What Drives Hotel Occupancy and ADR: A Demand Driver Framework for Hotel Feasibility Studies

Every hotel loan rests on two numbers, and most feasibility studies treat them as if they moved together. They do not. Occupancy is bought with generators that fill rooms. ADR is bought with scarcity. This analysis ranks the demand drivers behind both, separates the ones that are measured from the ones that are assumed, and shows how a lender-grade study should turn them into a projection that survives an SBA, USDA or rating-agency review.

A hotel is a 365-night auction. Each night the market clears at whatever rate the last available room will fetch, and the next morning the inventory resets to zero. That structure is why hotel underwriting is harder than any other income-property asset, and why the two headline metrics, occupancy and average daily rate, answer different questions. Occupancy asks who is coming. ADR asks how badly they need the room on the nights when rooms are scarce.

The distinction matters more in 2026 than at any point in the last decade. The U.S. hotel industry closed 2025 with occupancy of 62.3%, ADR of $160.80 and RevPAR of $100.19, the first full-year RevPAR decline outside a recession on record (1, 71). Nominal rate was at a record. Real rate was not. From 2019 to 2025, ADR rose 22.2% while the consumer price index rose 25.9%, which leaves real ADR 3.0% below 2019 and real RevPAR 8.1% below, once occupancy is included (71, 8). The industry has sold rooms at record prices and still lost pricing power.

The first eight months of 2026 reversed the headline. Occupancy ran at 64.3%, ADR at $166.25 and RevPAR at $106.91, up 4.9%, with June and July inflated by the World Cup and the July 4 anniversary (71). Strip those out and August was back to a 1.5% ADR gain against 3.4% inflation (3, 8). The CoStar and Tourism Economics forecast of +4.4% RevPAR for 2026 is a rate story: demand growth is forecast at only +1.7%, and 2027 is already penciled at +2.1% RevPAR with ADR up 1.6% (2).

That is the environment a feasibility study has to describe. A projection that shows a proposed hotel reaching 70% occupancy and above-inflation rate growth needs to say which generators deliver the rooms, which nights deliver the rate, and what happens to both when the largest generator stops.

Start with the mechanics, because they explain almost everything that follows.

Demand is price-inelastic at the property level. The best U.S. estimate, from 481 urban hotels across 22 major markets, puts own-price elasticity at about -0.13: a 1% rate increase costs roughly 0.13% of room nights (43). Cross-price elasticity, the share of guests who switch to a competitor when one hotel moves its rate, rises as you move down the chain scale, to 0.32 for limited service and 0.44 for economy (43). The more recent evidence runs the same way. Across more than 2,500 hotels from 2018 to 2021, properties that held or raised rate gained more RevPAR share than those that cut (45). Rate cuts rarely buy enough occupancy to pay for themselves. Projections that depend on discounting to fill a hotel are unsupported by the evidence.

Income elasticity runs the other way. Upper-priced hotels are far more sensitive to income than lower-priced ones: long-run income elasticity of about 1.01 in the top 50 markets for upper-priced product against 0.43 for lower-priced (44). The upper tier holds rate better on any given night and loses demand faster in a downturn. That trade-off sits at the center of the luxury-versus-economy divergence described below.

Rate is non-linear in occupancy. Daily regressions on San Diego, Austin and Louisville show room rates rising only gradually until a market reaches roughly 75% to 80% occupancy, then accelerating sharply (14). In San Diego a convention event lifts occupancy from 60% to about 73% at a soft baseline, from 75% to 82% at a normal one, and only from 80% to 85% when the market is already full. Above 80%, incremental demand shows up mainly as rate, because there are no rooms left to sell. STR's financial-performance analysts put peak gross operating profit margin at or slightly above 80% occupancy, and found that upper-upscale hotels are most likely to reach 90% occupancy when group is 30% to 40% of the mix (46).

The compression premium is large and measurable. A hotel REIT reported in 2011 that ADR on nights above 80% occupancy ran 7% ahead of the portfolio average, and 13% ahead on nights above 90% (47). In Chicago in early June 2026, overlapping events pushed weekend occupancy to 92.9% and lifted ADR 25.1%; downtown hotels reached 97.6% with weekend RevPAR up 41.8% (48). What the public record does not contain is a national figure for the share of annual ADR growth attributable to compression nights. Nobody has published it. It can only be built market by market from daily data, which is exactly what a feasibility study should do.

