The spatial inequality of Airbnb revenue in Amsterdam

(1)

The spatial inequality of Airbnb revenue in

Amsterdam

Livia Blonk 11116994 GIS methods

Sociale Geografie en Planologie Dhr. Dr. Rowan Arundel

Universiteit van Amsterdam 11.01.2019

(2)

2

Chapter 1: Introduction

Since its foundation in 2008, the worldwide popularity of Airbnb grew exponentially (Wachsmuth and Weisler, 2018). With more than five million listings in more than 81 thousand cities, the online platform for the rental and booking of private accommodations became one of the most popular ways to travel and discover new places worldwide (Airbnb Press Room, 2018a). In fact, according to the Dutch public broadcaster NOS (Mulder, 2018), increasingly more tourists coming to the Netherlands choose to book an Airbnb rather than a hotel room. At the same time, Airbnb became an easy way for hosts to earn important amounts of money that significantly impact their income (Oskam and Boswijk, 2016).

While the extra Airbnb revenue seems to help many families worldwide to make ends meet, in recent years some critical academics started to question whether low-income, less-educated and people belonging to ethnic minorities (people with a low socio-economic status, SES) also benefit from this business. The answer seems to be clear: even though, in some cases, people with a lower SES seem to participate more as hosts on Airbnb, they always seem to earn less revenue compared to people with a higher SES. In fact, both Airbnb participation and revenue are largely dependent on people’s possibilities and willingness to share their dwellings on Airbnb, which in turn are strongly related to their SES. In short, the opportunities to earn extra money through Airbnb are not equal for everyone, as they are bigger for people with a higher SES (Schor, 2016). As a consequence, this disparity in Airbnb revenue among different socio-economic groups can lead to a further increase in income and other types of inequality in society. According to previous research, this phenomenon is problematic as it can lead to several negative consequences (e.g. Tammaru et al., 2015).

Most of the previous studies, however, have been carried out in the USA, where the socio-economic differences between the people are quite substantial compared to the levels in other parts of the ‘Western’ world (OECD, 2017). Studies carried out in a non-American context, instead, either focussed only on Airbnb participation (such as Quattrone et al., 2016) or were focussed on other types of Airbnb-related consequences such increases in rental prices, income opportunities, nuisance or ethnic discrimination (e.g. Gurran and Phibbs, 2017; Laouenan and Rathelot, 2017). As no research to date on the relation between Airbnb revenue and SES has been carried out in a social context with lower levels of inequality, the aim of this research is to fill this knowledge gap. In order to do so, this study will try to find a relationship between the revenue through Airbnb and SES on a neighbourhood level in Amsterdam in 2017. While other researches have focussed more on the relationship between hosts’ SES and Airbnb participation and revenue, this research looks at the same relationship on a neighbourhood level. The reasons for which Amsterdam has been chosen as a case study and why the neighbourhood level has been chosen as a research unit, will be explained further in this paper. The main research question is formulated as follows:

‘To what extent is there a relationship between the revenue generated through Airbnb and the socio-economic characteristics of the different neighbourhoods in Amsterdam in 2017?’

(5)

5 This research is relevant for mainly two reasons. From a societal perspective, it can contribute to a better understanding of the consequences of Airbnb in the city of Amsterdam. In fact, since research has shown that Airbnb strengthens socio-economic inequalities in several other urban areas, it is important to know if this is happening in Amsterdam as well. From an academic perspective, this research contributes an in-debt study on the relation between SES and Airbnb revenue in a city with relatively low levels of inequality and socio-economic segregation like Amsterdam (Musterd, 2005).

The paper is structured as follows. This introduction is the first chapter. In the second chapter, the literature review of the relevant previous studies is exposed. In these reviews, the relations found in earlier studies between Airbnb participation and revenue and the socio-economic characteristics studied in this paper are explained. Besides, there is explained what Airbnb is, how it works, what its consequences can be in cities and what the current situation, legislation and discussions related to Airbnb are in the city of Amsterdam.

The third chapter will present the conceptual model, which gives an overview of the information discussed in chapter two, while the fourth chapter outlines the hypothesis that will be tested in this paper.

In the fifth chapter the data used in this research and the methodology are presented. Here the research question and the sub-questions are outlined. Besides, the reasons for which Amsterdam has been chosen as a case study and the reason for which the neighbourhood has been chosen as research unit are explained. Furthermore, the primary and secondary analyses carried out in this study are explained and the limitations of the Airbnb data and difficulties in measuring socio-economic characteristics are described. Finally, the chapter concludes by illustrating the operationalization table and by clarifying the calculations of the Airbnb revenue. The sixth chapter will present the results of the primary and secondary data analyses, which are in turn summarized into thematic maps and descriptions on the results of the regression analyses.

In chapter seven the results are discussed and the research questions are answered. Finally, chapter eight will conclude this paper by summarizing the main findings of this study. The literature and the appendix with further explanations and data tables can be found after the conclusion in this paper.

(6)

6

Chapter 2: Literature review

In this section, first the Airbnb concept will be introduced and the impact of Airbnb in cities will be explained. Afterwards, a section will be dedicated to the socio-economic, personal, legal and geographical factors influencing Airbnb participation and revenue1 according to several academic studies. Finally, the context of Airbnb in Amsterdam will be drawn with the current discussions and legal aspects on Airbnb in the city.

What is Airbnb

Airbnb is a worldwide online platform for the rental and booking of private accommodations (Airbnb Press, 2018a). On this platform, Airbnb hosts – people sharing (a room in) their dwelling – are brought in contact with Airbnb guests, people looking for short-term lodging. As reported by the company, the idea behind the Airbnb concept is that rather than sleeping in a commercial and monotonous hotel room, tourists and visitors can stay in local houses gaining a more authentic and diverse travel experience (Airbnb Press, 2018b). The offer on the online platform varies from apartments, houses and boats, to treehouses, villas and castles. Users can also book touristic experiences or reserve restaurants on the website, but this part of the business will be ignored in this paper (Airbnb Press, 2018b).

Airbnb is not the only existing company for the rental and booking of private accommodations. According to a study carried out by the Dutch newspaper Algemeen Dagblad (Nijman, 2018), Wimdu, Homeaway, Micazu, Couchsurfing and Talktalkbnb are similar companies, often working with different business models and not always dealing with money. However, these companies won’t be taken into consideration in this paper, as Airbnb is by far the largest and the most popular among the platforms and data on the activities on the other platforms are not publicly available (Wachsmuth and Weisler, 2018).

The Airbnb business model is quite simple and is explained on the Airbnb website as follows. On the demand side, guests can book unique accommodations for cheap prices in private accommodations; on the supply side, hosts are ‘economically empowered’ to share their personal spaces by becoming ‘hospitality entrepreneurs’ (Airbnb Press, 2018a). Hosts can set their own availability and prices and guests can book accommodations on a relatively safe platform which controls the listings2 and the guests through a system of ratings and reviews written by previous users. The hosts, guests, bookings and payments are supervised through screening systems and machine-learning technologies (Airbnb, 2018).

