The willingness to pay for shop diversity
Evidence from the Amsterdam housing market
July, 2017 Yasmin Buijs, 10647066University of Amsterdam (UvA)
MSc Business Economics: Real Estate Finance & Finance Master thesis
Abstract
The last couple of years, the number of tourism-‐related shops has increased enormously, especially in the city-‐centre of Amsterdam. This boosts the local economy, however, there has been a growing controversy about the side effect this increase might have. One of the most controversy side effect is that the diversity of shops is at risk, which could influence the surrounding house prices. This thesis investigates this by looking at the effect of different shop diversity measures on house prices using data over a period from 2008 to 2016. A hedonic price model is used for this analysis. The regressions produce both positive and negative results. The absolute measures suggest a positive linear effect of tourism-‐related shops on house prices, but the relative measure suggests a negative one. Nevertheless, the models give significant negative results for the non-‐linear variables from which the conclusion can be drawn that the relationship between tourism-‐related shops and house prices in areas with a medium number of tourism-‐related shops is negative. This relationship becomes even more negative when there is a high number of tourism-‐related shops. However, this difference in coefficients is not significant for the density variable. Furthermore, the gravity certification index show that distance to tourism-‐related shops is relevant, and that if the number of tourism-‐related shops reach a certain threshold, the house prices are negatively related to tourism-‐related shops.
Statement of originality
This document is written by student Yasmin Buijs, who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of contents
Abstract... 1
Statement of originality... 1
1. Introduction... 3
2. Literature review... 6
2.1 Tourism-‐related shops and externalities……… 6
2.1.1 Tourism-‐related shops and economy……… 6
2.1.1.1 Economy and house prices………. 7
2.1.2 Tourism-‐related shops as non-‐residential land use………. 7
2.1.3 Tourism-‐related shops and shop diversity………. 8
2.1.3.1. Shop diversity and house prices………. 9
2.2 Tourist-‐area lifecycle……… 9
2.3 Solutions……… 10
2.4 Measures for diversity……… 11
2.5 Relevance ……… 12
3. Methodology...13
3.1 Hypotheses... 15
4. Data…... 16
5. Results……... 21
5.1. Effect of the three diversity measures on house prices………. 21
5.1.1 Effect market share tourism related shops on house prices………. 21
5.1.2 Effect density tourism related shops on house prices……….. 24
5.1.3 Effect of the gravity certification index (GCI) on house prices………. 26
5.2 Summarizing the results……… 29
6. Robustness checks………. 30
6.1 Gravity certification index with different radii……….. 30
6.2 Density variable with different radii………. 32
6.3 The three diversity measures with interaction term……… 33
7. Implications and limitation... 35
8. Conclusion……… 36 References... 39 Appendix... 41
1. Introduction
The Netherlands, especially Amsterdam, is a popular tourist destination. The last decades, Amsterdam has experienced a significant growth of tourism, the number of foreign visitors for example has grown with 25% over the last 5 years (Municipality of Amsterdam, 2016). This increase in the number of tourists also leads to more capital inflow in tourist destinations. However, a research recently done by the weekly paper ‘de Groene Amsterdammer’ and the research platform ‘Investico’ claims that the economic advantages of tourism for Amsterdam are overestimated, and the costs underestimated. The profits end up at a small group of large entrepreneurs, which are often part of a foreign company. They argue that the idea that the bustle and nuisance are annoying but that the city benefits as a whole from the visitor’s, is a myth (Milokowski and Maas, 2017).
In the city-‐centre of Amsterdam the influence of Amsterdam’s growing popularity among tourists and visitors is increasingly visible in the retail and supply facilities. More entrepreneurs sell products that can be sold in large quantities and have a high profit margin, such as waffles, ice cream, and cheese. This potentially high profit margin combined with the big number of passers-‐by in the city-‐centre, makes the city-‐centre an attractive location for this type of entrepreneurs. The consumer, including the inhabitant of the city, is also looking for convenience and one-‐stop shopping. With the increased popularity of shopping at supermarkets, the traditional neighbourhood shop (baker, butcher, greengrocer) is under pressure. Furthermore, there is a lack of follow-‐up for older entrepreneurs. Due to the increased popularity of some tourist areas, these neighbourhood stores and facilities for residents and other specialty stores are further under pressure and disappear from some shopping areas. This is due to financially attractive acquisitions and entrepreneurs who adjust their range to the new opportunities created by the influx of large groups of tourists. Consequently, in some busy shopping areas, the typical diversity of the retail and catering sector is decreasing, and there is a one-‐sided supply of fast-‐food stores, facilities (for example leisure) and catering (Municipality of Amsterdam, 2017).
The sharp rise of especially tourism-‐related shops is being identified by a growing group of residents and entrepreneurs as a threat to the attractiveness of the city-‐centre. There are petitions and other initiatives in the (social) media that asks for attention. From the reports of the municipality of Amsterdam (‘Stand van de Balans’ and ‘sturen op een divers winkelgebied’)
can be concluded that residents are complaining about these changes and they argue that the city does not belong to the residents of Amsterdam anymore. In 2016, a survey was used as an instrument to measure the opinion of the residents about the diversity of shops. The survey responses indicate that 43% of the residents of the district Centre-‐West reason that the diversity of shops is too low. This dissatisfaction is probably related to the emergence of the Waffles-‐Ice-‐Nutella shops and other tourism-‐related shops and the increase of big chain stores in the city centre (Municipality of Amsterdam, 2016).
An attractive, mixed city is good for Amsterdam. Its attractiveness strongly determines the image, economic strength and international reputation. Tourism-‐related shops are not appealing to the city-‐centre of Amsterdam, and excessive dominance of these shops can have an opposite effect on the number of tourists and visitors. In the long-‐term this is detrimental to the attractiveness of the city-‐centre (Municipality of Amsterdam, 2017).
Since the increase in tourism-‐related shops, a debate is originated. There are reports written about the dissatisfaction of the different stakeholders, especially that of residents. Therefore, the aim of this thesis is to answer the following question:
What are residents willing to pay for shop diversity?
This is investigated by using high grade house price data that is made available by the NVM, and data of tourism-‐related shops which is made available by Locatus. With this data, a hedonic price model is used to investigate the research question. The main independent variables of interest are the tourism-‐related shops density measure, the tourism-‐related market share measure, and the gravity certification index of tourism-‐related shops. Expected is that small proportion of tourism-‐related shops have a positive effect on house prices, because it boosts the local economy, but when the diversity of the supply of shops is put at risk, it will negatively affect the surrounding houses.
The relation between the tourism sector and the economy has been investigated elaborately. Lots of research has found a positive linkage between tourism and economic growth in both the short and long term (Bimonte et al., 2012). On the one side, the tourism growth boosts local economy and make residents better off. On the other side, the tourism growth generates negative environmental and social externalities that make residents worse off (Biagia et al., 2012). Previous literature has evaluated the effect of local amenities and disamenities on tourism-‐related accommodations. Other studies focused on the effect of amenities such as
open spaces, public parks, natural areas, golf courses and other types of amenities on property values. Also, the effect of disamenities such as, noise, HVTLs, waste sites, crime and other types of disamenities on property values has been investigated. However, these studies are amenity-‐specific and are not looking at the effect of the diversity of amenities. Nevertheless, some reports have been written about the situation in Amsterdam, regarding the increase in tourism-‐related shops. They have concluded that the different stakeholders (the municipality, the residents, the entrepreneurs, and the real estate investors) are dissatisfied. However, this is based on surveys, where this analysis will use actual data to test if this dissatisfaction has an impact on house prices. Furthermore, this thesis is a contribution to the social discussion about tourism-‐related shops and can be useful by determining a policy.
The results imply that a consistent linear relationship between house prices and tourism-‐ related shops is missing: the density measures suggest a positive relationship, and the market share measure a negative one. However, the density and market share measure show that there is a negative non-‐linear relationship between tourism-‐related shops and house prices. When houses are in an area with a medium market share or density of tourism-‐related shops, house prices are, on average, 3.4%, lower than houses in an area with no tourism-‐related shops. The dummies for high market share and density of tourism-‐related shops are even bigger, and suggest that house prices in these areas are, on average, 4.2% lower in comparison with areas that have no tourism-‐related shops in the area. Furthermore, the gravity certification index (GCI) of tourism-‐related shops show that the relationship between this variable and house prices have the shape of a downward opening parabola. However, the value of the turning point is relatively high, which implies that only a few observations have a value that exceed the value of the turning point.
