Urban Competitive Vulnerability in Tourist Cities: An Integrated Framework and Empirical Evidence from Spain

Abstract

Urban competitiveness and vulnerability have traditionally been studied as analytically distinct dimensions, grounded in the assumption that competitive performance necessarily strengthens urban structures. However, empirical evidence from tourist cities reveals a paradox, as high levels of tourism competitiveness may coexist with cumulative processes of structural fragilisation. This article introduces urban competitive vulnerability—an urban system’s propensity to competitive erosion driven by internal fragility accumulation, even during high performance. Using panel data from six major Spanish tourist cities (Barcelona, Madrid, Valencia, Palma, Seville, and Málaga) over 2014–2024, we develop an integrated framework with four dimensions: tourism competitiveness, sectoral specialisation, territorial pressure, and governance capacity. We construct the Urban Competitive Vulnerability Index (CVI) and test four hypotheses using panel-data models with fixed effects and interaction terms. Results confirm significant positive relationships between tourism competitiveness and structural vulnerability (β = 0.540, p < 0.001). Sectoral specialisation increases vulnerability both directly (β = 0.504, p < 0.001) and indirectly through competitiveness (65.8% mediated effect). Tourist housing intensity significantly increases housing prices (β = 0.288, p < 0.001) and evictions (β = 0.125, p < 0.05). Cities with high prior vulnerability experienced more severe COVID-19 impacts (β = −3.688, p < 0.05) and slower recovery. While limited to Spanish cities, this study provides the first urban-specific framework for competitive vulnerability with direct implications for urban tourism planning and governance.

1. Introduction

In recent decades, cities have acquired a central role in global tourism dynamics, consolidating themselves as strategic nodes of attraction, cultural consumption, and economic accumulation. Urban tourism has evolved from a complementary phenomenon to become a structural component of contemporary urban systems, directly influencing their economic, social, territorial, and institutional configurations [1,2,3]. This process has intensified competition amongst cities to attract tourist flows, investment, talent, and international visibility, positioning urban competitiveness at the centre of academic and urban planning agendas.

This study focuses on six major Spanish tourist cities: Barcelona, Madrid, Valencia, Palma de Mallorca, Seville, and Málaga. The selection responds to specific criteria, including sustained high tourism intensity over the period 2014–2024, substantial variation in tourism models such as coastal versus inland and cultural versus beach-oriented configurations, diverse governance approaches, the availability of consistent longitudinal data, documented experiences of tourism-related urban conflicts, and representation of different urban scales ranging from 858,000 to 7.1 million inhabitants. Spain offers a particularly relevant context as one of the world’s leading tourist destinations (83 million international arrivals pre-COVID, with post-pandemic recovery estimated at 97 million by 2025), with cities experiencing intense tourism pressure and well-documented processes of touristification, gentrification, and residential displacement. The six selected cities represent approximately 40% of Spanish urban tourism activity (42% of hotel overnight stays) and encompass diverse tourism typologies, urban governance models, and regulatory responses to tourism challenges.

Traditionally, the competitiveness of tourist cities has been assessed through indicators such as growth, visitor volume, tourism revenue, accommodation capacity, and urban brand positioning [4,5,6]. These approaches, inherited from territorial competitiveness paradigms, have tended to associate urban success with economic dynamism and attractiveness, implicitly assuming that tourism growth is a desirable and sustainable outcome in itself. However, mounting empirical evidence suggests that the competitive performance of tourist cities can coexist with processes of structural deterioration that compromise their medium- and long-term urban sustainability.

In numerous urban contexts, tourism intensification has generated growing pressure on residential markets, urban infrastructure, and public services, as well as socio-territorial conflicts related to gentrification, touristification, and loss of urban accessibility for resident populations [7,8,9]. Particularly relevant is the expansion of short-term tourist accommodation facilitated by digital platforms, which generates direct competition for available residential housing stock [10]. This phenomenon transcends housing cost inflation and triggers processes of population displacement, social fabric fragmentation, and the erosion of community capital [11]. Territorial pressure constitutes a specific transmission mechanism of competitive vulnerability, converting tourism success into a factor in the expulsion of the resident population, the loss of functional diversity in central neighbourhoods, and the weakening of social networks, which are critical resources for urban resilience.

Simultaneously, the functional specialisation of urban economies in tourism activities has increased sectoral dependence, reducing productive diversification and heightening cities’ exposure to external disturbances, such as economic, health, or technological crises [12,13]. As studies on urban economic diversification have demonstrated [14,15], specialisation in a single sector reduces the productive system’s adaptive flexibility and concentrates labour and fiscal risks. In the case of tourism, this effect is amplified by the highly volatile nature of tourism demand and its sensitivity to external disturbances. Medium-sized cities with high tourism specialisation may exhibit greater structural vulnerability than metropolises with higher tourism activity but greater economic diversification.

The COVID-19 pandemic acted as a catalyst for these tensions, revealing the structural fragilities of urban models intensely oriented towards tourism [16,17]. Cities that had previously displayed high levels of tourism competitiveness experienced an abrupt collapse of economic activity, employment, and fiscal revenue, demonstrating that prior competitive success had not been accompanied by sufficient adaptive capacity. This episode revealed a suggestive paradox, whereby the cities that appeared to be the most competitive were, in many cases, also the most vulnerable.

Nevertheless, comparative evidence suggests that this paradox does not operate uniformly or deterministically. Institutional capacities for urban management emerge as critical mediating factors that can significantly modulate the intensive effects of tourism on urban vulnerability [12,18]. The effectiveness of regulatory instruments for tourist accommodation, land-use zoning, residential protection mechanisms, and intersectoral coordination amongst governmental areas constitutes institutional capacities that can attenuate tourism pressures on the urban system. Cities with similar endowments of tourism resources and comparable activity levels may experience divergent vulnerability trajectories depending on their governance capacities, suggesting that competitive vulnerability is not an inevitable outcome of tourism growth, but is critically conditioned by the management and regulatory capacities of the urban system.

Despite the relevance of this paradox, the academic literature has tended to treat urban competitiveness and vulnerability as analytically distinct dimensions. Studies on urban competitiveness have focused on identifying success factors, comparative advantages, and positioning strategies, generally from normative or prescriptive approaches [19,20,21]. Conversely, urban vulnerability analyses have examined cities’ exposure to economic, environmental, or social risks, frequently from resilience, risk management, or climate change adaptation perspectives [22,23,24]. However, the intersection between the two fields has been scarcely explored systematically. This conceptual disconnection limits understanding of contemporary urban trajectories, especially in tourist cities, where competitive performance can accelerate structural fragilities.

To address this empirically verifiable paradox, this article proposes that urban competitive vulnerability emerges from the interaction of at least four structural factors: the intensity of tourism performance, which can act as an amplifier of urban pressures; the degree of sectoral specialisation, which determines economic dependence on tourism; pressure on the territorial and residential system, which erodes social cohesion and urban absorption capacity; and institutional management capacity, which can modulate the effects of the preceding factors. This approach draws on recent advances in tourist destination studies, where competitiveness and vulnerability are increasingly integrated into unified analytical frameworks [25]. However, it proposes a specific adaptation to the urban context that attends to the functional and social complexity of cities. Urban competitive vulnerability is understood as an urban system’s propensity to experience erosion of its competitive position due to the accumulation of internal structural fragility, even in contexts of high economic and tourism performance.

To operationalise this approach, the study proposes an integrated analytical framework that combines urban tourism competitiveness indicators with structural vulnerability metrics, articulated through a bidimensional matrix that allows cities to be classified according to their risk and performance configurations. This framework is empirically applied through a longitudinal panel-data analysis for the period 2014–2024, capturing both structural heterogeneities across cities and their temporal evolution in response to relevant shocks, such as the global financial crisis, the expansion of digital platforms, and the COVID-19 pandemic. The analytical strategy combines the construction of composite indices of competitiveness and vulnerability with econometric panel data models that incorporate temporal and city-fixed effects, enabling the isolation of both structural heterogeneities across cities and the common effects of temporal crises. Additionally, interaction terms are introduced to capture moderating impacts, particularly the role of governance, and amplifying effects, especially the relationship between prior vulnerability and pandemic crisis impact.

The article contributes to the urban science literature in three principal dimensions. First, it develops a systematic theoretical integration of urban competitiveness and vulnerability, specifically applied to tourism, transcending dichotomous approaches that treat both dimensions as separate phenomena. Second, it proposes and empirically validates the first Urban Competitive Vulnerability Index through econometric analysis of longitudinal panel data, providing a replicable methodological instrument for diverse urban contexts. Finally, it provides robust evidence on the specific mechanisms through which tourism generates structural fragilities in cities, with direct implications for urban planning policies and tourism management.

