InnovationManagementandProductDevelopmentIRENESUÁREZPACIOSMasterofScienceThesis2020:TRITA-ITM-EX2020:169KTHRoyalInstituteofTechnologySchoolofIndustrialEngineeringandManagementMachineDesignSE-10044STOCKHOLM Impactsofpeer-to-peerrentalaccommodationinStockho

(1)

IN

DEGREE PROJECT COMPUTER SCIENCE AND ENGINEERING, SECOND CYCLE, 30 CREDITS

STOCKHOLM SWEDEN 2020,

Impacts of peer-to-peer rental

accommodation in Stockholm,

Barcelona and Rio de Janeiro

An exploratory analysis of Airbnb’s data

IRENE SUÁREZ PACIOS

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF INDUSTRIAL ENGINEERING AND MANAGEMENT

(2)

(3)

Impacts of peer-to-peer rental accommodation in

Stockholm, Barcelona and Rio de Janeiro:

an exploratory analysis of Airbnb’s data

Innovation Management and Product Development IRENE SUÁREZ PACIOS

Master of Science Thesis 2020: TRITA-ITM-EX 2020:169 KTH Royal Institute of Technology

School of Industrial Engineering and Management Machine Design

SE-100 44 STOCKHOLM

(4)

(5)

Abstract

As a part of the growing movement called the “peer-to-peer” economy, Airbnb has changed the short-stay rental market and has become one of the world’s largest booking websites for finding an accommodation to stay. The platform has also affected the economy of tourism around the world, so, given the importance of the subject, in this thesis study, the impacts that the Airbnb rental accommodation model has on clients of Stockholm, Barcelona and Rio de Janeiro is studied. In this way, it has been analyzed how factors such as price, location and seasonality affect Airbnb customers in these cities. To do this, the three cities were first analyzed individually and then compared, using data from the Inside Airbnb website from 2010 to now. This research has been carried out through an exploratory analysis using the R programming language. The study has been divided into three parts:

First, the Spatial Data Analysis has shown that Airbnb´s presence in all three cities has increased significantly in the past decade, growing from the most touristy parts of the city to surrounding areas. In addition, it has been observed that the largest number of Airbnb properties are apartments located near the city center and touristic places, which also are the most valued areas by Airbnb customers and the most expensive to rent a property. Secondly, a Demand and Price Analysis has been carried out. In this part, the demand for Airbnb listings has been estimated over the years since 2010 and across months. A significant increase in demand has been appreciated in the last decade, which also shows a seasonal pattern. In the three cases, the demand graph follows the city´s climate, showing the highest demand during the summer months, which corresponds to the most expensive period. Finally, through User Review Mining, customer opinion has been studied by applying text mining to reviews. In this part of the research, word clouds have been used to have a visual representation of the text data, showing the most frequent words and analyzing what makes customers feel comfortable and uncomfortable.

Key words: Airbnb, peer-to-peer economy, shared economy, exploratory analysis, text mining

(6)

(7)

Sammanfattning

I detta examensarbete har effekterna som Airbnbs hyresmodell har på kunder i Stockholm, Barcelona och Rio de Janeiro studerats. På detta sätt har det varit möjligt att analysera hur faktorer som pris, plats och säsongsvaror påverkar Airbnbs kunder i dessa städer. För att göra detta analyserades först de tre städerna individuellt och jämfördes sedan med data från webbplatsen Inside Airbnb från 2010 till nu. Denna forskning har genomförts genom en undersökande analys med programmeringsspråket R. Studien har delats in i tre delar:

För det första har den rumsliga dataanalysen visat att Airbnbs närvaro i alla tre städerna har ökat markant under det senaste decenniet och växte från att omfatta de delar av staden som är mest intressanta för turister till omgivande områden. Dessutom har det observerats att det största antalet objekt på Airbnb är lägenheter belägna nära centrum och platser intressanta för turister, som också är de mest värderade områdena av Airbnbs kunder och de som är dyrast att hyra i en fastighet. För det andra har en efterfrågan och prisanalys genomförts. I denna del har efterfrågan på Airbnbs registreringar uppskattats under åren sedan 2010 och över flera månader.

En betydande ökning av efterfrågan under det senaste decenniet har uppskattats, vilket också visar ett säsongsmönster. I samtliga tre fall följer efterfrågan förändringarna i stadens klimat och visar den högsta efterfrågan under sommarmånaderna, vilket också motsvarar den dyraste perioden.

Slutligen, i avsnittet Användarrecensioner, har återkoppling från kunderna studerats genom att använda textutvinning på recensioner. I denna del av forskningen har ordmoln använts för att få en visuell representation av textdata, som visar de vanligaste orden och analyserar vad som gör att kunderna känner sig bekväma och obekväma.

Nyckelord: Airbnb, peer-to-peer-ekonomi, delad ekonomi, undersökande analys, textutvinning

(8)

(9)

Acknowledgments

First, I want to thank Airbnb, since without their data the analysis carried out in this thesis would not have been possible.

Thank you to my two supervisors: Rafael Laurenti, for giving me the opportunity of doing this thesis at the Royal Institute of Technology and for your availability and guidance, and Joaquín Ordieres Meré, for agreeing to supervise my thesis remotely from the Polytechnic University of Madrid (UPM).

Doing this thesis would not have been the same without the friends that I have made in this Erasmus, especially Carmen and Sandra, who have become my Swedish family and who have encouraged and supported me at all stages of the project.

I also want to thank everyone who has been with me in the distance during this experience and that I have missed so much. To my friends from Majadahonda, my cousins from Rimor and my friends from university. Mainly Miguel, for all the fun times, for his support and for being the best friend.

Lately, I would like to thank my family for their love and support. To my parents, Aida and David for their advice and inspiration not only in this exchange but also during all my student years. Especially to my father, who has been an essential part of this project, for his patience and help.

Irene Suárez Pacios

(10)

(11)

Introduction

In this chapter, the background of the thesis is described along with the purpose and the delimitations.

1.1 Background

Airbnb was founded in San Francisco in 2007. A conference was being held in the city and two university graduates came up with the idea of offering three air mattresses on the floor in their apartment to conference delegates. The students used a simple website to advertise their apartment as an “AirBed & Breakfast” for those conference delegates who were looking for a cheap way to spend the night to avoid the high hotel prices. After this, the two colleagues recruited another friend to exploit their business idea and developed the website to advertise for tourists. In 2009, the website was relaunched as Airbnb.com [1]. In February 2011, the platform reached one million booking nights and by August 2016 there were more than 2 million listings in 34,000 cities and 191 countries worldwide [2].

