Last week, after the Bureau of Labor Statistics announced an unexpected decline in September's unemployment rate, former General Electric Chief Executive Officer Jack Welch questioned the credibility of the data.

In an attempt to influence the presidential campaign, he suggested, the Obama administration must have manipulated the bureau's report, which showed the jobless rate at 7.8 percent, a statistically significant drop from 8.1 percent in August. "These Chicago guys will do anything," Welch wrote in a Twitter message, "can't debate so change numbers."

Welch's comments demonstrate a stunning degree of ignorance and recklessness, and impugn his own credibility.

If he had been better informed, he might have said something useful about the data. Drawing from his own experience, he could have raised a legitimate point about how the employment statistics might be improved.

As CEO of GE, Welch increased the use of real-time statistics to improve GE's products and services. Such an approach could be used to strengthen our measures of the labor market and other macroeconomic indicators.

To be sure, the bureau's estimates, while unbiased, could be substantially improved. Every month, the agency gathers data from about 140,000 businesses and government departments as well as 60,000 households. Yet this is an outdated approach to compiling macroeconomic information; relying on household and business surveys means that we are not taking advantage of the commercial, administrative and other data being collected at an exploding rate.

The problems with the bureau's current approach are well- known. The BLS itself has trenchantly analyzed the shortcomings of its various labor-market surveys, and regularly revises its estimates - for instance, as more accurate unemployment- insurance records become available.

The agency has also published detailed statistics about the margin of error in its conclusions. One of the employment figures Welch focused on has a 90 percent confidence range amounting to more than 1 million people.

How can we do better? By also using the amazing volume of data being gathered elsewhere. Google, for example, makes information available about particular search terms that correlate with the state of the economy. The U.S. should harness such "big data" in its official statistics.

With increasingly powerful computing systems, the price of data storage plummeting and tight public resources, the government's statistical agencies should start combining the data collected for commercial and administrative purposes with government surveys to measure the state of the economy faster and better.

The Obama administration is investing $200 million to see if this approach could improve the quality of government data and lower the cost of obtaining it. As Alan Krueger, the chairman of President Barack Obama's Council of Economic Advisers, has said, "there is enormous potential to use these Big Data sets to cut survey costs and reduce respondent burden, to improve and expand our existing social and economic indicators and make them more timely, to assess the reliability of traditional survey data, to study network and GIS-related issues, and to answer a myriad of previously unanswerable questions."

Soon after making his rash comment, Welch admitted that he had no evidence for his charges that the BLS had, for political purposes, cooked the unemployment data. By then, of course, the damage had been done.

At this point, Welch should formally apologize to the dedicated public servants at the bureau and devote some time and resources to helping make government statistics more accurate by incorporating additional data from the private sector.


Jonathan Orszag, a senior managing director at Compass Lexecon and a senior fellow at the Center for American Progress, was an economic-policy adviser on President Bill Clinton's National Economic Council. Peter Orszag is vice chairman of corporate and investment banking at Citigroup and a former director of the Office of Management and Budget in the Obama administration.