The trouble with macroeconomics as a science is that its predictive power is terribly weak. There's lots of math based on unrealistic assumptions. Science is prediction, not explanation, wrote Fred Hoyle. A theory without predictive capability cannot lead to an engineering discipline or a control strategy. Yet economists would like their theories to affect public policy.

Whether there needs to be an intuitive model behind a theory is another matter entirely. If a theory has predictive power, it's in some sense good, although it may not be doing what you think it's doing. Artificial neural nets have the same problem - they can work, but there's no understandable model of what they're doing.