Introduction

“Touristification” is a concept recently used to understand the process in which various stakeholders interfere with the territory and transform it through tourist activities. The term implies a process of changes in the socioeconomic dynamics of a territory (Ojeda and Kieffer 2020). Touristification has also gained traction in the media, with connotations of tourism phobia and gentrification (Jover and Díaz-Parra 2022; Outón 2020). Its reach extends beyond public discourse into academia, particularly in anthropology and geography, where its conceptual and ideological interpretations diverge from its original intent (San Martín et al. 2019). Saarinen et al. (2017) perceive touristification as a synthesis of historical processes and power relations, aiding in understanding tourism within a territory. While various approaches have generated a proliferation of the concept, touristification is often perceived as just an independent phenomenon of tourist agglomeration (Fournier and Knafou 2013; Jansen‐Verbeke and Dewailly 1999). A critical gap in the current discourse is the lack of a cohesive theory and an empirical foundation for analysis, thus creating self-referential issues and leaving the concept open to ambiguous interpretations or even being criticised as an empty concept (Ojeda and Kieffer 2020). In this study, we aim to explore the concept of touristification from an ecological succession analogy.

It is imperative to emphasise that this research adopts a transdisciplinary approach to uniquely conceptualise touristification as an ecological succession process. This method not only examines its impacts on tourism ecosystems but also confronts the prevalent critiques of ideologically driven theories in the field, which often lack empirical substantiation (Ojeda and Kieffer 2020). Bridging this critical gap, our paper employs an ecological perspective to explore the effects of touristification, with a particular emphasis on retail diversity. The theorisation in this paper is the main scholarly contribution. The empirical examination serves as a critical test to validate the implications of our model of touristification on retail diversity. The empirical model adopts the methodology of Cheung and Yiu (2023), focusing on retailer dissimilarity as the dependent variable, diverging from the common practice in related literature that typically examines rental levels. This dual approach, blending theorisation with empirical verification, fortifies our understanding of the dynamics at play in tourist-impacted urban environments.

In this study, we argued that touristification could trigger a retail displacement effect within neighbourhoods, catalysed by the influx of tourists. This phenomenon occurs when the impact of mass tourism reshapes the social fabric of neighbourhoods, causing shops to cater predominantly to tourist preferences rather than those of local residents. Such a change can drastically reduce the retail diversity of a neighbourhood, often leading to a severe displacement of retailers who primarily serve the local residents. This displacement process is marked by the loss of affordable rental spaces for local shops and the eventual replacement of resident-oriented retailers with those focusing on tourists, as revealed from the interviews in Milano, González-Reverté and Benet Mòdico (2023).

From an ecological standpoint, touristification can be analogous to the ecological succession of non-indigenous species (i.e., tourist-oriented retailers) that influences the pre-existing ecological community (i.e., local retailers), leading to the loss of diversity. Therefore, we postulate that touristification acts as a socio-ecological process that reduces retailer diversity, as evidenced by the diminishing tenant mix in a retail environment. The COVID-19 pandemic has highlighted the vulnerabilities of such a reduction in retail diversity (Gómez-Varo et al. 2024). With travel restrictions in place and a drastic drop in number of tourists, the retail sector faced unprecedented challenges, with many businesses struggling to adapt to the loss of their local and primary customer base.

The paper is structured as follows. First, we review the literature on touristification in tourism studies. Second, we theorise touristification as an ecological succession process and hypothesise that large numbers of same-day shoppers will reduce retail diversity. Using Hong Kong as our case study, we empirically develop and test a refutable hypothesis from the theory of touristification. We hypothesise that ceteris paribus, a substantial increase in the number of tourists, especially same-day-in-town tourists (i.e., day-trippers) added to a retail ecosystem, can cause a potential disturbance in retail diversity in the local retail market. Finally, in the conclusions section, we discuss the need to develop destination resilience as a potential strategy to sustain tourism growth.

Literature review

Touristification: the intersection of tourism, gentrification, and urban transformation

Touristification, as suggested by Cocola-Gant (2018), serves as a heuristic device that recognises the role of tourism activities in driving gentrification. This process was often observed as a transformation of a middle-class neighbourhood into affluent enclaves characterised by a proliferation of corporate entertainment and tourism venues (Gotham 2005). In many global cities, tourism development is recognised not only as an impetus for economic growth but also as a primary economic driver. Concurrently, the escalation of tourism presents substantial challenges, reshaping resource allocation and altering the socioeconomic fabric of cities. The influx of tourists in a confined neighbourhood frequently sparks conflicts between residents and visitors, particularly in the context of space usage, with the retail landscape being a prominent battleground (Cheung and Li 2019).

Indeed, touristification tends to be a local manifestation, leading to socio-spatial segregation (Jouault 2018), urban conflicts (del Romero Renau 2018), and cultural levelling (Gravari-Barbas and Jacquot 2019). The process has been characterised as an “excessive negative impact of tourism” (Capocchi et al. 2019: 2), influencing the authenticity of cities and neighbourhoods, affecting the tourist experience and local quality of life, causing irritation and annoyance to residents (Milano et al. 2019) and even deprive a social right to the city (Diaz-Parra and Jover 2021). Provided that the process of touristification can displace the upper-middle-class or reduce their consumption (Nofre et al. 2018), more and more researchers regarded touristification and gentrification as two different processes (Cummings 2015; Romero-Frías and Leontidis 2020). In contrast with the gentrification process, touristification affects a neighbourhood’s residential sector and retail ecosystem. Typically, the process is referred to as a displacement of the local residents (and, in this study, local retailers), and the neighbourhood becomes less inclusive (and, in this study, less diverse).