The relatively low entry barrier at both the demand and the supply side made of Airbnb one of the most popular sharing economy concepts worldwide: with more than 5 million listings in 191 countries and 81.000 cities, there are more than 2 million people on average sleeping in an Airbnb every single night (Airbnb Press Room, 2018b). While guests are paying the price for the accommodations to the hosts, the company itself earns money by charging a percentage of the price of the reservation as a service tax. This model seems to be working: in May 2018 Forbes estimated the value of the company at 38 billion US dollars (Forbes, 2018).

1

Airbnb participation is seen in this paper as the number of dwellings shared on Airbnb in a neighbourhood, while Airbnb revenue is meant as the total amount of revenue (in euros) earned by Airbnb hosts by sharing their dwellings on Airbnb in a neighbourhood.

2

(7)

7

The economic effects of Airbnb

Founded in 2008 in San Francisco, according to Joe Gebbia (co-founder of Airbnb), the concept was first designed to help people to earn some extra money by renting a room with an (air)bed in their dwelling to tourists (Gebbia, 2016). Some authors, like Schor (2016), associate the initial success of the concept with the contemporaneous 2008 financial crisis, as Airbnb made it possible for thousands of people who lost their job at the time to gain some extra money to pay their bills. And still today, ten years later, Airbnb claims that it is helping households struggling economically to earn indispensable amounts of income that allows them to afford their own rents or mortgages (Airbnb, no date). Likewise, Chris Lehane (Airbnb’s Global Head of Policy and Communications) states that Airbnb is helping to spread wealth around the world, particularly among the people facing economic difficulties (Airbnb Press, 2018c). On behalf of Airbnb, Gene Sperling (2015), who has served as Director of the National Economic Council under presidents Bill Clinton and Barack Obama, also researched the positive economic effects of Airbnb. Sperling (2015) concluded that Airbnb helps to combat middle class income stagnation, as the Airbnb revenue of an average host seems to close the income gab created during the last 15 years. Furthermore, he states, the extra income through Airbnb can be extremely useful for American families who don’t have enough liquid savings to go through periods of economic crisis or unemployment, as “Airbnb provides a source of income that is not tied to their jobs and does not require drawing

on retirement savings, often with a tax penalty.” (Sperling, 2015: 9).

However, the positive effects described above originate from biased sources, for which they need to be critically addressed. In fact, the effects of Airbnb are not only positive. The online platform recently became subject to tons of discussions in all the major cities in the world (Zwam, 2018) and, as said before, increasingly more academics started to question to what extent the participation to Airbnb as a host is really possible for everyone (Schor, 2016; Gurran and Phibbs, 2017). In fact, increasingly more studies have shown that the relatively high entry barriers to become an Airbnb host discriminates people with – among others – lower education levels, non-Western Backgrounds and lower incomes (Schor, 2016). On the other hand, people with higher SES are facilitated to participate in the market as they have more access to the means required to become an Airbnb host (Gurran and Phibbs, 2017). This means that people from different socio-economic backgrounds have unequal access to Airbnb revenue which is, as will be explained in the section below, very problematic.

The impacts of Airbnb in cities

Positive effects

Airbnb has strongly facilitated city-trip tourism. Since tourism is a source of income for a city, it can be said that Airbnb generally boosts a city’s economy. Oskam and Boswijk (2016) state that the Airbnb hosts are the ones who are benefitting the most from the business, as they earn the direct rent-revenue. As mentioned before, Gurran and Phibbs (2017) argue that this is extremely useful for several households, who can pay their bills and rents thanks to Airbnb, allowing them to remain living in their house. Similarly, a study carried out in Berlin by Stors and Kagermeier (2017) found that several hosts share a room on Airbnb in order to compensate for the rising rental costs. The same authors also pointed at the fact that Airbnb can provide a financial security for flex-workers.

(8)

8 Furthermore, next to the Airbnb hosts, also other entities can benefit from additional income thanks to Airbnb-tourism. Among these there are restaurants, retail and entertainment businesses. However, even if Airbnb (2013) claims that Airbnb is ‘bringing tourism’s economic benefits to neighbourhoods and small businesses not

frequently visited by tourists,’ research has shown that this effect seems to be reduced as most expenditures still seem to benefit amenities in touristic neighbourhoods (Oskam and Boswijk, 2016).

Negative effects

The negative impacts of Airbnb are less visible but significant. In fact, Airbnb provides extra income in cities, but the income is not equally spread among the residents.

First, Airbnb is a big concurrent for hotels and its employees. According to Oskam and Boswijk (2016), the hotel industry is hurt by Airbnb, as it can hardly compete with the Airbnb prices and offers. Schor (2016: 269) analysed this phenomenon and found a ‘crowding-out’ effect caused by Airbnb: hosts are taking over the work of hotel employees, which is traditionally done by low-educated people. As Airbnb takes jobs away from the hotel industry, Airbnb can cause a decrease in jobs – and income – for less-educated people.

Second and most importantly, Airbnb generates revenue which is socio-economically and geographically unevenly distributed across the city (Wachsmuth and Weisler, 2018). As will be further explained, Airbnb participation and revenue in a neighbourhood depend on several factors, among which the accessibility of the residents on the platform (in terms of extra rooms, skills, legality, etc.) and the attractiveness of the dwelling for the guests (location, facilities, etc.). Several critical academics argue that there is strong relationship between people’s SES and Airbnb participation and revenue, as the accessibility of people on the platform and the attractiveness of the dwelling change according to people’s SES (Schor, 2016; Gurran and Phibbs, 2017). The higher the SES, the higher the possibility to earn significant amounts of money through Airbnb, and vice versa. SES therefore reduces the equal access to Airbnb revenue, which can strengthen the process of gentrification and the differences in income inequality among the population. This means that Airbnb can strengthen socio-economic differences in and between neighbourhoods (Oskam and Boswijk, 2016: 29). The way in which this happens and why it is problematic is explained in the coming paragraph.

Gentrification and income inequality

Airbnb can lead to gentrification. In fact, Airbnb revenue is likely to be higher in culturally or historically recognized neighbourhoods close to the city centre that are attractive for tourists (Wachsmuth and Weisler, 2018). This creates a new source of income in these neighbourhoods that the hosts will invest in their dwellings in order to increase their value and close. This will create and close a new form of ‘rent gap,’ the difference between the actual returns to property and the potential returns (Wachsmuth and Weisler, 2018: 1151). As a result, the overall prices and rents of the dwellings in that neighbourhood will rise. Consequently, the existing lower-income residents will be displaced by higher-income newcomers, a process called ‘gentrification’ (Wachsmuth and Weisler, 2018: 1151). Thus, less successful hosts or residents not participating on the platform are likely to get ‘priced out’ of their neighbourhoods as a consequence of this Airbnb-induced gentrification (Cansoy and Schor, 2016: 13). This means that low-SES households will find themselves increasingly spatially marginalized in society and will be forced to live in low-SES neighbourhoods with comparable household

(9)

9 incomes (Musterd and van Gent, 2016). As less-attractive neighbourhoods are not likely to get extra sources of income through Airbnb, these won’t experience this type of transformation and the vicious circle continues, for which the differences between neighbourhoods will increase substantially and residents will find themselves increasingly socio-economically segregated.

