Input-Output Models — Input-output (Leontief) models and balance equations - Finite Mathematics

SECTION 1

Historical Context & Motivation

Understanding how industries depend on one another has been a central challenge in economics since the Industrial Revolution. When a steel mill increases its output, it needs more coal, more iron ore, more electricity—and the suppliers of those inputs in turn need more inputs themselves. The input-output model, pioneered by the economist Wassily Leontief, provided the first rigorous mathematical framework for capturing these inter-industry dependencies in a single system of linear equations. Leontief's insight was deceptively simple: represent every sector of an economy as both a producer and a consumer, then use matrix algebra to determine the total production each sector must achieve to satisfy both internal demand and external (final) demand.

1874 • Walrasian General Equilibrium

Léon Walras formulated the first general-equilibrium model of an economy, establishing the idea that all markets are interconnected. His system, however, lacked a practical computational method for large-scale analysis.

1936 • Leontief's First Input-Output Table

Wassily Leontief published a 44-sector input-output table for the U.S. economy, demonstrating that inter-industry flows could be organized into a single matrix and solved with linear algebra.

1949 • Harvard Economic Research Project

Leontief expanded his model to over 400 sectors using early digital computers, proving the scalability of the matrix approach and attracting international attention from government planners.

1973 • Nobel Prize in Economics

Leontief received the Nobel Memorial Prize in Economic Sciences for the development of input-output analysis, cementing the technique as a cornerstone of applied economics.

2000s • Modern Applications

Input-output models are now used in environmental impact studies, supply-chain analysis, and carbon-footprint accounting, extending well beyond their original macroeconomic scope.

The central question that Leontief's framework addresses is this: given a set of final demands for goods and services, how much must each sector of an economy produce in total so that every sector's input requirements are also met? Answering this question requires a balance equation for every sector, and the elegant matrix formulation that results is the subject of this lesson.

SECTION 2

Core Principles & Definitions

Before diving into matrix equations, it is essential to understand the vocabulary and conceptual building blocks of input-output analysis. The model partitions an economy into n sectors (or industries). Each sector produces a single commodity and, in the process, consumes commodities produced by itself and by other sectors. The portion of each sector's output that is consumed internally by the economy's own industries is called intermediate demand, while the portion delivered to households, government, or export markets is called final (external) demand.

Technical Coefficient (aᵢⱼ)

The fraction of Sector j's total output that is consumed as input by Sector i. These coefficients are assumed fixed (constant returns to scale), forming the entries of the technology matrix A.

Technology (Consumption) Matrix A

An n × n matrix whose columns represent a sector's input recipe. Column j lists how much of each sector's output is needed per unit of Sector j's output. Each column sum must be strictly less than 1 for the economy to be productive.

Total Output Vector x

A column vector x = [x₁, x₂, …, xₙ]ᵀ where xⱼ is the total monetary output of Sector j. This is the unknown we solve for.

Final Demand Vector d

A column vector d = [d₁, d₂, …, dₙ]ᵀ representing the external demand for each sector's output—what consumers, government, and export markets require after all inter-industry needs are satisfied.

Leontief Inverse (I − A)⁻¹

The matrix (I − A)⁻¹ translates final demand into total production requirements. Its entries capture all direct and indirect multiplier effects rippling through the economy.

✦

KEY TAKEAWAY

Think of the technology matrix A as a recipe book for an entire economy. Each column is a recipe: to produce one dollar's worth of steel, you need a certain fraction of a dollar of coal, a fraction of electricity, and so on. The Leontief model asks: if customers order a specific menu of final goods (the demand vector d), how much of every ingredient must the whole kitchen produce—including the ingredients needed to make other ingredients? The answer is the total output vector x, and the master tool for computing it is the Leontief inverse (I − A)⁻¹.