Jansen‐Verbeke and Dewailly (1999) advocate that the concept of touristification devises a new geographical approach to better understand, along with tourist activities, the complex relationships between stakeholders, economic flows, appropriation of space, and transformations in the landscape. The concept of touristification has led to many studies that examine the causes and consequences of the process in tourism cities worldwide (Gotham 2018). One key effort of recent studies of touristification has been to differentiate it from gentrification patterns and trends, including the regulation of urban real estate markets (e.g., Airbnb, vide Gutiérrez et al. 2017; Cheung and Yiu 2022), the actions of transnational corporations, and shifting patterns of finance (Lees et al. 2013; López-Morales et al. 2016; Cocola-Gant 2018). Touristification research has mainly focused on residential displacement, touristic land uses and the elimination of mom-and-pop businesses by commercial entertainment corporations (Gotham 2005). The tourisification of hip-hop culture influences the interrelations between culture, tourists, and attractions and draws wider public attention (Xie et al. 2007). Consequently, the concept has been adapted to examine the connection between tourism, gentrification, and socio-spatial restructuring (Lees et al. 2013, 2016; Colomb and Novy 2016). More recent tourism research is concerned with the unprecedented commercialisation and corporatisation of culture and art festivals (Szmigin et al. 2017). Increasing commercialisation and touristification of festivals have led to the erosion of the values and significance of festivals, with the meaning of a festival gradually deteriorating into a commercial gimmick (Choi et al. 2021).

However, the connections between touristification and displacement of resident-serving retailers remain weakly theorised: this is perhaps unsurprising, given the seeming absence of studies on the impact of touristification on retail diversity. So far, Most academic discourse mainly focuses on the physical revitalisation of locations, concentrations of poverty, and resident-visitor irritants, but less on the impact of touristification on retail diversity. The over-emphasis on direct displacement underplays the wider impacts of touristification, especially on people and businesses who remain in the community (Chen et al. 2022). In the literature, very often, they considered the touristification impact on the retail sector as a tourism gentrification process (Barata-Salgueiro et al. 2017; Blázquez-Salom et al. 2019). Bridge and Dowling (2001: 99) assert that retail landscapes ‘provide a particularly sensitive indicator of the balance of forces in gentrified neighbourhoods.’ This inference echoes Smith and Williams’ (2013) argument that residential gentrification is intimately linked to the emergence of ‘modern “trendy” retail and restaurant districts’ which constitute ‘a visible spatial component of broader social transformation’.

Tourism-induced gentrification in the retail sector

Hubbard (2018: 295) suggests that ‘this type of [retail] gentrification story, involving the transformation of ‘local’ shopping streets into spaces of middle-class consumption, is one that occasionally captures media headlines, but remains little examined in urban studies.’ Indeed, such a high street retail transition is often slow and piecemeal (Kern 2015) and tends to excite less academic commentary than other gentrification associated with the super-rich’s colonisation of already-affluent neighbourhoods, i.e., ‘super-gentrification’ (Atkinson 2016). While such “tourism gentrification on retail sector” is sometimes referred to in urban studies, it is rare to find the discussion engaging the role of retail change in broader touristification processes and providing empirical evidence on the process (Hubbard 2018).

Recent studies, however, have begun to explore the long-run and short-run effects of international tourists’ retail shopping on domestic retail sales. Gholipour et al. (2021) applied an autoregressive distributed lag model to the retail sales data of 58 countries from 2004 to 2018; they found that retail industries in large and populous countries with popular tourist attractions are positively affected by a large number of international tourist arrivals and their retail spending. Retail shopping also has a positive and significant impact on retail sales in the long run. Perles-Ribes et al. (2018) compared the economic performance of holiday and residential tourism destinations in Spain and suggested a positive association between the economic development of destinations and their retail activities.

Across the globe, from Brixton in London to Brooklyn in New York City, from Shimokitazawa in Tokyo to Sodermalm in Stockholm, there are numerous instances where local shopping streets, once bustling with tourists, have undergone significant transformations. These areas, previously known for their traditional, immigrant, or family-owned stores, now increasingly mirror the ‘global ABCs of gentrification: Art galleries, Boutiques and Cafes’ (Zukin 2016, p. 31). Hong Kong, as a cosmopolitan tourism city, is no exception. Hong Kong, a vibrant tourist city, exemplifies this trend. Sheung Wan, hailed by Forbes as ‘one of Hong Kong’s coolest neighbourhoods’, epitomises this shift. Once a traditional area, it is now replete with artisan cafes, hidden bars, galleries, studios, and street art, manifesting an increasingly hip and expensive locale. This phenomenon, extensively documented in the literature on touristification, suggests that the upscaling of local shopping streets often marginalises local and low-income residents. The transition tends to benefit new businesses catering to a different demographic, while traditional community establishments gradually diminish, altering the socioeconomic fabric of these neighbourhoods.

