Inflation and money in Colombia: another P-Star model

Abstract

This document presents the estimation of a recent version of the P-Star model by Gerlach and Svensson (2003) and its predictions for Colombia (January 1980 to April 2005). The model is designed to explain the inflation gap (observed rate minus the target) based on the monetary gap and the output gap. According to the results, the output gap lacks significant effects while the monetary gap has significant positive effects on inflation.

Acknowledgements

We are very grateful to an anonymous referee for helpful suggestions which improved the article. We also wish to thank Tatiana Parra and John Jairo León, students from Universidad Javeriana and Universidad Nacional, respectively, for their collaboration and Hernando Vargas, and the members of the Board of Directors of the Banco de la República, for comments and suggestions to previous versions of this article. The opinions and calculations included in this document are those of the authors and do not compromise the mentioned institutions, their directors, or their boards of directors.

Notes

¹ Gómez, et al. (2002) present a detailed description of the design and implementation of the inflation target regime for Colombia.

² The same type of estimation was made using constructed opportunity costs based on the return of external assets and an anticipated rate of devaluation. In general, the results of these exercises were not expected since the opportunity cost did not appear to be significant in relation to the long-term. Additionally, the co-integrated relation does not appear to be stable. For more detail about these results, contact the authors.

³ See Phillips and Hansen (1990) and Hansen (1992).

⁴ The MAFE and RMSFE statistics are expressed in basic points of the inflation gap. For example, the MAFE models without a monetary gap for the period of one quarter is 1.043, which indicates that the mean error of prediction one step ahead is of 104.3 basic points. This means that if the observed inflation gap is 1%, the averaged prediction would be around 2.043%. In the MAPFE and RMSPFE cases, the statistics are measured in percentage terms: a MAPFE = 1.287 value indicates, on average, a difference of 128.7% between the observed gap value and its prediction.