Wednesday, February 17, 2010

What we don't know 

Commenting on Thomas Friedman's suggestion for a 50-page report on global warming to be called "What We Know", Russ Roberts reminds us to be careful about what we say about the stimulus as well:

Most of what Thomas Friedman thinks we know is based on multiple regression analysis trying to hold other factors constant other than human carbon emissions and making a variety of assumptions about the interactions between those factors along with the factors we cannot measure. That is hard to explain to a sixth grader. It can be done But it�s not knowledge. It�s an attempt to gain knowledge.

It is very similar to writing a report for a sixth grader on how the stimulus turned out. We have fewer jobs than we had before. That�s what we know. But even I, a skeptic, wouldn�t call that knowledge about whether the �stimulus� package worked or not. But I wouldn�t use the CBO estimate either. The CBO estimate is a repeat of the forecast it made before the legislation passed. We don�t know how many jobs were created or destroyed by the legislation.

Most of what we write about the effects of stimulus are just that, "an attempt to gain knowledge." A bureaucrat writes down some numbers. Reporters and bloggers find flaws. Econometric models estimate the effects, but those models were used to propose the policy put in place. It's not likely those models would go back and say the proposed plan didn't work: Econometric models aren't built to do that: If the model has as a premise that future government spending will create jobs, it isn't going to tell you that past government spending did not. Meanwhile, those in political opposition will look to find contradictions when none really exist. (GDP growth can lead employment growth.) And people get angrier and cynical.

There is nothing wrong with saying we don't know. It might have worked; it might not have. What we know is there are between three and four million fewer jobs than a year ago, and the deficit is larger. We want to know more. We are trying to know more. And if the volume of studies since 2000 of the Great Depression are any indication, we'll still want to know more a century from now.