Local Employment Dynamics:Partnership, Employment,

and Public-Use Data

Earlene Dowell and Heath HaywardLEHD Program

Center for Economic StudiesU.S. Census Bureau

2

Outline

Welcome and Introduction to LED Basics of LED Data: LODES and QWI QWI Explorer (Demo and Example Scenarios) OnTheMap (Demo and Example Scenarios) Advanced Scenarios/Coming Soon/Questions

and Feedback

3

Local Employment DynamicsPartnership

Then: Begun in late 1990s with a few states Goal to generate new labor market statistics from

existing records (UI and firm info) Now:

53 partner states/territories 3 data products 4 web-based data tools A culture of innovation and cost savings

4

What’s In a Partnership?

Sharing of costs (and data) Breadth of expertise Diversity of ideas and needs National scale and local knowledge But it requires commitment and

maintenance…

5

Building on State Inputs We combine state records with other

admin/census/survey data from the Census Bureau and other Federal agencies

We can then create public statistics on: Firms & Establishments Jobs & Workers By Firm and Person Characteristics

Without new respondent burden

Admin. Records & LED Infrastructure

QCEW*

Economic Survey Data

Business Register

UI* Wage Records

Federal Records

Demographic Census/Survey

Data

OPM*

Public-UseData Products…

QCEW = Quarterly Census of Employment and WagesUI = Unemployment InsuranceOPM = Office of Personnel Management

Linked National Jobs Data

Firm DataJobs Data

Person Data

• Job data cover over 95% of private employment and most state, local, and federal jobs

• Data availability: 1990-2014, start year varies by state, rolling end date

7

Person Data

Protecting Personal Information

Some records enter the Census Bureau with SSNs, some with other personally identifiable information

First, the SSN is replaced with a “protected identification key” (PIK).

The PIK is used for all further matching.

9

LED Data Products Quarterly Workforce Indicators (QWI)

Employment, Job Creation, Job Destruction, Hires, Separations, Turnover, Earnings

By industry, county, and worker characteristics LEHD Origin Destination Employment Statistics

(LODES) Employment and Workplace-Residence Connections Detailed geography + firm/worker characteristics

Job-to-Job Flows (Beta) Data being released over coming months

10

Data Product Infrastructure – Primary Unit of Analysis: Job

Association of: Worker–Employer–Year–Quarter Workers can have multiple jobs within a quarter

“Primary Job” (greatest earnings) - not defined separately in QWI, but is in LODES/OnTheMap

In contrast, most other surveys and censuses are: Household-based (ACS, CPS, Decennial), or Employer-based (QCEW, Current Employment Statistics)

Advantage of job-based frame – can produce tabulations by both worker and firm characteristics

11

Core Data Input: UI Earnings Records

UI = Unemployment Insurance Record of individual earnings for covered jobs

Administrative wage records, not UI claims data Collected for operation of state UI program

UI benefits are based on historical earnings Includes:

Total quarterly earnings for each job Firm identifier = State UI account number (SEIN) Worker identifier = Protected Identification Key (PIK)

Census identifier based on SSN

12

Job Coverage in UI Earnings Data Most private sector jobs covered

For-profit and not-for-profit classified together (as per QCEW standard)

State and local government also in system, though some reporting inconsistencies

Federal worker data from Office of Personnel Management (OPM) not yet available in QWI (have been incorporated into LODES/OnTheMap)

Self-employed not available Massachusetts data not available yet

13

Additional Data Inputs

UI wage records are linked to a variety of other data sources

Sources of establishment information: Quarterly Census of Employment and Wages (QCEW) Business Dynamics Statistics (BDS)

Sources of demographic information: Decennial Census Federal Tax Records Social Security Administration Records Other census and administrative records

This additional information enables tabulations by detailed worker and firm characteristics

Creating LODES Data: Sources

Confidential Data Sources: UI Wage Records QCEW Other Censuses and Surveys StARS

Public Use Data Sources 2000 Decennial (SF1, CTPP) TIGER/Line Shapefiles Previous year of OnTheMap

LEDInfrastructure

Files

Creating LODES Data: Processing

Unlike QWI, much of the processing is performed at the national level, starting with the definition of Primary/Dominant Job.

