Handbook of Methods > Current Employment Statistics - National

Handbook of Methods Current Employment Statistics - National Data sources

Current Employment Statistics - National: Data sources

The Bureau of Labor Statistics (BLS) collects data on employment, payroll, and paid hours from a sample of establishments each month. By federal law, participation in the Current Employment Statistics (CES) survey is voluntary, although it is mandatory by state law in New Mexico, Oregon, South Carolina, and in Puerto Rico. (Data for Puerto Rico is excluded from CES-national estimates.) Legal citations for state requirements are listed on the CES report form. The CES-State and Area (CES-SA) program uses the same sample and collection methods, thus references to CES apply to both CES-N and CES-SA programs.

To encourage participation, BLS uses a variety of collection techniques that are tailored to individual firm preferences. Data collection centers (DCCs) perform initial enrollment of each firm via telephone to collect the data for several months via computer assisted telephone interviewing (CATI), and where possible, convert respondents to web collection. Large multi-establishment firms and government agencies may be enrolled to provide data by the electronic data interchange (EDI) center. These respondents provide electronic files to BLS that include data from all their worksites. A small percentage of respondent data is collected by other methods.

Sample implementation

CES enrollment efforts begin immediately after a sample is selected from the unemployment insurance (UI) file, with collection generally beginning the first month after enrollment. Each quarter, about one-fourth of all new sample units selected are enrolled and data collection begins.

All newly sampled companies are classified into one of four industry groups for enrollment and initiation of data collection during a specific quarter. Use of the newly sampled units in CES estimates begins the first month following the quarter of enrollment and sample initiation. (See exhibit 1.) All birth units selected as part of the semiannual update are implemented in the last group, regardless of industry.

Exhibit 1. Industry groupings for the Current Employment Statistics (CES) quarterly sample rotation
Group	CES industry code	Major industry sector	Timing
Group	CES industry code	Major industry sector	Enrollment quarter	Estimation month
Group 1	10-000000	Mining and logging	1	April
	41-420000	Wholesale trade
	42-000000	Retail trade
	43-000000	Transportation and warehousing
	44-220000	Utilities
	55-000000	Financial activities
Group 2	20-000000	Construction	2	July
Group 2	70-000000	Leisure and hospitality	2	July
Group 3	50-000000	Information	3	October
	60-000000	Professional and business services
	80-000000	Other services
Group 4	30-000000	Manufacturing	4	January
	65-000000	Private education and health services
	Birth units for all private industries sampled from the third quarter of the longitudinal business database (LBD) that did not exist on the first quarter LDB

Data collection

Respondents submit monthly employment, hours, and earnings data from their payroll records to BLS by one of several transmission modes. Respondents report data for the pay period that includes the 12th of the month. BLS requests employee data for all paid workers including: all employees and women employees in all industries, production employees in goods-producing industries, and nonsupervisory employees in private service-providing industries. Total payrolls, commissions, and hours paid (including those for overtime and paid leave) are requested for all employees and for production and nonsupervisory employees. Overtime hours are requested for manufacturing industries only.

BLS provides a CES reporting form (BLS form 790 series) to all CES respondents except large multi-establishment respondents who report via electronic file. Six versions of the basic CES form are used, and each variation is tailored for the data items, concepts, and definitions for groups of major industries. Separate forms are used for mining and logging, construction, manufacturing, service-providing industries, public administration, and educational services. Except for a small percentage of respondents, the forms are used only for reference and are not returned to BLS.

Visit the BLS website for available CES data collection forms.

Primary collection methods

The majority of CES data are collected by one of three primary methods. The Data Collection section of the CES frequently asked questions webpage shows the most recent distribution of CES sample by data collection mode.

Computer assisted telephone interviewing (CATI)

Nearly all establishments reporting payroll data for CES were originally enrolled by one of four data collection centers using CATI. After initial enrollment, CES continues to collect data by CATI for several months before offering respondents the option of reporting by web. The initial CATI period provides time for CES to instruct respondents on the proper definitions of CES data items. Respondents that prefer CATI over web continue reporting by the CATI method.

The CES CATI system provides real-time screening of data that allows CES staff to ask follow-up questions at the time of collection to confirm the accuracy of data.

Web reporting

CES has two different web-reporting systems: one designed for firms reporting for either a single or a small number of worksites and a separate system for firms reporting for a larger number of worksites.

