Progress Report of US Census Bureau
2004-12-30 09:15:51
 

Eddie Salyers

Bureau of the Census, United States

 

1. Changes and Improvements in the Business Register

 

Over the past several years, the Census Bureau has developed a new Business Register (BR). It became fully operational in January, 2004. The Census Bureau¡¯s new BR has improved the economic census and survey operations.

 

The new Business Register differs in several significant ways. This affects how we use it. We are still learning additional ways of using the new structure. The new Business Register is a production Oracle database consisting of several sets of parent tables and their associated child tables. Parent tables are associated with different target universe: EIN_UNITS, SSN_UNITS, EMPLOYER UNITS, SURVEY UNITS, and NONEMPLOYER UNITS. The first two are primary administrative records sources that are collectively referred to as the Administrative Records Repository. Access to the Business Register target universes is made for statistical purposes through consolidated SAS data sets. One consolidated SAS data set for each target universe. Using SAS data for statistical access was designed to improve the efficiency of OLTP database processes by moving statistical users to SAS data sets. These SAS data sets are the only place that a complete set of information is brought together in a unified file. This allows their comparisons and quality check discussed below.

 

2002 Economic Census

 

The Business Register identifies the target population and provides the mailing list for the 5-year economic census.

 

In this capacity, it identifies the set of statistical units that represents the business population, as defined by reference period, statistical unit, industry codes, geographic codes, business size, and other factors.

 

In December 2002, we mailed Economic Census report forms to 5 million business locations with paid employees. Approximately 1.5 single location businesses (single units) received a short classification form, and an additional two million single establishment enterprises received one of the 500 plus Economic Census forms. All enterprises that own and operate more than one establishment (multiunits) were mailed census report forms for each location. Some 180,000 enterprises owning and operating almost 1.6 million different locations received individual census forms. For the remaining 1.5 to 2 million small, single location enterprises not mailed in the Census, we relied on tax data in lieu of direct reporting. In summary, the Economic Census provides detailed information on about half of the total number of business locations with paid employees, but these establishments account for over 85% of total economic activity as measured by receipts. The Economic Census also plays a critical role in updating and maintaining the organizational linkages within enterprises; it provides current information for all multi-establishment enterprises as well as identifying those single location enterprises that have become ¡°multiunits¡± since the previous Economic Census. As of September 30, 2004, the Census Bureau has released 577 of the 1825 economic census reports.

 

The Business Register has incorporated several new features that aided in the processing of the Economic Census

 

      Data for Employer and Nonemployer Businesses -- The new BR carries data for all active businesses (about 24 million records), including employer and nonemployer businesses. (Our previous register only retained information for employer businesses.) As a by-product, the Census Bureau can expedite the processing of the annual Nonemployer Statistics program. We released the Advance 2002 Nonemployer Statistics report in early May 2004. The Census Bureau¡¯s 2002 Nonemployer data took on great importance this Census, as several economists suggested that the decrease in payroll jobs was offset in part by an increase in self-employment (nonemployers).

      Additional Organizational Unit for Alternate Reporting Units (ARUs) ¨C To accommodate current business practices, we have added a new organizational unit, the ARU. An ARU is a consolidated unit established for the collection of multiunit enterprise data in certain networked industries, where financial data are not available by establishment. ARUs facilitated the collection of census and survey data for networked industries, while reducing respondent burden.

      Original and Corrected Data ¨C We retain the original and corrected data for several key data fields, including the employment, payroll, and receipts information. We retain the original data, the current ¡°best¡± data after analytical updates, and the previous data for those cases where the original data has been changed. These additional data assisted economic census analysts in their review and reconciliation operations.

      Edit and Impute Multiunit Data More Timely ¨C The Census Bureau edits and imputes for missing multiunit establishments and data items earlier. At the time when the data are received, we impute for missing multiunit data.

 

Current Surveys

 

Since its inception in the 1970s, the business register (formerly known as the Standard Statistical Establishment List or SSEL) has incorporated data for the establishment reporting units covered by our economic census, but its structures have not been able to accommodate data for the company-part reporting units (usually industry segments of multi-establishment enterprises) covered by many of our annual, quarterly, and monthly economic surveys. The new business register design implemented for 2002 resolves this shortcoming and makes it possible for us to integrate current survey reporting units into the register. We expect strengthened connections between the business register, the economic census, and current surveys to pay benefits that include improved program coordination, better measurement and management of company reporting burden, strengthened customer relationship management for key data suppliers, improved consistency of survey estimates, and a more company-centric approach to collection. A family of surveys that covers the services-producing sectors will be the first to be incorporated into the register starting late in 2005 as they make the transition to a new sample. We expect to add other surveys once this first phase is complete, probably after the 2007 Economic Census.