While this is already problematic itself, the consequences of Airbnb-led gentrification go further. In fact, this situation can lead to higher rates of income inequality among the residents of urban areas. The gentrified neighbourhoods – which already were attractive to tourists before their gentrification – will be even more attractive to Airbnb guests, as the value (and the quality) of the dwellings increased (Wachsmuth and Weisler, 2018). This means that less-attractive neighbourhoods are even more unlikely to be successful among Airbnb guests and earn Airbnb revenue. As a consequence, there will be residents that do substantially increase their incomes and residents who don’t. As most of the residents that do increase their incomes are likely to already have higher average household incomes and higher SES-levels, Airbnb revenue is augmenting incomes that were already relatively high. This means that Airbnb can severely impact income inequality in a city, and therefore strengthen the disparity between the socio-economic opportunities of different SES groups in society (Muster and van Gent, 2016). Several authors have pointed out that an increase in income inequality in society can lead to severe problems in the urban environment, for which it is an undesirable phenomenon (Tammaru et al., 2015; Cansoy and Schor, 2016; Musterd, 2005).

Factors influencing Airbnb participation and revenue

In this section, the main factors influencing Airbnb participation and revenue will be discussed. In order to do so, a few academic studies on the relation between Airbnb participation/revenue and SES are presented. Next to that, also the personal, legal and geographical factors influencing Airbnb participation and revenue are explained. As Wubetie (2017) states, the characteristics used to determine someone’s SES often differ in researches and depend on the availability of the data on the chosen case study. The socio-economic characteristics researched in this paper are the ethnic background, the education level and the household income. While personal choices are difficult to research, the legal and geographical aspects are researched by using owner-occupancy and the WOZ-value per m2. These two aspects are also seen in this paper as socio-economic characteristics as there are seen as indicators for people’s wealth; the reasons for that will be further explained.

Airbnb and hosts’ education

During the last three years, several scholars have researched the relationship between education and Airbnb participation. Interestingly, the results of these studies, carried out in different contexts, are very similar. Overall, it was concluded that the hosts benefitting the most from the business are highly educated people often having other sources of income (e.g. Schor, 2016).

Schor and Attwood-Charles (2017) found that in the USA, among all the Airbnb hosts, the high-educated ones were offering most of the listings and earning the largest amounts of money. Ke (2017) also found out that, in the USA, areas with high shares of highly-educated residents are associated with more Airbnb hosts. Schor and Attwood-Charles (2017) claim that the number of listings in less-educated neighbourhoods increased over time,

(10)

10 but that the latter did not attract many guests. Cansoy and Schor (2016) found that, in 104 metropolitan areas in the USA, education is highly influential in the Airbnb business, as areas with highly educated people present more listings, reviews, higher ratings and higher prices, suggesting that listings of highly educated hosts are more successful than the ones of low-educated hosts.

The reasons for this pattern can be explained as follows. According to a research carried out by Schor (2016) in the USA, highly educated providers are using platforms such as Airbnb to increase their earnings in order to reduce their education-related debt. Similarly, a study carried out in London showed that most Airbnb rooms are located in low-income but highly-educated areas (likely students), which could be explained by their need to earn extra money in order to pay their studies and living in an expensive city like London (Quattrone et al., 2016). Other authors claim that resources and skills of a certain level are required to enter the Airbnb market as a host (Schor and Attwood-Charles, 2017). In fact, the participation to Airbnb as a host requires specific know-how to organize and manage quite complex bookings (Cansoy and Schor, 2016; Ke, 2017) next to specific computer skills and internet access which mostly highly educated people possess (Gurran and Phibbs, 2017).

Airbnb and hosts’ income and wealth

Just like the relationship with education, also the relation between income and wealth and the participation to Airbnb on the supply side have been researched. Again, these studies found similar conclusions.

Before explaining their results, it might be interesting to look at the theoretical hypothesis of this relation. Cansoy and Schor (2016) argue that people are more likely to share their property when the marginal benefits of sharing are higher than the marginal costs of sharing. Since the marginal benefits are likely to be higher among people with lower incomes – because earning extra money is relatively worth more for them –, the areas with bigger shares of low-income groups are expected to count more listings compared to better-off areas (Cansoy and Schor, 2016). Similarly, Ke (2017) hypothesized that due to the relatively low entry barrier and potential high monetary gain, becoming a host on Airbnb would be particularly attractive for people with lower incomes. The results of the papers have proven this hypothesis to be true. In fact, areas where the median household income is low generally seem to count more Airbnb listings (Ke, 2017). However, this does not say anything on the revenue generated through these listings: according to research, areas where the median household income is higher tend to ask higher prices and earn more revenue through Airbnb compared to poorer areas (Cansoy and Schor, 2016; Ke, 2017). This means that even though the Airbnb participation in high-income areas is lower, the revenue through Airbnb in these areas is higher, suggesting that the Airbnb’s in high-income areas are more successful than the ones in low-income areas. The reason for this can be explained as follows.

A study carried out by Gurran and Phibbs (2017) in Sydney (Australia), explains how it requires an attractive Airbnb listing (attractive appearance, description, presentation, facilities and furnishing) in order to be able to earn significant amounts of money through Airbnb. This is unlikely to be the case in the listings of the people belonging to low-income groups, as the listings of the richer people are probably more attractive in terms of luxury, facilities and beauty (as they can afford themselves goods that poorer people cannot). Similarly, Ke (2017) argues that next to living in more attractive dwellings, people with higher incomes usually live in more attractive neighbourhoods compared to people with lower incomes: this will be addressed in the section ‘geographical aspects.’

(11)

11 Schor (2016) argues that sharing on Airbnb requires either a spare room or the possibility to sleep somewhere else while the dwelling is rented out. As it is unlikely that people with a low-income have an attractive underutilized room or apartment located in a neighbourhood where tourists are willing to come, it can be said that also in that sense people’s income and wealth status can be barriers to their possibility to earn important amounts of money with Airbnb (Ke, 2017).

When looking more in detail at the type of listings presented on Airbnb, it seems that there is some difference between single rooms offered on Airbnb and entire-home listings. While single rooms seem to be offered mostly in low-income but highly-educated areas (likely students), entire-home listings seem to be more common in wealthy areas (Quattrone et al., 2016). Next to that, most entire-home listings are offered in areas with high home-ownership rates, where dwellings are highly valuable in terms of prices (Quattrone et al., 2016). Similarly, Ke (2017) has shown that entire-home listings in higher income areas get more revenue compared to the ones in lower income areas.

Airbnb and hosts’ ethnic background

The relation between hosts’ ethnic background and Airbnb participation/revenue has also been researched by several authors. The event that triggered this wave of research was a study released by Airbnb in April 2016 entitled “Airbnb and economic opportunity in New York City’s predominantly black neighbourhoods” (Airbnb Citizen, 2016). In this study, the company claimed that the Airbnb business was helping black neighbourhoods in New York City, as the number of Airbnb guests between 2015 and 2016 in these neighbourhoods grew 78%, versus the 51% in the rest of the city (Airbnb Citizen, 2016).

However, critics of the company pointed out that the situation wasn’t quite like that. In fact, rather than helping the ‘black’ people to increase their earnings, Airbnb was facilitating the gentrification of these neighbourhoods, which made these neighbourhoods increasingly less affordable for the African-American people. Similarly, a study carried out a year later by Inside Airbnb3 showed that rather than the African-Americans, the ‘white’ hosts were earning the absolute largest amount of Airbnb revenue in New York (Wachsmuth and Weisler, 2018). Other authors came to the same conclusions. According to a study carried out in the USA, most Airbnb hosts are ‘ethnically white’ and native-born (Schor, 2016). Similarly, Attwood-Charles and Schor (2017) pointed at the fact that Airbnb hosts are usually disproportionally ‘white’ and that people with non-Western ethnic backgrounds often receive lower prices and lower ratings for their accommodations compared to those having a Western ethnic background. Laouenan and Rathelot (2017) found in their multiple case studies in the USA and Europe that within the same neighbourhood, hosts with a non-Western background tend to charge 3.2% less for comparable listings. However, despite the lower prices, the demand in these areas seemed to be lower (Laouenan and Rathelot, 2017). As a possible explanation, the authors suggest that non-Western hosts on average live in neighbourhoods that are less attractive for potential guests because they offer listings with less outside-options (balconies, terraces, gardens) compared to Western hosts (Laouenan and Rathelot, 2017). On the contrary, Cansoy and Schor (2016) found that ‘non-white’ areas tend to count more hosts and more reviews compared to other areas. However, also these authors concluded that their listings are rated worse and that they generate less income, which they see as a form of discrimination (Cansoy and Schor, 2016). Similarly,

3

Inside Airbnb is a website tracking down and aggregating publicly available data from the Airbnb website in order to make Airbnb data publicly available for researchers and policymakers with interest in the topic (Inside Airbnb, 2018).

(12)

12 Ke (2017) has shown that areas with higher fractions of ‘non-white’, foreign-born residents tend to have more single-room listings compared to other areas. However, these areas seem to supply less entire-home listings (Ke, 2017). According to Wachsmuth and Weisler (2018), in New York Airbnb demand is weak in racialized neighbourhoods lacking touristic and cultural attractions as well as areas where the public transport connection to the city centre is poorer. Moreover, in the USA, guests seem to complain more about the services provided by non-Western hosts compared to the ones of Western hosts (Attwood-Charles and Schor, 2017). Finally, as said before, Quattrone (2016) found that in London most Airbnb rooms are situated in student areas, where the education level is quite high, while the income is low. Interestingly, when looking at the ethnic background, he found that these hosts are predominantly non-UK born. This could suggest that foreign students (without specifying whether they are Western or not) are facing more economic difficulties in London, for which they are likely to share their rooms on Airbnb more often compared to UK native students.

An important aspect that needs to be taken into consideration when looking at people’s ethnic background, are their language skills. In fact, as several authors have pointed out, skills of a certain level are needed in order to be able to share on Airbnb (Ke, 2017; Gurran and Phibbs, 2017; Cansoy and Schor, 2016; Schor and Attwood-Charles, 2017). For people with a non-Western background, the ability to speak – in the case of Amsterdam – Dutch or English it is not taken for granted. However, this is a quite important competence on the Airbnb platform, as the hosts needs to be able to write an attractive description on the listing and communicate with the (potential) guests.

Other factors influencing Airbnb participation and revenue

Personal choices

Several authors have discussed the personal choices involved in the decision to become an Airbnb host. Ke (2017) states that the choice to become an Airbnb host does not only depend on one’s possibilities, but that it is also a personal choice. In fact, becoming Airbnb host requires time, energy, and the desire to share one’s personal space, which is not desirable and suitable for everyone, even though there is a monetary compensation involved (Ke, 2017: 15). This is an important extenuating factor, because having the possibility to share on Airbnb does not automatically imply the willingness to share on Airbnb. Similarly, Cansoy and Schor (2016) have pointed out that it is difficult to isolate the impact of people’s SES on their participation as Airbnb hosts, because the latter largely depends on people’s norms and preferences: for example, some people might don’t want to share their dwelling on Airbnb because they don’t trust unknown guests. Ke (2017) also states that the main motivation for people who are willing to share their dwellings, is the monetary compensation. Yet, not only the revenue is a motivation to participate on Airbnb. In fact, as Stors and Kagermeier (2017) state, Airbnb can be a way to have a flexible housemate, as hosts can choose the availability of the room they are sharing, and a way to socialize, as Airbnb allows to meet people from all over the world in a relatively easy way.

(13)

13

Legal aspects

Another important aspect that needs to be addressed is the legality of sharing one’s dwelling on Airbnb. As will be explained further in this paper, in Amsterdam only residents living in an owner-occupied dwelling are legally allowed to sub-rent their dwellings (if the VvE4 allows). As not everyone lives in an owner-occupied dwelling, there are many people who are legally not allowed to share their dwelling on Airbnb, even if they wanted to. Therefore, it is also important to look at the share of owner-occupied dwellings per neighbourhood, as that might be highly influential in the Airbnb participation rate.

Apart from being influential in the legal aspects, owner-occupancy is also strongly related to people’s wealth. In fact, richer people are more likely to have the financial means to buy a dwelling than poorer people. This means that people living in owner-occupied dwellings are likely to be wealthier than people living in private or social rented dwellings. Therefore, owner-occupancy is also seen in this paper as an indication of wealth, and as such as part of a neighbourhood’s socio-economic characteristics.

Geographical aspects

Finally, there is the geographical factor involved. A study released by Airbnb in 2013 claims that ‘73% of Airbnb

properties in Amsterdam are located outside the eight central tourist districts and 69% of Airbnb guests said they used Airbnb to explore a specific neighbourhood’ (Airbnb, 2013). However, several studies have shown that most

Airbnb guests prefer staying close to the city centre in attractive, historically and culturally interesting neighbourhoods (e.g. Wachsmuth and Weisler, 2018). This means that relatively newly-build neighbourhoods far away from the city centre, such as neighbourhoods in Amsterdam South-East, are less attractive for tourists: these neighbourhoods are therefore less likely to earn a lot of money through Airbnb (Ke, 2017). Unfortunately, it is quite difficult to estimate the ‘attractiveness’ of a neighbourhood for a tourist. However, there is a measure that strongly relates to the location/real estate properties of the dwellings and to the attractiveness of the neighbourhoods: the WOZ-value.5 As hosts living in dwellings with a high WOZ-value are likely to attract more Airbnb guests compared to hosts sharing dwellings with lower WOZ-values, the latter can be capable of influencing the success of an Airbnb listing on the platform. Therefore, also the WOZ-value will be taken into consideration in this research.