The structure of this thesis is as follows: section 2 gives a review of the existing literature followed by the relevance of this thesis; section 3 presents the methodology; section 4 describes the datasets that are used with some descriptive statics; the results, and robustness checks are discussed in section 5 and 6; and section 7 and 8 poses some further implications and concludes.
2. Literature review
This section begins by explaining, using different theories, which effect tourism-‐related shops might have on house prices
.
This is done by putting it into perspective with the situation in Amsterdam. Then some potential solutions to this problem are addressed. After that, the three different diversity measures will be discussed. This section is concluded by an analysis about the relevance of this research.
2.1 Tourism-‐related shops and externalities
2.1.1 Tourism-‐related shops and economy
Tourist destinations worldwide have experienced a significant growth in the influx of money and visitors. This growth has been the result of some socioeconomic changes, such as the increase in disposable income, longer periods off, more money is spent on leisure and the longer life expectancy. Furthermore, this tourism growth has been aided to the improved transportation, this is especially the case for international tourism (Biagi et al., 2015). Whereas tourist expenditures are an important source of funding for both the private and public sector as it generates income and employment for retail and retail area development. Lots of research has found a positive linkage between tourism and economic growth in both the short and long-‐term (Bimonte et al., 2012). Most of these studies argue that tourists and visitors generates economic activity directly in the form of output or sales, labour earnings and employment. They also use the ‘multiplier impact’ of tourism expenditures. This measures how many times money spent by a tourist circulates through a country’s economy (Frechtling and Horvath, 1999). Due to the multiplier effect, many studies agree that tourism generates an important source of income and boosts the economy. Whereas shopping is the second most important expenditure item for both domestic as international tourism, after accommodation. Although, shopping is not often mentioned as a primary reason for travel, it is of great economic importance to local merchants (Turner and Reisiger, 2001). Furthermore, several researchers have concluded that for many tourists a trip is not complete without having spent time on shopping (e.g. Hudman and Hawkins, 1989; Kent et al., 1983). Tourists often make their experience tangible by purchasing souvenirs and acquiring gifts for loved ones. Kent, Shock and Snow (1983) argued that ‘to be able to peruse, to examine, to feel and to think of the joys derived from purchasing certain merchandise is indeed pleasurable to
millions of people, and for them is a minor, if not a major reason for travel’. Therefore, it is more attractive for entrepreneurs to produce and sell products for tourists and visitors: where demand is, follows supply. Tourists are interested in souvenirs and quick bites that can be cheap and simple produced or prepared, so this will be supplied. Especially, selling of large amounts with a large profit margin, like waffles, ice and cheese, has become attractive.
2.1.1.1 Economy and house prices
Research in both the field of tourism economics and housing studies recognize that tourism and tourism-‐related activities can affect housing markets directly as well as indirectly. Directly via the ‘external’ demand that competes with the local for land, housing and services. Indirectly via the tourism-‐related amenities and disamenities that affect the market price of all surrounding houses. Kendall and Var (1985) suggest that the positive impacts of tourism and tourism-‐related activities are: more and better leisure facilities, more parks and gardens, and an increase in employment and business opportunities. However, tourism and tourism-‐ related activities might also simultaneously be the source of various sorts of disamenities that can be a large disadvantage to residents (Biagi and Detotto 2014; Biagi et al., 2012). These negative impacts include: crowding, congestion, noise, litter, property destruction, pollution, environmental degradation, general resentment to the wealth of tourists, loss of wildlife and ad hoc development (Kendall and Var, 1985). As such, the tourism-‐house price relationship is expected to be positive when tourism and tourism-‐related activities boost the local economy, or negative when the negative externalities that tourism-‐activity generates predominate.
2.1.2 Tourism-‐related shops as non-‐residential land use
Li and Brown (1980) have investigated the theory that accessibility at a micro scale increase the value of a house. They argue that proximity to some non-‐residential uses can be accompanied by external diseconomies such as congestion, noise, and air pollution that affect the value of residential properties. Their results show that the accessibility to non-‐residential land uses increase residential property prices, but that this effect is more significant for the accessibility to commercial establishments than for the accessibility to industries. Furthermore, the negative price effect of the related externalities is less significant for commercial establishments than for industries. This means that the accessibility to commercial areas predominates the externalities related with it, this is not the case for the
industrial areas. However, Kain and Quigley (1970) argue that the presence of commercial and industrial uses, results in a statistically significant negative effect on apartment rents and on the values of single-‐family houses. Stull (1975) argued that it is likely that a small proportion of commercial activity is desirable, because of the shopping convenience it affords. However, if this commercial activity reaches a certain threshold, it becomes a community liability rather than an asset, because the disamenities created, outweigh the extra shopping convenience. His results confirm this hypothesis, as he finds that as the proportion of commercial land increases the value of the single-‐family properties tend to rise, but if the proportion of commercial properties in a community exceeds 5% the value of single-‐family homes tends to fall. Grether and Mieszkowski (1980) used data from 16 market experiments in the New Haven, Connecticut metropolitan area. They measured the effect of non-‐residential land uses, such as industry, commercial, high-‐density dwellings, and high ways on the prices of nearby dwellings. However, they did not find a systematic relationship between non-‐residential land use and house prices.
