Airbnb &

Hotel Performance
An analysis of proprietary data in 13 global markets

Analyzed by:
Jessica Haywood
Patrick Mayock
Jan Freitag
Kwabena Akuffo Owoo
Blase Fiorilla
 Table of Contents

Executive Summary.............................................................................................. 3

Methodology and Data Sets.................................................................................. 6

The Data Dilemma................................................................................................ 7

The Accommodations Landscape.......................................................................... 8

Market Overviews................................................................................................. 16

Airbnb’s Impact on Hotel Compression Nights....................................................... 21

Airbnb’s Impact During Special Events.................................................................. 27

In Closing............................................................................................................. 34

Contact Information.............................................................................................. 35

Credits

Analysis Editing Design

Jessica Haywood Robert McCune Jon Edwards
Patrick Mayock Dan Kubacki
Jan Freitag Sean McCracken
Kwabena Akuffo Owoo Bryan Wroten
Blase Fiorilla Danielle Hess

Airbnb & Hotel Performance 2

 Executive Summary
Hotel performance continued to show strength while more hosts than ever were
renting their residences on Airbnb, according to STR and Airbnb data.

Performance data through July 2016 indicated hotels were following their normal
cyclical trajectory, hovering at or just below the peak at a time when Airbnb
listings outnumbered the world’s largest hotel company by nearly three units to
one. As of July, the U.S. hotel industry had recorded its 77th consecutive month
of revenue-per-available-room growth. During that same month, hoteliers sold
more roomnights (117 million) than ever before.

Those same dynamics largely held true in the following 13 global markets
analyzed in this report: Barcelona, Boston, London, Los Angeles, Mexico
City, Miami, New Orleans, Paris, San Francisco, Seattle, Sydney, Tokyo and
Washington, D.C. Airbnb provided more than two-and-a-half years of daily data
for each market, which STR analyzed and then compared to its hotel performance
data.

Among the most compelling findings:

• Airbnb occupancy generally was the highest in markets where hotels had
high occupancy.
• Hotel occupancy was significantly higher than Airbnb occupancy.
• While Airbnb’s share of total accommodation supply (i.e. Airbnb units and
hotel rooms) was growing, its share of market demand and revenues still was
generally below 4% and 3%, respectively.
• Airbnb guests typically stayed longer than the average hotel guest, with
roughly half of Airbnb roomnights coming from trips of seven days or longer.
• Airbnb’s share of business travel was substantially smaller than its share of
leisure travel.
• Hotel average daily rates generally were higher than Airbnb rates (e.g. $16
higher on average in our seven U.S. markets).
• Hotel ADR increased in all but one market (Paris) in the year ending July
2016. Airbnb rates decreased in eight markets and increased in five.

Supply might have been a contributing factor in that latter point, as the majority
of the markets analyzed in this report saw available Airbnb units increase by
more than 40%—and in some cases north of 100% . When analyzing growth
rates, however, it’s important to acknowledge the baseline. Airbnb is a relatively
new presence in many markets, and growth rates often are commensurate
with untapped potential. In other words, Airbnb has more room to grow in most
markets, whereas hotels have carved out an established presence over decades.

Airbnb & Hotel Performance 3

 The Airbnb demand story followed a similar narrative, with growth rates that
generally were above 60%. Harder to assess was the nature of this demand—
whether it was mostly incremental, as many of Airbnb’s most ardent defenders
have argued, or whether it represented a roomnight that otherwise would have
been spent in a hotel, as many critics have alleged.

While competitive pressures exist across the lodging landscape, it is difficult

to assess the interplay between different accommodation types with different
operating models. Hotel inventory is fixed; a hotel room tonight is a hotel room
tomorrow. Airbnb inventory can flex, as hosts take supply on and off the market
based on their willingness to make privately owned units available to the public for
rent. This analysis represents our attempt to normalize the data set, but it should
not be interpreted as a true “apples-to-apples” comparison.

Figure 1
Compression Nights for 7 US Markets 75 76
71
July YTD
61

43

27

15 16 17 15
13
4

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

That said, one can still glean meaningful insight and information from the data.
A review of compression nights, for instance, showed no noticeable impact
of Airbnb within the seven U.S. markets included in this report (Figure 1). The
number of nights in which occupancy exceeded 95% had increased steadily since
the Great Recession, reaching its peak of 76 during year-to-date July 2015. That
number dipped slightly to 71 through July 2016, which followed a softening of
several other macroeconomic indicators. Also at play was the 1.6% uptick in hotel
supply in those seven markets, which should not be underestimated. This looks to
be a small percentage increase on the surface given the high base of hotel supply
(92,871,685), but the total number of new supply added (1,487,458) to these
markets was actually substantial.

