AirBnB in Amsterdam

(1)

AirBnB in Amsterdam

Maarten M. Sukel maartensukel@gmail.com 10670602 Bachelor thesis Credits: 12 EC Bachelor Informatiekunde University of Amsterdam Faculty of Science Science Park 904 1098 XH Amsterdam Supervisor Dr. M. J. Marx ILPS, IvI Faculty of Science University of Amsterdam Science Park 904 1098 XH Amsterdam 2016-06-20

(2)

Abstract

The fast growth of AirBnB has a great impact on cities, creating both opportunities and obstacles. There is enough literature that advocates the need for new or adapted regulation, however it lacks information about AirBnB to base these regulations on. The main goal of this thesis is to give a clear view of the current state of AirBnB in Amsterdam.

For this thesis data has been scraped directly from AirBnB in Amsterdam. Data published by AirBnB will not be used. Also data about AirBnB in Berlin and Barcelona has been collected to compare these with the data of Amsterdam.

An interesting finding is that the listings in Amsterdam are mostly entire homes/apartments. In this thesis an overview of neighbor-hoods with a relatively high number of listings. Using available data about neighborhoods, it is discussed what kind of neighbor-hoods have more AirBnB listings. Interesting is that neighborneighbor-hoods with a low housing stock seem to have a relatively large amount of AirBnB listings. Neighborhoods with expensive real estate are also more expensive on AirBnB.

Amsterdam has a higher number of entire homes/apartments compared to Berlin and Barcelona. Also a small number of hosts owns a relatively large number of listings. When looking at these and several other factors there is an indication that AirBnB in Am-sterdam is more profit driven.

(3)

1 Introduction 3 1.1 Terms used . . . 3 1.2 Research question . . . 4 1.3 Overview of thesis . . . 4 2 Related Work 4 3 Data 5 3.1 Data collection . . . 5 3.2 Data correctness . . . 7 4 Methods 8 4.1 AirBnB in Amsterdam . . . 8 4.1.1 Type of residence . . . 8 4.1.2 Price . . . 9 4.1.3 Number of guests . . . 10 4.1.4 Reviews . . . 10 4.1.5 Hosts . . . 11 4.2 Neighborhoods . . . 11 4.2.1 Correlations . . . 15

4.3 AirBnB in Amsterdam compared to Berlin and Barcelona . . . . 19

4.3.1 Type of residence . . . 19 4.3.2 Price . . . 19 4.3.3 Number of guest . . . 20 4.3.4 Reviews . . . 20 4.3.5 Hosts . . . 21 4.3.6 Estimated revenue . . . 22 5 Results 23 6 Conclusions 23 6.1 Acknowledgements . . . 24 7 Reflection 24 8 References 25 A Appendix 27 A.1 Explanation BAG . . . 27

(4)

1 Introduction

AirBnB is a website launched in 2008 that allows people to rent out their home and to book from local hosts. According to AirBnB more than 2.000.000 rooms or residences have been offered in more then 34.000 cities around the world [1]. The fast growth of AirBnB is also reflected by the media attention it is receiving. Figure 1 shows the amount of Dutch media documents AirBnB has been mentioned in according to LexisNexis.

Figure 1: On the y-axis: Amount of mentions of AirBnB in Dutch media documents on LexisNexis. On the x-axis: months since 2011.

The fast growth of AirBnB has a great impact on cities, creating both oppor-tunities and obstacles. The opporoppor-tunities are mostly economic. City inhabitants can rent out their homes for visitors who are looking for a local experience. The obstacles are that homes can easily turn into hotels and thereby reduce the housing stock. The latter has a large influence on the quality of life in the cities.

Local and national governments are looking into ways to better understand the impact of and develop ways to regulate it. Yet, these governments have very little information about the number of residences and/or rooms that are being rented out through AirBnB. Even less is known about where, how and by whom these spaces are being rented out.

1.1 Terms used

Some terminology around AirBnB that will be used will be clarified.

A ’listing’ means a property or a piece of a property that is posted on AirBnB. Some different types of listings are entire homes/apartments, private rooms or shared rooms.

The term ’host’ is used to refer to a person that posted one or more properties to AirBnB.

A ’review’ is left behind by a guest. This can only happen after a guest rented a listing. It does not always happen when a listing is rented out since it is optional.

(5)

1.2 Research question

The main purpose of this thesis is to create a clear view of the situ-ation of AirBnB in Amsterdam. The following questions will be used to try to achieve this goal:

• How many and what sort of listings are there in Amsterdam?

• In what and in which kind of neighborhoods of Amsterdam are AirBnB listings?

• How does Amsterdam compare to Berlin and Barcelona when looking at AirBnB?

The main research question is:

• What is the current state of AirBnB in Amsterdam?

1.3 Overview of thesis

In section 2 some related work will be discussed. Section 3 is about how the data was collected and how the collected data will be checked for correctness.

In section 4 the data is analyzed and compared with the socio-economic conditions of the areas with AirBnB listings. Also the data about Amsterdam will be compared to the data of Berlin and Barcelona. In section 5 and section 6 the evidence collected with those methods will then be discussed in the form of results and a conclusion. Finally the thesis will be reflected upon in section 7

2 Related Work

In this section some related work will be briefly discussed.

Other work discusses regulation and policies for sharing economy software platforms as Uber and AirBnB [3]. It is suggested that law and regulation needs to be adapted to ensure that those platforms can operate legally. Similar work argues that when the market circumstances change dramatically regulation needs to be adapted as well [6].

These works give a clear view why the sharing economy should be regulated, but they do not give a data based view about AirBnB in Amsterdam. A data based view could help with insights for creating those regulations.

Two Finnish researchers interviewed 11 AirBnB-hosts and presented the results in 2014[4]. They did this interview to ’shed light on the behaviors and norms in play in a socio-technical system that fosters monetary transactions as a part of exchanges that require coordination and trust between the exchange partners’. Some conclusions are that hosts may lower the price to be able to choose out of a larger number of guests. Also when a listing has more positive reviews, the host may increase their price. This work is interesting because it gives possible explanations for the behavior of hosts.

In another study [15] the effect of an increase in AirBnB listings on hotel revenue in Texas was researched. This study answers the question that is partic-ularly interesting for others in the hospitality industry: How is AirBnB affecting

(6)

their sector’s revenues? To answer this question a data set of all Texas AirBnB listings between 2008 and 2013 was created and compared to tax revenues. The conclusion is that a 1% increase in AirBnB listings in Texas results in a 0.05% decrease in quarterly hotel revenues. Especially lower-end hotels and the ones that did not have business travelers were affected. This work is interesting be-cause it shows that AirBnB is not creating a new market without impact on others. It proofs that there is a negative impact on hotel revenue.