Just like owner-occupancy, also the WOZ-value is strongly related to people’s wealth. In fact, in neighbourhoods with high WOZ-values, the purchase and rental prices of dwellings are usually high. Therefore, it can be said that families living in neighbourhoods with high WOZ-values are likely to be better off in terms of wealth than the ones living in neighbourhoods with low WOZ-valued dwellings. In Amsterdam the situation is a bit different, because social housing is more or less equally spread around the city, for which there are poorer people (with low SES) living in dwellings with high WOZ-values (Gemeente Amsterdam, 2018a). However, being only about 30% of the total amount of dwellings, the remaining 70% of the dwellings are likely to be still inhabited by

4

VvE stands for ‘Vereniging van Eigenaren’ and is the association of owners. It will be explained further in this paper.

5

WOZ stands for ‘Waardering Onroerende Zaken’ and is the yearly valuation of real estate defined by a specific public office in the Netherlands. The WOZ-value largely depends on the quality, the facilities and the (attractiveness of) the location of the house (Rijksoverheid, 2018).

(14)

14 people with a corresponding wealth status. Therefore, WOZ-value is also seen in this paper as an indication of wealth, and as such as part of a neighbourhood’s socio-economic characteristics.

Airbnb in Amsterdam

In 2013 Airbnb released a study in which they claim that Airbnb is having a ‘positive economic impact on the

economy of Amsterdam.’ In fact, according to their study, ‘36% of the hosts say the income they earned via Airbnb helped them to make ends meet,’ ‘87% of Airbnb hosts rent the homes they live in to visitors on an occasional basis,’ and ‘the Airbnb community in Amsterdam helps residents and small businesses benefit from tourism […] and travelers from around the world to have a new, more sustainable and social way to visit one of Europe’s most vibrant cities’ (Airbnb, 2013).

However, the information released by Airbnb strongly runs apart from the news in the Dutch media on Airbnb and its effects in society. Whereas Airbnb is claiming to have extremely positive effects in Amsterdam, several researches carried out by common Dutch (digital) newspapers state that the situation is not as positive as Airbnb claims.

In an article published in NRC (Haanen, 2018), for example, it was stated that ‘holiday rental strongly contributes

to crowds and nuisance in the city center’ and that the number of tourists in Amsterdam ‘did not remain without consequences for the quality of life.’ According to a study carried out by Algemeen Dagblad (Couzy, 2018), in

2018 the number of Airbnb’s in Amsterdam reached 20.010 units, an increase of 8,5% compared to 2017: the newspaper called this amount a ‘record number’ and states that the conditions set by the municipality continue to be violated. According to a research carried out by Trouw (Heere, 2017), ‘thousands of houses in Amsterdam

are withdrawn from the housing market because they are permanently rented to tourists.’ Furthermore, in an

article published in the Volkskrant (Mebius, 2018) it was reported that the trade association of the hotels, restaurants and cafés in the Netherlands KHN (Koninklijke Horeca Nederland) blames Airbnb for ‘unfair

competition for hotels’ and calls for a ‘nation-wide policies’ to regulate the Airbnb business. Finally, in a study

carried out in 2016, the ING bank found that Airbnb is causing inflation in the housing prices, which can lead to increases up to 4% in Amsterdam (ING, 2016).

Legislation on Airbnb in Amsterdam

Just like several other cities in the world, also Amsterdam is trying to regulate Airbnb in order to minimalize its negative side-effects. The rules in Amsterdam for the so-called ‘vakantieverhuur’ (holiday rental), the law under which Airbnb falls, were established in 2017 as follows.

When a property is rented as holiday rental, it needs to be registered at the municipality with the Digi-D6 of the host in advance (Gemeente Amsterdam, 2016). When hosts register the holiday rental, they automatically agree on a few rules. For example, the host affirms to be the main occupant of the dwelling and to be registered at the municipality of Amsterdam on that address. For not reporting the rental of property, a host could be fined €6.000. This amount can be increased up to €20.500 if you have not complied with the other conditions for holiday rentals. People are allowed to rent out their dwelling for a maximum of 60 nights per calendar year for a

6

(15)

15 maximum of four guests per night. Renting out social rented property7 owned by a housing corporation is strictly prohibited, and in the case of private rental or home ownership, the host needs to have the permission from the landlord or from the Association of Owners (VvE). In practice, in almost all private rental contracts sub-renting is strictly prohibited, for which Airbnb is forbidden too. Therefore, as said before, it can be said that Airbnb is legally allowed only in the case of owner-occupancy, providing that the VvE – if existing8 – agrees on it. Finally, guests are prohibited to cause any inconvenience and need to pay tourists’ taxes, whereas hosts need to pay income taxes on the Airbnb revenue (Gemeente Amsterdam, 2016). Due to the tourist-overcrowded city centre the municipality has decided to decrease the maximum amount of nights to 30 from 2019 onwards (Gemeente Amsterdam, 2018b).

Airbnb and fraud in Amsterdam

Although the municipalities’ rules are clear, Airbnb-fraud seems to persist in Amsterdam. According to a research carried out by NRC (Noort, 2015), many hosts continue to rent out their dwelling for more than 60 days a year and to more than 4 people per time. Also, many hosts and guests are not paying income and tourists’ taxes or create different listings of the same dwelling on the Airbnb website (for example by registering different names) or they rent out their social rental dwelling, which is illegal (Noort, 2015). However, according to a study carried out by AT5 (unknown author, 2017), the biggest problems seem to be caused by commercial Airbnb’s. These dwellings are permanently rented out through Airbnb and, as nobody is actually living nor registered in those dwellings, can be seen as illegal hotels. Investors – the owners of these dwellings – make profit by buying dwellings and by permanently renting them out through Airbnb, illegally. In order to do so, these people create fake profiles: according to a study carried out by NRC (Loon and Niemandsverdriet, 2018), some of these investors own more than 100 Airbnb’s just in Amsterdam. In some cases, the guests renting the Airbnb’s never meet the owner, as they hold the key at companies such as Bnbmanager or 60days, which takes care of the key transfer, the cleaning, the correspondence with the guests, etc. In return, the company charges a percentage of the revenue, between 10 and 20%, depending on the services (Bnbmanager Amsterdam, 2018). The municipality of Amsterdam calls these companies ‘sleutelbedrijven’ (key companies) and their functioning is legal as long as the conditions set by the municipality are met (Gemeente Amsterdam, 2016). However, the same study of NRC (Loon and Niemandsverdriet, 2018) showed that ‘sleutelbedrijven’ themselves also own several properties, which they share through Airbnb under fake profiles: by doing so, they break the rules of both, the municipality and Airbnb itself. Finally, in an interview by AT5 (unknown author, 2017) with alderman housing Laurens Ivens, the latter states that ‘even though the implementation of the legislation on vacation rentals is very difficult,

Airbnb needs to be further regulated if the city wants problems with fraud to be solved’ and, as long as the

company itself doesn’t collaborate, ‘the municipality will have to take stricter measures, for example by carrying

out more controls and setting higher fines.’

