Many public organizations employ technologies of scrutiny such as peer review or quality assurance to improve their performance and decision-making. Such technologies may affect performance and decision-making directly, through scrutiny, and indirectly, through behavioural responses by agents within the organization. We examine one such technology in a large public sector organization in the UK. By comparing the distribution of project sizes before and after the introduction of a system of assurance implemented through a simple decision-rule, we document substantial manipulation by agents designed to avoid scrutiny. Furthermore, there is no evidence that project quality as measured by over- or under-spending and fidelity to the planned completion date is higher among reviewed projects, despite this observed manipulation. Our results suggest that organisations considering such a technology need to investigate both the naïve effect of the technology, and how agents will respond to its existence, setting a new organisational equilibrium.