2. Theoretical Framework and Hypotheses Development

The concept of urban competitiveness has its roots in regional and urban economics, particularly in Porter’s seminal works on territorial competitive advantage [19,26]. From this foundational perspective, cities are conceived as economic actors competing amongst themselves in globalised markets for the attraction of productive factors, investment, and high-value-added activities. Urban competitiveness is defined as a function of resource endowment, institutional efficiency, innovation capacity, and the quality of the business environment [27,28]. Camagni [20] significantly expanded this approach by incorporating intangible elements such as social capital, institutional quality, territorial cohesion, and collective learning capacity as determinants of sustainable urban competitiveness. This relational perspective recognises that competitiveness is not derived solely from static comparative advantages, but from dynamic processes of territorial capacity building that are activated through interaction amongst public, private, and social actors [29,30].

Concurrently, in the tourism field, destination competitiveness has been the subject of extensive academic production, crystallised in integrative models that combine natural and cultural resources, management factors, demand conditions, and institutional context [4,31,32]. The Crouch & Ritchie [31] model, in particular, established a comprehensive conceptual framework that identifies core resources and supporting factors that configure destination attractiveness and competitiveness, additionally incorporating qualifying determinants and destination management as mediating components.

The progressive adaptation of these models to the urban context has recognised that cities compete not only as general productive spaces, but also as complex tourist destinations, where the visitor experience is inextricably intertwined with residents’ daily life and with multiple overlapping urban functions [1,33,34]. This functional complexity distinguishes urban tourism from other tourism modalities, as it develops in densely inhabited, multifunctional, and socially heterogeneous territories [35]. The urban tourism literature has identified that the competitiveness of tourist cities depends on a multidimensional combination of factors: physical elements such as historical-cultural heritage; economic factors such as supply diversity; social dimensions including hospitality and security; and institutional capacities, especially multilevel governance and public–private coordination [36,37,38].

However, a growing body of empirical evidence has documented that intensive pursuit of tourism competitiveness can generate significant negative externalities. Peeters et al. [39] and Russo [38] proposed the concept of the “vicious cycle of tourism”, arguing that tourism saturation leads to deterioration of the resources that initially sustained the destination’s attractiveness. Subsequent studies have confirmed this dynamic in numerous European cities, identifying processes of infrastructural saturation, heritage degradation, and socio-territorial conflicts associated with overtourism [40,41]. The conceptualisation of urban competitiveness has undergone critical evolution, transitioning from approaches focused exclusively on growth towards more comprehensive perspectives that incorporate social and environmental sustainability [42,43,44]. Authors such as Harvey [45] have emphasised that urban competitiveness must integrate criteria of distributive equity, social cohesion, and capacity for urban system reproduction. Nevertheless, even these revised approaches tend to implicitly assume that competitiveness and sustainability can be aligned through adequate governance strategies, without sufficiently problematising that competitive success itself can generate systemic fragilities that accumulate latently [46,47].

2.1. Urban Vulnerability: From Exposure to Shocks Towards the Structural Fragility Approach

The concept of vulnerability has been extensively developed in disciplines such as risk geography, urban sociology, and environmental studies [22,23,48]. In its most general sense, vulnerability is defined as a system’s propensity or predisposition to suffer harm when threats materialise, understood as a function of three interrelated components: exposure, sensitivity, and adaptive capacity [49,50]. In the urban context, vulnerability has been analysed from multiple dimensions: economic, social, environmental, and institutional. Classical studies have documented how spatial concentration of economic activities, socio-spatial polarisation, and functional dependence increase cities’ susceptibility to systemic crises [46,51,52].

The predominant approach has been exposure analysis to specific threats [53,54]. More recently, the urban resilience approach has gained prominence, shifting emphasis towards the capacity to absorb impacts and transform creatively [24,55,56]. However, various authors have identified conceptual limitations. Vale [56] argues that emphasis on recovery capacity can render invisible the structural conditions that generate vulnerability. MacKinnon & Derickson [57] critique the implicit depoliticisation in many resilience discourses. From this critical perspective, vulnerability should not be conceived exclusively as a response to external shocks, but rather as an endogenous condition derived from the configuration of the urban system [48,58,59].

Hypothesis H1 (Vulnerability Amplification under External Shocks). 

Cities with higher levels of pre-crisis competitive vulnerability experience more severe impacts from external shocks and exhibit slower recovery trajectories in the immediate aftermath of the shock.

2.2. Urban Tourism, Sectoral Specialisation and Structural Fragility Mechanisms

In tourist cities, urban vulnerability manifests in particular forms. Economic dependence on tourism as the dominant activity generates functional specialisation that reduces productive diversification and increases urban system exposure [12,60]. Productive diversification theory holds that diversified urban economies exhibit greater adaptive capacity when confronting sectoral shocks [15,61]. Conversely, sectoral specialisation increases structural vulnerability by concentrating employment, investment, and the fiscal base in an activity subject to market cycles and exogenous factors [14,62].

In the specific case of urban tourism, this vulnerability is aggravated by multiple structural factors. Tourism is an activity susceptible to diverse external disturbances [63]. Furthermore, the intensive nature of tourism consumption of urban space generates specific pressures on scarce territorial resources [39,64]. The expansion of tourist accommodation through digital platforms has exponentially intensified these processes. Studies on Airbnb and similar platforms have documented their impact on reductions in residential supply, increases in rental prices, and alterations in the socio-demographic structure of central neighbourhoods [10,65,66]. This phenomenon of “tourism gentrification” [67] or “touristification” [7] erodes social capital, reduces functional diversity, and weakens community networks that underpin urban resilience [11,68].

Hypothesis H2 (Touristification and Residential Pressure). 

The intensity of tourist housing in urban space generates direct and specific pressures on the residential market—increasing housing prices and driving residential displacement—independently of aggregate competitive performance.

Tourism specialisation also generates labour structures characterised by high temporality, relatively low wages, and limited professional advancement opportunities [69,70]. This precarisation configures structural social vulnerability. Additionally, these processes generate growing conflicts amongst residents, visitors, tourism entrepreneurs, and public authorities [41,71]. The emergence of social movements against mass tourism reflects a crisis of social acceptability that constitutes a vulnerability factor for the sustainability of the urban tourism model [72,73].

The COVID-19 pandemic constituted an abrupt revelation, exposing the structural fragilities of urban models intensively oriented towards tourism that had been perceived as examples of competitive success [16]. These fragilities tend to accumulate gradually during favourable economic cycles [67].

Hypothesis H3 (Dual Role of Tourism Specialisation). 

Tourism specialisation increases urban vulnerability both directly, through structural economic dependence on a single sector, and indirectly, by amplifying tourism competitive performance and thereby intensifying systemic pressures on the urban system.

2.3. Towards the Concept of Urban Competitive Vulnerability: Theoretical Integration and Operationalisation

The integration of competitiveness and vulnerability into a single analytical framework constitutes a recent conceptual advance in the tourist destination literature. Pioneering studies such as those by Duro et al. [25] on Mediterranean coastal destinations and Barrera-Martínez et al. [74] on island destinations have empirically demonstrated that territories with elevated tourism performance can simultaneously exhibit growing structural vulnerabilities. These works have proposed transcending the traditional dichotomous vision. The integrated perspective recognises that, in contexts of high tourism intensity, competitive success can amplify structural vulnerabilities.

Transferring this approach to the urban sphere implies recognising the specificities of urban territory as a complex, multifunctional, and socially heterogeneous system. Cities are not merely tourist destinations, but living spaces where tourism interacts with multiple urban functions [1,35]. Urban competitive vulnerability is proposed as the structural propensity of an urban system to experience erosion of its competitive position due to the progressive accumulation of internal fragilities linked to its tourism development model, even in contexts of apparent high economic performance.

This definition incorporates three fundamental conceptual elements, emphasising the structural nature of vulnerability as arising from the internal configuration of the urban system rather than from external disturbances, highlighting the paradoxical coexistence between apparent competitiveness and latent vulnerability when conventional indicators of success obscure accumulating fragilities, and underscoring the systemic and multidimensional character of the phenomenon. Urban competitive vulnerability emerges from the interaction among multiple, mutually reinforcing fragilities, including economic dependence on tourism, pressure on residential markets, labour precarisation, deterioration of heritage assets, erosion of social cohesion and institutional weakness. Its empirical operationalisation requires the development of measurement instruments that simultaneously integrate competitive performance indicators and structural fragility indicators, through methodologies that enable the identification of urban configurations of differentiated risk.

Hypothesis H4 (Competitiveness–Vulnerability Paradox). 

