Validity and Value of I-O Models & IMPLAN

Input-Output Models & IMPLAN

Wassily Leontief won a Nobel prize in 1973 for his research on input-output (I-O) analysis.  Follow these links for more details on the history and assumptions of I-O models.  The IMPLAN I-O modeling system was originally developed within the U.S. Forest Service (USFS), and later expanded in partnership with the University of Minnesota, prior to becoming a privately-held company.  Click here for more details on IMPLAN’s history.  

IMPLAN has since expanded beyond single-region I-O modeling to multi-regional SAM modeling, taking into account household spending, commuting, and the trade of goods and services between regions across geographic levels.  In addition to the traditional economic data, IMPLAN models now also include data on environmental impacts (e.g., greenhouse gas emissions by industry), occupational impacts (e.g., occupations by industry, skills and education by occupation), area demographics, and more.  IMPLAN currently provides models for 80 countries and continues to expand geographic coverage. 

Both internationally and in the U.S., IMPLAN is used by private businesses, non-profit organizations, associations, academic institutions, and government agencies from the Federal to local levels.  IMPLAN’s customizability allows for the incorporation of expert knowledge into the system and analysis.  IMPLAN users are empowered with multiple training platforms, world-class educational resources and support, and assisted reporting.  

The Value IMPLAN Adds to Public Data

The Data

IMPLAN models incorporate data from numerous public data sources, which have varying levels of geographic coverage, levels of industry detail, non-disclosure practices, definitional frameworks, and release schedules.  Constructing a complete and consistent SAM data set therefore entails much more than aggregating data from the various sources; it requires in-depth knowledge of each raw data source and entails: 

• Providing estimates for non-disclosed (suppressed) values
• Providing estimates for non-census/non-survey years 
• Disaggregating data into finer geographic scales 
• Disaggregating data into finer industry detail 
• Projecting lagged raw data
• Making adjustments for various exclusions (e.g., certain types of employees)
• Estimating values for which there are no raw data at any geographic scale (e.g., domestic trade of goods and services, flows of commuters’ income) 

The following sub-sections highlight some of the specific ways IMPLAN adds value to public data for each of our data products. 

U.S. Data

Expanded Geographic Coverage

This article provides an overview of the data sources used to build IMPLAN’s U.S. models.  The lower the geographic scale, the fewer raw data sets available.  For example, the only raw data available to construct zip-code-level models is land area, population, number of households, and wage and salary employment by industry; IMPLAN must fill in all the rest – proprietor employment, labor income, industry output, profits, taxes, household spending, government spending, and more, using techniques honed over IMPLAN’s 50+ years.  

Estimation of Non-Disclosed Values

Public data sets that contain non-disclosures include the Bureau of Labor Statistic’s Census of Employment and Wages (BLS CEW), the Bureau of Economic Analysis’ Regional Economic Accounts (REA), the BLS’ Consumer Expenditure Survey (CES), the Census Bureau’s Annual Survey of Manufactures (ASM), the USDA’s National Agricultural Statistics Service (NASS) Census of Agriculture, among others. 

Adjustments for Exclusions, Definitional Differences, Misclassifications,  and Other Reasons

Making appropriate adjustments for exclusions, undercoverage, and definitional differences requires in-depth familiarity with data sources, attention to detail, and extensive quality control testing.  The BLS CEW data have several exclusions, some of are documented by the BLS and others of which have been discovered through comparisons to other data sources.   

The Census Bureau’s Retail Trade data exclude sales taxes and thus must be adjusted upward to obtain true industry output values for the retail sectors. 

Adjustments have also been made for incorrect NAICS code assignments (BLS CEW, Census Merchandise Trade), outdated county classifications (BEA REA), redacted data values not included in aggregate totals (NOAA commercial fish landings), and more.

Estimation of Domestic Trade Flows

There is no raw data source for county-to-county trade of goods and services on an origin-of-production basis.  IMPLAN’s county-to-county trade flow data are estimated by way of a double-constrained1 and calibrated2 gravity model, as summarized in this video and described in full detail this paper.  These trade flow data enable multi-region impact analysis. 

Estimation of Greenhouse Gas Emissions and Other Environmental Impacts

IMPLAN’s U.S. Environmental Data include some more up-to-date sources for environmental impact modeling than the main data source (the U.S. Environmentally Extended Input-Output (USEEIO) data).  For example, mineral extractions are controlled to more recent estimates of extractions from the U.S. Geological Survey (USGS). 

In IMPLAN’s regionally-specific Environmental data, the raw distributor data used to assign environmental flowables to regions are both more up-to-date than the raw data used by USEEIO, and also have non-disclosures accounted for such that data can be attributed to geographies more accurately. 

