Census.gov



Development of National Accounting of Address and Housing Inventory:

The Baseline Information for Post-censal Population Estimates

Ching-li Wang

Michigan Department of Management and Budget

Lewis Cass Building, 1st Floor

Lansing, Michigan 48909

(517) 373-2850

WangC1@state.mi.us

Paper prepared for the Estimates Methods Conference, U.S. Bureau of the Census, Federal Office Building #3, Suitland, Maryland, June 8, 1999.

Development of National Accounting of Address and Housing Inventory:

The Baseline Information for Post-censal Population Estimates

Ching-li Wang

Michigan Department of Management and Budget

Lewis Cass Building, 1st Floor

Lansing, Michigan 48909

(517) 373-2850

WangC1@state.mi.us

I. Introduction

The population estimation program to update population counts for all levels of geography in the nation will be very critical in the next decade. Advanced computing technology allows data users to compile and analyze various data sources with detailed geography for important decisions. Demand for population estimates for the detailed geography also increases. The upcoming American Community Survey, which will eventually replace the long census form in the future, requires the most updated housing units and population counts for all levels of geography to do sampling and to produce appropriate statistics.

Many population estimation methodologies have been used by the Bureau of the Census to produce subnational estimates - such as the Vital Statistics Method, Component Method II, Ratio-correlation Regression Method, Administrative Records Method, the Housing Unit Method, etc. So far, none of these methods have been able to produce the estimates for all levels of geography. Mainly it is because the current methodologies are constrained by the limitations of input data provided by agencies which collect or compile the data for the purposes other than population estimation. The census under-count problem also affects accuracy of the base year population for estimates. Despite how much the Census Bureau has done to test and improve the methods, the estimates are always affected by the problems with input data and the accuracy of census coverage. As a result, various methodologies have been developed to impute, adjust, or smooth the data. It seems that we have been struggling with census base and input data problems for population estimates and there is no way out. But, despite how imperfect the input data and census are, there is always one piece of information which is complete and accurate to a great extent. It is the addresses where the housing units are located and where people live.

People move around, but the addresses and housing units are not except the disaster like tornado in Oklahoma City and Wichita a few weeks ago, Hurricane Andrew in Florida a few years ago, or the current crisis in Kosovo. It is very hard for the Census Bureau to count everybody, but at least the Census Bureau can count all the addresses and housing units in the country. It also should be possible for the Census Bureau to collect new addresses and associated housing information after the census and maintain the most up-to-date address and housing counts on a regular basis.

The technology today allows us to do many things which we had never thought possible. With the technology, we should change the way we prepared population estimates in the past. The Bureau of the Census has developed a powerful Topologically Integrated Geographic Encoding and Referencing (TIGER) System. It provides the ability to assign individual addresses to geographic area codes and to the polygons (census blocks) formed by features, such as roads, streams, and so forth. So that, the census results can be tabulated for various geographic areas - such as states, counties, metropolitan areas, subcounty areas, places, census tracts, census blocks, school districts, congressional districts, voting districts, and Zip code areas. With the same process, the TIGER should be able to assign the new housing unit addresses to all geographic areas as in the census. Then, the population estimates based on the Housing Unit Method will be available for all levels of geography as shown in Figure 1.

Figure 1: Address and Housing As Baseline Information for Estimates

From the census, we should be able to know which addresses have returned the questionnaire or not, and whether the housing units at specific addresses are occupied or vacant. Combination of the census housing units and new housing units in the post-censal years, the housing inventory can be updated for the nation and all areas. The American Community Survey after census 2000 will provide the most up-to-date vacancy rate, persons per household. The annual survey of group quarter populations collected by the Federal-State Cooperative Program for Population Estimates also can be geocoded to all geographic areas. With these data, the Housing Unit Method should be able to produce the estimates for all levels of geography. The address and associated housing information are the baseline information for population estimates.

Population estimation based on the “Flexible Geography” has been in the Census Bureau’s agenda for sometime (U.S. Census Bureau, 1994). The Bureau also is exploring the possibility of using administrative records to produce estimates on various subjects by linking individual record to geographic location. Long (1996) indicates that it is very important to link administrative data on individual and establishments to a geographic unit based on the physical location in order to develop the demographic accounting (or economic accounting). He also emphasizes that linking data on population to geography is in many ways the key to population estimation methodology (Long, 1990: 17; 1996: 4). The National Research Council also recommended that the Census Bureau geocode several large federal administrative record systems and use them to frequently produce small-area estimates (Edmonston and Schultze, 1995). Following this direction, the Census Bureau has proposed and tested the use of TIGER to geocode IRS tax returns street addresses down to census blocks (Sater, 1993; Wetrogan, 1994, 1995). Therefore, we can see how important is the use of addresses for population estimates and how anxious the Bureau of the Census is planning to do that.