SECTION 3

Visual Explanation — The Flow of Goods

A clear way to visualize an input-output economy is as a directed flow diagram. Each sector is a node; arrows between nodes represent the flow of goods or dollars from a producing sector to a consuming sector. The diagram below illustrates a simple two-sector economy (Manufacturing and Energy) with arrows showing intermediate flows and external final demand leaving the system.

Each sector's total output (x₁ or x₂) is split among inter-industry flows (aᵢⱼ · xⱼ), self-consumption loops, and final demand (d₁, d₂) that exits the economic system.

In the diagram, notice the self-referencing curved arrows at the top of each sector box. These represent self-consumption—for example, the energy sector uses some of its own electricity to run its power plants. The horizontal arrows between the two sectors capture inter-industry flows: Manufacturing ships goods to Energy (aiding its production process) and Energy supplies power to Manufacturing. The remaining output flows downward as final demand to consumers. The fundamental accounting identity, the balance equation, states that for each sector, total output equals intermediate consumption by all sectors plus final demand.

SECTION 4

Mathematical Framework

We now formalize the ideas from the flow diagram into matrix notation. Consider an economy with n sectors. Let xⱼ denote the total output (in dollars) of sector j, and let dⱼ denote the final demand for sector j's product. The technical coefficient aᵢⱼ is defined as the dollar amount of sector i's output required to produce one dollar of sector j's output. These coefficients populate the technology matrix A, an n × n matrix.

BALANCE EQUATION (SCALAR FORM)

xᵢ = aᵢ₁x₁ + aᵢ₂x₂ + ⋯ + aᵢₙxₙ + dᵢ for i = 1, 2, …, n

Each sector's total output xᵢ equals the sum of what all sectors consume of its product (∑ⱼ aᵢⱼxⱼ) plus the external final demand dᵢ. This is the fundamental balance condition.

BALANCE EQUATION (MATRIX FORM)

x = Ax + d

Here x is the n × 1 total output vector, A is the n × n technology matrix, and d is the n × 1 final demand vector. The product Ax represents total intermediate demand.

To solve for x, we rearrange the balance equation by isolating x on the left side. Subtracting Ax from both sides gives x − Ax = d, which factors as (I − A)x = d, where I is the n × n identity matrix. Provided that the matrix (I − A) is invertible, we obtain the closed-form solution.

LEONTIEF SOLUTION

x = (I − A)⁻¹ d

The matrix (I − A)⁻¹ is the Leontief inverse. It exists whenever A is a productive matrix—that is, when every column sum of A is strictly less than 1.

⚠

Why Column Sums Matter

If the j-th column of A sums to 1 or more, it means sector j consumes as much (or more) in inputs as it produces in output—leaving nothing for final demand. The productivity condition (all column sums < 1) guarantees that (I − A) is invertible and that x ≥ 0 whenever d ≥ 0, ensuring economically meaningful non-negative production levels.

SECTION 5

Anatomy of the Technology Matrix

To build intuition for the technology matrix A, let us examine its structure in a concrete three-sector economy consisting of Agriculture, Manufacturing, and Services. Each column of A is a production recipe: column j tells you how many cents of every sector's output are consumed to produce one dollar of sector j's output. The diagram below displays this matrix along with annotations explaining how to read its rows and columns.

The technology matrix for a three-sector economy. Each column is a production recipe; column sums below 1 confirm the economy is productive.

Several features of this matrix deserve attention. First, the diagonal entries (a₁₁ = 0.20, a₂₂ = 0.10, a₃₃ = 0.05) represent self-consumption—Agriculture uses some of its own crop as seed, Manufacturing uses some of its own steel in its machinery, and so forth. Second, the column sums (0.60, 0.45, 0.40) indicate that Agriculture is the most input-intensive sector: sixty cents of every dollar it produces is consumed as intermediate inputs. Finally, the remaining fraction of each column (1 − column sum) is the value-added share—payments to labor, capital, and profit—which must be positive for a viable economy.