Airbnb & Hotel Performance 4

 What’s more, U.S. hoteliers in those seven markets saw no degradation of their
rate premiums during that same period. ADRs on compression nights during
2016 were 35% higher than on non-compression nights, which represents a
record high.

Compression night performance in the six international markets followed a similar

track, although more volatility was observed in the aggregate given the disparate
economic and political factors at play.

With this analysis, it is our intent to provide the most comprehensive comparison
of Airbnb and hotel performance data ever reported to better inform the
conversations in this “street corner” business. It is not our intent to draw
correlation or causation. We will share more findings as we continue to examine
the available information and as new data becomes available.

Airbnb & Hotel Performance 5

 Methodology and Data Sets
Airbnb provided STR with market-level activity data from 1 December, 2013,
to 31 July, 2016, for the following 13 markets: Barcelona, Boston, London, Los
Angeles, Mexico City, Miami, New Orleans, Paris, San Francisco, Seattle, Sydney,
Tokyo and Washington, D.C. These markets and submarket definitions were based
on zip codes used by STR. Matching based on zip code wasn’t possible in Tokyo,
London and Mexico City, so Airbnb provided data based on how it defines these
markets.

While Airbnb provided a complete set of data for these 13 markets, STR excluded
data we deemed incomparable to hotels to provide a more relevant analysis for
our stakeholders. STR excluded shared and private rooms, limiting its analysis to
entire homes/apartments/etc. We also excluded listings that hold more than seven
people because groups of this size are the least likely to stay in hotel rooms.

Some charts and tables do not include every market due to the fact that either the
Airbnb or hotel data in those segments was insufficient for analysis.

All currency is reported in local currency.

STR was not remunerated in any way for its analysis, and its participation in this
analysis was not contingent upon developing or reporting predetermined results.

Airbnb & Hotel Performance 6

 The Data Dilemma
Airbnb-sourced data is preferable to scraped data, but it still presents challenges.
The Airbnb universe includes accommodations of all shapes and sizes, not to
mention a fundamentally different operating model. Thus, we acknowledge our
analysis is not a true “apples-to-apples” comparison. By applying the following
filters to the data set, however, we ensured the Airbnb performance data we used
provided the most meaningful comparison to our hotel performance data:

1. We removed Airbnb listings that aren’t actually available for rent. (Some
hosts create listings but never actually make them available).
2. We removed shared bedrooms and private rooms with shared living space,
because it is unlikely a typical hotel guest would view such a space as a
viable alternative to a hotel room.
3. We removed large Airbnb listings that hold more than seven people, because
groups of this size are unlikely to stay in hotel rooms.

Within the remaining listings, there was still the “Airbnb experience” factor, which
might include kitchens and other apartment-style amenities, meeting the host,
and other factors unique to the accommodation. For example, a small percentage
of Airbnb listings are eclectic accommodation types such as treehouses, camper
vans, yurts, castles and so forth that were not identifiable in our dataset.

Most reports about Airbnb rely on scraped data. Pulled directly from advertised
listings on the company’s website, scraped data can prove a useful indicator—but
one subject to numerous assumptions, which in turn can lead to unsupported
conclusions. This is why STR prefers to use data directly from Airbnb.

Figure 2

Data Scraping Methodology and Assumptions

Scrape Listings, Scrape Reviews Scrape Base Price Apply Average Stay
Calendars

Assumptions: Assumptions: Assumptions: Assumptions:

All listings shown are bookable Review rates are equal across Published price applies Length of stay is uniform
All calendars are up to date listing and geographies uniformly across listing and
All blocked dates are booked geographies

Source: Airbnb

Airbnb & Hotel Performance 7

 One of the biggest issues with scraped data is the challenge of determining
whether a listing is actually booked or simply unavailable for rent. Without being
able to make this distinction accurately, it is impossible to determine supply and
demand.

To see just how different scraped data is from real Airbnb data, we compared real
Airbnb data from a previous STR analysis on New York City to scraped data. For
the 12 months ending August 2015, the scraped data substantially overestimated
both demand and revenue. This suggests that estimates of demand and revenue
market share that are based on scraped data may indeed be overestimates as
well.

STR has used scraped data in the past, and we believe it can be a useful,
directional indicator of Airbnb’s growth and prices. But just as when
benchmarking hotel performance, source data is far more reliable.

