Resort Conversion: Underwriting the Capex Stack

The underwrite is a sequence, not a spreadsheet#

A value-add resort conversion collapses three separate bets into a single IRR line, and the discipline is to test each one before they are allowed to merge. The first bet is the basis: does the acquisition price, plus the full capital programme, sit below what it would cost to build the same asset from the ground up. The second bet is the capex programme itself, and the contingency that sits underneath it. The third is the stabilization curve, the number of years the operating business needs to climb from opening-day occupancy to the income the exit valuation assumes.

Each bet fails on its own terms, not on the others'. A cheap basis does not protect against a capex overrun. A well-scoped capex budget does not protect against a slow ramp. A fast ramp does not rescue a basis that was never cheap enough to begin with. Underwriting models that net these three into one blended return before stress-testing each individually are the ones that discover the error only after capital is committed.

This piece works through the sequence in order: basis, capex, contingency, financing, and the stabilization curve, using benchmark data that is public and sourced rather than deal-specific. No single number here is a substitute for underwriting a specific building. The purpose is to show the shape of the discipline, and where, on the public record, that discipline most often breaks.

Basis versus replacement cost sets the ceiling#

A conversion only clears the first bet if acquisition basis plus fully-loaded capex lands meaningfully under the cost of building the same asset new. That ceiling is not abstract. HVS's 2026 U.S. Hotel Development Cost Survey, drawn from 2025 construction budgets, puts the overall median new-build cost at roughly $213,000 per key, with full-service product at a median of $467,000 per key and luxury product exceeding $1.6 million per key — measured, from a published industry survey covering six U.S. product tiers (see table below).

Those figures are American, and the principal should treat them as a ceiling reference, not a European transaction price. No comparably authoritative, published per-key development cost series exists for the European resort markets Victaura underwrites in. What travels across markets is the logic, not the number: a standing structure with the right bones removes the shell, the foundations and much of the construction timeline from the cost stack, and that removed cost is where the conversion's basis advantage over ground-up development is manufactured.

The test is arithmetic, and it has to be run before the building is admired. If acquisition price plus capex plus contingency plus carry does not sit comfortably below a defensible replacement-cost estimate for the finished product tier, the deal is not a conversion with an embedded discount. It is an expensive way to buy a building that happens to be old.

The capex stack has three buckets, and they fail differently#

A conversion budget is not one number; it is three buckets that behave nothing alike under stress. The first is hard cost: structure, envelope, mechanical, electrical, plumbing, life safety, vertical circulation — the part of the budget most exposed to what a standing building is hiding behind finished surfaces. The second is FF&E and the interior fit-out, which is largely a brand-standard and design decision rather than a structural one. The third is soft cost: design, permitting, financing fees, pre-opening expense, and, where a flag is involved, the brand's Property Improvement Plan scope.

The FF&E bucket has published tiering, and it moves with product positioning. Industry cost guides compiling HVS and Nehmer benchmark data put total interior fit-out in the range of $15,000–$25,000 per key for select-service product, $30,000–$60,000 per key for upper-upscale, and above $80,000 per key for true luxury once finishes, FF&E and soft costs are fully loaded — an industry estimate, compiled from a secondary cost-guide source rather than a single primary institutional survey, and treated here with that caveat attached.

Hard cost is the bucket most exposed to inflation between the budget date and the contract date, and that exposure is measured, not assumed. Turner Construction's Building Cost Index averaged 1485 across 2025, a 4.1% increase over 2024, with the index accelerating through the year — measured, from Turner's quarterly published index. A capex budget priced twelve to eighteen months before a general contract is signed is, on this data, already understating hard cost by several points before a single wall is opened.

Contingency is sized to what cannot be seen#

Ground-up construction is a known unknown; a conversion is an unknown unknown wearing a known building. New construction proceeds from drawings the owner controls from day one. A conversion proceeds from a structure whose condition is discovered progressively, as finishes come off and systems are exposed — asbestos behind a 1980s ceiling, a slab that will not take the loads a new mechanical plant requires, a facade that a heritage authority will not let the plan touch. Industry practice commonly cites a contingency reserve of 10–15% of hard cost as the working standard, rising to 15–20% for structures built before 1990 — an industry estimate, compiled across renovation cost-guide sources rather than anchored to a single audited dataset, and disclosed here with that limitation attached.

A conversion underwritten at a ground-up contingency is not a discounted deal; it is a ground-up project that has quietly waived its own insurance. The discount the principal believes is captured in the basis is the same discount that funds the contingency the standing structure requires. Spend it twice — once assumed in the return, once actually needed on site — and the spread the conversion was underwritten to capture is the first thing to disappear, often before the building has a roof back on.

