• How do I compute the establishment entry and exit rate using variables in the BDS tables?   

The establishment entry rate and establishment exit rates can be computed using establishment counts provided in each BDS table. Establishment entry (exit) rates are defined as the count of establishment entrants (exits) in year t divided by the average count of employment active establishments in year t and  year t-1. However, this computation cannot be done by simply taking estabs_entry in year t divided by the average estab count in year t and year t-1. Due to year-to-year scope changes in estabs, the estab count in year t is not longitudinally consistent with the estab count in year t-1. In order to ensure that the count of year t employment active establishments are longitudinally consistent with the count of employment active estabs in year t-1, we need to apply the same scope of establishments for the pair of years. For year t, the longitudinally consistent count of employment active establishments in year t-1 can be derived as:

estabs year t-1 = estabs year t + estabs_exit year t – estabs_entry year t

We can then compute the establishment entry rate as:

estabs_entry_rate year t= 100*(estabs_entry year t/(0.5*(estabs year t + estabs year t-1))

Establishment exit rate can be computed in a similar fashion as:

estabs_exit_rate year t = 100*(estabs_exit year t/(0.5*(estabs year t + estabs year t-1)))

• What’s new about the 2021 BDS data compared to prior releases?

The 2021 release includes a number of changes and improvements to the BDS data. These changes and improvements are described in the release note document linked here: 2021 BDS Release Note.

• What is the most current year of data available?

 Currently the BDS covers the years 1978 through 2021.

• How often is the BDS updated?

The BDS is updated once a year. The timing of the BDS release is dependent upon the availability of the source files from the Business Register and the CBP program. The final BR files are typically available in October, one year after the reference year. The CBP program then completes its edits of the BR data files and publishes in April—about 6 months later. Finally, the BDS uses the BR and CBP files and completes its processing about 5 months later. So BDS data are typically released in September—2 years after the reference year.          

• Adding net flows and current employment does not yield the following year's employment. Why?

The discrepancies result from processing decisions required when working with longitudinal administrative data (or in this case linked cross sections). A rolling window of establishment attributes is used to determine whether a case should be considered in scope for the BDS tabulations. Establishments can and do switch from being in scope to out of scope. For example, an establishment may switch from an in-scope industry to an out of scope industry such as 52592 Trusts, Estates, and Agency Accounts. In this case, the establishment would contribute to the employment stock in one year but not the next. Moreover, as continuing establishments enter and exit the scope of the BDS, their employment is not registered as job creation and destruction.

In addition to scope changes, establishments may also move between by-variable categories across years. For instance, a continuing establishment with no employment change may be classified as firm size “1 to 4” in one year and then “5 to 9” in the next. Since firm size is the average of firm size in the current and prior year, an establishment that had 4 employees in one year and 5 in the following two years would meet this criteria. When establishments switch between categories, the employment stocks in one cell will rise while the other declines without necessarily involving changes in employment flows (job creation or destruction).

Despite these nuances, the longitudinally consistent measure of employment in t-1 for the t-1 to t flows can be retrieved by the variable Denom (which is the average of employment in t-1 and t defined in a longitudinally consistent fashion at the establishment level), along with reported period t employment. It can also be retrieved by using the reported Net Job Creation level for t-1 and t along with reported employment in period t. Overall, then, reported employment in year t is relevant for the flows between t-1 and t (and so reported employment in t-1 is relevant for the flows between t-2 and t-1).

• What is the difference between metro and metropolitan statistical area (MSA)?

Metro is a designation applied at the county level, while an MSA is a grouping of one or more metro-designated counties. In addition, the statistics for metro areas pool all metro counties in the U.S.

• How are startups defined? How can I identify them in the data?

Startups are firms with the age of 0. No previous employment is associated with these firms and all its establishments are de novo establishments.

• How can I identify establishment entry in the data?

Establishment entry (“estabs_entry”) is one of the variables tabulated in the BDS.  It is defined as the count of establishments born within the cell during the last 12 months, where “born” is defined as going from zero March 12 employment in year t-1 to positive March 12 employment in year t.

