Abstract
We develop a non-equilibrium statistical mechanics framework for inflation in which macroscopic dynamics emerge from interacting heterogeneous agents operating off-equilibrium. Starting from stochastic Langevin dynamics for micro-level inflation and adaptive expectations, we derive a non-linear Fokker–Planck equation governing the joint evolution of inflation and beliefs, and obtain an effective coarse-grained Ising representation of binary pricing decisions. Within this unified kinetic-to-spin framework, inflation is characterized by two macro-order parameters: the mean expectation level, capturing anchoring to nominal targets, and the cross-sectional dispersion, capturing the degree of collective synchronization. The resulting system exhibits a well-defined phase structure controlled by interaction strength and noise, with disordered, anchored-but-heterogeneous, and synchronized regimes emerging as stationary states of the underlying dynamics. Phase transitions between these regimes display characteristics of non-equilibrium critical phenomena, including diverging susceptibility, critical slowing down, and bifurcation-driven regime shifts. Empirical evidence from micro-level survey data supports the existence of structured coordination in inflation expectations consistent with the predicted phase organization. Overall, inflation is interpreted as an emergent ordering phenomenon in an interacting many-body system.