Intuitively, promoting a diverse and inclusive high street of inner-city shopping in contemporary urban planning is desirable. This argument is made on the basis that in different communities, retail change alters more than the goods and services available: it also profoundly changes the way different neighbourhoods are represented, experienced and lived, often alienating and displacing longer-term residents in the process. So, at a time when concepts of diversity and mixed-use shopping streets are ‘rendered in rather reductive-instrumentalist terms as being socially desirable’ (Griffiths et al. 2013: 036:3), it is equally important to define diversity and inclusion clearly. The touristification of neighbourhoods, marked by the replacement of local shops with chain stores, trendy pop-up barista coffee shops, and new retail outlets, often leads to a deeper touristification process. This is characterised by the provision of goods and services catering exclusively to tourists. Despite its significance, the change in retail composition has not been a prominent theme in the literature on urban or touristification studies. The current discourse remains ambiguous about the extent to which retail geography contributes to our understanding of touristification as a global phenomenon. This ambiguity suggests an interdisciplinary need for renewed dialogue on how consumption spaces interact with urban change and conflicts (Hubbard 2018). Recognising the importance of retail diversity in the context of touristification is essential. It enables a deeper appreciation of the rich dimensions of diversity and underscores the value of diversity within the community.

Touristification and species-area relationship

To conceptualise touristification as an ecological process, namely the species-area relationship (SAR) within the tourism-retail ecosystem, it is important to start with a foundational understanding of SAR and then draw parallels to retail environments. The Species-Area Relationship is a fundamental ecological principle stating that the number of species (S) increases with the area (A) sampled, following a power-law relationship, typically expressed as \(S=C{A}^{Z}\), where C and Z are empirical constants (Fig. 1). This relationship underscores how larger areas tend to support more species due to increased habitat diversity and the inclusion of more ecological niches.

Fig. 1: The retail species-area relationship (SAR).
figure 1

Notes: Extending the Species-Area relationship to the retail environment, this figure demonstrates how retailer diversity, akin to species richness, intensifies with the expansion of retail floor space up to a specific limit. When represented on a standard scale, the increase in retailer variety with area resembles a rectangular hyperbola. On a logarithmic scale, the relationship straightens, conforming to the equation log R = log C + Z log A, where ‘R’ signifies retailer diversity, ‘Z’ is the slope indicative of the diversity growth rate, ‘A’ denotes retail floor area, and ‘C’ is the y-intercept. Z values typically lie between 0.1 and 0.2 for smaller retail spaces, while larger retail environments may have Z values between 0.6 and 1.2, reflecting a more significant shift in the retail mix.

Full size image

In the context of a retail environment, the ‘species’ can be thought of as different types of retail outlets, and the ‘area’ is the retail district or shopping area. The diversity of retail types is expected to increase with the size of the area, analogous to diversity in ecological studies. In conceptualising the impact of touristification on retail ecosystems through the species-area relationship (SAR) framework, it becomes evident that retail dynamics mirror several ecological processes. Retail outlets often emerge in an unplanned manner, akin to the random dispersal of species across various habitats, with their success subject to the unpredictable whims of market forces, much like ecological population fluctuations. Similarly, unforeseen disruptions, whether through policy changes or construction activities, can temporarily alter the retail landscape, echoing the disturbances seen in natural ecosystems.

The retail environment also undergoes adaptive processes where outlets evolve to develop unique characteristics or to fill specific niches reminiscent of speciation in isolated environmental settings. Just as ecological niches are created and transformed by the comings and goings of species, the retail landscape is dynamically reshaped by introducing new retail concepts and closing existing ones, offering fresh opportunities for market entrants. Interactions among retailers, ranging from competitive to cooperative, sculpt the retail ecosystem, paralleling the dynamics among species within biological communities. The diversity of retail spaces—from bustling high streets to expansive shopping malls—contributes to the overall heterogeneity of the retail ecosystem, much like habitat diversity supports species richness. Moreover, a minimum viable area is essential for maintaining retail diversity, reflecting ecological principles where certain habitats cannot sustain biodiversity below a threshold size. The economic vitality of an area, influenced by consumer spending power, drives the diversity and success of retail outlets in a manner comparable to how energy availability underpins biological productivity in nature. The physical and geographical attributes of retail districts, including their layout and accessibility, play a crucial role in determining their attractiveness to both retailers and consumers, much like the geographical features that influence habitat desirability in ecology.

Species-area relationships (SAR) with non-native retail outlets

Incorporating the phenomenon of touristification and its consequential influx of tourists into the context of SAR in retail ecosystems mirrors the dynamics observed with non-native species in ecological studies. Just as human activities have facilitated the spread of non-native species across natural habitats, touristification acts as a catalyst for the introduction of non-native (or ‘alien’ in ecological literature) retail outlets into new geographic areas. The influx of tourists serves as a resource, much like the ecological resources that support the establishment and proliferation of non-native species in biological communities.

The SAR, traditionally applied to understand the distribution of biodiversity, can be adapted to explore how the diversity of retail outlets changes with the increase in geographic area under the influence of tourism. Similar to ecological systems where larger areas tend to support greater species richness, in touristified areas, larger geographic extents can accommodate a more diverse array of retail outlets. This diversity is not just a function of physical space but is also significantly influenced by the economic and cultural ‘resources’ that tourists bring, akin to the way alien species exploit new ecological niches.

However, the introduction of non-native retail outlets—those primarily aimed at serving tourists and often divergent from the local retail ecosystem—can alter the native retail landscape. This alteration mirrors the impact of alien species in ecological habitats, where the introduction of non-native species can disrupt local biodiversity and lead to shifts in community structure. In retail ecosystems, the establishment of tourist-oriented outlets may lead to a homogenisation of retail offerings, displacing local businesses and eroding the diversity of the retail landscape, similar to how invasive species can outcompete native species and reduce ecological diversity.

The SAR framework suggests that as the area influenced by touristification expands, the diversity of retail outlets initially increases. However, the nature of this diversity shifts, with a growing proportion of non-native outlets that cater to the transient tourist population rather than the resident community. This shift has parallels in ecological studies, where the richness of alien species in a given area can sometimes overshadow the richness of native species, especially in regions heavily impacted by human activities.

