Job Flows A. Introduction 1. Many economic concepts, including employment, can be analyzed from a flow perspective 2. At a minimum, the level of employment depends on flows into and out of employment 3. Each of these flows may react differently to alternative types of economic conditions and environments 4. Some policies apply to specific types of flows a. economic incentives to entice a plant to locate or relocate to an area b. plant modernization tax credits 5. Analysis of flows is also important to understanding business cycle dynamics and longer-run processes like deindustrialization B. Concepts and measurement 1. Consider discrete time intervals and in particular the period from t−1 to t 2. Will describe flows for sectors but could examine specific firms or aggregate up to the entire employment figure 3. Job creation in sector s, Csta. sum of all employment gains experienced by all business units in that sector that expand or start up over the period b. ∑+∈∆=SeeststEMPC c. job creation can be further divided into components associated with business expansions and start-ups 4. Job destruction in sector s, Dst a. sum of all employment losses experienced by all business units in sector s that contract or shut down over the period b. ∑−∈∆−=SeeststEMPD c. destruction can be further divided into business contraction and shut-down components 5. Net employment change a. this is just the difference between job creation and job destruction b. this figure is widely reported 6. Job reallocation a. sum of all business unit employment gains and losses (sum of job creation and destruction) b. ststSeeststDCEMPR+=∆=∑∈ c. measure of turnover or volatility 7. Excess job reallocationa. sum of all business unit employment gains and losses beyond those needed to achieve net employment changes b. within-sector excess reallocation: ststNETR− c. aggregate excess reallocation can be written and decomposed ( )∑∑−+−=−sststststttNETRNETNETNETR 1) first term on RHS is between-sector shifts 2) second term on RHS is within-sector shifts 8. Definitions above are stated in terms of filled jobs; however, they could be redefined in terms of particular worker-job matches; when this definition is used, we refer to them as worker flows C. Measurement issues 1. Because each of the measures compares employment at two periods of time, the measures require longitudinal data 2. Business units a. the best business units seem to be establishments (physical locations where work takes place) rather than firms b. firms can encompass many establishments c. the composition of firms can change through mergers and split-offs; this would add anestablishment to one firm and subtract it from another leading to more apparent job flows d. because establishments have a particular physical location, they are simpler to track over time (though there are some difficulties) 3. Time periods a. typically depend on the availability of data b. can range from quarterly (e.g., ES-202 data), to annual (Survey of Manufacturers, LRD), to every five years (e.g., Census of Manufacturing) 4. Rates a. flows are usually described in terms of rates b. some problems associated with the choice of a base 1) if t−1 is used, it will be difficult or impossible to form rates for start-up firms; will also exacerbate regression-to-the-mean problems 2) if t is used, it will be difficult to form rates for closing firms c. hybrid measure, the average employment between time t−1 and t, is typically used 1) measure is symmetric around zero 2) bounded by the range −2 to 2 d. let Zst = (EMPs,t−1 + EMPs,t) / 2, then the job construction rate is stststZCc = ; other rates would be similarly definede. some of the awkwardness (arbitrariness) here arises because we are using discrete-time measures for something that is actually a continuous-time, discrete-unit process 5. U.S. data sets a. Longitudinal Research Database (LRD) 1) company-level database originally designed to investigate R&D issues 2) universe is manufacturing establishments with at least 5 employees 3) primary sources are Census of Manufacturers (universe of plants every five years) and Annual Survey of Manufacturers (universe for large plants, probability sample for smaller plants) 4) data form an unbalanced, replenished panel 5) special statistical procedures need to be used to address the introduction and deletion of plants from the database b. Business Employment Dynamics 1) source of data is the Unemployment Insurance records in the ES-202 program 2) quarterly employment data on all UI-covered establishments, except government employees, private households, and establishments without any employees 3) monthly statistics reported by BLSD. Empirical findings 1. Magnitude of job flows a. evidence from several different studies suggests that the annual rates of job creation and destruction in the U.S. are each roughly 10 percent; quarterly rates are closer to 6-7 percent b. net jobs in manufacturing establishments in the U.S. have generally been declining by a little over one percent per year, while net jobs in all private sector establishment have been growing c. reallocation is substantially larger than net job flows—that is, there is lots of excess reallocation d. data from the BED indicate that reallocation rates have been trending downward in recent years; decrease has occurred in all flows e. international evidence also indicates that reallocation is substantially larger than net job flows f. similar patterns appear within countries for specific industries, though magnitudes differ across industries 2. Persistence a. can consider two definitions of persistence 1) percentage of newly created jobs that remain filled for N periods 2) percentage of newly destroyed jobs that do not reappear through N periodsb. roughly 70 percent of new jobs in the U.S. last at least a year, and just over half last at least two years c. nearly 5/6s of newly destroyed jobs in the U.S. fail to reappear within a year, and roughly 3/4s are still empty two years later d. figures are very similar across developed countries 3. Distribution of flows a. substantial proportion of flows are concentrated at the extreme ends of the distribution—that is, are associated with births and deaths of plants b. most other flows are clumped near the middle of the distribution—that is, are