Lag-length selection criteria: empirical results from the St. Louis equation
This article describes and compares six criteria for determining the lag length of finite distributed lag models. These criteria are employed to select the lag length of the distributed lag variables within the St. Louis equation using a computationally efficient procedure. The lag lengths chosen are tested against each other and against arbitrarily overfitted and underfitted specifications. The results suggest that Akaike's final prediction error criterion and Pagano and Hartley's procedure perform well relative to the other criteria considered.
|Date of creation:||1983|
|Date of revision:|
|Contact details of provider:|| Postal: P.O. Box 442, St. Louis, MO 63166|
Web page: http://www.stlouisfed.org/
More information through EDIRC
|Order Information:|| Email: |
When requesting a correction, please mention this item's handle: RePEc:fip:fedlwp:1983-008. See general information about how to correct material in RePEc.
For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: (Anna Xiao)
If references are entirely missing, you can add them using this form.