). If the cap binds, as it does at every reset in this article's example at 4% land growth, the FMV figure above the cap does not change the rent. Model the cap first, then check whether the FMV reset exceeds it.
• Discounting every rent stream at one rate. The ground rent stream has different risk characteristics than the property's NOI, and different rent structures differ from each other. Fixed and capped CPI rent is a contractual charge whose default can cost the tenant the leasehold, and it can be discounted near a senior secured credit rate. Uncapped FMV-reset rent moves with land values, so it carries real estate risk and deserves a higher rate or explicit scenario weighting.
• Assuming FMV resets are path-dependent, or that they are not. Under a pure percentage-of-land-value reset, as in Table 1, each reset is recomputed from then-current land value; the year-40 rent does not depend on the year-20 rent. Growth in the tail comes from land appreciation. Check the lease for a no-decrease floor (like the "greater of" test in the sample clause) or a cap measured from the prior reset, either of which makes resets path-dependent.
• Treating CPI as deterministic. According to Bureau of Labor Statistics CPI-U data, the index rose at about 2.5% a year from September 1995 to September 2025, but 12-month changes ranged from -2.1% (July 2009) to 9.1% (June 2022). In a lease with a 2% floor and 6% cap, the effective CPI is bounded, but the variance still matters for present value calculations.
• Modeling land appreciation equal to building appreciation. Land and improvements behave differently: improvements depreciate physically while land does not. Using a blended property appreciation rate for FMV land resets can misstate the reset amount; project land value on its own.
• Ignoring arbitration cost and delay. A contested FMV reset can run for years. Penn 1's reset date was June 17, 2023, and the rent was still not final more than three years later. What the tenant pays in the meantime depends on the lease and the parties, and the back-payment at resolution can create a significant cash flow event. Model the timing, not just the amount.

How to Model It

A properly structured ground lease rent model requires a dedicated tab in your Excel workbook, separate from the operating pro forma, that projects the ground rent trajectory across the full lease term. Here is what the tab should contain:

Rent escalation schedule

Row-by-row annual projection of ground rent for the full lease term (99 years or remaining term). Each year should show: the applicable escalation method (CPI, fixed, or FMV reset), the escalation rate or FMV calculation, any cap/floor applied, and the resulting rent. Color-code FMV reset years differently from CPI years; they are structurally different calculations.

FMV reset calculator

A separate section for each FMV reset date. Inputs: land value appreciation rate (toggle between scenarios), FMV rent factor (from the lease), cumulative cap calculation (if applicable). Output: the reset rent before and after any cap. The cap test should be a clear TRUE/FALSE flag: does the uncapped FMV rent exceed the cumulative cap?

Scenario matrix

A data table or scenario manager showing ground rent trajectories under at least three assumptions: low (2% CPI, 3% land appreciation), base (2.5% CPI, 4% land appreciation, matching Table 1 and Figure 1), and high (3.5% CPI, 6% land appreciation). Present the year-by-year rent and the NPV of each stream side by side.

Integration with pro forma

The annual ground rent from this tab should feed directly into your operating pro forma as a line item below the NOI line (or above, depending on your convention). The pro forma should show NOI before ground rent, ground rent, and net cash flow after ground rent. This makes the ground rent burden visible at every point in the hold period.

The test of a good ground lease rent model: change the land appreciation rate from 4% to 6% and see if the FMV resets, the cumulative cap test, and the pro forma cash flow all update automatically. If you have to manually recalculate any step, the model isn't properly linked.

Related Articles

This article is part of the ground leases underwriting series. Each article addresses a specific dimension of ground lease deal structuring:

• Ground Lease Underwriting: Leasehold vs Fee Simple: valuation methodology, cap rate adjustments, and the mechanics of splitting a fee simple value into leasehold and leased fee interests.
• Institutional Ground Leases and Reversion Economics: how Safehold and institutional landowners structure modern ground leases, reversion value modeling, and the economics of fee simple conversion.
• Subordinated vs Unsubordinated Ground Leases: how subordination status affects financing, LTV constraints, lender requirements, and SNDA mechanics.