The single establishment web reporting system is designed for respondents to enter payroll data for one establishment at a time. Each month, CES sends an email to web respondents on or around the 12th of the month as an advance reminder to report their data for the reference pay period. This email contains a link to the single establishment web reporting system along with the report numbers that belong to that respondent. This method is ideal for respondents that provide data for a small number of reports because each report must be entered separately and submitted before continuing to another report. This reporting system is designed for simplicity and contains only basic screening of data to alert respondents to potential errors.

The multi-establishment web reporting system is designed for firms that report for a large number of establishments. Each month, CES sends an email to web respondents on or around the 12th of the month as an advance reminder to report their data for the reference pay period. The email contains a link to the multi-establishment web reporting system, which lists all establishments provided by the respondent on a single page in a spreadsheet style format. This design allows for more efficient data reporting for larger firms. This web reporting system has more thorough data screening than the single establishment web reporting system and alerts the respondent to several different types of data anomalies.

After data are submitted using either web method, CES screens data thoroughly and contacts respondents with any questions about potential anomalies in the data.

Electronic data interchange (EDI)

The EDI center processes electronic files submitted by large firms or government agencies. Each month, these respondents provide data files using a consistent format. The files can contain payroll data from 50 to 100s or even 1,000s of worksites. Analysts in the EDI center are responsible for processing the files and reviewing the data for accuracy. EDI is an extremely efficient method of data collection for firms that report for 50 or more establishments.

Legacy collection methods

Several collection methods are no longer offered to new respondents but are still used for a small number of existing respondents. For these establishments that do not use the primary collection methods, data are collected using nonstandard electronic files that require custom processing.

Editing and screening of microdata

All reported data, regardless of method of collection, are edited by BLS to ensure the information is correctly reported and consistent with the data reported by the establishment in earlier months. The data are further edited to detect processing and reporting errors that might have been missed during collection. When questionable reports are discovered at any stage of the editing process, BLS contacts the respondent for clarification or correction. BLS staff prepare estimates of employment, hours, and earnings using the edited data.

The CES program tests all respondent data, collectively known as microdata, to generate accurate, timely, and relevant monthly employment, hours, and earnings estimates. These microdata screening tests compare all newly reported data to the respondent’s historically reported data. Data that fail these microdata screening tests are then reviewed by analysts to determine whether the microdata should be used in estimation.

All edit and screening tests for a given report use only that specific respondent’s historically reported and current data. The tests do not incorporate data from other respondents, estimates, or sources. All payroll, commissions, and hours are normalized to weekly equivalents based on a respondent’s reported length of pay period. CES derives additional data types for each respondent when possible based on reported data. For example, average hourly earnings for a respondent is derived by dividing reported payroll by reported hours. Exhibit 2 describes each data type or variable used in the edit and screening process.

Exhibit 2. Current Employment Statistics data types and other payroll variables
Abbreviation	Basic data types
AE	All employees
AE Commissions	Total commissions of all employees
AE Hours	Total hours of all employees
AE OT	Total overtime hours of all employees (manufacturing only)
AE Payroll	Total payroll of all employees
PE	Production and nonsupervisory employees
PE Commissions	Total commissions of production and nonsupervisory employees
PE Hours	Total hours of production and nonsupervisory employees
PE OT	Total overtime hours of production employees (manufacturing only)
PE Payroll	Total payroll of production and nonsupervisory employees
WE	Women employees
Derivative data types
AE AHE	Average hourly earnings of all employees (corresponding payroll divided by hours (payroll/hours))
AE AWOH	Average weekly overtime hours of all employees (corresponding normalized overtime hours divided by all employee count (OT/AE))
AE AWH	Average weekly hours of all employees (corresponding normalized total hours divided by all employee count (Hours/AE))
PE AHE	Average hourly earnings of production and nonsupervisory employees (corresponding payroll divided by hours (PE payroll/PE hours))
PE AWOH	Average weekly overtime hours of production and nonsupervisory employees (corresponding normalized overtime hours divided by production and nonsupervisory employee count (PE OT/PE))
PE AWH	Average weekly hours of production and nonsupervisory employees (corresponding total normalized hours divided by production and nonsupervisory employee count (PE hours/PE))
PE ratio	Production-and-nonsupervisory-employee-to-all-employee ratio (PE/AE)
WE ratio	Women-employee-to-all-employee ratio (WE/AE)
Other Payroll Variables - Length of Pay Period (LP)
AE Commissions LP	LP for commissions for all employees
AE LP	LP for all employees
PE Commissions LP	LP for commissions for production and nonsupervisory employees
PE LP	LP for production and nonsupervisory employees

Each reported or derived data type must pass a series of edit and screening tests to determine the validity of the record. The tests are divided into four sequential categories: strict edit tests, nonstrict edit tests, screening tests for all data types, and screening tests for specific data types. Exhibit 3 shows the data types that must be reported to produce specific CES estimates.