2.1.3 Tourism-‐related shops and shop diversity
However, on the other side of this contribution to the economic prosperity are the negative environmental and social externalities of tourism and tourism-‐related shops. Many academic literature and reports picture these negative effects, that can make residents, property owners and other entrepreneurs worse off. The most actual negative effect of the increase of tourism-‐related shops is the decrease in diversity of supply. Since the policy change in 2009, in which is stated that ice for direct consumption can be sold in a premise without hospitality destination, the number of ice-‐waffles-‐Nutella shops has increased rapidly. Furthermore, as mentioned earlier, the supply of other tourism-‐related shops, cheese and souvenir shops, has also increased. This increase has resulted in a debate about the diversity of shops in the centre of Amsterdam, where some are speaking of a ‘monoculture’. There is even a petition launched, called ‘Red de Winkels’ (Save the shops), to combat the degradation of the city-‐ centre1. The adjustments of supply to the change of demand due to the increase of tourists and visitors, leads to busy spots in the city-‐centre and the overrepresentation of shops and facilities that are mainly focusing on the ‘mass tourism’, with low-‐grade products and fast
consumption (Municipality of Amsterdam, 2017). A good mix of residents, workers and visitors is important for the attractiveness of Amsterdam. The breakdown of the balance to dominance of the tourists can push away residents by alienation and can affect the attractiveness of the city in the long-‐run (municipality of Amsterdam, 2017). Furthermore, panellists in the RMA report also reasoned that the municipality of Amsterdam should take measures to prevent a monotone supply, and decrease the number of tourism-‐related shops where waffles, ice, cheese, souvenirs are sold.
2.1.3.1 Shop diversity and house prices
Places with overrepresentation of shops and facilities that focus on tourists and visitors, combined with crowded streets, results in a higher risk of pollution and conflicts about the use of public spaces. This has a negative effect on liveability of the area. Perception is more important for tenants and homeowners nowadays, which also means an attractive living environment: public space, security etcetera (Municipality of Amsterdam, 2017). Much academic research argues in their turn that a less attractive living environment results in lower demand. Ellis (1967), Stegman (1969) and Richardson (1971) are one of those researchers that investigated this, and they confirmed the hypothesis that residential site choice is not just determined by accessibility but rather by environmental attributes of the area. Ceteris paribus, favourable environmental attributes increase demand for these particular houses and this boosts prices, conversely unfavourable attributes reduce the number of potential buyers and this reduce prices.
2.2 Tourist-‐area lifecycle
Butler (1980) introduced the tourist area lifecycle (TALC). The basic idea of the model is that residential areas move through five stages. The five stages are: exploration, involvement, development, consolidation and stagnation. Following stagnation, a tourism area may start a new development phase (rejuvenation), or may continue to stagnate or may decline. During the first stage, the exploration stage, the shopping district serves the needs of the residents. There are small number of tourists transiting through the city and perhaps purchasing incidental products. In the next stage, involvement, tourists are beginning to become a more visible part of the community. Retailers are still serving the locals but are enlarging their product mix to start targeting tourists. As tourism expands, the area move into the
development stage. In that stage the shopping districts becomes more attractive to tourists and targets both the locals and the tourists: the destination benefits from increasing rates of growth. In the fourth stage, the consolidation stage, the stores are fully targeting on tourists: they are selling mementos, nonessentials and niceties for everyday life. The shopping district is no longer attractive for the residents (Snepenger et al.,2003). Finally, the shopping district move in to the stagnation stage. In this stage the arrival numbers reach a peak, because the city is now seen as unfashionable. This makes it difficult to maintain the number of arrivals. After this stage, the city enters a decline or rejuvenates (Moore and Whitehall, 2005). The model of Butler is useful in proposing the potential of a decline if certain problems are not addressed. Butler argues that a decline may be the result from a lack of good overall management of tourism, or from an absence of long-‐term planning for the destination or from failure to recognize that there are limits to growth (Hovinen, 2002). From this can be concluded that it is necessary for the municipality of Amsterdam to take measures, otherwise the potential for a significant decline exists.