 

2. Expanded Use of Administrative Records

 

Administrative records offer an efficient, low-cost alternative to the direct canvass of millions of small businesses. We have made several enhancements that will improve processing efficiency and reduce costs:

 

      The IRS has agreed to provide quarterly employment for legal entities from Form 941, Employer¡¯s Quarterly Tax Return. We will begin receiving quarterly employment data beginning in 2005. Currently, the Census Bureau obtains Form 941 employment only for the first quarter. The additional information will improve our measures of employment on the BR, and will provide a greater history of a business¡¯ reporting pattern for imputation purposes.

      Form SS-4 is an Application for an Employer Identification Number. Information from Form SS-4 is used to assign industry classifications for business births. In the past, the IRS provided the Social Security Administration (SSA) with SS-4 forms for the assignment of industry codes (SSA, not IRS, had the clerical coding expertise.) Beginning in 2002, the IRS began transmitting the SS-4 data to SSA electronically. We requested these electronic SS-4 data directly from the IRS, and began receiving the data on a weekly basis beginning in July of 2004. The Census Bureau has developed an automated coding program that accurately assigns NAICS classifications to 60% of the SS-4 applications based on the written description of activity, the business name, and other information on the SS-4 record. With the automated system, the Census Bureau is able to assign the industry codes much sooner, which benefits our current surveys. The SSA has adopted our automated coding system. The residual uncoded SS-4 records from the automated system are clerically coded at SSA.

      The Census Bureau initiated the process to amend the IRS tax regulations to improve Census programs. We are requesting several new IRS data to enhance programs or replicate Title 13 data.

      For the first time, the Census Bureau will be obtaining the Predecessor/Successor Ids for quarterly payroll tax returns. The may permit us to identify duplicates in both entities and data. Preliminary investigation show us that we may be able to identify the single establishment entity that files a payroll tax return under two separate tax returns and that reports the same payroll tax twice. The could lead to elimination of duplicates from our County Business Patterns tabulation or duplication between the same entity appearing in both the County Business Patterns and Nonemployer Statistics tabulations.

 

3. 2002-Based North American Industry Classification System (NAICS)

 

The 2002 Economic Census implemented a NAICS revision that made changes in four principal areas. Updated classifications from the census will be incorporated in the Business Register as well.

 

      The revision extended Canada-Mexico-U.S. comparability for NAICS Sector 23, Construction. Because time and resources did not permit full development of comparable NAICS 1997 classifications, there was three country agreement only as to the boundaries of construction. The NAICS 2002 revision completes the effort for this sector and achieves comparability for its 5-digit NAICS industries.

      The revision also restructures NAICS Sector 51, Information, and establishes new classes for important new and emerging industries.

      The U.S. made refinements to national industries in NAICS Sector 42, Wholesale Trade, to more accurately reflect the differing production functions of wholesalers that take title to goods versus those that do not and to recognize emerging business-to-business electronic markets in the U.S. These changes did not affect three-country comparability.

      The U.S. also made refinements to national industries in NAICS Sector 44-45, Retail Trade, to improve treatment of department stores and nonstore retailers. Again, these changes did not affect three-country comparability.

 

4. Hardware Improvements

 

      New Oracle machines

      Distributed processing

 

5. New Quality Assurance Measures

 

Consolidated SAS data sets are the foundation of new Quality Assurance measures. These allow us to:

 

      Check the consistency of the Business Register from month-to-month - For each targeted universe a consolidated SAS data set is created monthly. For the monthly SAS data sets, a record-by-record, variable-by-variable comparison is done using SAS¡¯s PROC COMPARE. This does two things: It verifies the SAS data set was created correctly and it identifies how the target universe has changed from one month to the next. The future goal is to capture this information on a monthly basis to identify the natural variation of the Business Register from month-to-month and year-to-year.

      Check the internal consistency of the data for a Business Register target universe - The consolidated SAS data sets consolidate information from several database tables. This allows us to identify problems with linkages between type of records or problems with missing addresses. A business rule may require all single unit establishments to have a mail address. The SAS data sets check the implementation of that rule by using the LINKS record that tells what the address is to combine the address information with all the other information for the record.

      Check the consistency among Business Register target universes ¨C This allows us to check the consistency of administrative records with establishment records

 

Article Accessories
Related Resources