What is good economic theory?  What is “applied theory” vs. “theory”?  Different people and departments have different answers to these questions.  Attempts to define these terms and then categorize papers using the definitions is fraught with problems.  Good economic theory is like pornography: You recognize it when you see it.  I am led to this conclusion by reading a paper “Model Building vs Theorizing: The Paucity of Theory in the Journal of Economic Theory” in Econ Journal Watch. The authors define good theory using three questions: Theory of what?, Why should we care?, What merit in your explanation?.  They then categorize papers in JET in 2004 using the three questions.  They ask the questions in order: a paper must pass the previous question for the next even to be applied.  By “what”, the authors mean “some fact in the real world”, by “care” they mean to ask “is it important to explain the fact” and by “merit” they mean “have you given the best explanation or an original one”.

I claim that a good economic theory does not have to satisfy the first question.  Indeed the main paper the authors hold out as an example of good economic theory, Akerlof’s Market for Lemons, I would argue fails the first question.  Akerlof was not trying to explain some fact in the real world.  In fact the used car market does not break down completely and may work quite well (see a paper by Hendel and Lizzeri in AER 1999) but this does not undercut Akerlof’s theory as it was never meant to describe the used car market. Akerlof utilized the used car market to illustrate the idea of adverse selection.  I claim he definitely passes the “why should we care” question, not because to he explains some fact but because he identifies a fundamental issue that impacts all trade under common values and asymmetric information.  Whether it is a big effect or a small effect is an empirical question; adverse selection may face some countervailing effects that minimize its impact.  But Akerlof certainly explained the logic of adverse selection incredibly clearly.  To identify countervailing effects, you first have to understand adverse selection.  Akerlof’s model is totally original and hence he also passes the third test.

Nash equilibrium also fails the first question – it explains no fact.  It passes the other two (broadly defined) as without this notion of equilibrium or its variants it is impossible to study strategic interaction formally.  Much of the research published in JET might fall into this category – fact-free but with the potential to be important in social science research.  But identifying this papers with some defined criteria is impossible.

Finally, unlike the physical sciences, economics has a normative theoretical component which is prescriptive not predictive.  This also fails the first question but is important nevertheless.

So, certainly both non-theorists and some theorists have little patience for research that displays mathematical ingenuity but has no value as social science.  But defining this work exactly is impossible.  This sort of work is like pornography quite simple to recognize when one sees it.  Unlike pornography, it draws no large audience and is quite easy to ignore.