Also unlike QWI, data for previous years are not reproduced with every production cycle – LODES is constrained by a “confidentiality budget”

One year is processed at a time because of the requirement of a previous year of data for the synthetic method.

Creating LODES Data: Confidentiality Protection

Two main processes are taking place to protect OnTheMap data: Noise Infusion – Applied to

workplace totals. Synthetic Data Methods – Applied

to residential locations.

For more information, see: http://lehd.ces.census.gov/led/datatools/doc/OTMSyntheticData%2005262009-jma.pdfhttp://lehd.ces.census.gov/led/datatools/doc/SyntheticDataDiagram%2006082009_JMA.pdf

Releasing LODES Data Public Data Release has two file types: OD

Connects a home block with a work block. Gives one count of jobs for each home-work block

pair and for each combination of year, job type, and segment variables.

RAC/WAC Provides totals on residence/workplace side only. Gives one total for each worker/job characteristic

for each combination of year, job type, and segment variables.

Releasing LODES DataJobs are classified by:

Whether the job is a worker’s primary/dominant job.

Whether the job is in the private sector. Whether the job is sourced from OPM*.

The six Job Types are: All Jobs, Primary Jobs, All Private Jobs, Private

Primary Jobs, All Federal Jobs*, and Federal Primary Jobs.*

* The Federal Job Types are only broken out in the raw LODES data. Only four Job Types are available in OnTheMap

Note: A job in LODES are defined as Beginning of Quarter Employment, which means the worker was employed by the same employer in both the current (2nd) and previous (1st) quarter.

Releasing LODES Data Labor Market Segment

Jobs are aggregated by 10 “segments” determined by a worker’s/firm’s characteristics.

The 10 segments are: All Workers Workers by Age (29 or younger; 30-54; 55 or older) Workers by Earnings ($1,250/mo or less; $1,251/mo to

$3,333/mo; greater than $3,333/mo) Workers by Firm’s Industry (Goods Producing; Trade,

Transportation, and Utilities; All Other Services)

OD Data is reported for each segment. Segments are similar to but not the same as

characteristics.

Releasing LODES Data:A note about the Industry segments

Industry Segments are build from NAICS Industry Sectors* (“2-digit”) Goods Producing:

NAICS 11, 21, 23, 33-33 Trade, Transportation, and Utilities:

NAICS 22, 42, 44-45, 48-49 All Other Services

NAICS 51, 52, 53, 54, 55, 56, 61, 62, 71, 72, 81, 92

* See http://www.census.gov/eos/www/naics/ for more info.

Releasing LODES Data

Characteristics: (available only in RACAC d 3 Age Characteristics – Comparable to Age

Segments 3 Earnings Characteristics – Comparable to

Earnings Segments 20 Industry Characteristics (NAICS Sectors) 6 Race Characteristics 2 Ethnicity Characteristics 4 Educational Attainment Characteristics 2 Sex Characteristics

Releasing LODES Data:Downloadable Data

Main vs. Aux To make it easier for users to limit the

amount of data they need to download, the files have been split by state of employment and state of residence.

Main: Jobs of workers who are employed in the state and reside in the state.

Aux: Jobs of workers who are employed in the state and reside out of state.

Releasing LODES Data:Downloadable Data

Releasing LODES Data:Downloadable Data

LODES data is available on the LED website: http://lehd.ces.census.gov/data/#lodes

LODES data is also available for bulk download through the following http site: http://lehd.ces.census.gov/data/lodes/

See OnTheMap Data Technical Document (http://lehd.ces.census.gov/data/lodes/LODES7/LODESTechDoc7.0.pdf) for more detail.

26

QWI Measures 32 indicators on:

Employment Counts of jobs (Individual) Hiring and Separation counts and rates (Individual) Job Creation and Destruction (Firm)

Earnings Average earnings for selected job histories Total earnings

Some indicators are special-purpose measures used in calculation of various rates

Files and applications organized by state

27

QWI Aggregation Levels: Firm

Firm-level characteristics: Based on national-level firm, sourced from Business

Dynamics Statistics (BDS) Firm Age (years)

0-1, 2-3, 4-5, 6-10, 11+ Firm Size (employees)

0-19, 20-49, 50-249, 250-499, 500+

Firm Age and Size available only for private ownership Reduced detail on geography/industry tabulations (3-

and 4- digit NAICS are only available for state-level totals)

