The initial release of the 2018 BDS statistics in September 2020, (described here:  https://www.census.gov/programs-surveys/bds/news-updates/updates/2018-bds-release.html), contained an error in how the establishment entry rates and establishment exit rates were calculated.  No other BDS statistics were affected by this error, including the entry and exit establishment counts, as well as measures of job creation from establishment births and job destruction from establishment exits. The establishment entry and exit rate statistics have been corrected and new data tables with these revised statistics have been released, see:  https://www.census.gov/data/tables/time-series/econ/bds/bds-tables.html.

The error in the establishment entry and exit rates was due to the wrong denominator being used in the calculation. The correct denominator, consistent with the legacy tabulations and the methodological documentation, is the longitudinally consistent average count of employment active establishments in t and t-1. This denominator can be computed directly using each year’s establishment counts as:

estabs_denom_t = 0.5*((estabs_t + estabs_exit_t – estabs_entry_t)+estabs_t)

The term (estabs_t+estabs_exit_t-estabs_entry_t) is the longitudinally consistent measure of active establishments in t-1. The measure estabs_denom_t is analogous to the denom_t measure which is the longitudinally consistent average of employment in t and t-1. The corrected entry and exit rates are given by:

estabs_entry_rate_t=100*(estabs_entry_t/estabs_denom_t)

estabs_exit_rate_t = 100*(estabs_exit_t/estabs_denom_t)

Note that these computations do not rely on using lagged values of the estabs variable. To ensure the longitudinal consistency of the establishment denominator measure, we construct the longitudinally consistent establishments count in t-1 using the establishments count in t along with the counts of establishment entry and exit. If the longitudinally consistent measure of establishments count in t-1 is compared to the lagged value of the estabs variable there will be small differences. This is related to why the lagged emp variable is not exactly equal to the longitudinally consistent measure of employment in t-1 used in the denom variable. In both cases, this reflects that establishments may enter or exit the scope of the BDS tabulations between years.

For more information about how the establishment entry and exit rates are calculated, see the following FAQ:  https://www.census.gov/programs-surveys/bds/documentation/faq.html.