CEW Data Details

1. How does IMPLAN CEW differ from BLS CEW data?

Fully disclosed annual employment and income data is available at the U.S., state, and county level based on the Bureau of Labor Statistics Census of Employment and Wages (CEW) series formerly known as ES202. State employment services departments, as part of the Unemployment Insurance Program, collect the base data and pass it to the U.S. Department of Labor. 

All data elements in this series are disclosed. The non-disclosed elements have been adjusted through a procedure developed by IMPLAN. This data is provided at the full NAICS code level of detail (dependent upon the year of the data ordered). NAICS based data is available from 2001 forward (Prior to that, SIC based data is available for the years 1988 to 2000). The CEW dataset provides annual average wage and salary establishment counts, employment counts, and wage and salary workers data by county at the 6-digit NAICS code level. 

2. How is the CEW data provided?

Your purchased data will be emailed to you. The product includes two Excel spreadsheets.

• The actual regional data
• A spreadsheet that provides the NAICS descriptions for every 6-digit NAICS code
• Spreadsheets include both 2012 and 2017 NAICS codes
• Separate spreadsheets for privately-owned and government-owned establishments
• The data include establishment counts, wage and salary employment, and wage and salary income

3. How does CEW data differ from IMPLAN Data?

CEW data differs from IMPLAN data in a number of key ways. Here are the top 5.

CEW	IMPLAN
6-digit NAICS level detail	Agricultural and Services at 3-4 digit NAICS Manufacturing at 5-6 digit NAICS
Data includes only Employment, Wages, and Establishment Count	Data includes Output, Value Added, Labor Income, Employment (including both wage and salary workers and proprietors), Employment Compensation, Proprietor Income, Other Property Type Income and Taxes on Production & Imports Net Subsidies
Excel spreadsheet format	Analysis Model that includes Multipliers and tools for calculating impacts
Time Series 1998-current	IMPLAN data has experienced changes in Sectoring and data estimation methodologies which make time series estimates challenging. Data is available for 1996-2004, 2006-current.
Data include private and government information (at 3 levels Federal, State, and Local Governments).	Data includes private Industries, State & Local Government Education, State & Local Government Non-Education, Federal Government Defense, Federal Government Non-Defense, Capital Investment, Trade, 9 Household Income Categories.
Only establishments that pay Unemployment Insurance and federal civilian jobs covered by Unemployment Coverage for Federal Employees (UCFE) are captured.  The data set does not capture: self-employed persons railway employment, religious organizations, military, elected officials, shell fishing and fin fishing,  private education, or any other establishments that have their own social insurance program. Since most farm employment is self-employment, CEW data misses much of the farm data	IMPLAN data is controlled to BEA REA data sets and ultimately BEA US NIPA employment as these data sets attempt to capture all employment in the economy and thus allow us to provide a more complete picture of the economy. Proprietor employment includes: Non-employers Partnerships

4. How can the IMPLAN Employment be smaller than the reported BLS figures?

It is possible that IMPLAN's figure could be lower than the BLS', although usually the IMPLAN Employment is greater than the BLS CEW reported employment. IMPLAN data usually reports higher Employment values because we include an estimate for the number of proprietors in the region as well, or because more than one NAICS codes is incorporated into a Industry. However there are a few conditions under which the IMPLAN reported values may be less that BLS:

1. A number of Industries undergo redefintions of their Employment and Output values following the BEA redefinitions. For Industries where this occurs, it effectively redistributes the reported values within one Industry or NAICS subset and assigns it to another related Industry (e.g. a portion of hotel employment redefined to casinos and gaming).
2. The BLS table '0' combines private industry plus government activity whereas IMPLAN separates out these values. The following descriptors provide a breakdown of the tables.
   0 = Total Employment (government and private)
   1 = Federal
   2 = State
   3 = Local
   4 = International Government (Embassies, etc.)
   5 = Private
   8 = Total Government
   9 = Total Employment Excluding Federal Government

   The data we create will be for Ownership Codes 1,2,3, and 5.

5. Why might the CEW county numbers not sum to the CEW state total values?

The data is consistent within the county therefore subsectors add to their aggregates. However, disclosures for CEW data are only run within a county - there is no vertical checking to the state totals.

They are not controlled to state values for two reasons:

1. The existence of "county" 999 which is the BLS CEW dumping ground for employment that can not be located to a specific county - we leave county 999 out (2014 data and earlier).
2. The CBP data used to non-disclose CEW data can be highly variant from the reported CEW data, and we don't want to distribute those CBP "inconsistencies" to other counties.

6. Why isn't there correspondence between NAICS 23* and IMPLAN Construction Industries?

Construction Sectors are somewhat unique in that we create our construction Industries from Census descriptions rather than NAICS codes to assist our users, so that they do not need to construct a building from its component NAICS based parts and also because it is a Industry with high proprietors. Thus there is not a direct correspondence between the Industries for IMPLAN construction and the reported values by NAICS in CEW since the reported CEW figures are distributed to their respective IMPLAN construction Industries. Visit the article Construction: Building the Right Model for more details.

Updated January 16, 2020