Emsi uses the Bureau of Labor Statistics’ (BLS) Quarterly Census of Employment and Wages (QCEW) dataset as the basis for both industry and occupation job counts. Since QCEW is published in terms of industries rather than occupations, we transform industry jobs into occupation jobs by means of staffing patterns, which are based on a national OES staffing pattern. This national staffing pattern describes, at the national level, how industries are made up (“staffed”) by various occupations. Emsi regionalizes this staffing pattern to apply to more granular geographies, and applies regionalized staffing patterns to QCEW industry job counts to form occupation job counts. Click here to read more about how Emsi regionalizes staffing patterns.
A staffing pattern is a simple matrix that shows, by percentages, what occupations work in what industries. Here is a simplified staffing pattern for Hospitals:
Most staffing patterns have hundreds of rows, all of which combined total 100%.
The percentages in these staffing patterns can be applied to industry job counts to yield occupation job counts:
We multiply the total number of jobs in the industry by the percent represented by each occupation, yielding an occupation job count for each occupation in the industry. This process is repeated for each industry, then each occupation’s job counts from each industry are summed, resulting in the total occupation job count for that occupation in the region of study.
We can visualize the overall process for an imaginary region that has four industries and five occupations. From QCEW, we know job counts by industry for each of the four industries in the region:
Each industry has its own staffing pattern, showing how the industry’s jobs should be broken out among the five occupations. First, we find the staffing pattern for Industry 1, which shows how jobs are to be broken out to all five occupations:
This pattern shows that 75% of the jobs in Industry 1 go to Occupation 1, 10% of its jobs go to Occupation 2, and so on.
We then apply that breakout to Industry 1’s jobs, yielding Industry 1 jobs broken out into occupations:
We then do the same for Industry 2. Its staffing pattern shows that 67% of its jobs go to Occupation 3, and so on:
Applying that staffing pattern to Industry 2, we get jobs by occupation for Industry 2. Those jobs are added onto the jobs we already broke out from Industry 1:
The same process applies for Industry 3:
And Industry 4:
After the process is repeated for all industries in the region, all industry jobs have been converted to occupation jobs. The jobs for each occupation, assigned from each industry, are then summed, resulting in a total job count for each occupation:
This is Emsi’s process for creating occupation jobs using QCEW job counts by industry combined with OES staffing patterns. By using this process, Emsi produces total occupation job counts that exactly sum to total industry job counts for any given region.
One common question that occurs after reading the above is why Emsi doesn’t use job counts from the Bureau of Labor Statistics’ Occupational Employment Statistics (OES) dataset instead of QCEW. OES delivers job counts by occupation, and it would seem to make more sense to use that dataset. However, unless we use the same dataset for both industry and occupation job counts, we will show different total regional employment, depending on whether jobs are viewed from an industry or occupation perspective. By using one source for both, we maintain consistency in employment counts between industries and occupations.
So why does Emsi rely on QCEW job counts rather than OES job counts? The Bureau of Labor Statistics’ Occupational Employment Statistics (OES) dataset is the most robust source for occupation data in the United States. OES is a survey of approximately 1.2 million businesses, surveyed over the course of three years. Each year, the BLS publishes OES data that contains the responses given over the last three years. The BLS estimates that OES survey coverage is about 57%* of the employed workforce in the United States. OES data is published in groups of counties (MSAs or non-metropolitan areas).
By contrast, QCEW is based on unemployment insurance records that businesses are required to submit quarterly (and each quarter contains monthly employment counts for the quarter). The universe of businesses that submit unemployment insurance records covers 95% of the employed workforce in the United States. QCEW is published for every county in the United States. QCEW is superior to OES in frequency, coverage, and granularity; hence Emsi uses QCEW job counts rather than OES job counts as the basis for Emsi employment data.
* See BLS OES FAQ, Section A, Question 4