Mix sets the ceiling. In luxury and upper-upscale hotels, transient business is about 60% of room nights, group 35% and contract 4% to 5% (41, 42). Those segments carry very different rates. The largest U.S. lodging REIT's 2024 segment data imply contract rooms at about $206, group at about $281 and transient at about $338, so contract business ran roughly 39% below transient (42). Across the whole industry, transient leisure accounts for about 70% of occupancy (13). The segment a hotel fills with determines how much rate it can earn.

The practical rule for underwriting follows directly. A market averaging 70% occupancy with forty compression nights a year supports more rate growth than a market averaging 72% with five. The annual average occupancy figure in a feasibility study tells a lender almost nothing about rate until the distribution of nights behind it is shown.

The same mechanics explain why 2025 and 2026 have been two different industries depending on which end of the price ladder a hotel sits on.

The MMCG national series through August 2026 shows luxury and upper-upscale hotels at 69.0% occupancy, $288.78 ADR and $199.36 RevPAR year to date, up 6.9% (71). Upscale and upper-midscale hotels, the heart of the select-service book that SBA and USDA lenders actually finance, ran at 68.5% occupancy and $152.99 ADR, with RevPAR up 4.0% after a 1.8% decline in 2025 (71). Midscale and economy hotels ran at 56.0% occupancy and $87.66 ADR, with RevPAR up 1.1% after falling 3.4% in 2025 and 1.2% in 2024 (71).

The midscale and economy series is the one lenders should study. Demand in that class fell 4.4% in 2023, 2.1% in 2024 and 2.5% in 2025, and supply shrank every year because owners removed or converted rooms faster than developers added them (71). ADR in 2025 was $86.16, against $75.86 in 2019: a 13.6% nominal gain over six years in which prices rose 25.9% (71, 8). In real terms the economy segment has lost about a tenth of its rate. Only luxury has posted ADR growth above inflation in 2026 (2).

The forecasters describe the mechanism plainly. Households earning $200,000 or more account for 25% of travel spending but only 11% of households, and "financial pressures on consumers are a limiting factor for consistent rate growth among lower scales" (2). Midscale and economy RevPAR growth in 2026 comes from demand, not rate. For a proposed limited-service hotel in a secondary market, the base-case assumption should be ADR growth at or below inflation, and the sensitivity case should hold rate flat in nominal terms.

Every hotel feasibility study lists demand generators. Few distinguish between generators whose effect on occupancy and rate has been measured and generators whose effect is asserted. The distinction is the difference between a projection a lender can test and one it has to take on faith.

The ranking below runs from best-documented to least.

Convention centers and event venues. This is the only generator class with recent, primary, daily-data measurement. Event days lift occupancy by 5 to 13 points depending on the baseline, with the rate effect dominating in full markets (14). A study of nine convention headquarters hotel openings found competitive-set demand grew by an average of 37.8% against supply growth of 34.4%, which is the clearest evidence in the literature that a specific kind of new supply induces demand rather than splitting it (15). The caution is in the public-sector projections: Washington, D.C.'s new convention center delivered fewer than 275,000 room nights in 2010 against a projection of nearly 730,000, Austin delivered 149,000 against 314,000, and the researcher who compiled those cases concluded that getting half the projected business "is about the norm" (58). Convention demand supports both occupancy and rate, is midweek and shoulder-season, and is bankable at roughly half of what the convention authority projects.

Stadiums and college football. Also measured, also rate-driven, and almost entirely weekend. When Nashville's stadium hosted concerts, nearby hotels ran $369 ADR at 87% occupancy, against $238 at 66% for games (16). A ten-year study of a major college football town found higher ADR and occupancy within three miles of the stadium on home-game dates, with the effect fading between three and twenty miles, and found that team rankings move hotel rates while win-loss records do not (18). Florida State home games add about $81 to Leon County ADR on the day before and after (19). The size of a single game weekend in a small market is startling: Columbia, South Carolina's RevPAR rose 44.2% in one September 2025 week on an 83% weekend gain, then fell 45.5% the following week when the schedule moved (21). The first campus-hosted College Football Playoff round produced RevPAR 32% below an average regular-season home game, because the December calendar had nothing else in it (17). Sports demand compresses a handful of weekends. Extrapolating those weekends into an annual average is the most common error in college-town studies.

Youth and amateur sports tourism. Large in aggregate, thin at the property level. The industry's 2026 report counts 124.3 million room nights and $111.2 billion in direct spending in 2025, and 65% of destinations name sports events their largest room-night generator (29). The figures are self-reported by sports commissions, the scope changed between editions, and the demand is weekend, summer and family-rate. It fills rooms; it lifts ADR only on tournament weekends.

