|
Chapter 13.
Employment Projections
Projections
Data Base
The BLS approach to employment projections
requires a large amount of specialized data, much of
which is assembled especially for this purpose. On the
other hand, all of the data used originates in existing
general purpose statistical surveys and programs. The job
of BLS in this case is to adapt these sources to the
particular needs of the forecasting system and, most
importantly, to meld them into a coherent overall picture
of the industrial and occupational structure of the
economy
The data requirements of the system fall into four
broad areas: population and labor force, GDP by detailed
product, interindustry data, and industrial and
occupational employment. Data on the size and work status
of the population defined by age, sex, race and ethnicity
is available through the decennial population census and
the monthly Current Population Survey (CPS). Only minor
adjustments are required to provide the demographic data
base which supports the labor force projections. These
sources, particularly the CPS which is a monthly survey
of households, are also extremely important in assembling
other types of data as discussed below.
Final demand by detailed product is mostly extracted
from the U.S. National Income and Product Accounts.
Because the accounts are constructed primarily by the
commodity flow method they are relatively rich in product
detail. The BLS system uses time-series data on somewhat
over 200 product groupings derived from this source.
The U.S. input-output accounts are the source of two
critical items, the use and make tables which define the
structure of production and the bridge tables which
allocate the final demand product groups to detailed
commodities. Input-output is the area of greatest
difficulty in data development. The U.S., like many other
countries, does not produce detailed annual input-output
tables, and its benchmark tables (released every 5 years)
are not available for years after the survey period. The
latest table (currently 1987) is generally updated to a
more recent year or years by estimating the row and
column sums and then applying a modified a balancing
procedure commonly referred to as RAS4 to adjust the base matrix to
the new sums. While this is not an ideal procedure it
does provide a more up-to-date starting point for the
projections and can yield information about trends in the
technical coefficients which is useful for projecting
them into the future. The bridge tables are an integral
part of the input-output accounts and share the same
problem of lack of timely data. The approach to updating
them is also similar except that explicit adjustments are
used in place of the more mechanical RAS balancing
technique. In addition to updating the input-output and
bridge tables, numerous adjustments are made to provide
consistency with the industrial and occupational
employment data described below.
Time-series data on industry and commodity output is
maintained for each sector in the BLS system. These, of
course, are key variables in the projection system and
are also used in updating the benchmark input-output
table. The series are based on a wide variety of basic
source data such as annual production and sales surveys
conducted by the Bureau of the Census and data on tax
collections from the Internal Revenue Service. The
contribution of BLS is to assemble these myriad sources
of data into a set of industry output measures which are
consistent both over time and with the benchmark
input-output table.
Time series on employment and hours by industry are
derived from three BLS sources for different groups of
workers: The Current Employment Statistics survey (or
establishment survey) for nonagricultural wage and salary
employment, production worker employment, and weekly
hours; the Current Population Survey, (or household
survey), for agricultural employment except agricultural
services, self-employed and unpaid family worker jobs and
hours, and private household workers; and unemployment
insurance data for employment in agricultural services.
Data needed to develop the occupational staffing
pattern matrix are derived from several sources.
Information on the occupational distribution of wage and
salary workers by industry (staffing patterns) is derived
from the Occupational Employment Statistics (OES) survey
conducted by State employment security agencies under a
BLS-State cooperative program for all but a few
industries. The OES survey is conducted on a 3-year
cycle, with roughly a third of the economy covered each
year. About 775 detailed occupations in more than 350
industries are surveyed, nearly all at the 3-digit
Standard Industrial Classification (SIC) level. In
developing the base-year matrix, occupations having fewer
than 5,000 workers are generally aggregated into similar
larger occupations or appropriate residuals. Also, most
industries employing fewer than 50,000 workers are
aggregated into residuals within the same 2-digit SIC, if
their staffing patterns are comparable to the residual.
In some industries adjustments are made to OES survey
staffing patterns because some occupations are not listed
separately in the survey questionnaire, but are included
in a residual category. To develop economy-wide
employment estimates for these occupations it is
necessary to disaggregate data from OES survey residuals.
Data from the decennial census are used for these
adjustments.
Adjustments also have to be made to staffing patterns
derived from sources other than the OES survey. For
example, the occupational classifications used to
classify Federal Government workers are more detailed
than those used in the matrix. Similarly, estimates of
occupational employment for self-employed workers, unpaid
family workers, and for workers in the private household
industry and agriculture, except agricultural services,
are derived from the CPS and must be adjusted to make
them comparable to the occupational classification used
in the matrix.
Once these data have been assembled, they are arrayed
in a matrix that shows occupational employment
distributed in percentages by industry. Because these
percentages are derived from surveys conducted in
different years, they are applied to total industry
employment estimates for the base year in order to
develop occupational employment estimates.
Maintaining and improving this data base is a prime
concern to BLS. When the projection effort at BLS first
got underway very little of this information was readily
available and a considerable effort was needed to define
and assemble the basic data on which the projections are
based. BLS continues to devote significant resources to
this task since improvement in methods and techniques
invariably involves developing new or refined data
series.
Footnotes
4 Ronald E. Miller and Peter D. Blair, Input-Output
Analysis: Foundations and Extensions (Prentice-Hall,
Inc., Englewood Cliffs, NJ, 1985), 276-294.
Next: Technical
References
|
Bureau of Labor Statistics
