A0391
Title: Wald-type test for conditional moving average unit root
Authors: Ryota Yabe - Shinshu University (Japan) [presenting]
Abstract: A Wald-type test is introduced for the unit root moving average model of order 1 (MA(1)). The MA model with a unit root is a widely studied time series model, and its unit root test is essential in various applications, such as testing the stationarity of AR processes and examining cointegration. The LM type test is very commonly used but has the drawback that its limiting power is not very high for distant alternative hypotheses. The KPSS test, which is based on this test, also exhibits the same limitation. A new Wald-type test derived from the auxiliary equation is introduced. This test was initially proposed in a prior study for a long memory process. This test is particularly useful for complex models where the limiting distribution of the estimator of the parameter of interest is unclear. The implementation of this test can be achieved by utilizing an auxiliary equation in conjunction with an OLS estimator. The results indicate that in both finite and infinite samples, the performance of our proposed test surpasses that of the LM test. Furthermore, the limiting power function of our test is consistently near the limiting power envelope function.
