UPOREDNA ANALIZA POSLOVNIH CIKLUSA PRIMENOM HP I HAMILTONOVOG FILTERA
Abstract: This paper examines the performance of the Hodrick-Prescott (HP) filter and the Hamilton filter in decomposing time series into trend and cyclical components, using monthly industrial production index data for selected Central and Eastern European countries from 2000 to 2025. The findings suggest that the cycles extracted using the Hamilton filter exhibit greater persistence and a higher degree of cross-country coherence. However, the analysis does not provide strong evidence of superior robustness of the Hamilton approach under variations in input parameters or in real-time estimation, despite claims that Hamilton filter “achieves all the objectives sought by users of the HP filter with none of its drawbacks” (Hamilton (2018)). Our study aims to enhance understanding of modern filtering techniques in business cycle analysis while highlighting some methodological challenges that researchers should consider when extracting time series components.
Rezime: U radu su analizirane performanse Hodrik-Preskotovog (HP) i Hamiltonovog filtera pri dekompoziciji vremenskih serija na trend i cikliˇcnu komponentu, koriste´ci meseˇcne podatke indeksa industrijske proizvodnje za zemlje Centralne i Istoˇcne Evrope u periodu 2000–2025. Rezultati pokazuju da ciklusi dobijeni Hamiltonovim filterom ispoljavaju vec´u perzistentnost i viši stepen usklad¯enosti ciklusa med¯u posmatranim zemljama. Med¯utim, nalazi ne potvr ¯ duju da je ovaj filter robusniji na promenu ulaznih parametara i na ocenjivanje u realnom vremenu, što bi trebalo da je njegova kljuˇcna metodološka prednost u odnosu na HP filter (Hamilton (2018)). Naše istraživanje može da doprinese boljem razumevanju novijih pristupa u analizi poslovnih ciklusa i da ukaže na pojedine metodološke izazove u formiranju izvedenih serija.
srpski
2025
© All rights reserved
Keywords: Hamilton filter, HP filter, business cycles.
Ključne reči: Hamiltonov filter, HP filter, poslovni ciklusi.