Airbnb & Hotel Performance 8

 The Accommodations Landscape
Unlike the hotel industry, in which descriptive attributes often are drawn from
fixed real estate, Airbnb’s portfolio is more fluid and difficult to categorize.
Listings come in various shapes and sizes, with availability constantly evolving as
hosts take inventory on and off the market.

Despite this elusive quality, Airbnb is not beyond measure. Data provided from
Airbnb on 13 global markets revealed new insights on the platform’s size and
makeup.

As of November 2016, Airbnb reported having just over 3 million listings

worldwide. By raw number of listings, this was nearly three times the size of the
newly combined entity of Marriott International and Starwood Hotels & Resorts
Worldwide (Figure 3).

Figure 3
Largest Lodging Companies by Rooms/Listings

Airbnb 3M

Marriott International 1.1M

Hilton Worldwide 774K

Intercontinental Hotels Group 717K

Wyndham Worldwide 677K

AccorHotels 519K

Choice Hotels International 511K

Best Western Hotels & Resorts 296K

Carlson Hospitality Group 177K

Hyatt Hotels Corporation 168K

But recall that only a portion of Airbnb’s listings are actually comparable to hotel
rooms. To estimate the fraction of hotel-comparable rooms within Airbnb’s global
inventory, we extrapolated based on data provided by Airbnb for 13 markets.

Airbnb & Hotel Performance 9

 Within this dataset, we calculated the fraction of Airbnb supply that comprises: 1)
units not available for rent, 2) shared rooms, 3) private rooms, and/or 4) units that
can host large groups (seven or more people). We then applied this fraction to the
total number of Airbnb listings globally to estimate that only 1.08 million listings
(35% of total Airbnb listings) are effectively available and reasonably competitive
with hotels. This estimate is just under the total number of Marriott rooms. Figure
4 shows this “corrected” view, per our methodology.

Figure 4
Largest Lodging Companies by Rooms/Listings
Less Unavailable, Shared Rooms, Private Rooms and Large Listings

Marriott International 1.1M

Airbnb 1.0M

Hilton Worldwide 774K

Intercontinental Hotels Group 717K

Wyndham Worldwide 677K

AccorHotels 519K

Choice Hotels International 511K

Best Western Hotels & Resorts 296K

Carlson Hospitality Group 177K

Hyatt Hotels Corporation 168K

Demographics differ between Airbnb and hotels

Over the past year, the average age of a guest booking on Airbnb was 35, and
53% of all guests were female, according to Airbnb. According to the AH&LA’s
“Lodging Industry Trends 2015” report, 50% of all hotel travelers were aged 35-
54, and males made up the bulk at 63%.

Airbnb & Hotel Performance 10

 Airbnb guests are primarily leisure-oriented
The alternative accommodations provider in July 2015 launched “Airbnb for
Business” to target more corporate travel. By 2016, Airbnb business travel had
tripled in size, and more than 50,000 companies had used Airbnb for Business,
according to Airbnb. The company estimates that 10% of its bookings are for
business travel, though this may include travelers combining business and leisure
travel.

While STR does not track business travel, an analysis of weekday (traditionally
business-focused) versus weekend (traditionally leisure-focused) travel provides
a good proxy.

Hotel occupancy typically was highest mid-week for the 13 markets, while
Airbnb’s occupancy was highest on the weekend (Figure 5). This supports the
logic that Airbnb is more leisure-focused than hotels. The only exception is San
Francisco, where Airbnb occupancy was the highest on Wednesday and Thursday.

Figure 5

US Occupancy by Day of Week

July 2016, 12-Month Moving Average

81% 82% 82%

76% 78% 79%

67%

52%
48% 49%
43% 42% 45%
42%

Sunday Monday Tuesday Wednesday Thursday Friday Saturday

Hotel Airbnb

US Occupancy by Day of Week, 07/16 12MMA

Airbnb & Hotel Performance 11

 US Rate by Day of Week, 07/16 12MMA
Figure 6

US Rate by Day of Week

July 2016, 12-Month Moving Average

$188 $187
$183
$178
$171 $171 $170 $166
$166
$158 $154 $153 $154 $156

Sunday Monday Tuesday Wednesday Thursday Friday Saturday

Hotels Airbnb

In both hotels and Airbnb, average daily rate was the highest on the days with
the highest occupancy (Figure 6). The largest variance in ADR between hotels
and Airbnb occurred on Tuesday and Wednesday, which makes sense given these
were the most occupied days for hotels and least for Airbnb. The pricing premium
was reduced to $5 on the weekend, when competition for leisure travelers is
arguably most intense. Recall that for the Airbnb data, we included listings that
can accommodate up to six people, which skewed Airbnb rates higher than if we
only considered listings accommodating one or two guests.

