Model selection in under-specified equations facing breaks
Although a general unrestricted model may under-specify the data generation process, especially when breaks occur, model selection can still improve over estimating a prior specification. Impulse-indicator saturation (IIS) can 'correct' non-constant intercepts induced by location shifts i...
| Main Authors: | , |
|---|---|
| Format: | Working paper |
| Published: |
University of Oxford
2010
|
| _version_ | 1797086007295737856 |
|---|---|
| author | Hendry, D Castle, J |
| author_facet | Hendry, D Castle, J |
| author_sort | Hendry, D |
| collection | OXFORD |
| description | Although a general unrestricted model may under-specify the data generation process, especially when breaks occur, model selection can still improve over estimating a prior specification. Impulse-indicator saturation (IIS) can 'correct' non-constant intercepts induced by location shifts in omitted variables, which surprisingly leave slope parameters unaltered even when correlated with included variables. However, location shifts in included variables do induce changes in slopes where there are correlated omitted variables. IIS acts as a 'robust method' when models are mis-specified, and helps mitigate the adverse impacts of induced location shifts on non-constant intercepts and equation standard errors. |
| first_indexed | 2024-03-07T02:15:53Z |
| format | Working paper |
| id | oxford-uuid:a239b145-e63b-4bd6-8de6-ee4df89aeb91 |
| institution | University of Oxford |
| last_indexed | 2024-03-07T02:15:53Z |
| publishDate | 2010 |
| publisher | University of Oxford |
| record_format | dspace |
| spelling | oxford-uuid:a239b145-e63b-4bd6-8de6-ee4df89aeb912022-03-27T02:18:39ZModel selection in under-specified equations facing breaksWorking paperhttp://purl.org/coar/resource_type/c_8042uuid:a239b145-e63b-4bd6-8de6-ee4df89aeb91Bulk import via SwordSymplectic ElementsUniversity of Oxford2010Hendry, DCastle, JAlthough a general unrestricted model may under-specify the data generation process, especially when breaks occur, model selection can still improve over estimating a prior specification. Impulse-indicator saturation (IIS) can 'correct' non-constant intercepts induced by location shifts in omitted variables, which surprisingly leave slope parameters unaltered even when correlated with included variables. However, location shifts in included variables do induce changes in slopes where there are correlated omitted variables. IIS acts as a 'robust method' when models are mis-specified, and helps mitigate the adverse impacts of induced location shifts on non-constant intercepts and equation standard errors. |
| spellingShingle | Hendry, D Castle, J Model selection in under-specified equations facing breaks |
| title | Model selection in under-specified equations facing breaks |
| title_full | Model selection in under-specified equations facing breaks |
| title_fullStr | Model selection in under-specified equations facing breaks |
| title_full_unstemmed | Model selection in under-specified equations facing breaks |
| title_short | Model selection in under-specified equations facing breaks |
| title_sort | model selection in under specified equations facing breaks |
| work_keys_str_mv | AT hendryd modelselectioninunderspecifiedequationsfacingbreaks AT castlej modelselectioninunderspecifiedequationsfacingbreaks |