State and Local Government Statistics at a Crossroads (2007)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

3 Data Users and Uses T here is a wide variety of uses and users of the Governments Division’s state and local government statistics. Not only do consumers use the data that emanate from the quinquennial Census of Governments and the related annual and quarterly surveys, but they also use the data that are developed under reimbursable arrangements with other federal agencies. In keeping with the panel’s charge, in this chapter we consider the uses of data from the division’s base programs, which comprise the ongoing Census of Governments and related surveys. The chapter focuses on three major groups of users—federal government agencies, public inter- est groups, and academic users—each of which was represented in the data users meeting and the workshop held by the panel. It also discusses issues for data users that came to light from panel members’ own experience. The chapter concludes with the panel’s conclusions and recommendations for responding to user needs. One important application of the Governments Division data is not discussed here, except in the context of uses by state and local govern- ment public interest groups. Individual state and local governments must determine how they measure up to other governmental bodies in terms of tax burden, expenditures by function, employees, payrolls, and revenues. In general, the Governments Division data are the only national source of data on local government organization, finances, and employment, as well as the most comprehensive and comparable source of data on state govern- ment organization, finances, and employment. 38 

DATA USERS AND USES 39 FEDERAL GOVERNMENT USERS The Governments Division data on state and local government finances and employment, along with the other Economic Census and current eco- nomic survey data collected by the Census Bureau, serve as primary sources for computations of critically important economic time series. The Bureau of Economic Analysis (BEA) uses the government statistics data as input to the gross domestic product (GDP) and other components of the national income and product accounts (NIPAs), as well as to the regional economic accounts. The Federal Reserve Board (FRB) uses the data as input to its quarterly and annual flow of funds accounts, which show not only the stock but also changes in the assets and liabilities of households, businesses, farms, major financial sectors, foreign borrowers and investors, the federal government, and state and local governments, as well as information on the instruments used by these sectors in their borrowing, lending, and invest- ing transactions. The Centers for Medicare and Medicaid Services (CMS) of the U.S. Department of Health and Human Services use the government statistics data as input to the annual national health expenditure accounts. Uses of the data by BEA, FRB, and CMS are discussed separately below. In addition to their use for these essential economic time series, the state and local government data have other uses by federal agencies that accord them political sensitivity and importance. In the years from its estab- lishment as an independent, bipartisan intergovernmental agency in 1959 to 1997, the U.S. Advisory Commission on Intergovernmental Relations (ACIR) was a major user of state and local government statistics to sup- port its studies of key aspects of intergovernmental interaction. The ACIR served as a repository of experience and information on intergovernmental structure, finance, process, and practice and conducted studies to identify emerging governmental issues, trends, and turning points. The state and local data were essential for the work of the ACIR, and, for a time, the interface between the ACIR staff and the Governments Division staff was exceptionally close and symbiotic. The ACIR also utilized and added value to the Census Bureau’s state and local government data in the preparation and publication of an an- nual volume entitled Significant Features of Fiscal Federalism. The volume, which was published by the nongovernmental American Council on In- tergovernmental Relations for a short time after the demise of the ACIR, provided significant analytical and comparative information from BEA and the Office of Management and Budget that enriched the Government Division’s data. The ACIR also published a series of reports on distressed communities that appeared under various titles. The distressed community reports contained a statistical analysis of the characteristics of governments and the populations of central cities and suburbs and relied heavily on Governments Division data. 

40 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS Although there is no longer the equivalent of an ACIR to drive demand for and use of the state and local governments data, several federal agen- cies use these data in defining units of government under federal legislation, assessing the impact of public policies and in formulas for local reimburse- ments and grants that govern the transfer of billions of dollars, so the stakes are still high for accurate, objective data. The U.S. Department of Housing and Urban Development (HUD) uses the data in its State of the Cities Data System (U.S. Department of Housing and Urban Development, 2007). HUD also has used Governments Division quarterly data on state and local government property tax revenues as an input to computation of operating cost adjustment factors for calculating rent adjustments under various housing programs. The U.S. Department of Labor (DOL) uses the data to administer the Family and Medical Leave Act and the Fair Labor Standards Act, specifically to determine under the Fair Labor Standards Act if an agency is a “public agency.” The data are also used by congressional committees, the Congressional Budget Office, the Congressional Research Service, and various executive branch agencies to assess the effects of legislative proposals on state and local governments and for specific analyses of federal actions. Bureau of Economic Analysis The Census Bureau considers as the most important applications of the Governments Division data to be the Bureau of Economic Analysis uses. BEA is an agency of the U.S. Department of Commerce. Along with the Census Bureau, it is part of the department’s Economics and Statis- tics Administration. BEA produces the gross domestic product, balance of payments, state and local personal income, gross domestic product by state, input-output accounts, and other economic statistics. BEA conducts research and analysis, develops and implements estimation methodologies, and disseminates statistics to the public. It collects some source data itself but relies heavily on other agencies, in particular, the Census Bureau, for most source data. BEA uses the data from the state and local government statistics programs as a primary source for the state and local components of the NIPAs and the regional economic accounts and for elements of other economic statistics, such as input-output accounts. BEA underscored the importance of the Governments Division data to   Statement by Thomas Mesenbourg, associate director for economic programs, U.S. Census Bureau, at the panel’s June 2006 workshop.   addition to detail from the input-output accounts, Tables III-1 to III-5 of BEA Method- In ology Paper 5 identify the other primary data sources currently used to prepare the state and local estimates in the NIPAs (U.S. Bureau of Economic Analysis, 2005). 

DATA USERS AND USES 41 the panel by noting the size of the state and local government component of the national accounts. Consumption expenditures and gross investment by state and local governments, not including government transfers, currently account for 12 percent of GDP and 63 percent of total government expen- ditures, including federal defense and nondefense components. The state and local government component exceeds not only the federal government component, but also such components as exports. The state and local government data, however, because of their lack of timeliness, are also one of the major sources of revision in the GDP esti- mates. Over the period 1983–2002, the average revisions to the estimates of state and local government purchases of goods and services were typi- cally larger than the average revisions to all other eight major components. Only the equipment and software investment and the export components had larger revisions (Fixler, 2004). The BEA time series estimates of GDP and other elements of the NI- PAs are widely cited in the media and used by many analysts and decision makers in the public, private, and academic sectors. Government agency users of the BEA aggregate estimates of state and local government expen- ditures include state and local governments, the Congressional Budget Of- fice, the Board of Governors of the Federal Reserve, the U.S. Department of the Treasury, the U.S. Office of Management and Budget, and the U.S. Government Accountability Office. Thus, much of the use of Governments Division data is indirect. For example, the FRB uses the NIPA data on net saving, real investment, and consumption expenditures as the basis of its review of economic and financial developments in its semiannual Monetary Policy Report to Congress. The reason that the FRB and other agencies seek out BEA time series rather than rely on the original Governments Division data is because of the value added by BEA. Most importantly, BEA translates the source data into the national accounts structure and concepts and transforms them to a calendar-year basis. BEA also provides updated quarterly estimates as the source data are released, and it derives price indexes and real constant dollar measures for the state and local government component on the same basis as other BEA series. One of the two main regional account series is the gross domestic product for states (formerly called the gross state product). This series is the sum of the value added for all industries in the state, which equals the sum of compensation of employees, plus taxes on production and imports less subsidies, plus gross operating surplus (the balance available to the   Based on presentation by Dennis Fixler, U.S. Bureau of Economic Analysis, at the panel’s June 2006 workshop.   U.S. Census Bureau, Statistical Abstract of the United States: 2006 (2005:Table 650). 