2.3 Solutions
To reach a solution for encouraging the diversity of shopping or prevent impoverishment of the supply, a good cooperation between stakeholders is essential. The four most important stakeholders for a shopping area are: the municipality, the entrepreneurs, the real estate investors, and the residents. The municipality has various interests in a well-‐functioning shopping area. These interests are physical, social and economic: the municipality is held responsible for an overall viability, a good spatial planning, security, local economic development and stimulating employment. A decline in product range may have a negative impact on (all) the municipal interests. This makes the municipality an important stakeholder. Entrepreneurs are also important stakeholders as they have a very strong interest in the proper functioning of the shopping area. An attractive shopping area increase the likelihood of achieving their two main objectives: continuity of operations and profit maximization. Another important group of stakeholders is the real estate investor. They are also interested in maintaining or increasing the attractiveness of the shopping area, because both the amount and continuity of the rents and the value of the property depends on this attractiveness. Finally, the residents of the area. They have both as residents and as consumer direct interest in the good quality of life in the shopping area. As consumer because they want a good price
and a great diversity of products and both residents and consumers have interest in a well-‐ functioning, good accessible and safe shopping area (Municipality of Amsterdam, 2017). As mentioned earlier, a good mix of residents, workers and visitors is important for the attractiveness of Amsterdam. To limit a decrease in diversity of shops, not only the municipality, but also the other stakeholders should take their responsibility, because achieving a certain tenant mix is not easy. Instruments that the municipality can use to limit a decrease in diversity is the adjustment of the zoning plan. This is only possible for spatial reasons or to promote the quality of life. Furthermore, they could buy some properties in the city-‐centre of Amsterdam to get influence on the tenants. Real estate investors or owners can select their tenants based on the zoning plan, in addition, they can, if desired, use lease incentives. Residents influence the shopping area by buying more locally at several (small) shops. Furthermore, entrepreneurs could change the shopping area by influencing the shopping behaviour of consumers through consumer loyalty, street branding, and changing their supply to serve the needs of the residents and consumers. Another instrument that the can be used is by encouraging and stimulating stakeholders in a specific area to jointly develop a vision and plan about the tenant mix. Some of these instruments are deployed. In November 2016, all stakeholders that have interest in the maintaining and increasing the attractiveness of the city-‐centre in Amsterdam signed ‘het binnenstadakkoord’ (the city-‐centre agreement). The purpose of the agreement is to develop a common vision, that combines the interest of all the stakeholders, which can create a vital, diverse, viable and attractive supply of shops (Municipality of Amsterdam, 2017). Furthermore, in March 2017 the municipality decided to adjust the zoning plan: they want to determine per street which shops can establish their selves (‘De jacht op Nutella’s wordt nu toch geopend’, 2017).
2.4 Measures for diversity
To measure the effect of a decrease in diversity of shops on the surrounding house prices, a similar approach as Linn (2013) is used. He has analysed in his paper the effect of voluntary brownfields programs on nearby property values and as brownfields have some similarity with tourism-‐related shops, his analysis is useful. The similarity between brownfields and tourism-‐ related shops is that both have multiple entities spread over the city-‐centre and that there are multiple entities located within a certain area. He used two measurements to investigate the impact of brownfields. The first one measures the density by counting the number of
brownfields within a given radius and the second one is a gravity index which measures the sum of the inverse distance of all brownfields: this index places greater weight on nearby sites compare to those further away. Where the first measure assumes that distance is irrelevant,
the second measure assumes that the distance influences the property value inversely. These two measures are both used. Where the density measure will use a one kilometre radius around the property, as assumed is that beyond one kilometre the effect will be negligible, and the surface of the circle will otherwise capture a too big part of the city. A radius below one kilometre is also assumed to be improper as this will bias the results of houses with a low number of tourism-‐related shops in their surroundings. Furthermore, there could be a measurement error in the coordinates of both the houses and the tourism-‐related shops which implies that a radius below one kilometre could be imprecise.
