Input‑Output analysis is a macroeconomic technique that maps how industries in an economy supply inputs to — and buy outputs from — one another. It uses an “I‑O table” of interindustry transactions to quantify how a change in final demand for one sector (for example, more bridge construction) ripples through suppliers, their suppliers, workers, and ultimately consumer spending. The method was developed by Wassily Leontief (Nobel Prize, 1973) and is widely used for impact assessment, planning and policy evaluation.
Key takeaways
– I‑O analysis measures interindustry linkages and estimates direct, indirect and induced effects of changes in final demand.
– The core calculations rest on a technical‑coefficients matrix (A) and the Leontief inverse (I − A)−1.
– It is useful for quick, transparent impact estimates (output, employment, income) but assumes fixed production recipes and fixed prices.
– For effects that involve substitution, price changes or capacity limits, a Computable General Equilibrium (CGE) model or other dynamic approach is more appropriate.
Understanding input‑output analysis
– I‑O table: a matrix showing purchases by each industry from every other industry (rows and columns) plus final demand (households, government, exports) and value‑added (wages, profits, taxes).
– Technical coefficients (A): derived by dividing each column entry by that industry’s total output; Aij = input of industry i per unit of output of industry j.
– Leontief inverse: L = (I − A)−1. Multiplying L by a change in final demand (∆y) gives the total output change across all industries: ∆x = L · ∆y.
– Multipliers: summary numbers derived from L (output, employment, income multipliers) that convert a unit change in final demand into broader impacts.
3 types of economic impact
– Direct: the immediate change in output or jobs in the industry where the final demand shock occurs (e.g., construction jobs paid to build the bridge).
– Indirect: supply chain effects — extra orders to suppliers (steel, cement, machinery) and their downstream hiring and purchases.
– Induced: additional household spending resulting from wages and income earned by workers from the direct and indirect activities (e.g., workers buying groceries, housing, services).
Practical example (bridge project)
1. Estimate final demand: local government plans $100 million in bridge construction (final demand vector with $100m in “construction” sector).
2. Use the regional or national I‑O table to construct A and compute L = (I − A)−1.
3. Compute total output change: ∆x = L · ∆y. This yields sectoral output effects (construction, steel, transport, retail, etc.).
4. Convert output changes to jobs: multiply ∆x by employment coefficients (jobs per $ of output) for each sector.
5. Decompose impacts: the direct effect equals the construction sector’s initial $100m (and its direct jobs); indirect effects are the supply chain components from L excluding the direct column, and induced effects can be estimated by adding a household consumption link (via a Social Accounting Matrix or an induced consumption coefficient).
Result: an estimate of total economic activity and jobs created by the $100m bridge project (direct + indirect + induced).
Explain Like I’m 5
Imagine building a lego house. To make the house, you need blocks, windows, and glue. The people who make blocks need plastic, and the people who deliver glue need trucks. If you ask someone to build a house, many other people suddenly have work to do. I‑O analysis is a way to count how many extra people work when you build that house and where those people work.
What are the advantages of input‑output analysis?
– Captures detailed interindustry linkages; shows how one sector’s change affects others.
– Transparent and algebraically straightforward (A matrix and Leontief inverse).
– Uses official national/regional accounts and is reproducible.
– Fast and useful for many policy questions: infrastructure, stimulus, industry shocks, environmental footprinting (CO2, water).
– Produces easy‑to‑interpret multipliers for output, employment and income.
What are the limitations of input‑output analysis?
– Fixed technical coefficients: assumes no substitution between inputs when relative prices or technology change.
– Constant returns and fixed proportions: ignores economies of scale and capacity constraints.
– Static in time: it does not model price adjustments, trade responses or time dynamics.
– Aggregation bias: sectors are grouped; heterogeneous firms within sectors are treated as identical.
– Data vintage and regional accuracy: I‑O tables are published infrequently and regional tables may require location quotients or adjustments that introduce error.
– Induced effects require a Social Accounting Matrix (SAM) or additional assumptions; simple I‑O models may under‑ or overstate these.
