Monday, May 28, 2012

A Decent Denominator

In this post I discussed a how a few measures people use for talking about stimulus and austerity. While I think there are definite risks for some of the things I pointed out, to a large degree it is my own preferences around clear exposition more than something that is actually "wrong".

One thing I didn't like was expressing things as a percent of GDP. That's fine for talking about a lot of things, but doesn't work as well when we're talking about the business cycle because both the denominator and the numerator are changing (and even worse - it's precisely the relationship between the two that we're usually interested in!). You can get some perverse results from that.

One thing I've never seen before - but should work pretty well - is measuring things like governmenet spending as a percent of potential GDP. That should no give you any perverse results and should be more easily interpreted.

Krugman has an example here.


  1. the problem with 'testing' predictions against potential gdp is that potential gdp is a theoretical construct which assumes the validity of certain theories that underlay the predictions. The logic of the output gap makes assumptions about structural conditions prior to the crash which are the very things in dispute in a lot of cases.

  2. Daniel Kuehn: Have you ever heard of the Observatory for Economic Complexity? It's a joint MIT-Harvard institute that has developed a measure that is supposedly more accurate than World Bank estimates for predicting GDP growth.


All anonymous comments will be deleted. Consistent pseudonyms are fine.