Exhibit 3. Sample required to produce basic estimates, shown in the order estimated
Basic estimate		Required data types^⁠1
1	All employees	AE
2	Women employees	AE, WE
3	Average weekly hours of all employees	AE, AE Hours
4	Average hourly earnings of all employees	AE, AE Hours, AE Payroll, AE Commissions^⁠2
5	Average weekly overtime hours of all employees (manufacturing only)	AE, OT
6	Production and nonsupervisory employees	AE, PE
7	Average weekly hours of production and nonsupervisory employees	AE, PE, PE Hours
8	Average hourly earnings of production and nonsupervisory employees	AE, PE, PE Hours, PE Payroll, PE Commissions^⁠2
9	Average weekly overtime hours of production and nonsupervisory employees (manufacturing only)	AE, PE, PE OT
^⁠1 CES acronyms are defined in exhibit 2. ^⁠2 Commissions are used in estimating average hourly earnings only if they are paid at least monthly or more frequently. Both commissions and payrolls are normalized to weekly equivalents and are summed to get total payroll for estimation purposes.

Strict edit tests

Several conditions known as strict edits must be met for microdata to be used in CES estimates. For example, the number of women employees cannot be greater than the total of all employees reported for the reference pay period. All microdata are processed to identify strict edit errors. If a microdata record fails any of the strict edit tests, all data types used in the specific test fail, become excluded from estimation, and are returned to the data collection group for correction. See exhibit 4 for a comprehensive list of strict edit tests used by CES.

Exhibit 4. Strict edit tests
Data types^⁠1	Condition required (microdata will be eligible for use in estimation)
AE, AE payroll, AE hours, AE OT, PE, PE payroll, PE hours, PE OT, WE	AE must be reported
PE, PE payroll, PE hours	If PE payroll and PE hours reported, then PE must be reported
AE payroll, AE hours	If AE hours reported, then AE payroll must be reported
AE hours, AE payroll	If AE payroll reported, then AE hours must be reported
AE payroll, AE hours, AE commissions	If AE commissions reported, then both AE payroll and AE hours must be reported
AE payroll, AE hours, AE OT	If AE OT reported, then both AE payroll and AE hours must be reported
PE payroll, PE hours	If PE hours reported, then PE payroll must be reported
PE hours, PE payroll	If PE payroll reported, then PE hours must be reported
PE payroll, PE hours, PE commissions	If PE commissions reported, then PE payroll and PE hours must be reported
PE payroll, PE hours, PE OT	If PE OT reported, then both PE payroll and PE hours must be reported
AE payroll	If AE payroll reported, then AE LP factor is required
PE payroll	If PE payroll reported, then PE LP factor is required
AE commissions	If AE commissions reported, then AE commissions LP is required
PE commissions	If PE commissions reported, then PE commissions LP factor is required
AE, AE payroll, AE hours	If AE or AE payroll or AE hours equals zero, then all must equal zero
AE, WE	AE must be greater than or equal to WE
AE, AE commissions	If AE commissions positive, then AE must be greater than zero
AE, AE OT	If AE OT positive, then AE must be greater than zero
PE, PE payroll, PE hours	If PE or PE payroll or PE hours equals zero, then all must equal zero
PE, PE commissions	If PE commissions positive, then PE must be greater than zero
PE, PE OT	If PE OT positive, then PE must be greater than zero
AE, PE	AE must be greater than or equal to PE
PE payroll, AE payroll	AE payroll must be greater than or equal to PE Payroll
AE commissions	If AE commissions LP indicates 'no commissions' are paid or AE commissions is paid less frequently than once a month, then AE commissions should not be reported
PE commissions	If PE commissions LP indicates 'no commissions' are paid or PE commissions are paid less frequently than once a month, then PE commissions should not be reported
PE hours, AE hours	AE hours must be greater than or equal to PE hours
AE OT, AE hours	AE hours must be greater than AE OT
PE OT, AE OT	AE OT must be greater than or equal to PE OT
PE OT, PE hours	PE hours must be greater than PE OT
AE	AE should be less than 199,999
PE	PE should be less than 199,999
AE AWH	AE AWH must be less than or equal to 168 hours
AE AHE	AE AHE in any private-sector industry, except food services and drinking places, must be no lower than 50 cents below the federal minimum wage
AE AHE	AE AHE in food services and drinking places must be no lower than 70 percent of the federal minimum wage
PE AWH	PE AWH must be less than or equal to 168 hours
PE AHE	PE AHE in any private-sector industry, except food services and drinking places, must be no lower than 50 cents below the federal minimum wage
PE AHE	PE AHE in food services and drinking places must be no lower than 70 percent of the federal minimum wage
^⁠1 CES acronyms are defined in exhibit 2.