Airbnb’s share of weekday demand grew steadily over the past three years and
was highest in Los Angeles, but was still less than 5% of total weekday demand in
every U.S. market we examined (Figure 7). Most cities hovered in the 1-3% range.

Airbnb & Hotel Performance 12

 Figure 7

Airbnb’s Share of Weekday Demand

July YTD

4.7%

3.8%

3.3%
3.1%

2.6% 2.6%

2.0% 2.1% 2.1%

1.8%
1.6% 1.6%

1.1% 1.2% 1.2%

1.0%

0.7% 0.7%
0.5% 0.6%
0.4%

Boston MA Los Angeles/ Miami/ New Orleans LA San Francisco/ Seattle WA Washington DC
Long Beach CA Hialeah FL San Mateo CA MD-VA

2014 2015 2016

While Airbnb’s share of weekday demand was growing, hotels in most markets
still saw weekday demand growth. Boston and New Orleans were the only markets
that saw a slight decline in hotel weekday demand year-over-year.

Additionally, weekday ADR was up in all markets except Miami. Four markets
(Barcelona, Tokyo, Los Angeles and San Francisco) saw hotel weekday ADR
growth of 5-10%.

Airbnb & Hotel Performance 13

 Guests stay longer with Airbnb
Figure 8 shows the percentage of Airbnb roomnights sold by trip length for seven
U.S. markets.
Figure 8

Trip Length Distribution for 7 US Markets

July 2016, 12-Month Moving Average

17.3%

53.5%
29.2%

1-6 days 7-29 days 30+ days

Please note: This chart does not show the percentage of total trips booked. Rather, it’s the
percentage of total roomnights sold. Of all Airbnb roomnights sold, 29.2% were part of a trip that
lasted between seven and 29 days, while 17.3% were part of a trip that lasted a month or longer.
Combined, 46.5% of all Airbnb roomnights sold were part of a “long-term stay.”

STR currently does not track length of stay. However, we do have an extended-
stay indicator that applies to approximately 30 brands in the U.S. that focus on
attracting hotel guests for an extended amount of time. In the seven U.S. markets
included in this report, 9% of all roomnights sold were attributed to extended-stay
brands. This figure does not account for guests that stay for an extended period in
non-extended stay hotels, but it is a good directional indicator.

Thus, the data suggests that travelers who use Airbnb typically have longer
lengths of stay on average than do typical hotel guests.

In Los Angeles, Mexico City and Sydney, more than half of all Airbnb roomnights
sold were part of trips that lasted a week or longer (Figure 9). New Orleans
had the smallest percentage (20%) of roomnights for trips of a week or longer.
Interestingly, hotel occupancy growth in extended-stay properties in New Orleans
was down 8%, which indicates the demand for extended-stay nights declined in
that market.

Airbnb & Hotel Performance 14

 The number of roomnights sold on Airbnb that are part of stays of a week or longer
declined 5% in the 12 months ending July 2016, which indicates that Airbnb
trip lengths in all markets studied are decreasing. Meanwhile, the share of hotel
demand attributable to extended stays has remained relatively stable.

Figure 9
Percent of Airbnb Demand for 7+ Days

Tokyo 35.9%

Sydney 56.9%

Paris 41.8%

Mexico City 56.2%

London 48.3%

Barcelona 39.5%

Washington DC-MD-VA 45.7%

Seattle WA 37.4%

San Francisco CA 47.7%

New Orleans LA 20.3%

Miami FL 49.7%

Los Angeles CA 50.9%

Boston MA 43.0%

The only market that saw the number of Airbnb extended-stay nights grow was
San Francisco. Notably, only 4% of all hotel supply in San Francisco was classified
as extended stay, which is the smallest of any U.S. market studied.

While Airbnb does have a large percentage of extended-stay demand, these hotels
in the U.S. still saw healthy growth. July 2016 year-to-date demand was up 5%,
which was above the U.S. average of 1%. ADR growth for extended-stay hotels
also was higher than the U.S. average at 4%.

Airbnb & Hotel Performance 15

 Market Overviews
While analyzing the data in aggregate helps paint the larger picture, further clarity
is gained by looking at the individual brush strokes on a market-by-market basis.

But first, a mathematical word of caution: Growth rates of many key performance
indicators for Airbnb are much larger than those of hotels. There is a simple
reason for this: Since Airbnb represents a tiny fraction of the inventory of hotels,
Airbnb data starts at a low base—hence, high growth rates.

