Thursday, November 12, 2009

Ginning up fake numbers on the stimulus

The trend continues of local journalists digging into the numbers provided by the White House of how many jobs the stimulus provided and finding the numbers full of holes. The Boston Globe has found strongly inflated numbers for Massachusetts.
While Massachusetts recipients of federal stimulus money collectively report 12,374 jobs saved or created, a Globe review shows that number is wildly exaggerated. Organizations that received stimulus money miscounted jobs, filed erroneous figures, or claimed jobs for work that has not yet started.

The Globe’s finding is based on the federal government’s just-released accounts of stimulus spending at the end of October. It lists the nearly $4 billion in stimulus awards made to an array of Massachusetts government agencies, universities, hospitals, private businesses, and nonprofit organizations, and notes how many jobs each created or saved.

But in interviews with recipients, the Globe found that several openly acknowledged creating far fewer jobs than they have been credited for.

One of the largest reported jobs figures comes from Bridgewater State College, which is listed as using $77,181 in stimulus money for 160 full-time work-study jobs for students. But Bridgewater State spokesman Bryan Baldwin said the college made a mistake and the actual number of new jobs was “almost nothing.’’ Bridgewater has submitted a correction, but it is not yet reflected in the report.

In other cases, federal money that recipients already receive annually - subsidies for affordable housing, for example - was reclassified this year as stimulus spending, and the existing jobs already supported by those programs were credited to stimulus spending. Some of these recipients said they did not even know the money they were getting was classified as stimulus funds until September, when federal officials told them they had to file reports.

“There were no jobs created. It was just shuffling around of the funds,’’ said Susan Kelly, director of property management for Boston Land Co., which reported retaining 26 jobs with $2.7 million in rental subsidies for its affordable housing developments in Waltham. “It’s hard to figure out if you did the paperwork right. We never asked for this.’’

The federal stimulus report for Massachusetts has so many errors, missing data, or estimates instead of actual job counts that it may be impossible to accurately tally how many people have been employed by the massive infusion of federal money. Massachusetts is expected to receive an estimated $1 billion more in stimulus contracts, grants, and loans.
Remember these were the numbers that the White House's own guy in charge of compiling the data joked that he'd scrubbed the numbers so hard that he had "dishpan hands."

And these are the results that they came up with when they were really trying to fix all the mistakes. Imagine that.

This follows a history of the Obama administration's economists and statisticians coming up with very suspect numbers. Samuel Staley had a critical take (subscription only) on CEA head Christine Romer's use of deceptively precise statistics to sell the stimulus package in the first place.
Beginning with a report written by Christina Romer, the current chairwoman of the President’s Council of Economic Advisers, and Jared Bernstein, now Vice President Biden’s chief economist, the White House trotted out numbers based on statistically naïve and simplistic methods in order to drum up support for the stimulus package.

Romer and Bernstein’s initial report, released prior to the president’s inauguration in January 2009, estimated that a hypothetical $775 billion stimulus plan that included government spending and tax cuts would generate jobs in a “reasonable range” of 3.3 to 4.1 million by the end of 2010. Perhaps in an effort to preserve some academic credibility, Romer and Bernstein were somewhat circumspect in their analysis: “Our estimates of economic relationships and rules of thumb are derived from historical experience and so will not apply exactly to any given episode. Furthermore, the uncertainty is surely higher than normal now because the current recession is unusual both in its fundamental causes and severity.”

If the story stopped here, few people would cry foul. Romer and Bernstein’s analysis was simply one more report issued as policymakers and academics scrambled to figure out what to do about the economy. Analysis from all quarters is useful in determining macro policy, particularly during periods of extraordinary uncertainty.

But it didn’t stop there, and that’s where questions of professional ethics arise. A well-known principle of statistics is that the accuracy of estimates falls the more specific, narrow, or infrequent the data become. Predicting economy-wide trends is easier than predicting what will happen to individual households. Even predicting statewide economic performance is tricky. Predicting local trends from macroeconomic analysis is almost impossible. These processes are fraught with all sorts of methodological land mines, not the least of which is the idiosyncratic nature of local economic growth.
But Romer and Bernstein didn't stop there. They went on to predict with fake accuracy how many jobs would be created by the stimulus in every single congressional district. There is absolutely no way that any self-respecting economist or statistician could predict with such precision how the stimulus would work in specific districts. But they needed those numbers in order to pressure politicians to vote for the stimulus.
If these estimates seem fishy because they fall into a very narrow range (between 6,500 and 9,200 for all districts in California), they should. Essentially, the job growth is distributed based on population with little or no analysis of or adjustment for local conditions, historical trends, and other economic strengths and weaknesses. Not surprisingly, no congressional district in the nation would be worse off than before the stimulus.

It’s almost as if White House economists were betting no one would examine their results closely. And they were right. Only a few conservative and libertarian bloggers (including the Reason Foundation) noticed the statistical sleight of hand.
And now they're compounding their original sin of ginning up fake numbers by ginning up more fake numbers about how many jobs have been "created or saved" when everyone acknowledges that there is absolutely no way to count "saved" jobs.
At best, the White House can be accused of adopting a naïve approach to economic analysis and statistics. At worst, its economic forecasting was a cynical misuse of applied economic analysis to manipulate people into supporting a short-term political goal. Given the high caliber of his economic team, the American public may legitimately ask whether the latter interpretation is more accurate. Either way, this misuse of data and analysis isn’t the kind of change the American people voted for when they bought into presidential Candidate Obama’s campaign message to “turn the page” on politics as usual.
The bill is coming due now. We aren't seeing the jobs that the stimulus supporters promised us. And now we're seeing that they have absolutely no way of counting the jobs that were supposedly created or saved by the stimulus. And the only debate is whether these people like Romer and Bernstein who had respectable reputations before they took on their present days are just being ignorant or if they've totally sold out.