28

QWI Aggregation Levels: Establishment

Establishment-level characteristics: Geography

State totals County, Metro, Workforce Investment Board (WIB) areas

Industry All industries NAICS Sectors, Sub-sectors (3-digit), Industry groups (4-digit)

Ownership All (Public + private) Private-only

All crossings of these characteristics reported (with a few exceptions for firm age and firm size)

29

QWI Aggregation Levels: Employee Age/Sex

Age (years) 14-18, 19-21, 22-24, 25-34, 35-44, 45-54, 55-64, 65-99

Sex Male, Female

We use the age categories specified in the Workforce Investment Act (WIA)

Data comes from a variety of sources (Decennial Census, surveys and administrative records)

30

QWI Aggregation Levels: Employee Education

Education categories: Less than high school High school or equivalent, no college Some college or Associate degree Bachelor’s degree or advanced degree Educational Attainment Not Available (age 24 or younger)

Valid only for individuals age 25 and up Reflects person’s maximum education level

Crossed by Sex in QWI tabulations Sourced from decennial census where available;

otherwise, imputed using multinomial logit model

31

QWI Aggregation Levels: Employee Race/Ethnicity

Tabulated according to categories defined by the Office of Management and Budget: Race

White alone African-American or Black alone Asian or Pacific alone Native Hawaiian or Other Pacific Islander alone American Indian or Alaska Native alone Two or More Races

Ethnicity Hispanic or Latino Not Hispanic or Latino

Race and Ethnicity are cross-tabulated in public QWI data Use data from Decennial Census where available; otherwise, impute using

Census file provided from Social Security Administration (SSA)

32

Concept: Employment History Jobs are linked across time Diagram illustration:

Diagram Legend:

Reference quarter t Earlier quarters (-), Later quarters (+)

RED: positive earning BLACK: zero earning COMBINED: earning in ANY ONE of the quarter GREY: quarters not referenced

Measure -5 -4 -3 -2 -1 t +1 +2 +3 +4 +5

33

Overview: Employment MeasuresMeasure -5 -4 -3 -2 -1 t +1 +2 +3 +4 +5

Count of Jobs (Flow Employment)EmpTotal

Beginning-of-quarter EmploymentEmp

End-of-quarter EmploymentEmpEnd

Full-quarter (Stable) EmploymentEmpS

Full-quarter (Stable) Employment, previous quarterEmpSpv

34

Overview: Earnings Measures for Employment Counts

Measure -5 -4 -3 -2 -1 t +1 +2 +3 +4 +5

Average Monthly Earnings for Beginning-of-quarter JobsEarnBeg $

Average Monthly Earnings for Full-quarter JobsEarnS $

Total Reported EarningsPayroll $

All income amounts reported for UI wages Mix of full-time and part-time jobs (not adjusted for

hours) Average earnings are based on quarterly wage record,

divided by 3 (monthly estimate)

35

Overview: Worker Flows Measures – Accessions

Measure -5 -4 -3 -2 -1 t +1 +2 +3 +4 +5

All Hires (Accessions)HirA

New HiresHirN

RecallsHirR

End-of-quarter HiresHirAEnd

Full-quarter HiresHirAS

New Hires into Full-quarter EmploymentHirNS

36

Overview: Worker Flows Measures – Separations

Measure -5 -4 -3 -2 -1 t +1 +2 +3 +4 +5

SeparationsSep

Beginning-of-quarter SeparationsSepBeg

Separations from Full-quarter EmploymentSepS

Separations from Full-quarter Employment, next quarterSepSnx

37

Overview: Earnings for Worker Flow Measures

Measure -5 -4 -3 -2 -1 t +1 +2 +3 +4 +5

Average Monthly Earnings for Hires to Full-Quarter EmploymentEarnHirAS $

Average Monthly Earnings for New Hires to Full-Quarter EmploymentEarnHirNS $

Average Monthly Earnings for Separations from Full-Quarter EmploymentEarnSepS $

All are based full-quarter counts, which are less biased by jobs that began or ended part-way through the quarter

Average earnings are based on quarterly wage record, divided by 3 (monthly estimate)

38

Hiring Rate End-of-Quarter hires divided by the average of Beginning-of-

Quarter and End-of-Quarter employment

HirAEndRt =

Bounded by 0% and 200% How can I use this?

