"To make sense of the data an economist must have a theory which stands on its own feet and does not originate from the data itself. The key point is that it must originate from something real that cannot be refuted."