Adaptive Learning in Sticky-Information Economies

Resumo/Abstract:

In this presentation, one develops a macroeconomic model in which the formation of expectations plays a central role and where the behavior of the agents is characterized by the presence of rigidity on information acquisition and processing (under this assumption, expectations about current events depend on information obtained in the past). Two possibilities relating the formation of expectations in contexts of information stickiness are explored: in a first moment, we consider that expectations are rational and, afterwards,
one takes into consideration adaptive learning, according to which the agents in the economy behave as econometricians (they collect information about the past evolution of the variables and they estimate the value of the parameters that are considered relevant).
Under rational expectations / perfect foresight, the model of price formation with information stickiness presents long-run conventional results: there is a fixed-point and this is stable; independently of the initial value of the in?ation rate, or of any other relevant macroeconomic variable, there will be convergence towards the steady-state value that the variable will asymptotically assume. The learning assumption introduces an enormous variety of eventual long-term outcomes, namely it emerges the possibility of endogenous fluctuations; for some parameter combinations, the found cycles have a low periodicity, while under other parameter choices it will not be possible to identify any periodicity. The absence of periodicity is associated with the existence of chaos.
The nonlinear results obtained under learning have implications at the level of economic policy: in the assumed model, a stable growth of the price level is not guaranteed; the inflation rate can fluctuate endlessly, in a more or less volatile way, around the long-run steady-state without ever converging to the equilibrium value.

Outros seminários / Other seminars: Programa completo / Full programme