MCCORMICK SRI: GOING DEEP WITH CENSUS DEMOGRAPHIC AND ECONOMIC DATA EMPLOYMENT AND UNEMPLOYMENT ESTIMATES FROM THE U.S. DEPARTMENT OF LABOR, BUREAU OF.

MCCORMICK SRI: GOING DEEP WITH CENSUS DEMOGRAPHIC AND

ECONOMIC DATA

EMPLOYMENT AND UNEMPLOYMENT ESTIMATES FROM THE

U.S. DEPARTMENT OF LABOR, BUREAU OF LABOR STATISTICS (BLS)

Tom R. Rex, Associate Director, Center for Competitiveness and Prosperity Research,

L. William Seidman Research Institute, W. P. Carey School of Business,

Arizona State University00:00

Overview

• The employment and unemployment estimates produced by the BLS for states and substate areas result from a cooperative effort with a state agency in each state.

• Most of the state and substate estimates can be obtained from either the BLS or the state agency.– The state agency website may provide the data on a

more timely basis than the BLS website and may provide more detail than is available from the BLS.

– However, the websites of the various state agencies are not consistent.

00:15

Employment EstimatesFrom the Current Employment Survey

(CES) and the Quarterly Census of Employment and Wages (QCEW)

1:11

Summary of CES and QCEW• Employment is an estimate of the number of jobs, not

the number of people employed.– No distinction is made between a part-time and full-time

employee.– Employment is estimated for the pay period including the 12th of

each month.

• The estimates are limited to civilian wage and salary employees.– Those self employed, active military personnel, and most or all

farm workers and household workers are excluded.

• Employment and wages are reported by employers by establishment.

• Employers report the industry by establishment, but more than one type of work may be done in a facility.– Industries are classified using the North American Industry

Classification System (NAICS).

1:30

Comparison of CES and QCEW

CES• Sample of 440,000

establishments• Most wage & salary workers• Monthly employment and

wages released in following month

• Wages only, and only for production workers

QCEW• Census of 9,000,000

establishments• 99.7% of w&s workers• Monthly employment and

quarterly wages released 6 months after end of quarter

• Wages, bonuses, tips, etc. for all w&s workers

3:30

The CES Program• This was previously called the “790” program.• The national estimates are made separately from

the states and do not equal the sum of the states.• The estimates are subject to both sampling and

nonsampling error.• The month-to-month change in employment at

continuing businesses is derived from the survey.– A model is used to estimate employment at new

firms.• The estimates for the last year and most of the prior

year are “benchmarked” each February to data from the unemployment insurance (UI) program.– For states and substate areas, the magnitude of the

benchmark revisions may be large.4:40

National CES Estimates

• The estimates generally are released on the first Friday of each month.

• The estimates are revised in each of next two months; annually, estimates for the last year and most of the prior year are benchmarked.

• Due to sampling error, monthly changes in employment of less than 96,400 cannot be considered to be significant.– The 90% confidence interval of any monthly change

in total nonfarm employment is +- 96,400.

6:28

Monthly Change in National TotalNonfarm Employment in 2011

90% Confidence Interval Estimate Lower Limit Upper Limit Jan 68 -28 164 Feb 235 139 331 Mar 194 98 290 Apr 232 136 328 May 54 -42 150

7:44

Access to National CES Data www.bls.gov/ces/

• “CES News Releases”• “CES Tables and Charts”• “CES Databases”

– “Employment, Hours, and Earnings - National”: “one-screen data search”

• Multiple measures can be selected, relating to employment, hours worked, and earnings.

• Supersectors and then industries may be selected.• The default output is monthly data from 2001 through

2011, but data are available back to 1982.• Various formatting options are available, including the

calculation of the numeric and percentage change over time.

• “CES Publications”

8:44

Demo CES data

10:00

Subnational CES Estimates

• Estimates are available for regions, states, and metro areas from the BLS; county data are released by the state agency.

• The region and state estimates generally are released on the on the third Friday of each month; the metro area data are released 10 days later.

• Estimates are revised in the next month; each February, estimates for the last year and most of the prior year are benchmarked.

• The sampling error is directly related to the size of the sample and the percentage of the universe covered.

• The availability of sectoral and industrial detail varies by state and by substate areas.

• Due to very large sampling error, 44% of the substate sectoral data are model based using the survey results, an ARIMA projection, and the state data for that sector.

14:53

Access to Subnational CES Data www.bls.gov/sae/

• “Latest SAE News Releases”• “SAE Databases”

– “Employment, Hours, and Earnings - State and Metro Area”: “one-screen data search”

• Supersectors and then industries may be selected.• The available measures (relating to employment, hours

worked, and earnings) vary with the geography and the sectors/industries selected -- only employment may be available.