– Risk of misinterpretation: multipliers are context‑specific and should not be applied indiscriminately.
Why is the input‑output model important?
– Policy analysis: estimate local or national economic impacts of public investments, taxes, trade shocks, disasters, or industry closures.
– Planning and business strategy: firms assess supply chain dependencies and regional effects of expansions or moves.
– Environmental accounting: widely used to trace sectoral resource use and pollution embodied in goods and services (e.g., carbon footprints).
– Quick first‑order estimates: it provides a transparent, data‑driven baseline before employing more complex dynamic models.
Step‑by‑step: How to conduct an I‑O analysis (practical workflow)
1. Define scope and objective
• Geographic boundary (national, state, county, metropolitan).
• Sectors to include and level of aggregation.
• Impact metrics: output, employment, value‑added, earnings, emissions.
2. Collect data
• Obtain the latest regional/national I‑O table (e.g., BEA for US, national statistical offices, UN, or commercial providers).
• Get employment and value‑added per‑dollar output coefficients; household consumption patterns if you’ll model induced effects.
3. Prepare and map sectors
• Reconcile your project or policy categories with the sectors in the I‑O table (concordance).
• Price/base year harmonization: ensure all data are in the same year dollars.
4. Build technical coefficients (A)
• For each producing industry j, compute Aij = transaction from i to j divided by total output of j.
5. Compute Leontief inverse
• L = (I − A)−1. Check that (I − A) is invertible and that results are economically reasonable.
6. Define final demand shock(s)
• Create ∆y vector representing the change in final demand by sector (e.g., +$100m in construction).
7. Compute impacts
• Output impacts: ∆x = L · ∆y.
• Employment impacts: multiply ∆x by sectoral employment‑per‑$ output coefficients.
• Value‑added/income impacts: multiply ∆x by value‑added or wage coefficients.
8. Estimate indirect vs induced
• For induced effects, use a SAM or add a household consumption row/column to the I‑O table and recompute multipliers, or apply a separate induced consumption multiplier (document assumptions).
9. Sensitivity and robustness checks
• Run alternative scenarios (different project sizes, local purchase rates, imports leakage).
• Test with different sector mappings and price years. Report uncertainty ranges.
10. Report and interpret
• Present direct, indirect and induced results, multipliers, assumptions and data sources.
• Discuss limitations and when a CGE/dynamic model might be preferable.
Tools and resources
– Public data: BEA (U.S.) I‑O accounts and RIMS II multipliers; Eurostat and national statistical agencies; UN I‑O tables.
– Software: spreadsheets are sufficient for small analyses; R (packages like io, chet), Python libraries, and commercial packages (IMPLAN, REMI) for regional studies.
– Academic references: Leontief’s foundational work and textbooks on input‑output economics for methodology.
Best practices and cautions
– Use the most recent, regionally appropriate table; adjust for local purchase coefficients if necessary.
– Be explicit about assumptions (imports, idle capacity, fixed prices).
– Avoid using multipliers out of context (different regions, different time frames).
– Where substitution, price effects, or long‑run adjustment matter, consider CGE models or dynamic extensions.
The bottom line
Input‑Output analysis is a practical, transparent way to quantify how an activity or policy change propagates through an economy via supply chains and spending. It’s especially valuable for first‑order impact estimates (output, jobs, income) and for highlighting interindustry dependencies. However, its simplifying assumptions — fixed technical coefficients, static prices and no substitution — mean results should be treated as indicative, not definitive. Carefully document data and assumptions, run sensitivity checks, and consider more sophisticated models when the question requires behavioral responses, price changes or dynamic adjustments.
Sources and suggested further reading
– Investopedia: “Input‑Output Analysis”
– Nobel Prize: Wassily Leontief biography and award — /
– U.S. Bureau of Economic Analysis: Input‑Output Accounts
– United Nations Statistics Division: Input‑Output tables and resources — /
– Run a small illustrative calculation (with hypothetical coefficients) showing the direct/indirect/induced decomposition for the $100m bridge example; or
– Point you to specific I‑O tables for a chosen country or region and help set up the A matrix and Leontief inverse. Which would you prefer?