7

A social rented property is a subsidized dwelling for people with lower incomes in the Netherlands. This allows people with lower incomes to be able to pay their rents and often to live in locations otherwise denied to them because of the high rent prices (Gemeente Amsterdam, 2018a).

8

VvE’s (associations of owners) exist in the Netherlands only in multi-family residential dwellings; no VvE’s exist for single family houses (detached houses). The reason for that, is that in multi-family residential dwellings common decisions on the maintenance of the dwelling need to be taken with all the owners, which is not necessary in single family houses

(16)

16

Chapter 3: Conceptual model

To get an overview of the information provided in chapter 2, a conceptual model has been created which helps to get a better understanding of the relationships between the factors analysed in the research. The conceptual model below (figure 1) is based on the conclusions of the academic articles and the legal aspects set by the municipality of Amsterdam and summarizes the relationship between SES and Airbnb revenue.

(17)

17

Chapter 4: The hypothesis

As the academic studies discussed in the literature review have shown, there seems to be a relationship between host’s SES and Airbnb participation and revenue. While most of the other studies have taken the individual Airbnb hosts as research unit, this research takes the neighbourhood level as research unit. This means that, instead of looking at the revenue and SES of individual hosts, this study will look at the Airbnb participation, revenue and SES of the neighbourhoods and of the hosts on a neighbourhood level. The verification of this approach will be discussed further in the section ‘the neighbourhoods as research units’, in chapter 5.

The hypothesis is thus formulated based on the neighbourhood as research unit. Because neighbourhood variables and statistics are directly related to the inhabitants and hosts, the results of the research also represent the average characteristics of the hosts on a neighbourhood level, for which the hypothesis also applies to the hosts. This means that when a neighbourhood with a big share of highly-educated residents is expected to present high rates of Airbnb revenue, it is expected that also the hosts in that neighbourhood on average earn more Airbnb revenue compared to the hosts in neighbourhoods with lower shares of highly-educated residents).

Despite the differences in research units, the findings of the papers discussed in chapter 2 can be used to draw the hypothesis for the case of Amsterdam. However, it is important to remember that the research contexts of those studies and this study are very different: five out of the eight studies are carried out in the USA, except for Quattrone et al. (2016) who researched in London (UK), Gurran and Phibbs (2017) who studied the case of Sydney (Australia) and Louenan and Rathelot (2017) who carried out only a part of their research in Europe. The differences between the levels of wealth and income inequality in the USA and in other parts of the Western world are quite substantial (OECD, 2017). Furthermore, Amsterdam is not very segregated compared to other cities in the world and the socio-economic differences in and between neighbourhoods in Amsterdam are relatively weak (Tammaru et al., 2015; Musterd, 2005). Besides, Airbnb legislation is different in Amsterdam and in other cities in the world (Airbnb, 2017). Altogether, these aspects could lead to big differences in people’s possibilities and willingness to share on Airbnb. Therefore, it is important to keep in mind that the results of these studies might not reflect the reality in Amsterdam.

The conclusions of the papers outlined in the literature review and the hypothesis that is tested in this paper are explained in the coming paragraphs.

Education is likely to influence both the participation and the revenue of a host on Airbnb. In fact, all the authors state that highly-educated people are more likely to participate on the Airbnb platform as hosts. Besides, they also seem to get more revenue compared to low-educated hosts. Therefore, the hypothesis is that also in Amsterdam Airbnb participation and revenue are likely to be higher in the neighbourhoods where the average education level is higher.

The majority of the studies have concluded that most Airbnb hosts are likely to have relatively low incomes. However, hosts with higher incomes seem to earn more revenue overall. For this reason, the hypothesis is that

(18)

18 also in Amsterdam Airbnb participation is higher in neighbourhoods with lower average household incomes, while the Airbnb revenue is higher in the neighbourhoods where the average income is higher.

While there seems to be some disagreement on the relation between Airbnb participation and ethnic background, there seems to be a consensus on the fact that Western hosts are more successful on Airbnb, as they get more demand, higher ratings and higher revenues. Therefore, while we can’t draw a clear hypothesis on the participation of non-Western hosts, it is likely that the Airbnb revenue in Amsterdam is lower in the neighbourhoods with higher shares of non-Western residents.

Furthermore, the neighbourhoods with high shares of owner-occupant dwellings are expected to present higher rates of Airbnb participation because of legality (discussed in chapter 2), and the neighbourhoods with higher WOZ-values are expected to earn more Airbnb revenue because they are likely to be more attractive for guests in terms of dwellings and location.

Finally, there is an interesting point suggested by Quattrone et al. (2016). In his study in London, he found that most Airbnb rooms are located in low-income but highly-educated areas, which are likely student areas. Unfortunately, the author doesn’t specify whether these areas are predominantly Western or Non-Western and he doesn’t say how much revenue these Airbnb’s earn. Nevertheless, the hypothesis can be drawn that also in Amsterdam low-income but highly-educated neighbourhoods are likely to have relatively high Airbnb participation rates.

Generally speaking, it is expected that neighbourhoods with a high ‘general’ SES are likely to participate more on Airbnb and to earn more revenue compared to the neighbourhoods with a low ‘general’ neighbourhood SES.

(19)

19

Chapter 5: Data and methodology

Methodology

This study aims at finding the relationship between Airbnb revenue and socio-economic characteristics on a neighbourhood level in Amsterdam in 2017. For reasons that will be further explained, also the relationship between Airbnb participation and socio-economic characteristics on a neighbourhood level will be researched. In order to do so, the hypothesis described above will be tested by means of a quantitative approach. In this chapter, first the research questions will be outlined. Then, the case study, research unit and analysis will be explained. Finally, the data and the calculations on Airbnb revenue will be presented.

The research questions

The main research question is:

‘To what extent is there a relationship between the revenue generated through Airbnb and the socio-economic characteristics of the different neighbourhoods in Amsterdam in 2017?’

In order to answer this question as good as possible, a complete picture on the SES of neighbourhoods in Amsterdam and of the Airbnb activities in the city are needed. Therefore, next to the mean research question, this research tries to answer the coming sub-questions:

1. What are the socio-economic characteristics of the different neighbourhoods in Amsterdam in 2017? 2. What is the ‘general’ socio-economic status of the different neighbourhoods in Amsterdam in 2017? 3. How are the Airbnb’s distributed among the different neighbourhoods of Amsterdam in 2017? 4. How is Airbnb revenue distributed among the different neighbourhoods of Amsterdam in 2017? 5. How is Airbnb revenue distributed among the Airbnb hosts of the different neighbourhoods in

Amsterdam in 2017?