“What fraction of the workforce are starting or returning to new jobs?”

Measure -5 -4 -3 -2 -1 t +1 +2 +3 +4 +5

HirAEndEmp

EmpEnd

39

Separation Rate Beginning-of-Quarter separations divided by the average

of Beginning-of-Quarter and End-of-Quarter employment

SepBegRt =

Bounded by 0% and 200% How can I use this?

“What fraction of the workforce are leaving their jobs?”Measure -5 -4 -3 -2 -1 t +1 +2 +3 +4 +5

EmpEmpEnd

SepBeg

40

Turnover Rate Measure of worker reallocation (“churn”) Measure of employment volatility

Incorporates both hires and separations

TurnOvrSt= If a firm of 100 individuals has 10 separations, and replaces

them with 10 hires => 10% turnover How can I use this?

“Which age group has the most employment volatility?” “Which industry has the highest employment churning?”

Measure -5 -4 -3 -2 -1 t +1 +2 +3 +4 +5

HirAS

SepSnx

EmpS

41

Overview: Firm-Based Measures:Flows, Creations, Destructions

Measure Description Category

FrmJbGn Job Creation (Gain)Creations

FrmJbGnS Full-quarter Job Creation

FrmJbLs Job Destruction (Loss)Destructions

FrmJbLsS Full-quarter Job Destruction

FrmJbC Net Job Flows

FlowsFrmJbCS Net Full-quarter Job Flows

HirAEndRepl Replacement Hires

HirAEndReplR Replacement Hire Rate

42

Measuring Firm-Level Worker Flows Firm job flows display dynamics at the

establishment level Job creation

Establishments that grow over the quarter Establishment births

Job destruction Establishments that shrink over the quarter Establishment deaths

Net Job Change = Job Creation – Job Destruction

43

Firm Job Flow Measures (1 of 2) Calculated at establishment level

Job Gain (FrmJbGn) Difference between End-of-quarter and Beginning-of-quarter employment

(EmpEnd - Emp) zero if negative

Job Loss (FrmJbLs) Difference between Beginning-of-quarter and End-of-quarter employment (Emp -

EmpEnd) zero if negative

Net Job Flows (FrmJbC) Difference between End-of-quarter and Beginning-of-quarter employment

(EmpEnd - Emp) Can be positive (net job creation) or negative (net job destruction)

-5 -4 -3 -2 -1 t +1 +2 +3 +4 +5

Emp

EmpEnd

44

Firm Job Flow Measures (2 of 2) Full-Quarter measures are defined similarly:

Full-Quarter Job Creation (FrmJbGnS) Difference between Full-Quarter employment (EmpS - EmpSpv)

zero if negative

Full-Quarter Job Destruction (FrmJbLsS) Difference between Full-Quarter employment (EmpSpv - EmpS)

zero if negative

Full-Quarter Net Job Flows (FrmJbCS) Difference between Full-Quarter employment (EmpS - EmpSpv)

Can be positive (net job growth) or negative (net job destruction)-5 -4 -3 -2 -1 t +1 +2 +3 +4 +5

EmpSpv

EmpS

45

Replacement Hiring Hiring and Job Creation are not necessarily equal:

Job Creation means more end-of-quarter employment than beginning-of-quarter employment at a firm But – there may be high levels of “churn” at firms, even without net

employment growth

To capture this, we define replacement hires: Replacement Hires (HirAEndRepl) are hires in excess of job

creation:HirAEndRepl = HirAEnd – FrmJbGn

The Replacement Hiring Rate (HirAEndReplR) is replacement hires as a percentage of average employment:

HirAEndReplR =

46

QWI Estimates:Source of Replacement Hires

Y200

6Q1

Y200

6Q2

Y200

6Q3

Y200

6Q4

Y200

7Q1

Y200

7Q2

Y200

7Q3

Y200

7Q4

Y200

8Q1

Y200

8Q2

Y200

8Q3

Y200

8Q4

Y200

9Q1

Y200

9Q2

Y200

9Q3

Y200

9Q4

Y201

0Q1

Y201

0Q2

Y201

0Q3

Y201

0Q4

Y201

1Q1

Y201

1Q2

Y201

1Q3

Y201

1Q4

Y201

2Q1

Y201

2Q2

0

2

4

6

8

10

12

14

16

Job creation End-of-quarter hires

Empl

oym

ent (

Mill

ions

)

