Job-to-Job Flows (J2J): New Public Use Data On Worker .

Job-to-Job Flows (J2J):New public use data on worker flows across jobsLEHD WorkshopMarch 2016Erika McEntarferLead EconomistLongitudinal Household Dynamics (LEHD) ProgramCenter for Economic StudiesU.S. Census BureauJoyce HahnAnalystAbt. Associates& LEHD ProgramU.S Census Bureau1

Job-to-Job Flows (J2J) fills an important data gap:In 2000, about ½ of all hires were workers moving fromone job to another. Most job vacancies are not for entry-level workersMost job moves are moves ‘up the job ladder’ ½ of wage growth for young workers is from job change(Topel & Ward, 1992). Procyclical worker reallocation from lower paying tohigher paying firms (Haltiwanger, Hyatt, &McEntarfer, 2015; Kahn & McEntarfer, 2014).Better understanding of worker moves across industriesand labor markets2

Anticipated J2J data users:Federal policy makers interested in the overall health ofthe labor market 70% of decline in hires in Great Recession was declinein job-to-job moves.State governors, economic development, and labormarket analysts concerned about losing workers to job opportunities inother states, more information about own in-migrants better targeting trade adjustment labor trainingAcademic and non-profit researchers interested in the reallocation of workers in response todemand shocks3

National Job-to-Job Flows:Steep decline in job change in last two recessions8%Hires from persistentnonemploymentSeparations to persistentnonemployment7%Share of employment6%5%4%Job-to-Job flows(separations and hiresfrom employment)3%2%J2J job-to-job hire rateJ2J job-to-job separation rateJ2J separations to persistent nonempJ2J hires from persistent nonempNote: Source: Job-to-Job Flows, national data. Shaded regions indicate NBER recession quarters. All data are seasonally adjusted.4

Within months of initial beta release, J2J appeared in the 2015Economic Report of the President:5

Overview of Presentation:Introduction to Job-to-Job Flows (J2J)Key J2J statistics and how they compare to otheravailable dataTaking the data for a drive: Where are North Dakota mining workers coming from?Where are Louisiana teachers going?Where did all the manufacturing workers go?Walkthrough of how to generate examples above6

Key J2J Files:Count and rate files: Hires and separations, by whether or not the hire/separation wasa job-to-job flow or an employment flow. Currently available by(more detailed tabs forthcoming in later releases): National and stateBy worker demographicsBy industry sector, firm age and sizeSeasonally adjusted and not seasonally adjusted dataOrigin-Destination (OD) files: For job-to-job flows only: characteristics of origin and destinationjobs. Currently available by (more detailed tabs forthcoming inlater releases): Origin State and Industry by Destination State and IndustryOrigin State to Destination State by worker demographicsOrigin State and (Age/Size) by Destination State and (Age/Size)7

Key J2J Statistics:Job-to-job hires/separations: Hires and separations from one job to another, with little or nononemployment between job spells EEHire/EESep: Hires and separations, job change occurs within thequarterAQHire/AQSep: Hires and separations, job change occurs across thequarterJ2JHire/J2JSep: Sum of EE and AQ, our preferred measure ofhires/separations associated with job changeHires/separations from/to persistent nonemployment: Hires and separations to/from longer nonemployment spells NEPersist/ENPersist: Hires and separations where worker is not employedat either end of the quarter prior to hire/after separationNEFullQ/ENFullQ: Subset of NEPersist/ENPersist, hires and separationswhere worker is not employed for the entire quarter prior to hire/afterseparation8

Conceptually, a flow chart (hires):No. (residual of MainJob Ends-MainSep)New Main Hire into Manufacturing firmDid worker hold adifferent main job at thestart of this quarter?Did worker leavethis job during thequarter?YesYesNo job held at start of quarterDid he hold anothermain job at the start ofthe previous quarter?A within-quarter jobchange (EEHire)YesAn adjacent-quarterjob change (AQHire)No job held at start of previous quarterHire from persistentnonemployment (NEHire)9

How do J2J compare to other related series?:10

Comparison to JOLTS: Layoffs8%J2J Separations to persistentnonemploymentShare of employment7%6%5%4%JOLTS Layoffs3%2%J2J separations to persistent nonempJOLTS layoff rateNote: Shaded regions indicate NBER recession quarters. All data are seasonally adjusted. These J2J tabulations do not include plannedadjustments to the J2J series to account for partially-missing geography early in the time series.11

Comparison to JOLTS: Quits8%Share of employment7%JOLTS Quits6%5%J2J Job-to-Job separations4%3%2%J2J job-to-job separationsJOLTS quit rateNote: Shaded regions indicate NBER recession quarters. All data are seasonally adjusted. These J2J tabulations do not include plannedadjustments to the J2J series to account for partially-missing geography early in the time series.12

J2J separations-to-employment vs. CPS employer-toemployer flows9%8%CPS job-to-job flows(Fallick & Fleischman)7%6%5%Job-to-Job separations4%3%2%J2J job-to-job separation rateCPS job-to-job rateNote: Shaded regions indicate NBER recession quarters. All data are seasonally adjusted. These J2J tabulations do not include plannedadjustments to the J2J series to account for partially-missing geography early in the time series.13

Taking the J2J data for a drive:Example 1: Where are North Dakota miningworkers coming from?14

Net migration of out-of-state workers into the NorthDakota mining sector: 2010-2014Source: J2J prototype origin-destination data. J2J data is not yet available for Massachusetts and Kansas , data for all other states ispresent. Net migration of out-of-state workers is hires into the North Dakota mining sector of workers who recently held a job in a15different state, minus flows of North Dakota mining workers to jobs in that state.

