I think it's because building great products isn't just a matter of execution. It's also, to some extent, an art.
I have trouble imagining someone being effective and competent with absolutely no passion. I will grant, though, that there are competent people who are more passionate, and those who are less so. I think a successful organization could have a blend of people across the passion scale, but I don't think one could build a successful product business without at least a few people who are passionate about what they're doing. Maybe in other kinds of businesses you could; I don't know.
Of course it gets trickier when the more competent one seems the less passionate. Then it's a question of how much to weight the two dimensions (along with whatever others one is considering). I have no magic formula here; I think it depends on the circumstances.
There seems to be a subtext here that work sample tests give no indication of passion. I don't think that's true; when I've given people programming exercises, I think the craftsmanship of the code (is it well-modularized, well-documented, nicely indented, etc.) is indicative of their passion. Competence, to me, is more about whether the code works.
So yes, I'm using craftsmanship as a proxy for passion, so I suppose next you'll ask, why not just go for craftsmanship? The point we seem to be coming to is that passion is difficult or impossible to measure directly, so why not stick to qualities that are more measurable?
I don't think everything important is easy to measure.