Organizing to Count: Change in the Federal Statistical System. - book reviews

Business Economics, Jan, 1996 by Michael J. Boskin

In this book, Janet Norwood brings her insight I and experience to bear on the issue of the best way to organize the collection and dissemination of federal statistics. The U.S. statistical system is decentralized, with major pieces strewn across various agencies and different cabinet departments. Oversight and appropriations is likewise spread among different Congressional appropriation subcommittees. Norwood identifies this fragmentation as the first serious issue. She notes, "We are unwilling to invest the power and resources required for coordination and oversight of the statistical system, and yet we have a tendency to multiply agencies to produce statistics."

The second key issue is that a great deal of data collection is now being done outside the government agencies with primary statistical responsibilities, which causes problems in review and quality control. As Norwood states, "...both public and private users of data find it increasingly difficult to get the data they need, and, when they are successful in their search, too often the data seem impossible to use."

The third major issue is a byproduct of the fragmentation; interaction with other countries and international organizations is also fragmented.

Norwood properly notes that, while most of the attention has focused on the budgets of statistical agencies, "we have problems in allocating the dollars already appropriated and ... we could do a much better job with what we have." Organizing to Count is therefore an analysis of whether radical reorganization of the statistical system, or at least some important changes therein, "could make the system function more effectively and efficiently."

All this might seem an arcane minor issue, even to readers of Business Economics, but it surely is not. As Norwood notes, "information produced by the federal statistical system affects Americans in every aspect of their lives." From price statistics used to index government benefit programs and tax rules to labor markets, productivity, income, and other core economic statistics, used by policymakers and the public to gauge the performance of the economy to population, health, environment, and other statistics used for other public and private purposes, these data affect millions of daily decisions. Hence, their accuracy, timeliness, and reliability are a matter of great national concern.

Norwood weaves an easy to follow trail through the federal statistical system, agency by agency. She also reviews the recommendations for or against consolidation of sixteen committees, commissions and study groups starting in 1903 and ending with the Boskin Working Group in 1990.

Norwood makes a strong case for a new consolidated independent statistical system, the Central Statistical Board (CSB), headed by a Chief Statistician. The next level down would be a set of five directors: one each for statistical policy, censuses and surveys, the decennial census, labor statistics, and the national accounts. Organizationally, each of the major topical and activity issues would fall under one or another of the directorates. For example, household surveys would be under the Director for Censuses and Surveys; the maintenance of lists would be under the Director for Statistical Policy; price and employment data under the Director for Labor Statistics; the GDP and International Accounts under the Director for National Accounts. Norwood believes the new CSB "would make it possible to develop and carry out a comprehensive, systematic effort to combine surveys and develop economies of scale." The CSB would function as an independent commission outside current cabinet departments. In addition to the work of the Census, BLS and BEA, it would coordinate work done by seven other statistical agencies and by programmatic agencies outside the CSB.

Norwood makes a most persuasive case that the new CSB would solve the major issues she addresses: fragmentation, review and quality control, and international interaction. She also develops a less sweeping fall-back proposal: the National Statistical Improvement Act. This would standardize confidentiality, protection and exchange of microdata, standardize appointments, standardize statistical agency status within the cabinet departments; codify procedures for release of economic indicators; and strengthen the role of OMB's chief statistician.

All of these objectives are laudable , and no doubt some version of a National Statistical Improvement Act would be desirable. In fact, it may even be necessary once the fallout from the current round of budget proposals plays itself out.

The one thing lacking in the book is a serious discussion of the problems of the data themselves. Our statistics are falling further and further behind in their ability to measure economic performance. Changes in the CPI substantially overestimate changes in the true cost of living, partly because of quality change and new products. These factors also render our measures of real output suspect, etc.