Nonstrict edit tests

Nonstrict edit tests, shown in exhibit 5, are designed to identify microdata reports that are possible, but highly unlikely. For example, one test condition determines if average hourly earnings are 25 times greater than the minimum wage. If a test condition is true and no explanatory comment code has been entered, the data type fails and, therefore, the microdata record fails. An analyst then reviews the failed record to determine whether follow up with the respondent is necessary or if the record can be used in estimation.

Exhibit 5. Nonstrict edit tests
Data types^⁠1	Test condition
AE, AE hours, AE LP	AE AWH greater than 65 hours
AE payroll, AE hours	AE AHE greater than 25 times the federal minimum wage
PE, PE hours, PE LP	PE AWH greater than 65 hours
PE payroll, PE hours	PE AHE greater than 15 times the federal minimum wage
AE, AE OT, AE LP	AE AWOH greater than 25 hours
AE, AE hours, AE OT, AE LP	AE AWOH greater than one-half AE AWH
PE, PE OT, PE LP	PE AWOH greater than 25 hours
PE, PE hours, PE OT, PE LP	PE AWOH to be greater than one-half PE AWH
PE	PE equals zero and AE greater than zero
^⁠1 CES acronyms are defined in exhibit 2.

Screening tests for all data types

All microdata records that pass the first two categories of edit tests are then screened by data type for unusual percentage and level changes. (See exhibit 6.) Failure of all screening tests for the data types that the respondent has reported results in that data type record being sent to an analyst for review. If any data type passes any one of the screening tests, that data type record passes to the final category of screening tests. If a respondent has not reported all of the historical data required to perform a screening test, that test does not factor into the results as a pass or failure.

Exhibit 6. Screening tests for all data types
Condition required	Variables tested^⁠1
Condition required	AE	WE	WE ratio	AWH	AHE	AWOH	PE	PE ratio	PE AWH	PE AHE	PE AWOH
1-month percent change less than X_1i	✓		✓	✓	✓	✓		✓	✓	✓	✓
2-month percent change less than X_1i	✓		✓	✓	✓	✓		✓	✓	✓	✓
1-month percent change differs by no more than X_2i percentage points from the 1-month percent change 12 months ago	✓		✓	✓	✓	✓		✓	✓	✓	✓
2-month percent change differs by no more than X_2i percentage points from the 2-month percent change 12 months ago	✓		✓	✓	✓	✓		✓	✓	✓	✓
1-month percent change differs by no more than X_2i percentage points from the 1-month percent change 13 months ago	✓		✓	✓	✓	✓		✓	✓	✓	✓
1-month percent change differs by no more than X_2ipercentage points from the 1-month percent change 11 months ago	✓		✓	✓	✓	✓		✓	✓	✓	✓
1-month change is less than K_1i	✓	✓		✓	✓	✓	✓		✓	✓	✓
2-month change is less than K_1i	✓	✓		✓	✓	✓	✓		✓	✓	✓
12-month change is less than K_1i	✓	✓		✓	✓	✓	✓		✓	✓	✓
1-month change is less than a tolerance value defined by the product of the average maximum and minimum 1-month change, the critical T value, and the D factor^⁠2	✓		✓	✓	✓	✓		✓	✓	✓	✓
12-month change is less than a tolerance value defined by the product of the average maximum and minimum 12-month change, the critical T value, and the D factor^⁠2	✓		✓	✓	✓	✓		✓	✓	✓	✓
Comment code indicates changed basis of reporting	✓	✓		✓	✓	✓	✓		✓	✓	✓
^⁠1 CES acronyms are defined in exhibit 2. ^⁠2 T_critical is based on the 0.975 percentile of a standard t-distribution. In order to process the statistical t-tests, a respondent must have reported data for the prior 6 months, at a minimum. D factors are used to provide estimates of the variance using a respondent's range of employment values.

