Deirdre McCloskey’s Sins

Deirdre McCloskey has made a name for herself by critically examining the logic and rhetoric of economic arguments. She has a nice pamphlet summarizing her claims called “The Secret Sins of Economics.” It’s written for non-economists and nicely makes three points:

1. Some alleged problems of economics are virtues. For example, using math to describe and analyze economic behavior is actually good because math allows you to clearly deduce conclusions from premises.

2. There are some drawbacks to economics that are annoying, but acceptable. For example, economists assume people are always chasing profits. Her response is to say that this narrow focus tends to yield interesting insights. McCloskey identifies other drawbacks of economics and economists, such as professional arrogance, but asks that we forgive those because they really aren’t that bad.

3. McCloskey identifies two horrible, unforgivable economic sins – (a) economists tend to prove qualitative mathematical theorems whose conclusions depend on arbitrary, qualitative premises and (b) statistical analyses routinely confuse statistiscal and substantial significance.

These are pretty weighty charges – that much theoretical economic work is just a useless game and economists (among others) make routine statistical errors no decent statistics undergrad would ever make.

How to respond? I’m not a professional economist – so I can’t speak for the economics profession, but I think the second charge – misunderstanding of significance – is right on target. I’ve told students and colleagues many times that “not significant” does not mean “no effect.” It simply means that you can’t automatically reject the hypothesis that, according to an arbitrary standard, there is no effect, which is different than saying there really is no effect. As McCloskey says, significance is simply a measure of confidence in the effect’s measurement. The whole situation is quite bad. For a summary of anti-significance test views, see the book “What if there were no significance tests?”

This first charge doesn’t bother me too much. All academic endeavors must engage in thought experiments. In fact, bizarre, unrealistic thought experiments can lead to some great insights. But what McCloskey, I think, really focuses on is the lack of empirical discipline. That is to say, when you come up with the premise of your theorem, it should be well justified.

When I studied math, there was a real difference between how mathematicians did it and how physicists did it. Math people are purely concerned with what is logically possible (does B really follow from A?) but physicists employ “physical intuition” – a sense of what assumptions were appropriate, a gut feeling developed from doing lots of experiments and observation. That’s why a lot of physics seems mysterious to mathematicians – the math looks familiar, but why did the physics people choose mathematical model X over Y? McCloskey’s point could be rephrased as saying that economists should move away from the mathematician’s style of modelling (proving what is logically possible) to the physics style of modeling (developing models inspired and constrained by observation and experiment).


Comments for this post are closed