ABSTRACT
Economists have great interest in measuring the business cycle inherent to economic time series; see Baxter and King (1999), Kaiser and Maravall (2001), Stock and Watson (1999), and references therein. A popular method for estimating the cycle is the Hodrick-Prescott (HP) filter (Hodrick and Prescott 1997). Since the HP is designed for nonseasonal series, it can only be sensibly applied to seasonally adjusted data. This serves as the context of this chapter: cycle estimation for univariate time series. Given that it is not uncommon for economists to start with seasonally adjusted data, which is typically produced without taking cyclical effects into account, it is natural to ask the
K12089 Chapter: 5 page: 109 date: February 14, 2012
K12089 Chapter: 5 page: 110 date: February 14, 2012
Modeling and
question: how is the estimation of the cycle affected? If there is additional signal extraction error due to the prior activity of seasonal adjustment, can this error be quantified? We set out to provide answers to these important questions.