The Airbnb booking website allows the customer to search for accommodation based on destination, travel dates and number of guests. The website returns a list of available accommodations that can be filtered by different attributes such as prices, type of place, number of rooms… Apartment information is also available with descriptions, photos and reviews from previous guests. The platform allows tourists to rent different types of accommodation: from small rooms to an entire apartment or house, 57% of Airbnb listings are entire apartments and homes, 41% are private rooms and 2% are shared rooms [11]. In some cases, the host may be living in the space at the same time as the rental or may be absent. Payments are made through the website and the platform charges guests a fee under 13% of the booking subtotal and hosts a 3% fee [3].

Airbnb is part of the growing movement called the “shared” or “peer-to-peer” economy, which uses online and mobile technologies. This new model competes with traditional and physical businesses since this economy generally implies consumers maintaining access to goods and services without owning them [2]. Airbnb has changed the traditional model by providing an online marketplace that enables the rental of spaces from one ordinary person to another: it connects people who own idle accommodation assets (hosts) with those looking for a place to stay (guests) via digital marketplaces. The rise of Airbnb represents an innovation within the tourism accommodation industry, whose rapid growth has been enabled by two key factors: technology innovations and supply-side flexibility [4].

One of the main features that explain the success of Airbnb is its prices: its accommodations tend to be cheaper than traditional accommodation. Airbnb hosts can offer very competitive low prices because the hosts´ primary fixed costs, such as rent and electricity, are already covered.

Moreover, there are no labor costs or they are minimal and hosts are usually not fully dependent on the revenue that they get from Airbnb. In addition to the economic aspects, Airbnb offers its hosts a compelling experience value proposition “Live like a local” [5], the company offers an

(14)

opportunity to experience local life, meet real neighbourhoods in non-touristic areas, interact with the host and have helpful local advice. Hosts may also appreciate staying in real homes over a hotel and having access to a kitchen, washing machine, dryer and other residential amenities [1].

Authenticity is also seen as one of the motivators for customers to choose Airbnb [6].

This new form of accommodation has changed the short-stay accommodation market, having an impact on the tourism economy. Airbnb, aware of this impact, has invested many resources in studies to analyze its real impact on the rental market. The platform insists that they expand the tourism market instead of competing directly with hotels. [2]. The company believes that they contribute positively to the tourism sector, since their hosts stay longer, spend more money and bring activity to local neighbourhoods (according to Airbnb, in many cities, over 70% of the spaces they offer are outside the central hotel districts [4]). Furthermore, since Airbnb listings are more scattered than hotels, hosts may be more likely to spend money in neighbourhoods that do not typically see much economic activity [1]. Airbnb also can absorb peak demand in destinations where the capacity of its hotel rooms is unable to sustain the tourist season or major events.

As for Airbnb, researchers have begun to examine its relationship to more traditional forms of hospitality. Several independent studies suggest that the platform has a negative impact on the income of local hotels and they state that Airbnb is expected to reduce hotel rates and revenues [2].

They consider that Airbnb is good for tourism but bad for hotels and that a large part of Airbnb listings compete directly with the traditional offer of hotels, since they share spaces [7]. However, other researches, such as Varma et al. investigation [8], suggest that hotels and Airbnb listings are quite complementary since their customers tend to be different, so they concluded that Airbnb hardly disrupts the industry.

The growth of Airbnb has been linked to protests against the high tourist presence in cities such as Barcelona, Amsterdam and Berlin. Residents’ annoyance with tourists has increased since in many cases, the high demand for Airbnb apartments has led to the displacement of residents and increased rental costs [5]. The discomfort with the platform is also related to the regulation debate, Airbnb has been accused of unfair competition in many sectors of the hospitality industry and there have been many discussions about how taxes should be imposed on the platform. When guests stay in traditional accommodations, they have taxes related to this. However, Airbnb does not charge them, so customers can generally avoid paying the taxes that are generally charged to this sector, which gives Airbnb rentals a competitive advantage over traditional accommodation. Taxes, along with local laws in the cities where Airbnb has listings, have led the platform to have legal battles against local governments and face many legal issues [1].

1.2 Purpose

The main objective of this project is to study the impacts of peer-to-peer rental accommodation in different scenarios using Airbnb data from 2010 until now. This thesis will focus on three different cities: Stockholm, Barcelona and Rio de Janeiro. The aim is to understand the effect that different patterns, such as location, price and season, have on customer’s demand for Airbnb properties.

In addition, customer opinions will be studied by analyzing customer reviews and feedback drawn from the Airbnb dataset. In this way, it will be possible to know what customers mention most frequently and what makes them comfortable and uncomfortable.

To carry out this study, an exploratory analysis will be performed using the R programming language. Thus, it will be possible to deepen into what affects Airbnb customers in these cities and gain new knowledge about the Airbnb rental market. Some of the questions that are intended to be answered through the analysis are:

(15)

• How is the evolution of Airbnb´s spatial penetration?

• Which locations in each city are highly rated by guests?

• How do prices of listings vary by review score and location?

• What type of properties are there in each city? Do they vary by neighborhood?

• How does the demand for Airbnb rentals fluctuate throughout the year to study the seasonality of demand? And between years?

• Are the demand and prices of the rentals correlated?

• Are there any common themes that can be identified from the reviews? What aspects of the rental experience do people like and what aspects do they dislike?

After the analysis of each of the cities, a comparison will be made between the results of the three.

With this comparison, the common and the different points will be obtained. In this way, it will possible to identify the common aspects that attract and deter customers.

This study is conducted for academic purposes so that it can contribute to understanding the impacts of a ”peer-to-peer” economy case study. This research involves an important analysis, since the platform has changed the market for short-term rentals that affects the tourism economy, so it may be interesting for many professionals and institutions such as the platform itself, economists, for research related to these fields, tourism booklets or local governments of the analyzed cities.

1.3 Delimitations

• This master thesis is carried out for twenty weeks, following the requirements of the Degree Project Course in Innovation Management and Product Development of KTH Royal Institute of Technology and the requirements of the Master Thesis in Industrial Engineering and Management of the Polytechnic university of Madrid (UPM).

• This study will solely incorporate analysis of three cities, which are Stockholm, Barcelona and Rio de Janeiro.

• The data collected for the analysis dates from 2010 until now (February 2020).

• Apart from the analysis of Airbnb variables, to complete the study, further research regarding the context of the city is needed. In this way, it is necessary to select several geographic, economic and social variables that affect the demand for Airbnb, since there are a large number of factors that directly or indirectly influence customers. This selection will be made following the criteria of the student after the necessary documentation and the acquisition of the necessary knowledge on this topic.

• To study demand, the ”number of reviews” variable is used as an approximation of the number of bookings made over the past year, since the dataset does not have data on the bookings made over the years. It is assumed that about 50% of guests review the host and their stay on the listings.

• Occupancy rates will be studied using data from the ”calendar” table, which provides data for the next year. This will be used as an estimate of the occupancy, as no occupancy data is available from previous years.