Commercial airports. Hotels physically connected to terminals posted a 5 to 5.8 point occupancy premium and a $24 to $27 ADR premium over other full-service airport hotels in the only published comparison, which is now dated (28). Air passenger counts are a valid correlate of airport-submarket demand. No published elasticity converts enplanements into room nights, so airport demand in a study is a share-of-comp-set calculation, not a ratio.

Hospitals and medical campuses. Real, steady, weekday and discounted, and almost entirely undocumented at the ratio level. There is no published figure for room nights per licensed bed from any appraisal, consulting or academic source. What exists is market-level. Rochester, Minnesota, home of the Mayo Clinic, sold 1,348,839 room nights in 2025, more than 52,000 above 2019, with a weekday-heavy pattern that is the reverse of most markets (26). Year-to-date 2026 occupancy there was 58.6% at a $134.42 ADR (26). A world-class medical generator supports a 6,000-room market at low-60s occupancy at best, because developers respond to irreplaceable demand with new supply. A 225-room dual-brand across from the Mayo campus sold for about $131,000 per key in January 2025 (27). Operators in Boise report hospital-driven hotels running above 80% against a 70% market, mainly on traveling clinicians on extended stays (25). Hospital demand is bankable for occupancy and should be underwritten at patient and family rates, not at transient rates.

Universities. One peer-reviewed study across 27 university-dominated cities found lodging demand more stable than the national average, with occupancy and ADR below it; university-town ADR ran at 76.4% to 87.3% of the U.S. level (22). Move-in weekend takes Ann Arbor to 90.1% occupancy against a 69.3% seven-month average, but lifts ADR only from $88.70 to $94.01; Chapel Hill shows the same shape, about 20 points of occupancy for 6% to 8% of rate (20). The cautionary case is a set of three university conference-center feasibility studies that projected the university would supply more than half of occupancy; after opening, it supplied less than a quarter (23). No published study gives room nights per enrolled student, and none documents hotel performance after an enrollment decline or campus closure.

Military installations. Documented through hotelier interviews, not segmentation data. Columbus, Georgia operators attribute 55% to 65% of demand to Fort Benning; Fort Gordon is credited with up to 70% of Augusta demand (24). The rate ceiling is the federal per diem, which the General Services Administration held flat for fiscal 2026 at a $110 standard lodging rate for the first time in five years (12). Military demand is weekday base occupancy with graduation-weekend spikes and a hard rate cap. There is no published series of hotel performance after a base realignment, which should be read as an evidence gap rather than evidence of resilience.

Energy and construction crews. The best boom-and-bust evidence in the literature, and the generator a lender should fear most. Oil and gas tracts accounted for 25% of new U.S. hotel supply from 2010 to 2015 (30). Williston, North Dakota is the canonical case and is covered in its own section below.

Casinos. The only measured case is Atlantic City. When four casinos closed in 2014, non-casino lodging rose from 24% to 29% of Atlantic County rooms, and to 33% after a fifth closure in 2016; the displaced business stayed largely inside the surviving casino hotels, whose per-property performance improved (38). A Federal Reserve working paper using foot-traffic data finds large and permanent spillovers from tribal casinos to nearby restaurants and hotels (39). Comp-room displacement of non-casino hotels has never been quantified in a public source.

Interstate interchanges and drive-to demand. The weakest evidence of any generator that appears in nearly every rural feasibility study. No published equation converts annual average daily traffic into room nights, and no published capture rate exists for highway hotels. The bypass literature, from Texas and Kansas, finds small but statistically significant declines in business volume in bypassed towns, with motel owners reporting major effects, but none of it isolates occupancy or ADR (40). Drive-to markets recover first in a shock and plateau early. In 2025, RevPAR in the Top 25 markets fell 0.6% while all other markets rose 0.1% (52). In the second quarter of 2026, occupancy in every location type was still below 2019, with resorts the furthest behind at 5.4 points (5). Traffic counts in a study are context for a transient share, not a demand model.

Large single employers. Office utilization correlates with full-service urban occupancy, but there is no published per-employee ratio, and no published hotel-performance follow-up to any of the large plant closures of the last three years.

The pattern across the ranking is consistent. Generators that create scarcity on specific nights (conventions, concerts, football, tournaments) are measured and lift rate. Generators that create steady weekday base demand (hospitals, universities, military, crews) are documented through operator interviews and fill rooms at capped or discounted rates. A study that assigns a per-bed or per-student ratio is reporting an analyst's judgment, and should say so.

Three demand types deserve separate treatment because they trade rate for stability, and because they are disproportionately represented in the hotels that SBA and USDA lenders finance.