42 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS state that allows it to reimburse bond holders, among others). It relies on the Census of Governments and the annual and quarterly finance surveys as primary data sources for taxes on production and imports for all industries and for gross operating surplus for government enterprises. The other major regional account series produced by BEA comprises the state personal income estimates. This series uses data from the Census of Governments and the surveys of annual government finances, tax collec- tions, employment, and public employee retirement systems. BEA’s industry accounts also use the Governments Division data for the state and local highway construction component of gross output and for the commodity composition of intermediate purchases for the input-output accounts. BEA Adjustments to Governments Division Data BEA must adjust the Governments Division data because information developed for purposes of fiscal and financial decision making does not conform precisely to the accounting and conceptual basis for the national economic accounts (see Box 3-1). One adjustment has to do with the cal- endar—the national accounts are on a calendar-year basis but state govern- ment financial data are on a fiscal-year basis, typically July 1 to June 30, and local governments have a wide variety in the timing of their fiscal years. To convert the more standard fiscal years to calendar years, BEA must use a 2-year average for the most part. BEA further modifies the state and local government data on receipts and expenditures to account for coverage differences, netting and grossing differences, and timing and other differences. These adjustments, in total, are quite consequential. In 2003, the adjustments for accounting and con- ceptual differences amounted to about a 30 percent reduction in state and local government revenues and expenditures from the original Governments Division estimates. BEA Uses of Employee Retirement, Employment, and Other Data in Producing Personal Income Data In response to an inquiry from the panel, BEA explained its need for data from the Governments Division public employee retirement system survey. Employer contributions to state and local government retirement systems are used for the estimates of compensation of the employees com- ponent of GDP and state personal income. Moreover, state personal income   These adjustments are presented in NIPA Table 3.19. Details are presented in Methodology Paper 5: Government Transactions (U.S. Bureau of Economic Analysis, 2005). 

DATA USERS AND USES 43 BOX 3-1 BEA Adjustments to Governments Division Data on State and Local Government Revenues and Expenditures in the National Income and Product Accounts (NIPAs) Coverage Adjustments The following coverage adjustments are made to revenues: Employee Retirement Plans—NIPAs include in the household sector, Census in the government sector. Unemployment Insurance—NIPAs treat as a federal program, Census treats as state programs. Estate and Gift Taxes—NIPAs treat as capital transfers, Census treats as current receipts. Tribal governments—NIPAs treat them as governments, Census does not. Imputations (e.g., imputed interest received)—NIPAs incorporate them, Census does not. The following coverage adjustments are made to expenditures: The adjustments are similar in concept to those made for revenues. The NIPA household sector includes retirement program-related expenditures of state and local governments. The NIPAs include imputations. The NIPAS define net investment as gross investment less consumption of fixed capital for government; Census does not include consumption of fixed capital and treats investment expenditures as current expenditures. Netting and Grossing Adjustments The NIPA expenditures are net of related sales revenue; sales are therefore subtracted from expenditures for government enterprises and general government. The NIPAs subtract insurance claims revenue. The NIPA expenditures for insurance services equal premiums plus premium supplements minus normal losses. The NIPAs include imputations for employer contributions to own social insurance funds. Adjustments for Timing and Other Differences The NIPAs put corporate profits taxes on an accrual basis. The NIPAs make a timing adjustment for differences in the treatment of out-of- court tobacco settlement payments. The NIPAs reflect adjustments to revenues and expenditures that arise from the quarterly interpolation to obtain calendar-year totals. SOURCE: U.S. Bureau of Economic Analysis (2005). 

44 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS estimates are used directly to allocate over $215 billion of federal funds for Medicaid and other programs that use the Medicaid funding formula. Aside from retirement system survey data, the only other local gov- ernment statistics that BEA uses in producing personal income estimates are the data on locally imposed taxes on individual income, motor vehicle licenses, and personal property. These items do not have a substantial im- pact on state personal income, and the national estimates are derived from samples, not complete accounts. For local income taxes, BEA might be able to get data online from the states that are affected if the tax data were not available from the Governments Division, but the availability of annual and interim financial reports varies dramatically across local governments. BEA also uses the Governments Division employment data in the mea- surement of GDP to allocate compensation across functions of government and to estimate full-time-equivalent employment for state and local govern- ments, which is used to estimate real compensation. The regional program uses the Census of Governments and the Annual Survey of Employment to derive state-level estimates of the employment and wages and salaries of students and their spouses who are employed by public institutions of higher education in which the students are enrolled. There is no other na- tional or state source for information on student workers at state institu- tions of higher education. In 2004 the national wage estimate for student workers was about $5 billion. BEA Suggestions to Enhance the Governments Division Data BEA would like to receive the Governments Division data on a more timely basis. BEA also reported to the panel that the following enhance- ments to the Governments Division data would be helpful: • the addition of information on defined-contribution retirement plans of state and local government workers; • more data on social benefit programs; • industry detail for tax data; and • data on intangible capital, such as innovative property (e.g., R&D) and economic competencies as well as software and other informa- tion technology. In some instances, BEA could take some reductions in data without harming the estimates. For example, although BEA considers data from the school system finance survey that the Governments Division conducts for the U.S. Department of Education to be critical, some of the details in 

DATA USERS AND USES 45 the survey could be reduced without harm to the measurement of GDP. This survey covers units that account for about one-half of total spending by local governments. FRB The FRB flow of funds accounts measure financial flows across sectors of the economy, tracking funds as they move from sectors that serve as sources of capital, through intermediaries (such as banks, mutual funds, and pension funds), to sectors that use the capital to acquire physical and financial assets. In constructing these accounts, the FRB needs inputs that are consistent and timely, much in the same way that BEA needs data for the national accounts (Teplin, 2001). The FRB uses three of the Governments Division surveys for its flow of funds accounts—the Annual Survey of State and Local Government Fi- nances, the Annual Survey of State and Local Government Public Employee Retirement Systems, and the Quarterly Survey of Finances of Selected Pub- lic Retirement Systems. Annual Survey of State and Local Government Finances Many of the inputs to the flow of funds accounts representing state and local government finances (excluding retirement plans) are taken directly from the national income and product accounts and not from the Census Bureau sources. With the exception of some differences in the definition of saving, the adjusted NIPA data fit into the FRB framework. However, information on the assets and liabilities of state and local governments are not readily available, so the FRB must benchmark its quarterly estimates of total financial assets of state and local governments to the Governments Division annual survey. The FRB uses year-to-year growth in the “cash and security holdings, other than insurance trust funds” ag- gregate value for all state and local governments as the benchmark growth rate measure for state and local government assets in its accounts. To this measure of state and local government assets, the FRB adds NIPA estimates of taxes receivable and its own estimate of trade receivables. Because of the time lag in the release of the annual Governments Division survey data, the FRB projects the growth rate and revises it once the actual growth rate becomes available. The FRB makes extensive use of the Comprehensive Annual Financial   BEA, Response to Panel Questions, July 26, 2006.   Discussion of the FRB uses of the Governments Division data is based on Paul Smith’s presentation at the panel’s June 2006 workshop. 

