Using statistical slight of hand to accomplish political goals is one of the fastest ways to create cynacism in the political system. No better example of that may be playing out as the press reports the so-called job creation from the stimulus plan.
Take the following lead paragraph from a USA Today article (October 27, 2009) on the release of the state job creation data:
States have reported using stimulus money to create or save more than 388,000 jobs so far this year, buttressing the Obama administration's claim that the $787 billion plan has had a significant impact on the economy.
That total, based on a USA TODAY review of reports from 33 states and Puerto Rico, includes teachers, construction workers, and others whose jobs were funded by stimulus money awarded to states. The administration plans Friday to release reports from all 50 states, providing the broadest accounting yet of the stimulus plan's impact.
What's the problem with this?
The numbers of jobs created or "saved" are simply counts provided by state agencies spending stimulus money. They simply record the number of people hired under the contract or for the project. They are not the result of investigative follow up, or a consistent methdology for identifying real jobs created or saved. (Indeed, these methodological problems have plagued economic development program evaluations for decades as states have claimed jobs were created by various tax incentive programs but no real way to verify the accuracy of the numbers.)
As a practical matter, it's virtually impossible to distinguish between a job saved or created. (I discuss this problem in a forthcoming feature article in the 2 November 2009 issue of National Review titled "Naive Statistics.") How do you know, for example, that an construction company hiring a worker for a sidewalk repair project is hiring someone that would not have gotten a job somewhere else? Or that the worker would have been laid off and unable to find another job? In truth, you can't, and the agencies reporting these job counts don't try. (It would, in fact, be very costly to do this.)
At least some people are honest about this, as the USA Today story reports at the end of the article:
The states' reports suggest the biggest impact has been at schools. Twenty-three states that have reported school job numbers said more than 156,000 jobs had been created or saved.
Carol Bingham, director of fiscal policy for the California Department of Education, estimated the stimulus saved about 20,000 teaching positions. But she and others warn that precisely counting saved jobs has proved almost impossible. "It was intended to be a count. The way it was done, I think it's going to end up being an estimate," she said.
Indiana officials reported that the stimulus had created or saved about 13,000 school jobs. Asked whether he had any idea how many layoffs the plan had prevented, state Education Department spokesman Cam Savage replied: "I really don't."
There's the rub: If the jobs reporedly created or saved are estimates, and you don't really know how many people would have been layed off without the money, how can anyone claim that the program is working? (See also my blog post on 16 October on this subject here.)