The statistics corner: it takes a long time to change statistical programs - The Business Economist at Work

Business Economics, Oct, 1993 by Joseph W. Duncan

IN THE NEXT ISSUE, my presidential address from the 1993 NABE annual meeting will be published. As readers of this column would expect, my title is "Statistics for Decision Making in the Next Millenium." One of the elements in preparing for the twenty-first century is we must recognize that it takes a long time to implement improvements in statistical programs. To illustrate just how long it takes, this column presents the following three examples.

DEFINING POVERTY

In 1964 the definition of poverty was specified by the Social Security Administration (SSA). It defined in dollar terms the income level that would be adequate to provide families with a nutritionally adequate food plan as devised by the U.S. Department of Agriculture. Income levels were defined for various sizes of families using factors such as the age and sex of the family head and the location of residence (urban vs. rural). In April 1973 a review of poverty statistics was initiated by the Office of Management and Budget (OMB).(1) The review stated:

"In 1969, as a result of the deliberations of a federal interagency committee, two modifications in the definition of poverty were adopted: (1) the method of adjusting poverty threshold for annual cost of living fluctuations was based on the changes in the Consumer Price Index (CPI) rather than on changes in the cost of food included in the Agriculture Department's economy food plan; and (2) the farm poverty thresholds were raised from 70 to 85 percent of the corresponding nonfarm levels. No changes were made either in the conceptual or the statistical base of the poverty threshold at that time. The statistical base today remains the 1955 survey of food consumption."(2)

The Committee on National Statistics of the National Academy of Sciences is currently reviewing the problem of defining poverty in today's context.

SURVEY OF INCOME AND PROGRAM PARTICIPATION

In the late 1970s, OMB authorized funding for the Department of Health, Education, and Welfare (at that time) to undertake research on procedures for implementing a survey that would improve information on income sources and amounts, program participation, and labor force activity in an integrated framework. The survey was originally scheduled to become operational in January 1981, but all funds for the Survey of Income and Program Participation (SIPP) were deleted from the federal budget in 1980 and again in 1981. A special effort by the Census Bureau and other staff in the Executive branch persuaded the Administration and Congress in the summer of 1982 to restore full funding for SIPP in the budget of the Census Bureau. Originally, the plan had been to have the survey sponsored by the Social Security Administration and conducted by the Census Bureau with costs divided between them. Restoration of funds permitted data collection for the survey to begin in October 1983. The original design included several samples (or panels). The first panel in 1984 was followed by a second panel in 1985. During 1985 and 1986 measurement of both samples proceeded in tandem and the first panel was discontinued in 1986, a period in which a third panel was initiated. This rotation process was designed to permit shifts in the content to provide detailed examination of specific questions in individual panels. The SIPP questionnaire includes:

1. A core section for each interview that contains questions about income sources and amounts, program participation, and labor force activity.

2. Fixed topical modules in each panel that ask questions about assets and liabilities, income taxes, annual income, program eligibility, and personal histories.

3. Variable topical modules that are added when the length of the survey and funding permit that have included questions on health care, child care, and child support.

SIPP is an ongoing household survey, but the use of overlapping panels, changing topics, delays in publishing data, and funding cutbacks and subsequent sample reductions have hampered the development of time series data. It should be noted that the planning process for this new survey effectively took nearly a decade before data collection for the first panel. Now, one decade later, the results are finally becoming a valuable resource for understanding the income dynamics of the population, particularly the low-income segment of the population. In early 1993, the Committee on National Statistics completed an assessment of SIPP with recommendations for improvements, but it will be several more years before they are implemented.

THE CONSUMER PRICE INDEX

In 1978 the Bureau of Labor Statistics completed the fifth and most complex revision of the Consumer Price Index. The new series was scheduled to begin in January 1978. The revision of the index at that time was based upon a survey of consumption patterns that permitted reweighting and redesigning the Index. For example, the number of areas in which prices were collected in the revised CPI increased to eighty-five categories from fifty-six in the preceding Index. The revised Index was based on weights from the Consumer Expenditure Survey conducted for the years 1972 and 1973. At the time, this was viewed as the most comprehensive revision of the Consumer Price Index.