• There are data that contain missing values and, because of the difficulty in working with them, the rows that contain them will be removed from the analysis.

(16)

• Because the reviews contain opinions in multiple languages, it is necessary to filter and delete non-English comments. This has to be done because the word clouds that will be created to display the most common words in reviews must be in English. Deleting non-English comments means removing less than 10 % of words, as the percentage of English words in reviews is higher than 90 % in the three cities.

• There is no scientific literature to compare with the results obtained from the analysis of the cities, since the study of the Airbnb rental model in Stockholm, Barcelona and Rio de Janeiro has not been carried out before.

(17)

Chapter 2

Frame of Reference

In this chapter, the frame of reference is presented. It summarizes the context of the three cities chosen for the study. The existing knowledge and the research of this section are later used in the analysis to interpret and discuss the results.

2.1 Stockholm

Stockholm is the capital and the most populous urban area of Sweden, with a population of 974,073 inhabitants in the municipality [9]. The city spreads across a Baltic Sea archipelago of fourteen islands. As can be seen in the map in figure 2.1, the Municipality of Stockholm is divided into 14 urban districts. The actual urban map of Stockholm is very influenced by the ”General plan for Stockholm 1952”. This plan was created to decentralize Stockholm and solve its housing shortage after the Second World War. It meant the creation of new suburbs built along the subway lines with commercial and public services, which allowed the population to live in the surroundings and get to the center to the city in a short period. This innovative urban model was called ”ABC City”

(Work, Housing, and Centrum) [10].

Nowadays, due to high levels of immigration, a leading hub for high-tech start-up scene and having the continent’s highest birth rates, Stockholm is Europe’s fastest-growing capital, with a population that has grown by almost a quarter of a million in just seven years [11]. The capital is considered the “European Silicon Valley” in the production of successful high-tech startups [12], as it currently holds around 22,000 tech businesses (Stockholm has birthed successful global brands like are Spotify, Skype, and µTorrent) [13]. This scenario means the arrival of people to work in these companies, who cannot assure their new employees a place to stay and find it difficult to host their workers since there are strict housing regulations designed to stop firms renting apartments en masse in the same block [11].

This, together with strict building regulations and a lack of investment over de last decades, has led to a housing storage in the city [11]. Thus, the Swedish National Board of Housing, Building, and Planning has estimated that, to meet demand, around 600,000 apartments need to be built in Sweden over the next nine years [14]. All this explains the long waiting lines for rent-controlled housing. The average waiting time to get a rental apartment in 2016 was nine years and between fourteen and sixteen in the most popular neighborhoods [15]. Moreover, sharing a room becomes also very difficult since over half of households in Sweden (52 % of all households) are single-person properties [16].

As stated, the Swedish rental market is regulated. It is regulated the amount of rent the landlord can take, under what circumstances an apartment can be rented, for how long and the high taxes associated with it [17]. Under these market conditions, Airbnb plays a controversial role. There have been many discussions about Airbnb and its effect on Stockholm´s complex housing market.

However, it is legal for tourists to rent an apartment through Airbnb [18]. It is profitable for

(18)

property owners to rent the apartment through online portals like Airbnb, as they can get a higher amount of renting per night instead of the traditional way of renting the apartment per month [19].

Notwithstanding, Airbnb seems the ultimate temporary solution to deal with the long waiting queues. The platform has helped the tenants to solve their accommodation challenges in an overcrowded market. Furthermore, it seems a suitable solution in this city, which is highly hi-tech driven.

• Älvsjö

• Bromma

• Enskede-Årsta-Vantör

• Farsta

• Hägersten-Liljeholmen

• Hässelby-Vällingby

• Kungsholmen

• Norrmalm

• Östermalm

• Rinkeby-Kista

• Skärholmen

• Skarpnäck

• Södermalm

• Spånga-Tensta

Figure 2.1: Stockholm neighbourhood map.

Wikipedia. Stockholm Municipality Map.

Retrieved March 4, 2020, from:

https://es.wikipedia.org/w/index.php?title=Estocolmo&oldid=124459340

2.2 Barcelona

Barcelona is a city on the northeast coast of Spain. It is the second-most populous municipality in Spain, with a population of 1.6 million [9]. As can be appreciated in figure 2.2, the city is divided into 10 administrative districts. This city is recognized for its innovative urban planning. The transformation of Barcelona began in the second half of the 19th century with the ”extension”

(Eixample) from the old town (Ciudad Vella) to the surroundings. It was designed by the architect Ildefons Cerdà, who created this new area characterized by long straight streets and a strict grid pattern crossed by wide avenues [20]. In the following years, the adjacent areas swallowed

(19)

across San-Martí and Saints-Montjuïc [22]. At present, Barcelona is undergoing a third wave of transformation, with the initiative “22@distric” to convert San-Martí into a technology and knowledge-driven economic powerhouse zone [20].

Nowadays, tourism involves early twelve percent of Barcelona´s economy [23], but it was not until the Summer Olympics that its international profile expanded to the current levels: Barcelona is currently considered the fifth most visited city in Europe and twentieth in the world [24] (the city receives almost 12 million visitors in total [25]). Many factors have contributed to the strong demand of the city: the support and encouragement of tourism from the Government; the successful advertising campaigns of Barcelona to the international market as a fun European destination, with good weather, beaches, lively nightlife and a wide variety of museums and architecture; the rise of budget airlines like Ryanair and the availability of a wide range of properties through short-rental portals such as Airbnb. Moreover, Barcelona leads the world ranking of congresses held [26]. Its location “at the entrance of Europe”, which makes the city accessible by sea, air, and road, and its Mediterranean climate with mild temperatures throughout the year (see climograph in Appendix C), make this city a good option to celebrate these type of events. Two examples of these are the Smart City Expo World Congress and the Barcelona Games World.

• Ciutat Vella

• Eixample

• Gràcia

• Horta-Guinardó

• Les Corts

• Nou Barris

• Sant Andreu

• Sant Martí

• Sants-Montjuïc

• Sarrià-Sant Gervasi

Figure 2.2: Barcelona neighbourhood map.

Wikipedia contributors. 2019, December 28. Districts of Barcelona.

In Wikipedia, The Free Encyclopedia. Retrieved March 4, 2020, from:

https://es.wikipedia.org/w/index.php?title=Estocolmo&oldid=124459340

In this scenario, the growth of Airbnb plays an important role as a platform to host millions of tourists [23]. Airbnb´s entry in 2009 took place in the middle of the global financial collapse, which was received by many unemployed citizens as an easy way to receive cash. By 2010, the Government liberalized the rules related to short-term vacation rentals, which resulted in thousands of new licenses for apartment owners. This number quadrupled over the next four years [23]. But this climate changed radically in the following years: many residents began to show signs of discontent,

(20)

as they were tired of the “excessive tourism” and the so-called “binge-drinking tourism” [27]. All this, added to the one million illegal beds in the territory lead to complaints and protests from the citizens of Barcelona [28].

