Old settings are likely to be reused before new ones for the same reason it's easier to make a sequel of a bad movie than a new one: name recognition. People are more comfortable buying into something they either already know something about or have heard about than something completely new. That's not to say that they can't come up with a good new setting (I don't care for Eberron, but it obviously became one of the most popular very quickly), just that they already have existing material that will be easier to sell.
New character concepts should be done as sub-classes, except in the case of setting specific needs (e.g. the artificer). The only new class I'd like to see that's setting neutral would be the psion/mystic, and maybe a swordmage type (fighter/wizard to paladin's fighter/cleric). I think 3E & 4E showed us the downside of class glut, and I don't want 5E to follow down that route. Subclasses work better IMO, because they usually only impact a few levels, while leaving the basic aspect for the class the same.
New monsters are always useful, but they should be carefully thought out. Making random new monsters just for the sake of having them isn't helpful, since many DMs need to have some reason for them to exist in their world if they want to use them. Certain monsters, like the Terrasque (sp?), can have an epic impact on a setting merely by existing.