6. To what extent is there a relationship between Airbnb participation and the socio-economic characteristics of the different neighbourhoods in Amsterdam in 2017?

7. To what extent is there a relationship between Airbnb revenue per host per neighbourhood and the socio-economic characteristics of the different neighbourhoods in Amsterdam in 2017?

8. To what extent is there a relationship between Airbnb revenue and the ‘general’ socio-economic status of the different neighbourhoods in Amsterdam in 2017?

9. Is Airbnb participation higher among students in Amsterdam in 2017?

(20)

20

Difference between socio-economic characteristics and socio-economic status

Socio-economic characteristics are seen in this paper as the five tested variables composing the socio-economic status (SES) of a neighbourhood. These variables are (1) the average annual standardized household income, (2) the share of owner-occupancy, (3) the WOZ-value per m2, (4) the share of low-education people and (5) the share of people with a non-Western ethnic background.

Some research questions – among which the main research question – focus on the socio-economic characteristics of the neighbourhood, while other focus on the socio-economic status of the neighbourhood. The difference between the two types of questions is explained as follows.

The questions dealing with the ‘socio-economic characteristics’, look at the relationship between Airbnb participation/revenue and the five variables listed above individually.

The questions dealing with the ‘general’ socio-economic status, instead, look at the relationship between Airbnb participation/revenue and the general SES measure that has been created in this paper, which is a combination of the five socio-economic characteristics described above into a single number. The difficulties encountered when measuring socio-economic status and the calculation of the ‘general’ SES measure used in this paper are explained further in this chapter.

Amsterdam as case study

In this research, the city Amsterdam in the year 2017 has been chosen as case study. The year 2017 has been chosen as it was the most recent year with complete data at the time this paper was written. As previously stated, according to Inside Airbnb (2017), Amsterdam counts more than 20.000 Airbnb’s. However, the active9 Airbnb’s in 2017 were 12183: this means that 2.84% of the dwellings of Amsterdam were actively shared on Airbnb in 2017 (Inside Airbnb, 2017; OIS Amsterdam, 2017). As said before, the number of Airbnb’s and the Airbnb-related problems seem to be increasing every year, for which Airbnb is a hot topic on the agenda of Amsterdam (e.g. Haanen, 2018).

Besides, as mentioned in the introduction, no research to date on the relation between Airbnb participation and revenue and SES have been carried out in a relatively more ‘equal’ and less segregated city like Amsterdam (Musterd, 2005). This is especially interesting in this city, because the spatial distribution of Airbnb’s in the different neighbourhoods in Amsterdam is clearly uneven (see figure 2) and no research on Airbnb has been carried out in Amsterdam so far. The growing popularity of Airbnb, the countless discussions on its consequences and the particular socio-economic context of Amsterdam make this city an interesting case to research.

9

(21)

21 Figure 2: the spatial distribution of Airbnb’s in Amsterdam.

(22)

22

The neighbourhoods as research units

By researching on a neighbourhood level, one can compare the socio-economic variables and the Airbnb revenue of the different neighbourhoods and find out whether there is a relation between the two factors. Researching the relationship between these variables on a neighbourhood level makes sense because the neighbourhood level captures who is living in the area. This means that the neighbourhood includes all the data of the residents living and the Airbnb activities happening in the neighbourhood at an aggregate level. The SES of a neighbourhood is the mean of all the SES-related data of the residents, while the Airbnb participation and revenue of a neighbourhood are respectively the sum of the Airbnb’s and of the Airbnb revenue of the hosts living in the neighbourhood (divided per the number of dwellings in order to be able to compare the different neighbourhoods).

Airbnb participation and revenue make sense to research on a neighbourhood level as the amount of Airbnb’s and revenue say something about the total Airbnb activities in that neighbourhood. By comparing the Airbnb participation and revenue between the neighbourhoods, it is possible to say something about the differences in (success of the) Airbnb participation and find relationships with the SES of the neighbourhoods. The comparison is possible as all the neighbourhood participation and revenue numbers have been ‘corrected’ for the size of the different neighbourhoods by dividing the numbers through the number of dwellings in that neighbourhood. Besides, by dividing the revenue through the number of Airbnb’s, it is possible to calculate the average revenue of a host on a neighbourhood level and compare the average host revenues of the different neighbourhoods in Amsterdam.

The SES makes sense to research on a neighbourhood level because people with a similar SES often live in the same neighbourhood. In fact, in most urban areas in the world, there is a phenomenon called ‘segregation’, which is the ‘spatial separation of specific population subgroups within a wider population based on the

socio-economic characteristics of the residents’ (Knox et al., 2010: 492). Obviously, there are always variations

between people in a neighbourhood (for example there can be high and low-educated people living in the same building). However, generally speaking these differences can be neglected on a neighbourhood level, as most residents are likely to live in a neighbourhood with a SES corresponding to their SES (Tammaru et al., 2015). Amsterdam is no exception because also this city is socio-economically segregated, even if its segregation levels are lower compared to other cities in the world (Musterd, 2005).

For all these reasons, performing the analysis on a neighbourhood level –instead of individual level- in Amsterdam makes sense. Obviously, the accuracy of the research can always be questioned; these aspects will be discussed further in chapter 7. The neighbourhoods of Amsterdam and their names are presented in the map and table below (figure 3, table 2). The ten highlighted neighbourhoods are the ones that are referred to in this paper as ‘city centre’: these neighbourhoods are situated in the central district (stadsdeel Centrum).

(23)

23 Figure 3: the neighbourhoods of Amsterdam.

Table 1: the names of the neighbourhoods of Amsterdam

0 Burgwallen-Oude Zijde 33 Oostelijk Havengebied 66 Oostzanerwerf 1 Burgwallen-Nieuwe Zijde 34 Zeeburgereiland/Nieuwe Diep 67 Kadoelen

2 Grachtengordel-West 35 IJburg West 68 Waterlandpleinbuurt

3 Grachtengordel-Zuid 36 Sloterdijk 69 Buikslotermeer

4 Nieuwmarkt/Lastage 37 Landlust 70 Banne Buiksloot

5 Haarlemmerbuurt 38 Erasmuspark 71 Noordelijke IJ-oevers West

6 Jordaan 39 De Kolenkit 72 Noordelijke IJ-oevers Oost

7 De Weteringschans 40 Geuzenbuurt 73 Waterland

8 Weesperbuurt/Plantage 41 Van Galenbuurt 74 Elzenhagen

(24)

