Input-Output Analysis: The Intuition

Imagine a simple economy with just two industries: Steel and Cars. To make steel, you need steel itself (for machinery) and cars (to transport workers). To make cars, you need steel (for the body) and cars (for the assembly line). Each industry consumes some of its own output and some of the other's output just to keep running.

Now suppose the government wants to produce 100 extra cars for public transport. You can't just build 100 cars — you first need more steel to build them, and more cars to transport the steelworkers. But to get that extra steel, you need even more steel for the steel plant, and more cars for those workers. This creates a chain reaction: every new unit of output demands inputs from both industries, which in turn demand more inputs, and so on.

Input-Output Analysis is the mathematical tool that captures this entire chain in one clean calculation. It answers the question: If we want a certain final output (say, cars for the public), what total output must each industry actually produce, after accounting for all the inter-industry demands?

The Precise Statement

We model an economy with $n$ industries. Let:

• $x_i$ = total output of industry $i$ (what it must produce in total)
• $f_i$ = final demand for industry $i$'s product (what goes to consumers, government, exports — not to other industries)
• $a_{ij}$ = input coefficient — the amount of industry $j$'s output needed to produce one unit of industry $i$'s output

The input coefficient $a_{ij}$ is the key. It tells us the technology: "To make one car, you need 0.3 tons of steel and 0.1 cars' worth of transport." These coefficients are assumed fixed in the short run.

Now, the total output of industry $i$ must satisfy two things:

1. The intermediate demand from all industries (including itself) that need its product as an input
2. The final demand from outside the production system

The intermediate demand from industry $j$ for industry $i$'s product is $a_{ji} x_j$ (industry $j$ needs $a_{ji}$ units of $i$'s product per unit of its own output $x_j$). Summing over all $j$ gives the total intermediate demand for $i$'s output.

So the balance equation for each industry $i$ is:

In matrix form, let $\mathbf{x}$ be the column vector of total outputs, $\mathbf{f}$ the vector of final demands, and $\mathbf{A}$ the $n \times n$ matrix with entries $a_{ij}$ (where $a_{ij}$ is the input from industry $j$ to produce one unit of industry $i$). Then:

This is the Leontief Input-Output Model. The matrix $(\mathbf{I} - \mathbf{A})^{-1}$ is called the Leontief inverse. Its entry $(i,j)$ tells you the total output required from industry $i$ to deliver one unit of final demand from industry $j$ — including all the ripple effects.

Why It Works

The equation $\mathbf{x} = \mathbf{A}\mathbf{x} + \mathbf{f}$ is a linear system. Rearranging:

If $(\mathbf{I} - \mathbf{A})$ is invertible (which it is for a productive economy — the Hawkins-Simon condition), then:

The Leontief inverse captures the infinite chain: the initial demand $\mathbf{f}$ requires direct inputs $\mathbf{A}\mathbf{f}$, which require further inputs $\mathbf{A}^2 \mathbf{f}$, and so on. The geometric series $\mathbf{I} + \mathbf{A} + \mathbf{A}^2 + \cdots$ converges to $(\mathbf{I} - \mathbf{A})^{-1}$ when the economy is productive.

A Concrete Example

Suppose a two-industry economy (Steel = 1, Cars = 2) with coefficients:

• To make 1 ton of steel: need 0.2 tons of steel, 0.4 cars
• To make 1 car: need 0.3 tons of steel, 0.1 cars

So $\mathbf{A} = \begin{pmatrix} 0.2 & 0.3 \\ 0.4 & 0.1 \end{pmatrix}$.

Final demand: 50 tons of steel, 100 cars. So $\mathbf{f} = \begin{pmatrix} 50 \\ 100 \end{pmatrix}$.

Compute $\mathbf{I} - \mathbf{A} = \begin{pmatrix} 0.8 & -0.3 \\ -0.4 & 0.9 \end{pmatrix}$.

Its inverse (check the determinant $0.8 \times 0.9 - (-0.3)(-0.4) = 0.72 - 0.12 = 0.6$):

Then total outputs:

So steel must produce 125 tons (not just 50) and cars must produce 167 cars (not just 100) to satisfy both final demand and the inter-industry needs.

The Big Picture

Input-Output Analysis is a linear, static model. It assumes fixed proportions in production (no substitution between inputs) and no capacity constraints. It is used for:

• Planning: What should each sector produce to meet a target?
• Impact analysis: What happens to the whole economy if demand for one product changes?
• Environmental accounting: How much pollution is generated per unit of final demand?

The core insight is that everything depends on everything else — and the Leontief inverse is the mathematical tool that untangles that web in one clean step.