It's a bit buried, but the gist appears to be (italics mine):

"And that’s it — there’s simply very little data on how well the Myers-Briggs (and other type-based personality tests) measures personality and even less on how it might predict job performance."