Did the Census Bureau Really Fake the Jobs Report?

The New York Post makes a shocking claim. But even if it's right, the fraud was too imperceptibly small to make any a difference. 


Remember this?

Back in the waning days of the 2012 election, former GE CEO Jack Welch caused a brief stir when he casually suggested via Twitter that the Obama administration was manipulating the monthly jobs report for political gain. There didn't seem to be much substance behind the allegations.

Late last night, however, New York Post columnist John Crudele published an article alleging he had found evidence that the jobs report really was cooked. In 2010, Crudele reports, the Census Bureau caught an employee named Julius Buckmon, who conducted surveys around the Philadelphia region, faking data used to calculate the national unemployment rate. According to Crudele, the Census only led a perfunctory investigation of Buckmon. And it may have been ignoring a much larger, more pernicious problem:

a knowledgeable source says the deception went beyond that one employee — that it escalated at the time President Obama was seeking reelection in 2012 and continues today.

“He’s not the only one,” said the source, who asked to remain anonymous for now but is willing to talk with the Labor Department and Congress if asked.

Buckmon says that his superiors had instructed him to fabricate survey answers if he couldn't collect enough of them to hit a desired response rate. “It was a phone conversation — I forget the exact words — but it was, ‘Go ahead and fabricate it’ to make it what it was,” Buckmon told Crudele.

Crudele's reporting may well be airtight. At this point, I'm certainly not in a position to poke holes in his facts. However, his conclusion that the final unemployment rate was being "manipulated," which is to say, substantively changed, is a little premature.

To get why takes some background, but I promise to make it brief. The Bureau of Labor Statistics bases its monthly jobs report on two different surveys. The First is the household survey, which is actually run by the Census (this is the one Crudele thinks was manipulated). It calls Americans and asks about their work status to calculate the unemployment rate. Second is the payroll survey. It's conducted entirely by the BLS and state agencies and asks companies how many new workers they hired or fired each month.

When you hear the economy created 185,000 jobs in a month, that's the BLS. When you hear the unemployment rate fell, that's the Census.

But here's the key bit: Both surveys estimate the total number of employed Americans. And while they might show different rates of job growth in any given month, over the long run, their changes tend mirror each other, as shown in the graph below. (Red is the household survey, blue is the payroll survey).

If the Census were really cooking its books to falsely lower the unemployment rate, you would expect its figures to diverge from the honest BLS survey. But that didn't really happen. (The data is shown by quarter, to smooth some of the month-to-month variation).

According to both the payroll and household surveys, job growth averaged about 0.1 percent per month from 2010 through the election. The household survey showed slightly faster job growth than the payroll survey in 2012, but slightly slower growth in 2011. Ultimately, if the household survey really was adulterated with fake data, the impact seems to have been too subtle to notice.

Again, Crudele may have uncovered an instance of actual incompetence or wrongdoing at Census. The BLS has reported it to the Commerce Department's inspector general for further investigation. But the real scandal with the jobs report probably isn't that the statistics are fake. It's that the numbers are growing far, far too slowly.