Tourism competitiveness is positively associated with urban structural vulnerability, such that cities with higher levels of competitive performance systematically exhibit greater structural fragility, even during periods of high economic performance (the competitiveness–vulnerability paradox).

Hypothesis H5 Exploratory (Social Mechanisms of Competitive Vulnerability). 

The social mechanisms through which competitive vulnerability is produced, experienced, and contested in tourist cities can be more fully captured through mixed-method research designs that combine quantitative panel analysis with qualitative inquiry—including urban ethnographies, semi-structured interviews, and process tracing—than through quantitative approaches alone. This hypothesis is not empirically tested in the present study, but constitutes a priority avenue for future research derived from the limitations identified in the preceding analytical framework.

3. Materials and Methods

3.1. Data Sources and Research Design

This study employs a balanced longitudinal panel design to analyse six Spanish cities of high tourism intensity during the period 2014–2024. Case selection responds to variation criteria across three dimensions: (1) demographic size (from 858,000 inhabitants in Palma de Mallorca to 7.1 million in Madrid); (2) tourism specialisation (from 0.4% of total employment in Madrid to 7.4% in Palma); (3) territorial configuration (insular, coastal, and inland cities). The cities to be analysed are Barcelona, Madrid, Málaga, Palma de Mallorca, Seville, and Valencia, which together account for 42% of hotel overnight stays in Spain and constitute the principal nodes of the national urban tourism system.

Data have been collected and analysed from high-quality official sources with consistent temporal coverage. Tourism variables (arrivals, overnight stays, occupancy, and accommodation capacity) are obtained from the Hotel Occupancy Survey and the Tourist Apartment Occupancy Survey of the National Statistics Institute (INE), with monthly periodicity, aggregated at the annual level. Economic and labour variables (municipal GDP, average income, activity rates, unemployment, and employment by sector) are derived from the INE, the State Tax Administration Agency (AEAT), and Social Security registers. Residential and urban variables (housing and rental prices, holiday housing stock, evictions, resident population) are extracted from the Ministry of Transport, Mobility and Urban Agenda, cadastral registers, and tourist accommodation platforms (administrative registration data).

The analytical period 2014–2024 has been selected as it enables capturing the tourism intensification phase driven by digital platforms post-2013, includes the exogenous COVID-19 shock (2020–2021) as a natural experiment to examine differential resilience, and analyses post-pandemic recovery trajectories through 2024. The resulting panel comprises N = 66 city-year observations (6 cities × 11 years), providing sufficient temporal and cross-sectional variation for robust estimation of estimated effects using econometric panel techniques.

It is important to note that the sampling strategy follows a purposive variation logic rather than a statistical representativeness criterion. The six cities were selected to maximise theoretical relevance by maximising variation across dimensions of scale, specialisation, and territorial typology [75,76]. Results should be interpreted as evidence of causal mechanisms operating within the selected cases, not as statistical estimates generalisable to all Spanish cities. Limitations and directions for extending the analysis are discussed in Section 6.4.

3.2. Operationalisation of Variables: Construction of Composite Index

The operationalisation of the theoretical concept of urban competitive vulnerability requires the construction of multidimensional composite indices that capture the phenomenon’s complexity. We follow the methodological approach of Nardo et al. [77] for composite indicator construction, which includes the theoretical foundation of dimensions, the selection of observable variables, scale standardisation, weighting based on conceptual relevance, and statistical validation of internal consistency.

3.2.1. Tourism Competitiveness Index

The Tourism Competitiveness Index (TCI) operationalises the intensity and efficiency of urban tourism activity through five standardised components, weighted according to their theoretical relevance for capturing the competitive performance of the tourism system:

$$\begin{array}{cc}\hfill {TCI}_{it}& =0.30 x Z_{arrivals\_capita}+ {0.30 x Z}_{overnights\_capita}\hfill \\ & + {0.20 x Z}_{beds\_capita}+ {0.10 x Z}_{occupancy }\hfill \\ & + {0.10 x Z}_{productivity}\hfill \end{array}$$

(1)

where Z expresses standardisation through z-scores (mean = 0, standard deviation = 1) for each variable. Arrivals and overnight stays per capita (30% weighting each) capture tourism activity volume and intensity, normalised by resident population and corrected for urban size. Accommodation beds per capita (20%) measure the relative infrastructural capacity of the system. The average weighted occupancy rate (10%) reflects operational efficiency. Tourism productivity (10%), calculated as the number of overnight stays per sector worker, reflects labour efficiency. The resulting index exhibits high internal consistency (Cronbach’s α = 0.87) and correlates strongly with alternative measures of urban tourism competitiveness in the literature (r > 0.82).

3.2.2. Competitive Vulnerability Index

The Competitive Vulnerability Index (CVI) constitutes the principal dependent variable. It operationalises the structural fragility of the urban-tourism system through four theoretically grounded sub-dimensions, each capturing a specific mechanism of vulnerability generation: structural dependence, territorial pressure, socio-economic fragility, and volatility.

Structural dependence captures vulnerability derived from sectoral specialisation and economic concentration in tourism activities:

$$\begin{array}{cc}\hfill {CVI}_{structural\_dependence\_it}& =0.40 x Z_{tourism\_employment\_share}+ {0.30 x Z}_{tourism\_workers\_capita}\hfill \\ & +{0.30 x Z}_{tourism\_services\_employment\_ratio}\hfill \end{array}$$

(2)

where tourism employment share corresponds to the percentage of tourism employment over total employment; tourism workers per 1000 inhabitants corresponds to tourism workers per 1000 inhabitants; and tourism/services employment ratio corresponds to the percentage of tourism employment over services employment.

Territorial pressure captures tensions on urban space and the residential market derived from touristification:

$$\begin{array}{cc}\hfill {CVI}_{territorial\_pressure\_it}& =0.30 x Z_{housing\_pressure}+ {0.25 x Z}_{holiday\_homes\_intensity}\hfill \\ & +{0.25 x Z}_{price\_rent\_ratio}+ {0.20 x Z}_{eviction\_rate}\hfill \end{array}$$

(3)

where $Z_{housing\_pressure}$ corresponds to total tourist accommodation places + holiday dwelling places/population × 1000; $Z_{holiday\_homes\_intensity}$ corresponds to holiday dwellings/population × 1000; $Z_{price\_rent\_ratio}$ corresponds to price per m² of housing/average monthly net income; and ${ Z}_{eviction\_rate}$ corresponds to evictions/population × 1000.

Socio-economic fragility captures the impact on social cohesion and the living conditions:

$${CVI}_{socioeconomic\_fragility\_it}=0.40 x Z_{unemployment\_rate}+ {0.35 x Z}_{rental\_effort}+{0.25 x Z}_{housing\_price}$$

(4)

where $Z_{rental\_effort}$ corresponds to rental price per m² × 70 m²/average net income per person × 100, representing the percentage of income devoted to renting a standard dwelling.

Volatility captures temporal instability and abrupt changes in key variables:

$${CVI}_{volatility\_it}=0.35 x {\mid Z}_{\mathsf{\Delta }housing\_price }\mid + {0.35 x \mid Z}_{\mathsf{\Delta }arrivals}\mid + 0.30 x \mid Z_{\mathsf{\Delta }population}\mid$$

(5)

The weights reflect the relative theoretical relevance of each dimension within the conceptual framework of competitive vulnerability. The index presents adequate statistical properties: acceptable internal consistency (Cronbach’s α = 0.74), low multicollinearity between sub-indices (VIF < 3.5), and substantial variation both temporally and cross-sectionally (within-SD = 0.82, between-SD = 1.23).

3.2.3. Weighting Rationale and Sensitivity Analysis

The weighting structures of both composite indices follow a theory-first approach consistent with the study’s deductive logic [13,77]. For the TCI, arrivals and overnight stays per capita receive the highest weights (0.30 each) because they directly capture the volume and intensity of tourism activity, the core dimensions of competitive performance in the destination management literature [31]. Accommodation capacity (0.20) reflects structural supply-side endowment. Occupancy and productivity (0.10 each) capture operational efficiency. For the CVI, structural dependence receives the highest weight (0.30) because it constitutes the fundamental mechanism through which tourism generates systemic fragility by reducing productive diversification [25]. Territorial pressure and socioeconomic fragility are assigned equal weights (0.25 each) as primary transmission mechanisms. Economic volatility receives a lower weight (0.20) as it represents a symptomatic expression of fragility rather than an underlying causal condition.

A purely data-driven weighting scheme (PCA) was not adopted as the primary approach because PCA-derived weights are sensitive to the specific composition of cities and years in the sample. Two alternative weighting schemes are tested: equal weights (0.25 per dimension) and PCA-derived weights. Correlations between the leading indices and the two alternative schemes range from r = 0.89 to r = 0.94 (all p < 0.001), and Cronbach’s α = 0.82. These results confirm robustness to the specific weighting decisions. Full details are reported in Appendix A.