However, to implement these agenda and recommendations to achieve the Census Bureau’s vision of population estimation program based on “Flexible Geography” and to link the administrative records to geography, it requires an accurate, complete, and up-to-date Master Address File (MAF). Without the updated MAF, many administrative records with new addresses can not be geocoded to the appropriate geography.

The Bureau of the Census is updating the Master Address File and TIGER for the 2000 census. To ensure the accuracy and completeness of the MAF, the Census Bureau is carrying out the LUCA (Local Update of Census Addresses) program for local and tribal governments to review the address list. By the time the census questionnaires are delivered, the MAF should be the most complete and up-to-date for the census. The question is that soon after the census, the address file will be out-of-date because more new housing units will be built. What can we do to update the address file to the most current date? This is what I would like to talk about in terms of developing “A National Accounting System of Addresses and Housing Inventory.”

By definition, the national accounting of addresses and housing inventory is to collect new housing unit address information on a regular basis, as the “daily entry” in the accounting practice. The system will produce a summary of additions, subtractions, and totals of housing units on a regular basis or at any time as needed. Based on the TIGER, the system will geocode new housing addresses and update the master address file and housing inventory. Since the key to update the Master Address File and TIGER line files is to collect new housing addresses, it is necessary to Enhance the Address Gathering system. So that, we can produce Local Estimates for all levels of geography. Therefore, I would like to introduce the EAGLE (Enhanced Address Gathering for Local Estimate) in the National Accounting of Addresses and Housing Inventory to carry the population estimation program to the next century.

This paper will present the concept of the national accounting system of addresses and housing inventory as the baseline information for population estimates. Then it will discuss the issues concerning data collection of new housing addresses and geographic update of MAF and TIGER Line files under a federal-state-local cooperative program.

II. Conceptualization of the National Accounting of Addresses and Housing Inventory

The National Accounting of Addresses and Housing Inventory is to update the address list and associated housing information on a regular basis. The basic components in the national accounting of addresses and housing inventory include (1) new housing addresses as the input, (2) the geocoding with TIGER as the process, and (3) updated Master Address File and Housing Inventory as the output, as shown in the Figure 2.

Figure 2: National Accounting of Addresses and Housing

Once new housing unit addresses are imported, there are two major phases of operation in the system. One phase, from left to right, begins with geocoding and produces the output after the new addresses match the addresses in MAF or can be coded to the lowest geographic level (i.e. census blocks). In this phase, the system will automatically update the MAF and housing inventory to tabulate the housing unit summary for all levels of geography - from nation to census blocks and other geographic areas such as congressional districts, voting districts, school districts, and zipcode areas. This phase is more computerized automation process.

Another phase of operation, from top to bottom, is to process new addresses which do not match the addresses in TIGER or can not be coded to the lowest geographic level (Census Blocks). When the new address does not match the address list in the system, it is necessary to verify its accuracy and make corrections for errors such as those badly recorded or misspelled and return to TIGER for processing.

If the new address still cannot be matched with the address list after verification and correction, there are two conditions for the new address depending on whether it can be coded or not. If the new address can be geocoded to the lowest geographic level, it is a new address on the existing road or street so that the address range in the TIGER is available to provide codes. In this case, the new address will go back to the system and proceed to the first phase to update MAF and housing inventory.

If the new address can not be coded, it is very likely that the address is on a new road or street or in a new subdivision. Then, it is necessary to update the TIGER/line file in order to incorporate the new geographic location and features of the new housing unit address to the TIGER system. In this case, the system will involve two major tasks simultaneously - the MAF update and TIGER/line file update. Finally, the new address goes back the system and proceeds to the first phase of the operation.

Overall, the major tasks in the system come down to two major program areas - (1) data collection of new addresses, and (2) geographic update of TIGER/line file. Presumably, the MAF and TIGER/line file for the census should be nearly perfect in terms of coverage and accuracy. If we build on this existing database and collect the information about new housing unit addresses, we should be able to prepare the population estimates for all levels of geography available in the TIGER system based on the Housing Unit Method. Focusing on the additions and subtractions of new addresses and housing units will be more economical then geocoding of the addresses from other administrative records which are collected for the proposes other than population estimates. The key issue is where to collect new housing addresses.

