The Statistics Corner: Information from Tax Returns: The Statistics of Income Program of the Internal Revenue Service: Part 2
Petska, Tom, Scheuren, Fritz, Business Economics
This article is the second of two reviewing the projects and operations of the Statistics of Income (SOI) program of the Internal Revenue Service; the first part was published in the April 1992 issue. Although SOI is a relatively small and obscure statistical agency, SOI data are, nonetheless, a part of the bedrock of the U.S. statistical system and central to the understanding of the economy as a whole. Here we conclude with a description of short-term modernization of work processes and an overview of longer-term innovations geared toward shaping the future.
THE 1980s were a time of major change in SOI. As a relatively small part of IRS, statistical work has sometimes been given less attention and a lower priority than needed. In spite of growing requirements, the SOI budget has fluctuated in this period; currently, it resides below inflation-adjusted levels of a decade ago.
In 1980 the SOI program consisted of twenty-six projects; now, in 1992, this number has more than doubled to nearly sixty. This two-fold growth in programs was accompanied by a parallel four-fold increase in the amount of data extracted from the various tax and information returns. The growing SOI workload was absorbed in the early 1980s by efficiency gains and sample size reductions. By 1984, however, conventional means of increasing efficiency became (self) limiting. Gradually, other ways of achieving productivity gains were adopted, including a total quality organization (TQO) strategy.
SOI's statistical processing has historically been separate from the mainline processing of tax returns for administrative purposes. SOI operations begin by sampling from tax or information returns in the basic tax administrative (or master file) system, which offers a sampling frame that enables efficient sample designs to be used. After returns are sampled, data elements already captured for adminstrative purposes are used as a starting point in statistical processing. These data are augmented with other items from tax returns, tested for consistency, and identifiable errors or inconsistencies are resolved.
Until a few years ago, the basic SOI information processing was conducted in a "batch-mode" manner, in which several sites often had roles in processing data for each tax return. A system such as this denied "ownership" and accountability for field processors and was not conducive to maintaining high levels of quality.
To address these processing problems, SOI built its own network of minicomputers solely for statistical processing. This new system, which has just been fully implemented, uses an on-line, real-time approach, so that all data processing is completed in a one-pass operation. In addition to reducing handling costs and removing overlapping responsibilities, accountability and ownership improve, because one person is now responsible for assuring the validity of all data processing for any tax return.
The SOI minicomputer network electronically links all SOI national office and field operations, so that data can be efficiently transferred between locations. This capability enables "experts," wherever located, to monitor processing better and to accelerate efforts to attain higher standards of quality.
Many other recent innovations in statistical techniques have been introduced or enhanced by the minicomputer network. SOI has an ongoing interest in developing improved methodologies. Towards this goal, research conducted each year on these methodological improvements appears in papers given at professional meetings, such as the annual meetings of the American Statistical Association. Routinely, these papers are compiled and published in an SOI methodology series.(1)
RESTRUCTURING MANAGEMENT PRACTICES
Long-term initiatives have been underway in SOI to improve "quality" management in SOI. During the 1980s, conventional quality control techniques for detecting errors were slowly replaced by quality improvement techniques designed primarily to prevent errors. …