ℹ

Open vs. Closed Models

The model presented here is an open Leontief model: final demand d is given exogenously. In a closed model, households are treated as another sector whose consumption depends on its income (i.e., labor payments). The closed model absorbs final demand into an enlarged matrix, creating a homogeneous system (I − A*)x = 0. We focus on the open model because it is the standard formulation in finite mathematics courses.

SECTION 6

Worked Example — Two-Sector Economy

Consider a simplified economy with two sectors: Goods (G) and Services (S). The technology matrix and external demand vector are given below. We wish to find the total production vector x = (I − A)⁻¹d.

GIVEN DATA

A = [ 0.1 0.2 ; 0.3 0.2 ], d = [ 24 ; 38 ]

A is 2 × 2; column 1 sums to 0.4, column 2 sums to 0.4 — both less than 1, so the economy is productive.

Finding Total Output x = (I − A)⁻¹d

1
Step 1 — Compute I − A

Subtract A from the 2 × 2 identity matrix: I − A = [ 1−0.1 0−0.2 ; 0−0.3 1−0.2 ] = [ 0.9 −0.2 ; −0.3 0.8 ].

I − A = [ 0.9 −0.2 ; −0.3 0.8 ]
2
Step 2 — Find the Determinant of (I − A)

For a 2 × 2 matrix [a b ; c d], det = ad − bc. Here: det(I − A) = (0.9)(0.8) − (−0.2)(−0.3) = 0.72 − 0.06 = 0.66. Since det ≠ 0, the inverse exists.

det(I − A) = 0.66
3
Step 3 — Compute (I − A)⁻¹

The inverse of a 2 × 2 matrix [a b ; c d] is (1/det) × [d −b ; −c a]. Thus: (I − A)⁻¹ = (1/0.66) × [ 0.8 0.2 ; 0.3 0.9 ]. Dividing each entry by 0.66:

(I − A)⁻¹ ≈ [ 1.2121 0.3030 ; 0.4545 1.3636 ]
4
Step 4 — Multiply (I − A)⁻¹ by d

x = (I − A)⁻¹d. Compute each component: x₁ = 1.2121 × 24 + 0.3030 × 38 = 29.09 + 11.51 = 40.61. x₂ = 0.4545 × 24 + 1.3636 × 38 = 10.91 + 51.82 = 62.73 (values in millions of dollars, rounded to two decimals).

x ≈ [ 40.61 ; 62.73 ] (millions $)
5
Step 5 — Verify the Solution

Check: Ax + d should equal x. Ax = [ 0.1(40.61)+0.2(62.73) ; 0.3(40.61)+0.2(62.73) ] = [ 4.06+12.55 ; 12.18+12.55 ] = [ 16.61 ; 24.73 ]. Adding d: [ 16.61+24 ; 24.73+38 ] = [ 40.61 ; 62.73 ] ✓. The balance equation is satisfied.

Verified: x = Ax + d ✓

Interpretation: to satisfy external demands of $24M in Goods and $38M in Services, the Goods sector must produce a total of approximately $40.61M and the Services sector $62.73M. The difference between total output and final demand ($16.61M and $24.73M, respectively) is consumed as intermediate inputs within the economy itself.

SECTION 7

Strengths and Limitations

Aspect	Strengths	Limitations
Analytical power	Captures all direct and indirect inter-industry effects in a single matrix operation; highly systematic and scalable to hundreds of sectors.	Assumes fixed technical coefficients (constant returns to scale); no substitution between inputs when relative prices change.
Data requirements	National statistical agencies regularly publish input-output tables, providing a ready-made empirical foundation.	Compiling accurate tables is expensive and time-consuming; data can be outdated by the time tables are published.
Policy use	Enables scenario analysis: 'If demand for automobiles rises by 10%, what happens to steel, rubber, and glass production?'	Cannot model capacity constraints, unemployment, or market disequilibrium; the model always yields a feasible equilibrium.
Mathematical elegance	Reduces a complex web of interdependencies to a single matrix inverse, making computation straightforward once A is known.	The inverse (I − A)⁻¹ may be expensive to compute for very large n; iterative methods are sometimes required.

