Hi all,
I am studying on the survival of firms in export market for my PhD thesis by using firm level data which includes firm characteristics such as age and size and also the characteristics of export destination (export partner's gdp, population, distance etc). My data is a panel in which each observation indicates the export performed by a firm to a specific export partner in that year. Each firm-partner observation is repeated once in a year. If firm did not exported in a specific year or a partner, there is no observation.
I can not share my data because firm level data is not allowed to be shared without the permission of statistical institute of Turkey. But you can find a very small example of how the data looks like. The number in id variable is the firm_id and the next three letters is iso3 code of export destination.
I know that i need to transform the data so that it can be suitable for the survival analysis. The event in my data is quit exporting and repeated for some firms within my the time interval so i have counting process style. Some of my variables are also time-varying (both firm and destination characteristics) and changes each year.
My problem is how should i transform the observations of a firm for instance exports between 2009-2011 and quits in 2012 and then never exports. For example, in my fake dataset id 1 exports to France in 2015 and quits in 2016 and never exports again until the the last year of my dataset which is 2019.
I know that the event variable will be 1 in 2016. But after that how should ı code the remaining years? Should i include only the years that export is performed and the year of quit in my data? I considered that if i code the event variable as 0 for the years that firm does not export than actually it might look like firm exports. If i code them as 1 then it will not be suitable for my definition of event which is stop export? So, how should i transform my data?
Thank you so much for your interest and help in advance,
Cemre,
I am studying on the survival of firms in export market for my PhD thesis by using firm level data which includes firm characteristics such as age and size and also the characteristics of export destination (export partner's gdp, population, distance etc). My data is a panel in which each observation indicates the export performed by a firm to a specific export partner in that year. Each firm-partner observation is repeated once in a year. If firm did not exported in a specific year or a partner, there is no observation.
I can not share my data because firm level data is not allowed to be shared without the permission of statistical institute of Turkey. But you can find a very small example of how the data looks like. The number in id variable is the firm_id and the next three letters is iso3 code of export destination.
I know that i need to transform the data so that it can be suitable for the survival analysis. The event in my data is quit exporting and repeated for some firms within my the time interval so i have counting process style. Some of my variables are also time-varying (both firm and destination characteristics) and changes each year.
| year | id | export |
| 2009 | 1FRA | 1 |
| 2010 | 1FRA | 1 |
| 2012 | 1FRA | 1 |
| 2013 | 1FRA | 1 |
| 2015 | 1FRA | 1 |
| 2009 | 1USA | 1 |
| 2010 | 1USA | 1 |
| 2011 | 1USA | 1 |
| 2010 | 1ITA | 1 |
| 2011 | 1ITA | 1 |
| 2010 | 2DEU | 1 |
| 2011 | 2DEU | 1 |
| 2012 | 2DEU | 1 |
| 2017 | 2DEU | 1 |
| 2010 | 2FRA | 1 |
| 2011 | 2FRA | 1 |
| 2009 | 2GBR | 1 |
| 2010 | 2GBR | 1 |
| 2011 | 2GBR | 1 |
| 2009 | 3ITA | 1 |
| 2010 | 3ITA | 1 |
| 2009 | 4AIA | 1 |
| 2009 | 4FRA | 1 |
| 2012 | 4FRA | 1 |
| 2011 | 4GER | 1 |
| 2009 | 5DEU | 1 |
| 2012 | 5DEU | 1 |
| 2018 | 5DEU | 1 |
| 2019 | 5DEU | 1 |
I know that the event variable will be 1 in 2016. But after that how should ı code the remaining years? Should i include only the years that export is performed and the year of quit in my data? I considered that if i code the event variable as 0 for the years that firm does not export than actually it might look like firm exports. If i code them as 1 then it will not be suitable for my definition of event which is stop export? So, how should i transform my data?
Thank you so much for your interest and help in advance,
Cemre,
