Middlesex University Research Repository
An open access repository of
Middlesex University research
http://eprints.mdx.ac.uk
Quattrone, Giovanni ORCID logoORCID: https://orcid.org/0000-0001-9219-8437, Proserpio,
Davide, Quercia, Daniele, Capra, Licia and Musolesi, Mirco (2016) Who benefits from the
"sharing" economy of Airbnb? Proceedings of the 25th International Conference on World Wide
Web. In: WWW 2016: 25th International Conference on World Wide Web, 11-15 April 2016,
Montreal, Canada. ISBN 9781450341431. [Conference or Workshop Item]
(doi:10.1145/2872427.2874815)
Final accepted version (with author’s formatting)
This version is available at:
https://eprints.mdx.ac.uk/24527/
Copyright:
Middlesex University Research Repository makes the University’s research available electronically.
Copyright and moral rights to this work are retained by the author and/or other copyright owners
unless otherwise stated. The work is supplied on the understanding that any use for commercial gain
is strictly forbidden. A copy may be downloaded for personal, non-commercial, research or study
without prior permission and without charge.
Works, including theses and research projects, may not be reproduced in any format or medium, or
extensive quotations taken from them, or their content changed in any way, without first obtaining
permission in writing from the copyright holder(s). They may not be sold or exploited commercially in
any format or medium without the prior written permission of the copyright holder(s).
Full bibliographic details must be given when referring to, or quoting from full items including the
author’s name, the title of the work, publication details where relevant (place, publisher, date), pag-
ination, and for theses or dissertations the awarding institution, the degree type awarded, and the
date of the award.
If you believe that any material held in the repository infringes copyright law, please contact the
Repository Team at Middlesex University via the following email address:
eprints@mdx.ac.uk
The item will be removed from the repository while any claim is being investigated.
See also repository copyright: re-use policy:
http://eprints.mdx.ac.uk/policies.html#copy
Who Benefits from the “Sharing” Economy of Airbnb?
Giovanni Quattrone
Dept. of Geography
University College London, UK
g.quattrone@cs.ucl.ac.uk
Davide Proserpio
Dept. of Computer Science
Boston University, USA
dproserp@bu.edu
Daniele Quercia
Bell Laboratories
Cambridge, UK
quercia@cantab.net
Licia Capra
Dept. of Computer Science
University College London, UK
l.capra@ucl.ac.uk
Mirco Musolesi
Dept. of Geography
University College London, UK
m.musolesi@ucl.ac.uk
ABSTRACT
Sharing economy platforms have become extremely popular in the
last few years, and they have changed the way in which we com-
mute, travel, and borrow among many other activities. Despite their
popularity among consumers, such companies are poorly regulated.
For example, Airbnb, one of the most successful examples of shar-
ing economy platform, is often criticized by regulators and policy
makers. While, in theory, municipalities should regulate the emer-
gence of Airbnb through evidence-based policy making, in prac-
tice, they engage in a false dichotomy: some municipalities allow
the business without imposing any regulation, while others ban it
altogether. That is because there is no evidence upon which to draft
policies. Here we propose to gather evidence from the Web. Af-
ter crawling Airbnb data for the entire city of London, we find out
where and when Airbnb listings are offered and, by matching such
listing information with census and hotel data, we determine the
socio-economic conditions of the areas that actually benefit from
the hospitality platform. The reality is more nuanced than one
would expect, and it has changed over the years. Airbnb demand
and offering have changed over time, and traditional regulations
have not been able to respond to those changes. That is why, fi-
nally, we rely on our data analysis to envision regulations that are
responsive to real-time demands, contributing to the emerging idea
of “algorithmic regulation”.
Categories and Subject Descriptors
J.4 [Social and Behavioral Science]: Miscellaneous
General Terms
Sharing economy, regulation, policy
1. INTRODUCTION
In the last few years, we have seen the proliferation of sharing
economy platforms. These platforms leverage information technol-
ogy to empower users to share and make use of underutilized goods
Copyright is held by the International World Wide Web Conference Com-
mittee (IW3C2). IW3C2 reserves the right to provide a hyperlink to the
author’s site if the Material is used in electronic media.
WWW 2016, April 11–15, 2016, Montréal, Québec, Canada.