46 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS Reports (CAFRs) provided by state and local governments to the Single Audit Reporting Program in the Governments Division to determine how to allocate assets across categories. For this purpose, the agency takes a sample of the largest governments and sorts through their reports, including footnoted information, to determine portfolio allocations. Annual Survey of State and Local Government Public Employee Retirement Systems The FRB uses most of the asset categories for investments and cash on hand and on deposit that are available at the aggregate, national level from the Governments Division survey of public employee retirement systems. It benchmarks total asset levels to the survey estimates after subtracting real property assets and adjusting corporate bonds from market to book value. The FRB uses the CAFRs for the largest 100 retirement funds to better adjust the asset categories. Quarterly Survey of Finances of Selected Public Employee Retirement Systems The FRB uses the quarterly growth rates for asset categories in order to move its time series forward between annual Governments Division sur- veys. The categories in the quarterly surveys are broader than the annual survey; the FRB therefore adjusts the quarterly data in two steps. After the FRB adjusts the quarterly estimates to be consistent with the annual estimates, it uses prior adjustments to make the annual estimates consis- tent with the flow of funds accounts. The FRB extrapolates estimates until quarterly survey estimates become available. FRB Suggestions to Enhance the Governments Division Data The FRB expressed interest to the panel in the addition of data items to the survey of state and local government finances to include more detail by asset categories, estimates of total assets held in pooled accounts and how these assets are invested, and greater access to the Governments Divi- sion electronic CAFR database. he FRB would also like the Governments T Division to change the asset categories in the public employee retirement   CAFRs are provided to the Governments Division by state and local governments as part of the federal single audit process for recipients of federal grants. This program, including the division’s information clearinghouse role, is overseen by the Office of Management and Budget and reimbursed by major grant-making federal agencies (see http://www.census.gov/ econ/overview/go1400.html). CAFRs are also required for compliance with standards of the Governmental Accounting Standards Board (GASB) as discussed in Chapter 6. 

DATA USERS AND USES 47 systems annual and quarterly surveys to match the flow of funds categories. The pooling of investment funds by state and local governments makes it difficult for the FRB to determine how much is in the investment accounts and where the money is invested. It would also like the division to conduct separate annual and quar- terly surveys of state and local government defined-contribution retirement plans, for which information is currently missing from the flow of funds ac- counts. The FRB has, in fact, developed an idealized form for such a survey and has discussed it with the Governments Division. From the perspective of the FRB, this initiative needs to continue, as the Governments Division survey collects data only on defined-benefit plans. Finally, like the BEA, the FRB would like more timely data from the division. Centers for Medicare and Medicaid Services Since 1964, the U.S. Department of Health and Human Services has published an annual series of estimates of total national health expenditures. These estimates, termed national health expenditure accounts (NHEAs), are compiled with the goal of measuring the total amount spent by residents of the United States to purchase health care goods and services during the year. Also included are the amounts invested in medical sector structures and equipment and in noncommercial research to secure the provision of health services in the future. The NHEAs, which are produced by the National Health Statistics Group in the Centers for Medicare and Medicaid Services, are generally compatible with the NIPAs, although they provide a more complete picture of the health care sector of the nation’s economy in one set of statistics. Three primary characteristics of the NHEAs flow from this framework. First, the NHEAs are comprehensive because they contain in a unified structure all of the main components of the health care system. Second, the accounts are multidimensional, encompassing not only expenditures for medical goods and services, but also the sources of funds that finance these expenditures. Third, the accounts are nationally consistent because they apply a common set of definitions that allow comparisons among categories and over time. The NHEAs measure expenditures on health care goods and services by type of service provider and by source of funds. Service providers are clas-   Memorandum from Stephen Heffler and Arthur Sensenig, National Health Statistics Group, Office of the Actuary, Centers for Medicare and Medicaid Services to Thomas J. Plewes, July 27, 2006. The appendix to this memorandum contains an introduction to the national health expenditure accounts, which includes a table of the most recent NHEA estimates. Also in- cluded in the Appendix is the description of the government public health activity estimates and the methodology used to prepare them. 

48 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS sified by establishment—for example, hospital services, physician services, and nursing homes. The source of funds classification is broadly broken down into private and public sources of funds, and the public source of funds category is further broken down into federal sources and state and local sources. In more detailed NHEA tables, estimates by source of funds are further broken down into specific federal, state, and local types of pro- grams—Department of Veterans Affairs, Department of Defense, Medicare, Medicaid, etc. (U.S. Centers for Medicare and Medicaid Services, 2007). Uses of Governments Division Data for the NHEAs Governments are involved not only in funding care for individual citizens, but also in organizing and delivering publicly provided health ser- vices, such as epidemiological surveillance, inoculations, immunization or vaccination services, disease prevention programs, the operation of public health laboratories, and other such functions. In the NHEAs, spending for these activities is reported in a category called “government public health activities,” which accounted for 21 percent of total health expenditures in 2004. State and local government public health activity expenditures are primarily for the operation of state and local health departments. CMS estimates state and local government spending for health departments by using data from the Census of Governments and from the Annual Survey of State and local Government Finances. The most recent year’s estimates are prepared by extrapolating the prior year’s estimates by the change in total state and local government expenditures. Federal payments to state and local governments are deducted to avoid double-counting, as are ex- penditures made through the Maternal and Child Health Program and the Crippled Children’s Program. The National Health Statistics Group in CMS is also an indirect user of many other state and local government statistics from the Governments Division. It uses BEA estimates that are based on the division’s data as input to estimates of other categories in the NHEAs, such as hospital care. CMS Suggestions to Enhance the Governments Division Data CMS’s suggestions and concerns regarding the Governments Division data for the most part center on the Government Finance and Employment Classification Manual, which is produced by the Governments Division and used to code its federal and state and local government statistics. The version of the Manual available on the division’s website at the time of CMS’s input to the panel was dated December 2000; in October 2006, the 

DATA USERS AND USES 49 Governments Division published a revised manual that was implemented in its annual surveys beginning in 2005 (U.S. Census Bureau, 2006b). CMS identified four basic issues with the expenditure classifications in the Manual (2000 version) that impede its ability to produce needed esti- mates: limited functional detail for health services; functional codes that do not match international codes; the overly broad definition of function code 32 (other health, now “health”); and the treatment of medical transporta- tion. These issues remain with the 2006 version of the Manual. Because of the Manual’s coding structure, CMS must rely on other sources—primarily the federal budget—to estimate expenditures of government health care providers by function and type of service. The situation is even more prob- lematic for state and local health care expenditures because there are so few viable alternative sources. Limited functional detail.  On one hand, the functional groupings of gov- ernment expenditures in Chapter 5 of the 2006 Manual do not include an aggregate “medical” or “health care” functional grouping. On the other hand, the available specific codes are limited in number and overly broad in scope. There are only seven relevant codes: 1. Code 28, federal veterans’ health—includes federal payments for health care services to veterans other than in hospitals. 2. Code 32, health (formerly “other health”)—includes a wide range of services (see below). 3. Code 36, hospitals (combines the previous code 36 for other public hospitals together with code 38 for “other hospitals”)—includes most payments for health care services in public and private hos- pitals, excluding federal veterans’ hospitals. 4. Code 37, federal own hospitals—veterans. 5. Code 39, federal other hospitals—veterans. 6. Code 74; public welfare, vendor payments for medical care—in- cludes Medicaid and other means-tested medical assistance. 7. Code 79, public welfare, other—includes a wide range of services, such as energy, day care, housing, and other means-tested public as- sistance, along with some health care services, such as the costs of administering medical and case assistance and payments to public hospitals other than under Medicaid. There are no other functional groupings identifying health care or medical services. CMS stated to the panel that this level of detail is not sufficient to produce estimates of government medical care spending by function, by provider type, or by nature of service. In part, the problem of limited functional detail would be overcome 

