Input-Output Models

Sandip Sarkar

Department of Economics and Finance, BITS Pilani KK Birla Goa Campus, Goa, India.

Note: Lecture notes Mathematical and Statistical Methods Course: 2024.

1 Preliminaries
We assume that there are n (n > 1) industries such that each industries output is used
by other industries as inputs. Furthermore, a part of the output may also goes to the
market as final demand in an open sector (say domestic demand). Wassiely Leontief was
interested in the following question:

Question 1. What level of output each of the n industries produce, in order that it will
be just sufficient to satisfy the total demand of the product?

We make the following assumptions:

Assumption 1. Each industries produce a single and a homogeneous product.

Assumption 2. Each industry uses a fixed input ratio or factor combinations in the
production process.

Assumption 3. Production in every industry is subject to constant returns to scale.

Thus if the production function is given by y = f (L, K), then multiplying L and K by a
positive constant θ we have: θy = f (θL, θK).

The three assumptions enables us to ensures that in order to produce one unit of
the j th commodity, the input required for the ith commodity is a constant, which will be
denoted as aij . Thus for production of one unit of good ‘j ’ we need a1j units of good 1 as
input. We also need a2j of good 2, a3j of good 3,.., anj of good n as inputs for producing
good j.
Note that we always denote aij is how much the ith commodity is used for producing 1
unit of the j th commodity. We assume that prices are given and consider dollar as the unit
∗ Email id: [email protected]

1
of the output. Hence, aij is the value of commodity i required as input required to produce
1 unit of commodity j. For example, if steel industry (industry s) produces output worth
1000$ and uses 100$ of coal (industry c) as inputs, then acs = 100/1000 = 0.1. You can
consider acs either as unit free, or consider in dollar. For example, 0.1 dollar (10 cents)
of coal is required to produce 1 dollar worth steel. Or you may also interpret that 10%
of the output of steel is required for the purchasing coal as inputs. For an n industry
economy the input coefficients can be arranged in a matrix as follows:
 
a11 a12 a13 · · · a1n
 
 a21 a22 a23 · · · a2n 
 
A= .

.. .. .. 
 (1)
 .. . . ··· . 
 
an1 an2 an3 · · · ann

Formally aij is defined as the amount of input from industry i required to produce 1
dollar worth of steel. The j th column of the matrix A may also be interpreted as inputs
of different commodities required for production of 1 unit of commodity j (or 1 dollar
worth of commodity j). The elements in the principal diagonal of the matrix A (i..e, aii is
nothing but the input demand for good i required for producing 1 unit of good i). Thus
it is the portion of the output that is used as inputs for itself.

2 The Open Model

The column sum of all elements defined in matrix A must be less than 1. For sake of
argument, lets assume that the sum of all the elements in column j is greater than 1
n
P
(i.e., aij > 1) then it is not economically feasible, because to produce 1 dollar worth
i=1
of good j we need more than 1 dollar input costs. Furthermore, we also must have some
other costs, known as the primary inputs (labor, raw materials, managerial costs, etc).
n
P
Therefore ∀j ∈ {1, 2, .., n} we must have aij < 1. The volume of primary inputs for
i=1
n
P
producing commodity j is given by 1− aij . Note that a part of the primary inputs may
i=1
also incorporate profits for the industry. If industry 1 just satisfies the total demand, i..e,
it meets the input requirements for n industries, and the final demand for open sector
(say domestic demand), then it must satisfy the following equation:

2
 x1 = a11 x1 + a12 x2 + ... + a1n xn + d1 (2)

To illustrate equation 2 consider a hypothetical case with three industries: railways,

coal, and electricity. Let the total output generated by railways is 100$, out of which 20%
is used for self consumption, 30% of the output is used by the coal industry, 20% is used
by electricity, and 30$ goes for domestic use. Denoting railways, coal, and electricity,
as industry 1, 2, and 3, respectively, we can write a11 = 0.2, a12 = 0.3, and a13 = 0.2.
Therefore, equation 2 in this special case requires: 100 = 0.2∗100+0.3∗100+0.2∗100+30.
We write the followings set of equations for the n industries.

x1 = a11 x1 + a12 x2 + ... + a1n xn + d1

x2 = a21 x1 + a22 x2 + ... + a2n xn + d2

.. (3)
.

xn = an1 x1 + an2 x2 + ... + ann xn + dn

In terms of matrix notation we can rewrite the above set of equations as follows:

(I − A)x = d (4)

where I is an n × n identity matrix, A is the n × n input matrix defined in 1, d and

x are n × 1 vectors.1
The matrix I −A is also known as the Leontief matrix. Note that if the Leontief matrix
̸ 0) then [I − A]−1 exists and we can solve for the optimum
is non-singular (i.e, |I − A| =
values of output for different industries, the problem Leontief himself was interested in
(i..e, answer to Question 1). Thus the optimum solution is given by:

x∗ = (I − A)−1 d (5)

We now illustrate how we can solve for x∗ in the following numerical example.

   
d1 x1
 d2   x2 
1
More specifically, d =  .  and x =  . . Note that the order of LHS and RHS of equation 4 is
   
 ..   .. 
dn xn
n × 1. Other formulations like x(I − A) will have matrix conformability error.

3
Problem 1. Consider the flowchart below of an input-out model for three industries:
railways, coal, and electricity. We denote railways these three industries as industry 1,
2, and 3, respectively. Write the input matrix, Leontief matrix, and solve for x∗ .

