Accounting for the Discrepancy between Household and Establishment

Accounting for the Discrepancy between Household and Establishment Employment Presentation for NBER Labor Workshop April 16, 2004 Steven N. Braun Collaborators: Karen Dynan (CEA), Namita Kalyan (CEA), Tess Scharlemann (CEA) Special thanks to: Mary Bowler (BLS), Patricia Getz (BLS), Bruce Fallick (FRB), Matt Weinzierl (CEA), Signe Wetrogran (Census)

Supply-Side Identity for Real GDP

Could an error in the Census net immigration projection account for the employment discrepancy? If so, how big an error is implied? The CPS measures rates only. These rates must be multiplied by a population estimate to get employment and many other labor-market variables. The estimate of household employment is derived as: The 2000 decennial census provides our best estimate of population.

But population growth during the inter-census years is a forecast based on limited information—and could be a source of error—as it was during the 1990 s when population growth was underestimated (Nardone et all 2003). To solve for the population growth that would eliminate the 3 -year (3. 0 million) change in the household to payroll discrepancy, substitute the change in payroll employment (-2. 3 million) for the 0. 2 million increase in payroll-concept household employment. Further, because forecasting the natural rate of increase in the 16 and over population is fairly easy (how many 15 -year olds become 16? ), attribute all of the forecast error to mistakes in the net immigration assumption.

Notes on Effect of Employment-to-Employment Transfers on 3 -year Employment Discrepancy Fallick-Fleischman Estimates Employment-to-Employment Transitions Source: As derived from the CPS. Notes: See Bruce Fallick and Charles Fleischman (2004) Working-age population 2001 (thousands) 215, 311 Decline in monthly job changes (thousands) -968

Distribution of Paycheck Periods Decline in the Number of Job Changers with Two Payroll Entries Not Listed as Dual Job Holders <1>Probability job-to-job transfer will result in 2 payroll entries without being counted as a dual-job holder in the CPS. A person who works at 2 jobs during the CPS reference week will be counted as a dual job-holder.

Formation of Limited Liability Companies Note: Represented states comprise 56% of the total U. S. population. Source: Collected by Bill Hobbs from state government sources.

Military Reserve Mobilizations Source: available from Do. D website • The call-up of a reserve soldier (who is not replaced) decreases payroll employment, but has no effect on household employment (because the civilian non-institutional population is not decreased—although perhaps it should be). • If the reservist is replaced, payroll employment will be unchanged, but household employment will increase. • In either case, the discrepancy (between household and payroll) will increase.

Hopes for Correction from the 202 -CES Benchmark • 202 data through September 2003 indicates that the CES is tracking the 202 data well. • Possible pro-cyclical divergence during 2003: Q 4 and 2004: Q 1. • BLS birth-death model does not include cyclical effects. • It uses deaths to infer births. This works, but why?