Friday, July 29, 2005

To the econ dorks in the room...

Have you ever seen a model with an Rsquared of 1? Most likely not, so is there ever a case that you don't have omitted variable bias?

Have you ever seen a true instrumental variable that you couldn't pick apart?

This is the result of 8 econ students/dorks having a bbq & drinking too many Sierra Nevadas.

2 Comments:

Blogger Joseph Schultz said...

I've seen an R-squared of one - but it involves predicting the number of cups in a gallon of milk.

In practice, you'll always have some bias. The trick is to minimize it as much as possible - or just state that it's there and ignore it.

7:52 PM  
Blogger Priscillia said...

Yeah, I'm with you Joseph, but there were some perfectionists in the group that weren't satisfied so I wanted to inspire them with some model that explained 100%. I think my friends are doomed for insanity.

9:24 PM  

Post a Comment

<< Home