Chapter 3 from Concepts and Methods of the U.S. National Income and Product Accounts

comprises public domain material from the Bureau of Economic Analysis, U.S. Department

of Commerce.

This reading is optional, the students will read it only if they are interested to know more than is

required by the course.

3-1

CHAPTER 3: PRINCIPAL SOURCE DATA

(Updated: November 2011)

Source data as determinants of initial release and revision schedules

Source data for the current quarterly estimates

Source data for the annual revisions

Source data for the comprehensive revisions

Source data are the information BEA uses to prepare the NIPA estimates, and

estimating methods are the steps BEA takes to transform the source data into these

estimates. The interaction of source data and estimating methods determines the

accuracy, reliability, and relevancy of the accounts.

The data that BEA uses are collected from a variety of sources and are usually

collected for purposes other than for incorporation into BEA’s estimates. Data collected

by federal government agencies provide the backbone of the estimates; these data are

supplemented by data from trade associations, businesses, international organizations,

and other sources. The Government data are from a number of agencies, including the

Commerce Department’s Census Bureau, the Labor Department’s Bureau of Labor

Statistics (BLS), the Treasury Department, the Office of Management and Budget, and

the Agriculture Department. “Administrative” data are data that are tabulated by federal

government and by state and local government agencies as a byproduct of administering

their programs—such as processing corporate tax returns, regulating public utilities, and

issuing building permits. “Statistical” data are data collected by the federal statistical

agencies, such as the Census Bureau and BLS. These data consist of periodic economic

and population censuses and a wide range of sample surveys, such as those that collect

data on manufacturing and trade, employment, and prices. The relatively few surveys that

BEA conducts cover international trade in services and international direct investment,

both by foreign companies in the United States and by U.S. companies in foreign

countries.

The source data available to BEA are not always ideal for the preparation of the

NIPAs. BEA must develop methods that transform the best available data into estimates

that are consistent with the NIPA concepts and framework and that fill gaps in the

coverage of the source data. (See “Chapter 4: Estimating Methods.”)

Source data as determinants of initial release and revision schedules

The availability of the source data is an important consideration in determining

the schedules for the initial release and the subsequent revisions of the NIPA estimates.

One factor affecting availability is the speed with which the source data are collected,

compiled, and released. Another factor is whether the source data are part of a statistical

CHAPTER 3: PRINCIPAL SOURCE DATA

3-2

program that, over time, provides more complete or otherwise better coverage—for

example, if the sample is larger or if more detailed information is collected for an annual

survey than for the monthly surveys.

In general, the most comprehensive source data for the expenditure components

of GDP are available at the 5-year intervals associated with the economic census of

establishments conducted by the Census Bureau. The economic census is the primary

data source for BEA’s input-output accounts, which are used to “benchmark” the NIPA

estimates for the quinquennial census years—most recently, 1997 and 2002. Related

annual surveys are drawn from samples of the establishments covered in the economic

census; these surveys generally collect less detailed data than those collected in the

economic census. Many of the annual surveys are supplemented by monthly surveys that

involve smaller samples and that collect less detailed data than the annual surveys. 1 In

addition, responding to the censuses and annual surveys is generally mandatory, while

responding to most of the quarterly and monthly surveys is voluntary.

The data from the monthly surveys are available first, and they provide much of

the information that is used to prepare the initial, or “current,” quarterly (and for a few

components, monthly) NIPA estimates. These estimates are subsequently revised as

additional reports become available from the monthly surveys. Annual revisions, which

are timed to incorporate newly available annual source data, are usually carried out each

summer. Comprehensive revisions, which incorporate the most complete source data as

well as other improvements to the accounts, are carried out at about 5-year intervals.

The following sections describe the most important federal government source

data that are used for the current quarterly estimates and for the annual and

comprehensive revisions of the NIPAs. In the preparation of the estimates, these sources

are augmented by a wealth of information from other public sources and from private

sources, such as trade associations.

Source data for the current quarterly estimates

Data from Census Bureau monthly surveys are among the primary sources for the

current quarterly estimates (table 3.1). For the most part, the samples for these voluntary

surveys are drawn from the economic census, from the corresponding annual surveys,

and from the Business Register; the samples are updated periodically to account for new

businesses (“births”) and for businesses that discontinue operations (“deaths”). 2

1 Many of the annual and monthly surveys are based on “probability sampling” (sometimes known as

“scientific sampling”). In this process, establishments are first placed into various “strata” on the basis of

their size. Depending on the distribution of establishments, an establishment in the largest strata could have

a 100-percent probability of selection and thus have a sampling weight of 1—that is, the establishment

would represent only itself. An establishment in a smaller stratum would have a smaller probability of

selection, say 1 percent, but in that case the establishment would have a sampling weight of 100—that is,