Learning about monetary regime shifts in an overlapping wage contract model (Q2366871)

This paper uses stochastic simulations to explore the finite-sample effect of learning during the transition to convergence in an overlapping wage contract model. In this model, the reaction function of the monetary authority is defined as a function of the deviations of inflation and unemployment from their target values. Agents to not know the true parameters of the monetary authority's reaction function, although they know the model structure and parameters. With each new observation of the economic variables, agents update their estimates of the reaction function and resolve the model using rational expectations. The simulations here allow for several classes of regime shifts in the reaction function, including one-shot, slowly phase-in, quickly phase-in and multiple regime shifts. Agents use ``unadorned'' RLS (recursive least square), window RLS and testing RLS to estimate the unknown reaction- function parameters. The main findings from the simulation results are as follows: (1) Learning affects the time-series behavior of output. (2) Policy is most effective (non-neutral) during the transition period when agents are learning the reaction function parameters, that is, when they least closely approximate rational expectations. (3) Transition periods are sufficiently long: They are longer or at least as long as peak-to-trough reference cycles or presidential election cycles.