The most recent
Abstract
We study the behaviour of three quantitative sample surveys and a non sample inflation expectation report for Colombia. We found that expectations in Colombia; (i) are not strongly, i.e. a la Muth, rational because they show cross-section disagreement, (ii) expectations, however, show some features of weak rationality, (iii) expectations disagreement is time varying and relate to inflation, inflation changes and the output gap, thus suggesting a staggered information flow to agents, (iv) the forecast error loss function employed by agents is not symmetric and increasingly penalizes higher expectations than finally observed inflation as the horizon grows, and (v) this fact also explains the stylised fact that observed expectation share with theoretical rational expectations that expectations look like lagged versions of inflation that dampen with the horizon. The latest finding also arises from a very general econometric set up we develop in this paper. These results imply that the effect of weakening the rational expectations assumption in Colombian monetary policy models should be assessed, especially when compared to sticky information and heterogeneous agents choosing non Mean Square forecast Error losses.