Another possibility to measure the effect of a decrease in diversity of shops, is to use the market share of tourism-‐related shops. The last couple of years, the total number of retail shops in the city-‐centre of Amsterdam has decreased (Onderzoek, Informatie en Statistiek, 2014). When using the market share of tourism-‐related shop as proxy for diversity of shops, the trends in the total number of retail shops are taken into account. As the data of Locatus only contain the total number of shops per four-‐digit zip code, the market share of tourism-‐ related shops is measured per four-‐digit zip code.
2.5 Relevance
Since the significant increase in tourism-‐related shops, a debate is originated. There are reports written about the dissatisfaction of the different stakeholders, especially that of residents. However, this dissatisfaction is based on surveys. To investigate whether this leads to a different behaviour, actual data should be used. Using a survey to gather empirical data, has some disadvantages over using actual data. One of the important disadvantages is that responses to surveys may not reflect the true beliefs, attitudes, or behaviours of the respondents (Salant and Dillman, 1994). By using data of actual transaction prices, it can be measured if residents think the area is less attractive. If that is the case, there would be less demand which results in lower transaction prices. This is relevant for real estate investors, but also for policy makers. When the house price effect of a lack of diversity of shops is known, they can take that into account when constructing their policies.
3. Methodology
The focus of this thesis is to investigate the effect of the increase of tourism-‐related shops on the surrounding house prices in Amsterdam. The analysis uses a hedonic price model including control variables such as house characteristics and fixed effects. The hedonic price model is mostly used in the existing literature for measuring effects on the housing market. Examples of studies that rely on the hedonic price model are Theebe (2004), Koster and Ommeren (2015), and Dekkers and Van der Straaten (2009). The hedonic approach deducts values by analysing observed market data, like property transaction prices. By doing a regression, a hedonic price model analyses the contribution of physical and locational property characteristics to property transaction prices (Theebe, 2004). The diversity of shops is measured in three ways: the density, the market share and the gravity certification index of tourism-‐related shops. The density variable measures the number of tourism-‐related shops in a radius of one kilometre of the sold property in the year of the transaction. The market share of tourism-‐related shops divides the number of tourism-‐related shops by the total number of shops in the retail sector with the same four digits in their zip code as the property i in the year of the transaction date t. The gravity certification index (GCI) of tourism-‐related shops is constructed as the sum of the inverse distance of all tourism-‐related shops. The summation is taken over all shops within one kilometre, under the assumption that tourism-‐related shops further away do not affect house prices. The dependent variable is the natural logarithm of the house transaction prices. log 𝑝𝑖𝑡 refers to the natural logarithm of the transaction prices
of property i at time t and Diversityproxy refers to one of the three measures for the diversity of shops. The hedonic model looks as follows:
log 𝑝𝑖𝑡 = 𝛼+ 𝛾Diversityproxy𝑖t + 𝛽’𝑥𝑖𝑡 + 𝜂𝑗 + 𝜃𝑡 +𝜖𝑖𝑡 (1)
Where 𝛼 is a constant, 𝛾𝑖𝑡 is the main independent variable of interest, and measures the
house price effect of the lack of diversity of shops in that area. Whereas different house characteristics can influence the house price, 𝑥𝑖𝑡 is included in the model and represents a
vector of house characteristics and 𝛽’ measures the impact of these characteristics.
House characteristics are variables such as the size of the house, the number of rooms, house type dummies, dummies for garage, garden, maintenance quality, whether the house is listed as cultural heritage, and construction year dummies (Dröes & Koster, 2016). Furthermore,
house prices can be influenced by neighbourhood characteristics, because neighbourhood residential properties share locational amenities. For example, the same police and fire departments, access to the same public schools, the same accessibility to transportation networks, and proximity to the same externalities (Basu and Thibodeau, 1998). Therefore, level six zip code fixed effects are included, and 𝜂j measures this effect2. Additionally, to
control for time trends, 𝜃𝑡 is included and measures the year and month fixed effects. Finally,
𝜖𝑖𝑡 is an identically and independently distributed error term. The error term is clustered on a
neighbourhood level to capture the effect of differences between neighbourhoods on the house prices. This clustering correct for heteroscedasticity and serial correlation. Clustering on six-‐digit zip code would give a problem as there are often only a few observations per zip code, this would result in an underestimation of the standard errors and an overestimation of the significance (Correia, 2015). Clustering on neighbourhoods gives enough observations per neighbourhood in the sample and assumed is that there is enough independence across clusters, but correlation within clusters.
