Effort–Reward Imbalance and Turnover Intention Among Luxury Hotel Frontline Employees: The Moderating Role of Perceived External Employability

Employee retention in luxury hotels may depend on both perceived employment reciprocity and outside job opportunities. This study examined the association between effort–reward imbalance (ERI) and turnover intention and the moderating role of perceived external employability among 303 frontline employees in six South Korean five-star hotels. Nine retained questionnaire items were assessed using principal axis factoring and parallel analysis, the heterotrait–monotrait ratio of correlations, the Fornell–Larcker criterion, and a specification curve across every admissible item subset. Mean-centered composite scores were analyzed using ordinary least squares regression with demographic controls, with HC3 standard errors used for sensitivity assessment. ERI was positively associated with turnover intention at mean employability (B = 0.893, p < 0.001). The interaction was positive but small (B = 0.0303; ΔR2 = 0.0011), with conventional p = 0.0046 and HC3 p = 0.0475 in the controlled model; HC3 p increased to 0.0619 without controls. A dominant common factor explained 66.86% of standardized item variance, and the ERI–turnover intention correlation was 0.974. Discriminant validity was not supported: heterotrait–monotrait ratios were 1.153 for ERI and turnover intention, 1.109 for ERI and employability, and 1.076 for employability and turnover intention, and the squared ERI–turnover intention correlation (0.949) exceeded the average variance extracted of both composites. Repeating the analysis across all 20 admissible item combinations produced no specification with an acceptable ratio (minimum 0.938). The hypotheses are therefore not described as supported, because they presuppose that the composites measure separable constructs and these data do not show that. The estimates describe the association among recorded composite scores. Independently validated measures and longitudinal replication are required before the proposed boundary condition can be tested.