In London data based research about AirBnB has been done [12]. Some conclusions relevant to this research are that AirBnB listings in London tend to be in areas that are attractive and accessible by public transportation. Also these areas tend to have residents who are young, employed and born outside of the UK. Some similar methods are applied in this thesis to collect and analyze data.

In the past AirBnB has publicized data about their listings and hosts in New York. In that publication AirBnB voluntarily shared city data on a wide scale on how its hosts use the online platform. However, AirBnB made sure it was a flattering picture by removing about 1,000 listings. They where exposed because they mis-represented the one-time purge as a historical trend by the AirBnB research website InsideAirBnB.com[7]. For this thesis no data and information that has been released by AirBnB will be used to draw any conclusions. More about InsideAirBnB and the used data can be found in section 3.

3 Data

In this section it will be explained how the data for this thesis has been collected and how correct this data is.

3.1 Data collection

For collecting the data a scraper has been made. The scraper works by using the AirBnB search engine. For example the url https://www.airbnb.nl/s/ amsterdam gets scraped. However since AirBnB has a limit of 300 listings per search query this alone is not enough. To create search queries that show all listings in Amsterdam additional filters have been added. The most important one is price range. Starting from a price of e9 per night to e250 per night a query was made for every single number. After e250 it goes in ranges of more than e1 to improve scraper speed. When a query still finds more then 300 listings an additional filter gets added, this additional filter goes through all neighborhoods as displayed on AirBnB. With this additional filter the 300 listings limit was no longer an obstacle.

Some example queries:

• https://www.airbnb.com/s/Amsterdam?price_max=9&price_min=9 • https://www.airbnb.com/s/Amsterdam?price_max=40&price_min=40

• https://www.airbnb.com/s/Amsterdam?price_min=100&price_max=100&neighborhoods= Oud-West

The information was scraped directly from the HTML. The columns that have been collected are shown in table 1. In figure 2 an example of a listing page that was used for collecting the information is shown. Both information

(7)

Minimal value Maximal value Type Name

index 1 11103 int64

neighborhood inf Zeeburg object

host A.U. Taia object

reviews 0 332 int64

datetime scraped 04/29/16 12:19:16 05/01/16 20:46:55 object

borg 95 object

buildingtype Zomerhuis/Cottage object

id 20168 12779731 int64

member months 0 87 int64

loc Amsterdam, NH, Nede... Amsterdam-Zuidoost,... object

zipcode 1111 object

latitude 52.2916 52.4523 float64

roomtype Priv´e Kamer Priv´e Kamer object

price 17 549 float64

number of beds 0 1 int64

minimalstay 1 25 float64

number of bedrooms 0 0 int64

number of guest 0 10 int64

avg reviews per year 0 172 float64

hostid 10008585 999447 object

name TRANQUIL GARDEN B... Cosy houseboat ... object url https://www.airbnb.... https://www.airbnb.... object

weeklydiscount 9% object

monthlydiscount 9% object

longitude 4.69071 5.01362 float64

extrapersons Geen kosten object

cleaning cost 16 object

member since april 2011 september 2015 object description inf we provide a new lu... object street A Bataviastraat vrolikstraat object streetl a bataviastraat zuiderkerkhof object

Table 1: The columns and their minimal value, maximal value and description.

visible for the user like the name of the listing as information not directly visible like the longitude and latitude is scraped. The information not directly visible is still present in the HTML. For example the coordinates are used to create a map like the one in figure 3.

The scraper collected the data between 2016-04-29 12:13 and 2016-05-01 20:53. It took 3 days and 10 hours. The reason it took over 3 days is that AirBnB limits the amount of requests. When this limit is reached it throws a 503 error. When this happens the scraper goes into wait mode. In total the scraper collected 11.116 listings of which 8.684 are localized in Amsterdam, when looking at the displayed location on AirBnB. In table 2 it is shown how often the different city names occurred in the scraped data. An explanation for the different names could be that they have been created when a different system was used.

As well for Berlin and Barcelona data has been scraped. For scraping Berlin and Barcelona the same method has been used as for Amsterdam. Barcelona has been scraped between 2016-04-22 23:47 and 2016-04-25 04:19. Berlin has been scraped between 2016-04-18 01:07 and 2016-04-22 18:39.

(8)

Listings

Amsterdam, Noord-Holland, Nederland 8260

Amsterdam, NH, Nederland 365

Amsterdam-Zuidoost, Noord-Holland, Nederland 34

Amsterdam, Nederland 25

Table 2: Different ways Amsterdam is named on AirBnB.

Figure 2: An example of a list-ing page where the information was scraped from.

Figure 3: Some data, like the lo-cation data, has been collected by looking at the raw HTML.

3.2 Data correctness

http://insideairbnb.com/ is a website that publishes data about AirBnB. When looking at press releases from AirBnB about the number of hosts in Amsterdam the data of InsideAirBnB seems ”in the right ballpark when ex-trapolating back” [13]. To check the correctness of the data collected it has been compared to other data available from InsideAirBnB from January 2016 [5]. In table 3 it is shown what kind of overlap is found based on the listing id. There are 5696 listing ids that occur in both the data used for this research as in the data from InsideAirBnB. There was also a number of ids that did not overlap. It could be that the reason for this difference is that the data from InsideAirbnb is from January 2016 and the self scraped data is from May 2016. Also it is possible that a listing has been re added changing its id. Another reason for this is that the scraper of InsideAirBnB runs multiple times a year and also checks on older listings ids to see if they are still there. It could be possible that AirBnB hides certain listings in search results that are for example fully booked.

When getting a Jaccard similarity score on the description, name and host name between the listings that are unique for the self data and the InsideAirBnb data 78 have a higher similarity than 0,7. These listings are sometimes of a host with multiple similar listings. Some of these have been re added. Reasons for this could be bad reviews. Another reason to re add a listing is to prevent getting caught renting the residence more nights then allowed [9]. The number of reviews can be used to determine a minimal number of stayed nights.

In table 4 several scraped values are compared from both data sources. The data of InsideAirBnB contains more listings, however the scraped data will be used since more is known about how it is collected. Also the same method can be used for collecting data about Berlin and Barcelona.

