Developing Uniform Measures of State Government Activity: Context, Classification and Census Bureau Data

The U.S. Census Bureau measures state and local government activity through the Census of Governments and related surveys. The data produced from these efforts are standardized across states and are the only nationwide dataset that allows for comparability both across states and on a national basis. Even with this standardization, the activities of 50 different state political systems present unique challenges to those who collect and use these data. We will introduce the principles by which the Census Bureau classifies governments and their activities. Additionally, through specific case studies, we will illustrate the ways states differ in their operation and in how they conduct the business of public service. Through these illustrations, we will offer a perspective that enables data users to delve into the data with a more thorough and accurate understanding, allowing them to formulate analyses more accurately. 

  Download the Article in PDF / E-Reader Compatible Format

About the Authors

Elizabeth Accetta is a section chief in the Public Sector Frame and Classification Branch in the U.S. Census Bureau’s Economic Statistical Methods Division. She holds a Masters Degree in Public Administration from the University of Pittsburgh and a Bachelor’s Degree in Economics from the University of Nevada Las Vegas.

Joseph Dalaker is a section chief in the Data User Outreach and Education Office of the U.S. Census Bureau’s Economy-Wide Statistics Division. He holds a Master of Public Policy degree from the University of Michigan and a Bachelor’s degree cum laude in Government and Economics from Cornell University

The 10th Amendment of the U.S. Constitution, similar to Article II of the Articles of Confederation, limited the power of the federal government so as to protect state autonomy. Because of this amendment, one can think of the states as 50 autonomous governments defining their own political processes and governmental structure within their boundaries, with divergent results. Therefore, those who use the U.S. Census Bureau’s statistics on governments should consider the ways in which a state government may have made different decisions about which public services get provided, how they get provided, what administrative tools are put in place to provide them and the structure of local governments within its borders.

In order to create uniform datasets on the organization, employment, and finances of state and local governments, the Census Bureau uses a system for classifying the functions, or activities, performed by governmental employees and by various financial transactions. This classification system also is supported by principles for identifying and defining which organized entities can properly be called “governments.” The Census Bureau conducts legislative research, and when necessary, contacts state or local officials, to keep track of the creation of new governments, the abolition of existing governments and governmental mergers, and to determine whether certain organizational entities meet the Census Bureau’s definition of a government for the purpose of measuring public sector activity. These principles enable the Census Bureau to create a uniform dataset even though  states are diverse in their  organizational arrangements and in how they conduct public business.

All the same, it is wise for data users to be aware of the Census Bureau’s classification principles and to appreciate the diversity among state governments. That is, data users should bear in mind how states are organized and conduct their business to understand the Census Bureau’s statistics that could be misleading if not viewed in light of that larger context.

The rest of this article will discuss the diverse ways in which state governments conduct their business, with an eye toward helping the users of Census Bureau public sector data better understand how the data are affected by those differences. We begin by discussing the Census Bureau’s definition of a government, which affects how we classify entities as independent governments or as dependent on another “parent” government. We then explain how the states’ choices of ending dates for their fiscal years affect the way users should analyze the fiscal impact of historical events. To illustrate further, we will discuss how the different ways states establish and work with dependent agencies can affect the dollar amounts shown in a variety of fiscal transactions, and that those amounts do not always agree with the amounts reported by the state governments themselves because the Census Bureau sometimes classifies the dependent agencies differently from state authorities. Last, we illustrate how states’ choices in the way they raise revenue, and how states classify that revenue, can complicate state-to-state comparisons. Through these illustrations, we hope that the reader will avoid making hasty conclusions when looking at state data, avoid treating all states identically, and instead appreciate the differences by which state governments operate and analyze Census Bureau statistics with that broader context in mind.

The Census Bureau’s Criteria for Classifying Governments

The Census Bureau defines a government as follows: A government is an organized entity which, in addition to having governmental character, has sufficient discretion in the management of its own affairs to distinguish it as separate from the administrative structure of any other governmental unit. To be defined as a government, any entity must possess all three of the attributes reflected in the foregoing definition: Existence as an organized entity, governmental character, and substantial autonomy.1 

The three attributes are further defined in the Census Bureau’s reports on Government Organization from the Censuses of Governments.2 In brief: Existence as an organized entity is demonstrated by: 

… The presence of some form of organization and the possession of some corporate power, such as perpetual succession, the right to sue and be sued, have a name, make contracts, acquire and dispose of property, and the like. … [S]ome entities not so specifically stated by law to be corporations [nevertheless] do have sufficient powers to be counted as governments. 

