Hello everyone,
I am studying the survival of firms in the exported destination between 2009-2019. My data is annual and I have many time-varying covariates. Because I am using the data of Turkish Statistical Institute, I have the total list of all exporting firms. I am planning to estimate survival by using cloglog model however first I want to estimate it with Cox to also present the results to my PhD advisor even if it is not the most proper method for the structure of data. I am interested in first spells between this period so far but then I will include multiple spells assign other methods suitable for it such as shared fraility.
My question is the time period 2009-2019. But there are some firms whose spells start later. They might be exported to same destination before 2009 but I do not have information for that. The only think that I know is they are not exporting from 2009 until the starting year of the spell cause otherwise it would have seen it in the data. As far as I understand these type of entries are named as delayed entry. When I search, in some sources, I have seen that they are using time0 option for these kind of data and in some there is entry option. I am planning to use the code below but still I am not sure. Besides, do I also need to transform my t_start and t_end variable as 0,1,2 as coding 2008=0 instead of years 2008,2009?
Thank you so much.
Cemre
I am studying the survival of firms in the exported destination between 2009-2019. My data is annual and I have many time-varying covariates. Because I am using the data of Turkish Statistical Institute, I have the total list of all exporting firms. I am planning to estimate survival by using cloglog model however first I want to estimate it with Cox to also present the results to my PhD advisor even if it is not the most proper method for the structure of data. I am interested in first spells between this period so far but then I will include multiple spells assign other methods suitable for it such as shared fraility.
My question is the time period 2009-2019. But there are some firms whose spells start later. They might be exported to same destination before 2009 but I do not have information for that. The only think that I know is they are not exporting from 2009 until the starting year of the spell cause otherwise it would have seen it in the data. As far as I understand these type of entries are named as delayed entry. When I search, in some sources, I have seen that they are using time0 option for these kind of data and in some there is entry option. I am planning to use the code below but still I am not sure. Besides, do I also need to transform my t_start and t_end variable as 0,1,2 as coding 2008=0 instead of years 2008,2009?
Code:
stset t_end, fail (fail) exit (t_end .) id(id) time0(t_start)
Code:
firm_countypair t_start t_end fail interval 1 2008 2009 0 1 1 2009 2010 0 2 1 2010 2011 1 3 2 2008 2009 0 1 2 2009 2010 0 2 2 2010 2011 0 3 2 2011 2012 1 4 3 2009 2010 0 1 3 2010 2011 0 2 3 2011 2012 1 3 4 2009 2010 0 1 4 2010 2011 0 2 4 2011 2012 0 3 4 2012 2013 1 4 5 2009 2010 0 1 5 2010 2011 0 2 5 2011 2012 1 3
Cemre