(9)

Number of listings

Unique scraped data 2988

Overlap 5696

Unique InsideAirBnB 5666

Table 3: The overlap between ids of listings of the data scraped for this thesis and the data from InsideAirBnB.

Data InsideAirBnB Data Self

Percentage of complete apartment/residence 80.21% 80.7%

Average number of reviews 14.38 14.66

Average amount of minimal stay 2.69 2.432

Table 4: Some information about the data scraped for this thesis and the data from InsideAirBnB.

4 Methods

In this section evidence will be collected about AirBnB in Amsterdam to help answer the research questions.

This section is divided into 3 parts. In section 4.1 several aspects of the collected data about Amsterdam will be discussed. In section 4.2 the data will be placed in neighborhoods and compared with other data available about those neighborhoods. Finally in section 4.3 AirBnB in Amsterdam will be compared to AirBnB in Barcelona and AirBnB in Berlin.

4.1 AirBnB in Amsterdam

To give a view of AirBnB’s current state in Amsterdam several aspects will be discussed: The type of residence, the price, the number of guests, the reviews, the host and how the listings are distributed between neighborhoods.

4.1.1 Type of residence

The three different types of listings according to AirBnB [2]: • Shared rooms:

”Shared rooms are for when you would like a comfy, communal experience, and don’t mind sharing a space with others. When you book a shared room, you’ll be sharing your bedroom and the entire space with other people. These rooms work best for the flexible traveler looking for new friends and a budget-friendly stay.”

• Private rooms:

”Private rooms are great for when you prefer a little privacy, but still value a local connection. When you book a private room, you’ll have a bedroom

(10)

to yourself, but will share some spaces with others. With a private room, you’ll be able to wake up to greet your new friends in the kitchen and have the freedom to bid them adieu at bedtime.”

• Entire homes/apartments

”Entire homes/apartments are best if you’re seeking a home away from home with complete privacy and the freedom to cook breakfast in your pajamas. With an entire home/apartment, you’ll have the whole space to yourself. You can be your own host, make your own dinner, and remember to treat your listing with the respect and courtesy you would at your own home.”

The most interesting type is the entire home/apartment since this means the building is no longer being used as a residence but it is being rented out while the owner is absent. The private rooms are also interesting because with these the home owner will likely be present during the guests stay.

It has to be taken into consideration that this is just something a user on AirBnb can enter. When looking at all the listings that say they are an entire home/apartment and using a Jaccard similarity on their description, host and listing name to compare them to all other listings over 826 listings had a simi-larity of more than 0,7. Some of these are the same listing with a different price. But also a lot of them seem to be more then one listing at the location.

80,7% of residences offered on AirBnB in Amsterdam is of the type ’Entire home/apt’. 18,75% is offered as a private room. There are almost no shared rooms on AirBnB in Amsterdam. The distribution of roomtypes together with their price is visualized in figure 4.

4.1.2 Price

The average price of a listing in Amsterdam at the time of scraping wase130,00 a night. Private rooms had an average price of e52,96 a night. The average price for an entire home/apartment wase134,63. In figure 4 the distribution is visualized.

Figure 4: The distribution of prices. On the y-axis the frequency and on the x-axis the prices.

The prices seem normally distributed with some peaks around, for example e150 and e200. The price is set by the host.

(11)

4.1.3 Number of guests

In figure 5 you can see that most of the listings offer to 2 guests.

Figure 5: Distribution plot of the number of guests per listing. For example, more than 50% of all listings is offered for 2 guest.

Out of all the Amsterdam listings 8,2% offers a place to stay to more than 4 guests, this number is interesting because it is not allowed to rent out to more than 4 guests [9].

4.1.4 Reviews

In the collected data about Amsterdam there is a total of 127.345 reviews. The average number of reviews per listing is 14,7. The median is 5.

In figure 6 the distribution is shown.

Figure 6: Density plot of the number of reviews for AirBnB in Amsterdam.

The number of reviews is important for a host. A host with more reviews can ask a higher price [4]. However the risk of getting caught illegally renting a residence grows because the number of reviews can be used as evidence that a residence has been rented out more than allowed.

35,56% of the listings had zero reviews, these listings were probably not very active.

(12)

4.1.5 Hosts

In Amsterdam there are 7452 unique hosts. 7,66% of those have multiple listings when looking just at the unique accounts. It is possible that they had multiple accounts so this percentage could be higher. The 10 hosts with the most listings are good for 5,1% of the listings. The 50 hosts with the most listings are good for 8,74% of the listings.

The number of hosts that have more listings is important. This, because when a single host has a large number of listings it means that these are groups or businesses renting out residences on a large scale. More about this later on in section 4.3.5.

4.2 Neighborhoods

The neighborhoods AirBnB uses are appointed after the user enters an address. These neighborhoods are not of any use since it is not known how they are appointed and often they seem incorrect. Also it is not possible to correctly combine these neighborhoods with other data.

The street that is displayed seems to be most correct and will be used for localization. The reason for this is that the street is also where the guests go to. Using the coordinates was also an option, however these can be moved by the user to a possibly random location and there is a possibility that they are randomized after the user selected the location.

Out of the 7008 entire homes/apartments listings in Amsterdam, 6878 are combined with a street. They are located in at least 1506 streets of Amsterdam. Out of all listing types 160 did not display a street, so that information was not scraped. The addresses were cleaned by lower casing and correcting common errors. For the ones that were still not usable, a Jaccard similarity of 0.8 was used to find the most similar street name. After this 95 listing where still not matched, these were dropped.

When looking at the streets with the highest price per guest per night the most expensive street is Lindengracht. In table 5 the 20 most expensive AirBnB streets are displayed.

(13)

Street Price per guest per night Price per night Listings

Lindengracht e73.60 e185.94 19

Prinsengracht e69.58 e207.81 74

Keizersgracht e69.17 e225.79 53

Bloemgracht e66.85 e213.93 15

Bosboom Toussaintstraat e60.69 e166.91 12

Van Beuningenstraat e60.30 e155.07 14

Bloemstraat e60.11 e172.33 15

Nieuwe Looiersstraat e59.45 e178.36 11

Willemsparkweg e58.96 e187.29 17

Brouwersgracht e58.04 e192.47 38

Herengracht e57.83 e194.54 33

Plantage Muidergracht e57.26 e203.61 18

Rozengracht e57.20 e178.75 16

Amstel e57.08 e200.53 39

Westerstraat e56.50 e182.53 13

Lange Leidsedwarsstraat e56.31 e152.85 14

Kloveniersburgwal e56.15 e217.13 15

Singel e55.98 e264.55 40

Oudezijds Voorburgwal e55.87 e180.31 22

Nicolaas Maesstraat e55.65 e263.09 11

Table 5: The 20 most expensive AirBnB streets when looking at price per guest per night.