ACM 978-1-4503-4143-1/16/04.
http://dx.doi.org/10.1145/2872427.2874815.
and services. Services covered by the sharing economy range from
transportation to accommodation to finance. One of the most com-
pelling example of the sharing economy is Airbnb, a peer-to-peer
accommodation website. Airbnb defines itself as “A social website
that connects people who have space to spare with those who are
looking for a place to stay”. The company, founded in 2008, grew
exponentially in the past few years, and by now it lists over 1.5 mil-
lion properties, with a presence in over 190 countries and 34,000
cities. By the end of 2014, the company had more than 70M nights
booked.
1
The explosive growth of the sharing economy has led regulatory
and political battles around the world. Proponents of the sharing
economy argue that it will bring many benefits, including extra in-
comes from the users of such services, better resource allocation
and utilization, and new economic activities for cities and munici-
palities.
2
On the other side, detractors argue that the negative exter-
nalities generated by the sharing economy far outpace the benefits.
Most of the critics denounce the sharing economy for being about
economic self-interest rather than sharing, and for being predatory
and exploitative. Indeed, the predatory aspect of such economy
has already seen its first victims: after Uber entered the New York
City market, the price of taxi medallion fell down by about 25%,
3
and in [21] the authors show that Airbnb entry in the state of Texas
negatively impacted hotel revenue.
Because of such negative externalities, the sharing economy and
its regulation have become highly popular policy topics. Many mu-
nicipal governments are attempting to impose old regulations on
these new marketplaces without much thought about whether these
laws apply to these companies, and without a complete understand-
ing of the benefits and drawbacks generated by these new services.
Furthermore, such a debate has resulted into little academic work,
as we shall see in Section 2. We aim to fill this gap by perform-
ing the first socio-economic analysis of Airbnb adoption. We do
so by using the city of London as case study. London is particular
well-suited because of its high diversity in socio-economic and ge-
ographic terms, and of its enthusiastic adoption of Airbnb (by June
2015, London had over 14,000 Airbnb properties listed). We show
which areas benefit from Airbnb, and how the insights related to
1
See: http://www.reuters.com/article/2015/09/28/us-
airbnb-growth-idUSKCN0RS2QK20150928
2
Airbnb itself released several studies quantifying the positive eco-
nomic impact of the company in many cities around the world. For
more details see: https://www.airbnb.com/economic-
impact
3
See: http://www.nytimes.com/2015/01/08/upshot/new-
york-city-taxi-medallion-prices-keep-falling-now-
down-about-25-percent.html
that inform policy making. More specifically, we make two main
contributions:
• We crawl Airbnb data in London from 2012 to 2015 and
study the adoption of the platform across the UK census ar-
eas in the city (Section 4).
• We analyze such data (Section 5) and contrast the socio-
economic conditions of the areas that benefit from Airbnb
to those of the areas that do not (Section 6).
We then conclude by putting forward five recommendations on
how Airbnb might be regulated based on our insights (Section 7).
2. RELATED WORK
Our work relates to the growing literature on the regulation of
the sharing economy. Research in these area comes from many
disciplines, from law to economy to policy. In [3], the authors,
after enumerating the efficiencies that the sharing economy pro-
vides for both service providers and consumers, discuss regulation
and policies for such software platforms. They suggest the need
to adapt law and regulations to allow those platforms to operate
legally. This will ensure that service providers, users and third par-
ties are adequately protected from any harm that may arise. Of the
same opinion are the authors in [8]. They argue that when market
circumstances change dramatically – or when new technology or
competition alleviates the need for regulation – then public policy
should accordingly evolve. Einav et al. [4] provide a discussion
about licensing, employment regulation, data, and privacy regula-
tion of the sharing economy. They do so by considering the cur-
rent regulations adopted by a few municipalities, and discussing
the pros and cons. In [12], the author critiques the existing regu-
lation of Airbnb. [20] presents a taxonomy of “sharing”, including
formality and gratuity, and examines doctrinal responses to sharing
situations. [16] compares Uber’s efficiencies with its regulatory ar-
bitrage. [15] analyzes the challenges of regulating the sharing econ-
omy from an “innovation law perspective” by arguing that these in-
novations should not be stifled by regulation, but should also not be
left totally unregulated. [2] argues for self-regulatory approaches
and reallocation of regulatory responsibility to parties other than
the government. Finally, [7] studies how financial incentives are
mediated by hospitality and sociability in Airbnb.