In response to the difficult situation, the Catalan Government, which has the competence of tourism in the region, launched an operation against illegal tourist apartments and tried to slow down Airbnb activity. They published a new law by which any rented apartment for visitors had to be registered in the Tourism Registry and have a permit [29]. They also took measures as a way to protect the locals and the hotel industry from what was considered “unfair competition”. To do this, in July 2015, Airbnb was punished with two fines of 30,000 euros (318,516 SEK each). However, since according to the municipal administration, Airbnb continued offering apartments without license numbers, they decided to apply the maximum fine of 600,000 euros (6,370,321 SEK)[30]. Given this situation, Airbnb appealed against the fines, and the platform, aware of the growing hostility began working more closely with local governments. Among other things, an action plan was introduced to identify hosts who were breaking rental laws and to close down unlicensed properties. At present, this situation is still unresolved. However, the supply of illegal apartments on offer has been reduced by 95%; 4,900 have been shut down and 6,500 fines issued [23].

2.3 Rio de Janeiro

Rio de Janeiro is the capital of the state of Rio de Janeiro and the second-most populous municipality in Brazil. Home of more than 6 million people [9], and founded on the east of Guanabara Bay, this city is located in the southeastern region of Brazil, on the Atlantic coast. The municipality of Rio is administratively subdivided into nine subprefectures presented in the map of the figure 2.3, containing a total of 160 neighbourhoods (see the table in Appendix D).

The city was founded in the 16th century around what is now Centro Histórico e Zona Portuária, following the Spanish model of town planning (Laws of the Indies) and building organized and orthogonal streets. In the following years, the city grew and spread inland, surrounding the

“morros” (small mountains). During the 19th century, due to the growing concern of the under- development of the city compared to other European cities, many changes were made imitating their urban planning [31]. In this way, Rio has witnessed many modifications from its origins to the present day: the development of the city has always influenced from first world countries that have mixed with its traditional Carioca culture, forming the so-called ”Carioca urbanism” (characterized by the union between the grid and the curve in the architecture and the city plan of the city) [32].

This mixture and the development attempt explain the irregular distribution of the urban areas of the city, which presents rich and touristic neighbourhoods and poor and dangerous zones, both very close to each other but very differentiated between them. The poor settlement slums in the city are known as “favelas”, and, as can be seen in Appendix E, they are scattered across the city.

In recent decades, the city has experienced significant changes, but it still faces major problems and imbalances related to urban, social, and safety issues. By 2000, the city faced extreme levels of inequality, with an extremely functional division between different districts, showing an insufficient urban infrastructure, water, sanitation, and urban transport in the city´s poor neighbourhoods.

Furthermore, the city was undeveloped as a tourism and business destination [33]. The Summer Olympics Games in 2016 supposed many changes driven by the goal of recovering the urban quality, infrastructure level, and social cohesion [33]. To host millions of visitors for the Games, almost 60,000 hotel rooms were available in the market [34], and Airbnb offered about 33,000 rooms and apartments in the Brazilian city [35]. Moreover, the city´s brand after the Games was positively affected, increasing interest in visiting the country [36].

(21)

millions of visitors is the Carnival of Rio, which takes place in late February. A survey published by Riotur shows that that commerce, hospitality, and services raised 3.78 billion of Brazilian reals (8,5 billion SEKS) during the four days of festivities in 2019 (around 7 million people enjoy this event) [37]. The city also hosts ”Rio’s international film festival” in November, which is one of the biggest in Latin America and ”Festas Juninas” in June, which is one of the most important folkloric festivals in Brazil [38].

Currently, Airbnb offers around 34 thousand listings across the city, helps many families earn extra income, and spreads business to parts of the city that generally do not see much economic activity. Last year, host income and guest spending generated economic activity of 160 million dollars (approximately 1,5 billion SEK) [3].

Figure 2.3: Rio de Janeiro neighbourhood map.

(22)

Chapter 3

Method

In this chapter, the methodology chosen for the elaboration of the thesis is described and discussed.

3.1 Research Design

The study was carried out for 20 weeks in which the master’s student worked full time. The author of this project is an exchange student, so this thesis was carried out through collaboration between the Machine Learning Department of the Royal Institute of Technology KTH, the university where the student carries out the exchange, with the Industrial Management Department of the Polytechnic University of Madrid (UPM), the student’s home university. The author of the thesis was assisted by a supervisor from KTH and a supervisor from UPM.

Regarding the research field of this study, the peer-to-peer economy is currently widely studied.

This explains why a thesis on a practical case of this economy is very interesting, since the Airbnb rental market has changed the short-term accommodation model. As explained in the introduction (see Purpose section), the objective of the research is to analyze the impacts of Airbnb on customers in three different cities. The cities considered in this study had to be cities where this specific analysis has not been done before. The expansion and penetration of Airbnb is widely studied and the platform model has been studied in cities such as London, New York and Paris. Taking this into account, the cities chosen for the study are Stockholm, Barcelona and Rio de Janeiro:

• Stockholm is chosen for its peculiar situation regarding the Swedish controlled rental market, which makes the study unique and interesting. Moreover, it is the city where KTH University is settled and therefore, the city where this thesis has been carried out.

• The selection of Barcelona is made because it is a very touristic city that has witnessed a very controversial situation between the Catalonian Government and Airbnb. The high presence of Airbnb in the city has led to many protests, conflicts and a fine from the local government, which has changed the way Airbnb works in the city.

• Rio de Janeiro implies an interesting field of study since it represents the case of Airbnb in a South American country, very different from the other two cities chosen.

Furthermore, the student has a special link with the three cities, which was necessary to deeply understand their context and urban map: the author knows Stockholm and Barcelona very well, and could ask for help to understand certain aspects of Rio de Janeiro, since the KTH supervisor is from there. This was essential for comparing and verifying descriptions obtained from literature and the internet and explains why these cities were chosen and not others on other continents such as Asia or Africa.

The study began with a literature study, in which aspects related to Airbnb and the cities of

(23)

of study. Once the necessary knowledge was obtained, the gathering of data and its analysis started.

Most of the Airbnb information used in the study was taken from the Inside Airbnb web page:

http://insideairbnb.com/get-the-data.html. It hosts an available, independent and non-commercial set of tools that contains data with information about Airbnb in different cities around the world.

Moreover, further investigation regarding the geographic, social and economic context of the cities was completed to develop a deep understanding of the different impacts of the peer to peer rental model. Once all the data was collected, it was examined using the R programming language and software environment.

