ScholarWorks@숙명여자대학교

초록

The first order autoregressive (AR(1)) process has been useful for analyzing serially correlated time series data. As a non-stationary time series, we discuss explosive AR(1) model focusing on the error structure generating the model. We have considered various errors such as Gaussian white noise, iid errors, martingale differences and stationary errors. It is noted that asymptotics are heavily influenced by the error structure of the model. The explosive bifurcating AR(1) process is also discussed. Based on the author’s works in the literature, this article reviews the explosive AR(1) process in terms of the error structure, bringing some perspectives to the future topics in this area.

키워드