There's no accounting for job creation

by on January 13, 2010 · 3 comments

An Associated Press story posted at 9:31 pm last night by Brett J. Blackledge notes that the Obama adminsitration “has abandoned its controversial method of counting jobs under President Barack Obama‘s economic stimulus, making it impossible to track the number of jobs saved or created with the $787 billion in recovery money.” Recipients of federal money were originally supposed to report how many jobs they had “saved or created.” Now, they’re just supposed to report how many people got paid with the money. 

The “saved or created” approach was an attempt to estimate how the Recovery Act spending affected employment. Unfortunately, it was impossible for this approach to deliver what it promised, as I pointed out in testimony to Congress last May. To know how many jobs the Recovery Act spending saved or created, we need to control for other factors that affect employment. We also need to control for the employment effects of the borrowing (and future taxing) that pays for the spending.

This is what good “macroeconomic” analysis does. It is not what employers do when they put people on the payroll.  Asking employers who received federal money to tell us how many jobs were created or saved is asking them to give us information that they do not have and cannot acquire.

Under the new approach, employers will report quarterly how many people they are paying with the federal money. Since the reports will be quarterly, some jobs may be counted more than once — I guess as much as four times a year. According to the AP story, some Republicans are complaining that the new approach will lead to even larger and more misleading job numbers.

The numbers will likely be higher. But they will not necessarily be misleading, as long as makes clear that these are simply reports on the number of people paid with the federal money, not an attempt to measure the number of jobs created.

Estimating jobs created is work for serious macroeconomists who try to control for other factors affecting the results. There’s of course plenty of room for disagreement over how to do this, as this commentary by my Mercatus Center colleagues Garrett Jones and Veronique de Rugy demonstrates.

The administration’s decision demonstrates once again the old adage that data is not knowledge.

Previous post:

Next post: