Hey

I am investigating the effects of different investor types (A, B and C) on the failure rate of a company (cross sectional data). investorA, B, and C are dummies which equals one if the first round investor was A,B, or C. Right now, I am performing a simple probit regression where “close” is a dummy if the company failed.

So I do:
Code:
probit close investorA investorB firstroundfunding, vce(robust)
I got the advice that I can or should use a cox hazard model.

Of course, every company have different years for their first funding round. So, I calculate the difference between the first funding round and the closing date in month (time). If the company is still in business, I used the end of my observation (last month). So, if the company was founded July 2018 and is still in business “time” would be 12. On the other hand, if the company failed in December 2018, “time” would be 6.
Is this right so far? On the web I just find examples for medical studies where the entry date was always the same.

Then I perform:

Code:
stset time close
stcox investorA investorB firstroundfunding, vce(robust)
First of all, I want to mention that failing isn’t really a common thing (~7%). Is this a problem?

Next can I add year dummies? Maybe first funding during the financial crisis was not a good idea.

Code:
stcox investorA investorB firstroundfunding i.firstfundingyear, vce(robust)
I would appreciate your advice

Kind regards
Alex