3.2 Data Collection

As stated before, the data was gathered from the mentioned webpage and literature, web pages and scientific articles. In this section, the approach of how data was gathered and its classification is explained.

3.2.1 Airbnb Data

Airbnb data was gathered from the Inside Airbnb webpage, which periodically publishes snapshots of Airbnb listings. In February 2020, data from the three cities was downloaded. The dataset is comprised of three main tables:

• Listings: detailed listing data showing attributes for each listing, such as prices, neighbourhoods, latitude and longitude.

• Reviews: detailed reviews given by the guests.

• Calendar: provides details about booking for the next year by listing.

3.2.2 Geographic, Social and Economic Data

To obtain a deeper understanding of each city, several groups of variables have been analyzed, capturing their geographic, social and economic context. The sources of this data include OpenStreetMap (https://www.openstreetmap.org), Google Maps and a variety of official city websites.

3.3 Data Analysis

When all the data was gathered, the data analysis began. In order to conduct the complete study, first of all, a description of the data was performed, followed by an analysis of its quality and an implementation of changes (to create the desired visualizations). Finally, the exploratory data analysis was performed.

3.3.1 Description of Data

In this section, the data content is described in greater depth. The table 3.1 contains information about the dataset used in the study.

(24)

Category Metric Source Description

Airbnb Airbnb

parameters

insideairbnb.com Airbnb parameters that characterize each of the listings

Geography

Distance to

the center GoogleMaps Distance from each of the

neighbourhoods to the center of the city

Points

of interest GoogleMaps Main tourist places or places of interest in the city

Public transport

GoogleMaps and

OpenStreetMap Public transport accessibility and frequency

Hotel distribution

GoogleMaps Distribution of hotels in the city

Climate climate-data.org Climograph of the city

Social Popularity News magazines Popularity and trends of the

neighbourhoods

Safety Police documents Crime, pickpockets and other risks factors related to the personal safety in a neighbourhood

Economic

Median income and median household value

Various sources

Median estimate of household income and value in a neighbourhood

Table 3.1: Dataset summary.

Airbnb Data

The Airbnb dataset of each city is composed of three tables:

• Listings. The parameters extracted from the listings table, used for the analysis are:

– ID: discrete variable that represents the identification number of the listings. This variable is present in the three tables.

– Name: categorical variable that contains the name that the property owner has given to the listing.

– Host since: date when the listing was uploaded to Airbnb website for the first time.

– Neighborhood: categorical variable with information of the neighborhoods where the listings are situated.

– Latitude: discrete variable that indicates the latitude of the property.

– Longitude: discrete variable that indicates the longitude of the property.

– Property type: categorical variable that shows the property type of the listing, such as apartment, loft, house...

– Price: continuous variable that shows the price of the listing.

(25)

• Reviews. The attributes of the reviews used in the analysis are:

– Listing ID: discrete variable with the listing identification.

– Date: day, month and year of when the review was written.

– Comments: textual variable that contains a written review of the host.

• Calendar. This data contains information on the bookings for the next year by listing.

– Listing ID: identifier of the listings.

– Date: day, month and year of the booking.

– Available: categorical variable that shows if the listing is available or not in the specified date.

A general overview of the data shows that there are:

• 6,328 unique Airbnb listings and 121,462 reviews in total in Stockholm. The first rental was up in March 2009 in Hägersten-Liljeholmen.

• 20,411 unique listings and 740,992 reviews in total in Barcelona. The first rental was up in September 2008 in Guinardó (Horta-Guinardó).

• 33,715 unique listings and 316,056 reviews in total in Rio de Janeiro. The first rental was up in March 2009 in Copacabana (Zona Sul).

Geographic, Social and Economic Data

• Geographic data

– Distance to Center. This variable is found to be one of the most significant factors that explain the presence of Airbnb in an area [39]. For simplicity, the center of the city is considered the commercial, cultural and historical heart of the city. This way, the distance to the center is calculated as the approximate distance between the city center and the center of the neighbourhood under study.

– Points of Interest. In this research a point of interest is considered a specific location that might be useful or interesting for tourists. Examples of points of interest include museums, town halls and churches. In a recent study conducted in London [39], a correlation between ”distance to center” and the ”tourism factor” has been detected, which has a great positive significance on the number of Airbnb offerings in that area.

This phenomenon is also expected to be present in the cities analyzed, so that areas with a high concentration of points of interest, which have touristic attraction, will also have a high presence of Airbnb properties. To analyze the points of interest of a neighbourhood, GoogleMaps is used for each city, creating a map with this information to be able to visualize it.

– Public transport. Transport links are a key component in the property prices of an area.

Some tourists spend a lot of money on centric listings, however, many people choose to move away from the center to have cheaper apartments. For those who prefer cheaper listings, the location that they choose is highly influenced by the area´s connectivity to the city center. This is why the Airbnb website has added an option to check the public transits close to the selected listing. Each city has different public transport modalities, nonetheless, since the three cities studied have a metro network, this mode of transport has been chosen as an indicator of the strength of the public transport in the city.

(26)

– Hotel distribution. As previously mentioned, it is unclear if there is a relationship between Airbnb listings in an area and hotels, there are many studies and research on this topic, but there is no unanimous response. However, the author of the thesis wanted to contribute to the analysis of the relation between both. As there is no public dataset available for the number of hotels in all cities, the hotel data has been extracted from Google, searching for ”city name” and ”hotels”. In this way, taking into account the available data, the relationship between the location of hotels and Airbnb properties has been analyzed to determine if they are located in the same spaces and share locations in the cities studied.

– Climate. The relationship between climate and demand is studied. The assumption is that customers tend to travel when weather conditions are good.

• Social data

– Popularity. The popularity of a neighbourhood can affect the demand for properties in a certain area, not only for Airbnb listings but also for people looking for a place to live.

As stated earlier, one of the aspects that Airbnb offers its customers is the experience of living like a local. This could become a better experience if the neighbourhood has a rich local life and good services, which is generally related to those areas where people want to live and where tourists want to have their accommodation.

– Safety. This parameter characterizes the personal safety of Airbnb users in case they want to stay in a listing situated in a particular neighbourhood. Issues such as criminality of the area or the presence of pickpockets are taken into account in this factor.

• Economic indexes

– Median income and median household value. These measures provide an indicator of the socio-economic structure of a city. The median income and the median household value of a neighbourhood affect the aesthetic of the area, the style and type of the buildings, the size of the properties, the average price of the houses... Additionally, a study on Airbnb in London showed that income is related to Airbnb since there are more people with low income that join Airbnb as hosts, possibly using the extra income generated from Airbnb to support themselves [40].

