The Respective Effects of Knowledge and Non-Knowledge on Policy Learning in the European Liberalisation Policy Process of Belgian Network Industries
Policy actors involved in decision-making processes interact and gradually accumulate evidence about policy problems and solutions. As a result, they update their policy beliefs and preferences over time. Existing studies adopting such a “learning” approach (Dunlop & Radaelli, 2013, 2018; Moyson & Scholten, 2018) to knowledge use in policy processes provide mixed evidence. While some suggest that policy actors do acquire and use knowledge (see Riche et al., 2017), many others are more skeptical and argue that policy learning is very unlikely, because of its numerous obstacles (e.g., Birkland, 2006; Leach et al., 2014; Moynihan, 2008; Fischer, 2016). Among these obstacles, “non-knowledge” such as emotions (Dunlop & Radaelli, 2017) or cultural values (e.g., Jenkins-Smith et al., 2014) is often pinpointed. Yet, there is no study providing an empirical assessment: what are the respective effects of knowledge and non-knowledge on policy learning?
To address this question, I adopt the conceptualization of belief systems in the advocacy coalition framework (ACF: Sabatier & Jenkins-Smith, 1993; Jenkins-Smith et al., 2018), in which policy actors’ preferences are modelled by factual beliefs (i.e., knowledge) and normative beliefs (i.e., non-knowledge) at a “deep core” level (i.e., applicable to all policy subsystems) and a “policy-core” level (i.e., specific to one policy subsystem). The paper relies on a web survey conducted among 413 policy actors (from 38 organizations) involved in the European liberalization policy process of two Belgian network industries: the railways and the electricity sector. Policy learning is measured with a “simple gain scores” approach (Moyson, 2018) and related with several sets of questions operationalizing the factual beliefs (or “knowledge”) and normative beliefs (or “non-knowledge”) of policy actors, based on multilevel regression analyses (to account for the effect of organizations).
The findings suggest that factual and normative beliefs explain the nature of policy actors’ preferences more than their evolution (i.e., learning: hypothesis 1). Second, deep core beliefs are more influential than policy core beliefs (hypothesis 2). While these findings are very consistent with the ACF, the surprise lies in a detailed analysis of the effect of each factual and normative belief on policy preferences and policy learning: the relations are not uniform, which is not expected by the ACF. For example, while some beliefs do not influence policy preferences as such, they can be a factor of their evolution, and the other way around. Last but not least, normative beliefs are more influential than factual beliefs (hypothesis 3). While this gives credit to a skeptical view on knowledge use, again, a refined analysis of the effect of each belief brings a more nuanced answer to the research question.
Overall, this paper concludes that the respective role of knowledge and non-knowledge in policy learning and policy processes is less a matter of quantity (how much knowledge and non-knowledge?) than of quality (what are the influential beliefs, exactly?). In doing so, From an ACF perspective, the paper also provides a better understanding of the structure of belief systems and of their role in policy learning. I conclude with some implications.