ABSTRACT
This article reviews recent conceptual debates on cross-national and cross-sectoral policy-learning in the political science literature. It proceeds from the argument that the existing literature is characterised by the absence of a comparative assessment of the risks and potentials of different strategies of policy learning. This sin of omission does not only have significant implications for the study of policy learning but also for its practice. The authors use the normative concept of improvement-oriented learning to assess the risks and potentials of three learning strategies: imitation, Bayesian updating and deliberation. They observe that the distribution of risks and potentials is most advantageous in deliberative learning strategies, but that imitation is the most risky learning strategy, and Bayesian updating ranges somewhere in-between.