3.2.4. Touristification Index

The Touristification Index (TFI) is specifically constructed to capture the transformation of urban space through tourism and constitutes the primary instrument for testing H2 (Touristification and Residential Pressure):

$${TFI}_{it}=0.30 x Z_{beds\_per\_capita}+ {0.40 x Z}_{holiday\_homes\_intensity}+ {0.30 x Z}_{house\_pressure}$$

(6)

For the analysis of COVID-19 impact, impact variables are constructed (percentage declines in 2020 vs. 2019 for arrivals, employment, and GDP) alongside resilience variables (index = [Level₂₀₂₃ − Level₂₀₂₀]/|Decline₂₀₂₀|), where values above 1 indicate complete recovery and values below 1 indicate partial recovery.

3.3. Panel Model Estimation and Validation

The panel structure of the data allows us to control for unobserved heterogeneity across cities using fixed effects (FE) models, which constitute our main specification. The baseline model for testing H4 (the competitiveness-vulnerability paradox) takes the form:

$${CVI}_{it}= {\beta }_0+ {\beta }_1{TCI}_{it}+ {\beta }_2 X_{it}+ {\alpha }_i+ {\delta }_t+ {\epsilon }_{it}$$

(7)

where i indexes cities, t indexes years, ${\alpha }_i$ represents city fixed effects (time-invariant unobserved heterogeneity), ${\delta }_t$ captures time fixed effects when specified, $X_{it}$ includes time-varying controls, and ${\epsilon }_{it}$ is the idiosyncratic error term.

The within transformation of fixed effects eliminates α_i and allows estimation of β₁ through:

$${CVI}_{it}- {CV\stackrel{-}{\mathrm{I}}}_i = {\beta }_1({TCI}_{it} - {TC\stackrel{-}{\mathrm{I}}}_i) + {\beta }_2(X_{it}- {\stackrel{-}{\mathrm{X}}}_i) + {\epsilon }_{it}$$

(8)

where ${CV\stackrel{-}{\mathrm{I}}}_i$ = T⁻¹ ${\sum }_t$ ${CVI}_{it}$ denotes the temporal mean for each city. This transformation exploits within variation (changes in each city over time) whilst controlling for permanent differences between cities. The coefficient β₁ thus captures the estimated effect of changes in tourism competitiveness on changes in vulnerability, net of unobserved heterogeneity.

The choice between pooled OLS, fixed effects, and random effects models is made through formal tests. The F-test for poolability tests (Appendix A) the null hypothesis that individual fixed effects are jointly equal to zero; rejection (p < 0.001 in all models) indicates significant heterogeneity across cities that must be controlled for. The Hausman test compares fixed- and random-effects models. Under random effects, E (${\alpha }_i$|$X_{it}$) = 0 is assumed, whilst fixed effects permit arbitrary correlation. In all main models, the test favours fixed effects, indicating that unobserved heterogeneity correlates with the regressors.

Heteroscedasticity and autocorrelation problems are addressed using city-level cluster-robust standard errors, which allow for arbitrary correlation of errors within each city over time and for cross-sectional heteroscedasticity. This specification generates larger standard errors than conventional estimates. We formally verify heteroscedasticity using the Breusch–Pagan test (LM = 16.70, p < 0.001) and autocorrelation using the Durbin–Watson statistic (DW = 1.64, within the acceptable range of 1.5–2.5, indicating the absence of severe autocorrelation). Multicollinearity is assessed through variance inflation factors (VIF). All values are below the critical threshold of 10, and most are below 5, indicating the absence of problematic multicollinearity that could inflate standard errors or destabilise estimates.

A note on potential endogeneity is warranted, with two sources of concern meriting particular attention. The first concerns reverse causation (TCI ← CVI). Whilst theoretically conceivable, the temporal logic of the hypothesised direction—whereby supply-side investment (TCI) precedes the accumulation of territorial and social pressures (CVI)—supports TCI → CVI as the dominant pathway. The robustness of the TCI coefficient across all specifications (β = 0.525–0.540) further reduces the plausibility of reverse causation as the sole explanation. The second concern relates to time-varying omitted-variable bias. Whilst fixed effects eliminate time-invariant confounders, city-level factors that evolve may still bias estimates. Future research should employ instrumental variable approaches, natural experiments exploiting regulatory reforms such as the introduction of short-term rental quotas, or regression discontinuity designs. These represent priority directions for Section 6.4.

To test economic specialisation (H3), a mediation analysis is conducted to examine whether the effect of tourism specialisation on vulnerability operates directly or through tourism competitiveness (TCI) as a mediating variable. Four sequential fixed effects models were estimated:

Effect of specialisation on mediator (TCI):

$${TCI}_{it}={ \gamma }_0 + a·{Specialisation }_{it}+{ \gamma }_2 X_{it} + {\alpha }_i + {\epsilon }_{it}$$

(9)

Effect of mediator (TCI) on outcome (CVI):

$${CVI}_{it} = {\varphi }_0+ b·{TCI }_{it}+ {\varphi }_2{X}_{it}+ {\alpha }_i+ {\epsilon }_{it}$$

(10)

Total effect of specialisation on outcome (without mediator):

$${CVI}_{it} = {\theta }_0 + c·{Specialisation }_{it} + {\theta }_2{X}_{it} + {\alpha }_i + {\epsilon }_{it}$$

(11)

Direct effect (including specialisation and mediator):

$${CVI}_{it} ={ \omega }_0 + c^{\prime }·{Specialisation }_{it}+ b^{\prime }·{TCI}_{it}+{ \omega }_2 {X }_{it}+{ \alpha }_i+ {\epsilon }_{it}$$

(12)

Thus, complete mediation is confirmed if a is significant (specialisation affects TCI), if b is significant (TCI affects CVI), if c is significant (total effect exists), and if c’ loses significance or is reduced when the mediator is included. The indirect effect is calculated as a × b, and the mediated proportion as (a × b)/c. The significance of the indirect effect is assessed using the Sobel test, which estimates the standard error of a × b via the delta method.

To test the amplification of vulnerabilities facing external shocks (H1), a panel model with temporal interactions is estimated, analysing the variation in each city before, during, and after the shock:

$${Arrivals }_{it}={ \beta }_0+ \sum _{t=2020}^{2023} {\beta }_t \left({CVI}_{pre} x D_{year=t}\right)+ \sum _{t=2020}^{2023}{ \gamma }_t D_{year=t }+ {\mathsf{\beta }}_4{\mathrm{X}}_{it}+{\mathsf{\alpha }}_i+{\mathsf{\epsilon }}_{it}$$

(13)

where ${CVI}_{pre}$ denotes the average CVI for each city in the period 2014–2019 (prior accumulated vulnerability), $D_{year=t}$ are dummies for each COVID/post-COVID year, and the ${\beta }_t$ coefficients capture the differential effect of prior vulnerability in each specific year. This specification allows identification of whether more vulnerable cities experienced more severe impacts in 2020–2021 and slower recoveries in 2022–2023.

Complementarily, a cross-sectional analysis with 6 cities is employed, regressing impact measures (percentage decline in tourism, employment, GDP) on pre-COVID CVI and controls:

$${Impact}_{{COVID}_i} = \alpha + {\delta }_1{ CVI}_{{pre}_i}+ {\delta }_2 {TCI}_{{pre}_i}+{ \delta }_3 {Controls}_i+{ u}_i$$

(14)

Whilst this strategy presents statistical power limitations due to the reduced number of cities, it provides estimates of the total effect of prior vulnerability on the magnitude of impact.

4. Results

4.1. Descriptive Statistics and Sample Characterisation

The empirical analysis is conducted on a balanced panel of 6 Spanish tourist cities observed during the period 2014–2024, yielding a total of 66 city-year observations. Table 1 presents the descriptive statistics of the key variables employed in the study, organised into four fundamental analytical dimensions (Appendix B).

Table 1. Descriptive statistics of key variables.

