Mr. Chairman and members of the Subcommittee:

Thank you for the opportunity to discuss with you today the Bureau's Fiscal Year 2000 budget request.

OVERVIEW

The Bureau is requesting direct appropriations of $3,071,698,000 for domestic discretionary spending; of that, $156,944,000 is for current statistical programs, and $2,914,000,000 is for periodic census programs. This request is $1,731,749,000 more than was appropriated in Fiscal Year 1999.

Other than for Census 2000, we are requesting only modest adjustments for the programs that provide the base data for the federal statistical community and a broad spectrum of public users. I strongly urge as you enter your deliberations on requirements for these programs that you not sacrifice them to meet Census 2000 requirements. Any erosion of funding for these producers of our critical economic and demographic data about and for the American people will have devastating effects on their availability and accuracy. We simply cannot afford erosion of these programs; however, we certainly prepared our estimates fully cognizant of the fiscal pressures the census places on the federal budgetary process.

DEMOGRAPHIC MEASURES

While many people associate the Census Bureau only with the decennial census or with population estimates and projections, there is a vast array of familiar information the Bureau collects and provides to policy makers on the socioeconomic, demographic, and housing characteristics of the Nation's people. Our primary surveys in the Current Demographic Statistics program are the Current Population Survey, or the CPS, and the Survey of Income and Program Participation, often referred to as the SIPP.

The CPS is the primary source of information on the labor force status of the population, and is funded as a joint program with the Bureau of Labor Statistics (BLS). It measures how much of the labor force is employed and unemployed, and refines employment data by various characteristics such as age, sex, ethnic origin, veteran status, and other demographic variables. In addition, the March Annual Demographic Supplement to the CPS provides the official estimates of annual income, poverty, and health insurance coverage.

The other survey, the SIPP, provides detailed data on trends in income, poverty, health insurance coverage, program participation and eligibility, disability, taxes, non-cash benefits, and general demographic characteristics, by interviewing a representative sample of people for four years. These survey estimates of economic well-being are both more comprehensive and more accurate than those from the CPS, including information on benefits received from government programs and how they affect economic well-being over time.

We are requesting $52,867,000 in FY 2000 for the Current Demographic Statistics program.

OTHER DEMOGRAPHIC ACTIVITIES

The Demographic Surveys Sample Redesign activity updates the samples for the major recurring household surveys to reflect the latest decennial census and thereby maintains the accuracy of the results; we are requesting $4,805,000 in FY 2000 for this activity.

Although relatively small in comparison to many of the Census Bureau's programs, the Intercensal Demographic Estimates program is a significant demographic activity. Between decennial censuses, decision makers rely on the program for annual population estimates for states, counties, cities, and towns. During non-decennial years, these estimates are the basis for the distribution of more than $100 billion in federal funds and are used to set caps on the amount of tax exempt revenue bonds a state may issue. This program supports the Census Bureau's mission to collect and provide timely, relevant data about the people and economy of the United States. The FY 2000 request for the Intercensal Demographic Estimates program is $5,300,000.

ECONOMIC STATISTICS

Fiscal Year 2000 is the sixth and last year in the six year 1997 Economic Census funding cycle. The focus of activity for the FY 2000 will be on the publication and dissemination of information collected and processed in the previous two years. Since the Economic Census data will be organized according to the new North American Industry Classification System (NAICS), the Bureau will have to prepare several new reports illustrating how NAICS affects the economic statistics. No data are published that disclose the operations of individual businesses.

By the end of the fiscal year, we will have released all but a handful of the 1,200 reports that are generated as part of the 1997 Economic Census. We provide data in a variety of different media, principally via CD-ROMs and the Internet.

FY 2000 also is the first year in the 2002 Economic Census cycle. We will begin planning activities related to design and content of the forms for 2002. We also will contact respondents to develop new means to reduce reporting burden. The FY 2000 request for the Economic Censuses is $46,444,000.

FY 2000 also is the first year in the five-year cycle of the 2002 Census of Governments. In FY 2000, our efforts will concentrate on planning and preparation for the content, coverage, methodologies and procedures appropriate for the several component surveys and programs that will constitute the census. We are requesting $4,594,000 in FY 2000 for the Census of Governments.

On a current basis the Census Bureau conducts more than 100 annual, quarterly, and monthly surveys of key national economic statistics--including services, construction, manufacturing, wholesale and retail trade, and transportation. Over the past few years, the Census Bureau has reviewed its current surveys and discontinued those that were not of the highest priority. For the Current Economic Statistics program, the Census Bureau is requesting $100,118,000 in FY 2000.

INNOVATION

The innovative work funded by the Electronic Information Collection (EIC) activity allows the Census Bureau to improve the way it conducts business by taking advantage of modern, efficient technologies.

The EIC program also provides Census Bureau implementation of the new International Trade Data System (ITDS). ITDS will facilitate the electronic interchange of trade data among Federal agencies and significantly reduce the reporting burden imposed on the business community by eliminating redundant data collection. The FY 2000 request for the Electronic Information Collection program is $7,939,000.

