Does tenure belong in the model when women have shorter service on average?

I am rebuilding the pay equity model for a Canadian credit union of about 1,700 employees, and I am stuck on a specification question rather than a data one.

Tenure is a strong predictor of pay here, and including it reduces the adjusted gap from 4.4% to 2.0%. But women's average service is 5.1 years against 7.3 for men, partly because we lose women at higher rates between years three and six. So tenure is doing a lot of the explaining, and tenure itself is not neutral.

My instinct is to report both versions and say why. Is that what others do, or do you pick one and defend it? And does anyone use time in grade or relevant experience instead of company service?