Dimensions	Mean	Std.Dev (SD)	Min	Max
DEPENDENT VARIABLE
Dimension A: Competitiveness and vulnerability
TCI	0	1	−1.12	3.94
CVI	0	1	−1.35	3.01
CVI_Dependence	0	1	−0.59	2.26
CVI_Pressure	0	1	−0.59	0.45
CVI_Fragility	0	1	−0.82	0.48
CVI_Volatility	0	1	−0.86	1.24
EXPLANATORY VARIABLES
Dimension B: Tourism variables
Total arrivals	6.86	4.03	1.12	13.29
Overnight stays	21.57	17.24	3.21	63.04
Tourism specialisation	2.87	2.31	0.35	7.35
Accommodation places per capita	18.32	14.67	3.12	52.34
Dimension C: Socio-residential variables
Price per m2 housing	2.168	0.784	1.090	4.442
Holiday homes	12.5	15.3	1.2	48.9
Unemployment rate	16.53	6.12	7.50	32.78
Resident population	3.24	2.16	0.86	7.09
CONTROL VARIABLES
Dimension D: COVID impact
Fall in tourist arrivals	−72.6	8.2	−86.6	−63.0
Years until recovery	2.83	0.75	2.00	4.00
Resilience index	1.06	0.04	1.00	1.12

Source: Authors’ own elaboration based on data analysis. Note: TCI and CVI indices standardised (mean = 0, std dev (SD) = 1). Statistics calculated for the period 2014–2024.

The composite indices constructed (TCI and CVI) present, by methodological construction, a mean of zero and a standard deviation of one. Regarding ranges, the TCI oscillates between −1.12 and 3.94, whilst the CVI varies between −1.35 and 3.01, evidencing significant structural heterogeneity amongst the cities analysed.

The socio-residential variables reveal structural tensions. The average housing price stands at €2168/m² (SD = 0.784; range: 1090–4442), reflecting both regional differences and the effects of tourism pressure. The proportion of holiday homes has a mean of 12.5% (SD = 15.3%), reaching 48.9% in the most extreme cases of residential touristification. The unemployment rate averaged 16.5% (SD = 6.1%; range: 7.5–32.8%), evidencing persistent labour vulnerabilities.

The impact of COVID-19 demonstrates the severity of the sectoral shock: mean fall in arrivals of −72.6% in 2020 (SD = 8.2%; range: −86.6% to −63.0%), mean recovery time of 2.83 years (SD = 0.75; range: 2–4 years), and resilience index of 1.06 (SD = 0.04; range: 1.00–1.12), indicating a recovery slightly above pre-pandemic levels.

4.2. Amplification of Vulnerability Under the COVID-19 Shock

The first hypothesis (H1) posits that cities with higher levels of pre-pandemic competitive vulnerability experienced more severe impacts during the COVID-19 crisis and slower recovery trajectories, thereby confirming that accumulated structural vulnerability conditions adaptive capacity in the face of external shocks. Table 2 presents three complementary empirical approaches to testing this hypothesis.

Table 2. Differential effects of the COVID-19 shock.

	(1) Temporal Interactions	(2) Correlations (2020 Impact)	(3) Resilience Index
CVI_pre x 2020	−3.688 ** (1.248)	-	-
CVI_pre x 2021	−2.186 *** (0.559)	-	-
CVI_pre x 2022	35.29 (58.62)	-	-
CVI_pre x 2023	283.67 *** (4.444)	-	-
CVI_pre (direct effect)	-	-	−0.002 (0.069)
TCI_pre	-	-	0.027 (0.119)

Source: Authors’ own elaboration. Notes: *** p < 0.01, ** p < 0.05, * p < 0.10. Tourist arrivals per capita (expressed per 1000 inhabitants) (1) Fixed-effects panel models with clustered standard errors; dependent variable: tourist arrivals per capita. (2) Pearson correlations with the percentage decline in arrivals in 2020. (3) Cross-sectional OLS regression; dependent variable: resilience index.

The temporal interaction model (1) estimates a fixed-effects panel including interactions between pre-pandemic competitive vulnerability (2019) and year-specific dummy variables (2020–2023), using tourist arrivals per capita as the dependent variable. This specification allows assessment of whether the effect of pre-existing vulnerability on tourism performance varies systematically during and after the pandemic shock.

The results reveal a temporal pattern consistent with hypothesis H1. In 2020, the interaction coefficient is negative and statistically significant (β = −3.688, p < 0.05), indicating that cities with higher pre-pandemic vulnerability experienced significantly steeper declines in tourist arrivals. A one-standard-deviation increase in pre-pandemic CVI is associated with an additional reduction of 3.688 arrivals per capita, amplifying the contractionary effect of the shock.

This effect persists in 2021 (β = −2.186, p < 0.001), albeit with reduced magnitude, confirming that pre-existing vulnerability influenced both the immediate impact and the early recovery phase. The 2022 coefficient is no longer statistically significant (β = 35.29, p > 0.10), suggesting convergence in trajectories. By contrast, the interaction term for 2023 becomes positive and highly significant (β = 283.67, p < 0.001), potentially reflecting more substantial rebound effects in cities that had previously experienced deeper contractions.

The correlation analysis (2) further supports the amplification hypothesis, revealing a positive and statistically significant association between pre-pandemic vulnerability and the magnitude of the tourism collapse in 2020. More vulnerable cities experienced proportionally larger declines in tourist arrivals.

The resilience model (3), which examines the determinants of the recovery index (ratio of arrivals in 2023 relative to 2019), shows no statistically significant effects of either pre-pandemic vulnerability (β = −0.002, p > 0.10) or competitiveness (β = 0.027, p > 0.10). This finding nuances the hypothesis: while vulnerability strongly conditioned the severity of the initial shock and early recovery (2020–2021), it did not determine medium-term recovery outcomes by 2023.

4.3. Touristification and Residential Pressure

The second hypothesis (H2) posits that tourism intensification generates specific pressures on the residential market and the urban territorial system, constituting a direct mechanism of vulnerability that operates with a certain degree of independence from aggregate competitive performance. Table 3 presents the results of three complementary model specifications examining these dynamics: territorial pressure, housing prices, and eviction rates.

Table 3. Territorial pressure and residential mechanism.

Variables	(1) CVI_Pressure	(2) Housing Prices	(3) Eviction Rate
TFI	55.71 ** (20.55)	-	-
Holiday housing intensity	-	0.288 *** (0.051)	0.125 ** (0.048)
TCI	0.079 (0.041)	−0.116 * (0.069)	0.034 (0.052)
Population (log)	1.496 * (0.855)	-	2.347 *** (0.734)
Income per capita	-	0.176 (0.267)	-
Time	−0.026 (0.016)	0.065 (0.057)	−0.012 (0.028)

Source: Authors’ own elaboration. Notes: *** p < 0.01, ** p < 0.05, * p < 0.10. Robust standard errors clustered by city in parentheses.

The residential pressure model (1) examines the effect of touristification on territorial pressure (CVI_Pressure) and reveals a positive and statistically significant relationship (β = 55.71, p < 0.05). A one-unit increase in touristification is associated with a substantial increase in territorial pressure (β = 55.71), a magnitude that underscores the strength of this mechanism. Tourism competitiveness (TCI) does not reach statistical significance (β = 0.079, p > 0.10), suggesting that territorial pressure responds primarily to the intensity of tourist use of urban space rather than to overall competitive success. Population size exerts a marginally significant positive effect (β = 1.496, p < 0.10), reflecting greater land scarcity in dense metropolitan contexts.

The housing price model (2) analyses the impact of holiday housing on real estate prices and confirms a positive and highly significant effect (β = 0.288, p < 0.001), indicating a substantial increase in residential market pressure in more touristified cities. This economically substantial magnitude empirically confirms the tensions associated with touristification and tourism-driven gentrification. The coefficient for TCI is marginally negative (β = −0.116, p < 0.10), indicating that, after controlling for holiday housing pressure, aggregate competitiveness does not automatically translate into higher housing prices.

The eviction model (3) examines effects on eviction rates, a direct indicator of residential exclusion. Holiday housing intensity exerts a positive, statistically significant effect (β = 0.125, p < 0.05), confirming that touristification not only inflates housing prices but also drives tangible processes of residential displacement. Population size shows a strong positive effect (β = 2.347, p < 0.001), reflecting the higher intensity of housing conflicts in larger cities. TCI remains non-significant, reinforcing the interpretation that adverse social outcomes operate primarily through specific territorial pressure mechanisms.

Overall, the results provide strong support for hypothesis H2. Touristification produces specific residential effects that constitute direct mechanisms of vulnerability and operate independently from general tourism competitiveness. The conversion of residential housing into short-term tourist accommodation emerges as a critical vector of territorial pressure, with observable consequences for housing affordability and residential exclusion.

4.4. Economic Specialisation and Mediation Through Tourism Competitiveness

The third hypothesis (H3) posits that tourism specialisation increases urban vulnerability both directly and indirectly, through its prior effect on tourism competitiveness. This formulation recognises that sectoral specialisation operates simultaneously as a pull factor—enhancing competitive performance—and as a source of structural fragility, by reducing economic diversification and increasing dependency on tourism-related activities.

The mediation analysis (Table 4) follows the standard analytical approach, sequentially estimating paths a, b, c, and c′ (see Appendix B for technical details).

Table 4. Mediation analysis of tourism specialisation.

