Drawing on the empirical evidence that heterogeneity in inflation expectations is persistent and endogenously time-varying, we embed two inflation forecasting strategies—one based on costly ex ante perfect foresight, and the second based on costless ex ante extrapolative trend-following—in a macrodynamic model. Drawing also on the empirical evidence that inflation forecast errors may have to exceed some threshold before agents abandon their previously selected forecasting strategy, we describe agents as switching between forecasting strategies according to evolutionarily satisficing learning. Convergence to a long-run equilibrium consistent with output growth, unemployment and inflation at their natural levels may be achieved even if heterogeneity in inflation forecasting strategies (with predominance of the extrapolative foresight strategy) is an attractor of an evolutionarily satisficing dynamic perturbed by mutant agents. Thus, in keeping with the empirical evidence, heterogeneity in strategies to form inflation expectations (with prevalence of bounded rationality) can be a stable long-run equilibrium.

The paper studies the nature of expectations formation rules for deterministic economic laws with an expectations feedback within the framework of dynamical systems theory. In such systems, the expectations formation rules, called predictors, have a dominant influence. The concept of a perfect predictor, which generates perfect-foresight orbits, is proposed and analyzed. Necessary and sufficient conditions are given for which local as well as global perfect foresight is possible. The concept is illustrated for the general linear model as well as for models of the cobweb type. For the standard overlapping generations model of economic growth, the existence of perfect predictions depends strongly on the savings behavior of the agents and on the technology.