III. Data Source of New Housing Unit Addresses

For Census 2000, the Census Bureau works with U.S. Postal Service to merge the Delivery Sequence File (DSF) with the 1990 Census’ Address Control File (ACF) to derive the MAF. After the census, of course, the Census Bureau can continue to work with USPS to update the address file. But, there is no information about the housing units on the specific addresses in the DSF. The DSF includes old addresses. The result from the LUCA program indicates that about 6.5 million addresses were added and 2.2 million addresses were corrected from the LUCA’98 alone. There will be more new addresses and corrections when the LUCA’99 is completed. In other words, the DSF is still far behind for updating the address list and contains errors. For the purpose of population estimates, we need the housing unit information besides addresses. Therefore, it is necessary to seek alternative source for new addresses and housing information.

The building permits, mobile home permits, and demolition permits have been used by the Bureau of the Census and many agencies to prepare subcounty population estimates. It would be possible to use this data source to update Master Address File and TIGER line file. The Manufacturing and Construction Statistics Division of the Census Bureau has routinely collected the information about new housing units authorized by building permits. If we collect the addresses and related housing information from this data source, the new housing addresses will be available for any geographic areas defined in the TIGER system. Since this data source reflects future housing construction and current conversions, the information will be far ahead of any source of data in updating MAF.

The Census Bureau sends Form C-404 to local governments every month to collect building permits, zoning permits and local public construction. The information reported in the Form C-404 includes type of housing unit - single family house detached, single family house attached, two-family building, three and four-family buildings, and five-or-more family buildings. It also includes demolitions and razing of buildings by the type of structure. In addition, the form includes non-residential building permits as a separate item. The additions, alterations, and conversions information are also included in report. If the Bureau of the Census collects data from the original permits with address or location of the proposed housing units, all of these housing information can be tabulated to all levels of geography through geocoding. Once the Census 2000 data become available, detailed information about the housing characteristics associated with the specific addresses will be available. Combination of the census data and new housing data based on these permits should provide a direct estimate of housing units for all levels of geography in the post-censal years.

However, there are several problems related to this data source. According to the Manufacturing and Construction Division, for the nation as a whole, about 5 percent of all privately owned housing units are constructed in areas not requiring building permits and about 3 percent of the single-family houses built in permit-issuing places are built without a permit. In addition, the mobile home information is not included in the Manufacturing and Construction’s building permits data collect. The demolition data collected by Manufacturing and Construction Division has been discontinued (Bryan and Devine, 1999). In some states, the reporting rate of building permits submitted to the Bureau is very low. For example, about two thirds of local units of government in Ohio do not report the building permits to the Census Bureau.

In viewing these problems, we need to develop an new housing unit addresses collection system. The information of new housing construction and addresses is very localized - controlled by local agencies. The Federal-State Cooperative Program will be very important to coordinate the collection of addresses and housing information for the National Accounting of Addresses and Housing Inventory. It is necessary to enhance the Federal-State Cooperative Program, which will be discussed later.

IV. MAF/TIGER Update and Geocoding of New Addresses

After the new housing addresses are collected, the major task in the National Accounting System of Addresses and Housing Inventory is to geocode the new addresses and update MAF/TIGER files. Before the geocoding can be done, it is necessary to have a complete address list. By geocoding the new addresses, the Master Address File and TIGER/line file can be updated to accommodate the new addresses. First of all we need to know how the Bureau of the Census compiles the census address list and how much efforts have been devoted to the development of Master Address File.

A. Census Address List

The Census Bureau uses a nationwide list of addresses to support many operations such as mailing census forms, interviews, and data tabulation. The census address list is a comprehensive file of all housing unit addresses occupied or vacant. A census address list record contains the mailing address and the Zip code of each housing unit and the geographic location of the housing unit. In nonhouse number and street name areas, the Census address list records contain an identifier as spot number on the map. The Census address list also contains a permanent record identification numbers.

To create the Census 2000 address list, the census Bureau match the address list form the 1990 census, the Address Control File (ACF) against the latest version of the Delivery Sequence file (DSF) from the United States Postal Service. Then link the Census address list to feature segments in the Census Bureau’s TIGER database. The TIGER database includes the geographic coordinates and name of all streets, water features, railroads, and other linear features, as well as the boundaries of all jurisdictions and statistical areas used to tabulate the census data. The TIGER database also includes address ranges along streets that have house number and street name addresses, and the hierarchy of census geographic area codes, from the state level down to individual census blocks. By linking the census address list to the TIGER data base, the Census Bureau is able to identify the street segment upon which an individual address exists, and thus determine the geographic codes that apply to that address.