Contingency and the Turner cost-index trend are not independent variables, and underwriting them as if they were is the common error. A static contingency percentage, fixed at signing and never revisited, does not account for a hard-cost index moving at over 4% a year through the construction period. The discipline is to size the contingency against the specific building's condition, then re-test it against realistic cost escalation for the length of the build — not to treat 10% as a number that ages well.

The financing bridge prices the same risk twice#

A lender underwriting a hotel construction loan is pricing the same unknowns the sponsor's contingency line is meant to absorb, and the terms say so. Conventional bank construction financing for branded select-service and upper-upscale assets was pricing at roughly 6.25%–7.25% all-in yield through 2025–2026, at loan-to-cost of 55%–70%; private construction lenders were offering more aggressive advance rates of 70%–75% loan-to-cost, but at 11.0%–12.5% interest — industry estimate, compiled from lender-facing financing guides rather than a primary regulatory data series. U.S. hotel loan originations reached roughly $27 billion in the first half of 2025 on the same reporting, a signal that construction and acquisition debt for the sector was flowing, not frozen.

The equity check is the first-loss position on exactly the gap a thin contingency creates. At loan-to-cost of 55%–70%, the sponsor is funding 30%–45% of total project cost in equity before a single room generates revenue; first-time hospitality sponsors were seeing equity requirements closer to 30%–35%, on the same sourcing. If hard cost runs over and the contingency line is exhausted, the shortfall does not go to the lender. It goes to the equity, at the exact moment in the project — mid-construction, pre-revenue — when raising incremental capital is hardest and most expensive.

This is why contingency sizing is not a conservatism exercise; it is a financing-structure decision. A budget that treats contingency as padding to be trimmed for a cleaner headline return is, in effect, asking the equity to underwrite construction risk that the lender has already declined to price. The two numbers — loan-to-cost and contingency — should be sized together, because a shortfall in one becomes, immediately, a call on the other.

Stabilization: the three-year assumption, tested#

The assumption that a hotel stabilizes in three years is old enough to have been tested academically, and the test mostly holds — with conditions. A frequently cited study circulated by the International Society of Hospitality Consultants, examining hotel occupancy performance against the standard three-year stabilization assumption used in appraisal and lending, found the assumption broadly reasonable as a market-wide convention, while flagging that individual asset types deviate from it meaningfully — measured, an empirical academic study, though dated relative to the current cycle and not resort-specific.

HVS and USALI-aligned feasibility practice builds this ramp explicitly into a ten-year projection rather than assuming a single stabilized year from opening. The standard approach projects occupancy and average rate rising from an opening-year base to a stabilized year, then holds that stabilized performance across the balance of a ten-year hold — an industry-standard methodology, not a fixed timetable, and one that varies by asset type: smaller, select-service properties and hotels near strong, existing demand generators tend to stabilize in two to three years, while larger full-service and resort properties, more dependent on building a client base from scratch, more often plateau in the third or fourth full operating year.

A conversion complicates this further, and the complication is directional rather than measured. A repositioned or rebranded resort is not simply a new hotel; it is a known location carrying an unfamiliar product, an unfamiliar operator, or both. Whether that shortens the ramp, because the location has pre-existing demand awareness, or lengthens it, because the market has to unlearn the prior positioning before it accepts the new one, is asset-specific. Underwriting a repositioned resort to the fast end of the stabilization range without a specific reason to expect it is the ramp-side equivalent of underwriting a conversion at ground-up contingency.

What the ramp-up curve is being asked to deliver#

The stabilization curve for a repositioned luxury resort is not underwriting to the hotel market broadly; it is underwriting to a segment that has been behaving differently from the rest of the index. Full-year 2025 STR data showed overall U.S. hotel occupancy and RevPAR softening, while the luxury segment posted RevPAR growth of roughly 3% for the year, driven entirely by rate rather than occupancy gains — measured, STR full-year U.S. performance data. In shorter, more volatile weekly windows across the year, luxury and upper-upscale properties recorded RevPAR gains above 10%, coinciding with a reported roughly 20% rise in group demand in those tiers — measured, but a weekly figure, and one that should not be annualized or treated as a steady-state run rate.

A single documented case illustrates the scale of what a well-executed renovation can move, and the scale of the caveat attached to it. A post-renovation RevPAR gain of 28.2% has been cited, referencing SEC-filed disclosure, for a renovated Extended Stay America portfolio asset — broker-reported, a single-asset, non-luxury data point surfaced through a secondary industry source rather than the underlying filing itself, and not a segment benchmark. It is evidence that renovation-driven RevPAR gains of that magnitude exist in the U.S. record; it is not evidence that any specific resort conversion will replicate it.