Furthermore, just as ecological studies investigate the mechanisms underlying the SAR for alien species, understanding the dynamics of retail diversity in touristified areas requires examining the processes that govern the introduction and success of tourist-oriented retail outlets. Factors such as the volume of tourist traffic, the spending power of tourists, and the cultural preferences of the visiting population can all influence which non-native retail outlets establish successfully and proliferate. As a result, touristification will act as a barrier that influences the diversity of the retail ecosystem, as shown in the inhospitable matrix SAR curve shown in Fig. 2.

Fig. 2: Comparative analysis of species-area relationship (SAR) in touristified vs. non-touristified retail ecosystems.
figure 2

Notes: This figure illustrates the expected effects of touristification on the Species-Area Relationship (SAR) in retail ecosystems. The solid line represents the SAR in an unmodified retail ecosystem. In contrast, the red-dashed line depicts the SAR in a touristified area, highlighting increased dynamics. The green dash-dotted line shows the SAR in areas with minimal touristification, indicating different levels of impact on retail diversity and ecosystem structure.

Full size image

Empirical test of touristification’s impact on retail diversity

In this section, we present empirical findings that examine the influence of touristification on retail diversity, specifically focusing on the tenant mix as a proxy. This approach, while methodologically similar to the previous work that used rent as the dependent variable (Cheung and Yiu 2023), differs significantly in its focus and implications. Unlike rent, which primarily reflects economic returns, the tenant mix offers a direct measure of retail diversity, thereby providing novel insights into the socioeconomic dynamics of touristification, and its implications on destination resilience.

Drawing an analogy from ecological studies, we theorise touristification as a process requiring tourist activities to overcome various barriers akin to species in the progress of ecological succession. Utilising the framework proposed by Richardson et al. (2000), tourists are viewed as ‘nutrients’ in an ecosystem, fostering the growth of tourist-oriented retailers as introduced species at various stages of the success pathway. To progress through each stage of ‘succession’, these retailers surpass barriers, eventually leading to their ‘naturalisation’ and potential ‘succession’ within the community.

We explore the multifaceted nature of touristification, incorporating elements of capital reinvestment and retailer relocation (Davidson and Lees 2005). This is approached through the lens of retail tenant mix, offering a unique perspective on how touristification manifests in localised retail environments. The ecological analogy extends to understanding the impact of touristification on species (retailer) diversity in these environments. Our hypothesis posits that as touristification overcomes the barriers on the succession path, the resultant dominance of tourist-oriented establishments leads to the suppression of local shops, thus diminishing the retail diversity enjoyed by existing residents.

Therefore, this study advances the hypothesis that touristification, analogous to an ecological process, negatively affects retailer diversity, as indicated by changes in the tenant mix. This hypothesis is distinct from many previous focuses on retail rents, offering fresh insights into the broader socioeconomic impacts of touristification on local retail ecosystems.

Background of Hong Kong tourism

The reduction in retail diversity during periods of overtourism has shown significant long-term impacts on the resilience of retail sectors in destinations like Hong Kong. As tourist numbers surged, many local retailers, particularly those not catering directly to the tourist market, found it increasingly difficult to compete, leading to a more homogenised retail landscape dominated by tourist-oriented businesses. The Covid-19 pandemic has further highlighted the vulnerabilities of such a model. With travel restrictions in place and a drastic drop in tourist numbers, the retail sector faced unprecedented challenges, with many businesses struggling to adapt to the sudden absence of their primary customer base.

The situation in Hong Kong serves as a compelling case study of these dynamics. Prior to the pandemic, Hong Kong’s retail sector thrived on the influx of tourists, particularly from mainland China, drawn by the city’s status as a shopping paradise. However, the pandemic’s impact, coupled with stringent measures to control its spread, led to a significant decline in tourist arrivals, leaving many retailers facing empty stores and plummeting revenues. This shift has underscored the importance of retail diversity as a buffer against economic shocks. Retailers that once catered predominantly to tourists found it challenging to realign their business models to serve the local population or adapt to the new market realities.

Moreover, changing perceptions among mainland Chinese tourists, who have been pivotal to Hong Kong’s retail sector, have further complicated the recovery of the destination. With a growing preference for other destinations and experiences, alongside adjustments in consumer behaviour influenced by e-commerce and the search for more authentic experiences, the retail landscape in Hong Kong is in a predicament. Experience of the city highlights the need for a more diversified retail sector that can withstand the ebbs and flows of tourist numbers, suggesting that resilience in the face of such shocks depends on a balanced mix of retail offerings that cater to both tourists and local residents alike. This situation presents a critical lesson for destinations worldwide on the risks of over-reliance on tourism-driven retail models and the importance of fostering a diverse retail ecosystem. As Hong Kong navigates its post-pandemic recovery, the emphasis on diversifying its retail offerings could not only help cushion the blow from future disruptions but also create a more sustainable and resilient economic model for the long term.

