The concept of gentrification, while born in the urban context of 1960s London [3], was adapted to rural settings by scholars who observed similar processes of class-based transformation in the countryside. Rural gentrification describes the influx of affluent, typically urban-origin, middle-class households into rural areas, leading to rising house prices and the cultural and economic displacement of established, often working-class, residents [4]. Unlike urban gentrification, which frequently involves the redevelopment of disinvested building stock, rural gentrification typically centres on the appropriation of existing housing and the cultural transformation of village life.

A primary driver of this process has been the demand for second homes. For decades, the purchase of cottages and farmhouses by wealthy urbanites for weekend and holiday use has been a contentious issue in desirable rural locations across the UK, from Cornwall to the Lake District. Research has consistently shown that high concentrations of second homes reduce the available housing stock for permanent residents, drive up prices beyond the reach of local incomes, and contribute to the seasonal decline of village services and community cohesion [1], [5]. These "part-time" residents can lead to what has been termed the "death of the English village," where local shops, pubs, and schools close due to a lack of year-round patronage, leaving behind sterile commuter and retirement dormitories.

This process creates profound forms of housing exclusion. As property prices decouple from local wage levels, younger generations are forced to move away to find affordable housing, severing family ties and community bonds. Key workers in lower-paid but essential sectors, such as agriculture, social care, and hospitality, also find themselves priced out of the communities they serve, creating critical labour shortages [6].

The rise of the platform economy, exemplified by Airbnb, introduces a new and powerful dynamic into this context. Airbnb can be understood as a technological accelerant for the processes described above. It dramatically lowers the transaction costs and increases the visibility for property owners wishing to enter the short-term tourist rental market. What might once have been a cottage used as a family's private second home can now be operated as a commercial, high-turnover holiday let, accessible to a global market of tourists. This "platformatization" of rural housing [2] has several distinct effects. First, it intensifies the commercial logic of property ownership. A property is no longer just a home or a second home, but a financial asset to be optimised for maximum yield. The potential revenue from high-season tourist lets often far exceeds that available from a long-term tenancy, creating a powerful financial incentive for landlords to convert properties from the residential to the tourist market [7]. This directly reduces the supply of long-term rental properties, a critical tenure for those unable to afford to buy.

Second, Airbnb extends the geography of tourist accommodation. Whereas traditional holiday accommodation was concentrated in licensed B&Bs, hotels, or designated holiday parks, Airbnb allows any residential property to become a de facto hotel. This "touristification" can spread throughout a village, fundamentally altering its character and sense of place. The constant churn of temporary visitors can erode neighbourly bonds and community trust, while local services may reorient towards tourist needs rather than those of residents [8].

Empirical research on Airbnb's rural impact is nascent but growing. A study in the Scottish Highlands found that the growth of STRs was a primary driver of a severe housing crisis in popular tourist areas like the Isle of Skye [9]. Similarly, research in rural US tourist towns has linked Airbnb's expansion to rising rents and increased housing precarity for local service workers [10]. However, few studies have applied rigorous econometric methods to model this relationship, and even fewer have considered the spatial spillover effects within a rural region.

Sussex encapsulates the tensions between conservation, tourism, and local housing needs. Its landscape is heavily protected by planning designations (e.g., the South Downs National Park), which restricts new housing development and inflates the value of existing stock. Simultaneously, its tourism economy is a vital part of its economic fabric. The question is one of balance.

This study fills a critical research gap by providing a systematic, quantitative analysis of Airbnb's role in the Sussex rural housing market. By moving beyond anecdotal evidence and case studies, it aims to quantify the impact of STRs on house prices and rental availability. Furthermore, by employing a spatial model, it will illuminate the interconnected nature of the crisis, demonstrating how market pressures flow across parish boundaries. This evidence-based approach is essential for informing local and national policy responses that can navigate the complexities of the platform economy while prioritising the housing needs of rural communities.

The dataset was assembled from a variety of public sources, aggregated at the parish level:

I. Airbnb data: Sourced from Inside Airbnb, which provides publicly available, scraped data on Airbnb listings.

Data collected included the number of listings, the proportion of 'entire home' listings, and host data (e.g., multi-listing hosts indicating commercial operation). The primary independent variable, AirbnbDensity, is measured as the number of active 'entire home' listings per 100 dwellings in each parish. This metric is chosen as 'whole home' listings represent the most direct removal of a housing unit from the residential market. II. Housing market data: Median house sale prices at the parish level were obtained from HM Land Registry.

Data on median monthly private rental costs and the number of properties listed for long-term rent were sourced from major property portals (e.g., Rightmove, Zoopla) and cross-referenced with Valuation Office Agency (VOA) data.

III. Socio-economic and geographic controls: A range of control variables was included to account for other factors influencing rural housing markets. Parish-level population estimates, median household income, and employment data (e.g., proportion employed in agriculture) were sourced from the Office for National Statistics (ONS). Geographic variables, such as the parish's distance to the coastline and whether it falls within a National Park or AONB, were calculated using GIS data. Proximity to primary transport links (e.g., mainline train stations) was also included. The key variables for the model are described in Table 1.

Median sale price of residential properties HM Land MedianPrice (£). Registry Median monthly rent for a two-bedroom VOA / property MedianRent property (£). portals

Number of properties available for long-LTR_Stock Property portals term rent per 1,000 dwellings.

Independent variable Number of 'entire home' Airbnb listings per AirbnbDensity Inside airbnb 100 dwellings.