The mere right to exist is not sufficient. Where a former government has ceased to operate— [that is, it] receives no revenue, conducts no activities, and has no officers at present—it is not counted as an active government.3 

The presence of language in a state’s laws describing the establishment, merger or disincorporation of a government is an important indicator of an entity’s existence or lack thereof. However, quite often further research is needed to determine the entity’s actual status, including, for example, contacting the local officials for information on the entity’s activities and relationships with other governmental bodies.

Governmental character is a somewhat circular term when used to define governments, but nonetheless does denote that the entity must provide services, or wield authority or bear accountability that is of a public nature. The Census Bureau’s reports provide examples of indicators of governmental character, any of which may be sufficient, but not all are necessary:

Governmental character exists when officers of the entity are popularly elected or are appointed by public officials. A high degree of responsibility to the public, demonstrated by requirements for public reporting or for accessibility of records to public inspection, is also taken as critical evidence of governmental character.

Governmental character is attributed to any entities having power to levy property taxes, power to issue debt for which the interest is exempt from federal taxation. … However, a lack of either of these attributes … does not preclude a class of units from being recognized as having governmental character, if it meets the indicated requirements as to officers or public accountability. Thus, some special district governments that have no taxing powers and provide electric power or other public utility services also widely rendered privately are counted as local governments because of provisions as to their administration and public accountability.4 

Substantial autonomy consists of two components: fiscal independence and administrative independence. As described in the Census Bureau’s reports:

Fiscal independence generally derives from the power of the entity to:

  •  Determine its budget without review and detailed modification by other local officials or governments;
  •  Determine taxes to be levied for its support;
  •  Fix and collect charges for its services; or
  •  Issue debt without review by another local government.

Administrative independence is closely related to the basis for selection of the governing body of the entity. Accordingly, a public agency is counted as an independent government if it has independent fiscal powers and additionally: 

  • Has a popularly elected governing body;
  • Has a governing body representing two or more state or local governments; or
  • Even in the event its governing body is appointed, performs functions that are essentially different from those of, and are not subject to specification by, its creating government.5

An entity must demonstrate both fiscal independence and administrative independence to be considered a government. Census Bureau reports also provide examples of situations in which an entity may be considered to be a dependent agency of another “parent” government, owing to the makeup of its board—as being composed mainly of officials of the parent government—or provisions that its plans or budgets be reviewed and potentially altered by another government.6 

These dependent agencies—synonymously referred to as “subordinate agencies”—can present a challenge to data users in two ways. First, the Census Bureau may consider some entities to be dependent upon—and thus part of—another government in ways the entities or parent governments themselves may not recognize or agree with. It should be kept in mind that the Census Bureau endeavors to measure public sector activity without omission or duplication, and as such, the dependent agencies need to be included with their parent government in order to measure public sector activity both completely and consistently. 

Second, as discussed in the Census Bureau’s report Individual State Descriptions, dependent agencies: 

… can be involved in a wide variety of activities— school systems, universities, utilities, toll highways, hospitals, etc. … 

Contrasting examples of the existence of dependent agencies are found in New York City and Chicago. Almost all local government services in New York City are classified as part of the city government. … As a result, New York City government includes over 100 dependent agencies. By contrast, many of these services in Chicago are classified as functions of independent special district governments.7

Cautionary Examples of Using Census Bureau Statistics on Governments 

Fiscal Years

Not all governments end their fiscal years on the same date, nor can the federal government mandate to more than 90,000 governments that they end their fiscal years on a given date. As a result, there are a variety of fiscal year end dates for the governments throughout the nation. Even at the state level, there are differences in dates. For the most part, 46 of the 50 states end their fiscal year on June 30. The other four states have varying ends for their fiscal years:

  • New York ends on March 31;
  • Texas ends on Aug. 31; and
  • Alabama and Michigan end on Sept. 30.

Due to the general differences in fiscal year end dates, to collect financial data for a given fiscal year, the Census Bureau defines a reference period for the census and surveys on government finances. The defined reference period for a survey cycle begins on July 1 of year A and ends on June 30 of year A+1. The reference period would be known as “Fiscal Year A+1.” However, there are exceptions to this reference period. The states of Texas, Alabama and Michigan will include the fiscal years that extend beyond the reference period; similar instances for local governments include the District of Columbia and local school systems in Alabama, Nebraska and Texas.

Given the variation in fiscal years, a user needs to take caution when trying to use Census Bureau statistics on governments to analyze the effect of a time-specific event. A good example would be trying to analyze the government spending during and after a large-scale hurricane. A high category hurricane travels up the Gulf Coast hammering through Mississippi, Alabama and Florida, leaving the states with wind damage and flooding. How did the state governments respond to the natural disaster? How much federal money did these governments receive?