3.3.2 Analysis of Data Quality

Previous to analyze the Airbnb data, it was necessary to perform several transformations on the dataset in order to be able to work with it and create the desired visualizations. Generally, there were no major inconsistencies or mismatches in the data, however, the following actions were taken:

• Initial verification

To verify the validity of the data, 10 random listings were selected for each city and their presence on the original Airbnb platform and the accuracy of its information were double-checked.

The following image shows an example of some of the information available on the Airbnb webpage and some data obtained from the Inside Airbnb website. It can be appreciated that the data from both sides match.

(27)

Name Host Number of guests

Number of beds

Number of baths

Location

score Neighbourhood Wifi Chic apartment

nearly Sagrada Familia VI

Aline 6 4 1 4.8 Sant Martí Yes

Figure 3.1: Screenshot of a property from the Airbnb website.

• Format changes

The majority of changes made in the data consisted of changing the format on the columns to obtain them in a desirable way:

– Some changes were made regarding the language of the review comments. The dataset contains reviews in multiple languages, so it was filtered to remove the comments that were not in English.

– The date format was changed to transform it from a mm-dd-yyyy format to obtain the day, month and year separately.

– Some adjustments were required to get the price column in integer values, removing de

”$” symbol and the comma separator.

(28)

• Missing values

The dataset had null values, so the rows that contained them were dropped. In order to visualize these missing values a vismiss plot was used. This R tool allows us to obtain the percentage of missing values and determine in which observations they are located. The image presented below shows an example of one of the representations made to analyze the missing values of several variables used in the exploratory analysis. Most of the data is available as the total percentage of missing values is 2.7% and most of these non-available data are found in the ”location score” variable.

Figure 3.2: Representation of missing data.

3.3.3 Setting of the Exploratory Analysis The exploratory analysis is divided into three sections:

• Spatial Data Analysis, in which questions regarding the impacts of Airbnb in different locations in the cities will be answered. This section is comprised of the following parts:

– Study of the evolution of the number of Airbnb listings over the years.

– Spatial visualization of the distribution of Airbnb properties in each city using the Leaflet tool of R, which allows the student to visualize the location of the Airbnb listings in the map of the city. With this tool, it is possible to geographically visualize the different Airbnb properties in the city to get an idea of how the listings are distributed in the neighbourhoods. This way, by clicking on each listing, information related to the property can be obtained with details such as the listing name, hostname, price of the property, room type and neighbourhood. The following figure shows an example of some listings situated in the neighbourhoods of Östermalm and Norrmalm in Stockholm, it can be appreciated that one of the listings has been selected and its attributes are presented.

(29)

Figure 3.3: Image of the interactive graph of the Airbnb listings of Norrmalm and Östermalm.

– Creation of a map with pointers indicating the main places of interest of the city. This visualization will be used to understand the distribution of the principal touristic places in the city and if there is a correlation between the location of Airbnb properties and them.

– Study of the correlation between the location of Airbnb properties and hotels in the city.

– Analysis of the areas that Airbnb customers value the most. This study will be carried out using the Airbnb users’ rating of their stay, classifying the data to obtain the average location score for each neighbourhood.

– Investigation of the prices in the different areas of the city, which will be conducted calculating the average price of the listings of each neighbourhood.

– Counting of listing type by neighbourhoods. A bar chart will be made to study the relationship between property type and neighbourhood. Since there are plenty of types of listings it will be selected those that are predominant.

• Demand and Price Analysis. The aim of this part is to study the demand for Airbnb listings over the years since 2010 and across months of the year to understand seasonality.

Moreover, a relation between price and demand will be established, in order to be able to answer whether prices of listings fluctuate with demand. To study the demand, the ”number of reviews” variable has been used as an approximation of the number of bookings made over the past year, since the dataset does not have data on the bookings made over the years. It has been assumed that about 50% of guests review the host and their stay on the listings.

The following points will be studied in this section:

– Study of the average occupancy rate by neighbourhood in each city. The occupancy rate is calculated as follows:

Occupancy rate = T otal booked properties in a neighbourhood T otal properties in a neighbourhood ∗ 100

Booked properties are properties that had been rented for the next year when the data was extracted from the Inside Airbnb dataset in February 2020. It is considered that there are generally spontaneous or last-minute bookings, so the occupancy will be higher than the calculated, especially in the long-term. This way, the minimum average occupancy rate for next year and for every neighbourhood will be studied. This can also be used as an estimate of the occupancy map for the coming year, as no occupancy data is available for previous years.

(30)

– Evolution of the number of listings over the years. A graph will be made to study the evolution of demand using the data extracted from the reviews table.

– Seasonality in demand. For this part, three graphs will be presented, each with data from a different years (2017, 2018 and 2019). In these representations, the demand across months will be analyzed using data from the reviews table.

– Variation of Airbnb prices throughout the year to investigate if the prices of the listings follow a pattern. For this, the daily average prices of the listings over the years, located in the calendar table, will be used.

– Variation of Airbnb prices across the week. This will be studied using a box plot of average prices by day of the week.

– Occupancy Rate by Month. Occupancy over the next year will be estimated using the number of bookings for next year obtained when the data was extracted from the dataset in February 2020. In order to perform this study, the percentage occupancy for each day will be calculated, analyzing what percentage of apartments have been already booked. This estimation allows the student to get an idea about the evolution of the occupancy and can be used as an estimation of the seasonal demand since there is no available data from the occupancy of the previous years. As it happened in the study of the occupancy rate by neighbourhood, the rates obtained are an aproximation and will not be the final occupation since it is considered that there are generally spontaneous or last-minute reservations, which means that the occupancy will be higher than the calculated, especially in the long-term.

• User Review Analysis. To analyze the customer´s opinion, text mining is applied to the customer reviews and feedback from the Airbnb dataset. This analysis is performed using the tm package of R, which provides the text mining framework needed.

– Creation of a word cloud to represent the most frequent words in the reviews. This will be done using the word cloud generator package wordcloud of R, in which the words have a size proportional to their relative frequency.

– To analyze customer opinions more deeply, two more word clouds have been built with words that express what makes customers feel “comfortable” and “uncomfortable”.

This can be carried out using word vectors. Word vectors place any given word in an n-dimensional space, where the proximity of any two words in this vector space is proportional to their “similarity”.

The analysis will be performed using R, which is an open-source programming language suitable for analysis. It is very useful for this project since it allows rapid prototyping and works with the datasets to design machine learning models.

(31)

Chapter 4

Results and Discussion

In this chapter, the answers to the questions of interest mentioned in the introduction and the exploratory analysis of the Airbnb data described in the method are detailed through a variety of different figures, maps, graphs and visualizations. First, each city is studied individually and then a comparison is made between them. For each city, the study has been divided into three sections:

• Spatial Data Analysis

• Demand and Price Analysis

• User Review Mining

4.1 Stockholm

The first city analyzed is Stockholm. To begin with, a rough description of some data of the city is presented, which will be interesting for the comparison of this city with Barcelona and Rio de Janeiro:

• As mentioned before, Stockholm´s population is 974,073 [9] and there are 6,328 unique Airbnb listings in the city. To get an idea, this means that in Stockholm, there is one Airbnb property for every 154 citizens.