For comparing the number of listings with the total number of residences all the listings were put into 482 neighborhoods of Amsterdam and there corre-sponding neighborhood combinations, for this [11] [8] and [10] have been used. Data from 2015 has been used since 2016 was not available. 344 of the 482 neighborhoods have entire homes/apartments listings in them.

To see how many listings there are in Amsterdam a listing density has been calculated for every neighborhood and for every neighborhood combination. This has been calculated by dividing the number of listings that are shown as entire homes/apartments in the neighborhood with the number of addresses in the neighborhood that are registered as a residence: ’Addresses that have been registered in the ’Basisregistraties Adressen en Gebouwen’ (BAG) that are residences, except living boats’. Although not all entire homes/apartments are really a residence, see section 4.1.1. When comparing different neighborhoods it still gives a clear view.

In figure 7 the 20 neighborhoods combinations with the highest density of entire homes/apartments is shown. Only neighborhoods with more than 10 listings are taken into account.

Also 1387 out of 1628 private rooms have successfully been put in their respective neighborhood and divided by the total number of residences in the area to get a private room density. For putting private rooms into neighborhoods the same process has been used as for the entire homes/apartments. In figure 8 the 20 neighborhoods with the highest private room density are shown.

(14)

Figure 7: The neighborhood combinations with the highest density of entire homes/apartments. Only neighborhoods more than 10 entire homes/apartments listings are taken into account. For example: in ”E22 Vondelbuurt” about 7% of the number of residences is equal to the number of entire homes/apartments listings in ”E22 Vondelbuurt”.

Figure 8: The neighborhood combinations with the highest density of private rooms. Only neighborhoods with more than 10 private rooms listings are taken into account. For example: in the neighborhood combination 2% of the num-ber of residences is equal to the numnum-ber of private rooms in the neighborhood combination.

In figure 9 the neighborhoods and their corresponding entire homes/apartments listing density have been plotted on a map of Amsterdam. The city center has the highest listing density. It has to be taken into consideration that the surface has no meaning on this map and only the colors of a neighborhood represent the listing density.

(15)

Figure 9: The neighborhoods and their entire homes/apartments listing den-sity. Neighborhoods that have less then 100 residences according to the BAG [8] are grey. Also neighborhoods with no data are grey.

In figure 10 the difference in average prices between the neighborhoods is shown. The city center seems most expensive. The expensive neighborhoods on the edge of the city can be explained by a low number of listings in those neighborhoods.

(16)

Figure 10: A map of Amsterdam with the average listing price in each neigh-borhood. The prices are in euros.

4.2.1 Correlations

In this section there will be looked at socio-economic conditions and how they correlate with data collected about AirBnB in Amsterdam. When looking at the neighborhoods it becomes possible to correlate with other data. All data used is from the BAG Amsterdam [8]. The exact definition of the used data can be found in the Appendix A.1.

When using a Pearson correlation the following correlating variables have a correlation stronger than 0.3 with listing density:

Ratings

• VKparkeren-r -0.342988

In neighborhoods where residents gave a high score to parking facilities the listing density is lower. This is probably because the neighborhoods with bad parking facilities are located in the city center.

• Lomganggroepenb-r 0.301080

The score people in a neighborhood gave to how different groups are in-teracting with each other has a correlation with listing density. This could be explained by the increased interaction with tourists.

• ORwoningenmooi-r 0.330272

The score that has been given to the outside ’prettiness’ of residences in the area. The higher the score the higher the listing density. This could

(17)

be explained by the fact that the most tourist areas are the ones that are considered the most ’pretty’.

• Lthuisvoelen-r 0.342540

The score that people gave to how at home they feel in their neighborhood is higher in neighborhoods with a higher listing density.

• Vveiligavond-r 0.352526

The neighborhoods in which a higher score has been given for feeling safe at night also have a higher listing density.

• Vveiligdag-r 0.386156

The neighborhoods in which a higher score has been given for feeling safe during the day also have a higher listing density.

• Lbuurt-r 0.371080

The score residents gave to their own neighborhood is higher in neighbor-hoods with a higher listing density.

• Lverwachtingbuurt-r 0.334168

The score residents gave to how the neighborhood is developing is higher in neighborhoods with a higher listing density.

From these ratings can be concluded that in the neighborhoods where resi-dents feel more positive about their own neighborhood there is a higher listing density.

• Bhvestc-p 0.318158

The number of industrial companies correlates with the listing density. The reason for this is that in neighborhoods with a lot of industry often there are not as much residences creating a high listing density.

• Wwoz-m2 0.379300

The estimated value of a square meter is higher in the neighborhoods with a higher listing density.

• BHwp-1000inw 0.413433

In the neighborhoods with a higher listing density a larger part of the population has a job.

• BHwpc 0.469483 and BHwpc-p 0.546705

In the neighborhoods with a higher listing density a larger part of the population has a job working in industrial companies.

• Wbezet 0.559201

Neighborhoods with a higher listing density also have a higher residence occupancy rate.

• SRcultuur-1000inw 0.576118

(18)

• SRsport-1000inw 0.579097

Neighborhoods with a higher listing density have more sporting facilities. • BHwinkDG-1000inw 0.586925

Neighborhoods with a higher listing density have more shops with daily goods.

When a Pearson correlation is done with the average listing price in a neigh-borhood the following correlations are higher than 0.3 or lower than -0.3.

• Lbuurt-r 0.311661

The score residents gave to their own neighborhood is higher in neighbor-hoods with a higher average listing price.

• Lthuisvoelen-r 0.314374

Where residents gave a higher score to feeling at home the listings have a higher price.

• Wonderhoudwoning-r 0.359171

The score that is given to the maintenance state of the residents own home.

• ORomgevingmooi-r 0.370145

The score of the outside of the living environment. This score is higher when the average listing price are higher.

• Lomganggroepenb-r 0.418690

The score people gave to how different groups in a neighborhood live together. This score is higher when the average price for listings is higher. • Wwoning-r 0.419763

In neighborhoods where people rate their own residence with a higher score the prices of the listings are higher.

• ORwoningenmooi-r 0.428777

The score that is given to how beautiful the houses in the neighborhood look. In neighborhoods where the houses are rated higher, also a higher average listing price occurs.

• ORonderhoudwoningen-r 0.432436

The score that is given to how residences in the neighborhood are main-tained. This score is higher when the average listing price is higher.

The above correlations between how nice and safe residents find their neigh-borhood and the average listing price in the neighneigh-borhood indicates that when residents find their neighborhood more appealing, a higher price is or can be asked.

(19)

• Iwwb -0.342918 and Iwwb12 -0.337542 and Iwwb-p -0.335744 and Iwwb12-p -0.308981

Neighborhoods in which more residents receive government aid have a lower average price.

• Wcorhuur -0.324874

In neighborhoods with more housing corporation’s residences have a lower average price.

• BeveenouderHH -0.309616

Neighborhoods with more one parent families have a lower average price. • Bev65NW -0.306309

Neighborhoods where there are more citizens older than 65 with a non-western background have a lower average price.

The above correlations show that neighborhoods with housing for these groups are neighborhoods where listings have a lower average price.

• Wwoz-gem 0.314247

Neighborhoods in which the average house price is higher also have a higher average price on AirBnB.

• BHvest-bank 0.326156

In neighborhoods with more banks the prices of listings are higher. • Bhvestl-p 0.327558

In neighborhoods with more real estate businesses the prices of listings are higher.

• number of guest 0.697764

A higher number of guests has a strong correlation with the price. The more guest allowed in a listing, the higher the price. This is a very logical correlation since often a price is per guest.

The most interesting of above is that a higher house price in a neighborhood correlates with a higher price on AirBnB.

When correlating the private rooms interesting correlations that differ from the above correlations with entire homes/apartments are the following:

• BevNstedeling-p 0.434411

The number of new urbanites is higher in neighborhoods with more private rooms.

• BevWest-p 0.423138

Neighborhoods with more private rooms have a higher percentage of West-ern residents.

(20)

4.3 AirBnB in Amsterdam compared to Berlin and Barcelona

In this section Amsterdam will be compared to Berlin and Barcelona. From Berlin 10.279 listings have been scraped. From Barcelona 10.368 listings have been scraped. The Berlin data has been scraped before it became forbidden to rent out entire homes/apartments through AirBnB without a license on the first of May 2016. When the data was filtered to make sure only listings in the city are used, Barcelona had 5.467 listings within the city boundaries, Berlin had 9.903 and Amsterdam had 8.684.

4.3.1 Type of residence

With 80,7% being entire homes/apartments Amsterdam has a much higher per-centage of listings that are an entire home/apartment that Barcelona or Berlin. Berlin is second with 59,63% percent and Barcelona is third with 52,53%.

Figure 11: The difference in type of listings in the cities Amsterdam, Berlin and Barcelona

4.3.2 Price

When looking at the mean of the prices Amsterdam is the most expensive with an average price of e130. In Berlin the average price is e63 and in Barcelona the average price is e91. In figure 12 the distribution of the price is shown. Berlin has a lot of low priced listings. Barcelona has both low priced, medium priced and high priced listings. Amsterdam has a lot of medium and high priced listings. The low prices in Berlin are partly because there are a larger number of private listings. Also the prices in Berlin are lower overall.

(21)

Figure 12: The difference in prices in the cities Amsterdam, Berlin and Barcelona. On the y-axis is the frequency a price occurred, color is what city and on the x-axis is the price in euros in the beginning of May 2016.

4.3.3 Number of guest

In Barcelona 38,4% is offered for 2 guests. In Berlin this is 48,64%. In Amster-dam this is 53%.

In Amsterdam 8,18% of listings offers a place for more than 4 guests. In Berlin this is 10,93%. In Barcelona it is much higher: 22,74%. This is interesting because it is not allowed to rent to more than 4 guests in Amsterdam [9].

In figure 13 it is shown for all number of guest.

Figure 13: On the x-axis: the number of guest, on the y-axis: the percentage that this number is guest is offered for.

4.3.4 Reviews

When looking at table 6, listings in Amsterdam have the highest number of reviews per listing. Also the median is higher. Listings in Amsterdam seem to be receiving more reviews than listings in Barcelona and Berlin.

When looking at the number of listings with zero reviews Amsterdam rela-tively has a lower number of these listings compared to Barcelona and Berlin.

(22)

City Zero reviews% Mean of reviews Median of reviews

Amsterdam 35.56% 14.66 5

Barcelona 45.37% 10.73 3

Berlin 44.92% 12.56 3

Table 6: Some information about the number of reviews per listing in each city.

Figure 14: Density plot of reviews in Amsterdam, Barcelona and Berlin.

4.3.5 Hosts

Amsterdam has the lowest percentage of hosts that have more than one listing, 7,66%. For Berlin this number is 8,64% and for Barcelona this number is 18,78%. Interesting is looking at the top 1% of the hosts with the highest number of listings. In Amsterdam they have 8,51% of the listings. In Barcelona they have 9,75% of the listings and in Berlin they have 4,88% of the listings.

In figure 15 the top 50 hosts of all three cities are shown. In Amsterdam there is a very small number of hosts that have a large number of listings. These hosts all seem to be ’key company’ that help rent people out there homes while they are absent for a fee. In Barcelona there are a lot of hosts with more than one listing, however these seem to be more distributed than in Amsterdam. A report of Penn State [14] concludes that in 12 large cities in the USA multi-unit hosts are good for almost 40% of the revenue made with AirBnB. In Amsterdam the multi-unit hosts at least have a large part of the cities listings.

(23)

Figure 15: The 50 hosts with highest amount of listings. For the three separate cities. On the x-axis the hosts and on the y-axis the number of listings per host.

4.3.6 Estimated revenue

For the data collected about each city the minimal number of stayed nights has been calculated. This is the sum of the number of reviews multiplied by the minimal stay for each listing.

• Amsterdam: 265.938 minimal stayed nights • Barcelona: 104.900 minimal stayed nights • Berlin: 264.341 minimal stayed nights

It is not know when these stays were. When multiplying the minimal number of stays with the price of each of the listing also a revenue can be calculated. The following minimal revenue’s can be calculated for each city:

• Amsterdam: e31.722.680 • Barcelona: e7.114.446 • Berlin: e17.303.715

This is not a correct number about the revenue made in Amster-dam. This is a minimal revenue based on the scraped data. It is not know when this revenue is made.

It has to be taken into consideration that minimal nights and prices can be changed at any time so these numbers are not very accurate. Also listings could be missing in this information. The price could have been different at the time of the actual stay. However these numbers can be used for comparing the cities. When comparing Barcelona with Amsterdam about four and a half times more revenue is made in Amsterdam.

