[ Your Name ]  would like to inform you about this article on
Complexity Digest 2003.02 - 01
http://comdig.unam.mx/index.php?id_issue=2003.02#10115
13-Jan-2003

[ Your Message ]



Modelling Methodology And Forecast Failure, The Econometrics Journal
 









Abstract: We analyse by simulation the impact of model¡Vselection strategies
(sometimes called pre¡Vtesting) on forecast performance in both constant¡V and
non¡Vconstant¡Vparameter processes. Restricted, unrestricted and selected models
are compared when either of the first two might generate the data. We find
little evidence that strategies such as general¡Vto¡Vspecific induce significant
over¡Vfitting, or thereby cause forecast¡Vfailure rejection rates to greatly
exceed nominal sizes. Parameter non¡Vconstancies put a premium on correct
specification, but in general, model¡Vselection effects appear to be relatively
small, and progressive research is able to detect the mis¡Vspecifications.
Source: Modelling Methodology And Forecast Failure[
http://www.ingenta.com/isis/searching/ExpandTOC/ingenta?issue=infobike://bpl/ectj/2002/00000005/00000002&index=3&WebLogicSession=PhwmltfuF2sfKxDPSMwV|3033548102239292324/-1052814329/6/7051/7051/7052/7052/7051/-1
], Clements M.P. - mailto: m.p.clementswarwick.ac.uk, Hendry D.F.,, The
Econometrics Journal, Vol. 5, No. 2, pp: 319-344(26), Dec. 2002 Contributed
byPritha Das

You can discuss this article on Articles Forum
http://comdig.unam.mx/topic.php?id_article=10115