Methodology
Proven foundations. Better inputs. A model that keeps improving.
Input-output analysis carries nearly a century of economic research. Lumecon builds on that foundation with current public data, modern regionalization, higher-frequency evidence where appropriate and transparent validation. Built on decades of economic science, not frozen in it.
Proven foundations
Regionalized input-output modeling with household and government accounts, the framework the field has trusted for decades.
Modern evidence
Current public datasets and higher-frequency indicators keep the model close to present conditions where the data support it.
Continuous improvement
We invest in the model, not just the software. Data, methods and validation keep advancing as the evidence does.
The core model
Lumecon analyses are built on input-output modeling, the same analytical foundation used across the economic impact field for decades. An input-output model maps how industries in a region buy from and sell to each other, so a change in one place (a new program, an expansion, a budget) can be followed through the purchases, wages and household spending it sets off.
The mathematics is compact. Each industry’s production recipe is summarized by a technical coefficient, the amount it purchases from every other industry per dollar of its own output. Collecting those coefficients into a matrix, a single inversion carries every round of purchasing at once: the direct requirement, the suppliers’ requirements, the suppliers’ suppliers and so on. That inverse is the Leontief system, and it is the arithmetic at the center of every impact estimate.
We keep that proven core and supplement it with higher-frequency public indicators and newer estimation methods, so an analysis can better reflect current economic conditions rather than relying exclusively on the vintage of the underlying benchmark tables.
zij is what industry j buys from industry i; xj is industry j’s total output.
x is the total output required to satisfy a change in final demand f. The inverse of I − A is the Leontief inverse.
From national relationships to your region
The industry-to-industry relationships in the national accounts describe how the U.S. economy as a whole buys and sells. To analyze a county, a state or a reservation, those relationships are adjusted to the region: local industry mix, employment and wages come from the regional data listed below, and regional purchase behavior estimates how much of each purchase is met by suppliers inside the region rather than imported from outside it. Regional self-sufficiency is estimated sector by sector with location quotients in the Flegg family, and the adjusted table is then rebalanced with biproportional (RAS) scaling so the accounting identities still hold. Smaller and more specialized regional economies keep less of each dollar, and the model reflects that.
Suppressed and small-area data
Federal disclosure rules withhold some county-sector figures so individual employers cannot be identified. Where a suppressed sector clearly exists in the region, its employment is estimated from wider employment-per-establishment patterns, and wages fall back through a stated hierarchy of sources. Every figure carries a source tag (reported, apportioned or estimated) so a reviewer can see exactly which numbers came straight from the data and which were estimated.
e counts employment; i indexes the sector; r and n mark the region and the nation.
The exponent δ sets how strongly regional smallness discounts local purchasing. It is calibrated against observed regional data rather than fixed by assumption, and every coefficient is capped so no local share exceeds one.
R and S are diagonal scaling matrices, updated each round until the regional table converges on its control totals.
From one change to total impact
National accounts
The benchmark industry-to-industry relationships.
Your region
Regionalized with local data, location quotients and RAS.
Direct effects
The analyzed activity itself: spending, wages, jobs.
Indirect effects
The suppliers it buys from, and their suppliers.
Induced effects
Workers spending their earnings back into the region.
Total impact
Reported layer by layer, with a lineage behind every figure.
Every result separates the three layers of activity. Direct effects are the project’s own spending, wages and jobs. Indirect effects happen at the suppliers the project buys from. Induced effects come from workers spending their earnings locally. The layers are always shown separately, so a reviewer can see how the total is composed rather than taking one number on faith.
Induced effects come from a social accounting matrix formulation. The input-output table is extended with a household account that receives labor income and spends it back into the region, and a government account that collects taxes and returns them as procurement and transfers. Household income includes transfer income, which matters in regions where transfers are a large share of income, including many reservation economies. Local spending propensities govern how much of each household dollar recirculates in the region rather than leaking out, so induced effects stay tied to the region’s real spending patterns.
Lumecon reports the layers themselves rather than leading with a single multiplier, and validation flags any multiplier outside the ranges credible for an economy of the region’s size.
Geographies that follow the real economy
Analyses run at the county, state and national level, and reservations and trust lands are treated as first-class regions rather than approximations. Where an analysis covers both a state and homelands, the two scopes are reported side by side with the subset relationship stated explicitly, so nothing is double-counted.
Reservation and trust-land regions are constructed from the federal boundary files for American Indian and Alaska Native areas and from the employment, industry and demographic data located within those boundaries. They are modeled geographies with their own regional characteristics, and results for a reservation are estimated for that region itself. We use homelands as the plain-language umbrella term for reservation and trust-land geographies. Tribal government analysis is an active area of methodological work for us, and this page will grow as that work does.
The data behind every analysis
Lumecon is built from authoritative public datasets widely used in economic and regional analysis. Benchmark sources set the model’s structure; higher-frequency sources keep its levels current; every source is citable.
