Public accountability mechanisms are among the most important means with which governments seek to guard and improve the performance of public organizations. In recent years, more and more resources are invested in accountability. Unfortunately, however, this has not lead to noticeable increases in performance. On the contrary, research documents a plethora of accountability-failures. Accountability is often time-consuming and expensive, provokes strategic reactions by managers and professionals, and may harm their motivation. The key issue is: how can public accountability become more effective? This paper seeks to answer this question by connecting two largely separated strands of research: public administration research on real-world organizations and experimental research on the effects of different forms of accountability on individual decision-making. Important and relevant findings from experimental research will be used to analyze the effects of accountability. Experimental research has mapped out how different attributes of accountability are likely to have various – positive and negative – effects on individual decisions. These insights are of acute relevance to our study of public organizations, although they need to be translated to real-world settings. The paper develops the Calibrated Public Accountability-model (CPAmodel) from experimental research findings. The CPAmodel describes how six important attributes of accountability are likely to affect decision-making. One of these attributes is for instance whether the accountability ‘forum’ focuses on processes or results, which has profound effects on agents. In studying the effects of accountability on public organizations, a key factor is the enormous variety in operational tasks. Studies suggest that public sector accountability is currently not attuned to task specificities. This makes the quest for the calibration of accountability mechanisms to tasks, in order to realize proportionally appropriate accountability settings, highly relevant. The applicability of the CPAmodel will then be demonstrated in an analysis of two dissimilar forms of accountability.