In areas without house number and street name addresses, the Census Bureau will create the Census address list through a field operation called Address Listing. During address listing, the address listers will visit all the housing units in the designated areas to obtain the mailing address, location description, and telephone number, if available. They also will record the geographic location for these residences, and note the location of the housing unit on a map.

B. Local Update of Census Addresses (LUCA)

The census address list and related maps are the foundation for a complete and accurate census. In order to ensure the completeness and accuracy of the census address list, the Census Bureau is carrying out the address list review program, called “Local Update of Census Addresses” or (LUCA). The Census Address List Improvement Act of 1994 (Public Law 103-430) allows the Census Bureau for the first time in history to provide local and tribal governments the list of residential housing unit addresses for review. The address list review provides an opportunity for local and tribal government to review information on the Census Address list before it is used to deliver questionnaire for Census 2000. The 1998 LUCA program for the city-type address local units has completed and 100% field check was underway in the past few months. The 1999 LUCA for non-city-type address area has begun and expected to complete by the summer.

From the technical document, we can see how thoroughly the Census Bureau tries to make sure the Census Address List is complete and accurate. However, participation in the address list review is voluntary. When specific jurisdictions express an interest in participating, the Census Bureau will produce the necessary materials such as address list for specific areas and related maps. Once the Census Bureau delivers these materials, that government will have three months to review and provide its updates to the Census Bureau.

Newly constructed housing units are more likely to be missed than the houses built years earlier. Local and tribal governments are instructed to pay more attention to where new construction activity is occurring and use that information to ensure that these new addresses of housing units appear on the census address list. Local participants are also instructed to verify the map on the boundaries of jurisdiction and streets or roads to make sure they are correct.

The participating government units will go over the new housing, conversion of single-family housing unit to multi-family use, mobile home parks, apartment buildings and possible changes in boundaries. The information locally collected includes new housing construction or building permits, housing inspection records, planning or zeroing records, annexation records, etc. Training workshops are also held for the participants to have hand-on experience in using the Census address list, Census Bureau maps, the summary listing provided by the Bureau, and how to add the address and correct the boundary line.

After participants return corrected census address list and maps to Regional Census Center, the Census Bureau processes corrections and provides summary results and updates Census Address List/TIGER data base. Then, the Census Bureau conducts 100% field check of the corrected addresses to validate census list and sends the feedback materials to participants for review. If there are any disagreement, both participants and Regional Census Center reconcile to make sure every address is correct. Finally, U.S. postal service letter carriers check Census address list and Census Bureau updates the list/TIGER database again before the addresses are used for delivery of Census 2000 Questionnaires. In this process, the address file and TIGER are updated simultaneously.

C. Future Master Address File Update and Housing Inventory Update

As we can see, the Census Bureau has spent so much effort to compile the census address file and worked with local and tribal governments to carry out the Local Update of Census Addresses (LUCA) to ensure its completeness and accuracy. Through LUCA program, many participants have had the experience in handling census address list, census maps, and compilation of local address list. With such a tremendous effort and experience from the LUCA, why don’t we take one more step forward to update MAF and TIGER on a regular basis after the census?

If the Census Bureau continues a review program similar to LUCA to incorporate the new housing unit addresses from building permits and other available sources such as the Annual Boundary and Annexation Survey, the MAF/TIGER could be up-to-date all the time. The Bureau will not have to wait for another 10 years and start over again. There are about 1.5 million new housing units in the United States every year. As of May 14, 1999, according to Census Geographic Division, about 17,000 local units of government have participated in the LUCA’98 or LUCA’99 program. The LUCA’98 has already added 6.5 million new addresses and made 2.2 million corrections. If the one time effort with such a tremendous amount of task is spread through out 10 years by working on 1.5 million new addresses annually, the MAF and TIGER will be always up-to-date whenever needed, even for the next census. To update MAF with new housing information on a regular basis will be much easier and less burden on local agencies than to compile the entire MAF for the census every 10 years.