Comparative analysis of the Leontief input-output framework

🔍

CONTEXTUAL INSIGHT

The Leontief model's greatest virtue—its linearity—is also its greatest limitation. In engineering terms, it is like a circuit-analysis model that assumes all resistors are ohmic (constant resistance): tremendously useful for a first-order design, but unable to capture nonlinear effects such as overheating or saturation. In economics, these 'nonlinearities' include price-driven input substitution, economies of scale, and supply-chain bottlenecks. Despite these caveats, input-output analysis remains the backbone of national accounts and environmental impact assessment worldwide.

SECTION 8

Connections to Advanced Theory

The Leontief model sits at the intersection of linear algebra, economics, and applied mathematics. Understanding it opens the door to several more sophisticated frameworks that extend or generalize its assumptions.

Leontief (This Lesson)	Advanced Extension
Fixed coefficients aᵢⱼ (no substitution)	Computable General Equilibrium (CGE) models allow price-dependent substitution via production functions.
Single-period, static model	Dynamic Leontief models introduce capital coefficients and time-varying investment, solving differential or difference equations.
Purely monetary flows (dollars)	Environmentally extended IO (EEIO) models append satellite accounts for CO₂ emissions, water use, and waste per dollar of output.
Domestic economy only	Multi-region IO (MRIO) models link national tables through trade matrices, tracking global supply chains.

From the basic Leontief model to advanced extensions

From a purely mathematical standpoint, the study of non-negative matrices (Perron–Frobenius theory) provides the rigorous conditions under which (I − A)⁻¹ exists and is non-negative. Specifically, if A is a non-negative matrix with spectral radius ρ(A) < 1, then I − A is invertible and (I − A)⁻¹ = I + A + A² + A³ + ⋯, a convergent Neumann series. The k-th term Aᵏ captures the k-th round of indirect requirements: A is direct inputs, A² is inputs-of-inputs, A³ is inputs-of-inputs-of-inputs, and so on. This geometric-series interpretation provides both computational insight and economic intuition about how demand multiplies through successive layers of the supply chain.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

Explain in your own words why the column sums of the technology matrix A must each be strictly less than 1 for an economy to be called 'productive.' What would it mean economically if a column sum equaled or exceeded 1?

Show answer

Each column sum represents the total dollar amount of intermediate inputs consumed per dollar of output in that sector. If a column sum equals 1, the sector uses up its entire output as inputs—nothing remains for final demand (consumers, government, exports). If the sum exceeds 1, the sector consumes more than it produces, implying negative value-added and an economically infeasible situation. The productivity condition (all column sums < 1) ensures that (I − A) is invertible with a non-negative inverse, guaranteeing that every non-negative final demand vector d yields a non-negative production vector x.

Question 2: BASIC CALCULATION

Given A = [ 0.2 0.3 ; 0.4 0.1 ] and d = [ 12 ; 18 ], compute I − A and its determinant. Is the economy productive?

Show answer

I − A = [ 0.8 −0.3 ; −0.4 0.9 ]. The determinant is (0.8)(0.9) − (−0.3)(−0.4) = 0.72 − 0.12 = 0.60. Since det ≠ 0, the inverse exists. The column sums of A are 0.6 and 0.4, both strictly less than 1, so the economy is productive. With det = 0.60 > 0 and a productive A, we can proceed to find x = (I − A)⁻¹d for any non-negative d.

Question 3: INTERMEDIATE

Using the data from Problem 2 (A = [ 0.2 0.3 ; 0.4 0.1 ], d = [ 12 ; 18 ]), find the total output vector x = (I − A)⁻¹d. Verify your answer by checking that x = Ax + d.

