Abstract
Key performance indicator (KPI) systems translate international ranking methodologies into institutional management practice, but the performance-measurement literature has long warned that indicators used for high-stakes decisions induce behavioural responses that can corrupt the very processes they monitor. This paper isolates the behavioural risk category from a five-category economic risk framework developed for ranking-oriented university governance, and asks how a KPI architecture can be designed to resist metric gaming from the outset rather than police it after the fact. Drawing on the design of a three-level (university-faculty-department) KPI system aligned with the Times Higher Education (THE) World University Rankings methodology, developed for New Uzbekistan University and adopted in the practice of the National Quality Assurance Agency for Education (implementation certificate No. 03/18-144 of 16 December 2025), the paper identifies four design features with gaming-resistant properties: the pairing of outcome KPIs with process KPIs at every level; quality-weighted rather than count-based publication indicators; differentiated monitoring frequencies matched to indicator dynamics; and transparent cascading of indicators to responsible units. The behavioural risk category is argued to be the most underestimated yet most consequential threat in ranking strategy implementation: unlike macroeconomic or technological risks, it is generated by the management system itself. The paper contributes a design-stage treatment of a risk that the rankings literature typically discusses only as an observed pathology.