50 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS if the Census Bureau functional codes were to match international codes. CMS expressed concern that the two-digit function codes in Chapter 4 of the Manual do not appear to be compliant with the Classification of Func- tions of Government (COFOG) maintained by the United Nations Statistics Division in that they do not have sufficient detail. The COFOG 2-digit division code 07 (health) is not replicated in the Manual (2-digit division codes are the highest level of aggregation in COFOG), nor are the COFOG groups (3-digit codes) and classes (4-digit codes) that pertain to health. At a minimum, 3-digit COFOG detail for health would be needed for CMS to produce sufficiently detailed estimates of government public health activity by function and nature of service from the Governments Division data. Definition of function code 32: health. The definition of function code 32 is “Provision of services for the conservation and improvement of public health, other than hospital care, and financial support of other govern- ments’ health programs.” The long list of examples includes some functions that are not recognized as health care in other classification schemes—the System of National Accounts (SNA), the System of Health Accounts (SHA), the COFOG system, and the definitions in the NHEAs. Among the func- tions that are not universally recognized as health care are • rabies and animal control; • abatement of mosquitoes, rodents and other vermin; • functions of the Environmental Protection Agency (EPA) at the federal level; • health-related inspection and regulation (e.g., inspection of restau- rants, water supplies, food handlers, nursing homes, etc.); • environmental activities in support of regulation of air and water quality; and • cleanup of hazardous waste sites. Medical transportation. In the Manual, ambulance services are included in two separate function codes, depending on whether the service is organized as part of a fire department (included in function code 24, local fire pro- tection) or as an independent service (included in function code 32, other health). Lack of specific detail, as well as inclusion of the same function in two different codes in the Manual, makes it impossible for CMS to estimate expenditures on medical transportation provided by governments. It is important to be able to break out expenditures for medical transportation provided by governments, since CMS estimates that this function costs between $35 and $70 billion annually. 

DATA USERS AND USES 51 Implications of Classification Issues Lack of detailed data on state and especially local government expen- ditures on government public health care services, antiterrorism spending, and medical transportation is seen as a problem, not only for CMS, but also for the public health community. Insufficient detail leaves the public health community without a uniform metric for assessing state and local public health infrastructure, severely limiting and complicating the task of emergency planning. The lack of data affects the reliability of estimates of surge capacity, public health expenditures for pandemic flu preparedness, and similar purposes. Since the terrorist attacks on September 11, 2001, and the anthrax attacks in fall 2001, public health issues have grown in importance, and CMS has received more requests for information on per capita spending on public health by county or state in order to make estimates of surge capacity, public health expenditures for pandemic flu preparedness, and similar purposes.  CMS is limited in its ability to respond to these requests because, as stated above, the Governments Division data are not detailed enough to break out public health and antiterrorism efforts from all other health expenditures. PUBLIC INTEREST GROUPS Public interest groups use the Governments Division data on state and local government organization, finances, and employment as the basis for tracing changes in the well-being of their constituents, for analyzing general government trends, and for research, advocacy, and lobbying purposes. Seven organizations invited by a subgroup of the panel participated in a May 2006 meeting to express their thoughts on the work of the Govern- ments Division: the Federation of Tax Administrators; the Governmental Research Association; the International City/County Management Asso- ciation; the National Association of Counties; the National Association of State Budget Officers; the National Association of State Retirement Administrators; and the National League of Cities. The National Confer- ence of State Legislatures, which was also invited but could not attend the meeting, provided comments afterward. The panel’s June 2006 workshop included additional presentations from the Governmental Research As- sociation, the International City/County Management Association, and the National League of Cities. Each organization differs in its needs and uses of the Governments Division data. Appendix E is a summary of their presentations on data needs and uses. Since the Governments Division data do not meet all the needs of public interest groups, some groups have designed their own surveys to capture 

52 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS additional information. While they acknowledge that the quality of their data is not as high as that from the Governments Division, they often collect data at finer levels of detail and, for the most part, are able to make data available in a timely manner. The International City/County Management Association (ICMA), a professional association of appointed local officials, including county man- agers; city, town, and village managers; department heads; and other local government officials, rarely uses the Census Governments Division data directly.10 However, ICMA does use the Governments Division’s surveys and data to develop the samples for their own surveys. In order to obtain more current data, ICMA conducts five to seven an- nual surveys of their constituency, which includes all county governments and municipal governments (cities, towns, and villages) representing popu- lations of 2,500 or more. A primary reason driving ICMA to develop its own surveys is to allow for comparison across governments, because local governments want to benchmark and make very specific comparisons. For example, two areas of interest for comparisons are revenues and expendi- tures for solid waste management and parks and recreation. The response rates for ICMA surveys are low—generally between 20 and 40 percent. Lo- cal governments have a low response rate and commission forms of govern- ment generally do not respond. Most of the data are released in September following the end of the previous fiscal year for which they are collected, and the ICMA tries to maintain longitudinal consistency, although ques- tions are sometimes added to address current policy issues. The National League of Cities (NLC) also conducts surveys of its own—notably the Annual City Fiscal Conditions Survey, which has been conducted annually since 1987.11 The NLC uses its survey to supplement the revenues and expenditure data from the Governments Division. A sec- ond purpose of the survey is to capture the items that reflect the biggest concerns for politicians and policy makers in agreeing on a budget, which typically involve the general fund budget. The questions in the NLC survey are similar to those on the Governments Division surveys, although they are more detailed. NLC surveys have some problems with accuracy and response rates, since the surveys are self-reported and voluntary. The over- all response rates for the Annual City Fiscal Conditions Survey have been between 30 and 50 percent, although for larger municipalities the response rates have been about 70 to 80 percent. About 70 percent of all municipal 10  Based on presentations by Evelina Moulder, director of survey research and information management, International City/County Management Association, during the panel’s meeting with data users, May 9, 2006, and its workshop, June 22, 2006. 11  Based on presentations by Christiana Brennan, National League of Cities, during the panel’s meeting with data users, May 9, 2006, and by Michael Pagano, National League of Cities and the University of Illinois at Chicago, at the panel’s workshop, June 22, 2006. 