The Data Processing Systems program is used to maintain the Bureau's information technology infrastructure. This program buys or rents the hardware and software needed by the Bureau's general purpose computing facility, and is requesting $22,997,000 for FY 2000.

Survey Development and Data Services conducts necessary research into survey design and estimation, time series analysis, error reduction, privacy protection and special experimental and evaluation studies; these activities continually improve the Bureau's survey and census activities, and requests $3,959,000.

GEOGRAPHIC SUPPORT

The Bureau has developed an integrated and automated computer based Geographic Support System (GSS). The GSS requires large volumes of information from many external sources to establish and maintain accurate geographic boundaries, address ranges, and other map information within the Topologically Integrated Geographic Encoding and Referencing (TIGER) data base and accurate addresses in the Master Address File (MAF). This information must be updated on a periodic and regular basis to meet the needs of the Economic Census, the Current Demographic Statistics, and the Intercensal Demographic Estimates activities We are requesting $33,406,000 for Geographic Support activities.

While the Census Bureau will, in FY 2000 as in all years, continue its regular current programs and cyclical censuses, we have no new program initiatives this year; we need not tell you that the lion's share of our resources and our energy will be devoted to the data collecting and processing activities for the 2000 decennial census in the Census Year itself! So let me continue with Census 2000.

DECENNIAL CENSUS

For the past decade, the Census Bureau has been looking ahead to one day: April 1, 2000. We have been preparing to take a family portrait of all the people of our Nation on that day--how many of them there are, who they are, where they live, how they live.

We have been preparing to take this portrait so that it will not only be the clearest (that is, most accurate) portrait possible, but also the most cost-effective census possible. We are also determined that Census 2000 will impose as light as burden as possible on the 270 million or so men, women, and children whom we are counting.

I realize, as you do, that the budget request you've been provided thus far reflects a funding need that now has been overtaken by events. While the estimates for the other Bureau programs will not change, and, in fact, reflect only modest adjustments over Fiscal Year 1999 levels to keep pace with rising costs of operations, the estimate for Census 2000 will change significantly.

On January 25, 1999 the Supreme Court announced their 5 to 4 decision that the Census Act does not authorize the Bureau to use statistical sampling methods in Census 2000 to produce the population count used for apportioning Congressional representation. While this ruling was disheartening to those who have worked so hard over the last seven years in an attempt to design, plan and conduct a census that is both more accurate and less costly than prior censuses, our charge now is to make every effort to achieve these same goals without statistical methods for apportionment purposes. Consistent with the Supreme Court ruling, we also plan to conduct an "Accuracy and Coverage Evaluation," or ACE, to determine the undercount and produce more accurate numbers for every other purpose. It will be very difficult. Particularly now that we're a mere 372 days from Census Day 2000 and the census design debate continues.

Obviously, as a result of the Supreme Court decision, it was necessary to make significant changes in our original plans that will result in changes to the budget estimates as presented to you. Let me summarize for you where I think we are and what I think you can expect:

• We've prepared and released an Operational Plan for a Census without sampling for non-response follow-up. This Operational Plan outlines significant modifications in five areas:
  • methods to improve public response,

  • methods for conducting nonresponse followup operations without sampling,

  • methods to improve coverage in lieu of the use of sampling techniques,

  • enhancing quality assurance and training of already planed methods, and,

  • conducting the Accuracy and Coverage Evaluation to evaluate the census operations and produce numbers for non-apportionment purposes.

• We've prepared a new Master Activity Schedule to which we are currently adding more detail, for a census without sampling for non-response followup to incorporate changes in the areas I just noted. This schedule is the real work plan for the daily, if not hourly, activities of the thousands of permanent and temporary census workers, and draws on lessons learned in last year's Dress Rehearsal.

• We're currently working on revised budget estimates, together with a detailed explanation and justification of the changes which will be required. The Census Bureau expects to complete this task during the spring. While availability of revised estimates on this schedule may not be ideal for the appropriations process, it is the earliest I believe that we can produce these estimates with the same care and certainty of those developed over the past seven years.

• Because the change in methodology required by the Supreme Court decision affects both the timing and magnitude of census operations, we expect to require additional funds for FY 1999, and we will also alter our previous estimates for FY 2001.

• For a number of reasons that I will explain further, the task of accurately enumerating every American in Census 2000 will be substantially more difficult and costly than the 1990 Decennial Census, and the results may not be any more accurate.