Estimation of Occupation-Level Impacts

IMPLAN’s Occupation Data are compiled from several raw data sources, taking the best features from each.  Occupation codes are harmonized such that they roll up to one another by digit – consistency that does not exist in raw source data.  Special Standard Occupational Classification (SOC) codes related to agricultural sectors are preserved in IMPLAN’s data while deprecated elsewhere.  IMPLAN also produces occupation codes for military occupations, which are unique to IMPLAN and do not exist in any raw data.  Knowledge, Skills and Abilities (KSAs) and Core Competencies data are also estimated by IMPLAN for several occupations for which the source O*NET data do not exist.  

Time Series Data

Because the data production processes can change somewhat from year to year, depending on the availability and quality of raw data sources and any possible  improvements that may be developed subsequent to quality control discoveries,  IMPLAN re-estimates its entire set of models, from 2001 to the current data year, every five years, using current best practices, revised raw data in some cases, and more current raw data in other cases, yielding a truer time series that can be used for statistical analysis, trend analysis, shift-share analysis, and more.  

Canada Data

IMPLAN's Canada models transform Statistics Canada's raw input-output tables into a ready-to-use economic impact framework, adding value in several ways. IMPLAN provides sub-provincial regional detail by using specialized and locally available data to disaggregate provincial numbers. This allows users to model impacts at a finer geographic scale than a province, or to develop a custom region. IMPLAN has also expanded the industry scheme beyond what is available from Statistics Canada into several hundred detailed sectors and incorporated detailed tax data, for estimation of local, provincial, and federal tax impacts. Finally, the IMPLAN Canada models extend beyond Statistics Canada's core accounts by incorporating Canadian environmental data (resource use, emissions, and other environmental flows by industry) and detailed occupational data (employment by occupation within each industry), enabling users to assess environmental and workforce impacts alongside standard economic effects, all packaged into a consistent, annually updated dataset that's ready to run in software rather than reconciled by hand from raw government tables.

International Data

IMPLAN transforms fragmented, inconsistent international statistics into a single, unified, and highly reliable engine for global economic impact analysis, using OECD Input-Output Tables as the backbone. IMPLAN combines the OECD I-O Tables with the OECD Inter-Country I-O database (enabling multi-regional trade analysis between countries), the OECD Trade in Employment database (allowing for estimation of employment impacts), and the OECD Global Revenue Statistics database (supporting estimation of tax impacts). IMPLAN ensures that every data element is available for each and every country in IMPLAN International, filling data gaps with International Labour Organization (ILO), World Bank, and national statistics data. The result is a more complete, analysis-ready international dataset than any single OECD source provides on its own.

Europe Data

The IMPLAN Europe data (currently in development) uses the FIGARO (Full International and Global Accounts for Research and Input-Output analysis) tables as the foundation and framework for its models. IMPLAN is collecting and collating numerous data sources, including Eurostat regional accounts, OECD, and national statistics offices, to produce sub-national Europe data at several different geographic levels. For European Union member countries, the sub-national geographical classification will be based on the nomenclature of territorial units for statistics (NUTS) system.  For the United Kingdom, the sub-national classification system will be based on the International Territorial Levels (ITL) system.  While some regional data (employment by industry, gross value added by industry, and gross domestic product) is publicly available at the most detailed geographic level, these data are subject to non-disclosures and are only available at high-level industry aggregates. 

The IMPLAN Modeling System

IMPLAN has long been trusted as a robust tool for users to estimate economic impacts within an I-O framework.  IMPLAN boasts an intuitive user interface that delivers logical direction for accessing extensive regional economic data, creating scalable and customizable impact analyses, and interpreting project results.  For decades, IMPLAN has been the tool that economists and researchers turn to when they want to measure the impact of an economic event on a region, adding value for both I-O experts and those just getting started.

IMPLAN Guides provide a starting point for setting up even the most complex of impact analyses.  They cover common use cases such as construction and operations, as well as mixed use development, forward linkages, and more.  These Guides provide step-by-step instructions with helpful education along the way so new users know not only how to set up these types of analyses but also why certain best practices for project configuration and result reporting are important. 

For the more seasoned economist, IMPLAN provides a vast amount of data for use in I-O analysis and beyond.  On a project level, IMPLAN has more detailed customization options than ever before with the addition of the Industry Impact Analysis (Detailed) Event and the expansion of the Customize Region Menu.  Additionally, powerful capabilities such as MRIO utilize interregional trade and commuting data to quantify the economic effects across many regions with the click of a button.  IMPLAN’s dedicated Support Team serve users as another valuable resource to help answer any and all questions related to the software and data. 

Verifying Impact Results

Because other changes to the economy occur in addition to and simultaneous to the one being analyzed, such that the analysis cannot be done in an “all else held constant” experimental environment, the estimates of indirect and induced effects could never be proved or disproved.  Nonetheless, there have been attempts to compare different modeling systems and to evaluate the accuracy of regional economic impact estimates.  The IMPLAN data receive constant scrutiny from both internal and external stakeholders, upholding IMPLAN’s reputation as the gold standard for I-O data. 

RELATED ARTICLES

Methodology of IMPLAN

Data Sources & Estimation Methods

Written July 28th, 2026