Contract business is a block of rooms sold to one company for an extended period at a significantly discounted rate: airline crews at airport hotels, construction and utility crews in project markets, permanent guests in soft markets. The REIT filings are candid that contract rates "may be utilized by hotels that are in markets that are experiencing consistently lower levels of demand" (42). The discount is deep. Luxury and upper-upscale airport hotels sold contract rooms at $111.15 in 2018, below the national upper-midscale average at the time (42). The underwriting consequence is that contract rooms committed at discount cannot be yielded on compression nights, so a contract-heavy hotel earns less rate precisely when the market would pay the most. Rating agencies treat a large contract account as a major-account concentration and credit it only to the extent it is durable (63).

Government demand is the per diem rate floor, and in 2025 it fell out from under the select-service segment. Government per diem room nights were down 9% nationally through April 2025, with declines in 76% of 334 markets, and 30-day forward government bookings in the Washington region down 44% (13). By summer the national decline was 11% and the Washington decline 26% (13). Marriott told investors that two-thirds of its government revenue sits in select service and that government is about 4% of regional room nights, which was enough to turn its U.S. select-service and extended-stay RevPAR negative in the second quarter of 2025 (73). A feasibility study for a hotel near a federal installation should show the government share explicitly and run the projection with per diem flat, because that is the current policy.

Extended stay is the segment where the trade is most visible and most favorable. Extended-stay hotels ran 77.8% occupancy in August 2026 against 65.5% for comparable classes, a 12.3-point premium that has held between roughly 10 and 14 points for years and peaked at 18.3 points in the fourth quarter of 2020 (49). In March 2020, when industry RevPAR fell 49%, economy extended-stay RevPAR fell 10% to 15% (49). The cost of that stability is rate: extended-stay ADR was $122.17 in August 2026 against a national $161.78 (49, 3). The risk is supply. Extended-stay room count reached about 645,000 at mid-2026, up 4.6% in a year, the largest gain in six years, and roughly 40% of the entire U.S. construction pipeline is extended-stay product (49, 50). In 2025 extended-stay demand rose and RevPAR still fell 2.2%, because supply grew faster (49). A lender should credit extended stay with a 10 to 14 point occupancy premium and then stress the submarket's extended-stay pipeline specifically, since that is the one segment where supply growth runs an order of magnitude above the national rate.

Generator analysis describes who comes to a market. The macro overlay describes whether they can afford to.

Labor. Payrolls rose by 29,000 in September 2026 against a consensus of 84,000, with the prior two months revised down by a combined 60,000; unemployment was 4.2% and average hourly earnings rose 3.0%, the slowest since 2021 (9). Wages growing at 3.0% against inflation of 3.4% squeeze exactly the households that fill economy and midscale hotels (9, 8).

Fuel. The 2026 conflict in the Middle East pushed regular gasoline to a $4.48 monthly average in May; it was still $4.35 in early October, $1.23 above a year earlier, with the West Coast above $5.70 (10). The Energy Information Administration's annual forecast is $3.91 for 2026 against $3.10 in 2025 (10). No primary source has quantified the effect on interstate and drive-to hotel demand. The inference is that it is one reason economy is the only chain scale still losing demand, but it is an inference, not a measurement.

International inbound. Total arrivals fell 5.5% to 68.3 million in 2025, with Canadian arrivals down 20.9% (11). Canadian return trips from the United States fell for a thirteenth consecutive month in January 2026 before turning positive on base effects in the summer (74). Overseas arrivals were still down 7.0% in July 2026, and the World Cup did not produce the expected inbound lift (11, 2). Inbound spending fell 2.4% to $175 billion in 2025, and the industry forecast of $178 billion for 2026 is running ahead of the year-to-date results (70). For most SBA and USDA hotel markets this is a second-order variable. For gateway and border markets it is the first.

Events. The World Cup beat forecasts on rate and missed on volume. Host-market RevPAR rose 18.7% in the first tournament fortnight while host-market demand fell 1.1%, and three host cities lost a combined 13.5% of demand as room-block cancellations and displaced business travel outweighed visiting fans (72). New York's final-week ADR rose 24% to $351.18 (4). CBRE cut its estimate of the tournament's contribution to annual RevPAR to half a point to eight-tenths of a point (6). The lesson for underwriting carries beyond 2026: an event in a full market generates revenue through price and displaces other demand, while an event in a soft market generates net new room nights. The same event is worth different things to different hotels.