DATA USERS AND USES 53 spending is represented in the responses received, and supplemental infor- mation is collected from budgets that are posted online. About 80 percent of respondents use the NLC electronic portal to complete the survey, al- though the NLC sends a paper version to all cities in the sample. The National Association of Counties conducts surveys only to obtain specific information not gathered elsewhere. It recently collected special data for its report on counties in crisis. The National Association of State Budget Officers relies on its own fiscal survey and final expenditure report in order to break down the “all other” category from the Governments Division, which accounts for about one-third of all state government expenditures. The categories for the this expenditure report differ from the Governments Division categories. Finally, the National Association of State Retirement Administrators, the National Conference of State Legislators, and the Federation of Tax Ad- ministrators conduct surveys on public pensions, annual state expenditures, and taxes, respectively. RESEARCH INSTITUTIONS AND ACADEMIC RESEARCHERS Researchers in universities and public policy institutes use the Gov- ernments Division data for purposes similar to those of public interest groups—to conduct policy research and to communicate current issues of state and local government finance and organization to policy makers and the media. Academic researchers also use the data for in-depth analyses of the factors that affect the finances and employment of state and local gov- ernments, including the consequences of economic growth and decline. At its June 2006 workshop, the panel heard from four groups of researchers: (1) a group at Canisius College that analyzes overlapping governmental entities, particularly special districts that span counties and states, which are difficult to capture; (2) the Urban Institute–Brookings Institution Tax Policy Center, which is building a database of rankings to help users compare and contrast governments on a variety of dimensions; (3) researchers at the University of Wisconsin–Madison, who focus on in- tergovernmental grants; and (4) the panel’s own members, who identified research uses and data needs in several areas. Some research users, including the Urban Institute, are both users of and disseminators of data on state and local governments, which gives them additional valuable perspectives (see the following section). It should be recognized that these policy and academic users who made presentations are simply illustrative of a much broader group of academics and others who rely on these data and use them in the study of public administration and political science. The views of these users are also important and should be considered by the Census Bureau as it develops a research and development program for the future. 

54 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS The researchers who made presentations uniformly attested that the Governments Division data are valuable. They especially appreciated the fact that all of the information collected on state and local governments is available at the same place, which facilitates linking the states to their local governments. Such linkages are important because the responsibilities of different levels of government vary across states. Having an annual series on state and local government finances is crucial to measuring the effects of policies and general equilibrium effects from other economic conditions. The aforementioned gap in 2001 and 2003 state and local revenue data from the Governments Division came at a time of recession. While some analysts believe that property taxes acted as a savior for many state and local governments during that time, there are no data to support that theory because of the gap. Having this information on an annual basis, even in the aggregate, is critical for research on state and local governance. Research on Governmental Organization Researchers at Canisius College use the Governments Division data for analysis of such issues as central city and suburban fiscal disparities. They also use state and local governments data as input to the urban public fi- nance component of the Department of Housing and Urban Development’s State of the Cities Data System. The system includes information on 21,000 governments in 101 metropolitan areas. Researchers credit the fiscal infor- mation from the Governments Division for consistently defined categories for the ability to compare across states and time. Prior to 1997, researchers who wanted information on government organization mainly used three Governments Division publications—the Government Organization Compendium, the Government Finance and Employment Classification Manual, and the Governments Integrated Di- rectory. Currently, this information is available electronically in public use formats of the individual unit file and the county area file. The Govern- ments Division data are critical to making meaningful comparisons among governmental units and consistent determinations between independent and subordinate governments with classifications that are consistently applied across states, even when the rationales within states vary. The only way to make comparisons across states and over time is with Governments Divi- sion data. For the types of analysis conducted by researchers at Canisius College in support of the HUD data system, a central issue is that multiple govern- ment and fiscal entities typically overlie any given city or suburban area. The city is a fiscal artifact, as central cities and overlying governments of- ten are linked fiscally to county governments, school districts, and special districts. For example, in Columbus, Ohio, 14 school districts span the 

DATA USERS AND USES 55 Columbus metropolitan area, which includes the city, a couple of counties, and several special districts. Moreover, the special districts that span coun- ties and sometimes states will not necessarily be identified in geographic area breakouts. Examining the work that has been done with Governments Division data reveals several areas warranting improvement. For example, the Gov- ernments Division allocates all of the activity of multistate and multicounty special districts to one county area (a geographic concept). For example, the Washington Metro Transit Authority, which spans Washington, D.C., Northern Virginia, and suburban Maryland, is allocated to Washington, D.C. Some users would find it very helpful to have computerized informa- tion on the counties that are served by a special district. For survey years after 2002, the Governments Division stopped collecting data on service areas of special districts out of concern that special districts have become more complex and in recognition of the difficulty of equating the service area of a district with the counties in its legal jurisdiction. For example, the Intermountain Power Public Authority in Utah sells power to the city of Los Angeles. Given complexities like this and the Governments Division’s limited budget, it has stopped collecting these data. The disappearance of these data has made the analysis conducted by Canisius College for the HUD data system more difficult. Uses and Misuses of Rankings The Urban Institute–Brookings Institution Tax Policy Center database of state and local government information (described below) is designed to rank state and local governments on a wide range of dimensions, such as tax burden or per pupil expenditures on education. Some workshop partici- pants argued that rankings can be misleading and there is a risk they could be misused. Others suggest that rankings would be useful analytical tools if accompanied with the actual data that are ranked, and measures of central tendency, such as median, mean, and interquartile range. In reality, there is recognition that politicians, the media and the public like rankings and many use them to make decisions. It is therefore important that there are respected analysts and researchers, like those at the Urban Institute, who compile rankings as accurately, as promptly and with as many appropriate caveats as possible. Some participants go further to suggest that having rankings published by the Governments Division would give the rankings and the data behind them even more credibility, although they recognize that there would be a concern about the political and controversial nature of such a publication. In its role as honest broker, the Governments Division could potentially contribute to improving the rankings that are often made of government fi- 

56 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS nancial data, such as percentages of personal income. The choice of income series influences the rankings. It would be helpful if the Governments Divi- sion would specify the appropriate series for ranking purposes, elaborate its strengths and weaknesses, and indicate whether researchers should use the data as originally published or as revised. More generally, researchers who compile rankings need consistent and timely annual state and local data. Other issues related to compiling rank- ings are interstate consistency for user fees versus taxes and comparing state support for public higher education. Other Research Needs According to researchers at the University of Wisconsin–Madison, the Governments Division data on revenues and expenditures would be en- hanced if important background information were included on the effects of intergovernmental grants, including those made by states in funding public education and unconditional grants from states to municipal gov- ernments. Background information in the form of metadata, which is ac- companying data about the data, would also be helpful to inform analyses of the relationship between state aid and changes in economic and fiscal conditions. One example comes from the state of New Hampshire, which in 2000 passed legislation that required school districts to levy a minimum property tax; that tax was then treated by the Governments Division as an increase in the state property tax. Without this background information, the Governments Division data could be taken to indicate a large increase in the state’s share of education funding. More generally, panel members, as well as other researchers, com- mented on important changes and trends in state and local government finances and employment for which the current Governments Division data provide too little detail for analysis. For example, state and local govern- ments have played an increasingly important role in financing and deliver- ing social services, and the federal government has made major changes to federal-state responsibilities in this area. States often go beyond what the federal government supports or requires by spending money on “state-only” child care programs or child welfare programs. It is possible to measure spending by states on very broad functions, such as public welfare, by us- ing data from the Census of Governments and the annual surveys. It is also possible to measure spending by states on specific federally funded social service programs, such as the Child Care Development Block Grant, by us- ing data from the relevant agency, which is the Administration for Children and Families of the U.S. Department of Health and Human Services in this case. It is not possible, however, to measure spending on important areas of public welfare—what might be called subfunctions—using either data 

DATA USERS AND USES 57 source. Thus, it is not possible to know total state spending on child care (regardless of whether it is for a specific federal or state program). This is an example of an important public policy area for which the Governments Division data fall short. As another example, many policy researchers—and the press and the public—are interested in the extent to which state and local governments have privatized functions previously conducted by government workers. This, too, is an important policy area about which the Governments Divi- sion data can provide no current information. This is unfortunate, since the data were once available—the 1987 and 1992 censuses included data on whether or not local governments provided certain services and, if so, whether they were partially or totally contracted out (although it was not possible to tell whether the services were contracted out to a private pro- vider or to another government). These data were discontinued with the 1997 Census of Governments. Finally, there is widespread interest among public finance analysts for reinstating the data on taxable property values and assessment–sales price ratios that used to be collected in the Taxable Property Value (TPV) Survey. This survey was initiated with the 1957 Census of Governments in response to the directive in the U.S. Code (Title 13, Section 161) that each quinquen- nial Census of Governments includes “data on taxes and tax valuations of states, counties, cities, and other governmental units.” The TPV provided the only nationwide, state, and selected county assessment–sales price ratio statistics on a five-year basis, from 1957 to 1992. Most of the surveys in that time provided an inventory of the assessed values and numbers of locally assessed real property parcels and their distribution within six use categories, data on de facto assessment levels, and coefficients of dispersion for individual assessment jurisdictions, plus effective property tax rates for hundreds of such places.12 The TPV had two parts, one on the assessed values for local general property taxation and the other on the taxable real property assessment–sales price ratios. The first part provided information on real property and personal property taxa- tion, types of properties that are currently lumped together under property taxation. The TPV survey had the strength of national internal consistency, which permitted unbiased comparisons of actual assessment levels across state and local jurisdictions. The Census Bureau discontinued the TPV while conducting the 1992 Census of Governments because of significant budget restraints and the survey’s declining response rates. In addition, there was a change in defini- tion of a “representative sale” that confounded data collection and analysis. 12  National Tax Association, letter to Dr. Kenneth Prewitt, director, U.S. Census Bureau, March 23, 1999. This letter is included in this report as Appendix C. 

58 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS In responding to outside inquiries, the Census Bureau reported that it was “actively involved in research and testing new survey methodologies and technologies. It may develop that these efforts, in combination with appro- priate funding levels and improved cooperation for voluntary participation in our public sector surveys, will create a climate in which we can consider reactivating taxable property values in the future.”13 In the eight years that have passed since the TPV letter exchange, nothing has been done to test the collection of TPV, or to resolve issues with the definition of a representative sale (see Appendix C). Until the discontinuation of data on market values, researchers could calculate effective property tax rates relative to the property value. With the current antiproperty tax climate in many states, the decisions to limit property taxes are being made without reliable data. These data on effec- tive tax rates are needed for tax limitation decisions and for wide-ranging analyses of household and business finances and economic behavior. DATA USERS AS DATA DISSEMINATORS Several public policy organizations have taken the initiative to develop databases of state and local government statistics for internal and external uses. The panel heard from three of them about their efforts and the uses they make of the Governments Division data. The organizations repre- sented are the Urban Institute–Brookings Institution Tax Policy Center, the Public Policy Institute of California, and Governing Magazine. Urban Institute–Brookings Institution Tax Policy Center The Urban Institute has made a significant commitment to creating a data resource for the occasional user of state and local government financial information—for example, if users are interested in property tax revenues across all states over time, they can get that information from the Urban Institute instead of having to go to the Governments Division and look up the information state by state.14 The Urban Institute–Brookings Institution Tax Policy Center has been working on a data query system that is geared to neophyte users, people who are new to the Governments Division data and may be intimidated by them, although it will serve many of the needs of more sophisticated users as well. The Tax Policy Center engages in pub- 13  U.S. Census Bureau, letter to Robert D. Ebel, National Tax Association, May 3, 1999. 14  The work of the Urban Institute in developing this database was presented to the panel by Kim Rueben, senior research associate at the Urban Institute–Brookings Institution Tax Policy Center. This system can be accessed from the Tax Policy Center web page (http://www. taxpolicycenter.org/home/). 

DATA USERS AND USES 59 lic outreach and has frequent interaction with reporters looking for such information as state rankings. The Urban Institute data query system allows users to access informa- tion by state for each type of government and for the nation as a whole. The data are displayed in nominal or inflation-adjusted dollars and can be expressed in total dollars, per capita, percentage of personal income, percentage of general revenues, or percentage of total expenditures. Users can export the data to Excel, comma delimited, or HTML files. The system will link to the Governments Division website for metadata, such as data definitions, technical notes, and standard errors. Public Policy Institute of California The Public Policy Institute of California (PPIC) makes state and local government data available in a table format with the dual goal of providing historical information in a flexible format and increasing the profile of its research.15 Given its research areas, most of the data and research projects are based on comparisons involving California governments. The institute plans to include a tutorial on its web site with instructions for creating and using the tables and for using the Governments Division information as metadata. Through its efforts to increase the awareness of local policy makers who are unaware that their budgets are the basis for the Census of Governments data, PPIC (and similar organizations in other states) may indirectly improve compliance with Census Bureau requests for statistics. Governing Magazine While Governing Magazine is not an academic institution, its readers represent the same interests as those of users of the Urban Institute and PPIC data systems.16 The magazine provides one of the major sources of information tailored to the needs of users of state and local government information. The magazine uses Census Bureau data, from both the De- mographic Directorate and the Economic Directorate, with supplemental information from a daily review of about 75 local newspapers to publish a ranking of the 1,000 largest jurisdictions by revenue and a listing of the nation’s billion dollar governments in terms of revenue, including all 50 states, 15 cities, 25 counties, and 8 special districts. These rankings illumi- 15  The work of PPIC was presented by Tracey Gordon at the panel’s June 22, 2006, workshop. 16  Discussion of Governing Magazine based on Ann Jordan’s presentation at the panel’s June 22, 2006, workshop. 

60 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS nate state and local governance in ways that the raw data from the Census Bureau cannot. For example, where a government falls on the list is not only a func- tion of population or economic activity within the jurisdiction, but also a function of the responsibilities within that government. A special district government, Port Authority of New York and New Jersey, has greater expenditures than four states and two major cities, Baltimore and Boston. When comparing expenditures and other economic activity, it is important to tabulate data by several types of governmental organizations. The biggest frustration for Governing Magazine in using the Gov- ernments Division data is the timing of the data—for example, the 2006 Sourcebook published by the magazine used numbers for the 2003–2004 fiscal year. Given the age of the data, they often reflect the effects of poli- cies made on some prior elected official’s watch, and this fact is not always made clear by the media or candidates for political office. The magazine is aware of and uses similar data from other sources, such as the National Association of State Budget Officers, the National Conference of State Legislatures, and the Rockefeller Institute of Government; however, these sources collect more data on tax revenues than on spending, and they provide limited information on local governments. The consistency of the Governments Division data is the major reason why the magazine continues to use it as a primary source. In addition to the issue of timeliness, Governing Magazine would like to see more data on technology expenditures. Most of the magazine’s advertisers are engaged in information technology (IT), and the magazine has made an effort to apprise companies in all industries of the purchasing power of governments using expenditure data from the Governments Divi- sion. However, these data lack detailed information on IT expenditures. IT data are difficult to collect since these expenditures weave throughout departments and do not hold a single line item in a budget. The Governments Division has an effort under way to determine fea- sible methods for collecting accurate information on IT expenditures. The division has communicated with the Association of State Chief Information Officers and determined that there is commonality among the state govern- ments in how IT expenditures are reported. It appears that the best place to start in collecting this information is to work with the largest governments, but the division stresses the need for standards in this area. In addition to setting standards, the division would need to review the results of the In- formation and Communication Technology Survey recently conducted as a supplement to the Census Bureau’s Annual Capital Expenditures Survey. A review of the supplementary information would help the division under- stand better the IT infrastructure and to determine the best way to collect 