Show answer

(I − A)⁻¹ = (1/0.60) × [ 0.9 0.3 ; 0.4 0.8 ] = [ 1.5 0.5 ; 0.6667 1.3333 ]. Multiplying: x₁ = 1.5(12) + 0.5(18) = 18 + 9 = 27. x₂ = 0.6667(12) + 1.3333(18) = 8 + 24 = 32. So x = [ 27 ; 32 ]. Verification: Ax = [ 0.2(27)+0.3(32) ; 0.4(27)+0.1(32) ] = [ 5.4+9.6 ; 10.8+3.2 ] = [ 15 ; 14 ]. Ax + d = [ 15+12 ; 14+18 ] = [ 27 ; 32 ] = x ✓.

Question 4: APPLIED

A regional economy has three sectors: Mining (M), Energy (E), and Transport (T). The technology matrix is A = [ 0.10 0.20 0.05 ; 0.15 0.10 0.20 ; 0.05 0.15 0.10 ] and the final demand vector is d = [ 50 ; 40 ; 60 ] (in millions of dollars). Write out the system (I − A)x = d in full. Without computing the inverse, explain what economic information the entry in position (1,2) of (I − A)⁻¹ would represent.

Show answer

I − A = [ 0.90 −0.20 −0.05 ; −0.15 0.90 −0.20 ; −0.05 −0.15 0.90 ]. The system is: 0.90x₁ − 0.20x₂ − 0.05x₃ = 50; −0.15x₁ + 0.90x₂ − 0.20x₃ = 40; −0.05x₁ − 0.15x₂ + 0.90x₃ = 60. The (1,2) entry of (I − A)⁻¹ represents the total increase in Mining's output (both direct and through all indirect supply-chain rounds) required for every additional dollar of final demand directed at the Energy sector. It captures not only the direct mining inputs that Energy needs, but also the mining inputs needed by Transport to serve Energy, the mining inputs needed by Mining itself to serve Transport, and so on through all successive rounds of indirect requirements.

Question 5: CRITICAL THINKING

The Neumann series expansion states (I − A)⁻¹ = I + A + A² + A³ + ⋯ when ρ(A) < 1. Give an economic interpretation of the first three terms (I, A, and A²) of this series. Then argue why the series must converge if all column sums of A are less than 1.

Show answer

The identity matrix I captures direct final demand—each sector produces exactly what consumers order, with no inter-industry effects. The term Ad captures first-round indirect demand: the inputs each sector needs from other sectors to fulfil the original order. A²d captures second-round effects—the inputs needed to produce those first-round inputs. Each successive term Aᵏd represents the k-th round of supply-chain ripple effects. For convergence, note that if every column sum of A is at most c < 1, then the matrix norm ‖A‖₁ = max column sum ≤ c < 1. Since ‖Aᵏ‖ ≤ ‖A‖ᵏ ≤ cᵏ → 0, the partial sums I + A + ⋯ + Aᵏ form a Cauchy sequence and converge. This also means the spectral radius ρ(A) ≤ ‖A‖₁ < 1. Economically, convergence reflects the fact that each round of indirect demand is a smaller fraction of the previous round, since every sector adds value (consumes less than it produces).

SUMMARY

Lesson Summary

Summary

The Leontief input-output model represents an economy as a system of linear equations in which each sector is both a producer and a consumer. The technology matrix A encodes the technical coefficients aᵢⱼ—the fraction of sector j's output consumed by sector i. The balance equation x = Ax + d states that total output x must cover both intermediate demand Ax and final demand d. Rearranging yields the solution x = (I − A)⁻¹d, where (I − A)⁻¹ is the Leontief inverse.

The model requires A to be a productive matrix (all column sums < 1) so that (I − A)⁻¹ exists and is non-negative. Each entry of the Leontief inverse captures the total direct and indirect output required from one sector per unit of final demand for another. This framework, while assuming fixed coefficients and constant returns to scale, remains a foundational tool in economic planning, supply-chain analysis, and environmental accounting, and its mathematical structure connects naturally to the theory of non-negative matrices and Neumann series.