Control variables MedianIncome Median annual household income (£). ONS Dummy variable (1 if parish is in a National InAONB DEFRA Park/AONB, 0 otherwise).

Distance in kilometers from the parish DistToCoast GIS calculation centroid to the nearest coastline.

PopDensity Population per square kilometer. ONS Proportion of dwellings registered as second SecondHomeProp Local authorities homes (from Council Tax data).

Source: Author's compilation from Land Registry, VOA, Inside Airbnb, ONS, Local Authorities.

N=152 parishes.

Housing markets are inherently spatial; prices influence the price of a house in one parish in neighbouring parishes. To account for this spatial dependence, this study uses a spatial panel regression model. First, we test for the presence of spatial autocorrelation in our key variables using Moran's [11] I statistic. A significant result justifies the move away from non-spatial models like standard OLS.

The study employs the Spatial Durbin Model (SDM), which is a general model that includes both a spatially lagged dependent variable (endogenous interaction) and spatially lagged independent variables (exogenous interaction). The SDM is chosen because it provides unbiased estimates even if the true data-generating process is a Spatial Autoregressive (SAR) or Spatial Error Model (SEM).

The SDM with parish and time fixed effects is specified as:

Y_it = ρW Y_it + X_itβ + W X_itθ + μ_i + λ_t + ε_it, Where:

Y_it is the dependent variable (e.g., MedianPrice) for parish i in year t.

X_it is the matrix of independent variables.

W is the spatial weights matrix. A Queen contiguity matrix is used, where elements w_ij = 1 if parishes i and j share a border, and 0 otherwise. This matrix is row-standardised.

ρ (rho) is the SAR coefficient, capturing the effect of neighbouring parishes' dependent variable values. β represents the direct effect of the independent variables within parish i.

θ represents the indirect (spillover) effect from the independent variables in neighbouring parishes (W X_it).

μ_i and λ_t are parish and time fixed effects, respectively, controlling for unobserved heterogeneity across parishes and time.

ε_it is the idiosyncratic error term.

The model is estimated for each of the three dependent variables: MedianPrice, MedianRent, and LTR_Stock. This tripartite analysis provides a comprehensive picture of Airbnb's impact on the housing market, covering owner-occupied, private rental, and housing supply dimensions.

Table 2 presents the mean values for key variables across the 152 rural parishes in the study, comparing the start and end points of the study period.

The descriptive statistics paint a stark picture of the pressures on the rural Sussex housing market. Over the decade, median house prices and rents have surged by nearly 50%, far outpacing the growth in median income. Most strikingly, the stock of long-term rental properties has plummeted by over 40%. During this same period, the density of 'entire home' Airbnb listings has increased more than fivefold, a growth rate that dwarfs all other variables. This strong correlational evidence provides a prima facie case for the study's central hypothesis.

Moran's I [11] statistic was calculated for the key variables in 1950 to test for spatial clustering.

The results indicate strong, statistically significant positive spatial autocorrelation for both house prices and Airbnb density. This confirms that high-price parishes tend to be clustered near other high-price parishes, and the same is true for parishes with a high concentration of Airbnb listings. This spatial clustering underscores the necessity of a spatial modelling approach.

The SDM with fixed effects was estimated. The results are presented as direct, indirect (spillover), and total effects. Table 4 shows the results for MedianPrice.

Interpretation: The results strongly support the hypothesis. The direct effect of Airbnb density is highly significant, indicating that a one-unit increase in Airbnb density (one extra STR per 100 dwellings) is associated with an increase in median house prices of approximately £6,251 within that parish. The significant indirect effect shows that this also increases house prices in neighbouring parishes by £1,875. The total effect is substantial. As expected, being located in a protected landscape (AONB/National Park) carries a large price premium. The proportion of second homes also positively correlates with prices, suggesting that Airbnb is an additional, distinct pressure on top of this long-standing issue. Table 5 shows the results for the availability of long-term rental stock (LTR_Stock).

Interpretation: This model provides direct evidence of the housing stock conversion. The Airbnb density coefficient is negative and highly significant. A one-unit increase in Airbnb density is associated with a direct reduction of 0.85 long-term rental listings per 1,000 dwellings. This demonstrates a clear trade-off: as properties are shifted to the short-term market, the long-term rental pool for residents shrinks. The indirect effect is also significant, suggesting that a high concentration of STRs in one parish reduces LTR availability in neighbouring areas, possibly as landlords in the vicinity are inspired to follow suit.

The evidence strongly suggests that a laissez-faire approach to STRs is untenable if the goal is to maintain viable, balanced rural communities. A suite of policy interventions should be considered:

I. Mandatory registration and licensing: A compulsory register of all STR properties would provide local authorities with the data needed to monitor and manage the market. This could be linked to a licensing scheme, with fees used to fund enforcement and affordable housing initiatives.

II. Use of planning controls: The Central government could make it easier for local authorities to require a change-of-use planning application to convert a residential dwelling into a full-time STR. This would give councils the power to limit the number and concentration of STRs in areas under severe housing pressure. III. Council tax reform: Implementing significant council tax premiums on second homes and commercial STRs could disincentivise the practice and generate revenue. Some councils in Wales have already implemented premiums of up to 300%.

IV. Community-led housing: Supporting community land trusts and other forms of community-led housing can create permanently affordable homes that are protected from the open market and cannot be converted to STRs.