Introduction Conventional governance theory predicts that corruption deteriorates service quality and reduces citizen satisfaction; however, emerging evidence documents a counterintuitive pattern in which bribery exposure is associated with higher evaluations. This study examines this paradox in Peru, testing whether three adaptive mechanisms (expectation adjustment, cultural normalization, and pragmatic access) explain the bribery-satisfaction relationship across eight subnational territories with differentiated levels of institutional trust. Methods We analyzed 171,399 survey observations from five annual waves of Peru’s National Household Survey (ENAHO, 2019-2023). Multilevel mixed-effects regression models estimated the association between bribery exposure and citizen satisfaction with public services (health, education, and security), incorporating institutional trust as a moderating variable and controlling for sociodemographic and territorial characteristics. Results A significant positive association emerged between bribery exposure and satisfaction (β = 0.032, p 0.001), contrary to theoretical predictions. Substantial territorial heterogeneity was documented: four domains exhibited significant positive associations, three showed non-significant positive associations, and one displayed a non-significant negative association. Institutional trust significantly moderated this relationship (β = –0.038, p 0.001), with low-trust citizens displaying the strongest positive associations. The positive association intensified during the 2020-2021 crisis (52.4% increase) before weakening during recovery. Discussion These findings demonstrate that citizen adaptation to bribery is territorially contingent rather than uniform, consistent with adaptive mechanism explanations. Results challenge the assumption that anti-corruption interventions automatically improve citizen evaluations, underscoring the need for differentiated subnational policy strategies that address the cognitive normalization mechanisms underlying this paradox.
Gonzales et al. (Thu,) studied this question.