Regarding data structure in logit panel model

[shilpa gupta]

I am examining the micro level i.e. firm level determinants of External commercial borrowing in India by applying the logit panel model . The data set comprises of 545 firms for the years 2004-2020. I have further divided the dataset into 3 sub time periods a) 2004-2006: Pre-global financial crisis period, b) 2007-2009: Global financial crisis period and c) 2010-2020: post- global financial crisis period. While dividing the into 3 sub time periods, i encountered a problem. There are firms which have taken ECB in pre and post crisis periods but not in GFC period. Should i removed those firms from the GFC sub time period dataset ?

[shilpa gupta]

data is available since 2004 only.

[Carlo Lazzaro]

Shilpa:
as per FAQ, please share an excerpt/example of your dataset, so that interested listers can reply more positively to your query. Thank you.

[shilpa gupta]

This is dataset

[shilpa gupta]

This is dataset for global financial crisis period. there are firms who had taken ECB in pre and post crisis period but not in GFC. the dependent variable for those firms is zero. i want to ask should i remove these firms. however, these firms are part of entire study period of 2004-2020.

[shilpa gupta]

As per my knowledge if i remove those firms then result will be biased. i am right ?

[Carlo Lazzaro]

Shilpa:
1) do not remove those firms;
2) interact -i.GFC_ with firm dimension (turnover; equity; market share) and see what happens to your regression results.

[shilpa gupta]

Thank you