Linear Algebra › Section B · Matrix algebra › Topic B.6

The Leontief input-output model

To make cars, an economy needs steel; to make steel, it needs electricity; to make electricity, it needs steel again. The Leontief model untangles this web with one matrix equation and tells planners how much every industry must produce.

Industries feed each other

Every industry uses the products of other industries, and often its own, as inputs. A power company buys equipment from manufacturers, and manufacturers buy power. On top of this, households, government and export markets want goods for their own use. The economist Wassily Leontief turned this circular picture into a linear system and was awarded the Nobel Memorial Prize in Economic Sciences in 1973 for input-output analysis. The model is a direct application of matrix products and inverses.

The ingredients

Divide the economy into \(n\) producing sectors. The production vector \(\mathbf{x}\) lists each sector's output for a year, measured in dollars. The final demand vector \(\mathbf{d}\) lists what the non-producing part of the economy, called the open sector, wants from each sector. The consumption matrix \(C\) records the input needs: column \(j\) of \(C\) is the unit consumption vector of sector \(j\), listing the dollar value of input it buys from each sector to produce $1 of its own output. So entry \(c_{ij}\) is what sector \(j\) buys from sector \(i\) per dollar it produces.

Intermediate demand is Cx

If sector \(j\) produces \(x_j\) dollars of output, it needs \(x_j\) times its unit consumption vector in inputs. Adding these over all sectors gives

x_1\mathbf{c}_1 + x_2\mathbf{c}_2 + \cdots + x_n\mathbf{c}_n = C\mathbf{x}

which is exactly the column view of a matrix-vector product. The vector \(C\mathbf{x}\) is the intermediate demand: the output that the producing sectors themselves use up. Whatever is left over, \(\mathbf{x} - C\mathbf{x}\), is what can go to the open sector.

The Leontief equation

Leontief asked whether there is a production level that meets every demand exactly, with nothing short and nothing wasted. That means total output equals intermediate demand plus final demand:

\mathbf{x} = C\mathbf{x} + \mathbf{d}

Moving \(C\mathbf{x}\) to the left, and writing \(\mathbf{x}\) as \(I\mathbf{x}\), gives the linear system \((I - C)\mathbf{x} = \mathbf{d}\). If \(I - C\) is invertible, the production plan is \(\mathbf{x} = (I - C)^{-1}\mathbf{d}\). A solution only makes economic sense if every entry of \(\mathbf{x}\) is non-negative.

When a sensible solution exists

Suppose all entries of \(C\) and \(\mathbf{d}\) are non-negative and every column sum of \(C\) is less than 1. A column sum below 1 says the sector spends less than $1 on inputs per $1 of output, which a sector needs in order to make a profit. Under these conditions \(I - C\) is invertible, every entry of \((I - C)^{-1}\) is non-negative, and so \(\mathbf{x} = (I - C)^{-1}\mathbf{d}\) is a non-negative production plan. It is also the only plan that meets the demand exactly.

Worked example: a two-sector economy

In a simplified economy, each $1 of manufacturing output uses $0.50 of manufacturing and $0.25 of services, and each $1 of services uses $0.20 of manufacturing and $0.40 of services. Final demand is $50 million of manufacturing and $30 million of services. 1. The columns of \(C\) are \((0.5, 0.25)\) and \((0.2, 0.4)\), with sums 0.75 and 0.6, both below 1. 2. \(I - C\) has rows (0.5, −0.2) and (−0.25, 0.6), and \(ad - bc = 0.3 - 0.05 = 0.25\). 3. The \(2 \times 2\) formula gives

(I - C)^{-1} = 4\begin{bmatrix} 0.6 & 0.2 \\ 0.25 & 0.5 \end{bmatrix} = \begin{bmatrix} 2.4 & 0.8 \\ 1 & 2 \end{bmatrix}

4. In millions of dollars,

\mathbf{x} = (2.4(50) + 0.8(30), \ 1(50) + 2(30)) = (144, 110)

5. Check:

C\mathbf{x} = (72 + 22, \ 36 + 44) = (94, 80)

and \(\mathbf{x} - C\mathbf{x} = (50, 30)\), the final demand.