3 Hawkins-Simons condition
Note that all elements of the vector x∗ must be non-negative. Otherwise such solutions
is not economically feasible. Now, non-singularity of the Leontief matrix, ensures ex-
istence of x∗ . However, we need to incorporate additional restrictions to ensure that
∀i ∈ {1, 2, .., n}: we have x∗i ≥ 0 are non-negative. Hawkins-Simons provides a condition
which is necessary and sufficient for non-negative x∗i . The condition is formally stated
below:

Proposition 1. Hawkins-Simons condition: Given any

1. n×n matrix B such that all off-diagonal elements are negative, i.e, bij ≤ 0 whenever
i ̸= j.

2. n × 1 vector d, such that all the elements are non-negative (i.e., ∀i ∈ {1, 2, .., n} we
have di ≥ 0).

4
 3. For any vector x∗ we have Bx∗ = d.

All elements of x∗ is non-negative if and only if the leading principal minors of the
matrix B are positive.

Note that if we define B = I − A, then it is easy to check that all the off-diagonal
elements are negative. Thus Hawkins-Simon condition provides a necessary and sufficient
condition for all elements of x∗ defined in 5 to be non-negative.

3.1 Principal (Leading) Minors of a Matrix

In linear algebra, a minor of any matrix M is the determinant of some smaller square
matrix, by cutting cutting down the ith row and j th column of the matrix, where i and j
not necessarily the same. If i = j, the the underlying minor is referred as the principal
b11 b12 b13
minor. For example, for any B = b21 b22 b23 , the simultaneous deletion of three ith
b31 b32 b33
b11 b12 b11 b13
row and column, we have the following three principal minors: , , and
b21 b22 b31 b33
b22 b23
.
b32 b33
Since these matrices are of order 2 × 2 we refer this as first order principal matrix.
Similarly, we can generate first order principal minors by deleting any two rows and
columns, where b11 , b22 , and b33 . The third order principal minor for matrix B is |B|
itself.
The set of principal matrices that are naturally ordered are referred as the leading
principal minors. For the matrix B, we have the following three leading principal minors:
b11 b12
B = b11 B2 = and B3 = |B|.
b21 b22

Problem 2. Find all the principal minors and the leading principal minors of the fol-
x11 x12 x13 x14
x21 x22 x23 x24
lowing 4 × 4matrix: X =
x31 x32 x33 x34
x41 x42 x43 x44

5
 Here is the solution for the leading principal minors.
NB: For Hawkins-Simion condition we need to concentrate only on Leading
principal minors.

3.2 Economic interpretation of the Hawkins-Simon condition

Recall that the Hawkins-Simon condition states that the output vector (i..e, x∗ ) is positive
whenever all the leading principal minors of the Leontief matrix (i.e., |I −A|) are positive.
For the sake of interpretation we consider a n = 2, and thus we have
 
1 − a11 −a12
I −A=  (6)
−a21 1 − a22

Now the first leading principal minor is 1 − a11 , and 1 − a11 > 0 =⇒ a11 < 1. The
second leading principal minor is |I − A| > 0 =⇒ (1 − a11 )(1 − a22 ) − a12 a21 > 0. The
last condition can be further expressed as

a11 + a12 a21 + (1 − a11 a22 ) < 1 (7)

6
 Now we already have (1 − a11 a22 ) ≥ 0, hence, we must have

a11 + a12 a21 < 1 (8)

Economically a11 means that the direct use of the first commodity as input for the
first commodity itself. Interestingly a12 a21 is the indirect use of the first commodity as
input for the first commodity. This is because the first industry is giving a fraction a12 to
second industry. In return the first industry gets a21 as inputs from the second industry.
Thus a part of the output send to industry 2 comes back to industry 1, and eventually
we have the indirect usage of first commodity as a21 a12 .
Equation 8 implies that the amount of first commodity used as direct and indirect
inputs in producing a dollar worth of the output must be less than 1 dollar. Thus what
Hawkins-Simon condition requires is certain practicability and visibility restrictions in
the production process..

Problem 3. An economy has three industries: fishing, forestry, and boat-building.

ˆ To produce 1 ton of fish requires the service of α fishing boats.

ˆ To produce 1 ton of timber requires β tons of fish to feed the foresters.

ˆ To produce 1 fishing boat requires γ tons of timbers.

These are the only inputs required for each of these three inputs. Suppose there is no
external demands for fishing boats. Find the gross output for each industries, provided the
external demands for fish and timbers are df and dt , respectively. Illustrate the Hawkins-
Simons condition in this regard.

4 The Closed Input Output Model

In a closed economy the exogenous sector is incorporated within the model just as a new
sector. Therefore, the primary inputs as well as the final demand (i.e, di = 0) does not
matter. All goods produced are like intermediate goods used by other industries. The
system of equations in this context can be written as follows:

7
 x0 = a01 x1 + a02 x2 + ... + a0n xn

x1 = a11 x1 + a12 x2 + ... + a1n xn

x2 = a21 x1 + a22 x2 + ... + a2n xn (9)

..
.

xn−1 = an−11 x1 + an−12 x2 + ... + an−1n−1 xn−1

In matrix notation the above equation can be written as follows:

(I − A)x = 0n (10)

where 0n is a n × 1 vector with all its elements as 0.

Note that one solution of the above equation is given by x∗i = 0∀i, which is a trivial
solution. However, multiple non-trivial solution can be obtained if |I − A| = 0, which
Pn−1
is true, since in the absence of primary inputs we have a0j = i=1 aij . Check that the
matrix I − A has linearly dependent columns. Thus in this case we have homogeneous
equations where we get multiple solutions.