ABSTRACT
Building on theories of path dependence and policy feedback, this paper examines whether pension reform exhibits Reform Memory, the tendency for recent reform history to influence the probability of subsequent policy change. While existing research highlights the importance of historical legacies in shaping public policy, little systematic evidence exists on whether pension reforms themselves exhibit a measurable and predictable temporal structure. Using a newly constructed global database of 790 pension reform events across 183 countries from 2000 to 2024, we analyse the dynamics of major structural and parametric pension reforms. Employing rare-events logit models and a range of robustness exercises, we find strong evidence of a highly non-linear form of reform memory in pension policymaking. Countries that experienced a major reform within the previous 2 years exhibit a major reform probability of 10.1%, compared with only 1.0% among countries whose previous reform occurred more than a decade earlier. This reform-memory effect remains robust across alternative reform definitions, reform-intensity measures and regional subsamples. By contrast, conventional determinants, including population ageing, public debt, inflation and economic growth, display weak or inconsistent explanatory power. The findings indicate that pension reforms do not occur as isolated responses to contemporary pressures. Instead, reform activity exhibits strong temporal dependence, clustering within short-lived periods of elevated policy activity before reverting to longer periods of institutional stability. More broadly, the paper provides systematic cross-national evidence that pension systems evolve through reform cycles rather than a sequence of independent legislative events.