Consumer bifurcation. The forecast consensus for 2026 runs from +2.5% RevPAR (CBRE) and +2.9% (PwC) to +4.4% (CoStar and Tourism Economics), and the spread is almost entirely about rate at the top of the market (6, 7, 2). All three agree that 2027 is a low-single-digit year with hard June and July comparisons.

National supply is the one variable working in favor of every incumbent hotel. Net supply growth was 0.6% in 2025 and is forecast at 0.4% for 2026 and 0.6% for 2027, against a long-term average of 1.6% and a 2018 peak near 2% (2, 71, 52). Rooms under construction have sat near 140,000 for a year, and the March 2026 count of 136,990 was the fifteenth consecutive monthly decline (51). The full pipeline is near a record at roughly 767,000 rooms, but only 19% of it is in construction, the lowest share in twelve years (2). Elevated financing costs and construction inflation have stretched the typical opening to three years after a project enters the pipeline, and that horizon keeps extending (71).

The composition of what is being built matters more than the total. Lodging Econometrics counts 5,975 projects and 703,001 rooms in the U.S. pipeline at mid-2026, of which 1,081 projects and 133,216 rooms are under construction; upper midscale and upscale together are 59% of projects (50). In the MMCG national series, roughly 84,000 of the 145,000 rooms under construction in August sat in the upscale, upper-midscale and midscale classes, and that concentration has held for a decade (71). Luxury has the largest in-construction share of its existing base, at 4.5%, and economy the smallest, at 0.7% (51). Conversions now outnumber new builds: the conversion pipeline reached a record 1,567 projects and 152,044 rooms, more projects than are under construction from the ground up, and nearly 1,900 hotels converted brands in 2025 (50, 2).

Two findings from the academic literature govern how a study should treat a specific competing project.

First, impact is local, same-tier and especially same-brand. The Cornell study of 1,315 chain-affiliated Texas hotels found that a new same-brand hotel among an incumbent's ten nearest competitors cost it $47 to $66 per available room per quarter, about 2.7% of revenue at the upper estimate, against about $36 for a new hotel of another brand in the same tier (53). An FTC working paper confirms large within-brand cannibalization that brand proliferation attenuates but does not remove (54). The brand premium itself is eroding: the revenue advantage of chain affiliation over equivalent independents fell by more than half between 2000 and 2015, most of it attributable to online review platforms (60). CBRE finds soft-brand rooms growing at nearly ten times the rate of traditional brands, and that the brand family growing fastest by brand count posted the slowest median RevPAR growth, at 0.3% a year (59). More flags in a submarket do not mean more demand.

Second, new hotels ramp faster than the three-year convention. Across 3,494 U.S. new entrants from 2006 to 2009, new hotels opened with ADRs above incumbents and reached occupancy parity with their competitive set in seven quarters; branded, managed hotels reached RevPAR parity by the first quarter of year two (55). STR's own conference analysis puts a new-build hotel at 58% occupancy in its first month and at a 100% RevPAR index by month 17, with New York an outlier at month 35 (56). A separate Cornell study measured occupancy stabilization, defined as the point where occupancy stops rising by at least a point a year, at 3.08 years on average (55). The two are consistent: a new hotel matches its comp set in a year and a half, then keeps edging up into year three. A feasibility study that shows parity before month 17 is aggressive; one that shows a three-year build is conservative and defensible.

Single-generator concentration is the risk every rating agency names and no public formula quantifies. The best way to size it is the one market where the full cycle has been measured.

Williston, North Dakota had 7 hotels and 607 rooms in January 2010. By 2014 it had 23 hotels and 1,967 rooms, and by 2016 more than 2,300 rooms across 25 properties, most of them built against oil-field crew demand at contract rates (32). During the boom one Williston hotel ran 100% occupancy at about $120 a night (32). In January 2016 city-wide occupancy was 27%, against 62% a year earlier; ADR fell from $117 to $85 between June 2015 and June 2016; 22 boom-era hotels were running at 20% to 25% occupancy with about 1,800 rooms empty on an average night, and two hotels closed (31, 32). The rating agencies had seen it coming. Morningstar DBRS flagged a 248-key Williston extended-stay loan in August 2015 as dependent "primarily on demand generated directly by the oil and gas industry" (33). By December 2016 about two dozen CMBS loans on North Dakota properties were delinquent, including a Hampton Inn in Williston and a Hyatt House in Minot (34). In the Permian, an eight-hotel energy-corridor portfolio filed for Chapter 11 in July 2015, describing its business as "closely tied to the oil and gas labor market," and the Midland and Odessa metros carried the two highest concentrations of mining and construction payrolls of any U.S. metro while adding more than 2,700 hotel rooms since 2010 (75, 69). Provision expenses at banks in energy states peaked in 2016 (68).

The Williston numbers give a lender a usable downside case: a 35-point occupancy loss and a 27% ADR loss inside twelve months, in a market where supply had tripled into the boom. That is the stress to apply wherever the dominant generator is temporary or commodity-linked.

The same cycle is running now in Louisiana. Lake Charles hoteliers reported ADR up 20% and occupancy up 25% during the 2014 to 2015 petrochemical buildout, with the local visitors bureau predicting occupancy would reach 100% "when the boom is in full swing" and a parish planning board member warning that building for peak construction creates "a 10- to 20-year skewing of the market" (35). The federal housing analysis for the market later attributed slower job growth to the completion of two large facilities after 2018 (36). The second cycle began in 2025: a 6,500-job, 56-month LNG build and a 7,500-job, 30-month export terminal, with regional construction demand projected to peak in late 2026 (37). One of the three planned terminals was suspended in December 2025. Northeast Louisiana's data-center construction is projected to peak near 10,000 workers in early 2027, and data centers employ a few hundred people once built. There is no published hotel series yet for any of these markets, and there will not be one until it is too late to matter for a loan being underwritten today. The Williston series is the proxy.

The pattern generalizes beyond energy. A hotel whose largest account is a single construction project, a single military installation under realignment review, a single plant, or a single university's conference calendar carries concentration that the market will not price until it fails. The feasibility study is the only document in the credit file that can quantify it in advance, and it does so by reporting the share of room nights tied to the top generator and the top accounts, and running the projection with that share removed.

Everything above is input. What a lender receives is a projected occupancy and ADR for each of the first five operating years, and the method that connects the two is well established in appraisal practice even if few studies follow it with discipline.

Segment the accommodated demand. The competitive set's occupied room nights, taken from a STR or CoStar trend report, are divided into commercial, meeting and group, leisure, and, where material, contract, using management interviews and whatever segmentation data the brands hold (57). Each segment is then projected at its own growth rate, tied to the generators that feed it. A single blended growth rate is the first sign of a weak study.

Count the compression nights. With 24 to 36 months of daily comp-set data, the study counts nights above 80% and above 90% market occupancy, by day of week and month, and attributes them to recurring weekday corporate peaks or to a few events. Recurring peaks are bankable; a calendar of six tournament weekends is not. This is also where unaccommodated demand is quantified: the shortfall between demand on sellout nights and available capacity, cross-checked against rate spikes (14, 57). HVS studies have classified as much as 8.0% of base-year demand as unaccommodated where fill nights are frequent, and zero where the daily data show none (57). The study should state the sellout threshold it used and justify it. A market that only sells out in July does not support a year-round hotel.

Add induced demand only with a mechanism. Induced demand is room nights that will come to the market because of a new generator or because of the subject itself: a brand absent from the market, meeting space the market lacks, an extended-stay product where crews are currently housed in transient rooms (57). The nine-case convention-hotel evidence shows it is real; the convention-center projection record shows it is routinely doubled (15, 58). Rebranding an independent lifts revenue per room by about 4%, with roughly 60% of the gain attributable to the brand, which is a measured induced effect (59). A study that adds induced demand should cite the comparable project that produced it.

Allocate through fair share and penetration, by segment. Fair share is the subject's rooms divided by competitive-set rooms. Penetration is the share it actually captures divided by fair share, measured separately for each segment against competitors' historical penetration (57). The flaw in simple penetration is that fair share changes the moment new supply opens, which is exactly when impact analysis matters, so the better practice uses segment-level competitive indices held constant unless something property-specific changes (57). Every in-construction and final-planning project in the competitive set belongs in the denominator, weighted most heavily where it is the same brand family.

Show the ramp. Occupancy parity with the comp set at 17 to 21 months for branded select service, longer for independents and gateway markets; occupancy still edging up into year three (55, 56). ADR for a new build enters above the comp set and holds there if the product justifies it (55).

Reconcile ADR to the distribution, not the average. Project rate from the segment mix and the compression-night count. A hotel that will fill with contract and government business earns a rate ceiling, and the study should say what it is. A hotel in a market with frequent weekday compression earns a rate premium, and the study should show the nights that produce it.

Run the concentration scenario. Report the share of projected room nights tied to the single largest generator and the three largest accounts. Then remove the largest generator and show the DSCR. The Williston case sets the magnitude; the study sets the probability.