Variables	(1) Spec -> TCI	(2) TCI -> CVI	(3) Spec-> CVI	(4) Spec + TCI -> CVI
Tourism specialisation	0.615 *** (0.094)	-	0.504 *** (0.053)	0.460 (0.202)
TCI	-	0.540 *** (0.064)	-	0.192 (0.121)
Population (log)	−0.680 (1.592)	2.900 ** (1.337)	5.102 ** (1.993)	4.832 (3.604)
Time	−0.014 *** (0.004)	0.016 (0.033)	−0.013 (0.029)	−0.008 (0.041)

Source: Authors’ own elaboration. Notes: *** p < 0.01, ** p < 0.05, * p < 0.10. Robust standard errors clustered by city in parentheses. Indirect effect (a × b): 0.615 × 0.540 = 0.332 ***. Proportion mediated: 65.8%. Sobel test: z = 5.23, p < 0.001.

The results reported in column (1) indicate that tourism specialisation exerts a positive and highly significant effect on tourism competitiveness (β = 0.615, p < 0.001). This confirms that cities with a higher concentration of tourism-related activity tend to exhibit superior competitive performance, thereby validating the implicit assumption underlying many urban tourism development strategies.

Column (2) replicates the core result identified previously, confirming the positive effect of tourism competitiveness on urban vulnerability (β = 0.540, p < 0.001), corresponding to path b in the mediation model. Column (3) estimates the total effect of tourism specialisation on vulnerability, and reveals a positive, statistically significant relationship (β = 0.504, p < 0.001). However, this specification does not distinguish between direct and mediated effects.

Column (4) provides the key interpretative insight by simultaneously including both predictors. The direct effect of tourism specialisation is substantially reduced (β = 0.460). It loses statistical significance (p > 0.10), while the coefficient associated with competitiveness also weakens (β = 0.192), suggesting strong collinearity between the two processes and supporting partial mediation.

The calculated indirect effect (a × b = 0.332) and the Sobel test (z = 5.23, p < 0.001) confirm the mediation is statistically significant. The mediated proportion reaches 65.8%, indicating that approximately two-thirds of the total effect of tourism specialisation on urban vulnerability is transmitted through its prior impact on competitiveness.

These findings confirm that tourism specialisation constitutes a dual determinant of vulnerability. It generates direct structural fragilities associated with sectoral dependence while simultaneously enhancing competitive performance, thereby intensifying urban pressures through indirect mechanisms. The indirect pathway is quantitatively dominant, suggesting that vulnerability materialises primarily through the intensification of tourism success and its systemic consequences. While specialisation increases competitive performance, this same success operates as a transmission channel for structural vulnerabilities. Consequently, vulnerability does not arise inevitably from specialisation per se, but rather from the intensive growth model that specialisation enables and which, in the absence of adequate regulatory mechanisms, exerts cumulative pressures on the urban system.

4.5. Analysis of the Competitiveness-Vulnerability Paradox

The fourth hypothesis (H4) of this study posits a positive and significant relationship between tourism competitiveness and urban structural vulnerability, thereby configuring the so-called competitive vulnerability paradox. Table 5 presents the results of the estimated model.

Table 5. Analysis of the Urban Competitive Vulnerability Index.

Dependent Variable: CVI
Variables	(1) Base	(2) Controls	(3) Time	(4) Size
TCI	0.540 *** (0.064)	0.540 *** (0.064)	0.538 *** (0.065)	0.525 *** (0.072)
Population (log)	2.900 ** (1.337)	2.821 ** (1.289)	2.803 ** (1.301)	-
Time	0.016 (0.033)	0.018 (0.034)	0.089 (0.124)	0.014 (0.033)
Time2	-	-	−0.004 (0.007)	-
Income per capita (log)	-	0.123 (0.245)	-	-
Medium-sized city	-	-	-	0.287 (0.423)
Large city	-	-	-	0.156 (0.389)

Source: Authors’ own elaboration based on data analysis. Note: *** p < 0.01, ** p < 0.05, * p < 0.1 (Appendix C).

The base model (1) estimates the relationship between the TCI and the CVI, while controlling for population size and temporal trend. The results confirm hypothesis H4, yielding a positive, statistically significant TCI coefficient (β = 0.540, p < 0.001). An increase in one standard deviation in tourism competitiveness is associated with an increase of 0.540 standard deviations in vulnerability. This result implies that cities with higher levels of tourism performance tend simultaneously to accumulate greater structural fragilities, confirming the paradoxical coexistence between competitive success and systemic vulnerability.

Population size also shows a significant, positive effect (β = 2.900, p < 0.05), suggesting that larger cities experience higher levels of vulnerability due to greater urban complexity and intensified territorial pressures. The temporal variable does not reach statistical significance, indicating the absence of a common linear trend during the period analysed.

5. Discussion

This study provides systematic empirical evidence on the competitiveness–vulnerability paradox in Spanish tourist cities, confirming that superior tourism performance can coexist with—and actively generate—structural fragilities. These findings challenge implicit assumptions in the urban competitiveness literature, which has often treated tourism growth as an intrinsically desirable objective without adequately examining its broader system-wide implications for urban sustainability. In this section, we interpret the main results in relation to existing scholarship, identify specific causal mechanisms, contrast our evidence with international cases, and discuss the study’s limitations. The subsections follow the order of theoretical centrality rather than the sequential order of the results: we begin with the competitiveness–vulnerability paradox (H4) as the core conceptual contribution, proceed through the two transmission mechanisms (H3 and H2), and conclude with the shock amplification analysis (H1), which functions as an empirical corroboration of the accumulated fragility argument.

5.1. The Competitiveness-Vulnerability Paradox: Theoretical Implications

The core result—the positive and robust relationship between tourism competitiveness (TCI) and competitive vulnerability (CVI) (β = 0.540, p < 0.001)—fundamentally calls into question dominant models of urban and destination competitiveness. Seminal conceptual frameworks [4,26,31] have emphasised visitor attraction and revenue generation as indicators of success, implicitly assuming that tourism growth produces economic robustness. Our findings demonstrate that this assumption is problematic, since a one-standard-deviation increase in competitiveness is associated with a 0.54-standard-deviation increase in vulnerability after controlling for urban heterogeneity and temporal effects.

This pattern suggests that vulnerability is not merely exposure to external shocks (as in the conventional conceptualisations of [22,23], but rather an endogenous condition emerging from the accumulation logic operating during expansionary cycles. The result aligns with critiques in the critical urban resilience literature [56,57], which argue that the conventional emphasis on recovery capacity can obscure the structural conditions that generate vulnerability. In our case, competitive “success” functions as a vector of fragilisation, as pressures on housing markets, sectoral specialisation and labour precariousness accumulate precisely under conditions of high tourism performance.

The robustness of this relationship—remaining essentially invariant to the inclusion of economic controls, quadratic time terms, and urban size effects—suggests that the paradox operates across cities of different scales and development levels. The stability of the coefficient (β = 0.54) across specifications indicates that the association is unlikely to be an artefact of obvious omitted variables; rather, it reflects a structural feature of intensive urban tourism development.

5.2. Transmission Mechanisms: Specialisation, Mediation and Thresholds

The mediation analysis reveals the causal architecture through which tourism specialisation generates vulnerability. Two-thirds of the total effect (65.8%) operates indirectly: specialisation increases competitiveness (β = 0.615, p < 0.001), which subsequently amplifies vulnerability (β = 0.540, p < 0.001). The remaining third corresponds to a direct effect associated with sectoral dependence per se. This decomposition has important interpretative implications.

First, it indicates that vulnerability does not stem simply from tourism “monoculture”, but more fundamentally from the intensive growth model that specialisation enables. Agglomeration economies and operational efficiencies, as documented by Maskell & Malmberg [30], do materialise competitive advantages; however, these same dynamics intensify urban pressures when they operate in the absence of effective regulation. The relationship is therefore not mechanical, as cities with similar levels of specialisation may display divergent vulnerability trajectories depending on their regulatory and institutional capacities.

Second, the predominance of the indirect pathway suggests that policies aimed solely at reducing specialisation (i.e., diversification strategies) may be insufficient if they do not simultaneously address the pressures generated by existing competitive intensification. Effective management requires intervention at multiple points within the causal chain, including the containment of tourism growth where necessary, the protection of residential housing stock and the strengthening of sectoral diversification.

A further question our analysis cannot definitively resolve is whether there are touristification thresholds beyond which vulnerability accelerates in a non-linear manner. The quadratic time specification (Table 5, column 3) does not capture significant effects, but this reflects aggregate temporal patterns rather than intensity thresholds per se. Future research could explore non-linear specifications (e.g., splines, logarithmic transformations) or regime-switching models to identify tipping points at which urban pressures become qualitatively different. The overtourism literature suggests the existence of such thresholds [40,41], but their precise identification requires dedicated methodological approaches.

