On Wednesday, a New York Post economics and business writer's column was headlined "Census ‘Faked’ 2012 Election Jobs Report" New York Post original story: Just two years before the presidential election, the Census Bureau had caught an employee fabricating data that went into the unemployment report, which is one of the most closely watched measures of the economy. And 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. One of the employees was let go in 2010, but there are other sources implied in the story, that the process was pervasive. This eventually led to a Census Bureau statement: The Census Bureau takes allegations of fraud by its employees very seriously. Fabrication of data by an employee is grounds for disciplinary action, including dismissal and possible criminal action. We have no reason to believe that there was a systematic manipulation of the data described in media reports. As a statistical agency, the Census Bureau is very conscientious about our responsibility to produce accurate Current Population Survey data for the Bureau of Labor Statistics and all other surveys we conduct. We carefully cross check and verify the work of our staff to ensure the data's validity, including random quality control monitoring. That monitoring process includes reinterviewing respondents, and rechecking the data an employee has submitted, looking for red flags that indicate possible fabrication, such as abnormally short lengths of interviews or higher survey completion rates that are out of sync with normal survey collection productivity levels. That is why when we learned of the allegations of fabricated Current Population survey results, we immediately reported them to the Office of the Inspector General. Business Insider put it in perspective:
  • The headline doesn't deliver in the story what it said it would
  • The article does not say there was a conspiracy
  • The September 2012 unemployment report looked weird, but with data since published, it looks on trend
  • The problem was focused in one region
Business Insider is waiting for more information. The unemployment data are bad enough without fabrication of data. Anyone who reads the unemployment reports for the past two years has been frustrated with the unemployment rate improving almost every month solely because of contraction of the workforce. That is, the unemployment rate has been improving more because people are giving up the job searches than there are people being newly employed. These data are from the household survey which includes freelance and self-employed workers. We will not know what happened for quite a while, and speculation that one employee can go adrift on their own based on what they think someone in the administration wants is not as unbelievable as it might seem in light of the recent IRS scandal. In that case, it was not the faking of data, it was the deliberate and selective slowdown of the bureaucracy. That, too, started with the blame of a single employee, that time in Cincinnati. It is unlikely that this is the case here. It's hard to have one area's survey data be out of line with its own trends and that of other areas. Besides that, the Bureau of Labor Statistics has another survey which tracks payroll workers. Those data are always checked against Social Security tax filings which are (gasp!) actual data that counts up the actual real people with payroll jobs. While payroll data do not include freelancers and self-employed, even the data from their quarterly tax payments can be added to the BLS payroll data to help determine if survey results from the offices in question make no sense or look suspicious. This is not much ado about nothing. The Bureau of Labor Statistics, no matter what the administration, has always been a diligent data collector and manager, whose work represents some of the best statistical work in the world. To have that tarnished is a great disappointment because their data have always been relied on by government, businesses, and citizens. While there has always been room for interpretation, the BLS data and process is well-documented to a high degree. Those who have methodological disagreements with the BLS have enough data from the BLS itself to determine the extent of their disagreements. I hope these matters are resolved soon and that the investigation shows it is one or a few employees or at worst one office. In the meantime, what should users of BLS data do? There are other sources, such as ADP and Gallup, that have been publishing data about employment for years. Perhaps now is a good time to review their efforts and consider adding them to the process of determining the health of national employment. # # #