While the above works do an excellent job in defining the bases
upon which the sharing economy should be regulated, none of them
does so upon empirical evidence of what the sharing economy re-
ally is, how it has been adopted, and who benefits from it. By con-
trast, this work argues for evidence-informed policy making, and
provides answers to the above questions by empirically investigat-
ing Airbnb adoption.
3. OVERVIEW
Where are Airbnb listings located? This is one of the most fre-
quently asked questions by municipalities, hoteliers and travellers.
To start answering it, we crawled extensive data about Airbnb prop-
erties (from 2012 to 2015) and hotels for the city of London (which
the next section will describe in detail), and we simply map the
presence of hotels and Airbnb listings in the city. A clear distinction
that the Airbnb website makes is between entire home/apartment
(case where the whole home/apartment is rented) and private room
(case where only a private room is rented and all the other spaces of
the house are shared with others). Given that distinction, we sep-
arately map the offering of Airbnb houses (Figure 1b) and Airbnb
Airbnb Rooms
Year hotel_in_bnb_areas bnb_in_hotel_areas
2012 0.14 0.64
2013 0.14 0.67
2014 0.12 0.71
2015 0.12 0.71
Airbnb Houses
Year hotel_in_bnb_areas bnb_in_hotel_areas
2012 0.24 0.64
2013 0.23 0.64
2014 0.24 0.64
2015 0.24 0.63
Table 1: Fraction of London areas that have hotels and Airbnb
properties (rooms vs. houses).
rooms (Figure 1c), and contrast them to the offering of hotels (Fig-
ure 1a). Figure 1 shows that hotels have spotty coverage throughout
the city of London, and they are mostly concentrated in the center
and near the main airport (Heathrow) on the west side. Airbnb
houses have a heavy presence in the city center (like hotels), but
they also reach adjacent areas up to around 10 miles from the cen-
ter. Airbnb rooms massively cover – almost uniformly – the great-
est part of the city of London instead, including suburban areas.
To go beyond visual inspection, we compute the overlap between
Airbnb adoption and hotel adoption. Since each area can be cov-
ered at various levels of strengths by Airbnb and hotels, we adopt
the fuzzy logic functions. Specifically let bnb and hotel be two
fuzzy sets such that bnb
i
∈ [0, 1] and hotel
i
∈ [0, 1] denote, re-
spectively, the strength of Airbnb’s offering in area i and of hotels’.
The strength is zero if Airbnb listings (hotels) are totally absent
from area i, is one if they show maximum presence (with respect
to the entire dataset), and, otherwise, assumes intermediate values
proportional to the presence. Upon those two sets, we compute the
ratio of areas covered by Airbnb that are also covered by hotels,
and the ratio of areas covered by hotels that are also covered by
Airbnb as follows:
hotel_in_bnb_areas =
|bnb ∩ hotel|
|bnb|
bnb_in_hotel_areas =
|bnb ∩ hotel|
|hotel|
(1)
Where the intersection (bnb ∩ hotel) of two fuzzy sets is defined
by (bnb ∩ hotel)
i
= min{bnb
i
, hotel
i
}, and the cardinality of a
fuzzy set bnb is defined by |bnb| =
P
i
bnb
i
. We compute these
ratios for every year, from 2012 to 2015, for the two Airbnb list-
ings categories: rooms and houses (Table 1). Airbnb properties
(especially rooms) tend to be located in areas where there are ho-
tels. That has been true over the years and, from 2012 to 2015, has
increased for Airbnb houses as well (specifically, by 7%). On the
contrary, hotels do not tend to be in areas where there are Airbnb
properties. Therefore, we can safely conclude that Airbnb listings
cover a much broader city area than what hotels do.
Since the spatio-temporal dynamics behind Airbnb are quite unique
(and definitely different than those behind hotels), we set out to
study it in detail and answer four main questions:
RQ1 – What are the main socio-economic characteristics of areas
with Airbnb listings?
RQ2 – Are all types of listings equal? Is there any difference be-
tween, for example, Airbnb listings of rooms and those of
entire houses?
RQ3 – What is the temporal evolution of Airbnb listings?
(a) Hotels (b) Airbnb houses (c) Airbnb rooms
Figure 1: Heat maps of the number of hotels, Airbnb houses, and Airbnb rooms in each London ward. The darker the ward, the higher the
number. The legend reflects the actual (not normalized) numbers, which are thus comparable across the three maps.
RQ4 – Where do Airbnb customers actually go? That is, what
are the main socio-economic characteristics of areas where
Airbnb customers go?