24

10 Westelijk Havengebied 43 Westindische Buurt 76 Slotermeer-Noordoost

11 Bedrijventerrein Sloterdijk 44 Hoofddorppleinbuurt 77 Slotermeer-Zuidwest

12 Houthavens 45 Schinkelbuurt 78 Geuzenveld

13 Spaarndammer- en Zeeheldenbuurt 46 Willemspark 79 Eendracht

14 Staatsliedenbuurt 47 Museumkwartier 80 Lutkemeer/Ookmeer

15 Centrale Markt 48 Stadionbuurt 81 Osdorp-Oost

16 Frederik Hendrikbuurt 49 Apollobuurt 82 Osdorp-Midden

17 Da Costabuurt 50 IJburg Oost 83 De Punt

18 Kinkerbuurt 51 IJburg Zuid 84 Middelveldsche Akerpolder

19 Van Lennepbuurt 52 Scheldebuurt 85 Slotervaart Noord

20 Helmersbuurt 53 IJselbuurt 86 Overtoomse Veld

21 Overtoomse Sluis 54 Rijnbuurt 87 Westlandgracht

22 Vondelbuurt 55 Frankendael 88 Sloter-/Riekerpolder

23 Zuidas 56 Middenmeer 89 Slotervaart Zuid

24 Oude Pijp 57 Betondorp 90 Buitenveldert-West

25 Nieuwe Pijp 58 Omval/Overamstel 91 Buitenveldert-Oost

26 Zuid Pijp 59 Prinses Irenebuurt e.o. 92 Amstel III/Bullewijk

27 Weesperzijde 60 Volewijck 93 Bijlmer Centrum (D,F,H)

28 Oosterparkbuurt 61 IJplein/Vogelbuurt 94 Bijlmer Oost (E,G,K)

29 Dapperbuurt 62 Tuindorp Nieuwendam 95 Nellestein

30 Transvaalbuurt 63 Tuindorp Buiksloot 96 Holendrecht/Reigersbos

31 Indische Buurt West 64 Nieuwendammerdijk/Buiksloterdijk 97 Gein

32 Indische Buurt Oost 65 Tuindorp Oostzaan 98 Driemond

The data

The database used to define the SES characteristics of the neighbourhoods has been downloaded from OIS Amsterdam (2018), a data portal of the municipality of Amsterdam. The map of Amsterdam has been downloaded from the Dutch national data portal CBS (2017). The data on Airbnb has been downloaded from Inside Airbnb (2018), a website issuing publicly available information about Airbnb listings in well-known cities. This website does not directly provide the amount of revenue per host in 2017 but provides enough information to make an estimation of the revenue based on other variables. The data tables are presented in the appendix 7 and the exact calculations will be explained further in this paper. However, before presenting the calculations, the limitations of the Airbnb data and the difficulties in measuring socio-economic characteristics of a neighbourhood are explained.

Limitations on the Airbnb data

First of all, the data are not downloaded from the official Airbnb website but from Inside Airbnb, a website that tracks down and aggregates publicly available data from the Airbnb website. It would be preferable to use official data from Airbnb but, as Wachsmuth and Weisler (2018) point out, the company has always been quite secretive on its data. However, as the same authors have stated, Inside Airbnb is one of the most relied-upon

(25)

25 websites supplying data on the Airbnb activities, for which it still makes sense to use these data for this research (Wachsmuth and Weisler, 2018).

Second, as said before, Airbnb is the largest home-sharing platform, but not the only one. However, to date there are no websites such as Inside Airbnb providing data on the activities on these platforms, for which the latter cannot be taken into consideration in this research. However, this fact is important to keep in mind as it means that in Amsterdam there are probably more home-sharing activities that are generating revenue, which however cannot be quantified (Wachsmuth and Weisler, 2018).

Third, the longitudinal and longitudinal coordinates of the Airbnb’s are not precise. In fact, the exact locations of the dwelling have been anonymized by Airbnb for privacy reasons. This means that the (publicly available) information on the locations of the Airbnb’s has been randomly moved a couple of meters away on the map. The range in which this happens is within a radius between the 0 and 150 meters (Wachsmuth and Weisler, 2018: appendix p. 2-3). Since an Airbnb’s true location must lie in a surrounding buffer of 150 meters, some Airbnb’s laying close to the neighbourhood borders might have been erroneously positioned in the adjoining neighbourhood, affecting the overall calculations. However, since the coordinates are randomly obfuscated, we can assume that all the neighbourhoods are somehow affected by this problem in the same way for which, within this marge of error, it still makes sense to perform the analysis with this data.

Difficulties in measuring socio-economic characteristics of a neighbourhood

‘Socio-economic status’ is a rather abstract concept defining people’s position in relation to others within society (Tammaru, 2015). Being an abstract concept, the variables used to define socio-economic status differ from study to study. Wubetie (2017: 1) states that the most commonly used variables are measures of “material

capital (income, wealth, trust funds, etc.), human capital (skills, abilities, credentials, etc.) and social capital (instrumental relationships such as being friends with lawyers and doctors).” The variables used to measure the

socio-economic characteristics of neighbourhoods in this study are chosen taking Wubetie, the articles in chapter 2 and the legal aspects set by the municipality of Amsterdam as point of reference. As mentioned before, these variables are (1) the average annual standardized household income, (2) the share of owner-occupancy, (3) the WOZ-value per m2, (4) the share of low-education people and (5) the share of people with a non-Western ethnic background. The operationalization of the concepts is presented in the table below (table 3).

(26)

26

Operationalization table

Table 2: Operationalization of the concepts

Concept/factor Dimensions Variables Indicators

Airbnb

participation per neighbourhood

Number of Airbnb’s

Total number of Airbnb’s that were active10 during 2017

Total number of Airbnb’s per 100 dwellings per neighbourhood in 2017

Airbnb revenue per neighbourhood

Revenue through Airbnb

Total revenue through the rent of Airbnb

accommodations in 2017

Total estimated revenue in euro through the rent of Airbnb accommodations per dwelling per neighbourhood in 2017 Airbnb revenue per

Airbnb host per neighbourhood

Revenue through Airbnb

Total revenue through the rent of Airbnb

accommodations in 2017

Average estimated revenue in euro through the rent of Airbnb

accommodations per Airbnb host per neighbourhood in 2017

Socio-economic characteristics of a neighbourhood

Income Annual standardized household income11

Average annual standardized household income per neighbourhood in 2017 Wealth Type of property of the

dwellings

Share of owner-occupancy per neighbourhood in 2017

WOZ-Value of the dwellings Average WOZ-value per m2 per neighbourhood in 2017

Education Highest attained education level

Share of people with a low education level as their highest attained education level per neighbourhood in 2017

Ethnicity Ethnic background Share of people with a non-western ethnic background per neighbourhood in 2017 General SES of a

neighbourhood

Socio-economic status

Combination between the five socio-economic characteristics of a

neighbourhood listed above

Average of the percentiles of the different socio-economic characteristics (corrected for their positive/negative relation with SES) of a neighbourhood

10

The definition of ‘active’ is explained in the section ‘calculations on the Airbnb revenue.’

11

The reason for which the annual standardized household income has been chosen above other types of income measures, is that this one corrects for variations in household size, which gives a better sight in the available income.

The spatial inequality of Airbnb revenue in Amsterdam