Using HP Filtered Data for Econometric Analysis : Some Evidence from Monte Carlo Simulations

The Hodrick-Prescott (HP) filter has become a widely used tool for detrending integrated time series in applied econometric analysis. Even though the theoretical time series literature sums up an extensive catalogue of severe criticism against an econometric analysis of HP filtered data, the original Hodrick and Prescott (1980, 1997) suggestion to measure the strength of association between (macro-)economic variables by a regression analysis of corresponding HP filtered time series still appears to be popular. A contradictory situation which might be justified only if HP induced distortions were quantitatively negligible in empirical applications. However, this hypothesis can hardly be maintained as the simulation results presented within this paper indicate that HP filtered series give seriously rise to spurious regression results. ∗We would like to thank the participants of the Fourth Workshop in Macroeconometrics at the Halle Institute for Economic Research for their comments on a preliminary version of this paper. Furthermore, we are indebted to the participants of the Thirtieth Macromodels International Conference, in particular David Hendry, Søren Johansen, Katarina Juselius and Helmut Lütkepohl, for stimulating discussions and fruitful suggestions which helped to improve our paper. †Department of Economics, Law, and Social Sciences, University of Erfurt, P.O. Box 900 221, D-99105 Erfurt, Phone: +49/361/737-4598, E-Mail: [email protected] ‡Department of Economics, Law, and Social Sciences, University of Erfurt, P.O. Box 900 221, D-99105 Erfurt, Phone: +49/361/737-4591, FAX: +49/361/737-4599, E-Mail: [email protected]