When Canadians look for official statistics, there’s one agency, Statistics Canada, that provides basically all of them. In the United States, on the other hand, we have a hodgepodge of agencies—the Bureau of Justice Statistics for data on crime and law enforcement, the Bureau of Transportation Statistics for data on transportation, the National Center for Education Statistics for data on schools and academic performance, and so forth—13 “principal” federal statistical agencies in all.
Why is the American system so fragmented? There’s an element of history; presumably, each department established its own statistical agency to meet the department’s need. There’s also a large element of inertia—the House and Senate each assign the work on appropriation bills to subcommittees that are responsible for specific departments, so no individual appropriation subcommittee is responsible for looking at the system in its entirety.
How do other countries organize things? Most of them have moved toward a unified statistical agency, following the Canadian model. For example, in the UK the statistical agency is the Office for National Statistics; in France it’s Insee; Italy has Istat; and Australia has the Australian Bureau of Statistics. Japan’s statistical system, on the other hand, is like the U.S. system with multiple statistical agencies. Germany’s system is also decentralized, albeit along a different dimension. Most of the responsibility for collecting German data is assigned to the state government agencies, while the Federal Statistical Office‘s role is to coordinate with the states and to compile data needed specifically for federal purposes.
Has the U.S. ever considered consolidating its statistical agencies? Yes, many times. For example, in 2012 Obama recommended consolidating the Census Bureau, BLS, and BEA as part of a larger streamlining and consolidation plan. Although the plan went nowhere with Congress, he revived the plan in 2015 as part of his 2016 budget proposal.
If there isn’t any real momentum to consolidate the U.S. statistical agencies, do we have any idea of how much waste or duplication is entailed in having multiple agencies? While there aren’t any definitive estimates, my guess is that there’s less duplication than one might think. The main example of duplication is the fact that the Census Bureau and BLS each must maintain separate business registers—that is, the lists of businesses that are operating in the country from which samples of businesses are drawn to receive one of the surveys that the government uses to collect economic data. If the agencies were consolidated, or if legislation allowing for more data sharing between the agencies were enacted, they would be able to share a single business register, which would save the taxpayer several million dollars.
Other than the business registers, however, there aren’t a lot of other obviously duplicative statistical programs. Within OMB, a relatively small statistical policy office is responsible for coordinating statistical policy and approving all data collection requests. The head of that office is the U.S. Chief Statistician, a civil service position, which was recently filled by Nancy Potok (former Deputy Director of the Census Bureau).
The U.S. statistical agencies devote quite a bit of effort to coordinating their activities, and part of that is taking advantage of agency specialization. For example, the Census Bureau, the agency with the deepest survey operations, undertakes about $250 million a year in reimbursable survey work for other agencies. The statisticians and economists who work for the agencies maintain professional networks to share research and methodological information across agencies. And there are several cross-agency committees, such as the Federal Economic Statistics Advisory Committee, which focus on cross-cutting work.
Personally, I think the U.S. would be better off with a consolidated statistical agency. While the agencies currently do coordinate with each other, the coordination would be easier if it took place within one agency. Budgets could also focus on a common set of priorities, and consolidation would make it easier for staff to share data and methodological information. Washington policy makers are seldom able to focus their attention on making organizational changes that would streamline government and make it work better, but if they could, consolidating and streamlining the statistical system would make a lot of sense.
Perhaps you meant $250 million for the Census Bureau operations? $250 billion seems rather exorbitant, even by USG standards
Good catch, I’ve corrected it. I guess working in national accounts I got in the habit of talking about billions.
A useful step might be to consolidate all the statistical reports into a single website, so that someone looking for US government statistics could go to a single place for all of them. At the same time they could also be kept within the agency of origin.
This would improve access to the information without compromising the quality of the production process.
Your site is an important contribution.