Why Hotel PE Deals Fail and How Winners Find Profit

So here's what's been bugging me. Hotels reprice their rooms every single night. That's the most flexible asset in real estate. And yet hotel PE deals seem to blow up more often than office towers. Those two things can't both be true.

They can, actually, and the contradiction is the whole story. Preqin counted seventy-eight billion dollars raised for hotel assets in 2023 alone. And a 2021 McKinsey study found sixty-eight percent of hotel acquisitions missed their projected returns by more than two percentage points.

Sixty-eight.

Sixty-eight. Flexible pricing didn't save them. Because flexibility cuts both ways.

Meaning the hotel can charge up tonight...

And it can charge down tonight. An office lease locks in rent for years. A hotel re-leases the entire building, every night, at whatever the market says.

So the nightly repricing that sponsors sell as an asset—

Is the exact mechanism that exposes the deal. If your assumptions about that nightly market are wrong, the error shows up in the cash flow within days.

And that's what you'd call the Operating Assumption Gap?

That's the name for it, yes. It's the quiet space between what the underwriting model assumed — occupancy, room rate, capex per room — and what the hotel actually does.

So it's not fraud, it's not even sloppiness necessarily. It's a, a gap that just sits there.

Right. The Journal of Real Estate Literature lifecycle analysis estimates operating cash flow makes up thirty to seventy percent of a hotel's total holding-period return. So the assumptions driving that cash flow aren't a side detail. They're most of the deal.

Give me a concrete case where this played out.

Think of Blackstone buying Hilton in 2007. The Oxford case study reports it was financed with seventy-eight and a half percent debt, at twelve point four times EBITDA.

Twelve times. So there's no cushion for any gap between model and reality.

Almost none. And then the recession hit. Average global occupancy fell about twelve percent across the pandemic years later, but 2008 was its own shock — and that leverage left zero room for assumptions to be wrong.

So Hilton's the failure case?

That's the twist — it isn't. The case reports Hilton's management-and-franchise segment grew adjusted EBITDA twenty-five percent from 2007 to 2012. Operating improvement partly rescued a leverage mistake.

Huh. So the operating business is both the risk and the rescue.

Exactly. And that's the, the thing generalist PE keeps missing. A 2024 Journal of Financial Economics study compared thousands of matched hotels.

Matched how?

Staggered comparisons of comparable properties, plus repeat-sales regressions. Hotels bought by PE specialists ran net income margins five point four percentage points higher than the same hotels under generalist owners.

Five point four — and, sorry, what does that mean in plain terms?

For every hundred dollars of revenue, the specialists kept about five more dollars than the generalists did. Same buildings, different operators.

Did they earn it on revenue growth, then?

No — and this surprised me. The margin edge came from cost savings without compromising revenues. And capital gains over the holding period ran roughly sixteen percentage points ahead of generalists.

So general real-estate skill — buying well, financing well — that wasn't the driver?

The study is explicit about that. Generic real-estate expertise alone didn't explain it. Hotel-specific operating know-how was the first-order driver.

Which loops back to the gap. If a sponsor doesn't know hotels, the assumptions in the model—

Were written by someone who doesn't know what the building can actually do. And it's worse early on. The risk analysis finds operating risk is typically highest in a property's early years, before stabilized trading.

Which is exactly when value-add PE steps in. The 2025 CoStar piece on value-add strategies treats renovation and repositioning requirements as the risk source, not the fix.

Mm. So a limited partner reading a deal memo should ask — does this sponsor know this asset class, or just this asset?

And are there early warning signs in the assumptions themselves?

A few. For example, a study of over three thousand hotels found operating-profit volatility differs by brand affiliation, class, location, age, size. If the model treats every property as average, that's the gap, pre-loaded.

Wait — brand affiliation changed volatility?

A separate study of four thousand-plus hotels found brand-affiliated properties had steadier key metrics than independents, and the stabilizing effect was stronger for profit than for revenue.

So a branded asset can absorb a bad assumption better than the model thinks?

Or an independent one breaks faster than the model thinks. Same gap, opposite sign. Sorry. The point is the gap isn't one mistake — it's structural.

the model assumes an operating business the sponsor doesn't actually know how to run.

That's it. Hilton's leverage nearly killed it; operating expertise pulled it back. The sixty-eight percent who missed their numbers mostly lacked that expertise — and never saw the gap until the rooms stopped selling.

So if a sponsor hands over a pro forma showing a fifteen percent return, what actually breaks if I just believe it?

Everything after page one. A pro forma is a forecast, not proof. Valuation rests on estimates of future revenue, expenses, and sale proceeds.

Okay, but which estimate is the one that kills the deal?

Let's treat that as the investigation. Three suspects — occupancy, rate, and expenses. One of them is usually holding the knife.

Fine, detective. Walk me through the suspects.

Occupancy is rooms sold divided by rooms available. ADR is room revenue divided by rooms sold — the average rate on sold rooms.

So they measure different things. Which one moves the money?

Neither alone — that's why RevPAR exists. It's room revenue over rooms available, so it captures rate and volume together.

So RevPAR is the number to trust?

Careful. RevPAR is room revenue. It ignores food and beverage, other departments, expenses, and capital needs entirely.

So a hotel can look busy on RevPAR and still bleed cash.

Exactly. And here's the mirage part — HVS argues forecasts should be grounded in the property's own history and where it sits in its economic life cycle.

Rather than extrapolating one great year, you mean.

Yes. Take one unusually strong period, stretch it across ten years, and the model looks brilliant on paper. It's, it's a story, not a forecast.

How would I check whether the projection is grounded, then?

The Counselors of Real Estate framework splits it into steps — market occupancy, then the subject's competitive position, then projected occupancy.

Wait — the hotel's own position is a separate step from the market?

It has to be. CoStar's RevPAR Index compares a hotel against its competitive set, and one hundred means parity. If your hotel sits at eighty, the market forecast isn't your forecast.

So a rising market with a hotel that keeps losing share — that's a mirage right there.

That's a good one. CRE says you build a stabilized operating profile, sorry, before converting it into the multiyear projection of revenue, expenses, capital needs, and NOI.

Now the expense side. My instinct says revenue assumptions are where deals die.

Build the opposing case properly — revenue gets all the attention, and sponsors do love the top line. But HVS points out costs have different fixed and variable relationships.

Meaning if revenue drops, not everything drops with it.

Right. Treat every cost as moving in lockstep with revenue, and your projected margins are fiction the moment revenue stumbles.

Can you give me a concrete example of an expense that gets quietly mishandled?

The FF&E reserve — money set aside for furniture, fixtures, and equipment. Omit it and you've overstated distributable cash. RLJ Lodging Trust reported its reserve requirements generally run three to five percent of gross revenue.

Three to five percent of revenue. Every year.

And even then, HVS notes the deposit isn't automatically the hotel's total future capital need. RLJ held thirty‑one point nine million in those reserve accounts.

So the reserve is a floor, not a ceiling. Is there better capex data than a guess?

The 2023 ISHC capex study — Stephen Hennis co‑chaired the committee — covered more than nine hundred hotels and over thirty‑one hundred annual spending data points.

That's the benchmark a pro forma should be tested against.

And CRE says benchmark per‑room renovation costs against similar brands and markets, using the renovation history as an input.

What about the exit? Everyone assumes they sell at the end.

HVS is blunt — terminal value is an assumption, not a guaranteed price. The exit capitalization rate reflects the risk of owning the hotel over the projection, and it materially moves the sale value.

The discounted cash flow is as honest as its inputs.

And you shouldn't rely on a single model output. CRE recommends checking the DCF against direct capitalization, replacement cost, and comparable sales.

One more suspect I want to interrogate — the analyst's own confidence.

Good instinct. A study applied seventeen accuracy measures to over two thousand occupancy forecast instances and found the measures disagreed with each other. Even the scorecard is a risk.

So you can pick the metric that flatters you.

Exactly. And accuracy improves as the horizon shortens — which means the ten‑year column is where confidence goes to die.

So what's the anti‑mirage test, in one breath?

Make every assumption traceable — market, operations, capital, financing, exit — then test how value moves when each one changes.

And the Uniform System of Accounts gives you the comparable format to do it.

It does — the standard statement format, so the subject's data lines up against comparables. So next time you see a fifteen percent return on paper, you'll look for the reserve line. That missing three to five percent of revenue is sitting in plain sight.

So I was staring at a hotel P&L the other day and the thing that jumped out wasn't a number. It was the light switch. Hundreds of rooms, lights on all day. Nobody models a light switch.

Keep that switch in mind, because it's going to come back. But you've hit the real problem — a hotel is real estate and an operating business at the same time. Pricing, staffing, distribution — returns depend on those decisions, not just the building.

So how does an owner actually manage both halves at once?

Think of the discipline CBRE describes for asset managers — benchmark performance, investigate variances, approve capital with a rationale, then check whether results matched the plan. Four moves, repeated.

Benchmark, investigate, approve, verify. But benchmark against what? Everyone quotes RevPAR.

Careful there. RevPAR is ADR times occupancy. HVS points out it misses food and beverage, events, parking, spa. GOPPAR is the broader measure — total revenue minus operating expenses, per available room.

So a busy hotel on RevPAR can still be losing on GOPPAR.

Exactly. And compare against the competitive set — the RevPAR index tells you whether you're capturing your share of room revenue versus nearby rivals. Also review market segments, channels, and what bookings actually cost to generate.

Give me a concrete case where this whole playbook ran end to end.

Waikiki. Host Hotels & Resorts with operator Kokua Hospitality — a forty-five million dollar renovation, repositioning the Ocean Resort Hotel as Hyatt Place Waikiki Beach. Hotel Management reported total revenue up sixty‑eight percent and RevPAR index up eighty‑three percent.

Eighty‑three.

But notice the trap inside that story. A CoStar‑reported case saw its RevPAR index hit its lowest point during the renovation year — rooms out of order, guests avoiding the dust.

So the capital plan has to price in the disruption, not just the payoff. What else trips owners up?

The management agreement. Base fee usually a percentage of gross operating profit or total revenue, plus an incentive fee often tied to RevPAR or GOP‑margin performance against competitors. Those terms shape every incentive the operator has.

So if the incentive fee keys on RevPAR, the operator chases rooms and ignores—

Ignores the cost of winning them. That's why CBRE splits costs above and below the GOP line — managers have more influence above it, while below it costs are more fixed and ownership‑controlled.

Wait — back to my light switch. Does anyone actually price that?

ENERGY STAR says lighting is almost a quarter of electricity use in a typical hotel, and retrofits can cut lighting electricity by half or more depending on the starting point. Sorry. And a ten percent energy reduction is worth the same as raising ADR by sixty‑two cents at limited‑service hotels.

So a light switch is, in effect, a room‑rate decision. What about pricing itself — is everyone doing airline‑style dynamic pricing?

That was my assumption too, and it's wrong. A European study of over fifty‑two million daily prices found static pricing dominated — about eighty‑one percent of the sample. Systematic lead‑time yield curves were rare.

Eighty‑one percent static. So pricing discipline is an actual lever most owners rarely pull.

One paper tested a dynamic‑pricing method on hotel data and reported roughly six percent higher average revenue than the fixed‑price policy — in computer experiments, mind, not a guaranteed real‑world uplift.

Hedged, but real. Okay — so what evidence would actually convince me a sponsor can run this playbook?

Post‑project numbers that match the plan. Go find the renovation they finished three years ago and ask whether the promised GOPPAR and RevPAR index showed up on schedule.

And if the answers are vague, the building might be fine but the operator is the gap.

That's the whole playbook in one sentence. The switch, the fee, the plan, the check — four pillars, checked relentlessly.