Hellenic Business

Even if the U.S. government wanted to manipulate monthly jobs figures, it would be impossible to accomplish, said a former head of the U.S. government’s labor statistics agency.

Accusations that the government had manipulated the latest employment report spread across Twitter and other forums Friday after the U.S. unemployment rate fell to its lowest level since President Barack Obama’s inauguration. Among those questioning the better-than-expected report was former General Electric Chief Executive Jack Welch, who tweeted the suggestion that the “unbelievable jobs numbers” were fabricated to help Obama’s electoral chances in next month’s presidential election.

But, Keith Hall, who served as Commissioner of the Bureau of Labor Statistics from 2008 until 2012, said in an interview Friday that there is no way someone at the agency could change any of the data from its two monthly employment surveys. The significant improvement in the unemployment rate may reflect normal statistical errors in the sampling process, he said, but that has nothing to do with manipulation.

“There’s nothing wrong with the numbers,” said Mr. Hall. “The only issue is the interpretation of the numbers. The numbers are what they are.”

For September, the politically important unemployment rate fell to 7.8% in September from 8.1% the prior month, according to the Labor Department. That was the lowest level since January 2009 and well below the 8.1% forecast of economists surveyed by Dow Jones Newswires. The unemployment rate estimate is derived from a survey of households, which came up with an estimate that 863,000 jobs were added for the month.

But the separate establishment survey from which the official payrolls number is derived reported a more modest seasonally adjusted gain of 114,000 jobs in September. That was below the consensus forecast of 118,000, though the previous two months were revised higher.

Mr. Hall said the inconsistent reports reflect the different samples used in the two surveys, one focused on households the other on businesses. The establishment survey has a huge sample size of 141,000 business and agencies covering 486,000 worksites, whereas the household survey covers just 60,000 homes.

“The household survey is much smaller. When you look at something like labor force and employment levels, the uncertainty of those numbers is much larger,” said Mr. Hall. “Within two months, the household survey could show the unemployment rate eking back up.”

The two surveys, which are conducted by career civil employees who often serve through multiple administrations, will often move in different directions on any given month. For example, the revised July figures show the economy added a solid 181,000 jobs, but in that same month the unemployment rate rose to 8.3% from 8.2% in June. Over the longer term, however, they have historically shown a predictable convergence.

“It is quite possible that the positive news could be reversed when the October jobless number is reported four days before the election,” said Sung Won Sohn, an economist at California State University, Channel Islands.

The Wall Street Journal