Diabold Mariano Test
Posted: Sat Jul 22, 2017 12:56 am
I'm using Eviews 9.5 SV to perform the Diabold Mariano (DM) test. I have read the instructions, however, I cannot get the test to run properly therefore my problem is more operating the software than understanding the test.
*** Excel file and DM results attached ***
The following is the steps I'm taking;
1) Importing data from Excel = File - New - Workfile - unstructured / undated - 1279 observations.
2) Open the workfile = File - Open - Eviews Workfile.
This produces the screen featuring; c, forecast_1, forecast_2, resid.
3)Highlight 'forecast_1' then open the spreadsheet.
4)Highlight 'forecast_2' then open the spreadsheet.
At this point both forecasts are on different windows / spreadsheets. If I try to open them as a group then then 'forecast evaluation' option vanishes from the list.
5) Using either the forecast 1 or 2 windows, View - Forecast Evaluation, 'forecast data options' = forecast_1 forecast_2, select 'All', evaluation sample = 1279, training = 1 500.
6) Clicking 'Ok' then generates the DM test results.
However, I'm unconvinced by the results that state both forecasts have the same accuracy (impossible) and the statistics for forecast 1 are all zero values (0.0000).
Please advise
Ross
*** Excel file and DM results attached ***
The following is the steps I'm taking;
1) Importing data from Excel = File - New - Workfile - unstructured / undated - 1279 observations.
2) Open the workfile = File - Open - Eviews Workfile.
This produces the screen featuring; c, forecast_1, forecast_2, resid.
3)Highlight 'forecast_1' then open the spreadsheet.
4)Highlight 'forecast_2' then open the spreadsheet.
At this point both forecasts are on different windows / spreadsheets. If I try to open them as a group then then 'forecast evaluation' option vanishes from the list.
5) Using either the forecast 1 or 2 windows, View - Forecast Evaluation, 'forecast data options' = forecast_1 forecast_2, select 'All', evaluation sample = 1279, training = 1 500.
6) Clicking 'Ok' then generates the DM test results.
However, I'm unconvinced by the results that state both forecasts have the same accuracy (impossible) and the statistics for forecast 1 are all zero values (0.0000).
Please advise
Ross