Firm Age

[Shehzad Moin]

Hi, I am trying to generate a firm age variable in Stata. In my data, for some firms, I find negative or zero firm age when I use the command gen FirmAge = year - listing year. I can fix it, but the question remains whether the firm age should be 0 or 1 in the year of listing and the year of concern. For instance, if both the year and listing year are the same, say 2014, then mathematically firm age is 0, but should it really be considered as 0 or we need to adjust it to 1, as the firm has already been listed in the year 2014? Please suggest the right approach. Moreover, should I take log of firm age or use the firm age as it is in my regression model?

[Felix Bittmann]

Welcome to the forum! I think whether or not to have firms with age zero is just a matter of definition. From a statistical point of view, using the variant which allows zeroes is no problem as long as this variable is not logged. This relates to your second question: what are your arguments to apply the log? This can be beneficial if there are firms of very different ages in your dataset. Have you looked at the distribution? In any case, I think both of your questions are more economics questions than Stata questions. I recommend to have a look at published papers or books in your field so you can study the best practices.

[Clyde Schechter]

I think whether or not to have firms with age zero is just a matter of definition.

Absolutely, I couldn't agree more.

In examples from my own field, clinical epidemiology, when we look at the age of a person, we consider them to be age 0 between birth and their first birthday. On the other hand, when we calculate the length of stay for a hospitalization, if you are discharged on the same day you were admitted, that is considered a 1 day stay (probably because the hospital bills for 1 day!), and if you remain in the hospital past midnight of that day, even just for a fraction of a second, it is now a 2 day stay, etc. So it is a matter of adopting the definition commonly used in your context. As it is apparent you are working in economics or finance, that's out of my wheelhouse and I can't advise you, but I would bet there is a standard practice in those fields for the purpose.

However, I would be concerned about your finding that some of the calculated firm ages are negative. That would imply that the firm went out of business before it was listed. I'm no expert in finance, but I would be beyond astonished to learn that any stock exchanges will begin listing a firm that is already defunct. I would think these negative firm ages represent errors in the data, and I would recheck the data management that produced these results for errors, and if none are found, contact the source of the original data for an explanation of what is going on (and probably a corrected data set.)

[Shehzad Moin]

Dear Felix and Clyde
Thanks a lot for your responses. Kindly accept my apologies for the late reply. I had written my reply to your responses, but I was facing some technical error in posting my response. I tried a couple of times but due to consistent technical error, I postponed and got stuck in other tasks. However, your suggestions are worthwhile and I'll duly consider them. I am from accounting and finance, and my research is on corporate sustainability.
Thank you once again!