Wordpress

Recently, most of my posts have been going to the wordpress verion of this site. I haven't yet gotten a handle on how to synch the two - perhaps that will be forthcoming. In the meantime, new posts are here: http://calloftheloon.wordpress.com/

Manufacturing Statistics

I'm always skeptical when politicians start using economic statistics. Chiefly, this is because I don't believe they understand them. I don't mean that they don't understand them as in they don't know the precise definition of, say, GDP vs. GNP, but that they profoundly misunderstand what they are. To wit: economic statistics are an attempt to measure a latent, unobservable, platonic ideal. They are not themselves truth, but an attempt at measuring truth. So GDP is an estimate of the size of the economy, it is not itself the size of the economy.

Now great number of intelligent people spend a great deal of time constructing the best statistical definitions they can, ensuring that they are as consistent as possible over time in a changing world, and measuring the inputs as accurately as possible. This task is fundamentally incompatible with politics. Politics does not depend on getting the best measure but on getting the "right" measure. Right being whichever estimate best helps one's constituency.

A recent attempt to base politics on economic statistics is the current fetishization of "manufacturing." But what is manufacturing? I manufactured this blog post. You can print it out and manufacture a letter. Are the economic statistics themselves the output of economists who manufacture them? When you write laws around statistics, you quickly find that the statistics change to fit the laws. This is not good science. A good video (courtesy Greg Mankiw) makes the point by considering the question: Are hamburgers manufacturing?

(As an aside, I also think laws tying the budget to a measure of GDP are flawed for the same reason.)

Models

As an economist, I am a great fan of using mathematical models to describe the world and spend a great deal of my time in teaching explaining their advantages to students. They help clarify the thinking and force you to examine your assumptions. Furthermore, they allow you to make testable hypothesis which allows one to make use of the scientific method to determine the truth of some interesting fact about the world. Nevertheless, they do have their shortcomings. The classic, pithy phrase and rejoinder are:

"All models are wrong; but good models are useful."

I recently came across a good description of why one wants to use both model-based and stylized-fact-based approaches. It's from Alan Manning's chapter in the Handbook of Labor Economics, "Imperfect Competition in the Labor Market"

New developments are often thought of as departures from these canonical models. Although the use of very particular models encourages precise thinking, that precision relates to the models and not the world and can easily become spurious precision when the models are very abstract with assumptions designed more for analytical tractability than realism. So a model-based approach to the topic is not always helpful and this survey is based on the belief that it can be useful to think in very broad terms about general principles and that one can say useful things without having to couch them in a complete but necessarily very particular model.

Like any tool, models can be very useful, but much of the skill is in knowing the best tool to use for the problem at hand.