I don't think it's actually a good indicator because performance is affected by so many important factors—including things like the direct manager and the rest of the team.

What it is, rather, is the IBM effect in reverse. Regardless of how good a predictor it really is, common knowledge is that it really matters. If you add a random new worker to the team and it doesn't work out, well, these things happen. If you have someone with a classic "red flag" (most of which are either highly exaggerated or complete bullshit) and it doesn't work out, it's all your fault.

Of course, in reality, the same factors dominate in both cases—things don't always work out, but that's only to be expected. But woe if you can be blamed for it!