• Why are there entrants with establishment age greater than zero?

Age 0 establishments are de novo establishments in the economy based on the first year of positive employment in the week that includes March 12. They are establishments classified under Establishment Age = 0.  Establishments can be entrants with age greater than zero in year t if, for example, they have positive March 12 employment in year t-3, have zero March 12 employment in both year t-2 and year t-1, and then have positive March 12 employment again in year t.

• What is a firm  death?

Firm death identifies events where all the establishments associated with a particular firm and the firm itself cease all operations. Note that firm legal entities that cease to exist because of merger and acquisition activity are not classified as firm exits in these data.

• What are "Left Censored" Firms?

These are firms born before 1977 and for which we do not know their true age.

• Can the number of left censored firms increase?

It is not a given that the number of left censored firms and the number of establishments owned by such firms in a given year will be lower than those in the previous years. This is not hard to see once we understand that firms are legal units that can be rearranged, merged, or split into separate legal entities. Put simply, a left censored firm can be split into two left censored firms. Similarly, the acquisition or creation of new establishments by a left censored firms can result in a higher number of establishments belonging to left censored firms. Merger and acquisition activity can in principle then increase the number of left censored firms and the number of establishments owned by such firms. Note that this is not true for statistics on the number of establishments using establishment age. That is, the number of left censored establishments using establishments age must decline over time, and this is true in the BDS data.

• Why do I get different numbers for the same statistic in different tables?

There are a number of possible reasons for discrepancies between statistics in different tables and specifically between tables at higher levels of aggregation (e.g. economy-wide) versus tables at lower levels of aggregation (e.g. state, industry, etc.).

1.  Noise infusion: Because noise is added at the cell level for certain variables, including: job creation, job creation births, job creation continuers, job destruction, job destruction deaths, job destruction continuers, and net job creation, this can cause discrepancies in these statistics across tables.

2.  Rounding can cause some small discrepancies across tables.

3.  For the variables ‘firms’ and ‘firmdeath_firms’, these statistics are often smaller in the economy-wide table than as summed across tables at the lower geographic and industrial aggregations. This is due to the fact that in the BDS, geography and industry are defined at the establishment level instead of the firm level. So firms with multiple establishments operating in different geographies or industries will show up in the total firm data for each geography or industry in which they operate at least one establishment. The same holds true for the variable ‘firmdeath_firms’. This discrepancy does not apply to the other statistics in the BDS, however, because these statistics all represent tabulations of establishment-level data.

Related to this, it should be noted that in contrast to firm geography and industry—which as noted are assigned at the establishment level—firm size and age are assigned at the firm level and therefore represent the size or age of the entire firm, not just the part that operates in a given geography or industry. A given firm, therefore, will generally only show up in the total firms data in one age or size category in a given table. There are some exceptions to this rule for the firm size and initial firm size categories, as noted in the cases outlined in #4 below.

4.  For the variables ‘firms’ and ‘firmdeath_firms’, there can be discrepancies in these statistics between the economy-wide table versus the firm size and initial firm size tables. This difference is due to the fact that establishments are assigned a firm size based on the average size of their associated firm in (t) and (t-1). Initial firm size is based upon the size of the establishment's associated firm in (t-1) or, in the case of entrants, their size in (t). Importantly, an establishment's associated firm may change between periods. This means that multi-unit firms, due to merger and acquisition activity, may have establishments with different combinations of (t) and (t-1) firm identifiers and thus may be assigned to different firm size and initial firm size categories. For example, suppose a large multi-unit firm acquires a small continuing single-unit firm. The acquired single-unit establishment's initial firm size in (t) will be assigned based upon the size of that single-unit in (t-1) whereas the multi-unit's other establishments will be assigned an initial firm size based on the (t-1) size of the multi-unit firm.

5.  The fact that cells in tables at lower levels of aggregation are sometimes suppressed with (D) or (S) obviously means discrepancies can exist between these statistics at lower versus higher levels of aggregation.

 More Questions?

   » Complete list of answers at our Question & Answer Center