DATA USERS AND USES 61 information on IT. Additional field and cognitive testing and focus groups would be needed as well. CONCLUSIONS AND RECOMMENDATIONS Based on the foregoing discussion, gleaned from meetings with data users and reflecting the expertise and experience of its members, the panel makes four overall conclusions on the topics of usefulness of the Govern- ments Division data; the trust by users in the data; the importance of uses of aggregate time series and of micro-level data; and the priority needs of data users. The panel’s recommendations in this chapter address two of users’ three main priorities: providing needed detail and classifications, and maintaining time series. The third—and highest—priority for users is improving the timeliness of the information, which the panel addresses in Chapter 5. Overall Conclusions Conclusion 3-1: The data on state and local governments from the Census Bureau’s Governments Division are of broad national interest and importance. • The data serve a democratic nation built on principles of de- centralization and local control by maintaining a comprehensive source of information on state, regional and local governments that assist those institutions and public interest organizations—and through them, the voting public—to understand how individual governments compare with other governments on such important measures as tax burdens and expenditures on education, security, health, and other public services. • The data are necessary for comparative research and policy analy- sis of levels and trends on a wide range of important topics, such as the changing nature of local and regional government institu- tions, including the emergence of new forms of local governance; intergovernmental grants and transfers of funds; the layering of governmental functions among types of governmental units; the effects of changes in the economy on revenues and expenditures; and the burdens of property and other taxes. • The data are essential for economic time series that are widely used for public- and private-sector decision making, such as the national income and product accounts, the regional accounts, the flow of funds accounts, and the national health expenditure accounts. 

62 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS The panel notes that the full extent of the use of the Governments Divi- sion data is not easily documented or measured. The reason is that many uses of the data are indirect. The large numbers of users of measures of state and local government expenditures, revenues, assets, and debts in the NIPAs, the flow of funds, and the national health expenditure accounts, who are dependent on the Governments Division data, are often unaware of the fact that Governments Division data play an important role in the development of the measure of interest. Conclusion 3-2: Virtually all users of the Census Bureau’s Governments Division data, including federal agencies, public interest groups, and academic researchers, view the data as authoritative and valuable be- cause of the consistency of the data over time and across governments and the use of carefully specified standards and definitions for classify- ing governments and governmental activities. The importance of maintaining the Government’s Division role in pro- ducing model data on state and local government finances and employ- ment cannot be overstressed. The panel returns to this point in Chapter 6 in discussing management issues and challenges for the division and the Census Bureau. Conclusion 3-3: The Census Bureau’s Governments Division data serve two main communities: users of aggregate estimates (macrodata) for key economic time series, which include the federal agencies that pro- duce them, primarily the Bureau of Economic Analysis and the Federal Reserve Board, and users of data for individual state and local govern- ments (microdata) for research, policy analysis, and comparative rank- ings. While these two groups of users differ in some respects in their views of priority needs from the division, both groups benefit when the full range of needs is considered in establishing priorities. Management of the Census Bureau’s Economic Directorate stressed repeatedly to the panel that the Bureau of Economic Analysis is the most important user for the Governments Division data, followed by the Federal Reserve Board. While fully supporting the needs of these two agencies and other agencies that produce important economic time series, such as the Centers for Medicare and Medicaid Services, in the panel’s view, the Gov- ernments Division needs to better balance its offerings to reflect the needs of other users as well. Ideally, there is interplay between uses of macrodata and microdata from a statistical agency. In the case of statistics on state and local govern- ments, microdata analyses can bring to light anomalies in the data, such 

DATA USERS AND USES 63 as those that may stem from differing accounting classifications between reporting units of government that require investigation. The correction of these reporting anomalies may improve the aggregate estimates. Microdata analyses can also identify important structural shifts in the provision of government services and sources of revenue that may have implications for the kinds of aggregate estimates that are produced for the NIPAs and other key economic time series. In turn, macrodata series are important for microdata users to provide the broad context for their analyses. Conclusion 3-4: Users are in broad agreement about priority improve- ments they would like made in the Census Bureau’s Governments Di- vision data on state and local government finances and employment. Improving the timeliness of the data is of the highest importance, fol- lowed closely by improvements in the detail provided and in the clas- sification structure and avoidance of gaps in time series. The panel addresses improving the relevance of the detail and classifi- cation structure and maintaining time series below; it addresses timeliness in Chapter 5. Level of Detail and Classification Issues Federal agencies, public interest groups representing state and local gov- ernments, and the research community all have a wish list for improvements in the level of detail and classification of governments and governmental functions. As examples, the research community and public interest groups want more disaggregated data on state and local government finances and greater attention to the information on cities and special districts, while the Centers for Medicare and Medicaid Services want data for health expen- ditures that are more detailed and consistent with international classifica- tions. Not satisfied with the data available from the Census Bureau, some public interest groups conduct special surveys to obtain additional detail beyond that available from the Governments Division. A common theme underlying many user requests is the need for the Governments Division to adapt its classification structure and provide even more detailed information that illuminate important changes in types of expenditures and revenues and the provision of services by state and local governments. They ask, in other words, that the Census Bureau take ad- ditional steps to maintain the relevance of the data. There are a number of examples of the need for data to maintain relevance. Both BEA and FRB agree on the need to have information on defined-contribution employee benefit plans in addition to the long- provided information on defined-benefit plans. The growth of defined- 

64 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS contribution plans has been explosive in the private sector, and the public sector is moving, although more slowly, in this direction as well. As another example, the Governments Division expenditures data would be much more useful for tracking trends in the devolution of federal responsibility to the states if the data identified not only very broad functions, such as public welfare, but also subfunctions, such as day care and job training services. Employment data would be more relevant if there were information on the privatization of government services formerly performed by govern- ment workers, a need that is understandable in light of the perceived trend toward outsourcing. Balancing Relevance with Budgetary and Burden Considerations Given the wide variety of data desired, satisfying all of the specific requests for new and disaggregated data to maintain the relevance of state and local government statistics might well overwhelm the Governments Division’s budget. The panel recognizes that it is far easier to point out shortcomings in the data than to determine how they might best be cor- rected, or to accept the trade-offs that might be required in a constrained budget environment. Seemingly simple solutions do not always work as expected: for example, eliminating some details so that others can be added may not save much money, because if the details are in state and local gov- ernment computers, the marginal expense of collection may be trivial and there may be a reprogramming cost for not collecting them. Moreover, it may not be possible to save money by reducing sample sizes, because the Governments Division already appears to have done much of what may be possible here, and the savings from dropping governments in the annual surveys may be exceeded by reenlistment costs when they are included again in the 5-year censuses. Yet another consideration when adding new detail is the effect on re- spondent burden. For example, the data in the expenditure matrix of the finance survey are the most difficult to collect. The division has suggested collapsing the matrix to include fewer functions than the current 25, such as combining police and fire into a general public safety function or pro- viding total expenditures and not object expenditures for certain functions. Although this shortening of the list of data items would reduce burden, it would also interrupt the continuity of historical data series, provide less detailed information, and eliminate key information on functions used for a variety of analytical purposes and for BEA’s calculations of gross domestic product. Shortening the list of data items would also move the functional detail available for state and local governments further away from the clas- sifications recommended in international guidelines (see Chapter 6). 

