Tourism demand responses to geopolitical risk: historical validation, dynamic associations, and cross-country heterogeneity in Europe

Introduction The sign of the tourism response to geopolitical risk is well established, but its shape is not, since depth, duration, reversibility, and cross-destination variation remain unsettled. This study estimates the monthly path of foreign overnight stays around country-linked geopolitical-risk episodes for 12 European destinations, from 2012 to 2024. Methods Severe months are identified by a threshold set at the 97.5th percentile of each destination’s own risk history up to the preceding month, a past-only rule that avoids the future-information leakage of full-sample definitions, and a stratified sample of flagged episodes is matched to the historical record through a frozen documentary framework. A panel distributed lead–lag model with destination and month-year fixed effects and few-cluster inference traces the response. Results A severe-risk month is associated with a contemporaneous decline of 0.115 log points, approximately 11%, distinguishable from zero under an exact single-coefficient wild-cluster-bootstrap test, and a cumulative effect of −0.345 log-point-months over the first half-year. The path is non-monotonic, attenuating around the fifth month before turning negative again, and the simultaneous band over the lead–lag family isolates no individual horizon, so no sustained recovery is established. Magnitude and persistence depend on the severity definition (long historical benchmark, short adaptive benchmark, or episode onset) and on destination-specific seasonality. Responses vary sharply across destinations. Market diversity and government effectiveness enter only as exploratory correlates and do not support a protective-buffer reading. Discussion The study contributes a leakage-free threshold, a documentary validation of the risk measure, and a disciplined separation of primary from exploratory evidence.