Matias Vernengo — What killed theory? What theory?

So Noah Smith and Paul Krugman are again trying their hand at the history of economic ideas to understand what happened with the profession in the last three decades....

Naked Keynesianism
What killed theory? What theory?
Matias Vernengo | Associate Professor of Economics, University of Utah

Scientific models are (supposed to be) general descriptions of some segment of reality. Their connection with reality is buttressed by correct predictions. But theories are never finally confirmed, since general assertions regarding unbounded sets, and future information is unbounded, are always open to disconfirmation. A single solid disconfirmation is sufficient to falsify a hypothesis, whereas theories are not disconfirmed as such. When problems develop, at first ad hoc solutions are devised within the theoretical framework, but when problems mount or a single issue calls the whole way of looking into question, then theorists look for a better instrument for seeing, i.e., a new theory.

Theories are ways of seeing reality from a particular perspective. The closer they are to a general description that is borne out by testing, with major hypotheses surviving attempts at falsification, the more scientific they can claim to be. Regardless of past successes, anomalies often arise that the theory does not explain, and these become the problem set that defines the cutting edge of the discipline.

No model can be representational for other than a simple bounded system, where anomaly is ruled out. As complexity increases, the potential for representational modeling decreases owing to the uncertainty introduced by unknown unknowns that cannot be foreseen prior to their emergence from analysis of the system based on existing information.

The way of seeing embedded in a theory that attempts to generally describe a complex system is scientific to the degree that it enables accurate prediction and unscientific to the degree it doesn't. The latter condition arises owing to two chief reasons. First, emergence, and secondly, assumptions that are not empirically testable or which testing tends to call into question if not disconfirm. Emergence is not controllable, but assumptions are, at least to a degree.

To the degree that a theory is unscientific and the approach is uncritical, it can be described as mythological. And to the degree that a mythological way of seeing is accepted, the field is dominated by magical thinking.

However, a purely empirical approach focusing only on data is unsatisfactory for it offers no way of seeing the entire data set scientifically, i.e., in terms of a general description. It is not possible to have an integrated field of study without presuming some way of seeing, that is organizing raw data into information. Hence, ignoring the operative way of seeing in using a purely empirical approach invites magical thinking as the method and mythology as the POV.

This is a knotty issue in social science, and conventional economics is deeply affected by it. The assumption of the market as a "natural" phenomenon is a myth based on magical thinking, i.e., "the invisible hand," whose basis is "rationality," functioning as deus ex machina. Homo economicus is a mythological creature.