• The number of international visitors in 2018 was 2,604,600 [41] and it is estimated that this corresponds to half of the total number of visitors. Therefore, it is estimated that, in total, around 5 million tourists visit the city each year. This means that in Stockholm, there is one available Airbnb property for every 791 tourists.

4.1.1 Spatial Data Analysis

In this section, an analysis of the impacts of some variables using spatial visualizations is presented.

Evolution of the Number of Airbnb Properties

Intending to understand the large number of Airbnb listings in the city, the graph in figure 4.46a shows the evolution of the Airbnb properties in Stockholm over the years. In this graph, it can be appreciated that the number of Airbnb listings has increased significantly from 2010 to 2019: there were 31 properties in 2010; 2,901 in 2015 and there are currently 6,328. However, this rise is more significant from 2014 to the present.

(32)

Figure 4.1: Evolution of Airbnb listings over the years in Stockholm.

In order to visualize the rise of Airbnb properties over the years, an interactive graph has been created using the Leaflet tool in R:

(a) 2010 (b) 2015

(33)

Observing the figure 4.2c and the table below, which contains the number of Airbnb listings by neighbourhood at the moment, it can be seen that the maximum number of listings are clustered around Södermalm and Norrmalm, followed by Kungsholmen and Östermalm, which form the city center. Moreover, it can be appreciated that the mentioned central parts of the city have a higher number of properties and this number decreases as we move from the center towards the surroundings. This gradual reduction of properties follows the growth pattern that the city has followed (see frame of reference): Stockholm has grown from the old and central parts to the surroundings along the subway lines. This radial pattern is present in other aspects of the analysis and it will be explained further on.

Neighbourhood Number of listings

Södermalm 2032

Norrmalm 970

Kungsholmen 799

Östermalm 753

Hägersten-Liljeholmen 738 Enskede-Årsta-Vantör 592

Bromma 438

Skarpnäck 437

Rinkeby-Kista 304

Farsta 204

Hässelby-Vällingby 172

Älvsjö 151

Spånga-Tensta 88

Skärholmen 85

Table 4.1: Number of Airbnb properties by neighbourhood in Stockholm.

Places of Interest

The concentration of a high number of Airbnb properties around Södermalm, Norrmalm, Kungsholmen and Östermalm makes sense, especially considering that the principal touristic places are situated around these neighbourhoods, as can be appreciated in the following map, that represents the main places of interest of the city:

Figure 4.3: Principal places of interest in Stockholm.

(34)

Observing the maps 4.2c and 4.3, it can be said that the closer to the city center and the touristic places, the more Airbnb listings. The neighbourhood with the highest number of listings is Södermalm. This could be explained by its favorable geographical situation since it is very close to touristic areas. Furthermore, it is a very attractive area for Airbnb customers, since it is considered a trendy neighbourhood and the hub of creativity with many bohemian shops, cafes, vintage fashion stores and art galleries.

Presence of Hotels

Regarding the distribution of hotels in Stockholm (see figure A.1a in Appendix A), it can be seen that most of them are located in Norrmalm (in the center of the city and close to the main places of interest). Therefore, the largest number of hotels are located in Norrmalm, while the neighbourhood with the largest number of Airbnb listings is Södermalm. This makes sense, since Södermalm is a popular neighbourhood full of narrow streets with small bars and shops, whereas Norrmalm is a busy commercial area and the cultural center, characterized by wide streets crowded with people. In addition, Norrmalm has the second-highest number of Airbnb listings, which could indicate that, in the case of Stockholm, it is true that a large part of Airbnb listings are found to be in the location where hotels are situated.

Location Score by Neighbourhood

In figure 4.4, a representative map of the city with the average location score is shown. As can be seen in the representation, Norrmalm and Södermalm receive the highest location scores, followed by Östermalm and Kungsholmen. This is consistent with the analysis carried out in the previous section since these areas are located in the city center, which is very touristic and contains a high number of restaurants, shops and entertainment facilities.

It is worth mentioning that the distribution of the location score follows the expected radial pattern (related to the pattern of growth of the city, see Frame of references): the highest location scores are found at the center of the city and this score is reduced as we move away from the center to the surroundings following the metro lines. In many cases this also occurs with the number of stars of the hotels, most of the five-star hotels are situated at the center of the city (mainly in Norrmalm) and the hotels and hostels with a lower number of stars are generally located in the surroundings.

This means that, in general, the neighbourhoods where Airbnb top-rated properties are situated coincide with the location where most five-star hotels can be found.

Hence, the neighbourhood score decreases as we move away from the city center. However, the low scores in some distant neighbourhoods may be due, in addition to the distance factor, to the loss of popularity caused by several cases of violence. This is the case of Rinkeby-Kista, Älvsjo, Skärholmen and Enskede-Årsta-Vantör. These neighbourhoods contain the so-called ”vulnerable areas” or “especially vulnerable areas”, which are terms applied by police in Sweden to areas with high crime rates and social exclusion [42]. The Älvsjo case is an exception since, although it is considered a ”vulnerable area”, it is in the middle of the Stockholm rating scale. This could be caused because the ”Stockholm International Fair”, which is the biggest rentable facility in northern Europe, is located in this neighbourhood. This facility arranges trade fairs, international congresses, seminars, general assemblies and music events, which brings many people from all over the world and many of them stay in Airbnb properties.

Another factor that may influence the location score is the availability and frequency of public transport. The map introduced in Appendix B (figure B.1) contains the metro map of the city.

As can be seen, there are seven metro lines that follow an axial form: all lines go from one point of the city to another, passing through the central station (T-Centralen), which explains why the

(35)

Figure 4.4: Stockholm location score.

Price by Neighbourhood

As can be observed in figure 4.5, in many cases this map follows the previous map, since the highly rated location tend to be costly. Therefore, Söderman and Kungsholmen are the most expensive neighbourhoods (with an average price between 1223.84 and 1336.3 SEK), followed by Norrmalm and Östermalm (with an average price between 1111.37 and 1223.84 SEK).

Thirdly, Bromma has an average price between 998.9 and 1111.37 SEK. This is a high and medium-income residential neighbourhood, where the most expensive properties based on Stockholm´s average purchase price are located [43]. Fourthly, the properties situated in the south of the city, except for Skärholmen, receive an average price between 773.96 and 998.9 SEK, which coincides with the medium average score that these neighbourhoods receive.

The cheapest neighbourhoods (with an average price between 661.49 and 773.96 SEK) are those that are far from the center and also have the lowest location score: Rinkeby-Kista, Spånga-Tensta, Hässelby-Vällingby and Skärholmen.