Net in-state inflows into the North Dakota miningsector: 2010-2014Source: J2J prototype origin-destination data. Net in-state inflows is hires into the North Dakota mining sector of workers who recently16held a job in a different industry in North Dakota, minus flows of North Dakota mining workers to jobs in that industry in North Dakota.

Example 2:Where are Louisiana’s teachers going?17

Joyce was in AmeriCorps in the Baton Rouge areaafter Katrina She worked in the local schools There were serious problems with teacher retention There were concerns that Katrina evacuees that were teachersweren’t coming back to Louisiana That they found jobs and stayed put So in this example, we look at the J2J data to answer thequestion Joyce had back in 2007 How many teachers are leaving Louisiana for other education jobs?

Louisiana education workers leaving to take educationjobs out-of-state: 2005.3-2014.2Source: J2J prototype origin-destination data. J2J data is not yet available for Massachusetts and Kansas , data for all other states is19present.

Louisiana education workers leaving to take noneducation jobs out-of-state: 2005.3-2014.2Source: J2J prototype origin-destination data. J2J data is not yet available for Massachusetts and Kansas , data for all other states is20present.

In-state separations from the Louisiana educationsector: 2005.3-2014.2Source: J2J prototype origin-destination data. In-state separations is flows of Louisiana education sector workers to jobs in a different21industry in Louisiana.

Example 3:Where did all the manufacturing workers go?22

There was a surprisingly swift decline in U.S.manufacturing employment between 2000-201335%22,00030%Employment (in 005%0%10,000Manufacturing Employment (CES)Manufacturing Primary Employment (J2J)Share Manufacturing Employment

J2J: decompose employment decline into flows to otherindustries vs. flows to long nonemployment spellsChange in manufacturing employment2.00%1.00%Net employment gain from workers moving fromother industries to manufacturingNet employmentchange, manufacturing0.00%-1.00%-2.00%Net employment gain/loss fromcross-industry job-to-job moves (natl)Net employment gain/loss fromworkers moving in/out ofemployment (natl)Net employment change-3.00%-4.00%Net employment decline, workerseparations to long nonemploymentspells-5.00%24

Separation rates from manufacturing to otherindustries2.0%1.8%Separation Rate1.6%1.4%Blue: Separation rate of manufacturing workers tolow-wage services1.2%1.0%0.8%0.6%0.4%Red: Separation rate ofmanufacturing workersto construction jobs0.2%0.0%Separation Rate: Manufacturing Workers to Retail, Food Service, and Hospitality JobsSeparation Rate: Manufacturing Workers to Mining, Utilities, Transportation, and Agriculture JobsSeparation Rate: Manufacturing Workers to Jobs in Other IndustriesSeparation Rate: Manufacturing Workers to Construction Jobs

What happened to downsized manufacturing workerswho experienced longer nonemployment spells?:J2J OD currently available only for workers with lessthan 4-6 months of nonemployment between job spellsWe hope to be able to release OD data for workers withlonger nonemployment spells in later releases.For now, I can tell you a little about what happened tothem: For the 2000-2003 separators: 35% recalled to previous employer or found anothermanufacturing job 45% found jobs in other industries ½ of these after a nonemployment spell of over a year, mostlyappear to become general laborers and truck drivers 20% have no subsequent UI-covered employment.26

How to do the examples shown here: awalkthrough of how to use the beta dataA J2J data application is coming!But until it’s here things are a bit more basic27

To start – how to find the beta J2J data:28

Example 1:How to get the data to make this map29

First pull the industry sector level dataSelect LouisianaSelect the ‘OD’ fileSelect ‘CVS’ formatGrab the ‘industry sector’ file30

Keep geography, industry, year, quarter, geography orig,industry orig, EE, and AQHire. Delete all other columns.31

Then filter geography orig to obtain the origin state of interest.32

Then filter industry orig to obtain the origin industry of interest.Repeat with industry for the destination industry of interest.33

Then filter years to obtain the period of interest. (In the interest oftime, let’s choose only one quarter: 2005.1)34

Then sum EE and AQHire to get all Outflows for every state by yearand quarter.35

Example 2:How to make this graph8%Hires from persistentnonemploymentSeparations to persistentnonemployment7%Share of employment6%5%4%Job-to-Job flows(separations and hiresfrom employment)3%2%J2J job-to-job hire rateJ2J job-to-job separation rateJ2J separations to persistent nonempJ2J hires from persistent nonempNote: Source: Job-to-Job Flows, national data. Shaded regions indicate NBER recession quarters. All data are seasonally adjusted.36

First access the dataWe’re using nationaldata this timeGrab the ‘rates’ fileGrab ‘all demographics, all industries, all firmsize/age, seasonally adjusted’ file37

Confirm grabbed the correct file for your analysis38

Hide everythingexcept the variablesyou want to graphRecall that Main Hires is the sum of:*J2JHire (hires of workers changingjobs)*NEPersist (hires of workers frompersistent nonemployment spells)39

Red: decline in hires andseparations due toemployment flowsBlue: Decline in hires andseparations due to jobchangeApproximately 70% of thefall in hiring in the GreatRecession was due to a fallin job change40

Example 3:How to make this graphChange in manufacturing employment2.00%1.00%Net employment gain from workers moving fromother industries to manufacturingNet employmentchange, manufacturing0.00%-1.00%-2.00%Net employment gain/loss fromcross-industry job-to-job moves (natl)Net employment gain/loss fromworkers moving in/out ofemployment (natl)Net employment change-3.00%-4.00%Net employment decline, workerseparations to long nonemploymentspells-5.00%41