The honest reading of both data points together is that the tailwind is real and the multiple is not guaranteed. Luxury demand was, on the 2025 data, pricing power that the rest of the market did not have. A conversion underwriting is entitled to lean on that tailwind for the segment it is repositioning into. It is not entitled to borrow the outer edge of a single-asset renovation case as the base case for the stabilized year.

Exit timing and the capital chasing this trade#

A stabilization curve is only relevant if there is a buyer at the end of it, and the global exit environment for luxury resort product improved through 2025 into 2026 on the public transaction record. JLL's 2026 Global Hotel Investment Outlook reported global hotel transaction volumes up 22% from the 2023 trough, with the Americas leading at 27% growth in 2025 and EMEA up 4%, against a 20% decline in Asia Pacific — measured, published transaction-volume data from a primary industry source. JLL flagged luxury resorts and trophy assets specifically as a segment of rising institutional appetite, and expected large transactions above $250 million to increase materially in 2026.

This matters to the underwrite in a specific, narrow way: it is evidence about liquidity, not about price. A recovering transaction market means more counterparties are active and more capital is available to close a sale at the end of a hold period; it says less about the specific exit multiple a converted resort will achieve, which depends on the stabilized NOI actually delivered, the operator and brand in place at exit, and the buyer pool's appetite for that specific market at that specific time.

The prudent use of this data is as a floor on the liquidity assumption, not a ceiling on the return assumption. A model that assumes a ready buyer exists for a stabilized, well-branded luxury resort asset at the end of a five- to seven-year hold is, on the 2025–2026 record, assuming something that is currently true of the broader market. A model that assumes a specific exit cap rate or multiple without a defensible market comparable is assuming something this data does not support.

Where the model breaks#

The failure mode that matters is not any single bad assumption; it is the correlation between the assumptions that a spreadsheet, run once, does not show. A construction delay caused by a discovered structural issue does two things at once: it consumes the contingency the delay itself required, and it pushes the opening date later, which pushes the entire stabilization curve later against the same construction loan maturity. The two effects compound rather than add. A model that sensitizes contingency and stabilization timing independently, each against an otherwise-unchanged base case, is not stress-testing the deal; it is stress-testing two deals that do not exist.

Brand scope is a second, distinct failure mode specific to conversions rather than ground-up work. Where a franchise or management flag is involved, the operator's property-improvement and brand-standard requirements are set against the brand's current specification, which can tighten between the letter of intent and the signed franchise or management agreement, and which the sponsor does not fully control. A capex budget locked at acquisition against a brand standard that has not yet been finalized is a budget with an open liability, not a closed one.

The discipline that survives this is sequencing, not optimism. Price the contingency against the specific structure's known condition, not a category average. Stress-test contingency and stabilization together, because a delay touches both. Do not finalize the capex number until the brand standard is fixed in writing. And size the equity check to the possibility that all three of these move against the sponsor in the same construction cycle, because on the public financing terms cited above, that is precisely the risk the equity, not the lender, is holding.

Honestly disclosed#

The weakest links in this underwriting framework are named here rather than smoothed over. The FF&E per-key ranges and the 10%–15% contingency convention both trace to secondary cost-guide compilations rather than to a single primary institutional survey with disclosed methodology; they are treated throughout as industry estimates, and a specific building should be budgeted bottom-up by a quantity surveyor, not benchmarked off this range alone. The construction-loan terms cited are similarly compiled from lender-facing financing guides rather than a regulatory data series, and should be treated as directional for pricing conversations, not as committed terms.

Almost every hard number in this piece is American, and the resort markets Victaura underwrites in are not. HVS's development-cost survey, the Turner cost index, the construction-financing terms and the Extended Stay America renovation case are all U.S.-sourced. There is no consolidated, published European equivalent for conversion-specific capex or contingency benchmarking that this piece would anchor a European underwrite to. The figures here establish the shape of the risk and the order of magnitude; they do not substitute for a market-specific quantity survey and a local lender term sheet.

The stabilization and RevPAR data carry their own, narrower limits. The three-year stabilization study is an older academic paper, not a current-cycle dataset, and it is not resort-specific. The weekly STR RevPAR figures cited are short-window and should not be read as an annual run rate. The single-asset Extended Stay America renovation case is exactly that: one asset, in one segment, reported through a secondary source rather than the underlying SEC filing itself, and it is not a claim about what any specific resort conversion will achieve.

Skin in the game disclosure. Victaura, through its parent Greystone B.V. (Netherlands), holds an active operating position in prime resort property. Readers should assume commentary may be influenced by, or benefit, Greystone's position. This document is classified as marketing material under MiFID II Article 24(3). It is not investment advice.