A TEST FOR LINEARITY OF STATIONARY TIME SERIES

Abstract. A standard assumption that is often made in time series analysis is that the series conforms to a linear model. The object of this paper is to describe statistical tests for testing this assumption. The tests are constructed from the bispectral density function, and depend on the application of Hotelling T ². These tests are illustrated with two real time series and four simulated time series. Some guidelines about the choice of the parameters are also included.