This empirical study considers the high street retailing habitats of Hong Kong during its climax of over-tourism from 2016 to 2018 (Gössling et al. 2020). In 2018, total tourism expenditure amounted to HK$331.7 billion (US$42.5 billion) from more than 65 million visitors (HKTB 2021). In the same year, Hong Kong was ranked the most visited city globally (SCMP 2019), and it is one of the hottest destinations for shopping tourists. The landscape of high street retail underwent a profound transformation. Taking the excessive features of dispensaries in tourist areas as an example, one may be surprised by how many dispensaries are found on a single street. The widespread presence of pharmacies, distinguished by their brightly flashing neon lights and advertising signs, has become a defining feature of the city’s streetscape. Together with Yau Ma Tei and Mong Kok, Tsim Sha Tsui has formed the most pharmacy-populated district with over 16 drug retailers on every square kilometre of land. This is not because Hong Kongers have such a tremendous demand for medicines but rather because the increasing numbers of mainland Chinese tourists, including parallel traders, in recent years have fuelled the thriving of pharmacies in those tourist areas.

Over the years, the pharmacy business has grown to be more reliant on mainland Chinese tourists. The intertwined relationship is not surprising because more than 75 per cent of inbound tourists to Hong Kong are from Mainland China. Driving Chinese tourists’ pharmacy hunt in the city is their demand for drugs and healthcare products due to China’s less-trusted drug products and rapidly ageing society.

Shopping is always a relevant component of tourism, and in some cases, shopping even becomes the prime travel motivation – shopping tourism. The impacts of tourism on the retail sector can be substantial. During boom periods, tourism is an economic growth engine that boosts sales in the retail markets. For example, according to the World Tourism Organization, the total expenditures of international tourism amounted to US$1.585 trillion (World Bank 2020); in 2018, the proportion of retail spending is estimated to be about 54.5% (Neffke et al. 2018). Shopping tourism is one of the major growth engines in Hong Kong. Fig. 3 shows the year-on-year growth of Chinese visitor arrivals and retail sales of medicines, cosmetics and toilet requisites in Hong Kong. This reflects a strong association between the Mainland Chinese tourism industry and the retail sector of Hong Kong, especially its growth in the pharmacy business. However, most previous studies focus only on the revenues of retailers without investigating their impacts on retail diversity.

Fig. 3: Year-on-year growth of chinese visitor arrivals and retail sales of medicines, cosmetics, and toilet requisites.
figure 3

Notes: The vertical axis shows year-on-year growth rates in percentages, while the horizontal axis represents the month/year from 2011 to 2018. The blue curve reflects the growth rate of visitor arrivals from Mainland China to Hong Kong, and the orange curve represents the growth rate of retail sales volume (RSV) for medicines, cosmetics, and toilet requisites. Data sources: HKTB (2021) and CEIC (2023).

Full size image

Data and methodology

Rental data from 1782 shops in four central shopping districts in Hong Kong from 2016–2018 are collected (Retail in Asia 2019) when overtourism is at its climax in Hong Kong. Table 1 shows the summary statistics. The dataset reports the total rented shop floor area, and shop type during the three tenancy years.

Table 1 Summary statistics of the variables.
Full size table

Shop types are further categorised based on the level 3 categories (LEVEL 3) of shop types developed by Yiu and Xu (2012), which was developed on the basis of the UK government’s classification of retail businesses (Guy 1998) with an adaptation to fit the shopping landscapes in Hong Kong (Fig. 4). Fixed effects on the three tenancy years by incorporating time dummy variables (2016, 2017, 2018) are included to control the market conditions in the year when the tenancy is signed or renewed.Footnote 1

Fig. 4: Number of shops by trade type (in level 2 categorisation).
figure 4

Notes: This chart shows the distribution of shops by trade type, similar to tourist-oriented high streets. Food and beverage, health and beauty (pharmacy), clothing, footwear and accessories, and watches, jewellery and ornaments make up the top four categories, accounting for over 80% of total shops.

Full size image

Beyond characteristics of retail locations and the tenant mix, this study evaluates purchasing power and the extent of tourism through analysis of the median household income from the 2016 census and the hotel occupancy rates near the retail spaces. Nevertheless, identifying these neighbourhood traits necessitates a clearly marked boundary for urban retail zones. Unlike malls that possess physical confines, urban retail zones often blend seamlessly into adjacent commercial spaces without clear separations. Prior research has adopted various subjective methods for boundary delineation. For example, Zhang et al. (2020) examined how the mix of tenants influences retail lease prices in Dutch urban retail districts by setting a boundary based on retail density. They also posited a constant influence of tenant variety on lease prices, overlooking the potential for tenant mix to be shaped by the conditions of the retail area (a form of endogenous bias). This study broadens the application of the SAR model, previously applied within mall contexts, to urban retail environments, aiming to recognise retailer diversity.

This study utilises the k-means clustering technique, a machine learning approach, to establish the perimeters of urban retail sectors, following the methodology described by Cheung and Yiu (2022). This process involves grouping based on spatial coordinates, rental values, and street locations, effectively creating clusters with uniform rental prices within specific streets and areas. Through this technique, we have identified 50 distinct retail clusters (CLUSTER). Once the boundaries of these urban shopping zones are precisely defined, it becomes possible to quantify neighbourhood features like the retail space available for lease (SHOP_AREA – indicative of scale) and the diversity of retail types presented (NO_TYPE - reflecting tenant mix). Additionally, this study gathers data on the mean occupancy rate of hotels (HOTELOCC) and the median income of households (XINC_2016) within each retail cluster, emphasising that both hotel occupancy and household income fluctuate over time and are unique to the retail shops’ local area. To visually represent the spatial distribution and clustering of retail shops, we have included a map (Fig. 5). This map illustrates the k-means clusters on the Hong Kong map, highlighting the geographical spread and density of retail shops across the identified clustersFootnote 2.