5.3. Residential Touristification: Comparison with International Evidence

The substantial effect of the touristification index on housing prices (β = 0.288, p < 0.001) provides quantitative confirmation of dynamics documented qualitatively across multiple contexts. Holiday housing intensity shows a positive and highly significant association with housing prices—an economically meaningful magnitude with direct implications for housing accessibility.

This finding is consistent with scholarship on short-term rental platforms across different geographies. Wachsmuth & Weisler [10] show that Airbnb inflates rents in New York through two channels: the direct conversion of long-term rental units to short-term use and broader pressure in already-constrained markets. Our estimated coefficient (0.288) exceeds that reported by Ardura Urquiaga et al. [66] for Madrid (0.18), potentially reflecting that our panel includes cities subject to higher aggregate tourism pressure. In Lisbon, Sequera and Nofre [11] identify displacement processes in central neighbourhoods where touristification exceeds 15% of the residential stock—a threshold that Palma de Mallorca and Barcelona surpass in specific areas according to our data.

The positive effect on evictions (β = 0.125, p < 0.05) provides quantitative support for narratives of tourism-driven gentrification in Mediterranean cities. Importantly, this pattern is not exclusive to Southern Europe. Gotham [67] documents analogous processes in post-Katrina New Orleans, where tourism-oriented reconstruction displaced pre-existing populations. The apparent universality of the mechanism—converting residential space into tourism consumption space—produces inflation and displacement, suggesting it operates across contexts. However, its intensity varies with regulatory frameworks.

Compared with cities with stricter regulations, cities with less stringent regulations have partially mitigated these effects. Amsterdam introduced a 30-day cap on short-term rentals in 2017 and restricted new licences in saturated areas, achieving relative stabilisation of pressures (although constraints persist). Barcelona tightened regulations in 2021 by strengthening municipal licensing requirements and restricting new permits in high-density areas, with outcomes still under evaluation. While our data broadly capture the pre-intensive regulation phase, cross-sectional variability is consistent with the idea that cities with stronger institutional capacity (proxied by administrative effectiveness and sectoral coordination) may experience more moderated pressures.

5.4. The COVID-19 Shock and Differential Resilience

The pandemic impact analysis supports the concept of competitive vulnerability, but with important nuances. Cities with higher pre-pandemic vulnerability experienced sharper declines in 2020 and 2021 (β = −3.688 and β = −2.186, respectively). However, these differences attenuated substantially in 2022–2023, with the resilience model (Table 2, column 3) indicating that pre-crisis vulnerability had no significant effect on medium-term recovery.

This trajectory suggests that vulnerability conditions the severity of immediate impact and early adaptive response, but does not necessarily determine outcomes in the medium run. We interpret this as evidence that structural vulnerability amplifies acute shocks by reducing economic and fiscal buffers, while compensatory policies such as temporary employment regulation schemes and sectoral support, together with endogenous adjustments including the temporary reorientation towards domestic tourism and operational adaptations, can decouple pre-existing vulnerability from subsequent recovery trajectories.

This pattern contrasts partially with findings on urban economic resilience after 2008, where pre-crisis structural differences strongly predicted divergent trajectories [73]. The discrepancy may reflect the nature of the shock—COVID-19 as a time-bounded disruption to global demand versus the financial crisis as a prolonged credit and investment contraction—and the scale of public intervention, which was substantially larger during the pandemic.

A key methodological limitation is that our data aggregate entire cities, thereby obscuring intra-urban heterogeneity. It is plausible that highly touristified neighbourhoods experienced differentiated impacts that city-level indicators cannot capture. Spatially disaggregated analyses could examine whether concentrated touristification generates more persistent localised vulnerability.

5.5. The Role of Governance: A Mechanism Awaiting Empirical Validation

Throughout the manuscript, we refer to the moderating role of institutional capacity, yet this study does not empirically test that mechanism. Cross-sectional variability in CVI trajectories suggests that institutional factors matter—cities with similar levels of specialisation display divergent vulnerability paths; however, we do not operationalise or quantify governance capacity.

This limitation is non-trivial and requires transparency. Our claims regarding governance constitute logical inferences based on unexplained residual variation, rather than empirically validated findings. Future research should develop indices of institutional capacity (inter-sectoral coordination, regulatory effectiveness, planning quality) and incorporate TCI × governance interaction terms into panel models. A plausible hypothesis is that cities with stronger governance exhibit a flatter TCI → CVI slope, indicating that institutional capacity moderates (rather than eliminates) the paradox.

Constructing such indices presents challenges. Quantitative metrics (e.g., planning budgets, staff-to-population ratios) capture resources but not effectiveness. Qualitative indicators (e.g., existence of tourism management plans or zoning instruments) require systematic coding and may not reflect actual implementation. Mixed-method designs combining quantitative modelling with comparative case studies are likely to be the most productive route to elucidate this mechanism.

It is important to clarify the epistemological status of the governance argument. The moderating role of institutional capacity constitutes a theoretical inference grounded in cross-city variation in CVI trajectories, rather than an empirically validated finding. Cities with similar levels of tourism specialisation display divergent vulnerability paths—a pattern consistent with the moderating role of institutions—but this study does not operationalise, measure, or formally test governance capacity. This distinction is fundamental to the transparent interpretation of the paper’s contributions.

The core reason governance capacity is not empirically tested is the absence of longitudinal, city-level indicators of institutional effectiveness consistent across the 2014–2024 panel. Constructing a valid city-level governance index would require primary data collection (e.g., a Delphi study with local experts).

6. Conclusions

This study introduces and empirically validates the concept of urban competitive vulnerability, demonstrating that tourism success and structural fragility are interrelated rather than independent dimensions. Drawing on panel evidence from six Spanish cities over the period 2014–2024, the analysis provides systematic results that challenge conventional assumptions in the literature on urban competitiveness and tourism-led development.

6.1. Main Findings

The main results can be summarised in four empirical contributions.

First, we confirm the existence of a robust and persistent competitiveness–vulnerability paradox. Cities with higher levels of tourism competitiveness systematically exhibit greater structural vulnerability (β = 0.540, p < 0.001), controlling for urban heterogeneity, temporal trends, and economic development. This finding challenges the implicit assumption embedded in normative competitiveness models that treat visitor growth and revenue generation as sufficient indicators of urban sustainability. Competitive success can coexist with—and may even generate—fragilities that compromise the adaptive capacity of the urban system.

Second, we document that tourism specialisation operates as a dual determinant of vulnerability. The mediation analysis indicates that approximately two-thirds of the total effect, estimated at 65.8 per cent, is transmitted indirectly, as specialisation increases competitiveness and this, in turn, amplifies vulnerability. This suggests that urban fragility does not arise merely from tourism “monoculture”, but more fundamentally from the intensive growth model enabled by specialisation, which—absent effective regulation—exerts cumulative pressures on the urban system.

Third, we identify residential touristification as a specific and quantitatively substantial mechanism of vulnerability. The expansion of short-term holiday accommodation is associated with significant effects on housing prices (β = 0.288, p < 0.001) and eviction rates (β = 0.125, p < 0.05), consistent with processes of territorial exclusion. Holiday housing intensity shows a positive association with housing prices (β = 0.288, p < 0.001), indicating a substantial increase in residential market pressure in more touristified cities. This economically meaningful magnitude provides quantitative support to qualitative accounts of tourism-driven gentrification.

Finally, the concept is further corroborated through the analysis of the pandemic shock. Cities with higher pre-COVID fragility experienced sharper declines and slower recovery trajectories during 2020–2021. However, these differences attenuated substantially in 2022–2023, suggesting that vulnerability conditions the severity of the immediate impact but does not necessarily determine medium-term trajectories, potentially reflecting compensatory policies and endogenous adjustments.

6.2. Theoretical Contributions

This study makes three main conceptual and methodological contributions.

First, it proposes an integrative analytical framework that synthesises two strands of scholarship that have often evolved separately—urban competitiveness and territorial vulnerability—by conceptualising competitive vulnerability as an emergent property of urban systems intensively oriented towards tourism. This approach moves beyond the dichotomy between research focused on success factors and research focused on risk exposure, recognising their constitutive interdependence.

Second, it develops an empirical operationalisation through multidimensional composite indices (TCI and CVI) that capture specific mechanisms through which vulnerability is produced: structural dependence, territorial pressure, socio-economic fragility, and volatility. This operationalisation provides replicable methodological tools for comparative research and enables quantitative assessment of a phenomenon that has predominantly been explored through qualitative approaches.

