Annual measures of gross job gains and gross job losses: as a complement to the quarterly gross job flow statistics, annual gross job gains and losses statistics reveal the tremendous amount of churning that underlies the net growth of employment

Monthly Labor Review, Nov, 2004 by Joshua C. Pinkston, James R. Spletzer

As part of the existing process of linking establishments across consecutive quarters in the Business Employment Dynamics program, BLS and the States identify what are termed breakouts and consolidations. The term "breakout" refers to a single establishment splitting into multiple establishments, and the term "consolidation" refers to multiple establishments merging into a single establishment. Breakouts and consolidations may be actual economic events representing business expansions and contractions, or merely administrative reporting changes due to how an employer with multiple establishments within a State reports its data. Although BLS and the States continuously work with employers to obtain data at the establishment level, some employers with multiple establishments within a State report their total employment and wages in a consolidated manner. Occasionally, an employer reporting consolidated data will disaggregate its data to the worksite level (or, much less frequently, vice-versa).

Establishments involved in breakouts and consolidations need to be treated with care when constructing gross job gains and losses statistics. For example, an employer with multiple establishments in the State that disaggregates its data from a statewide level to a worksite reporting level would initially appear in the microdata to be a closing of an existing large establishment and the opening of several new small establishments. The record linkage system used in the Business Employment Dynamics program strives to identify the relationships between the establishments that are involved in all one-to-many breakouts and many-to-one consolidations. These establishments can then be treated as continuous, rather than as openings and closings, when constructing the quarterly gross job gains and losses statistics. (3)

Breakouts and consolidations cause additional difficulties when users attempt to create annual gross job gains and losses statistics. For example, if one wanted to accurately track establishments from March of one year to March of the following year, information on breakouts and consolidations from all quarters within the year needs to be taken into account in order to understand business survival and thus avoid spuriously defining openings and closings.

The annual gross job gains and losses statistics reported in this article are based upon an algorithm that takes into account information on breakouts and consolidations from all quarters within the year. Previous research shows that an algorithm that uses all information within the year is preferable to a more naive approach which takes two quarters of microdata that are 1 year apart and links establishments without accounting for breakouts and consolidations that occur within the year. Such a naive approach, relative to the algorithm used here, increases the annual gross job gains and losses statistics by roughly 7 percent to 9 percent. (4)

This article uses data from the first quarter of 1998 through the first quarter of 2002. The quarterly statistics that we present replicate the official (seasonally unadjusted) statistics from the BLS Business Employment Dynamics program. (5) Employment is defined as the number of workers covered by unemployment insurance and earning wages during the pay period that includes the 12th of the month. The gross job gains and gross job loss statistics use reported employment data in the third month of the quarter as the measure of the establishment's quarterly employment. Thus, employment growth for the second quarter refers to employment growth from March to June. To be consistent with much of the gross job flows literature, many of the annual statistics that this article presents measure employment growth from March of one year to March of the following year.