Heterogeneous and persistent inflation expectation formation in Costa Rica

Abstract

I analyze which forecast method (rational, adaptative, last inflation or the BCCR’s inflation target) is used by the informants of the Encuesta mensual de expectativas de inflación y tipo de cambio of the Central Bank of Costa Rica to form their inflation expectation. Following Branch (2004) I assume that the agents decide between different methods based on their forecast error. A logit model is used to estimate the probabilities that the agents assign to each method. Since the agents respond to the survey multiple times, I correct the estimation to incorporate the temporal dependence in the informants’ answers. The main result is that most of the informants use methods that require little information instead of forming rational expectations. Further, I present evidence that changes in the sample selection procedure have had important implications in the survey-based expectations. Therefore, it is not recommended to use them when it is required a rational expectations indicator or to make intertemporal comparisons.