Predicting Turning Points by Modelling Leading Indicators.pdfVIP

Predicting Turning Points by Modelling Leading Indicators Masaki Katsuura* Department of Economics Meijo University 1-501 Shiogamaguchi, Tenpaku, Nagoya, Aichi, JAPAN 468-8502 Phone: +81-52-832-1151 Fax: +81-52-833-4767 Email: katsuura@meijo-u.ac.jp Abstract This paper proposes a statistical model for business cycles. The purpose of this model is to construct a leading index, the structure of the model being simpler than the Stock-Watson model. Estimation is based on partial common principal component analysis using leading indicators. The model produces two useful series for forecasting turning points; one expresses the leading fluctuation of the business cycle as the usual leading index and the other is the turning point indicator. The model is estimated for US data using the five components from the Long Range Gauge of the Economic Cycle Research Institute. The estimated series performs well for forecasting official turning points. KEY WORDS: Business Cycle; Leading Indicator; Stock-Watson model; Partial Common Principal Component Analysis JEL Classification: C22; E32; E37 December 1999 * The author is visiting the School of Economics and Finance, Queensland University of Technology (GPO Box 2434, Brisbane, Queensland 4001, AUSTRALIA, phone: +81-7-3864 2951, fax: +81-7-3864 1500 email: m.katsuura@.au) until March 2000. I. Introduction The present US economy