Capturing the Shape of the Business Cycle with Autoregressive Leading Indicator Models
Email address: George.Athanasopoulos@BusEco.monash.edu.au
Keywords: Leading Indicator Models, Monetary and Financial Variables, BBQ.
JEL Classifications: C32, E32
In this paper, we examine if linear and non-linear autoregressive leading indicator models can reproduce the shape of the "business cycles" in Australia, New Zealand and the United States. Autoregressive leading indicator (ARLI) models were proposed by Zellner and Hong (Journal of Econometrics, 1989), and were generalized to non-linear models by Anderson and Vahid (Macroeconomic Dynamics, forthcoming). We examine if monetary and financial variables, such as money and the interest rate spread, are more suitable than real variables in capturing the cyclical characteristics of output (GDP). The success of different models in modelling the business cycles is gauged by the non-parametric procedure developed by Harding and Pagan (MIAESR Working Paper, 1999). Our preliminary findings indicate that the bivariate ARLI models of output and the spread between short-term and long-term interest rates can successfully capture the shape of the business cycle. In particular, they can capture the deviation of the actual path of the cycles from a triangular path, a characteristic that other models of GDP fail to reproduce.
PDF file of paper: athanasopoulos.pdf
Session: Empirical Analyses of Business Cycles
Time: Friday, 6 July, 3:30pm - 5pm