A study built this way produces a projection a credit officer can test line by line. It also produces one that an SBA, USDA, rating-agency or bank-examiner review will recognize, because those reviews are built on the same questions.

The four channels through which a hotel loan is reviewed ask the same thing in different language: is demand diversified, segmented, reconciled to the competitive set and robust to new supply, and if not, how much of the projected cash flow should be credited?

SBA under SOP 50 10 8.1. The standard operating procedure effective October 1, 2026 does not impose a blanket feasibility-study requirement on hotel loans (61). Hotels, motels and other lodging facilities appear on the SBA's limited or special purpose property list, which governs the borrower's equity contribution on a 504 loan rather than mandating a study (61). The study provisions sit in the 504 chapter, where SBA reserves the authority to request one "when it is needed to further understand the small business type and market conditions at the project location," and lists five circumstances that may trigger the request: market saturation by industry type and location, a unique market concept, a highly specialized project property, project size disproportionate to the community it will serve, and rapid growth of the applicant group with rising unseasoned debt (61). The 7(a) chapters do not mention feasibility; the general authority is 13 CFR 120.160(b), and in practice lender credit policy drives most hotel study requests (61). The consequence for a hotel study supporting an SBA file is that it should answer the saturation and sizing questions directly: how many rooms the market absorbs, how many are in the pipeline, and whether the proposed room count fits the community's demand. SOP 50 10 8.1 also requires 504 projects with total costs of at least $5,000,000 to submit the third-party lender's credit memorandum, which puts the lender's own demand analysis in front of SBA alongside the study (61).

USDA under 7 CFR Part 5001. The USDA rule is prescriptive where SBA is permissive. For Business and Industry guaranteed loans above $1,000,000 to a new business, "a feasibility study prepared by an independent qualified consultant acceptable to the Agency is required" (62). New business includes an existing operator opening in a new market or labor area, which captures most hotel developments. Community Facilities carries a parallel mandatory trigger. The study's content is fixed by Appendix A to Subpart D, which requires executive summary, economic, market, technical, financial, management, recommendation and qualifications sections; the market component calls for analysis of current and future market potential, competition and sales or service estimations, and the financial component includes sensitivity analysis (62). The lender, in turn, must include "a written evaluation of the feasibility study" in its credit evaluation, and projected increases in revenue or margins must be reasonable and substantiated (62). For a USDA hotel, the segmented demand build-up, the pipeline and the concentration scenario described above are not best practice; they are the content of the market section the regulation requires.

CMBS rating agencies. None of the four agencies publishes a mechanical penalty for a single employer, base or university. Concentration is priced through the cash-flow haircut and the capitalization rate. Morningstar DBRS imposes an occupancy cap "that typically falls anywhere from 75% to 90%, depending on the property's location and respective market dynamics," and states that new supply and "potential changes to major accounts at the property" influence the occupancy it will credit (63). In its rating releases the agency has placed an extended-stay portfolio concentrated in tertiary and suburban markets at a 10.71% capitalization rate, the higher end of its lodging range, against 8.50% for a Manhattan hotel with diversified tourist and commercial demand (63). In its 2020 analysis it stressed hotel values by 25% to 45% depending on demand segmentation and location, with corporate and fly-to demand at the top of the range (63). KBRA capitalizes its own stressed cash flow at a long-term, asset-specific rate; in one 2024 conduit with 14% lodging exposure its values came in 37.0% below the appraisals (64). Fitch's 2016 commentary is the clearest statement of the principle: "In single demand driven markets, the real estate cycle will be more volatile; it may also be shorter" (65). A feasibility study that quantifies the top-generator share and runs the loss scenario is anticipating the exact adjustment the agencies will make.

Bank examiners. The OCC's Commercial Real Estate Lending handbook instructs that "feasibility studies commissioned by the borrower may be biased and should be critically reviewed," that the bank "should conduct its own analysis of the project," that market analysis should consider "existing and anticipated comparable properties," and that "supply often overshoots the expected demand resulting in prolonged lease up" (66). It singles out hotels: "Successful hotel lending requires specialized knowledge," with demand affected by local business activity, consumer spending, travel costs and the dollar (66). The FDIC's examination modules direct examiners to assess "current and projected vacancy and absorption rates," the assumptions in "feasibility studies and market surveys," and the time to stabilized occupancy, and to adjust values when management's assumptions are "irrelevant or inappropriate" (67). The examiner's red flags map one to one onto the weaknesses this framework is designed to prevent: an unsupported ramp, ignored supply, a single generator, stale data, no reconciliation to the comp set, and borrower optimism.

