It would be just as accurate to say that longer work weeks predict wider income gaps. This kind of work is not causal inference.