Furthermore, to test if the relationship between tourism-‐related shops and house prices changes when the proportion of tourism-‐related shops changes, dummies for the market share and density measures are included. The independent variables of interest are divided into three dummy variables, to test for the non-‐linear effect of diversity of shops on surrounding house prices. In this way, it is possible to check whether the different classes of the independent variable have a different effect on the surrounding house prices. To avoid perfect multicollinearity, the ‘none’ dummy is excluded from the model. The inclusion of the dummies will give the following model:
log𝑝𝑖𝑡=𝛼+ 𝛾Mediumdummy𝑖t + δHighdummy 𝑖𝑡 + 𝛽’𝑥𝑖𝑡 + 𝜃𝑡 + 𝜂𝑗 + +𝜖𝑖𝑡, (2)
To test if a tourism-‐related shops are desirable till a particular threshold, a squared term is used for the gravity certification index. Furthermore, using a squared term has the advantage that it is independent of the bins that are chosen, which is not the case for the dummy variables. Nevertheless, a squared term is more sensitive to outliers.
2 The results remain the same when neighbourhood fixed effects on four-‐digit zip code level
are included (Table A3). The turning point of the GCI variable only comes at a slightly lower value.
If tourism-‐related shops are desirable till a particular threshold, the sign of the GCI𝑖t coefficient
should be positive, while the sign of the squared term should be negative. This would result in a downward-‐opening parabola, with a maximum at a positive proportion of tourism-‐related shops. This results in the following model:
log𝑝𝑖𝑡=𝛼+ 𝛾GCI 𝑖𝑡 + δSquareofGCI𝑖𝑡 + 𝛽’𝑥𝑖𝑡 + 𝜃𝑡 + 𝜂𝑗 +𝜖𝑖𝑡, (3)
3.1 Hypotheses
Using this approach, the hypothesis that the lack of diversity of shops due to the increase of tourism-‐related shops has negatively influenced the surrounding house prices in Amsterdam, can be tested. The existing literate argues that tourism can boost the local economy or that the negative externalities that tourism and tourism-‐related activities generate predominate. It is expected that more tourism-‐related shops bring more negative externalities as crowding, congestion and environmental degradation than positive ones. Especially, the effect of environmental degradation is expected to play an important role which is result of the decrease in the variety of shops. However, as Stull (1975) finds that a small proportion of commercial activity is desirable, but if it reaches a certain threshold when the disamenities created outweigh the extra shopping convenience, house prices will fall. In the same way, a small proportion of tourism-‐related shops is expected to have a positive effect on house prices, because it boosts the local economy, but when the diversity of the supply of shops is put at risk, it will negatively affect the surrounding houses. Therefore, the hypothesis is that a small proportion of tourism-‐related shops is positively related with house prices, however if the diversity of shop supply is at risk, house prices and tourism-‐related shops will become negatively related.
4. Data
This analysis uses two databases to answer the research question about the effect of tourism-‐ related shops on the surrounding prices in Amsterdam. The first database is made available by the Dutch Association of Realtors (NVM) and contains house price data from 2008 to 2016. The second database is made available by Locatus, which is a company that collects information about stores, shopping areas, and footfall. From this database data is collected about retail stores and tourism-‐related shops in Amsterdam. With this data, the market share, density and gravity certification index of tourism-‐related shops can be calculated. The dataset of Locatus covers all the tourism-‐related shops in Amsterdam. It contains 2,042 observations of tourism-‐related shops in Amsterdam of which 1,526 are located the city-‐centre of Amsterdam, over the period 2008-‐2016. Figure 1 and 2 show the density of tourism-‐related shops per four-‐digit zip code in Amsterdam in 2008 and 2016. These figures show that most tourism-‐related shops are clustered in the city-‐centre and that only a few areas in the city-‐ centre experienced a growth in tourism-‐related shops.
