Flexible time horizon planning, learning and aggregate fluctuations

Abstract

A dynamic general equilibrium model is constructed and analyzed under the assumptions that the representative decision maker is endowed with incomplete knowledge about his environment of interest along with limited cognitive skills. Under various beliefs about the long-run growth prospects consistent with Kaldor's, 1961 stylized facts, the benchmark infinite-dimensional decentralized planning problem collapses into infinite sequences of T>1 dimensional sub-problems which are each within the decision maker's limited computing range. Based on the time length of the planning horizon and the nature of expectations, the resulting framework in which both forecasts and plans may get revised in the light of new information encompasses at the limits the benchmark infinite horizon case with perfect foresight, a rolling infinite horizon case, a self-fulfilling static planning case and Day's, 1969 two-period heuristic case. Under exogenous price and quantity growth forecasts consistent with the perfect foresight steady-state balanced growth path, the time length of the planning horizon may have no effect on the resulting competitive equilibrium which may deviate dramatically from the optimal solution. When these growth forecasts are endogenous, the competitive equilibrium effects of the planning horizon time length depend on the characteristics of the information sets. The resulting model may help us to better understand how economies respond to shocks by unraveling new propagation mechanisms related to changes in optimism or pessimism about short-run and/or long-run growth prospects.