Amsterdam has almost double of the revenue of Berlin despite them having about the same number of listings and stayed nights. The prices in Amsterdam are simply higher resulting in this higher revenue.

(24)

5 Results

How many and what sort of listings are there in Amsterdam?

When looking at the collected data, 8.684 Amsterdam listings were found on AirBnB in the beginning of May 2016. However throughout the year there are more than this since not all listings are online at the same time.

The largest part of the listings that are being offered in Amsterdam are entire/homes or apartments, about 4 out of 5. The rest of the listings are private rooms. There are almost no shared rooms in Amsterdam.

The average price of listings in Amsterdam ise130,00 a night. For private rooms the average price is e52,96 a night. For entire homes/apartments the average price ise134,63. These prices are also from the beginning of May 2016. When looking at reviews 35,56% of the listings had no reviews, and the average number of reviews is 14,7. The median number of reviews in Amsterdam is 5.

When looking at the host there are not many with multiple listings, however there is a small number of hosts whom had a large number of listings. This indicates that there is a large number of small businesses behind renting out listings in Amsterdam.

In what and in which kind of neighborhoods of Amsterdam are AirBnB listings?

Neighborhoods located in tourist areas, especially the city center seem to have the highest listing density. Neighborhoods where residents give higher ratings to how nice and safe the neighborhood is also have a higher listing density. Also the neighborhoods that are expensive residential wise and with a high residence occupancy rate have a higher listing density.

How does Amsterdam compare to Berlin and Barcelona when look-ing at AirBnB?

Interesting is that the number of listings that are an entire home/apartment is higher in Amsterdam.

When looking at the difference in pricing Amsterdam is the most expensive of the cities with more listings in the middle and high segment. Berlin has more listing in the low segment, which can partly be explained by Berlin having more private rooms. But also Berlin is cheaper overall. In Barcelona the listings seem to be more distributed price wise.

Also the listings in Amsterdam seem to be receiving more reviews. The re-views can be an indication that the listings in Amsterdam have a higher number of stayed nights. However, it could also mean that the hosts in Amsterdam more actively try to get more reviews to attract more customers and/or raise their prices.

The small number of hosts in Amsterdam that have a large number of listings is much higher than in Barcelona and Berlin. This indicates that a large part of AirBnB in Amsterdam is ran by businesses that use the platform to make a profit.

6 Conclusions

Main research question for this thesis was ”What is the current state of AirBnB in Amsterdam?” The other research questions helped to get an answer on this.

(25)

From the collected evidence the following conclusions can be made.

The neighborhoods with a high listing density are the ones that have more expensive real estate. Neighborhoods with a high listing density also have a higher residence density. This is particularly interesting because of the already small housing stock in these areas. Since the listing density was calculated based on only entire homes/apartments the renting through AirBnB is likely to further decline the number of available space that can be used for living.

The tourist areas of the city are also the areas where relatively more listings are located.

When compared to Barcelona and Berlin, the prices, as well as the number of reviews for each listing, are higher in Amsterdam.

The small number of hosts with a large amount of listings is also an interest-ing difference between the cities. When combininterest-ing the large number of listinterest-ings that are an entire home/apartment with the small number of host that have a large number of listings it can be concluded that AirBnB in Amsterdam has the most profit-making character of these cities.

6.1 Acknowledgements

I would like to express my very great appreciation to Maarten Marx for fulfilling his role as supervisor as good as he did during the creation of this thesis. Also I would like to thank Juan-Carlos Goilo for the help he has given in writing this thesis. Finally, I’d like to thank everyone who has supported me in anyway.

7 Reflection

The biggest imperfection of this thesis is that because of the short time span the data collected is just a snapshot. However it should still be helpful in creating a better view of the current state of AirBnB in Amsterdam. In the future more data will be available which will allow an even better view of the state of AirBnB.

In section 4.1.1 and other sections where room types are used a problem is that the room type is user generated. It is possible users are not honest when filling in the type of listing they are renting out. However the larger part of these seem to be filled in correctly.

In section 4.2 some listings could not be matched with a neighborhood. The number of listings that could not be matched with a neighborhood is not large enough to be an issue. What could be an issue is that some data used for correlating with AirBnB data is of slightly different zones (for example Iwwb-p). However this is the most accurate data available.

Section 4.3 has the problem that when scraping the 3 cities this happened in linear order. It might possible the amount or the kind of listings that are found on for example a different week day is different.

For calculating the estimated revenues in section 4.3.6 it strongly has to be taken into account that these numbers are just a rough estimate. The prices are likely to be changed since the actual renting out has happened. Also the minimal number of stay could have changed. For getting a better estimate more data is required instead of a single snapshot.

(26)

8 References

[1] AirBnB. Airbnb about page. https://www.airbnb.com/about/about-us. Accessed: 27/5/2016.

[2] AirBnB. Airbnb room types. https://www.airbnb.com/help/ article/5/what-does-the-room-type-of-a-listing-mean. Accessed: 28/5/2016.

[3] Benjamin G Edelman and Damien Geradin. Efficiencies and regulatory shortcuts: How should we regulate companies like airbnb and uber? Har-vard Business School NOM Unit Working Paper, (16-026), 2015.

[4] Tapio Ikkala and Airi Lampinen. Defining the price of hospitality: net-worked hospitality exchange via airbnb. In Proceedings of the companion publication of the 17th ACM conference on Computer supported cooperative work & social computing, pages 173–176. ACM, 2014.

[5] InsideAirBnB. Amsterdam data. http://insideairbnb.com/ get-the-data.html. Accessed: 2016-04-01.

[6] Christopher Koopman, Matthew Mitchell, and Adam Thierer. Sharing economy and consumer protection regulation: The case for policy change, the. J. Bus. Entrepreneurship & L., 8:529, 2014.

[7] C. Murray and T. Slee. How airbnbs data hid the facts

in new york city. http://insideairbnb.com/reports/

how-airbnbs-data-hid-the-facts-in-new-york-city.pdf, 2016. Accessed: 2016-05-01.

[8] City of Amsterdam. basisbestand gebieden

amster-dam. https://www.ois.amsterdam.nl/online-producten/

basisbestand-gebieden-amsterdam. Accessed: 2016-05-01.

[9] City of Amsterdam. Faq renting. https://www.amsterdam.nl/ veelgevraagd/?caseid=%7B9B2C2273-F797-460B-AD20-05DFB9F6F39F% 7D. Accessed: 31/5/2016.