There are some exceptions to failures of screening tests for all data types. For example, if a respondent that previously reported only one set of payroll numbers suddenly provided a breakout of microdata for multiple worksites, it would likely fail the screening test for several data types. However, if the data were reported along with a code that indicated a change in basis of reporting, the data would be considered correct for the current month and would pass screening. The data will not be used in the current month of estimation but may be matched with the following month’s reported data if the subsequent month of data is reported with the same basis of reporting.

Screening tests for all data types use a large set of a variable (X) and constant (K) factors that vary by industry and by data type. Industry analysts adjust these factors if screening tests result in excessive amounts of microdata failures that are determined to be correct or excessive amounts of bad microdata passing through the screening tests. For example, data from a large, multiple establishment business or government agency may regularly fail screening tests. If the respondent regularly confirms the accuracy of their data, the industry analyst may adjust an X or K factor to allow the respondent’s data to pass. These X and K factors are not available to the public in order to prevent disclosure of CES respondents and their payroll information.

Screening tests for specific data types

The final category of screening tests are thresholds and conditions for the specific data type being tested. (See exhibit 7.) Failure of all tests results in a record being sent to an industry analyst for review. Tests not processed because of insufficient historical data for a piece of microdata are excluded from counts of passed and failed tests.

Exhibit 7. Screening tests for specific data types
Data types^⁠1	Condition required
AE	Comment code corroborates AE increase or decrease
AE, PE	The 1-month percent change in non-production worker employment is less than or equal to factor X₄
AE, PE	The 1-month change in non-production worker employment is less than or equal to factor K₂
AE, AWH	The relative 1-month change in AE is greater than K3 and the 1-month change in AWH is less than K₄
AE, AHE	The relative 1-month change in AE is greater than K5 and the 1-month change in AHE is less than K₆
AE, AWOH	The relative 1-month change in AE is greater than K5 and the 1-month change in AWOH is less than K₇
PE, PE AWH	The relative 1-month change in PE is greater than K3 and the 1-month change in PE AWH is less than K₄
PE, PE AHE	The relative 1-month change in PE is greater than K5 and the 1-month change in PE AHE is less than K₆
PE, PE AWOH	The relative 1-month change in PE is greater than K5 and the 1-month change in PE AWOH is less than K₇
^⁠1 CES acronyms are defined in exhibit 2.

Analyst review

An analyst reviews microdata that failed screening tests. The analyst considers the failed data, the respondent’s historical data, comment codes, and any information gleaned from data reconciliation contacts. Based on this information, the analyst may accept the microdata as originally reported for use in estimation or exclude the microdata from the sample and the current month's estimation. If excluded, the analyst may also request additional clarification or correction from data reconciliation contacts.

Confidentiality

BLS requires that all published data meet strict quality and privacy guidelines. These guidelines are designed to ensure that adequate sample exists to produce statistically sound estimates and to protect the confidentiality of survey respondents. CES estimates are subject to annual review to determine if they meet BLS publication and disclosure standards. Failure to meet the standards may stem from inadequate sample size, inadequate sample response rates, or dominance of the sample by a few reporters. The laws governing BLS confidentiality and disclosure tests are described below.

Law 29 U.S.C.2 authorizes BLS to collect payroll data for the CES survey. All BLS employees, agents, and partner statistical agencies pledge to use the information provided by respondents for statistical purposes only and hold the information in confidence to the full extent permitted by law. In accordance with the Confidential Information Protection and Statistical Efficiency Act of 2002 (Title 5 of Public Law 107-347) and other applicable Federal laws, responses are not disclosed in identifiable form without informed consent of the respondent. Per the Federal Cybersecurity Enhancement Act of 2015, Federal information systems also are protected from malicious activities through cybersecurity screening of transmitted data.

In addition to protection of raw data provided by respondents, estimates are also processed through a series of statistical tests to prevent potential disclosure of individual respondents and their payroll data. These sample adequacy and disclosure tests are confidential and cannot be released.

For more information about the laws that protect respondents to BLS surveys, visit the BLS confidentiality policy webpage.

Last Modified Date: February 02, 2024