Luxury tourism and purchase intention factors: a mixed approach applied to luxury goods

The study addresses how the COVID-19 pandemic has altered luxury tourism consumers’ purchasing behavior and what factors now drive purchase intention for luxury goods in a tourism context. It responds to managerial needs to understand complex, evolving consumer preferences that increasingly emphasize authentic, experiential, and socially-influenced luxury consumption. Two main literature gaps are identified: (1) the need to re-examine luxury purchase drivers after major social changes due to COVID-19; and (2) the lack of integrated analysis that jointly considers buyer vanity, CSR, and environmental/social influence factors. The research aims to develop and empirically test a model (combining PLS-SEM and fsQCA) to determine which determinants significantly affect purchase intention, and to identify configurations of determinants that lead to intention. Hypotheses test positive effects on purchase intention from perceived value (H1), WOM (H2), vanity (H3), price image (H4), CSR (H5), and social influence (H6).

Prior research shows luxury consumption in tourism shifting from product-centric attributes (exclusivity, status, quality) toward experiential and psychosocial elements. Early studies emphasized exclusivity, status signaling, and social influence (e.g., Park et al., 2010; Hung et al., 2011; Chan et al., 2014). Subsequent work increasingly highlights experiential/hedonic value (Yang & Mattila, 2016; Chen & Peng, 2018), quality and price (Zaidan, 2016), cultural/contextual differences (Correia et al., 2018; Han et al., 2018; Pham et al., 2020), and psychosocial processes including self-presentation and sustainability orientations (Li et al., 2021; Brandão & Cupertino de Miranda, 2022; Lai et al., 2022). The review suggests diminished focus on product exclusivity and a shift toward experiences that are harder to imitate, with social mechanisms (WOM, social influence) gaining prominence. The proposed model builds on luxury consumer culture theory and integrates UTAUT2 with constructs salient to luxury (price, social influence) and additional variables (WOM, vanity, CSR, perceived value) noted in the literature as relevant to purchase intention.

Design: Mixed-method approach combining PLS-SEM and fuzzy-set Qualitative Comparative Analysis (fsQCA) to capture both symmetric and configurational effects. Sampling and data collection: An online survey was administered via intercept to foreign tourists leaving luxury brand stores in luxury shopping areas of Mallorca, Ibiza, Barcelona, and Madrid (Spain). Inclusion criteria: foreign tourists over 18, repeat luxury goods buyers (≥2 times), intending to continue purchasing luxury goods during travel to Spain, and targeting brands listed in Brand Finance 2020 Luxury Brands Ranking. Data collection occurred April–June 2021. Convenience sampling was used, with control questions to verify eligibility. Total responses: 530; valid sample: 522. Measures: 7-point Likert scales (1=strongly disagree; 7=strongly agree). Constructs and sources: Perceived Value (PV; 3 items, Joung et al., 2016), Vanity (VAN; 4 items, Soh et al., 2017), Behavioral Intention (BI; 5 items, Soh et al., 2017), Price value (PRICE; 2 items, Shukla et al., 2015), CSR (2 items, Ramasamy et al., 2013), Social Influence (SI; 3 items, Venkatesh et al., 2012), WOM (4 items, San-Martín et al., 2015). Two additional questions assessed pandemic influence on luxury purchases and preference for online vs. in-person purchasing. PLS-SEM analysis: Conducted in SmartPLS 3 with bootstrapping (10,000). Measurement model assessed via loadings (>0.7), reliability (Cronbach’s alpha, composite reliability >0.7), convergent validity (AVE >0.5), discriminant validity (HTMT <0.85–0.90). Model fit via SRMR. CMB tested with a latent CMB variable; all VIF <3.3 indicated no common method bias. Endogeneity tested via Gaussian copulas; none significant (p>0.05). fsQCA: Followed Ragin’s three steps: calibration, truth table construction, and logical minimization. Constructs calibrated from 7-point Likert to fuzzy membership using percentile thresholds (90th=full membership, 50th=crossover, 10th=full non-membership). Two models examined: Model I: BI=f(PV, VAN, PRICE, CSR, WOM, SI); Model II: -BI ~ f(PV, VAN, PRICE, CSR, WOM, SI). Necessary condition analysis (consistency >0.9) and sufficiency analysis (solutions’ consistency and coverage) were conducted. Predictive validity assessed by splitting the sample into a subsample (to derive fsQCA solutions) and a holdout sample (to test solution consistency/coverage).