[10] City of Amsterdam. Nieuwe gebiedsindeling voor

amster-dam. http://www.ois.amsterdam.nl/nieuwsarchief/2015/

nieuwe-gebiedsindeling-voor-amsterdam. Accessed: 2016-05-01. [11] City of Amsterdam. sbk openbare ruimte. https://www.amsterdam.

nl/publish/pages/374843/20151008_sbk_openbare_ruimte.xls. Ac-cessed: 2016-05-01.

[12] Giovanni Quattrone, Davide Proserpio, Daniele Quercia, Licia Capra, and Mirco Musolesi. Who benefits from the sharing economy of airbnb? In Pro-ceedings of the 25th International Conference on World Wide Web, pages 1385–1394. International World Wide Web Conferences Steering Commit-tee, 2016.

[13] Tom Slee. Airbnb data collection: Methodology and accuracy. http:// tomslee.net/airbnb-data-collection-methodology-and-accuracy. Accessed: 2016-06-09.

(27)

[14] Penn State. From air mattresses to unregulated business: An analysis of the other side of airbnb. http://www.ahla.com/uploadedFiles/_Common/ pdf/PennState_AirBnbReport_.pdf. Accessed: 2016-10-6.

[15] Georgios Zervas, Davide Proserpio, and John Byers. A first look at online reputation on airbnb, where every stay is above average. Where Every Stay is Above Average (January 23, 2015), 2015.

(28)

A

Appendix

A.1 Explanation BAG

• VKparkeren-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Wat vindt u van het aanbod van parkeervoorzieningen in uw buurt? (1=ruim onvoldoende, 10= ruim voldoende).

Er wordt alleen gerapporteerd over gebieden met minstens 20 responden-ten.

• Lomganggroepenb-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Hoe gaan verschillende groepen mensen in uw buurt met elkaar om? (1=zeer onprettig en 10=zeer prettig).

• ORwoningenmooi-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Hoe beoordeelt u de woningen in uw buurt? (1=zeer lelijk, 10 =zeer mooi).

• Lthuisvoelen-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Voelt u zich thuis in uw buurt? (1= helemaal niet thuis, 10 = zeer thuis). Er wordt alleen gerapporteerd over gebieden met minstens 20 responden-ten.

• Vveiligavond-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Hoe veilig voelt u zich ’s avonds in uw buurt? (1=zeer onveilig, 10=zeer veilig).

• Vveiligdag-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Hoe veilig voelt u zich overdag in uw buurt? (1=zeer onveilig, 10=zeer veilig).

• Lbuurt-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Hoe tevreden bent u met uw buurt? (totaaloordeel) (1=zeer ontevreden, 10=zeer tevreden).

(29)

• Lverwachtingbuurt-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Hoe denkt u dat de buurt waar u woont zich de komende jaren zal on-twikkelen? (1=zeer negatief, 10 =zeer positief).

• Bhvestc-p

Aantal vestigingen met de sectiecode C: Industrie.

Een vestiging is een locatie van een door de Kamer van Koophandel geregistreerde onderneming waarin of van waaruit een economische ac-tiviteit wordt uitgeoefend voor minimaal 12 uur per week door minimaal ,,n werkzaam persoon.

Sinds 1 juli 2008 zijn ook vrije beroepen verplicht zich in te schrijven in het Handelregister. In 2008 leidde dit tot een toename van het aantal geregistreerde artsen en advocaten en in 2009 van de creatieve beroepen. • Wwoz-m2

WOZ-waarde per m2 • BHwp-1000inw

Aantal werkzame personen (banen van minstens 12 uur) in het gebied per 1.000 inwoners.

Sinds 1 juli 2008 zijn ook vrije beroepen verplicht zich in te schrijven in het Handelregister. In 2008 leidde dit tot een toename van het aantal geregistreerde artsen en advocaten en in 2009 van de creatieve beroepen. • BHwpc

Aantal personen die voor 12 uur of meer per week werkzaam zijn bij vestigingen met de sectiecode C: Industrie.

Sinds 1 juli 2008 zijn ook vrije beroepen verplicht zich in te schrijven in het Handelregister. In 2008 leidde dit tot een toename van het aantal geregistreerde artsen en advocaten en in 2009 van de creatieve beroepen. • BHwpc-p

Aandeel van het totaal aantal werkzame personen dat voor 12 uur of meer per week werkzaam is bij een vestiging met de sectiecode C: Industrie. Sinds 1 juli 2008 zijn ook vrije beroepen verplicht zich in te schrijven in het Handelregister. In 2008 leidde dit tot een toename van het aantal geregistreerde artsen en advocaten en in 2009 van de creatieve beroepen. • Wbezet

(30)

• SRcultuur-1000inw

Aantal cultuurvoorzieningen per 1. 000 inwoners. Het gaat om het aantal in het ARRA geregistreerde cultuurvoorzieningen met de SBI-codes 5914, 85521, 85522, 90011, 90012, 90013, 90041, 91011, 91012, 91019, 91021 en 91022.

Deze codes staan voor Bioscopen; Dansscholen; Kunstzinnige vorming van amateurs (geen dansscholen); Beoefening van podiumkunst; Producenten van podiumkunst; Circus en variete; Theaters en schouwburgen; Open-bare bibliotheken; Kunstuitleencentra; Overige culturele uitleencentra en openbare archieven; Musea; Kunstgalerieen en expositieruimten.

Sinds 1 juli 2008 zijn ook vrije beroepen verplicht zich in te schrijven in het Handelregister. In 2008 leidde dit tot een toename van het aantal geregistreerde artsen en advocaten en in 2009 van de creatieve beroepen. • SRsport-1000inw

Aantal in het ARRA geregistreerde vestigingen met de SBI-code 931 (Sport) per 1.000 inwoners.

Sinds 1 juli 2008 zijn ook vrije beroepen verplicht zich in te schrijven in het Handelregister. In 2008 leidde dit tot een toename van het aantal geregistreerde artsen en advocaten en in 2009 van de creatieve beroepen. • BHwinkDG-1000inw

Aantal winkelruimtes voor dagelijkse goederen per 1.000 inwoners. • Lbuurt-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Hoe tevreden bent u met uw buurt? (totaaloordeel) (1=zeer ontevreden, 10=zeer tevreden).

• Lthuisvoelen-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Voelt u zich thuis in uw buurt? (1= helemaal niet thuis, 10 = zeer thuis). Er wordt alleen gerapporteerd over gebieden met minstens 20 responden-ten.

