A Bayesian Approach to Predicting Cycles Using Composite Indicators

Autor: Paulo Picchetti
Rok vydání: 2018
Předmět:
Zdroj: Societies and Political Orders in Transition ISBN: 9783319900162
DOI: 10.1007/978-3-319-90017-9_20
Popis: This paper proposes a methodology for estimating the probabilities of recession starts and endings in the Brazilian economy. The model providing these estimations is a logistic regression using as covariates some transformations of composite leading and coincident indicators for Brazilian economic cycles. A very attractive feature of this approach is the avoidance of the need for extrapolating the information beyond the available sample, allowing for more reliable real-time assessments. It is shown that a Bayesian approach to the estimation of the model produces more robust and interpretable results.
Databáze: OpenAIRE