How to deal with missing data?

[oussama zen]

Hello every one,

I am coming back to you because I have one question.

among my database, I have a variable which is observed in the following years: 1960, 1970, 1980, 1990, 2000, 2010, 2013, 2015 and 2017.

I want to run an ARDL model made up of a set of variables including thiat variable.

Does anyone have any idea on how I can estimate my model with this kind of variable?

NB: Friends suggested that I do a multiple imputation to complete the series before running the ARDL.

by advance thank you.

[Carlo Lazzaro]

Oussama:
your friends are probably right.
What I would add is that you have to diagnose the mechanism underlying your missing data (MCAR; MAR; MNAR). See -mi- entry for further details and interesting references.

[oussama zen]

thank you dear carlo