Fig. 5: Spatial distribution of retail shops in Hong Kong.
figure 5

Notes: This map illustrates the k-means clusters across Hong Kong, showing the geographical spread and density of retail shops within the identified 50 clusters. Source of the map: Open Street Map.

Full size image

This study examines the fluctuations in hotel occupancy rates within neighbourhoods as an indicator of the touristification phenomenon. It introduces an innovative framework for retail rent analysis aimed at assessing how the influx of tourism influences the replacement of traditional retail establishments with those aimed at tourists, leading to a homogenised retail environment. The allure of a touristified area often draws in more businesses seeking to capitalise on the increased foot traffic, subsequently diminishing the variety of retail offerings. As a result, locales once defined by a rich array of local shops, eateries, and community-centric spaces may find themselves overshadowed by establishments geared towards tourists.

The SAR Model proposes that achieving a balanced tenant mix, or an equilibrium in retail diversity, necessitates a harmonious distribution of large anchor stores, mid-sized food and beverage outlets, and smaller retail units. Employing the SAR Model, this paper details how the diversity of retail types within each cluster is determined by the collective rented space (area), the median income of nearby households (energy), and the mean hotel occupancy rate (touristification) associated with each cluster. The calculation for the rented area of retail space, and the diversity of shop types present in each cluster are derived from our dataset. Meanwhile, data on the median household income levels within the proximity of these retail clusters are sourced from the 2016 Census data (CSD, 2021), and the hotel occupancy rates are collected from the Hong Kong Tourism Board.Footnote 3

Based on the level 3 classification of shop types, we extend the SAR model such that the balanced number of retail types \({m}_{k,t}^{* }\) of a shopping cluster k is related to the product of the total shops’ rented floor area \({S}_{k,t}\) of the cluster, the average household income of potential shoppers \({E}_{k,t}\) of the cluster, and the average hotel occupancy rate \({H}_{k,t}\) of the cluster to the power of \({\theta }_{i}\) within the habitat at time t (Eq. (1)).

$${m}_{k,t}^{* }={\alpha }_{1}\left({S}_{k,t}^{{\theta }_{1}}{E}_{k,t}^{{\theta }_{2}}{H}_{k,t}^{{\theta }_{3}}\right)$$
(1)

Equation (2) represents the log-log transformation of Eq. (1).

$${\mathrm{ln}}({m}_{k,t})={\alpha }_{1}+{\theta }_{1}{\mathrm{ln}}({S}_{k,t}){+{\theta }_{2}{\mathrm{ln}}({E}_{k,t}){+\theta }_{3}{\mathrm{ln}}({H}_{k,t})+\mu }_{k}$$
(2)

To test the touristification effects on retail diversity, we applied a differences-in-difference (DID) model as shown in Eq. (3) with an interaction term between \({\mathrm{ln}}({S}_{k,t})\) and \({\mathrm{ln}}({H}_{k,t})\). The expected sign of \({\theta }_{1}\) and \({\theta }_{4}\) are positive and negative, respectively, according to the power law and the touristification hypothesis that tourists reduce retailer diversity.

$${\mathrm{ln}}({m}_{k,t})={\alpha }_{1}+{\theta }_{1}{\mathrm{ln}}({S}_{k,t}){+{\theta }_{2}{\mathrm{ln}}({E}_{k,t}){+\theta }_{3}{\mathrm{ln}}({H}_{k,t})+{\theta }_{4}{\mathrm{ln}}({S}_{k,t})\times {\mathrm{ln}}({H}_{k,t})+\mu {\prime} }_{k}$$
(3)

Figure 6 summarises the workflow of this study, from the theorisation of touristification as a socio-ecological process to derive the hypothesis that touristification reduces retailer diversity. Data from the retail markets in Hong Kong are collected and cleaned. Shops are then clustered into high-street shopping areas by a machine learning algorithm and categorised into retail types. The extent of retail diversification and touristification in each cluster are identified for conducting the hypothesis tests using a DID model.

Fig. 6: Workflow diagram.
figure 6

Notes: This diagram summarises the workflow of the study, from theorising touristification as a socio-ecological process to deriving the hypothesis that touristification reduces retailer diversity. Data from the retail markets in Hong Kong were collected and cleaned, and shops were clustered using machine learning algorithms. Retail types and the extent of diversification and touristification were identified, leading to hypothesis testing using DID models.

Full size image

Results, discussions and robustness check

Baseline Model 1 in Table 2 shows the results of the SAR power law on retail diversity (tenant mix), with panel data of 60 clusters and 3 years. With the logarithm of the number of shop types in a cluster as the dependent variable, the total rented floor area of shops, and average household income level in each cluster are the explanatory variables, with a fixed time effect. The results support the SAR power law, such that the cluster’s habitat size (total rented floor area of shops) is positively associated with retail diversity. The effect of touristification is negative on retail diversity, but it is not statistically significant in the baseline model. The effect of household income level on tenant mix (energy level) is positive yet statistically insignificant.

Table 2 Results of SAR models.
Full size table

Models 2 and 3 in Table 2 test the moderating effect of touristification on the rented area impact on tenant mix, with and without a fixed time effect. Besides confirming the power law, the results also support the touristification hypothesis that clusters with a higher concentration of tourists negatively moderate the effect of shopping areas on retail diversity in Models 2 and 3. The results agree with the succession hypothesis that shopping tourists can displace resident-oriented retailers and replace them with tourist-oriented retailers in neighbourhoods; it can exacerbate retailer diversity.Footnote 4

The loss in retail diversity implies that residents nearby have fewer choices for buying local goods and services. Tourist-oriented commercial activities will also worsen their living environment. The increase in retail rents of the neighbourhood can be reflected in the selling prices of the retail goods (Cheung and Yiu 2023).

