Testing for unit roots in autoregressive moving average models: An instrumental variable approach

摘要: Abstract In this paper we propose an approach, based on instrumental variable estimator, for testing the null hypothesis that a process Y t is ARIMA( p , 1, q ) against alternative it stationary + 0, process. Our approach extension of procedure suggested by Hall (1989a) case = 0. We derive limiting distributions estimator when estimated model either (i) true model, (ii) with shift in mean included, or (iii) and linear time trend included. The performance test statistics investigated using Monte Carlo study. Generally speaking, criteria seem to perform as good better than existing methods specified correctly. However, if overspecified, then empirical levels are higher nominal level moderate-sized samples, whereas underspecified estimators inconsistent.