• Wonderhoudwoning-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Hoe beoordeelt u de staat van onderhoud van uw woning?

• ORomgevingmooi-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Hoe beoordeelt u de inrichting van de woonomgeving in uw buurt? (1=zeer lelijk, 10 =zeer mooi).

(31)

• Wwoning-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Wilt u aangeven in welke mate u tevreden bent over uw woning? (to-taaloordeel).

• ORwoningenmooi-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Hoe beoordeelt u de woningen in uw buurt? (1=zeer lelijk, 10 =zeer mooi).

• ORonderhoudwoningen-r

Gemiddeld rapportcijfer dat bewoners geven als antwoord op de vraag: Hoe beordeelt u de staat van onderhoud van de straten en stoepen in uw buurt? (1= ruim onvoldoende, 10 = ruim voldoende).

• Iwwb

Aantal uitkeringen WWB levensonderhoud (Meer dan 9) • Iwwb12

Aantal uitkeringen WWB levensonderhoud met grote afstand tot de ar-beidsmarkt (Meer dan 9)

• Iwwb-p

Aandeel 15 tot en met 64-jarigen dat WWB Levensonderhoud ontvangt (Meer dan 9 bijstandsgerechtigden).

De cijfers op buurtcombinatieniveau van 2014 en eerder zijn gebaseerd op de gebiedsindeling van 2010. Met uitzondering van het samenvoegen van Duivelseiland en het Museumkwartier zijn de grenswijzingen per 1-1-2015 buiten beschouwing gelaten.

• Iwwb12-p

Aandeel 15 tot en met 64-jarigen dat WWB Levensonderhoud ontvangt en een grote afstand tot de arbeidsmarkt heeft.

• Wcorhuur

Adressen die op de peildatum zijn geregistreerd in de Basisregistraties Adressen en Gebouwen (BAG) als woning met een eigendomsverhouding ’eigenaar-bewoner’.

• BeveenouderHH

Het huishoudentype ’eenoudergezin’ wordt bepaald op basis van de re-latiecodes tussen bewoners die op de peildatum in de Basisregistraties

(32)

Adressen en Gebouwen (BAG) zijn geregistreerd binnen een adres. Dit type huishouden bestaat uit een meerderjarige bewoner en een of meerdere ongehuwde bewoners die wettige kind(eren) zijn van de meerderjarige be-woner op dit adres.

• Bev65NW

Aantal Amsterdammers met een niet-westerse herkomst van 65 jaar en ouder

• Wwoz-gem

Door de gemeente (Dienst Belastingen) periodiek getaxeerde waarde van onroerende zaken in het kader van de Wet waardering onroerende za-ken (WOZ). De WOZ-waarde wordt vastgesteld op basis van het peiljaar voorafgaand aan de peildatum.

De cijfers op buurtcombinatieniveau over 2014 zijn gebaseerd op de ge-biedsindeling 2010. De grenswijzigingen per 1-1-2015 zijn buiten beschouwing gelaten.

• BHvest-bank

toegekend op basis van de activiteitencode (SBI) waarmee deze vestiging is geregistreerd bij de Kamer van Koophandel. Dit zijn vestigingen met de SBI-codes van financile instellingen (64).

De deelfunctie bank valt onder de hoofdfunctie kantoren.

Een vestiging is een locatie van een door de Kamer van Koophandel geregistreerde onderneming waarin of van waaruit een economische ac-tiviteit wordt uitgeoefend voor minimaal 12 uur per week door minimaal n werkzaam persoon. Het aantal vestigingen wordt bepaald op een peild-datum.

Sinds 1 juli 2008 zijn ook vrije beroepen verplicht zich in te schrijven in het Handelregister. In 2008 leidde dit tot een toename van het aantal geregistreerde artsen en advocaten en in 2009 van de creatieve beroepen. • Bhvestl-p

De deelfunctie ’politie en brandweer’ wordt aan een vestiging toegekend op basis van de activiteitencode (SBI) waarmee deze vestiging is geregistreerd bij de Kamer van Koophandel. Dit zijn vestigingen met de SBI-codes van landbouw, visserij, delftstoffenwinning en overig. Een vestiging is een lo-catie van een door de Kamer van Koophandel geregistreerde onderneming waarin of van waaruit een economische activiteit wordt uitgeoefend voor minimaal 12 uur per week door minimaal n werkzaam persoon. Het aantal vestigingen wordt bepaald op een peilddatum.

De deelfunctie politie / brandweer valt onder de hoofdfunctie Voorzienin-gen.

Sinds 1 juli 2008 zijn ook vrije beroepen verplicht zich in te schrijven in het Handelregister. In 2008 leidde dit tot een toename van het aantal geregistreerde artsen en advocaten en in 2009 van de creatieve beroepen.

(33)

• BevNstedeling-p

Nieuwe stedelingen zijn personen van autochtone of westerse herkomst in de leeftijd van 18 tot en met 54 jaar die na hun 18e levensjaar geregistreerd zijn in de gemeente Amsterdam.

• BevWest-p

De herkomstgroep westers bestaat uit personen die op de peildatum gereg-istreerd zijn in de gemeente Amsterdam met geboorte in een westers land of van wie een van de ouders is geboren in een westers land. Het geboorte-land van de moeder is bepalend voor de herkomstgroep, tenzij de moeder in Nederland is geboren. In dat geval wordt de herkomsgroep bepaald aan de hand van het geboorteland van de vader. Westerse landen zijn landen in Europa (exclusief Turkije en Nederland), Noord-Amerika en Oceanie, of Indonesie of Japan.

(34)

AIRBNB IN

AMSTERDAM

B A C H E L O R S C R I P T I E M A A R T E N S U K E L 2 3

TERMEN

•Listing •Host •Review 4

RESEARCH QUESTIONS

The main purpose of this thesis is to create a clear view of the situation of AirBnB in Amsterdam.

• How many and what sort of listings are there in Amsterdam?

• In what and in which kind of neighborhoods of Amsterdam are AirBnB listings?

• How does Amsterdam compare to Berlin and Barcelona when looking at AirBnB?

The main research question is:

• What is the current state of AirBnB in Amsterdam?

5

INHOUD

•Data

•AirBnB in Amsterdam

•Verdeling over buurten

•Amsterdam, Berlijn en Barcelona

•Conclusie 6

DATA

8

DATA

AIRBNB IN

AMSTERDAM

(35)