Replacement Hires

Data: QWI pooled across all available states

51

Choosing Among LED Data Products

Data Product

Why Choose It? Potential Drawbacks

QWI You need employment, hires, separations, turnover, or earnings by detailed industry or person characteristics, quarterly time resolution, or a relatively short data lag

No geography below county; no residential information

LODES You need employment for detailed or customized geography, or you need the residential patterns of the workforce

Annual time resolution; less detailed firm/person characteristics; significant data lag (temporary)

J2J You need to understand transitions of workers among jobs

Data product still under development*

*But feedback is welcome.

52

Choosing Data 1When should I be interested in using LED data compared to other available statistics?

Suppose I’m primarily interested in Employment

Do I need the latest national estimate available?

• Current Employment Statistics (CES) • Employment by industry - ‘the payroll survey’

• Current Population Survey (CPS)• Employment status and demographics - ‘the

household survey’

Some sub-state geographies are available concurrently through Local Area Unemployment Statistics (LAUS)

?

53

But suppose I need either sub-national employment data or statistics by detailed industry:

Quarterly Census of Employment and Wages (QCEW)• Employment by detailed industry, sub-state geography

and better employment coverage (6-month lag)Quarterly Workforce Statistics (QWI)

• Employment by detailed industry, sub-state geography, and worker demographics (age, sex, education, race) and fewer cell suppressions than the QCEW (9-month lag)

American Community Survey (ACS)• Employment status by more sub-state geographies than

CPS/LAUS (9-month lag)LODES/OnTheMap

• Employment at the block-level (>1 year lag)County Business Patterns (CBP)

• Employment at the zipcode-level (>1 year lag)

Choosing Data 2

?

54

Suppose I’m primarily interested in Hires/Separations/Turnover

Do I need the most current national data (1 month lag) or do I want to differentiate between quits and layoffs?• Job Openings and Labor Turnover Survey

(JOLTS)

Do I need sub-national data (state/county), data by worker demographics, or for detailed industries?• Quarterly Workforce Statistics (QWI)

Choosing Data 3

55

Suppose I’m primarily interested in Wages

State and Regional Wage Information by Occupation?• Occupational Employment Statistics (OES)

Wages by Detailed Industry and Geography? • Quarterly Census of Employment and Wages (QCEW)

Wages by Detailed Industry and Geography and by Worker Demographics? Starting Wages for New Hires by Industry and Geography?• Quarterly Workforce Statistics (QWI)

Choosing Data 4

56

Choosing Data 5Suppose I’m primarily interested in Commuting

Transportation mode, time to work, work at home?• American Community Survey (ACS) Commuting Data

Commuting for Detailed/Custom Areas or Multiple Jobholders? • LEHD Origin-Destination Employment Statistics (LODES)

57

Enough Slides!

Let’s open up the LED Applications and get our hands on the data

58

Quarterly Workforce Indicators (QWI)

• Detailed workforce dynamics, by worker characteristics and firm characteristics

• Popular uses: • Local workforce

demographics• Local industry workforce

trends• Workforce turnover, job

creation and destruction

59

2002Q2

2003Q1

2003Q4

2004Q3

2005Q2

2006Q1

2006Q4

2007Q3

2008Q2

2009Q1

2009Q4

2010Q3

2011Q2

2012Q145%

46%

47%

48%

49%

50%

51%

52%

53%

Actual

Constant 2002 Industry Shares

Percentage of Female Workers at New Firms, 2002 - 2012

Note: "New Firms" are firms of age 0 or 1. "Actual" is the percentage of female workers at new firms observed in the data. "Constant 2002 Industry Shares" measures the hypothetical percentage of female workers at new firms, assuming that the distribution of new firms across industries remained constant at their 2002Q2 levels. Data is for privately-owned firms and excludes workers in the following states: AZ, AR, DC, MA, MS & NH. Source: U.S. Census Bureau, Center for Economic Studies, Quarterly Workforce Indicators, 2013 Q3 Release

• Can see workforce composition by detailed firm characteristics

• Such as what share of the workforce at startup firms is female?