A low diversity of retailers can also be detrimental to destination resilience, as reflected in Hong Kong’s retail sector during and after the pandemic. This is the implication that a higher hotel occupancy rate leads to a lower diversity of retailers in a neighbourhood of the same shop area, i.e., the negative sign of \({\theta }_{4}\) for the interactive variable \({\mathrm{ln}}({S}_{k,t})\times {\mathrm{ln}}({H}_{k,t})\). Admittedly, the benefits from touristification for shop owners and retailers are only achievable when the number of tourists is high. After the pandemic outbreak in 2020, the number of tourists to Hong Kong plummeted by more than 95%. The annual change of retail sales fell by about 50% in early 2020. Retail rent fell by 9.2% in 2020. The retail space take-up was also sharply reduced by 108,400 sq. m., resulting in an unprecedented vacancy rate of 11.4% since data available in 1995 (RVD 2022). A lack of retail diversity has shown a sign of reduced destination resilience in this case study. In order to achieve destination resilience, a policy implication of this study is a further study on tactical measures to lead shopping tourists to tourism-compatible neighbourhoods by means of, say, a tourist tax on retailers similar to the bed tax on short-term accommodations (Cheung and Yiu 2022).

To further ensure the validity of our findings and confirm that they are not driven by the effects of our initial clustering method, we employed Density-Based Spatial Clustering of Applications with Noise (DBSCAN) to reanalyse the retail clusters.

In essence, the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) technique (Pedregosa et al. 2011) is another clustering method that excels in identifying groups based on the density of data points, i.e., retailers in our case. DBSCAN operates by defining a radius (ε) around each data point and counting the number of additional points within this radius. If a point has a predefined minimum number of neighbours (minPts) within this area, it is classified as a core point, around which a cluster is built. Points on the edge of these clusters, which have fewer neighbours but lie within the radius of a core point, are identified as border points, while all other points are considered noise. This method of clustering is particularly suited to our analysis of retail spaces in densely populated urban environments like Hong Kong, where retail clusters do not conform to regular, spherical shapes. DBSCAN’s flexibility in capturing clusters of arbitrary shapes and sizes, along with its robustness in handling outliers, provides a precise tool for delineating complex patterns of retail distribution influenced by touristification, ensuring our findings reflect the true nuances of urban retail clustering.

Thus, this alternative clustering approach served as a sensitivity test to validate our findings and confirm that they were not artefacts of the chosen clustering technique. Utilising parameters eps=0.015 and min_samples=4, this approach successfully delineated 85 distinct retail clusters, providing more granular insight into the impacts of touristification on retail diversity. In the dense retail environment of Hong Kong, DBSCAN offers some advantages. Its primary strength lies in its ability to identify clusters of arbitrary shapes and sizes, crucial for accurately capturing the varied and irregular spatial distributions of urban retail clusters. Unlike k-means, which assume clusters of similar sizes and globular shapes, DBSCAN adapts to actual density variations within the data, allowing for the detection of both high-density core areas and more sparse peripheral ones. Figure 7 shows the retail clusters using DBSCAN.

Fig. 7: Geographic distribution of retail clusters using DBSCAN.
figure 7

Notes: This figure shows the spatial distribution of retail clusters in Hong Kong identified using DBSCAN (eps = 0.015, min_samples = 4). Each point represents a retail location, and colour coding corresponds to different clusters, with a gradient from purple to yellow indicating cluster IDs from 0 to 85. The clustering reveals the density and geographical spread of retail areas, which assists in analysing the impact of touristification on retail diversity.

Full size image

The robustness checks, as demonstrated in the last column of Table 2, clearly show that DBSCAN not only clarified the cluster formations but also ensured their alignment with known commercial distributions across Hong Kong. The integrity of all variable signs was maintained, further confirming the consistency of our results. This methodological refinement has led to more robust findings, underscored by stronger significance levels, particularly in illustrating how touristification transforms local retail landscapes — akin to ecological succession.

As a result, the exploration of touristification through the retailer-area (or species-area) relationship offers a fresh and intriguing perspective on touristification. By likening the influx of tourists and the resultant shifts in retail ecosystems to ecological succession and species-area relationships, you have laid a foundation for a more detailed discussion about the socioeconomic impacts of tourism on urban neighbourhoods. The analogy of ecological succession in the context of touristification posits that just as non-native species can disrupt local ecosystems, an influx of tourists can also disrupt local retail environments. This disruption often manifests as a homogenisation of retail offerings, where specialised, locally-oriented shops are replaced by those catering to tourist preferences, akin to how non-native species can overshadow native flora and fauna, leading to a decrease in biodiversity.

The theory suggests that touristification, much like the ecological succession process, follows a predictable pattern where the ‘original inhabitants’ (in this case, local retailers and services) are gradually displaced by ‘newcomers’ (tourist-oriented businesses), altering the fabric of the neighbourhood. This displacement is not just a matter of changing shop fronts but reflects deeper shifts in cultural, economic, and social dynamics within the area. The species-area relationship further enriches this analogy by proposing that as the ‘habitat’ (the touristified area) expands, so does the diversity of ‘species’ (retail types), but only up to a point. Beyond this threshold, the unique characteristics of local retail ecosystems begin to diminish, leading to a more homogenised, less diverse retail landscape dominated by businesses catering to tourists.