Third, the study identifies and empirically documents specific causal mechanisms, distinguishing between direct effects of sectoral dependence and indirect effects mediated by competitive intensification, as well as territorial mechanisms associated with residential touristification. This causal decomposition moves beyond simplified narratives that attribute vulnerability exclusively to tourism growth per se.

6.3. Implications for Urban Policy

The findings have direct implications for urban planning and destination management in the following areas.

Regulating residential markets. The results justify targeted protective instruments, including caps on short-term rental licences, restrictive zoning in saturated areas, and change-of-use permitting requirements. Evidence from cities such as Amsterdam (a 30-day limit introduced in 2017) and Barcelona (regulatory tightening in 2021) suggests that regulation can moderate touristification pressures. However, aggregate effects on prices also depend on complementary social housing policies.

Economic diversification. The evidence underscores the importance of strategies to reduce structural dependence on tourism. Policies that strengthen complementary sectors (knowledge-based activities, creative industries, advanced services) can increase absorptive capacity in the face of sectoral shocks. However, such strategies may encounter tensions with urban land and housing market dynamics, where high land prices can displace activities with lower willingness to pay.

Monitoring systems. The findings support the establishment of urban observatories that integrate competitiveness, territorial pressure, social cohesion, and adaptive capacity, complementing conventional monitoring systems focused primarily on activity volumes. The proposed CVI offers an adaptable methodological tool to identify early accumulation of structural fragility across diverse urban contexts.

Collaborative governance. The evidence suggests that competitive vulnerability is not an inevitable outcome of tourism growth, but is critically conditioned by institutional capacity to manage systemic pressures and distributive conflicts. Participatory processes that build social legitimacy for management strategies and foster consensus on growth limits constitute critical institutional resources. Nevertheless, this study does not operationalise or empirically test governance mechanisms, making this a priority area for future research.

6.4. Limitations and Future Research Directions

This study has several limitations that condition interpretation and point to future research opportunities.

First, the geographical restriction to Spain limits generalisability. Although Spain constitutes a particularly informative setting—one of the world’s leading tourism destinations, diverse touristification experiences, and high-quality data—mechanisms may operate differently under other institutional and cultural conditions. Extending the analysis to Mediterranean (Athens, Rome, Dubrovnik), Asian (Bangkok, Kyoto), Latin American (Cartagena, Cusco), and North American (New Orleans, San Francisco) cities would help assess whether the paradox is universal or context-dependent.

Second, the observational design constrains causal inference. Although fixed-effects models control for time-invariant unobserved heterogeneity, robustness across specifications increases confidence, but residual endogeneity cannot be entirely ruled out. Natural experiments exploiting regulatory reforms (e.g., introduction of short-term rental quotas) or regression discontinuity designs (e.g., administrative thresholds related to city size) could provide cleaner identification.

Third, annual temporal aggregation obscures seasonality—a critical dimension of tourism vulnerability. Monthly data would allow assessment of whether vulnerability concentrates around seasonal peaks or persists. Likewise, city-level aggregation conceals intra-urban heterogeneity documented in neighbourhood gentrification research. Analyses using spatially disaggregated units (districts or census tracts) would enrich understanding of localised dynamics.

Fourth, the weighting of composite indices, while theoretically grounded, includes an element of arbitrariness. Systematic sensitivity analysis, varying weights, could assess robustness. Alternatively, dimensionality-reduction techniques (e.g., principal component analysis) could generate empirically derived indices to complement our deductive approach.

Finally, the period analysed (2014–2024) captures the intensification phase and the pandemic shock, but longer time series would be necessary to examine full growth–decline–recovery cycles. The absence of pre-2014 data limits our ability to assess whether post-platform intensification represents a qualitative regime shift or a continuation of previous trends.

6.5. Future Research Agenda

Extending the analysis to Mediterranean (Athens, Rome, Dubrovnik), Asian (Bangkok, Kyoto), Latin American (Cartagena, Cusco), and North American (New Orleans, San Francisco) contexts would enable evaluation of the extent to which the competitiveness–vulnerability paradox operates transversally or is mediated by institutional and cultural conditions.

Developing research designs with stronger causal identification—such as natural experiments exploiting regulatory reforms, regression discontinuity designs, or instrumental variable strategies—would strengthen causal claims beyond the limitations of observational panel data. Complementarily, flexible specifications such as splines and regime-switching models could help identify critical thresholds at which urban pressures become qualitatively different, thereby testing the hypothesis of non-linear acceleration in vulnerability beyond certain levels of touristification.

Operationalising and empirically testing institutional capacity as a moderator of the competitiveness–vulnerability relationship constitutes a high-priority research avenue with direct applied relevance. Constructing multidimensional governance indices and incorporating them via interaction terms in panel models would clarify the extent to which institutional capacity can modulate the identified paradox while addressing the methodological challenges inherent in measuring complex institutional constructs.

A further avenue, formalised in this study as Hypothesis H5 (exploratory), concerns the integration of mixed-method research designs that combine quantitative panel analyses with qualitative inquiry. Urban ethnographies, semi-structured interviews, and process tracing are particularly well placed to illuminate the social mechanisms through which competitive vulnerability is produced, experienced, and contested in specific urban contexts—dimensions that aggregate panel data cannot fully capture. Whilst H5 is not empirically tested here due to the scope and data constraints of the present study, its inclusion as a formal exploratory hypothesis reflects the conviction that the quantitative findings reported above constitute a necessary but insufficient basis for understanding the full complexity of the competitiveness–vulnerability paradox. Mixed-method designs that integrate the structural patterns identified through panel econometrics with the lived experiences of residents, local authorities, and tourism actors represent the most productive methodological frontier for advancing this research agenda.

6.6. Interpretative Synthesis

Overall, the findings suggest that urban competitive vulnerability is a systemic phenomenon rather than an occasional pathology of poorly managed cases. The TCI–CVI relationship operates consistently across the full range of cases analysed, controlling for alternative factors. This implies that the paradox emerges from the structural logic of intensive urban tourism development, whereby the very dynamics that generate competitive advantage through specialisation, agglomeration economies and operational efficiencies also exert pressure on urban systems when effective containment mechanisms are absent.

International comparisons reinforce this interpretation, as geographically and culturally distant cities exhibit convergent patterns of residential touristification and displacement. Variation in intensity reflects differences in regulatory frameworks and institutional capacity, suggesting that vulnerability is not inevitable but requires active governance. The cases of Amsterdam and Barcelona illustrate that regulation can mitigate—though not eliminate—pressures, yet effective implementation encounters political and economic resistance.

The COVID-19 shock exposed latent fragilities while also demonstrating that structural vulnerability does not fully determine future trajectories, as compensatory policies and endogenous adaptation can modify outcomes. This points to a degree of political agency, yet it also underscores the costs of allowing vulnerability to accumulate, given that effective responses required massive public intervention whose fiscal sustainability remains questionable in the context of recurrent shocks.

The emerging research agenda should prioritise: (1) identification of thresholds and non-linearities in the competitiveness–vulnerability relationship; (2) operationalisation and empirical testing of institutional capacity as a moderating mechanism; (3) geographical extension to assess the universality of mechanisms; (4) analysis of intra-urban heterogeneity using spatially disaggregated data; and (5) longer longitudinal studies capturing complete tourism cycles. Methodologically, combining quantitative panel analyses with comparative qualitative research would help capture both aggregate patterns and the social processes through which vulnerability is produced and experienced.

6.7. Final Reflection

The competitiveness–vulnerability paradox documented in this study raises fundamental questions regarding urban development models oriented towards intensive tourism growth. The evidence suggests that competitive success, in the absence of effective mechanisms for territorial regulation, economic diversification, and the protection of social cohesion, may generate development trajectories that are paradoxically fragile.

This observation does not imply an anti-tourism stance, but rather highlights the need to reconceptualise urban success beyond conventional performance indicators. Such a reconceptualisation requires the explicit integration of structural sustainability, adaptive capacity, and social cohesion into the evaluation of urban development pathways.

The central challenge lies in identifying institutional configurations that capture the economic benefits of tourism while simultaneously mitigating systemic pressures. The findings suggest that such configurations are indeed possible. However, they depend on political willingness to implement effective regulation, institutional capacity to coordinate sectoral policies, and collaborative governance processes that confer social legitimacy upon management strategies.

The tension between competitive growth and structural sustainability cannot be resolved through unilateral technical solutions. Instead, it requires deliberative processes that build collective agreement around acceptable limits to tourism growth and socially negotiated redistributive priorities. Ultimately, overcoming the competitiveness–vulnerability paradox depends on the capacity of urban societies to envision and construct development models that prioritise collective well-being and adaptive capacity over the maximisation of short-term economic growth.