I'm fitting a survival model onto a risk set of 10 years of teachers contract records. Among other variables I have a certification which teachers earns during their career, always after they have entered their profession (after their first year). My outcome variable is either leaving teaching or moving to a different school (is a competing risk model).

This certification, as an event, it can only happened after the first year. Is it better to consider this qualification a fix or time variant factor? As a fixed factor is quite easy to set; as time variant factor I wouldn't know how to set it up.

Would it be a missing variable till a teacher earn the certification, and then a fix covariate? For example, a certification is earn in the 4th year, and the teacher migrates to a different school in the 8th year, For a wide shape data frame, i'm imagining the following data pattern:

[NA NA NA NA 1 1 1 1 . . .] = certification earn in the fourth year, andmoving to a different school in the 8th year.

Other option, I've thought is to have it as a dummy, when is not present to fixed to zero, when the qualification is for high accomplishment to have it  as +1; and when the certification is lower to have as -1. For the previous example, then the data pattern would be:

[0 0 0 0 1 1 1 1 . . .]

If anyone has any hints, suggestions, tips or reference for this scenario, I'll be more than happy to hear them.

Thanks in advance,