Reconstructing macroeconomics

6 August, 2015 at 16:27 | Posted in Economics | 2 Comments

wrong-tool-by-jerome-awOlivier Blanchard says that, currently, DSGE models have “much too much in them to be fully understood”. There is a rationale for studying a model that we do not understand–if and only if it makes predictions that fit the world. If one has such a model that makes reliable predictions, studying it is a not-implausible road to understanding the world, because maybe, just maybe, an understanding of the model will carry an understanding of the world along with it as a bonus. And there is a rationale for taking models we understand and seeing where and how they fit the world in order to help us iterate toward a better model that fits better.

But is there a case for investigating models we (a) do not understand that (b) do not fit the world? Even if we were to reach the point of understanding the model and how it works, what would that gain us?

Brad DeLong

Just questions. And the answers are — No. Nothing.



  1. The inability or unwillingness of economists to make the traditional sharp distinction between analytical models and operational models — between analysis and synthesis in older terms — serves them poorly. No useful point can be made about method without at least some modicum of precision respectful of this fundamental distinction.
    The point of theoretical analysis is to identify or derive by logical means the necessary and sufficient elements in a system and their functional relationships. By itself, analysis “predicts” nothing; it only identifies the factors we might use in constructing an explanation. To “predict” outcomes in some concrete circumstance, requires auxiliary hypotheses, operational models, observation and measurement. It is the difference between a geometry and a surveyor’s mapmaking. A geometry is literally no-where, and navigation by pure geometry is not just impossible, it is non-sensical. One needs a geometry to construct a map, it is true, but whether one can navigate successfully is a test of the map or the skills of the navigator, and involves a host of concepts and measurements, additional to pure geometry. Only in exceptional circumstances would we regard a failure of the map to indicate a limitation of the geometry used to construct it.
    We do not imagine that the crash of an airplane disproves the theory of aerodynamics. Nor would we expect a manufacturer of airfoils to design one without testing it thoroughly in a windtunnel, making careful measurements.
    But, much of theoretical economics proceeds with great carelessness about introducing auxiliary hypotheses or making measurements. Economists talk as if their analytical models are themselves maps of the world, no matter how absurd the simplifying axioms. I would not object to representative agents or rational expectations in principle, except for this handwaving insistence that the theoreticians need to construct no operational bridge to the world, before pontificating on how the real, political world works. It is not the axiomatic assumptions that constitute the essential absurdity, but the severe neglect of appropriate operational modeling and empiricism.

  2. “Even if we were to reach the point of understanding the model and how it works, what would that gain us?”

    Not necessarily nothing. At least we would know what doesn’t work. That’s something.

Sorry, the comment form is closed at this time.

Blog at
Entries and comments feeds.