There are many factors that contribute to the difficulty of enumerating every American where they live, as well as the cost of conducting this highly integrated, complex mobilization of people and technology. First, there are more people and more housing units. In addition, the American people are more mobile than ever, so there is simply more work to be done. Second, rising labor costs and a different employee profile account for a large part of the estimated cost increase. Taking the census is extraordinarily labor intensive and under the current plan, we need to hire several hundred thousand more people than projected under the original plan. Many of the types of people who used to take temporary census jobs are employed elsewhere, resulting in the need to recruit, hire and train more people than ever before for part-time census employment, often as a second job. And, in order to be most productive, Census 2000 will need temporary workers with a high skill level, both to operate and maintain our office automation technology and to cooperatively have direct face-to-face and telephone contact with respondents. Because many of them will go to people's homes, and they must all protect the privacy of respondents, our employees must meet very stringent standards and background checks. In order for the Census Bureau to successfully compete with private industry for these skilled people, even for temporary employment, wages must be competitive, and therefore costs go up.

In addition to increased workload and labor costs, we must invest heavily in order to increase public awareness and the response rate--and to try to improve our enumeration of those who are traditionally hard-to-count. We are still finalizing the costs of a Census which does not include sampling for nonresponse followup. Because it was necessary to complete the detailed

Master Activity Schedule for the operating plan before calculating final costs, a detailed budget estimate for Census 2000 is not yet available.

The current request for Census 2000 is $2,789,545,000 in FY 2000. To keep this amount in the appropriate context, let me briefly explain what the request before you will fund:

• In addition to our 12 permanent regional offices, we have opened and will operate 12 temporary regional census centers in Atlanta, Boston, Charlotte, Chicago, Dallas, Denver, Detroit, Kansas City, Los Angeles, New York, Philadelphia, and Seattle. Each Regional Center will have about 20,000 square feet, and, at peak will employ an average of 150 people.

• At the peak of Census 2000, we had expected to create about 600,000 temporary jobs nationwide. Our original plan called for recruiting 2.7 million people to fill these jobs. Regardless of the plan, the Bureau will hire people to work in the neighborhoods and areas where they live.

• We will operate 476 temporary offices nationwide--almost 40 for each Regional Census Center. These offices will conduct the local data collection activities--principally, collecting data from households which do not mail back a questionnaire. Each temporary office will have about 8,300 square feet, and will employ about 50 people to support the field data collection activities.

• We will continue to hire former welfare recipients through the Administration's "Welfare to Work" initiative. A total of 4,000 former welfare recipients will be hired.

• Private contractors, in partnership with the Census Bureau, will conduct data capture activities to scan millions of census questionnaires and convert the data into machine readable form using state-of the-art imaging technology. Four centers--each approximately 200 thousand square feet--will be located in Pomona, California; Phoenix, Arizona; Baltimore, Maryland; and Jeffersonville, Indiana--an expansion of an existing Census Bureau facility in Jeffersonville. At their peak, each data capture center will employ approximately 2,000 people, in two shifts.

• The Census Bureau has awarded contracts to:
  • Lockheed-Martin, for purchase and set up of the DCS 2000 optical scanning solution for imaging and capturing data from census questionnaires;

  • TRW, to manage the 3 data capture centers (except Jeffersonville);

  • IBM , to develop the American Fact Finder, providing for release and publication of Census 2000 data over the Internet.

  • EDS, to establish decentralized phone centers to assist the public in completing census forms; and

  • Young and Rubicam, to develop and implement a paid advertising campaign.

• The Census Bureau will create 15,000 temporary jobs in questionnaire assistance centers, located in neighborhoods which will most benefit from such help, primarily those neighborhoods which are linguistically isolated.

• For non-English speaking residents, we will make questionnaires available in 5 languages other than English.

• And finally, in addition to enumerating the 50 States and the District of Columbia, we will also enumerate Puerto Rico, the U.S. Virgin Islands, and the U.S. Pacific Territories.

AMERICAN COMMUNITY SURVEY (ACS)

We have included in our fiscal year 2000 request $25,000,000 to continue the American Community Survey to conduct comparative nation-wide analyses between American Community Survey at the 31 sites we established during this fiscal year. The American Community Survey must take place concurrently with the 2000 census in order to compare and analyze the data and assess whether it will allow us to eliminate the long form from the 2010 Decennial Census, long a goal of this committee and one of the intended outcomes of the Survey.

The American Community Survey is already a success story. The ACS gives current, small-area data on demographic, socioeconomic, and household characteristics of the population. Gathering these data on a more frequent basis and forming partnerships with other Federal agencies and State and local officials, helps decision makers plan and evaluate public policies. For example, the Census Bureau has been invited to participate with the Department of Justice in a crime mapping and data-driven management initiative. The community policing perspective requires current, small-area data not only on crime but also on factors which create crime risks and protect against those risks. American Community Survey data on clusters of poor housing, school dropouts who are not employed, and people who walk to work tell local officials about crime risk factors so that they can intervene before neighborhoods experience serious problems.

CONCLUSION

As I have stated earlier, our FY 2000 request for the Decennial Census program must be revised to appropriately reflect the budgetary resources that will permit the Census Bureau to produce a population count for apportioning the House of Representatives consistent with the January 25, 1999 Supreme Court decision. The amounts requested for the non-decennial programs will not change.

Thank you for this opportunity to address our FY 2000 funding requirements. I will be glad to answer any questions you may have.