To continue the example using Figure A, let’s say that the event occurred in August 2014. A user wants to compare the government spending of Mississippi, Alabama and Florida to determine how each responded to the natural disaster. Mississippi and Florida have the fiscal year end date of June 30, but Alabama’s fiscal year ends on September 30.

When looking at Census Bureau government statistics for the 2014 fiscal year, and given the Census Bureau defined reference period, Mississippi and Florida would show data from their Fiscal year that ended on June 30, 2014. However, the data for this reference period for Alabama would include data through their fiscal year ending on Sept. 30, 2014.

Because of the differences in fiscal years, the effects of the hurricane in August 2014 would not be shown in the Mississippi or Florida data for the 2014 fiscal year. It would, however, be included for Alabama’s data. Therefore, a comparison between the three states for a single fiscal year would not be feasible. This is important to remember. A user would have a greater benefit by looking at the impact over at least two fiscal years of data. In this case, the differences in fiscal year periods alone would have an important effect, not to mention any long-range fiscal impacts such a storm might have.

Dependent Agencies

For the purpose of Census Bureau statistics, the term “state government” refers not only to the executive, legislative and judicial branches of a given state, but it also includes those agencies, institutions, commissions and public authorities that operate separately from the central state government but where the state government maintains administrative or fiscal control over their activities, as described earlier.

Consider the following example, which illustrates how recognizing the presence of dependent agencies can help one to understand the Census Bureau’s finance data. A data user looks to compare New Jersey’s debt reported in the state’s Comprehensive Annual Financial Report, also known as CAFR, to what is reported for the Annual Survey of State Government Finances (shown in Table A.) The user notices that the figures do not match. Why does it appear to the user that the Census Bureau is overstating the debt for the state?

When the user adds the general obligation bonds and revenue bonds for the state in the CAFR, they come to a figure of $23.9 billion. When that user consults the Census Bureau data, they see a figure for total outstanding debt of $64.2 billion. How can the difference be so great? The Census Bureau statistics gather outstanding debt from not only the central state government—as reported in the CAFR—but also from a number of other agencies the state considers to be “legally separate entities.” From the Census Bureau classification, however, these entities do not meet the criteria to be a separate, independent government and have shown either fiscal or administrative dependence, or both. Some of these agencies include such entities as the New Jersey Transit Corporation, New Jersey Turnpike Authority and New Jersey Educational Facilities Authority. 

State Government Tax Revenue

When analyzing tax data, especially when comparing tax data across states, it is essential to remember that the Census Bureau’s statistics on tax revenue reflect taxes a state collects from activity within the state and not necessarily from the residents of that state, because people, companies or other organizations from outside the state may be the ones performing the activities. As a result, analyses based on rankings or per capita statistics do not measure “burden,” nor lend themselves as easily to apples-to-apples comparisons, in the way a data user might expect. To understand the economic impact of taxes on states, it is not enough to compare only the total taxes collected by each state—one must also understand how those taxes are collected.

As an example, the following table highlights the total taxes collected for Florida and Alaska in 2013, the population estimates for each state and the tax revenue per capita. 

From this table, can a data user accurately state that the tax burden for a resident of Alaska is almost four times that of a person that lives in Florida? Not necessarily. The user should dig further into the source of the tax revenue in each state before establishing such a hypothesis. 

Alaska, for instance, does not have general sales taxes or individual income taxes, but it does collect severance taxes from companies that extract oil and natural gas. Like Alaska, Florida also does not collect individual income taxes. But unlike Alaska, Florida instead relies heavily on a general sales tax, which, because of its tourist industry, is partially supported by visitors from outside Florida. In that sense, both Alaska and Florida use “exported taxes”—taxes collected from people or organizations that may reside outside their state.

State-Specific Terminology

A state’s definitions of a certain activity will be set to fit their specific policy needs. As a result, we will see a variety of different terms to define the same activity or revenue. It is the responsibility of the Census Bureau to sift through the various terms and classify these activities according to standardized definitions.