DATA USERS AND USES 65 Long Lead Times and Maintaining Continuity One reason for the difficulty in implementing changes to maintain the relevance of the data is the long lead times that face the Governments Divi- sion when it seeks to make changes. The division is constrained to a fixed 5-year cycle for reviewing and introducing modifications to the content and other aspects of the Census of Governments, and, in a cascading manner, to the current survey programs. Although it is possible to update the annual and quarterly surveys independent of changing the quinquennial census, this is rarely done because of the risk of introducing data discontinuities between the census and the surveys. These long lead times constrain the Governments Division to continu- ing to collect the same data for some time even though there is evidence that more relevant data could be collected. For example, as this report was writ- ten in 2007, the next opportunity to incorporate changes in the Census of Governments will be in 2012. In order to make those changes, the revisions to data collection related to the organization phase must be decided by Oc- tober 2010, one year before the mail-out in October 2011; the revisions to data collected on the employment phase would have to be made by April 2011 to incorporate them in the October 2011 organization phase mail- ing; and the revisions to data collection on the finance phase would have to be decided in a window from January 2011 to October 2011, depend- ing on the content. Not all of these revisions are within the Governments Division’s control. Those relating to the annual education finance survey, for example, require negotiation with the sponsor, the National Center for Education Statistics. Even when changes in data collection can be made, there is a challenge in avoiding discontinuities in the data series that would adversely affect the time series required by key data users. Several methods are available for bridging the gap between old and new series. For example, the experience of the Census Bureau in bridging the potentially substantial discontinu- ity in series between the Standard Industrial Classification System (SIC) and the North American Industrial Classification System (NAICS) in the late 1990s is instructive. Beginning with 1998 data, the Census Bureau released County Business Patterns data classified by industry according to the NAICS, replacing the SIC categories. In order to facilitate comparison to earlier years, the Census Bureau developed estimates of the discontinued SIC data series for several years (U.S. Census Bureau, 2000a). Maintaining Relevance: Recommendation The panel credits that the Governments Division has been responsive to changing user needs in some respects. For example, the recent update of the 

66 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS Classification Manual added revenues from lotteries to the list of allowable categories for local governments in addition to state governments, and the division is also preparing a strategic plan, which provides for user input on data series (see Chapter 6). The panel thinks that the division needs to go further by developing a continuous process of identifying, evaluating, and following through on the most important needs of its users in both the short and the longer terms. To begin this process, the Governments Division should develop a roadmap that takes into account potential changes in the content and clas- sification system for the Census of Governments and annual surveys, the issues such changes raise, and the pros and cons of alternative changes with regard to relevance, historical continuity, burden, and needed resources. The roadmap should be based, in part, on input solicited from users and data providers. The feedback should be used to identify high-priority needs and preferred methods of response. This will not be an easy task and may consume considerable staff time over the next 2–3 years. It should have a goal of identifying changes that can be implemented not later than in the 2012 Census of Governments and subsequent annual surveys. If resources permit, implementation of priority changes could usefully begin with the annual surveys prior to the 2012 census to permit evaluation and the de- velopment of overlapping time series with new and old measures. A similar process should be repeated at regular intervals (see further discussion in Chapter 6). Recommendation 3-1: Over the next two to three years, the Govern- ments Division should seek input for and widely circulate a working paper that describes potential improvements to the detail and classifica- tion of the division’s data on state and local government finances and employment, the issues that each may raise, and the pros and cons of changes. Based on feedback from users, the division should develop a plan with well-justified priorities for improvements to be made in the 2012 Census of Governments and subsequent annual surveys. Maintaining Time Series In the state and local government statistics program, as with all federal government statistical programs, there is a tension between ensuring rel- evance, and thus the analytical value of the data, and maintaining historical continuity of the data series. Historical continuity cannot be maintained un- less some things remain constant, yet some data may no longer be relevant, while added or modified data may be required to shed light on new issues of analytical and public policy importance, such as the role of information technology in government operations. 

DATA USERS AND USES 67 In the past, the Governments Division has had to trade off historical continuity in data series, not necessarily to maintain currency but, rather, because of budget difficulties. For example, to generate cost savings to finance editing and processing improvements to the annual finance survey, the Census Bureau temporarily reduced the sample size for local govern- ments, for 2001 and 2003 only. The smaller sample was sufficient to de- velop estimates of local government finances for the nation as a whole, but it was not sufficient to develop estimates of local government finances for each of the states, resulting in the only breaks that have ever occurred in the history of these annual data series. The sample size reduction and as- sociated work to improve data processing required a carefully negotiated alteration in the production schedule, especially with several users in the Bureau of Economic Analysis.17 Not having state-by-state information for 2001 and 2003 was a mat- ter of great concern to researchers, who could not fully assess what was happening to the fiscal health of state and local governments during a time of recession or assess the different strategies adopted by governments to respond to revenue shortfalls. Many questions that could be answered with state-level data on local governments in prior recessions and recoveries could not be answered this time around, for example: Did state govern- ments that were particularly hard hit by the recession tend to cut back state aid to local governments? Which states were more likely to impose cutbacks, and did they vary by fiscal capacity, poverty level of the popula- tion, or region? How did state and local fiscal structures relate to response to the recession—for example, were expenditures during the recession more stable in states in which local governments relied more heavily on property taxes? Recommendation 3-2: The Governments Division should give priority to maintaining basic time series on state and local government finances and employment. It should avoid gaps and interruptions in basic time series, which undermine the ability of users to make consistent com- parisons over time and across jurisdictions. When new or modified content is introduced, the division should use such methods as over- lapping series or bridges between new and old series to assist users in making the transition. Taxable Property Values The demise of the Taxable Property Values Survey was of great con- cern to researchers and policy analysts in government finance. The panel 17  Correspondence with Henry Wulf, October 18, 2006. 

68 STATE AND LOCAL GOVERNMENT STATISTICS AT A CROSSROADS is cognizant of the major stumbling block to the reinstatement of the TPV Survey, namely, budget constraints. The Governments Division has esti- mated that the cost of resurrecting the survey is about $25 million, which represents virtually its entire budget. Nonetheless, the potential usefulness of the data for public- and private-sector decision making and for research to understand the finances and economic behavior of all sectors—govern- ment, household, and business—is great. The panel urges, as part of the division’s assessment of data needs and implementation strategies recom- mended above, that the Governments Division determine priority needs for this type of information and innovative methods that could be used to collect it. We recognize that collecting and categorizing this information is challenging. The basis for assessments has changed in many jurisdictions from fair market value to acquisition value and value-in-use, and differing governments may vary the basis of assessment for the same parcel. None- theless, the value of having the Governments Division collect such data is that the information provided would be consistent and comparable. Recommendation 3-3: In view of the importance of consistent, com- parable, objective data on property tax valuation and other features of property taxation by state and local governments, the Governments Division should carry out a program of research and testing to explore conceptually sound and cost-effective means of collecting these data, which could be in conjunction with, or independent from, the Census of Governments.