PLS-SEM measurement model: All item loadings >0.7; reliability and validity thresholds met. SRMR=0.047. Path coefficients to Behavioral Intention (BI) with p-values (one-tailed, 10,000 bootstraps):

• WOM → BI: 0.392, p<0.001 (significant)
• Price value → BI: 0.239, p<0.001 (significant)
• Vanity → BI: 0.132, p<0.001 (significant)
• Social influence → BI: 0.130, p=0.014 (significant)
• Perceived value → BI: 0.062, p=0.101 (ns)
• CSR → BI: 0.001, p=0.986 (ns) Explained variance: R2(BI)=0.549 (adjusted 0.543). Largest contributions to explained variance noted for WOM (~26.97%) and Price (~14.05%). fsQCA necessary conditions for BI: Perceived value (PVc) consistency=0.914; Vanity (VANC) consistency=0.926 (both necessary). Other conditions did not meet necessity threshold. fsQCA sufficiency (presence of BI): Seven solution configurations; overall solution consistency≈0.698 and coverage=0.768. Social Influence (SI) appears across all solutions as a core component, with WOM and PRICE frequently present; configurations demonstrate equifinality with various combinations (including presence/absence of CSR and PRICE) leading to BI. fsQCA sufficiency (absence of BI, -BI): Eleven solutions; overall solution consistency=0.944 and coverage=0.891. Predictive validity: Subsample-derived solution models showed high consistency (>0.80) and good coverage when tested on the holdout sample. Contextual findings: 90% reported not buying more luxury goods during the pandemic; 10% did. Purchase channel preference: 61% in-person vs. 39% online.

The findings indicate that, post-pandemic, purchase intention for luxury goods among tourists is significantly driven by WOM, price image, vanity, and social influence. CSR and perceived value do not directly predict intention in the symmetric (PLS) model. However, fsQCA reveals perceived value and vanity as necessary conditions, highlighting that while these factors may not individually exert strong direct effects, they are essential components within causal configurations that lead to intention. Social influence emerges as a core condition present across all sufficient solutions, underscoring the heightened role of social context and network effects in luxury purchasing. The insignificant direct CSR effect suggests an ongoing paradox between luxury and sustainability in consumers’ minds; nevertheless, CSR appears in some successful configurations, implying that its impact may depend on how it interacts with WOM and social influence and how effectively firms communicate CSR initiatives. The results support a shift from product exclusivity toward experiential and socially validated luxury consumption, with multiple viable strategic paths (equifinality) to stimulate purchase intention.

This study contributes an updated, post-pandemic perspective on determinants of luxury purchase intention in tourism by integrating PLS-SEM and fsQCA. Direct drivers (PLS) include WOM, price image, vanity, and social influence, while perceived value and CSR show no direct effects. Configurationally (fsQCA), perceived value and vanity are necessary conditions and, with social influence and WOM, form multiple sufficient paths to intention. The mixed-method approach exposes complexity and equifinality in consumer decision-making, offering managers alternative strategic combinations tailored to resources and contexts. Future research should further probe the luxury-CSR tension, refine conceptualizations of perceived value in experiential contexts, and test cross-cultural and longitudinal generalizability.

Key limitations include: (1) the subjective, context-dependent nature of luxury, mitigated but not eliminated by providing brand rankings as reference; (2) use of convenience sampling in Spanish luxury shopping locations and focus on foreign tourists, which may limit generalizability; and (3) cross-sectional design, which precludes causal inference over time. Additionally, recruiting confirmed luxury consumers required stringent screening, possibly introducing selection bias.