The exact linear dependency among age, period and birth cohort makes it impossible to recover the true parameters of Age-Period-Cohort (APC) models. We then propose to extract reliable information from APC models via the Shapley decomposition, a model-agnostic procedure from game theory that...