(36)

Figure 4.5: Stockholm prices score.

Types of Listings in Stockholm

In this section, the relationship between property type and neighbourhood is studied. Since there are many types of properties, the analysis focuses on the most frequent ones in the city. The plot in figure 4.49a shows the ratio of property type and the total number of properties in the borough.

Some key observations from the bar chart are:

• The top five property types selected for the analysis are apartments, houses, townhouses, lofts and condominiums.

• Apartments are the most common type of property in all of the neighbourhoods, with a presence rate between 51.5% (Älvsjö) and 94.7% (Östermalms).

• The second most common type of properties are houses, present in all the neighbourhoods at different levels. The neighbourhoods the highest presence of this type are Älvsjö with 44.7%

and Spånga-Tensta with 34.7%. It can be noted that, in general, houses are more present on the outskirts of the city and there is a lower percentage in the central neighbourhoods such as Kungsholmens, Norrmalm, Östermalms and Södermalms.

• The third most common properties are townhouses. The neighbourhood with the largest

(37)

• Lofts are present in the central neighbourhoods and the immediate surroundings of the city.

This makes sense since construction areas in the city center are generally smaller and more expensive than in the surroundings. This matches the results obtained and explains why lofts and apartments are more common in these central parts, whereas in the surroundings there are more houses.

• Södermalm, Norrmalm, Kungsholmen and Östermalm do not have available townhouses to be rented on the webpage. This is curious, since Gamla Stan, the old town island and most touristic part of the city in Södermalm, is famous for its colorful townhouses. Norrmalm has also many characteristic townhouses, which have become a popular image of the city. The explanation of the lack of townhouses on the Airbnb webpage from these neighbourhoods could be that these types of properties are family buildings that are not in the market since they are kept from generation to generation as a symbol of family tradition. Assuming this hypothesis, this type of property is not usually rented and, if they are, they are rented in the traditional way, as regular apartments and not as Airbnb properties.

• It is worth mentioning that, in Stockholm, due to its geography of islands, boats can be rented as apartments. This type of property is a minority and is only present in those neighbourhoods with access to lakes or the sea, such as Hägersten-Liljeholmens, Östermalms, Kungsholmens and Södermalms.

Figure 4.6: Types of listings in Stockholm.

(38)

4.1.2 Demand and Price Analysis

In this section, an investigation is conducted on some of the factors that affect the demand and prices of Airbnb properties in Stockholm. To carry out this study, the variable ”number of reviews”

is used as an indicator of demand, as explained in the method.

Occupancy Rate by Neighbourhood in Stockholm

To begin with, the average occupancy rate by neighbourhood is studied:

Neighbourhood Occupancy rate Enskede-Årsta-Vantör 85.33%

Skarpnäck 85%

Hägersten-Liljeholmen 84.9%

Kungsholmen 84.7%

Norrmalm 84.5%

Östermalm 83%

Södermalm 82.5%

Farsta 81.5%

Bromma 81.3%

Älvsjö 76.8%

Spånga-Tensta 76.25%

Skärholmen 72.3%

Hässelby-Vällingby 68.5%

Rinkeby-Kista 40.9%

Table 4.2: Occupancy rate by neighbourhood in Stockholm.

(39)

It can be seen that most neighbourhoods show an average occupancy rate between 80 and 85%, which is a high rate, since it means that, on average, more than 80-85% of Airbnb properties will be occupied. The neighbourhoods with these rates are those situated in the center of the city and in the immediate surroundings, which are also those with the highest number of Airbnb properties.

Taking into account that the number of Airbnb properties in each neighbourhood is considerably different (table 4.1) and knowing the average occupancy rate of each area (table 4.2), it is concluded that the areas situated in the city center will receive more visitors, followed by nearby neighbourhoods.

Demand Across Years in Stockholm

The following graph represents the evolution of the demand of Airbnb listings in Stockholm:

Figure 4.8: Demand across years in Stockholm.

Some key observations from the bar chart are:

• The number of reviews has increased over the years, which, as discussed earlier, indicates an increase in demand.

• It can be appreciated that from 2012 to 2014 the demand hardly grew. However, from 2014 to 2018 it grew steadily with a significantly higher slope than in previous years. This slope has doubled in the last two years and shows a growing trend.

• It can be seen that the demand follows a seasonal pattern: each year there are peaks and drops, which indicates that certain months are busier compared to the others. This phenomenon is studied in the following section.

(40)

Seasonality in Demand in Stockholm

To understand seasonality, the demand is studied through months for 2017, 2018 and 2019:

Figure 4.9: Seasonality in demand in 2017, 2018 and 2019 in Stockholm.

(41)

to be a pattern of how demand fluctuates across the year: the demand increases until August and then decreases, reaching its lowest point in December. Moreover, by looking at the attached climograph in Appendix C, which represents the monthly average temperatures and monthly average precipitations in the city, it can be seen that the demand for Airbnb properties follows a similar pattern. The extremely low temperatures and the frequent snow falls between December and April reduce tourism in these months, which increases considerably in summer due to the good weather.

Prices Across the Year in Stockholm

With the aim of studying how prices vary throughout the year, the image below shows a representation of the variation of prices over the months:

Figure 4.10: Seasonality in price in Stockholm.

As can be seen in the graph, the average listing price hardly varies around 1,100 SEK. The price begins the year decreasing until March when it starts to increase until mid-September. Then it decreases again until the end of the year. This way, the highest average price occurs in August and the lowest in March. The shape of the graph is similar to the demand graph, except in January and February when prices fall while the number of reviews (indicative of demand) increases. This seems contradictory to common sense, as prices are expected to decrease with lower demand. It could be due to the assumption that the number of reviews is a reflection of demand, which might not always be the case.

Prices Across the Week in Stockholm

As can be seen in the graph, Fridays and Saturdays are slightly more expensive compared to the other days of the week. However, it can be seen that the variation is small and the city hardly shows any difference between the days of the week.

(42)

Figure 4.11: Box plot of prices by day of the week in Stockholm.

Occupancy Rate by Month in Stockholm

In this section a calendar heatmap of the occupancy rate of the Airbnb properties for 2019-2020 is presented:

Figure 4.12: Occupancy Rate by Month in Stockholm.

Some observations of the heatmap are:

• The occupancy rate appears to increase throughout the year, starting from the lowest occupancy in January to the busiest in November. This way, January and February seem to be the calmest months while November seems to be the busiest.

• It seems like there are three occupancy intervals: from January to February with low occupancy, from March to May with a medium occupancy and from June to November with a high occupancy rate.

• On this occupancy map, the highest occupancy rate can be found in the last three days of November of 2019. The high occupancy these days has its explanation on the date the data