Because the new housing information is very localized, the cooperation and input from local agencies are critical. Although the LUCA has added and corrected so many addresses, only 44% of the eligible local units of government in the nation have participated in the address list review program. To what extent the LUCA participants will continue to provide new housing addresses and to what extent other local units of government will be interested in providing new housing addresses to the Census Bureau is the issue we need to explore. This requires an extensive coordination between Census Bureau, States, and Local Governments.

V. Enhance Federal-state-local cooperative programs

To implement the National Accounting System of Addresses and Housing Inventory, it is necessary to enhance the federal-state-local cooperative program in collecting new housing addresses and updating the Master Address file.

A. Establishing a Data Collection Network

The building permit issuing system is different from area to area. Some building permit issuing authorities are located at the county level. Some are located at Minor Civil Division level. Some local building authorities cover other jurisdictions. However, at least certain agencies in local areas issue building permits, though it is not uniform in terms of where the authority is located. In the areas without building permit, the local input becomes even more critical. Therefore, it is necessary to have a state-local coordination system to collect the data.

Every state has at least one participating agency of the Federal-State Cooperative Program for Population Estimates (FSCPE). It will be an ideal arrangement for the FSCPE to work with the Census Bureau to coordinate the collection of new housing address data. Based on the list of contacts available from the Manufacturing and Construction Division of the Census Bureau, the FSCPE members establish communication channels and identify the areas where are not covered by the building permits. Depending on the state-local working relations, every state creates a State-Local Cooperative Program to work with local agencies (Building Permits Authorities or other agencies maintaining new housing information) to collect the addresses and new housing information on a regular basis.

Currently, many local databases have been computerized to some degree. Whenever, the local building authorities issue the new permits, the new addresses along with housing structure information and issuing dates can be quickly fed into the system. One typical example of state’s effort in collecting building permits as such is the State of New Jersey. The New Jersey Department of Community Affair provides training, distributes free software for local offices to report building permit and certificate activity electronically. The transmitted information includes detailed housing information on each permit and certificate the local offices issue. The information also includes whether they were completed and whether they were issued certificates of occupancy. The building permits report in the State of New Jersey is far more detailed than the report released by the Manufacturing and Construction Division.

We cannot expect other states have the same data collection capacity as New Jersey. But, there must be a way to allow local authorities to submit the information with less effort in this information technology age. It is necessary to explore the possibility of a new housing address collection network which can automatically transmit the data to the system. It would be desirable if the Census Bureau provides certain amount of financial support for states to assist the local authorities in submitting the data electronically. This issues will be discussed later.

B. State’s Participation in MAF/TIGER Update

After states collect the new housing addresses, they can submit the information electronically to the Census Bureau as the current procedure of submitting input data in the FSCPE program. Then the Bureau can start to update MAF and TIGER. However, states have more knowledge about local conditions and have more information about changes in geographic boundaries. Whenever there are any data or geographic problems, the states are in a better position to solve them quickly. To make the National Accounting of Addresses and Housing Inventory more efficient is to strengthen the Federal-State Cooperative effort not only in data collection, but also in TIGER/Line file update.

The GIS technology along with powerful computers today allows data users to do mapping and geocoding right on the desktop. Any agencies with redistricting capability should be able to update MAF/TIGER. When the states have developed the GIS capability to update address list and TIGER/line file, the Census Bureau may consider an option to allow states to participate in the MAF/TIGER update. If all 50 states become capable of updating their own MAF/TIGER, then the Census Bureau needs only to work with 50 states instead of 39,000 local units of government to update maps individually. The Bureau’s annual Boundary and Annexation Survey and Tribal Review Program may be incorporated into the Federal-State Cooperative Program for MAF/TIGER update.

One of the examples of state’s capacity to update MAF and TIGER is Michigan. Building on the experience and expertise in redistricting project in the 1990s, Michigan has developed a full scale of Geographic Information System services. Due to the need of integrating various state’s database on a common geographic base and the necessity of using the GIS in analyzing geographic location for state programs, many state agencies in Michigan work together to form a project, called “Michigan Geographic Framework”. As the lead agency of the geographic framework, the Michigan Information Center develops a mechanism for maintaining a statewide base map and provides the digital base map for state government.

The geographic framework file integrates and conflates the attributes of the 1:100,000 scale TIGER/Line files to the 1:24,000 scale MIRIS (Michigan Resources Information System) base map. The base map consists of features not only in the current TIGER/Line files, but also in Michigan Department of Natural Resource’s MIRIS (Michigan Resources Information System), and enhanced linear referencing system from the MALI(Michigan Accidental location Index). The attributes in the Michigan Department of Transportation’s highway information from the MALI are also integrated to the framework file.

