Project Managers are expected to guess the future, a difficult task even in sciences like meteorology. As many projects finish late (due to expanded scope, underestimated durations, etc), it is tempting to pad estimates.
When is it okay (or not okay) to do so? Off the top of my head, it is risky to share padded estimates with project team members (as they will expand their work to fit the expanded timelines), and risky to share padded estimates with management (as they will add some padding of their own before passing the schedule upwards). On the other hand, planning a schedule with no padding is pretty much guaranteed to lead to a late project.