The finding also reminds us that the concept of tourism destination resilience originated in the context of ecological studies. The concept refers to the ability of an ecological system to return to a stable equilibrium after experiencing a disruption (Holling 1973). Since its conceptualisation, resilience has been applied with its meaning in many other disciplines. For example, resilience in the context of social-ecological systems refers to the ability of such systems to cope with disturbances (Hudson 2010). In tourism studies, resilience can be defined as the ability of a tourism ecosystem to resist, recover, and adapt from exogenous shocks, such as hazards, by preserving and recuperating its functions (Alvarez et al. 2022).

Conclusions

Touristification reshapes the retail landscape, creating both beneficiaries and sufferers. As touristification accelerates, retail property owners may see gains, but long-term residents face losses, exacerbated not merely by rental hikes but by the displacement of local shops that cater to their preferences. This paper elucidates how the transformation towards tourist-oriented retailing can inflict greater welfare losses on incumbent residents, particularly when locally cherished stores are replaced by generic tourist-targeted outlets. This change erodes the unique character of neighbourhoods, a loss deeply felt by established residents. The distinction between stores generating high consumer surplus and those easily substitutable underscores the unique value of local retailers over ubiquitous tourist-focused chains.

The anticipation that an influx of tourists might demand convenience-focused retail options does not presuppose that such changes universally degrade community welfare. Rather, the broader implication of our analysis is to furnish empirical indicators for when changes in neighbourhood dynamics—be they due to touristification or other urban developments—might significantly impact local welfare beyond mere alterations in living costs. This extends to various urban transformations, not limited to the advent of new infrastructure or significant business closures, influenced by the balance of unique versus generic retail establishments.

An empirical investigation, akin to analyses using Hong Kong data to examine touristification’s impact, could similarly assess touristification’s effects on retail diversity. Initial findings suggest that areas undergoing touristification likely experience a heightened turnover of tourist-oriented retail establishments, alongside a shift towards services primarily catering to tourists, such as restaurants, albeit without a stark transformation in the retail character towards service provision or chain dominance. This view of retail evolution in touristified areas highlights the dynamics between incoming tourist demand and local retail supply, supplementing the narrative of retail gentrification during the overtourism period.

Theoretical implications

In synthesising ecological concepts with urban studies, the SAR retail theory underscores the importance of considering the carrying capacity of urban areas for tourism, akin to the ecological carrying capacity of habitats. It raises critical questions about sustainable tourism and the need for policies that balance tourist influx with the preservation of local retail diversity and cultural identity. This ecological framing enriches the discourse on touristification and provides a tangible framework for empirical investigation and policy formulation. It invites further research into the mechanisms of touristification to develop strategies that promote resilience and sustainability in urban retail ecosystems. The work opens up exciting avenues for interdisciplinary collaboration, integrating insights from ecology, urban planning, and tourism studies to address the complex challenges posed by touristification. Also, it calls for a re-evaluation of how urban spaces are managed and conserved in the face of growing tourism, urging for a balance that respects both the needs of visitors and the rights and traditions of local communities.

Moreover, a growing transdisciplinary approach beyond sociology and geography, including economics, history, political science, and public health, to name a few relevant disciplines, emphasises the globalised nature of touristification. The theorisation of touristification from an ecological perspective is a heuristic device that helps to integrate past understandings of touristification with current tourism theories to improve empirical research on the causes and consequences of urban change and inform the regulatory and legal changes underpinning displacement and tourism development.

Practical implications

With an increased global prevalence of touristification and growing public concern surrounding its implications for urbanism development, it is warranted to investigate touristification in tandem with urban neighbourhoods. Conceptualising touristification in the retail environment is one of these attempts. Furthermore, research on touristification has focused on spatial differentiation, the class transformation of urban neighbourhoods, and the displacement of former residents by wealthier residents. Conceptualising touristification as an ecological succession allows us to examine the process in both micro and macro forms and to understand further the types of touristification occurring in cities worldwide. The conceptualisation also allows more parsimonious explanations of the causes and consequences of touristification.

Our study serves as a transdisciplinary theorisation of touristification using a socio-ecological perspective and seeks to synthesise a broader perspective, interconnections, and epistemology on tourism, retailing and urban management literature. The conceptualisation also helps destination management organisations guide their efforts to adapt and mitigate against the impacts of touristification by identifying chronic sources of vulnerability in the retail environment. This study also opens up a new research agenda on touristification and suggests that further studies should be conducted on the effects of tourism on gentrifying retail neighbourhoods.

Limitations and future studies

While this case study provides a novel theorisation of touristification from the perspective of ecological theory, some limitations remain. Due to the limited data on individual tourist purchasing powers and numbers of tourists at a neighbourhood level, this paper considers the changes in hotel occupancy rates of neighbourhoods as a proxy of the touristification process. Although we have used both hotel occupancy rates and household incomes to correspond to the tourist activities and their relative purchasing power in a neighbourhood, it is by no means a perfect proxy. As such, future research should verify the tourist arrivals and their spending when data is available. In addition, this study investigated changes in local retail shops in Hong Kong from 2016 to 2018. Three years may be considered relatively short even though this study period can avoid any impact of the pandemic. Hence, future studies should accumulate longitudinal data that can be used to accurately track touristification effects on retail while figuring out the key drivers of the process. A comparative analysis of steady-state versus dynamic models of resilience can also be considered in further studies (Hall, Prayag and Amore 2018). Furthermore, it is important to recognise that besides touristification, there are other factors contributing to the changes in retail diversity, such as gentrification (González 2017). Using a quantitative approach to analyse a politicised debate over the touristification effect may also be a limitation.