OMB records memo

It's Already on File: How Administrative Records Can Help Assess Mobility

Jeff tries to explain how government works to readers of Science.

The motto of the U.S. Census Bureau, the government’s de facto statistical agency, is “Measuring America—People, Places, and Our Economy.” As assistant director  for research and methodology at the Census Bureau, Ron Jarmin tries not only to improve how those measurements are done but also how the outside research community can make use of the data once they’ve been collected.

The rest of the government also collects vast amounts of data on Americans in the course of doing its job. But officials at the so-called mission agencies—labor, justice, treasury, education, health, housing, and so on—have historically paid relatively little attention to how researchers outside the government might be able to use their data.

That could be changing, however. A 14 February memo from Sylvia Burwell, then the director of the White House’s Office of Management and Budget (OMB), urges agencies to make better use of the massive amounts of economic and demographic data they routinely collect—so long as their actions do not jeopardize privacy or undermine the confidentiality of the records themselves. U.S. researchers who use government data hope that it’s the first step toward one-stop shopping for these so-called administrative records.

The new initiative is part of the administration’s broader effort to make the government work better and smarter. Specifically, officials hope the additional number crunching will allow them to better evaluate existing programs, address new challenges, and save money by reducing the need to collect data already in some other government file. OMB plans to help agencies get over the hump by highlighting successful practices across the government and providing model agreements to foster collaboration among departments.

The initiative also coincides with a new push to increase outside access to the government’s social and economic data.  The Council of Professional Associations on Federal Statistics, a coalition of organizations that rely on federal statistics, has funding from the Alfred P. Sloan Foundation for a survey of how agencies now manage administrative records, says Kitty Smith, the council’s executive director who for many years led the Economic Research Service, the chief statistical agency within the U.S. Department of Agriculture. The next step will be to ask scientists to identify the most valuable agency data sets, followed by discussions on what it would take to make them available. The OMB memo could grease the skids for such cooperative activities, Smith adds.

The OMB memo focuses on what it calls nonpublic administrative records. That includes everything from income and earnings data submitted to the Internal Revenue Service and the Social Security Administration to what individuals provide when they apply for housing, nutrition, education, agriculture, and any number of federal assistance programs.

Most of that information is collected under strict rules designed to protect the privacy of any named individuals or companies and to ensure that the records themselves remain off-limits to anyone without the proper security clearance. And Burwell, who last month was nominated to lead the Department of Health and Human Services, emphasizes the importance of “fully respecting privacy and protecting confidentiality” of those records.

The need to balance “the legal prohibitions and the opportunities” of working with administrative records is a long-running issue for federal agencies, says a senior OMB official, who agreed to talk with ScienceInsider on background. Statistical agencies have had much more experience in knowing where to draw the line, the official notes, citing as an example the cautious process the Census Bureau followed more than 2 decades ago in creating a network of secure data centers, where outsiders can access sensitive data under tight supervision.

The first center was opened within the bureau’s headquarters in Suitland, Maryland, the official explains, the second at the agency’s regional center in Boston. Eventually came a university-based center, and now there are more than a dozen operating around the country. “The point is that it was taken one step at a time,” the official says. “And so far everybody is very pleased with their success.”

The new directive assumes that agencies will take a similarly cautious approach when they undertake statistical analyses of their existing records. “The boon of technology is that we can do things we could never do before, like record matching and record linkages,” the OMB official says. “But the bane is that the same technology we use to good effect could be used by those of less goodwill, to do harm, or at least do the kinds of things that were not the intention of those who collected this information.”

“Our compact with the American people says that we will maintain the confidentiality of the information,” the OMB official emphasizes. “And the agencies are responsible for making sure that happens. It’s not just a legal framework, it’s a bond we have with the public.”

Social science researchers say they take that bond seriously, and live in constant fear that there might be a breach. “Researchers would always like less restrictive access,” says Miles Corak, an economist at the University of Ottawa who has helped develop data sets on social mobility while working for Statistics Canada. “But you have to appreciate that if there’s just one mistake, or one time that data are disclosed, it ruins it for everybody, forever. So you can understand why the government is so risk averse.”

U.S. social scientists say the biggest threat to expanded use of administrative records could come not from the agencies that manage them but from Congress. Some legislators think that the government is already collecting too much information, they note. Researchers worry that politicians might overreact to any move toward making more linkages by attempting to ban existing access, including those arrangements that have led to a better understanding of social phenomena.

Such an outcome “would be about as far away from evidence-based policy as you can get,” says Tim Smeeding, an economist at the University of Wisconsin, Madison. “It’s like saying, ‘Let’s just hide the evidence and not give people access to it.’ That makes no sense to me.”

*Correction, 4 June, 10:40 a.m.: Timothy Smeeding was incorrectly identified as a sociologist in an earlier version of the story.

See also: