SUppose that time on the job market improves wages, not directly but opportunistically, via the randomness of jobs.
The worker meets randomly with opportunities to improve his wage (in multiple ways). Naturally, it is not totally random, because he employs selection to perpetuate anything to his advantage.
This will give a unique longevity premium that might not be captured by common stats of tenure length etc. (for example, this does not need to have tenure on a specific job, or even acquired expertise, just perpetuating whatever advantage comes by rnadomly!)