|
How Do We Measure the
Labor Force Statistics? (pdf)
- Nationally,
labor force statistics are derived directly from the Current
Population Survey (CPS), which is conducted by the Census Bureau for
the Bureau of Labor Statistics (BLS). The survey is based on a
selected sample of households and includes the population 16 years of
age and over. The national CPS sample includes approximately 60,000
households.
- On
the first Friday of each month, BLS releases the national labor force
statistics for the preceding month. These figures reflect an estimated
count of the employed and unemployed persons for the nation, who
together, comprise the labor force. BLS also publishes a monthly
seasonally adjusted (the official estimate) and a not seasonally
adjusted unemployment rate. The unemployment rate represents the
percentage of the labor force that is unemployed.
- In
Rhode Island, the CPS includes about 1,200 sample households. However,
unlike the nation, the RI sample is not comprehensive enough to
reflect the State’s true labor force estimates. Therefore, BLS
developed a methodology for those states that could not use the CPS
sample data directly.
- In
the old days (prior to 1989), Rhode Island’s labor force figures
were computed manually using a BLS-approved methodology called the
“Handbook Method.” This multi-step process was similar to a
building block approach and relied heavily on data from two sources -
Current Employment Statistics (CES, another BLS program), and
administrative data records. Even though BLS developed this
methodology, the state analyst was still very involved in the
estimating process.
- However,
it was determined that the Handbook Methodology was not an ideal way
to produce estimates. The estimates were not actually comparable
state-to-state because this method did not fully account for
differences in state Unemployment Insurance (UI) laws and state
economies.
- In
1989, BLS introduced a new “Regression” Methodology for developing
labor force estimates for the 39 smaller states (including Rhode
Island) and the District of Columbia. At that time, the eleven larger
states used data directly from the CPS. BLS used more than ten years
of data from the CPS, the CES, and the UI system – sources which
covered employed and unemployed persons in all sectors of the economy
– to develop regression equation models.
- Advantages
for using model-based estimation includes a reduction in the number of
state-supplied inputs, consistency in the estimates from
state-to-state, and an automated approach using a PC to access the
individual state models. A disadvantage to the model-based approach is
the analyst became less involved in the actual production of the
estimates.
- Effective
with the estimates for January 2005, all states began using a
web-based model system developed by BLS. Web
STARS (State Time Series Analysis and Review System) represents
the third generation of regression models.
- One
of the major differences between the new Web STARS and the
prior STARS version is that states are now grouped into one of nine
Census Divisions. State estimates are benchmarked (a benchmark is a
reliable total to which less reliable estimates are controlled) to the
national CPS estimates of employment and unemployment each month via
the Census Division models.
- The
process of adjusting the monthly state estimates to the national total
via the Census Division is referred to as “real-time
benchmarking.” This simply means that the process occurs as part of
the normal monthly estimation (in real-time). Prior to the
implementation of Web STARS, the adjustment process occurred
only once a year.
- The
new Web STARS requires a state analyst to enter nonfarm payroll
employment and UI claims information into the web application each
month. Once the inputs are entered, BLS takes over the estimation
process. Final estimates cannot be produced for a state until all
states in the relevant Census Division have entered their inputs.
- In
addition to the “real-time benchmarking,” a modified annual
historical benchmarking still occurs at the end of the year. It
involves updating model inputs (revisions to employment and claims
data, which the state analyst enters directly into Web STARS).
It also involves updating the population controls, smoothing the model
estimates and controlling the state estimates to the revised monthly
historical benchmarked estimates at the Division level (BLS performs
these activities), which, in turn, sum to the monthly national CPS
estimates.
- The
models provide monthly state estimates of the civilian labor force,
the resident employed, the number unemployed, and the unemployment
rate for both the seasonally adjusted and not seasonally adjusted data
series. During the annual benchmarking process, state estimates are
revised historically. Typically, the seasonally adjusted series is
revised back five years and the not seasonally adjusted series is
revised back three years.
Note: The
methodology to produce the state labor force statistics (resident labor
force, resident employed, unemployed and the unemployment rate) was
developed by the Bureau of Labor Statistics (BLS). Currently,
all states adhere to this methodology. Although
the analyst enters data into the STARS web-based application, BLS is
ultimately responsible for generating the final estimates and for
seasonally adjusting the data.
|
FAQ
Home