• With the release of the 2007 BEA BM, IMPLAN decided to no longer tie itself so tightly to the BEA BM sectoring scheme, as this continued to decline in sector number.  Instead, we decided to keep virtually every BEA BM sector that had ever existed and only add new sectors as the BEA supplied data for them or we obtained data for them elsewhere.  However, it became apparent with the release of the 2012 BEA BM that not all BEA sectors could be persisted because in addition to being deprecated by the BEA, some sectors are also deprecated within the NAICS classification system, meaning that we no longer have CBP or CEW data for them; these are our most detailed data sources and without those data we would have no means with which to persist those sectors aside from using split ratios from old CBP or CEW data, ratios that would get very stale very quickly.  Thus, there are 3 cases in which sectors had to be combined in the new IMPLAN sectoring scheme.  Fortunately, there are more cases of new sectors (i.e., more sector detail) than cases of combined sectors (i.e., less sector detail).  Each of these cases is listed below.
    • Previous IMPLAN sectors 20 and 21 will comprise a single sector (20) in the new 546 sectoring scheme.  This is due to 2012 NAICS codes 211111 and 211112 being rearranged into NAICS codes 211120 and 211130, which do not have enough detail to keep these two sectors separate.
    • Previous IMPLAN sectors 26 and 27 will comprise a single sector (22) in the new 546 sectoring scheme.  This is due to 2012 NAICS codes 212231 and 212234 being combined into 2017 NAICS code 212230.
    • Previous IMPLAN sectors 287 and 289 will comprise a single sector (285) in the new 546 sectoring scheme.  This is due to 2012 NAICS codes 333911 and 333913 being combined into a single 2017 NAICS code (333914).
    • Previous sector 395 will be split into 10 sectors in the new 546 sectoring scheme: sectors 392-401.  There is sufficient 2017 NAICS code detail to allow for this.
      • Wholesale - Motor vehicle and motor vehicle parts and supplies
      • Wholesale - Professional and commercial equipment and supplies
      • Wholesale - Household appliances and electrical and electronic goods
      • Wholesale - Machinery, equipment, and supplies
      • Wholesale - Other durable goods merchant wholesalers
      • Wholesale - Drugs and druggists’ sundries
      • Wholesale - Grocery and related product wholesalers
      • Wholesale - Petroleum and petroleum products
      • Wholesale - Other nondurable goods merchant wholesalers
      • Wholesale - Wholesale electronic markets and agents and brokers
    • Previous sector 437 will be split into 2 sectors in the new 546 sectoring scheme: sectors 443 and 444.  There is sufficient 2017 NAICS code detail to allow for this.
      • Insurance carriers, except direct life
      • Insurance agencies, brokerages, and related activities
    • Previous sector 440 will be split into 2 sectors in the new 546 sectoring scheme: sectors 447 and 448.  There is sufficient 2017 NAICS code detail to allow for this.
      • Other real estate
      • Tenant-occupied housing
    • New State and Local Government institution[1]: Whereas there were 2 State and Local Government institutions in the old 536 sectoring scheme (State and Local Government Education and State and Local Government Non-Education), the new 546 sectoring scheme will have 3 State and Local Government institutions:  State and Local Government Education, State and Local Government Hospitals and Health Services, and State and Local Government Other Services.  As such, previous state government payroll sector 531 (Employment and payroll of state govt, non-education) will be split into 2 sectors (sectors 540 and 541) and previous local government payroll sector 533 (Employment and payroll of local govt, non-education) will be split into 2 sectors (sectors 543 and 544) in the new 546 sectoring scheme.  There is sufficient 2017 NAICS code detail to allow for this.1
      • * Employment and payroll of state govt, hospitals and health services
      • * Employment and payroll of state govt, other services
      • * Employment and payroll of local govt, hospitals and health services
      • * Employment and payroll of local govt, other services 
    • Cases of aggregation:
    • Cases of disaggregation:


Footnote 11 on page 7–7 of the 2006 “Concepts and Methods of the U.S. Input-Output Accounts” document states that the National Income and Product Accounts (NIPAs) value exports of goods and services at the value leaving the country, which is equivalent to purchasers’ prices (this is in contrast to the NIPA values for imports, which are expressed in producer prices).  Because input-output (I-O) models are based in producer prices, the two export control totals (one for goods, the other for services) must be converted to producer prices by shifting some of the NIPA control total for exports of goods to the NIPA control total for the exports of services, to account for the value of the transportation and wholesale services involved prior to exportation and thus included in the purchaser price of those exports. 

