This paper argues that the recent reversals in pension privatization hold interesting lessons for the research on international policy diffusion. First, diffusion research needs to account more explicitly for countries breaking away from the international trend, and to model social dynamics in more general. Second, the case of pension reforms shows how different forms of learning coexist and rather reinforce each other than constitute competing paradigms of policy diffusion. Instead, third, diffusion is often just one part of a larger story: the existence of policy cycles especially in structurally weak societies. To corroborate these three claims I give qualitative and quantitative evidence on the rise and current dip in the privatization of public pensions.