Quarterly Workforce Indicators (QWI)

60

Job-to-Job Flows Types of questions that can be answered:

How did the growth and decline in construction jobs in the last decade impact the ability of low-wage workers to move to better jobs?

Where are North Dakota’s oil boom workers coming from?

Download data from http://lehd.ces.census.gov/data/j2j_beta.html

61

Job-to-Job flows

62

Job-to-Job flows

63

QWI: Combining More than One Indicator to Create New Insights

64

QWI: Combining More than One Indicator to Create New Insights

65

QWI: Combining More than One Indicator to Create New Insights

66

Public Data Tools

All tools are free and available 24/7. Live Demonstrations of

QWI Explorer LED Extraction Tool (QWI) OnTheMap OnTheMap for Emergency Management

67

Web Addresses for Tools

QWI Explorer http://qwiexplorer.ces.census.gov/

LED Extraction Tool http://ledextract.ces.census.gov/

OnTheMap http://onthemap.ces.census.gov/

OnTheMap for Emergency Management http://onthemap.ces.census.gov/em.html

68

What Questions Can LED Answer?

Which industries in my region are hiring older workers?

Younger workers?

Workers without a high school diploma?

What do these jobs pay?

69

Questions

Where do the workers employed downtown live?

What share of workers employed in my community also live there?

What share of workers with a short commute have a college degree?

70

Questions

Where did ND’s oil and gas workers come from (industry/geography)?

Where did MI’s auto workers go to (industry/geography)?

71

QWI Explorer 32 Quarterly Workforce

Indicators Flexible Pivot Table/Chart

interface Data on detailed interactions

between firms and workers include employment, employment change (individual and firm), and earnings

Analyze/report by worker demographics: age, earnings, race, ethnicity, educational attainment, and sex

Analyze/report by firm

characteristics: NAICS

classification (sector, 3, 4), firm

age, and firm size Quarterly data very current

(9-12 months old) 49 states available (plus DC,

MA coming soon)

72

OnTheMap: Where workers live, and where they work

• This map shows LODES data of where residents of Vancouver, Washington work

• Popular uses: • local economic

development• business site

selection• emergency planning

73

OnTheMap

Where do workers live? Where do residents work? What are the commuter flows

of a particular area? Analyze/report by worker

demographics: age, earnings, race, ethnicity, educational attainment, and sex

Analyze/report by firm characteristics: NAICS Sector, firm age, and firm size

2002-2011 annual data 49 states available (plus DC,

MA coming soon) User-selected areas Based on Census Blocks Disclosure protection Flexible Inputs/Outputs

Recognized by United Nations as a major U.S. statistical innovation

74

OnTheMap: Block-level employment detail

Hurricanes, Floods,Winter Storms

Disaster Areas

Wildfires

Demographic & Economic Data

• Comprehensive Reports

• Real-time Data Updates

• Easy-to-use & Interoperable

• Historical Event Archive

• Flexible Analyses & Visualizations

New Public Data Service for Emergency Preparedness & Response

OnTheMap for Emergency Management

76

76

Hurricane Sandy - October 25, 2012

77

Real World Examples

Some brief examples of from our users…

78

LODES: An Examination of Maryland Enterprise Zones

79

LODES: An Examination of Maryland Enterprise Zones

80

QWI: Education and Employment in Utah

81

QWI: A Comparison of I-95 and I-270 Corridors

82

Kansas City, MO – Earnings Tax

Civic Council of Greater Kansas City 1% Earnings Tax on gross compensation for all

those living or working in KCMO In 2010 & 2011, ballot challenges to the tax

were brought to voters LODES and QWI from LED helped the

community focus on “issues and outcomes” and showed “tax and benefits are shared with non-KCMO residents.”

83

Takeaways

The LED Partnership provides unique data products and tools at a relatively low cost

LED data products (QWI, LODES, J2J) can give insight into local and regional economies and labor markets

LED’s web tools provide free, 24/7 access to a basic analytical platform for the data

84

Thank You! Local Employment Dynamics

lehd.ces.census.gov Contact

Patrick.Hayward@census.gov Earlene.KP.Dowell@census.gov

Tools QWIExplorer.ces.census.gov LEDExtract.ces.census.gov OnTheMap.ces.census.gov OnTheMap.ces.census.gov/em.html