• The default output is monthly data from 2001 through 2011, but data are available back to 1990.

• “SAE Tables”• “SAE Publications”

18:26

Demo CES data

18:30

Alternative Source of CES Data wpcarey.asu.edu/bluechip/jobgrowth/

• Provides easy access to annual change in employment for regions, states, and metro areas.

• Displays rankings.

• The annual change is available back to 1991.

• Totals, supersectors, and sectors can be selected.

• Easy to put job growth in an area into geographic and historical perspective.

• The data may not be updated until the day after the BLS releases the monthly subnational estimates.– However, an advance look at the metro data is available at

the same time as the regional and state data.21:36

Demo JobGrowthUSA site

22:44

QCEW Estimates

• This was previously called the “ES-202” and “Covered Employment and Wages” program.

• In addition to employment and wages, the number of establishments are reported.

• The average weekly wage is calculated by dividing quarterly wages by the average of the three monthly employment figures, then dividing by 13.

• Not all multi-establishment employers report by establishment.

26:06

QCEW Estimates (cont’d)

• Missing employer data are imputed by the state agency.

• Considerable data may be withheld due to the federal disclosure law.

• The QCEW is not designed to be used as a time series.– Administrative changes (an employer changing the

industry, location, or ownership) are not reflected in the data for prior periods.

– The exception is that the prior year data are adjusted for these changes for the geographic areas reported in the news release.

26:45

Access to QCEW Estimates www.bls.gov/cew/

• “QCEW News Releases”

• “QCEW Databases”– “State and County Wages”: “one-screen data search”

• Can be used for metro and micropolitan areas, states, and counties (to choose a county, first select the state), and the nation.

• The amount of available detail varies with the geographic area, including industry, ownership (private vs. level of government), and establishment size.

• Multiple measures can be selected: employment, number of establishments, total wages, average weekly wage, and average annual pay.

– “Flat Files”: ftp://ftp.bls.gov/pub/special.requests/cew/

• “QCEW Publications”29:20

Demo QCEW data

30:02

Unemployment Estimates From the Current Population Survey (CPS)

32:21

Summary of Unemployment Estimates

• The CPS provides monthly estimates of the labor force, employment, unemployment, and unemployment rate.

• The estimates are released with the CES employment figures.

• The estimates are derived from a survey of households.– The estimates are of the number of persons, not jobs.– Thus, the definition of CPS employment is different

from CES employment.• An “unemployed” person is defined as not having

worked during the reference week and having actively looked for work in the last four weeks.

32:33

National Estimates of Unemployment

• The estimates are directly from the Current Population Survey (CPS).– The CPS surveys 60,000 households.

– The standard error of the unemployment rate is 0.1.

• The estimates are released early in the following month (typically the first Friday).

• No revision is made to the CPS estimates in the next month.– At the beginning of each year, seasonally adjusted

data are revised for the five prior years; population controls also are changed at this time.

33:55

Access to National Unemployment Datawww.bls.gov/cps/

• “CPS News Releases”• “CPS Databases”

– “Labor Force Statistics”: “one-screen data search”• Users can search by gender, race, ethnic origin,

age, education, and marital status.• Multiple measures can be selected, including the

unemployment rate.

• “Featured CPS Tables”• “CPS Publications”

34:52

Subnational Unemployment Estimates

• State estimates usually are released on the third Friday of the following month.– Estimates for metro areas and cities (with a minimum

population of 25,000) are released 12 days later.• Each monthly estimate is revised in the following

month; annual benchmarking is done at the beginning of each year.

• According to the BLS:– “Below the national level, the CPS sample is too small

and cannot support reliable estimates of even total employed and unemployed for states on a monthly basis.”

– “The CPS sample does not cover all counties.”• Therefore, subnational estimates are derived from

models.

35:37

Subnational Unemployment Models

• States: “signal-plus-noise time-series model”– Current and historical data from the CPS, CES, and the

unemployment insurance (UI) program are used.

– The model has three components: variable coefficient regression, flexible trend, and flexible seasonal component.

• Labor market areas: “handbook method”– For employment, adjustments are made to the CES

employment data.

– For unemployment, UI data are used for part of the universe and an econometric model is used for the balance.

• Cities and many counties: disaggregation method

36:28

Access to Subnational Data (Local Area Unemployment Statistics) www.bls.gov/lau/

• “LAU Economic News Releases”• “Featured LAU Searchable Databases”

– “Local Area Unemployment Statistics”: “one-screen data search”

• Users can choose a state or region/division, then the area, then seasonal adjustment.

• Labor force, employment, unemployment, and unemployment rate are displayed.

• “Tables and Maps Created by BLS”

37:49