October 20, 2014

OPEN SOURCE IT ALL:

Open source the CBO! (Pascal-Emmanuel Gobry, 10/20/14, The Week)

[H]ow did the CBO know the fiscal impact of ObamaCare 10 years from now? Well, to forecast that with any certainty, the CBO would have to know the GDP 10 years from now, and the tax policy 10 years from now, and other critical factors. Of course, it's literally impossible to do these things. Which is why the CBO builds mathematical models with all sorts of assumptions into them that spit out a "score." Like pretty much all predictors of future things, the CBO has to do some (educated) guesswork. And indeed, in the case of ObamaCare, the task is so difficult that the CBO basically gave up earlier this year.

This is not to criticize the CBO or its work. While the efficacy of mathematical modeling in economics is overrated, it is also necessary. It is impossible to do this kind of forecasting without building certain assumptions into it.

It's good that we have the CBO, and it's good that it does the work it does. But there is an easy way to make it better.

Right now, the CBO's work is essentially a black box. The CBO spits out the "score," but nobody knows anything about the model that score is based on.

Like any model, the CBO's is built on questionable assumptions. As I said, that is a necessary part of the work. But it would also be helpful to the public to have the ability for other people to use different, equally legitimate assumptions and to put them into the CBO's model to get a different score.

It's just common sense. We should open source the CBO so that as many people as possible can look into the innards of its scores and arrive at their own conclusions, and perhaps debate other scores. The CBO's model isn't intrinsically better than anybody else's -- it's just the CBO's.

Don't stop with the CBO.  All government information should be publicly available.

Posted by at October 20, 2014 6:09 PM
  

blog comments powered by Disqus
« ...AND CHEAPER...: | Main | WALKING THE WALK: »