Nonlinear and nonstationary models in econometric analysis

点击翻译按钮获取中文摘要

Yajima investigated an effect of missing observations on estimation of nonstationary unit root processes and stationary long memory models.He considered the two estimations for unit root processe. The first one is a Yule-Walker type estimator and the second one is a least-squares type estimator.He derived the limiting distributions of these estimators. Next he introduced the third estimator, a sample-correlation coefficient estimator. Then he clarified their asymptotic difference if we apply them to estimate the autocorrelation function of both short-memory and long-memory stationary models.Kunitomo considered X-11-ARIMA,a seasonal adjustment method which has been recently developed by Bureau of Census, Department of Commerce in U.S.A.He clarified its theoretical properties and topics which should be solved in future. He also proved the limiting distribution of the maximum likelihood estimator of a simultaneous switching autoregressive model proposed by himself.