For example, medical provider taxes have provided a large funding mechanism for many years. They have been used by state governments to generate the state matching funds needed to

  • Hospital Assessments
  • Provider Fees
  • Quality Assessments
  • Provider Taxes

When classifying these aforementioned assessments, fees and taxes into our own set of definitions, the Census Bureau looks beyond the terms used for the revenue and into the definition of the assessment, fee or tax. In each case, the definition included some form of per unit charge for a hospital or medical provider. For example, State Y’s “Hospital Assessment Fee” is charged to hospitals at a rate of $350 per inpatient bed day. State X’s “Hospital Services Tax” is charged to hospitals at a rate of 3 percent of gross revenues. Two different terms, and two different definitions for the state, but the Census Bureau will classify them in the same tax category (Other Selective Sales and Gross Receipts Taxes). 

The Census Bureau tax category “Other Selective Sales and Gross Receipts Taxes” is defined as: 

Taxes on specific commodities, businesses, or services not reported separately above. For state governments, includes sales or use taxes based on sale price, where the authorizing legislation is separate from the state’s general sales and use tax law. 

Since that definition does not yield a starkly unambiguous explanation, one needs to look to the definition of the subcategory for “Selective Sales and Gross Receipts Taxes:” 

Taxes imposed on the sale of particular commodities or services or on gross receipts. … 

Here, one sees that the definition will include taxes on services and gross receipts. So while State Y is collecting a rate per inpatient bed day—a unit of service—and State X is collecting a rate per gross revenues—or gross receipts—by the Census Bureau definition, both will be included in “Other Selective Sales and Gross Receipts Taxes.” 

Discussion: Using the Data

As discussed and illustrated above, the Census Bureau’s data may not entirely be in accordance with the state governments’ own accounts of their activities, but they do reflect a consistent approach toward creating a unified dataset that accounts for all public sector activity without omission or duplication. In order to make the best use of the data, the data user should understand the basic principles by which the Census Bureau classified governments and their activities, and in so doing, become sensitive to the differences by which governments in different jurisdictions conduct their business. 

Is there any way to do better than just “becoming sensitive” to the differences among governments? That is, can the data be made completely consistent, so that all governmental units and activities can be treated as identical? The 10th Amendment to the U.S. Constitution implies that the answer is a resounding “no:” 

The powers not delegated to the United States by the Constitution, nor prohibited by it to the states, are reserved to the states respectively, or to the people.

States, therefore, have been long-recognized as having the powers to organize their governmental affairs, and for setting the rules by which they, and the local governments within their purview, can operate. At this point, a careful analyst might worry that conducting valid analysis depends upon having identical units to compare and analyze. This article argues that in many cases, governments are not truly identical units. 

What, then, should the analyst do to ensure that his or her work is valid? Users of statistics on governments should appreciate the diversity of state governments by framing their research with sensitivity toward the dimensions talked about in this article: the timing of fiscal years when looking at historical events; the role of agencies that the Census Bureau classifies as dependent or subordinate to the state; the degree to which revenue may be obtained from people or groups outside the state’s boundaries; and the Census Bureau’s classification system, which focuses on the purposes of activities rather than on specific programs, and the purposes of financial transactions rather than specific funds. To that end, two resources cited earlier can assist data users in their analysis: the publication Individual State Descriptions, see footnote 2; and the Government Finance and Employment Classification Manual, 2006 edition, see footnote 8.


The Census of Governments and its related programs provide a rich source of data on state and local governments in the United States, not only on the number of governments by type, but also additional detail on their organization, employment and finances. Census Bureau staff apply a standard set of criteria while classifying governments and their activities in order to provide what is perhaps the only complete and uniform set of data on the activities of governments in the United States.

In making their classification decisions, Census Bureau staff do need to be sensitive to the differences of governance among states and local governments, and the effects those differences have on the data. Similarly, data users can strengthen their analyses when they too take account of the broader context within which states provide public services, bearing in mind the examples discussed in this article. 

Disclaimer: This report is released to inform interested parties of research and to encourage discussion of work in progress. The views expressed are those of the authors and not necessarily those of the U.S. Census Bureau. 

Acknowledgments: The authors would like to thank Stephen D. Owens, Lisa M. Blumerman, Brigitte Wehrs, Erika Becker-Medina, Melissa Therrien, Joy Pierson, Franklin Winters, and Kevin Deardorff of the U.S. Census Bureau for their thoughtful review and comments on drafts of this paper. 


1 U.S. Census Bureau, 1957 Census of Governments, vol. 1 no. 3, Local Government Structure. p. 3.
2 U.S. Census Bureau, 2012 Census of Governments, Individual State Descriptions, p. v et passim., accessed 3 March 2015.
3 Ibid.
4 Ibid.
5 Ibid.
6 Ibid.
7 Ibid., p. vii.
8U.S. Census Bureau, Government Finance and Employment Classification Manual. p. 4-12.
9 Ibid.