MODELING LONG-RUN BEHAVIOR WITH THE FRACTIONAL ARIMA MODEL
成果类型：
Article
署名作者：
SOWELL, F
刊物名称：
JOURNAL OF MONETARY ECONOMICS
ISSN/ISSBN：
0304-3932
DOI：
10.1016/0304-3932(92)90016-U
发表日期：
1992
页码：
277-302
关键词：
摘要：
When modeling long-run behavior, fractional ARIMA models can give insights unobtainable with the nonfractional ARIMA models. As an application, the deterministic trend and unit root with drift models are nested in the fractional ARIMA model. This allows testing between the two models based on estimated parameter values. This test is applied to postwar US quarterly real GNP. The test concludes that GNP is consistent with both models. The estimated fractional parameter is significantly smaller than reported in Diebold and Rudebusch (1989). The difference is explained by bias in the previous estimates. Relationships with the cumulative impulse response and spectral density at frequency zero are noted.
来源URL：