Michigan Department of State’s Qualified Voter File (QVF) is geocoded and provides the opportunity to update street names, address ranges and 5 digit Zipcodes associated with the road network on a daily flow basis. Because of its completeness and accuracy, the Census Bureau’s Regional Office in Detroit will use the maps produced in the Geographic Framework projects to update addresses in the TIGER for the Census 2000.

With the development of the geographic framework, the Michigan Information Center also provides the services to geocode the major state database such as birth and death files, cancer and hospital data for the Department of Community Health. Michigan has developed the GIS capable of geocoding any administrative records and updating the addresses file and TIGER/Line file in Michigan. Once the geocoding process is completed, Michigan will be able to produce population estimates for various geographic levels. Michigan’s case in updating address file suggests that to update MAF on a regular base is feasible at the state level.

The need for state and local’s participation in the National Accounting of Address and Housing Inventory can be seen further from the address naming and numbering issue. The address naming and numbering system, in many cases, is different from one community to another. In some cases, the street names are changed simply to honor someone. One county in Michigan (Barry), for example, contains seven road naming and numbering systems. Because of problems in the address naming and numbering, many areas in Michigan begin to use E911 system for the standardization of numbering and for non-duplicated naming and numbering. With such local knowledge, one can quickly spot the errors in the address list and make corrections.

This is again an indication of how important state and local agencies is in contributing to the update of the MAF file. Therefore, the partnership between the Census Bureau and States and local should be an on-going process. The data collection network and MAF update project in the National Accounting System of Addresses and Housing Inventory can provide a mechanism for enhancement of such partnership.

VI. Funding and Cost-Benefit Issues

One of the critical issues concerning the development of the National Accounting of Address and Housing Inventory is funding. The National Research Council recommends that “if the cost estimates for continuously updating the master address file and associated geographic database are comparable to the cost of one-time updating just prior to the census,... the development should proceed. If the cost estimates are higher, then the clear advantages of continuously updated address system should be weighed against additional costs.” (Edmonston and Schultze, 1995: 174).

As mentioned before, the LUCA has added 6.5 million new address and 2.2 million corrections just for LUCA’98 alone. If we spread this amount in 8 years, it is about 1.1 million new address a year. There will be more when the LUCA’99 is complete. On the other hand, there are about 1.5 million new housing units every year. It is higher than the LUCA’s addresses to work on, but not too dramatic. In addition, the start-up cost for the program has been absorbed in the Census Address List development for census 2000. The marginal cost of continuously updating should not be higher than the one-time updating. Especially, when the continuously updating is based on the existing efforts on the part of the Census Bureau and the experience of local participants through LUCA, it would be more efficient than one-time updating which needs to start over again every 10 years.

Although USPS’s DSF is the primary source for the Census Bureau uses to update the master address file, the new housing addresses might take 6-18 months to show up on the USPS list. The results from LUCA have shown that the USPS’s DSF was much morse in term of updating new addresses. Therefore, by working on just new housing addresses from building permits will produce the address list more current than the USPS file. By that account, the continuously updating is clearly more advantage than the one-time updating. The benefit of continuous updating for all the operations in the Census Bureau is endless -- including the Population Estimation Program, on-going surveys, and geocoding of other federal data base. Therefore, the benefit are out-weighed even if the costs of continuously updating of MAF turn out to be higher.

A. Funding within Census Bureau for the Program

There are many potential financial sources in the Census Bureau to support the program. With the National Accounting of Address and Housing Inventory to update the MAF on a continuous basis, the funding allocated to prepare the update of MAF and TIGER for the 2010 Census can be used to finance the system. The budget for compiling housing construction also can be appropriated in part to support the project because the Manufacturing and Construction Divisions will not need to collect the Form C-404. All the data that division needs will be part of the output from the National Accounting of Address and Housing Units.

The products of the National Accounting of Address and Housing Unit to update Master Address File will be extremely useful to the on-going operation in the Census Bureau. The American Community Survey, Current Population Survey, Survey of Income and Program Participation, and other projects need the most updated address list for appropriate sampling. The budget for these projects could include the cost of updating sampling frame. In addition, to geocode major federal data systems for other federal agencies can provide a substantial funding for the program, just like the case in Michigan, where most financial support for the GIS projects (Address file update) comes from the geocoding projects for other state agencies.

