Science involves experiments. The best economists can do is take a few poorly recorded historical measurements of incomprehensibly complicated systems and try to force-fit them to their simple models.
Does it qualify as an actual prediction? Broken clocks are correct twice a day but...