Tokyo serves as a good example. The market’s Airbnb supply footprint roughly
tripled in the 12 months ending July 2016, a rate that dwarfed hotel supply
growth of 3.2% in the comparable period. Those percentages belie the actual
market dynamics, however. Tokyo’s baseline of 951,459 Airbnb rental nights
listed in the 12 months to July 2015 is much less than the hotel baseline of 38
million available roomnights.

In other words, Airbnb had considerably more room to grow, and each additional
unit listed represented a higher percentage increase than did each additional
hotel room.

Supply and demand

As mentioned previously, Tokyo reported the highest percentage growth, with
both Airbnb supply and demand more than tripling. This outpaced the growth rate
of the second fastest growing market, Mexico City, where supply and demand
both more than doubled.

Airbnb growth rates were muted in its more “mature” markets, such as its home
base San Francisco and also Paris. The larger baseline of units in these markets
softened the supply growth rates.

All markets, with the exception of Barcelona, reported supply growth of more than
30% in the 12 months ending July 2016. All markets reported demand growth of
more than 20%.

Airbnb & Hotel Performance 16

 Generally, Airbnb saw a favorable balance between supply and demand (Figure
10). Demand outpaced supply in all but three markets (Paris, Tokyo and San
Francisco), resulting in healthy occupancy growth.

Figure 10

Airbnb Supply and Demand Growth

July 2016, 12-Month Moving Average

13.1%
Greater Barcelona 36.2%%

62.1%
Greater London 67.9%

107.0%
Mexico City 119.2%

39.6%
Paris
33.0%

53.1%
Sydney 81.3%

252.4%
Tokyo 241.3%

62.7%
Boston MA 70.5%

41.3%
Los Angeles/Long Beach CA 61.5%

52.0%
Miami/Hialeah FL 60.6%

76.3%
New Orleans LA
79.3%

31.9%
San Francisco/San Mateo CA 25.1%

77.9%
Seattle WA 84.8%

53.7%
Washington DC-MD-VA 65.5%

Airbnb Supply PC Airbnb Demand PC

Because hotel supply and demand started at such a high base, year-over-year growth
rates for the 12 months ending July 2016 were much smaller compared to those
recorded for Airbnb.

While Tokyo’s Airbnb growth rates were the largest among the markets we examined,
the city’s hotel growth numbers were not. Miami reported the largest hotel supply
increase (4.1%), and Mexico City reported the largest hotel demand increase (5.5%).

Airbnb & Hotel Performance 17

 Unlike Airbnb, hotels saw a less favorable balance between supply and demand
(Figure 11). Seven markets reported supply increases that outpaced demand,
resulting in negative occupancy growth during the period. Two markets (Paris and
New Orleans), reported hotel demand decreases.

Given that hotel demand decreased and Airbnb demand increased in those two
markets, Airbnb likely did affect hotel demand. However, there are also other
market-level factors at play such as the recent terrorist activity in Paris and large
event shifts in New Orleans that also negatively impacted hotel demand. In all
other markets, both Airbnb and hotel demand increased at the same time, so
there seemed to be no universal relationship at work.

Figure 11

Hotel Supply and Demand Growth

July 2016, 12-Month Moving Average
5.5%

4.1%
3.3%

3.2%

3.2%

3.3%
2.4%

2.5%
2.4%

2.3%
2.3%

2.4%

2.0%

1.8%
1.6%

0.7%
0.7%
0.9%

0.9%
0.5%

0.5%
-9.0%

0.2%

-0.4%
0.0%
0.0%

Greater Greater Mexico Paris Sydney Tokyo Boston Los Angeles/ Miami/ New Orleans San Francisco/ Seattle WA Washington DC-
Barcelona London City MA Long Beach Hialeah LA San Mateo MD-VA
CA FL CA

Hotel Supply PC Hotel Demand PC

Occupancy and ADR

Hoteliers sold significantly more of their available inventory than did Airbnb hosts
during the 12 months ending July 2016.

In San Francisco, for instance, hoteliers sold more than eight out of every 10
hotel rooms available (84.6%), which was third highest among our 13 markets.
Airbnb hosts in that market, meanwhile, sold more than five out of every 10 units
available (51.8%), which also was third highest.

The above example underscores another trend: Airbnb occupancy was highest
in markets in which hotel occupancy was highest, which might suggest Airbnb is
accommodating incremental demand (Figure 12).

Airbnb & Hotel Performance 18