The National Accounting of Address and Housing Units is an important program to support census operation, population estimates, and carry out other missions of the Census Bureau. This is an investment for the Bureau of the Census and also for the country.

B. Funding Issue for the Federal-State Cooperative Programs

Because the information of new housing address is very localized, the federal-state-local cooperation in collecting the data is very critical as discussed above. However, the resource and expertise in state’s demographic functions vary substantially from state to state. To ensure a uniform quality of data collection requires certain resource including funding to support state’s operation. So that, all states can maintain a minimal level of capacity to collect data for the MAF update.

According to the Office of Management and Budget’s report, there are several federal-state-local cooperative statistical programs in which state and local agencies participate in the collection or compilation of nationally standardized statistics. These include Department of Agriculture’s Crop and Livestock Reporting Program, Department of Labor’s Current Employment Statistics Program (Labor Statistics), Department of Health and Human Service’s Cooperative Health Statistics System (National Center for Health Statistics), Department of Education’s National Cooperative Education Statistics System/Common Core Data Program (National Center for Education Statistics), Department of Justice’s Criminal Justice Comprehensive Data System, U.S. Geological Survey’s National Water Resources Research and Information System, and the Census Bureau’s FSCPE. Except the Census Bureau’s FSCPE, all other federal-state cooperative statistical programs have some kinds of funding (Coffey and Habermann, 1990).

Population data has been recognized as the basic information for all programs used by all levels of government and other public and private organizations to make important decisions. It is the most used information among the products produced by these cooperative statistical programs. Yet the Census Bureau’s cooperative statistical program is the only one that is not funded. Therefore, the development of the National Accounting of Address and Housing Inventory can be used as an important step for the Bureau of the Census to seek funding for the Federal-State Cooperative Program to collect new housing addresses and to update MAF/TIGER line files.

VII. Conclusion and implications

Is the development of the National Accounting of Addresses and Housing Inventory feasible? The ideas presented in the paper may cause many people to say that it is impossible because there are so many problems. This is exactly the same reaction we saw in the late 80s when the Census Bureau was developing the TIGER to digitize the nation’s geography from coast to coast. Now we can see how useful and powerful the TIGER is today.

The Census address list and related maps are the foundation for a complete and accurate census. To update MAF is the fundamental step to meet this need not only for the census, but also for all the surveys conducted by the Bureau which require an update-to-date address list for sampling. Thus, the National Accounting of Addresses and Housing Inventory to update the MAF can benefit not only the population estimates but also other operations in the Census Bureau including the census. It also can be used to geocode many federal administrative database, and state administrative database. Its benefit for multiple uses is endless.

The new housing unit address is critical to the Geographic Division to update the TIGER/Line file. By collecting the housing information and address through building permits meet the needs for most updated housing unit inventory for the Manufacturing and Construction Statistics. More importantly, the new addresses along with housing information allow the Population Division to produce population estimates for all levels of geography. For a long time, the Census Bureau is looking for a method which can produce the estimates for all levels of geography down to census tracts or blocks. The housing unit method based on the National Accounting of Addresses and Housing inventory should be able to pave the a way to achieve that goal.

The major tasks of the National Accounting of Address and Housing Inventory come down to two areas - Address data collection and MAF/TIGER update. The data to be collected in the system is the new housing addresses derived from building permits and other related sources which can provide addresses and housing information. Because the address information is very localized, it is necessary to enhance the federal-state-Local cooperative program. The opportunity for state and local units of government to get involved in the census process on a regular process will enhance the partnership between the Bureau and state and local agencies.

As the census questionnaire return rate has been decreasing over decades, the ability of local governments to help the Census Bureau identify hard-enumerated populations becomes very critical. Through collecting and reviewing addresses in their own localities, local governments would enhance their role in the census process. The involvement of local units of government in updating addresses and housing units would provide educational opportunity for localities to build up and maintain their capacity in collecting local information.

Many states and cities are very concerned about the accuracy of the census because of its political and fiscal impact. If they are given more responsibility to participate in the census process and consider the census as part of their responsibility, they will devote more effort to insure a complete census count in their own jurisdiction for their own interest. Therefore, incorporation of more state and local efforts into the census process is what the Census Bureau should consider to improve the census count. The development of National Accounting of Addresses and Housing Inventory to involve state and local governments will help achieve this goal.

