reconciling model employment across datasets

Comments

4 comments

  • Avatar
    IMPLAN Support
    Thank you for your post, and we apologize on the delay in the response. We feel as if we could better address your response if we had a better understanding of your concerns. To that end could you provide us some information about the variances that you are actually seeing that are concerning you? We understand and appreciate that you may not want to put exact numbers up, but even if you could provide us some information on what year of the IMPLAN data set you are working with, confirm or correct our assumption that you are working with Davidson County, and let us know what variances you are seeing between the IMPLAN numbers and those sets you are comparing them to, such as: IMPLAN values are 20% higher than CEW and 10% lower than CBP. This would give us better framework to address your specific question beyond what the document you posted describes. If you are seeing issues for both Employment Compensation and Employment, data on both, or clarification on which one is causing the concern, would also be helpful. Additionally, we do not know exactly how EMSI creates their numbers so we would not really be able to provide a good comparison of their data methodology with our own. BEA REA is the source, as you note, where we derive our Proprietor Income and Employment data and thus typically creates the largest adjustment. Thanks!
  • Avatar
    ccotton
    Happy to provide more info. Thanks for exploring the variance. Overall, the Implan model employment value is substantially higher (1.5 – 2x) than other data sources. Model year: 2014 Geography: Tennessee (95 counties) Implan industry code 434 employment is slightly less than 28,000 CBP 2013 NAICS 5222 employment: 13,190 Nonemployer 2013 NAICS 5222 all establishments: 311 CEW 2014 NAICS 5222 private employment: 16,521 (Gov’t emp 120) BEA/REA FT / PT wage and salary employment 522 (3 digit): 52,725 BEA/REA Total FT / PT employment 522: 56,634 Overall, the BEA/REA 522 figures are very similar to CBP/NES. However, NAICS 5221 (35,967) combined with 5223 (6,381) make up roughly 75% of 522 so what factors are causing the proprietor adjustment to favor 5222? I understand you may not be able to speak to EMSI data comparisons due to methodological issues. My understanding is that they use similar sources (QCEW, BEA, etc. plus census/ACS) to calculate self-employed estimates. For the purposes of having another data point in the conversation, the estimate for 5222 including non-QCEW and self-employed is just over 17,000. Please let me know if you need any additional info. I really appreciate you taking a look into this. Thank you!
  • Avatar
    IMPLAN Support
    Hello Christopher, The primary difference in the job count derives from the fact that Sector 434 includes both Naics 5222 and Naics 5223 the latter covering 5,176 jobs. The 2014 Tennesee CEW W&S Employment data for these two Naics codes are disclosed: 16,517 + 5,176 = 21,693. The 2013 TN REA Proprietor Employment data for the parent sector (Credit intermediation and related activities) to which IMPLAN codes 433 and 434 belong is also disclosed: 5,245. This figure gets split between IMPLAN sectors 433 and 434 based on the proprietors/establishment ratios described in #2, resulting in 4,892 proprietors in sector 434. Controlling all states to the U.S. increases this trivially to 4,895. Finally, the redefinitions process further increases sector 434’s employment by about 500 (coming from sector 396). Here are a couple of additional notes about data sources as well. 1. The CBP W&S Employment data are lagged a year so we only use it help disclose the current CEW data, which are current and also W&S Employment only. 2. The Census non-employer data are also lagged and are at the state level only. We use this data source together with the CBP data on Organizations with Employees by Ownership (also lagged and at the state level) to get a state-level average proprietors/establishment count by Naics code. These proprietor/establishment ratios are used to distribute the REA proprietor employment amongst the IMPLAN sectors that belong to that REA sector. Note, however, that the REA data are also lagged a year at the time we use it - so you will want to look at the 2013 REA figure when comparing to IMPLAN. Hopefully this helps
  • Avatar
    ccotton
    Very helpful. Thank you!

Please sign in to leave a comment.