A bivariate threshold time series model for analyzing Australian interest rates Original Research Article

In recent years, research in nonlinear time series analysis has grown rapidly. Substantial empirical evidence of nonlinearities in economic time series fluctuations has been reported in the literature. Nonlinear time series models have the advantage of being able to capture asymmetries, jumps, and time irreversibility which are characteristics of many observed financial and economic time series. As compared to the linear models, the nonlinear time series models provide a much wider spectrum of possible dynamics for the economic time series data. In this paper, we explore the use of nonlinear time series models to analyze Australian interest rates. In particular, we concentrate on the class of bivariate threshold autoregressive (BTAR) models. Monthly Australian interest rates from 1957.1 to 2002.8 are considered. The series under study are 2- and 15-year government bonds, representing short- and long-term series in the term structure of interest rates. A BTAR model is fitted to the observed vector series and the results show that the dynamic structure of the two interest rate series depends heavily on the status (expansion versus contraction) of the economy.