Literature often conceives policy learning as a force for common good, a journey towards continuous improvement aiming to create public value through improving the design of public policies. While research occasionally highlights that learning mishaps can have adverse outcomes, the dark side of policy learning and its contribution to public value destruction through subpar policy design remains remarkably under-explored and under-theorized. In this article, this gap is addressed through a learning failures lens. Drawing on policy learning and public value theories, two types of policy learning failures are conceptualized and their contribution to public value destruction through policy design is explored: misdirected learning design failures (non-deliberate and cybernetic), and deontological failures (deliberate and normative). These learning failures are highly influenced by several aspects pertinent to institutional and administrative arrangements, democratic traditions, and practices, as well as features of policy problems. This paper explores these failures across two empirical vignettes drawing on cases of policy design from Belgium and Hungary. In doing so, this article makes three main contributions. First, it offers a more robust a novel understanding of public value destruction, by positioning learning failures as a potent yet hitherto understudied value destruction mechanism. This breaks path with rather normative perspectives of learning, predominant in literature. Second, it also offers a novel perspective that traces the roots of policy design failures to policy learning, as a fundamental policy process behaviour (i.e., an ontology of policymaking). Third, it offers practitioners insights into how to mitigate the destructive potential of learning failures for policy design processes.

Policy narratives have long been a central topic in public policy research (Stone, 2012). In the last decade, the Narrative Policy Framework (NPF) as a specific theoretical framework dedicated to the analysis of policy narratives has emerged and found great scholarly attention (Jones et al., 2023a; Shanahan et al., 2011). The NPF takes a structured approach, by identifying and analyzing distinct narrative elements, including the setting of a narrative, its acting characters, the plot, and the narrative’s moral. While these foundational elements of policy narratives as well as narrative strategies have been explored widely and for diverse empirical contexts, the connection between narratives and policy learning as well the role of power in these contexts remains less studied.

This paper explores the connection between policy narratives and policy learning (as a specific type of policy change). We adopt a procedural perspective on narratives and learning by analyzing the “variation in policy narrative elements” that occurs during learning processes. Furthermore, we distinguish between narratives that are used in the context of policy- as compared to power-oriented learning processes – therefore posing the research question: How do policy narrative elements vary within policy- and power-oriented learning processes?

To answer this research question, a case study is conducted on the European Taxonomy for Sustainable Activities. It was introduced in June 2020 as a first classification system of economic activities worldwide, and extended by the Complementary Climate Delegated Act in 2022, labeling natural gas and nuclear energy as “transitional sustainable activities”. We argue that in this case, we can observe policy learning moving from policy- to power-oriented learning. In this learning process, we investigate the variation of policy narrative elements, using a dataset of 40 policy documents including parliamentary speeches from members of the European Parliament (EP) and press releases from political groups in the EP.

Regulatory impact assessment is a widely adopted instrument in evidence-informed policymaking, challenged by the complex task of combining expert knowledge with input from stakeholders and citizens. Social Multi-Criteria Evaluation (SMCE) is a methodological framework for designing and comparing policy alternatives in a consistent and transparent way, allowing for the integration of a plurality of technical impact dimensions and social perspectives. We systematically reviewed 20 years of literature on the empirical applications of SMCE since its formalisation in 2004. A total of 54 papers reporting on various case studies have been identified and analysed with the aim of drawing lessons for the use of SMCE in public policy design. The findings demonstrate the great flexibility of the approach: the studies represent a vast diversity of policy fields and geographical contexts, spanning 20 countries across four continents. The results indicate that SMCE allows for combinations with other approaches and adapts well to real-world settings, including variations in data collection methods and the degree of social actors' involvement. The authors recurrently mention the particular usefulness of SMCE in addressing conflictual situations by helping to expose different problem framings and fostering mutual learning, as well as dealing with missing, incomplete, or uncertain data. The reported challenges include the significant resource investment and the need to mediate power relations among the involved actors. Overall, the analysis suggests a promising potential for the SMCE to strengthen democratic nature of policymaking processes by encouraging structured public involvement, while maintaining a balance with expert assessment, in order to identify the most acceptable and effective compromise solution.