We base the amount of the shift on the latest BEA Benchmark I-O tables, which contain data on the exports of goods and services expressed in both purchaser and producer prices.  We had not previously made this adjustment and will be doing so beginning with the 2018 IMPLAN data, with the adjustment ratio being updated every 5 years with the release of new BEA Benchmark I-O tables.


For more background on this adjustment process which we apply to the BLS’ CEW data, please refer to our data release notes for the 2014 and 2016 data years

Upon investigation of a customer query, we determined that our method of applying state-level adjustment ratios to all counties within the state, while allowing us to use only fully-disclosed data for the calculation of the ratios, did not always yield reasonable estimates for some counties.  Thus, beginning with the 2019 IMPLAN data, we are incorporating two refinements to this adjustment process: 1) we now use ratios calculated from county-level data, regardless of disclosure code and 2) incorporated a more dynamic and less subjective method for capping very large adjustment ratios.

Depending on the disclosure codes for each county and sector, this change potentially affects employment and employee compensation (EC) estimates in the following IMPLAN sectors:

  1. Religious organizations
  2. Commercial fishing
  3. Junior colleges, colleges, universities, and professional schools
  4. Private households
  5. Support activities for agriculture and forestry


This data year marks the first year to incorporate the new 2017 Census of Agriculture. The Census of Agriculture is used to:

  • Estimate benchmark proprietor employment
    • Specifically, IMPLAN takes the BEA REA Farm proprietor employment for the year of the benchmark and distributes it among the 14 IMPLAN farm sectors using proprietor farm count per proprietor farm sector data from the Census of Agriculture. These benchmark-year (to make sure we don’t falsely lead people to think that these are from the Benchmark itself) proprietor employment estimates by farm sector are then combined with the BEA Benchmark output data by farm sector to calculate output-per-proprietor ratios by farm sector.
  • Estimate crop production for states not reported by ERS or NASS in their annual data sets
    • We supplement gaps in ERS and NASS state-level crop production data with estimates derived from the Census of Agriculture.
  • Distribute state level farm sector production to counties
    • State values are distributed to counties by using the ratio of county physical production to state physical production from the Census of Agriculture. For example, if County A has 10% of the state’s corn sales (or acres if sales is not available, and farms if acreages is not available), then it gets 10% of the state’s Annual Cash Receipts value for corn.


New non-citrus fruit state output values and county distributors:

  • In their annual data sets, the USDA’s Economic Research Service (ERS) and National Agricultural Statistics Service (NASS) have discontinued reporting for multiple non-citrus fruits (bananas, blackberries, boysenberries, figs, guavas, and pineapples). Therefore, we now use the 2017 Census of Agriculture’s values for the overall non-citrus fruit category (NAICS codes 111331, 111332, 111333, 111334, and 111339) for our state estimates, which has the following benefits:
    • Reduces the number of disclosures performed by us for those individual fruit categories (i.e., reduces the number of estimations made by us).
    • Avoids the need to project a continually growing and aging set of values for individual fruit commodities (i.e., avoids the reliance on a projection of the last reported value for a given specific fruit category that is no longer reported, a value which becomes more out of date with each passing year).
    • Simplifies our processes and reduces the risk of error.
  • These 2017 state values are projected to the data year based on ERS disclosed state-specific non-citrus fruit values (either one or multiple individual fruit categories, depending on data availability for that state and years), where available, else U.S. disclosed non-citrus fruit values. State to county distributors are derived using the Non-citrus totals (excluding berries) and Berry totals data from the 2017 Census of Agriculture.