4. DATASETS AND METRICS
To answer those questions, we need to collect information from
various data sources. On one hand, we need detailed records of
Airbnb properties; on the other hand, we need to collect socio-
economic data and derive neighborhood metrics from it.
4.1 Airbnb Data
We have periodically collected, since mid 2012, consumer-facing
information from airbnb.com on the complete set of users who
had listed their properties in the city of London for rental on Airbnb.
We refer to these users as hosts, and their properties as their listings.
Each host is associated with a set of attributes including a photo, a
personal statement, their listings, guest reviews of their properties,
and Airbnb-certified contact information. Similarly, each listing
displays attributes including location, price, a brief textual descrip-
tion, photos, capacity, availability, check-in and check-out times,
cleaning fees, and security deposits.
Our collected dataset contains detailed information on 14,639
distinct London hosts, 17,825 distinct London listings, and 220,075
guest reviews spanning a period from March 2012 to June 2015.
From this data we measure:
Airbnb offering per area (bnb_offering): the ratio between the
number of Airbnb listings registered in a given London area
over the surface of the same area in square kilometers. We
have also considered two types of normalization other than
surface – number of inhabitants and number of dwellings.
For all the three types, the results are comparable.
Airbnb demand per area (bnb_demand): the total number of Airbnb
reviews registered in a certain area of London over the size
of the area in square kilometers. We use reviews as a proxy
for demand, not least because it has been shown that people
leave reviews after staying at a place more than 70% of the
times [6].
4.2 Socio-economic Conditions
We used two different data sets that reflect socio-economic con-
ditions of London areas.
4.2.1 Census Data
We gather the 2011 official UK census data
4
containing demo-
graphic information about small areas defined by the UK Govern-
ment and known as wards. This includes the population density
of the area, how many young people live there, the number of ed-
ucated people, as well proxies concerning how pleasant a partic-
ular area is to live in (e.g., the percentage of green space). From
this dataset, we also collect housing information. This includes the
number of flats and houses present in an area, the number of proper-
ties sold, the number of dwellings that are owned rather than rented,
and the median house price. This information is useful to have an
accurate picture of the type of housing available in each London
area, as well as the fluidity of the housing market there. Most of
those metrics have been widely used. By contrast, a few have been
used in a limited number of papers and need to be illustrated:
Diversity of Ethnic Groups (ethnical_mixed). The idea for this
diversity index was taken from Chris von Csefalvay’s data
blog [19]. In the blog, the author describes a method of mea-
suring diversity in England and Wales with a metric taken
from mathematical ecology. This metric is calculated as the
Gini-Simpson diversity index
5
of the ethnic groups living in
each area. The census data contains five different categories
of ethnicity (number of white, black, Asian, mixed and other
individuals in an area). These five categories were used to
calculate the Gini-Simpson index. This index represents the
probability that two individuals chosen at random from an
area are of a different ethnicity (high values are associated
with multi-ethnic areas).
Bohemian Index (bohemian). We start from the work of Richard
Florida [5] on the effect of the bohemian, artistic and gay
population on regional house prices. The author found that
a newly derived “Bohemian-Gay Index” has a substantial ef-
fect on house prices. We can thus hypothesize that a similar
metric may have an interesting effect on the number or price
of Airbnb offerings. Unfortunately, since gender is not part
of the UK census information, we are not able to recreate this
metric. We therefore followed the same approach adopted by
Nick Clifton [1] that analyzed the creative class in the UK
instead. By following Florida’s work, Clifton computed a
4
See: http://data.london.gov.uk/dataset/ward-
profiles-and-atlas
5
The Simpson diversity index is a measure that reflects how many
different entries there are in a data set and the value is maximized
when all entries are equally high [18].
Category Metric Source Description
Airbnb
bnb_offering Airbnb website
Number of Airbnb properties per km
2
bnb_demand Airbnb website
Number of Airbnb reviews per km
2
Hotel hotel_offering Ordnance Survey
Number of hotels per km
2
Attractiveness
foursquare Foursquare
Number of Foursquare check-ins per km
2
transport Census
Score for accessibility to public transportation
attractions Ordnance Survey
Number of attractions and entertainment places
Demographic
young Census
Number of people aged between 20 and 34 years per km
2
income IMD from Census
Score for income
employment Census
Ratio of the number of employees over the area’s population
ethnical_mixed Census
Score for ethnic diversity
bohemian Census
Fraction of residents employed in arts, entertainment, and recreation
melting_pot Census
Percentage of non-UK born residents
education Census
Percentage of residents with MSc+
Housing
living IMD from Census
Score for living environment conditions
green_space Census
Percentage of green space over the total area’s surface
top_house_price Census
Percentage of dwellings in council tax band F-H (band of the highest median house price)
houses_vs_flats Census
Percentage of houses over houses plus flats
owned_vs_rented Census
Percentage of owned properties
house_price Census
Median house price
sold_houses Census
Number of properties sold per km
2
Table 2: Description of the variables used in our analyses.
