Making it a property rather than a field means it can be used on interfaces.
The exact implementation (although auto-properties don’t really have much implementation…) is also abstracted, so you could in the future base it on a combination of fields without breaking (compile) compatibility.