Real-Time vs. Full-Sample Performance of One-Sided and Two-Sided HP Filters. An Application to 27 EU Member States' GDP Data

Abstract

The paper makes a comparison of the results of the application of two-sided
and one-sided versions of the Hodrick-Prescott filter on GDP data concerning
27 EU Member States. Based on the results, the overall finding is that,
contrary to its assumed advantages, the one-sided filter does not help overcome
endpoint unbiasedness. Quite the opposite, it rather spreads and consolidates
the endpoint bias that plagues the two-sided version over the entire filtered data.
In addition, regression-based results on the influence of the second, third, and
fourth moments of the GDP acceleration rates on the differences between onesided and two-sided HP trends are presented.