cultural metric (the Bohemian Index) by only using the data
made available in the UK census. This metric is defined as
the fraction of people employed in arts, entertainment and
recreation.
Melting Pot Index (melting_pot). This is the second metric used
by Nick Clifton [1] to describe the creative class in the UK
and is the number of people born outside the UK divided by
the total number of people in the area.
4.2.2 IMD Score
We also collect the UK Index of Multiple Deprivation (IMD)
data
6
available at the level of small census areas known as Lower-
layer Super Output Areas (LSOAs). LSOAs are defined to roughly
include always the same number of inhabitants (around 1,500).
7
IMD is a composite score, comprising seven distinct domains: (i) in-
come, (ii) employment, (iii) health, (iv) education, (v) barrier to
housing and services, (vi) crime, and (vii) living environment. For
the purpose of our study, we collected the values of two indexes,
called income and living environment, as we hypothesize that these
two factors, jointed with the ones collected with the census data,
may have an impact on the number and type of Airbnb offerings.
4.3 Attractiveness
A traditional metric often used to describe London areas is trans-
portation accessibility (transport): the higher the value, the more
accessible the area by public transport. This metric is ready avail-
able from the UK Census. To capture more nuanced facets of at-
tractiveness of London areas other than transport accessibility, we
compute three further metrics from two other data sets.
4.3.1 Foursquare
Foursquare has been launched in 2009 and it is one of the most
popular location-based social networking website.
8
Using Foursquare,
6
See: https://www.gov.uk/government/uploads/system/
uploads/attachment_data/file/6871/1871208.pdf
7
See: https://www.gov.uk/government/statistics/
english-indices-of-deprivation-2010
8
See: https://foursquare.com/about
registered users that visit a location can “check-in” on the appli-
cation to share their real-time location with friends. In Decem-
ber 2013, Foursquare surpassed 45 million registered users and
currently male and female users are equally represented.
9
Janne
Lindqvist et al. studied why people check-in and found that indi-
viduals tend to use Foursquare to see where they have been in the
past and ultimately curate their own location history [11]. For this
reason, we hypothesize that, in cities where Foursquare has high
penetration such as London, the number of Foursquare check-ins
may be considered as an approximate measure of the attractiveness
of areas (i.e., areas where city dwellers prefer to visit and spend
time in). We use the official Foursquare API to crawl Foursquare
check-ins.
10
We perform this step between 04/03/2014 and
08/04/2014, resulting in the collection of 26,344,115 users check-
ins in the whole London metropolitan area. We then compute our
first measure of area attractiveness as the number of Foursquare
check-ins in a specific area over the area’s surface in square kilo-
meters. We denote this variable as foursquare.
4.3.2 Ordnance Survey
Ordnance Survey
11
is the national mapping agency for Great
Britain. OS mapping is usually classified as the more detailed map-
ping of the country and covers not only roads but also millions of
Point of Interests (POIs) of varying nature, from restaurants to hos-
pitals and hotels. Ordnance survey data is freely available.
12
We
downloaded the data in July 2015, collecting 513,786 POIs in the
whole metropolitan London area. For the purpose of this study,
we considered the number of Ordnance Survey POIs that fall under
one of the categories of “eating and drinking”, “attractions”, “re-
tail”, “sports and entertainment” – to capture London areas that are
covered by attractions – normalized by the size of the area in square
kilometers. We denote this variable as attractions.
4.4 Hotel Data
To study whether Airbnb offerings are located in areas with pres-
ence of traditional forms of accommodation, we consider the num-
ber of Ordnance Survey POIs that fall under one of the categories
of “hotels”, “motels”, “country houses and inns” normalized by the
9
See: https://en.wikipedia.org/wiki/Foursquare
10
See: https://api.foursquare.com
11
See: www.ordnancesurvey.co.uk
12
See: www.ordnancesurvey.co.uk/resources