Rounds of demand

There is a second way to see the answer. To deliver \(\mathbf{d}\), sectors first need inputs \(C\mathbf{d}\). Producing those inputs needs further inputs \(C(C\mathbf{d}) = C^2\mathbf{d}\), and so on in ever smaller rounds. Adding all the rounds,

\mathbf{x} = \mathbf{d} + C\mathbf{d} + C^2\mathbf{d} + C^3\mathbf{d} + \cdots

When the column sums of \(C\) are less than 1, the powers \(C^k\) shrink towards the zero matrix, the series converges, and

(I - C)^{-1} = I + C + C^2 + C^3 + \cdots

This also explains why the inverse has no negative entries: it is a sum of matrices with non-negative entries.

Worked example: adding the rounds

Use the same economy and \(\mathbf{d} = (50, 30)\). 1. First round of inputs:

C\mathbf{d} = (0.5(50) + 0.2(30), \ 0.25(50) + 0.4(30)) = (31, 24.5)

2. Second round:

C^2\mathbf{d} = C(31, 24.5) = (15.5 + 4.9, \ 7.75 + 9.8) = (20.4, 17.55)

3. Running total after three terms:

(50 + 31 + 20.4, \ 30 + 24.5 + 17.55) = (101.4, 72.05)

4. The rounds keep shrinking; after 25 terms the total agrees with \((144, 110)\) to within 0.01. The exact inverse is faster, but the series shows where the extra output comes from.

Reading the inverse

The entries of \((I - C)^{-1}\) have a direct meaning. Since \(\mathbf{x} = (I - C)^{-1}\mathbf{d}\), raising final demand for sector \(j\) by $1 raises the required output by column \(j\) of \((I - C)^{-1}\). In the example, an extra $1 million of final demand for services requires $0.8 million more manufacturing and $2 million more services. Diagonal entries are always at least 1, because a sector must make the extra unit itself, plus whatever the rounds of inputs add.

Common mistakes to avoid

First, mixing up rows and columns of \(C\): column \(j\) describes what sector \(j\) buys, not what it sells. Second, solving \(C\mathbf{x} = \mathbf{d}\) or \((C - I)\mathbf{x} = \mathbf{d}\) instead of \((I - C)\mathbf{x} = \mathbf{d}\). Third, forgetting that the model is in money: entries are dollars of input per dollar of output. Fourth, reporting a plan with negative entries; check the column-sum condition, and check \(\mathbf{x} - C\mathbf{x} = \mathbf{d}\) at the end.

Key takeaways

In the Leontief model, x is total output, d is final demand from the open sector, and column j of the consumption matrix C lists the inputs sector j needs per dollar of output. Intermediate demand is Cx, and balance requires x = Cx + d, that is (I − C)x = d. When C and d are non-negative and every column sum of C is below 1, x = (I − C)⁻¹d is the unique, non-negative plan, and (I − C)⁻¹ = I + C + C² + ⋯. Column j of (I − C)⁻¹ gives the extra output needed for one more unit of final demand for sector j.

Flashcards

Flashcard

Production vector x

Each sector's total output, in dollars

What the model solves for.

Test your understanding

Question 1 of 5

In the Leontief equation x = Cx + d, what does the vector Cx represent?

Question 2 of 5

Column 2 of a consumption matrix C is (0.15, 0.30, 0.10). How much does sector 2 buy from sector 3 to produce $200 of its own output?

Question 3 of 5

For a two-sector economy, (I − C)⁻¹ has rows (2.4, 0.8) and (1, 2), and the final demand is d = (10, 20). What production vector x is needed?

Question 4 of 5

With the same (I − C)⁻¹, rows (2.4, 0.8) and (1, 2), final demand for sector 2 rises by $5 million. How much more must sector 1 produce?

Question 5 of 5

C and d have non-negative entries, and every column sum of C is less than 1. Which statement is guaranteed?

← Previous topicNext topic →