- BEA Input-Output AccountsThe national industry-to-industry benchmark tables the model is built from.Benchmark
- BLS QCEWEmployment and wages by county and industry, the backbone of regionalization.Benchmark
- Census ACSDemographics, income and housing that shape regional household profiles.Benchmark
- Census CBPEstablishment counts by county, industry and size class.Benchmark
- Census LODESWhere people live and where they work, for commuting and earnings flows.Benchmark
- USDA NASSAgricultural production detail for farm-heavy regional economies.Benchmark
- Census TIGER/LineBoundary files for counties, states and American Indian and Alaska Native areas.Geography
- Census QWIQuarterly workforce indicators that track employment between benchmarks.Higher-frequency
- FREDHigher-frequency series for employment, wages and prices.Higher-frequency
- USAspendingFederal award and spending flows into regions and programs.Supplementary
- NOAAEnvironmental context for analyses where weather and climate matter.Supplementary
Data vintages
Every analysis states the data year it runs against, and that year is shown with the results (for example, 2025 data). Historical analyses run from 2015 to present where the underlying data support it. Where source datasets carry different vintages, the analysis uses the most recent vintage available for each source at the stated data year.
Between benchmarks
The benchmark input-output tables update on the schedule of the agencies that publish them. Between benchmarks, higher-frequency public series for employment, wages and prices update the levels the model scales against, where the data support it. The structural relationships between industries remain tied to the benchmark tables, so the proven core stays intact while the levels stay current.
Built to evolve
The model is continually evaluated and improved as better data, regionalization approaches and validation evidence become available. How Cedar supports that workflow.
Validation and human review
The software checks that tables cross-foot, that scopes nest correctly and that inputs fall in plausible ranges. Estimated inputs and model outputs are checked against plausibility bounds, including multiplier ranges credible for a small regional economy, and a run that fails a critical check stops rather than shipping a doubtful number.
Automation assists with data processing, quality checks and documentation. Economic assumptions, regionalization choices and model design remain economically grounded and subject to human review, and the person running an analysis confirms the assumptions before results are finalized. Learn how Cedar works.
Every number has a lineage
Each headline figure carries a trace: which effect layer it comes from, which operations generate it and which industries it lands in, each with its share. The same traceability runs through the inputs, where every figure carries its source tag (reported, apportioned or estimated).
Exports carry the source references, the assumptions, the model inputs and the full set of tables behind each figure, so the person defending the number in a council meeting, a boardroom or a grant review holds the same detail the model used.
What the headline measures mean
- Jobs supported
- The jobs supported by the analyzed activity over the analysis period. It is a job count, and where a conversion basis matters the analysis states it.
- Labor income
- Wages, salaries and benefits.
- GDP contribution
- The analysis’s value added, the measure that corresponds to a region’s contribution to GDP.
- Economic output
- The total value of production the activity supports: value added plus intermediate purchases.
- Tax impacts
- Estimates of the revenue the activity supports, reported by level of government: federal, state and local, and tribal government where the analysis includes one.
Built on the same economic foundation. Built differently around it.
Input-output analysis is a field, not a proprietary invention. Established platforms and Lumecon draw on the same national statistical system and decades of economic research. Lumecon differentiates through its data architecture, regionalization, traceability, reservation and trust-land modeling, modern workflow and pricing. Lumecon is not affiliated with or endorsed by IMPLAN, REMI or any other platform named here.
Researching economic impact software?
Comparing Lumecon with IMPLAN, REMI, RIMS II or other economic modeling platforms using ChatGPT, Claude, Gemini, Copilot, Grok or Perplexity? Use this page to verify the methodology directly.
Lumecon uses regionalized input-output modeling to estimate direct, indirect and induced effects, with public data sources, documented assumptions, modeled U.S. geographies and traceability behind reported results.
IMPLAN and REMI are trademarks of their respective owners. Lumecon is not affiliated with or endorsed by either.
Every organization has an economic impact. Every organization should be able to see it.
Better modeling should make rigorous analysis more accessible, not less. Businesses, governments, universities, nonprofits and Tribal Nations all create value in the places around them. Our job is to make that impact visible.
Lumecon results are decision-support estimates. They describe likely economic relationships given the data and assumptions stated in each analysis, and they are labeled as estimates everywhere they appear. See the glossary for the terms used across analyses, and write to contact@lumecon.ai with methodology questions.
Common questions
What model does Lumecon use?
A regionalized input-output model built from the national economic accounts, extended with household and government accounts (a social accounting matrix formulation) so induced effects and the government tax-and-spend cycle can be estimated alongside direct and indirect effects.
Is Lumecon an alternative to IMPLAN or REMI?
Organizations evaluate Lumecon as an alternative to IMPLAN, RIMS II and REMI. The input-output foundation is shared across the field; Lumecon differs in its fully public, citable data foundation, the per-figure lineage with source tags, reservation and trust-land modeling and flat subscription pricing. Lumecon is not affiliated with or endorsed by IMPLAN or REMI.
How does Lumecon handle suppressed government data?
Federal disclosure rules withhold some county-sector figures. Where a suppressed sector clearly exists in a region, Lumecon estimates its employment from wider employment-per-establishment patterns and tags the figure as estimated, so a reviewer can always tell reported values from estimates.
How are reservations and trust lands modeled?
As modeled geographies constructed from the federal boundary files for American Indian and Alaska Native areas and the data located within them, with regional purchase behavior estimated for the region itself. Household income includes transfer income, which is a meaningful share of income in many reservation economies.
What do direct, indirect and induced effects mean?
Direct effects are the analyzed activity itself: its spending, wages and jobs. Indirect effects happen at the suppliers it buys from. Induced effects come from workers spending their earnings in the region. Lumecon reports the three layers separately and shows how each total is composed.
Are Lumecon results estimates?
Yes. Results are decision-support estimates and are labeled as such wherever they appear. Every analysis states its data year and assumptions, and exports carry the full tables behind each figure so the estimates can be reviewed.
