Non-parametric direct multi-step estimation for forecasting economic processes
We evaluate the asymptotic and finite-sample properties of direct multi-step estimation (DMS) for forecasting at several horizons. For forecast accuracy gains from DMS in finite samples, mis-specification and non-stationarity of the DGP are necessary, but when a model is well-specified, iterating th...
| Main Authors: | , |
|---|---|
| Format: | Working paper |
| Published: |
University of Oxford
2004
|
| _version_ | 1797062871247486976 |
|---|---|
| author | Hendry, D Chevillon, G |
| author_facet | Hendry, D Chevillon, G |
| author_sort | Hendry, D |
| collection | OXFORD |
| description | We evaluate the asymptotic and finite-sample properties of direct multi-step estimation (DMS) for forecasting at several horizons. For forecast accuracy gains from DMS in finite samples, mis-specification and non-stationarity of the DGP are necessary, but when a model is well-specified, iterating the one-step ahead froecasts may not be asymptotically preferable. If a model is mis-specified for a non-stationary DGP, in particular omitting either negative residual serial correlation or regime shifts, DMS can forecast more accurately. Monte Carlo simulations clarify the non-linear dependence of the estimation and forecast biases on the parameters of the DGP, and explain existing results. |
| first_indexed | 2024-03-06T20:51:48Z |
| format | Working paper |
| id | oxford-uuid:37d63022-24b9-4eba-a3e0-0a47210993f7 |
| institution | University of Oxford |
| last_indexed | 2024-03-06T20:51:48Z |
| publishDate | 2004 |
| publisher | University of Oxford |
| record_format | dspace |
| spelling | oxford-uuid:37d63022-24b9-4eba-a3e0-0a47210993f72022-03-26T13:46:21ZNon-parametric direct multi-step estimation for forecasting economic processesWorking paperhttp://purl.org/coar/resource_type/c_8042uuid:37d63022-24b9-4eba-a3e0-0a47210993f7Symplectic ElementsBulk import via SwordUniversity of Oxford2004Hendry, DChevillon, GWe evaluate the asymptotic and finite-sample properties of direct multi-step estimation (DMS) for forecasting at several horizons. For forecast accuracy gains from DMS in finite samples, mis-specification and non-stationarity of the DGP are necessary, but when a model is well-specified, iterating the one-step ahead froecasts may not be asymptotically preferable. If a model is mis-specified for a non-stationary DGP, in particular omitting either negative residual serial correlation or regime shifts, DMS can forecast more accurately. Monte Carlo simulations clarify the non-linear dependence of the estimation and forecast biases on the parameters of the DGP, and explain existing results. |
| spellingShingle | Hendry, D Chevillon, G Non-parametric direct multi-step estimation for forecasting economic processes |
| title | Non-parametric direct multi-step estimation for forecasting economic processes |
| title_full | Non-parametric direct multi-step estimation for forecasting economic processes |
| title_fullStr | Non-parametric direct multi-step estimation for forecasting economic processes |
| title_full_unstemmed | Non-parametric direct multi-step estimation for forecasting economic processes |
| title_short | Non-parametric direct multi-step estimation for forecasting economic processes |
| title_sort | non parametric direct multi step estimation for forecasting economic processes |
| work_keys_str_mv | AT hendryd nonparametricdirectmultistepestimationforforecastingeconomicprocesses AT chevillong nonparametricdirectmultistepestimationforforecastingeconomicprocesses |