Figure 1 -‐ Density tourism-‐related shops per PC4 in 2008
100-150 tourism-related shops 75-100 tourism-related shops 25-75 tourism-related shops 10-25 tourism-related shops 2-10 tourism-related shops 1 tourism-related shop No tourism-related shops
Figure 2-‐ Density tourism-‐related shops per PC4 in 2016
Another thing that becomes clear is that every sector experienced a large growth in the city-‐ centre during the last years (Table 1). For example, Waffle-‐Ice-‐Nutella shops went from 12 shops to 40 shops in the city-‐centre, which is an increase of more than 230%. Table 2 shows the growth in tourism-‐related shops for 5 four-‐digit zip codes in Amsterdam; only these zip codes are tabulated as they only experienced a significant growth during 2008-‐2016. Table A1 in the appendix shows the number of tourism-‐related shops for all four-‐digit zip codes. The tabulated zip codes are in the city-‐centre of Amsterdam, where most tourism-‐related shops are located. Especially the areas with 1017 and 1012 in their zip code have experienced a substantial increase in tourism-‐related the last years, this is also shown in figure 1 and 2.
Table 1 -‐ Descriptive statistics: number of tourism-‐related shops in city-‐centre of Amsterdam
Number of tourism-‐related shops
2008 2009 2010 2011 2012 2013 2014 2015 2016 % change Cheese 14 15 15 14 22 24 29 35 39 179% Souvenirs 108 115 121 131 132 138 132 127 139 29% Waffle-‐Ice-‐ Nutella 12 11 15 18 11 20 26 23 40 233% Total 134 141 151 163 165 182 187 185 218 63% 100-150 tourism-related shops 75-100 tourism-related shops 25-75 tourism-related shops 10-25 tourism-related shops 1-10 tourism-related shops 1 tourism-related shop No tourism-related shops
Table 2 -‐ Descriptive statistics: growth in tourism-‐related shops in city-‐centre of Amsterdam over the period 2008-‐2016
PC4
1011 1012 1015 1016 1017
Number of tourism-‐
related shops 2008 6 89 4 7 25
Number of tourism-‐
related shops 2016
9 132 7 16 46
Percentage change 50% 48% 75% 129% 84%
Figure 3 visualizes the numbers of Table 1 and shows that the total number of tourism-‐related shops in the city-‐centre of Amsterdam increased significantly. It went from 119 shops in 2007 to 226 shops in 2017, which is an increase of 63%. This figure also shows that souvenir shops are the biggest part of the tourism-‐related shops, however, as Table 1 shows, this sector experienced the smallest relative change in number of shops. While Waffle-‐Ice-‐Nutella shops have experienced the largest relative increase of 233%.
Figure 3 – Number of shops per branche in the city-‐centre of Amsterdam
From the above figures, it can be concluded that the number of tourism-‐related shops in the city-‐centre of Amsterdam have increased in all sectors. Figure 4 shows that the total number of shops in the city-‐centre of Amsterdam have decreased with more than 300 shops, which
0 50 100 150 200 250 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 # of sh op s
is approximately 6%. This has resulted in an increase of the market-‐share of tourism-‐related shops: it went from 2% to 4%.
Figure 4 -‐ Total shops and market share tourism-‐related shops in the city-‐centre of Amsterdam
Table 3 shows the descriptive statistics of the main independent variables and of the database that is supplied by the NVM. The database provides a variety of variables, including information on transaction prices and house characteristics such as house size, house type and construction year. The sample period of 2008-‐2016 gives 67,953 observations in Amsterdam after removing some outliers. Observations with outliers in transaction price, house size, and number of rooms are deleted. Observation with a transaction price below 50,000 euros are dropped. Furthermore, all houses with more than 25 rooms are dropped. Finally, some observations with a living space of 1 square metre are excluded.
From this database are also some control variables created. These include dummies for house type and construction year brackets. Furthermore, parking and cultural heritage dummies are created and these equal one if the house offers a parking facility or is marked as cultural heritage. The maintenance dummy indicates the sum of the ratings from inside and outside maintenance and if this is more than 15, the dummy equals one. The inside and outside maintenance variables are both on a scale from 1-‐10. Finally, some year and neighbourhood fixed effects are included. The latter include level six zip code identifiers.
Using both datasets, the main independent variables are created and divided into three dummy variables to test for the non-‐linear effect. The first variable, the market share of tourism-‐related shops, has a value of one when the percentage of this variable is 0%. This was
0,0% 0,5% 1,0% 1,5% 2,0% 2,5% 3,0% 3,5% 4,0% 4,5% 5500 5600 5700 5800 5900 6000 6100 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 % m ar ke t s ha re # of o bs er va tio ns