Tourism, institutions and the environment: global evidence on rural sustainability and cross-regional heterogeneity

Purpose This study re-examines the tourism–environment relationship as context-dependent rather than universally portable, testing nonlinear tourism intensity, conditioning by the democratic institutional context and cross-regional heterogeneity. Design/methodology/approach Using 2,780 country-year observations from 133 countries (2000–2020), two-way fixed-effects models estimate tourism–emissions curvature and its conditioning by V-Dem electoral democracy. Region-specific models with support, influence and small-cluster checks test coefficient heterogeneity; rural typologies, validated k-means clusters and an entropy-weighted Sustainable Rural Tourism Index (SRTI) describe profile heterogeneity. Findings Global coefficient signs resemble an inverted-U, but the quadratic term is insignificant. Regional tourism coefficients differ significantly, yet no regional inverted-U survives the support and sensitivity checks: apparent turning points rest on single high-intensity island or city economies. The tourism × electoral-democracy interaction is only suggestive (p = 0.090). Country profiles differ, but the tourism-intensive profile's ranking reverses when tourism is not scored as an asset. Practical implications Tourism firms and destination managers should not treat destination maturity or broad democratic improvement as substitutes for operational decarbonisation. Energy, transport, supplier and destination-level actions should be adapted to regional and structural profiles, used as guidance rather than proof of strategy effectiveness. Social implications Rural sustainability depends on protecting shared environmental assets and aligning tourism expansion with stakeholder participation, local resource systems and enforceable environmental governance. Originality/value The study reframes established tourism–environment findings as boundary-condition propositions rather than as a universal tourism–Environmental Kuznets Curve. It integrates regional curvature, democratic-context conditioning and rural segmentation while separating fixed-effects inference from descriptive SRTI diagnostics.