We calculate this ratio for all IMPLAN farm sectors when a new BEA Benchmark is released; these ratios are used in our annual data process (see our farm data document for more details on how the ratios are used).  Prior to the current (2012) Benchmark, we simply used the ratio of Benchmark EC (before redefinitions, bridged to IMPLAN sectors) to benchmark-year CEW wage and salary employment (also bridged to IMPLAN sectors).  However, upon calculating the new ratios this year, it became apparent that the Benchmark has a different distribution of EC across the  farm sectors relative to that of CEW wage and salary income data, which would result in strikingly different EC per wage and salary income ratios across farm sectors, as well as negative supplements to wages for some farm sectors.  Thus, we now re-allocate CEW wage and salary employment to farm sectors according to the distribution implied by the Benchmark EC data. 

In summary:

  1. We first calculate ratios of wage and salary income to wage and salary employment for each farm sector from CEW. 
  2. We then convert those wage and salary income per wage and salary employment ratios to EC per wage and salary employment ratios using a ratio of Farm Total EC to Farm Total wage and salary income from the BEA’s REA data series.
  3. Next, we distribute that wage and salary employment according the Benchmark EC so that the distribution of wage and salary employment now matches the Benchmark rather than CEW, the latter of which lacks full coverage of non-proprietor farm employment.
  4. Finally, we use these sector-specific values to distribute the BEA REA Farm Total wage and salary employment among the 14 IMPLAN farm sectors, since the BEA REA data provide full coverage of farms but no distribution to specific farm sectors.


Prior to DY 2018, we were distributing final demand for the new construction sectors geographically using data points other than output.  However, in order to have equivalence between final demand and output for new construction sectors, we should be distributing final demand for these sectors based on their output; as of DY 2018 we are doing just that.


Beginning with the 2018 IMPLAN data, we now incorporate current, but more aggregate, BEA Output data as controls for our projections of the more-detailed but lagged BEA Output data that we’ve always used, thereby giving us greater confidence in those output values.


The BEA BM (and IMPLAN) define retail output as retail margin + sales taxes collected on gross sales.  However, the annual retail trade survey (ARTS) data on retail sales and margins do not include sales tax.  Thus, we have always made an adjustment to bump the ARTS values upward to account for sales tax.  However, it was discovered in DY 2017 that our adjustment process was not quite accurate, so in DY 2018 (when we needed to update the adjustment ratios anyway to incorporate the new BM), we corrected the calculation of said adjustment ratios.  This change also affects the calculation of the margins as a percentage of sales that enter into the structural matrix. 


The 2018 IMPLAN data are the first to incorporate the Census Bureau’s Preliminary 2017 CFS data on commodity shipment characteristics. The Census Bureau releases new CFS data every 5 years, first as a small preliminary data release and then later as a larger more-detailed data set.  IMPLAN incorporates the CFS data over two years:

  • In the first year (2018 IMPLAN data in this case), the preliminary data are used to revise estimates of average miles shipped for shippable commodities (which are used as a calibrator for the gravity model) and estimates of shipment-modes by industry (used to determine which combination of impedances to use for each commodity in the gravity model).
  • In the following year (2019 IMPLAN data in this case), IMPLAN incorporates the later-released detailed CFS data to further revise both sets of estimates.


[1] Due to limitations of the desktop version of the IMPLAN software, this change is only available in the latest online version of the IMPLAN application. 



BEA Benchmark & The New 546 Sectoring Scheme



BEA Input-Output Accounts

Comprehensive Update of Industry Accounts Now Available

Measuring the Nation's Economy: An Industry Perspective | A Primer on BEA's Industry Accounts

National Agricultural Statistics Service (NASS)

USDA Census of Agriculture

USDA Economic Research Service (ERS)



Written November 7, 2019