Documentos de Trabajo
Permanent URI for this collection
Browse
Browsing Documentos de Trabajo by Subject "C10-General"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
- El Filtro BAXTER-KING, Metodología y AplicacionesThis document briefly explains the Baxter-King’s filter methodology, which has become a very useful tool to analyze economic cycles and as a method for trend extraction.Some mathematic properties of Hodrick-Prescott filter are also reviewed, in order to understand the difference between this method and Baxter-King’s. I applied the filter to monthly, quarterly and annual series. Although Baxter-King and Hodrick-Prescott filters do not differ much in their results, the former has the advantage that it allows the researcher to specify the kind of information she wants to isolate from the series.
- Parámetro de suavizamiento del filtro Hodrick-Prescott para Costa RicaThis paper updates the estimation of the smoothing parameter in the Hodrick-Prescott filter for the Costa Rican GDP series previously performed by Segura and Vásquez (2001). These authors use the method of Marcet and Ravn (2003) to determine these values. However, this method assumes that the standard value of 1600 for quarterly series is appropriate, assumption that has been questioned. For this reason, in order to obtain alternative estimates of this parameter we apply the methods of modified HP filter of McDermott (1997) and optimal filtering of Pedersen (2002). These two methods provide values of the smoothing parameter well below of those obtained with the first method and so, the output gaps display a lesser degree of variability and amplitude compared to the gaps obtained by the application of various univariate filters. On the other hand, the output gaps obtained using the optimal values suggested by Marcet and Raven’s method display greater correspondence with those implied by the univariate filters and exhibit better predictive performance in the period 2013 - 2016. Therefore, even though the methods of McDermott and Pedersen provide lambda values which theoretically are more appropriate than the usual standard values, from a practical point of view they are less informative. The optimal values found are 26, 1800, 22200 for annual, quarterly, and monthly GDP series, respectively; and 26400 for the IMAE index of economic activity.