The addresses and housing units are the basic source of information for the census operation. The Census Bureau works very hard to ensure that the Master Address file and housing counts are complete. This is the data Bureau of the Census has full control to ensure its completeness. Counting people involves many dynamic factors. Its completeness heavily depends on whether people want to be counted or where they are during the census, which the Census Bureau has no full control. If the census bureau cannot develop and maintain a complete master address file and housing inventory, the Bureau’s credibility in collecting the data will be questionable. The National Accounting of Addresses and Housing Inventory to maintain a complete up-to-date Master Address File, TIGER line files, and Housing units for all levels of geography will strengthen the Bureau’s credibility to provide the baseline information for population and economic data.

In the 80s, when the Bureau was developing the TIGER system to digitize the census geography from coast to coast, many people saw it as an impossible task. The TIGER has become one of the most innovative systems in the Census Bureau. Now it is time for the Census Bureau to take one more step to best utilize this system to develop a National Accounting System of Addresses and Housing Inventory. The TIGER has done the ground work to provide a powerful tool for linking data to geography. Now it is time for the EAGLE (Enhanced Address Gathering for Local Estimates) in the National Accounting of Address and Housing Inventory to fly from coast to coast and everywhere in the country to produce better population estimates and to carry out the Census Bureau’s missions further in the 21st century.

References

Bryan, Thomas and Jason Devine, “Sub-county Population Estimates Method of the U.S. Bureau of the Census.” Presented at the Annual Meeting of Population Association of America, New York, March 25, 1999.

Coffey, Jerry and Hermann Habermann. “Federal-State-Local Relationship: Where are we?- Where are we going?”, Office of Information and Regulatory Affairs, office of Management and Budget, 1990 Joint Statistical Meetings, Anaheim, CA August 6-9, 1990.

Duncan, Joseph W. et al, A Framework for Planning U.S. Federal-Statistics for the 1980's, Office of Federal Statistical Policy and Standards, U.S. Department of Commerce, July 1978, Chapter 22, “ Federal-State-Local Cooperative Systems of Data Collection.”

Edmonston, Barry and Charles Schultz (ed.) Modernizing the U.S. Census. Washington, D.C.: National Academy Press, 1995.

Long, John F. “The subnational population program of the U.S. Bureau of the Census; past, present, and future” Paper presented for presentation at the American Statistical Association, August, 1990. .

Long, John F. “A theoretical Perspective on Administrative Records and Demographic Accounting” Paper presented at the Annual Meeting of the Population Association of America, New Orleans, May 9-11, 1996.

Michigan Department of Management and Budget, Michigan Information Center, “”Michigan Geographic Framework”, . 1999

Prevost, Ronald C. And Clark, Annetta 1995, “Continuous measurement procedures and the Statistical representation of persons residing in group quarters”, Paper presented at the annual meeting of the Population Association of America, San Francisco, California, April 6-8, 1995

Prevost, Ronald C. “Administrative Records in Official Statistics - How can our research on Administrative Records in Census 2000 lead to greater use in 2010?” Presented to Council of Professional Associations of Federal Statistics, November 15, 1996.

U.S. General Accounting Office, “Federal and State Officials Views on the Operations of Five Federal/State Cooperative Statistical Programs (GGD-80-71), Memo to Office of Federal Statistical Policy and Standards, Department of Commerce, June 2, 1980.

Sater, Douglas K. “Geographic Coding of Administrative Records - Past Experience and Current Research” Census Bureau, Population Division Working Paper No. 2, April, 1993.

Sater, Douglas K. 1995 “Differences in location of Households and Tax Filing Units”, Presented at the Population Association of America Annual Meeting, April 6-8, 1995.

U.S. Bureau of the Census. “Building Permits Survey Documentation.” , 1999

U.S. Bureau of the Census. “Program fro Integrated Estimates (PIE): Initial Vision Statement.” Population Division, FSCPE meeting in Miami, Florida, May 2-3, 1994

U.S. Bureau of the Census, Local Update of Census Addresses (LUCA): the LUCA 98 Booklet and Local Update of Census Addresses (LUCA), Technical Guide , March,1999.

U.S. Bureau of the Census. ‘The Census Bureau’s Local Update of Census Addresses: 1999 Address list review opportunity for areas that do no have house number and street name mail delivery.” 1999.

Wetrogan, Signe I. “ Monthly Activity Repot for August 1994", U.S. Bureau of the Census, Population Division, 1994.

Wetrogan, Signe I. “Status of IRS DATA and Other Administrative Data”, Presentation at the 1995 FSCPE meeting in San Francisco, April, 1995

-----------------------

[pic]

[pic]

[pic]

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download