Five conclusions follow for any hotel study written in the fourth quarter of 2026.

Rate growth should be underwritten below the headline. Real ADR is still below 2019, the 2026 recovery is concentrated in luxury and in two event months, and the 2027 consensus is low single digits with hard comparisons in June and July (1, 2, 6). For select-service and economy product in secondary markets, inflation-level ADR growth is the base case and flat nominal rate is the sensitivity.

Occupancy should be underwritten from the distribution of nights, not the average. A study that reports annual occupancy without the count of compression nights, the day-of-week shape and the segment mix has not described the market's pricing power.

Generator ratios should be labeled as judgment. No published standard converts hospital beds, students, soldiers or traffic counts into room nights. The analyst's bottom-up estimate from interviews and generator records is the right method, and it should be presented as that, with the university conference-center case as the reminder of how far institutional demand can be overcounted (23).

Supply should be analyzed at the submarket, by brand family. The national 0.4% figure is context. The specific projects in the competitive set, their tier and their flag are what move the subject's fair share and penetration (53, 54).

Concentration should be quantified and stressed. The share of room nights from the largest generator, the loss-of-generator DSCR, and the extended-stay pipeline in crew and project markets are the three numbers that distinguish a study that will survive the Williston cycle from one that will not.

The purpose of a feasibility study in a hotel credit file is not to confirm the sponsor's pro forma. It is to tell the lender which of the two numbers on the first page is bought with generators, which is bought with scarcity, and what each is worth when the market is tested. That is the analysis MMCG Invest delivers in every hotel study prepared for SBA, USDA and conventional lenders.

What is the difference between a demand generator and a demand driver in a hotel feasibility study?

A demand generator is a specific source of room nights in a market: a hospital, a convention center, a military base, an interstate interchange. A demand driver is the broader force that determines how much business those generators produce and what rate it pays: employment, income, fuel prices, air capacity, consumer confidence, supply. A study identifies the generators to build occupancy by segment and applies the drivers to project growth and rate.

Which demand generators lift ADR, and which only lift occupancy?

Generators that create scarcity on specific nights lift ADR: conventions, concerts, college football, tournaments. Generators that create steady weekday base demand lift occupancy at capped or discounted rates: hospitals, universities, military installations, construction crews, airline contracts and government travel. The measured threshold at which rate accelerates is roughly 75% to 80% market occupancy (14).

How much occupancy should a feasibility study credit to a hospital or a university?

There is no published ratio of room nights per hospital bed or per student, so the credible method is a bottom-up estimate from interviews with the institution and the existing hotels, cross-checked against the comp set's segmentation. Market-level evidence shows that even a world-class medical campus supports low-60s occupancy once supply responds, and that university conference-center studies have projected more than twice the institutional demand that materialized (26, 23).

Does SBA require a feasibility study for a hotel loan?

Not automatically. SOP 50 10 8.1, effective October 1, 2026, lets SBA request a study on 504 projects where it sees market saturation, a unique concept, a highly specialized property, a project disproportionate to its community, or rapid applicant growth (61). Most hotel studies in SBA files are required by the lender's own credit policy, and the study should answer the saturation and sizing questions SBA names.

Does USDA require a feasibility study for a hotel loan?

Yes, for Business and Industry guaranteed loans above $1,000,000 to a new business, which includes an existing operator entering a new market. The study must follow Appendix A to Subpart D of 7 CFR Part 5001, and the lender must file a written evaluation of it (62).

How should a lender underwrite a hotel whose demand depends on one employer or project?

Quantify the share of room nights tied to that generator, remove it, and test debt service coverage. The Williston, North Dakota record shows a 35-point occupancy loss and a 27% ADR loss within twelve months when the generator stopped, and rating agencies price the same risk through occupancy caps and higher capitalization rates (31, 32, 63).

How long does a new hotel take to stabilize?

Branded select-service new builds reach occupancy parity with their competitive set in about seven quarters and a 100% RevPAR index by month 17 to 21; absolute occupancy keeps rising into year three (55, 56). Independents and gateway-market hotels take longer. A projection showing parity before month 17 is aggressive.

Is 2026 hotel performance a good base for a 2027 projection?

No. The 2026 gain is concentrated in luxury and in the World Cup and July 4 months, and the forecasts for 2027 are +2.1% RevPAR with ADR up 1.6% and difficult June and July comparisons (2). A study should underwrite 2027 from the 2027 consensus, not from the 2026 run rate, and should treat real ADR, which is still below 2019, as the relevant measure of pricing power (1, 8).