Parkinson's law, if you have not heard it before, states that Work expands so as to fill the time available for its completion. That means that if you estimate a task to take one day to complete it will take one day to complete. This is a potential problem with using hours as estimates for your tasks. because if your team is very defensive/pessimistic when doing the estimates you will get less things done than if your team is optimistic. The only upside is that an over estimating team will feel happy that they always finish things in time while a team that constantly under estimates will feel bad for not being able to complete everything planned. But from the outside (and supported by Parkinson's law) the under estimating team will complete more things than the over estimating team.

But having an unhappy team because of constant under estimation is not good either so what should you do? The obvious alternative is; don't estimate your tasks (at least not in time). Having small tasks even of different sizes will be just as good for tracking progress with burn-downs as if you use hours to estimate remaining time. But even while that may be a little bit better than using hours Parkinson is still lurking bringing down your velocity. And working on on a budget is dangerous since you don't know if you last task will be completed "on budget". Actually i think that for most teams at least a few tasks takes longer than estimated and if everything else is completed on budget you will never be able to complete everything on time.

I think the only thing that can work against Parkinson's law is a mature team where everybody wants to complete more and more tasks since it is a proof of improvement (sounds a lot like a Kaizen mind doesn't it). And with a mature team where there is a hunger to complete as much as possible all the time it doesn't really matter how you estimate your tasks. And if it doesn't matter i suggest you make it as easy as possible so skip the hours.