LC Abueg: mathematical economics

Chapter 5
Input-output models
Lectures in
Mathematical
Economics
L Cagandahan Abueg
De La Salle University
School of Economics

A model of open trade

External trade
Example. Consider three countries
1, 2, and 3. In each country,
consumption C is a fraction of
national income Y, i.e.,
Ci = ci Yi , ci ∈ [0,1), i = 1,2,3

chapter 5: input-output models 1

LC Abueg: mathematical economics

External trade
The imports of i from j, denoted Mji
is a fraction of national income of
i, i.e.,
Mji = mji Yi , mji ∈ [0,1), i = 1,2,3
Total imports of i is, therefore
Mi = m1i Yi + m2 i Yi + m3 i Yi
= mi Yi

External trade
Note that in the above,
m1i + m2 i + m3 i = mi
We also impose the hypothesis that
0 ≤ mi < ci < 1, i = 1,2,3
since empirically, the marginal
propensity to import is less than the
marginal propensity to consume.

External trade
Observe that the exports of j to i
are the imports of i to j. We then
have the following equations:
X1 = M12 + M13
= m12 Y2 + m13 Y3
X2 = m21Y1 + m23 Y3
X3 = m31Y1 + m32 Y2

chapter 5: input-output models 2

LC Abueg: mathematical economics

External trade
The equilibrium condition for each
country i is
Yi = Ci + Ii + Gi + X i − Mi
where G denotes government
spending and I denotes the level
of investment. Take country 1: we
have the equation
Y1 = C1 + I1 + G1 + X1 − M1

External trade
Note that
C1 = c1Y1, M1 = m1Y1
X1 = m12 Y2 + m13 Y3
Thus

Y1 = (c1 − m1)Y1
+ m12 Y2 + m13 Y3 + I1 + G1

External trade
Similarly, for country 2,
Y2 = m21Y1 + (c2 − m2 )Y2
+ m23 Y3 + I2 + G2
and for country 3,

Y3 = m31Y1 + m32 Y2
+ (c3 − m3 )Y3 + I3 + G3

chapter 5: input-output models 3

LC Abueg: mathematical economics

External trade
In matrix notation,
 Y1   Y1   I1 + G1 
     
Y2  = A Y2  + I2 + G2 
Y3  Y3  I3 + G3 
where
c1 − m1 m12 m13 
 
A =  m21 c2 − m2 m23 
 m31 m32 c3 − m3 

External trade
Let
 Y1   I1 + G1 
   
x =  Y2  , d = I2 + G2 
 Y3  I3 + G3 
Then
x = Ax + d
x − Ax = d
(I − A)x = d

The Leontief model

chapter 5: input-output models 4

LC Abueg: mathematical economics

Leontief model
Definition. An input-output model*
is an economic model which
analyzes linkages among industries.
The output of every industry is
produced by using inputs from
other industries, called intermediate
inputs. Inputs outside these
industries but used in production
are called primary inputs.
* W.W. Leontief [1936]. “Quantitative Input-Output Restrictions in the
Economic System of United States. Review of Economics and
Statistics, vol 18 no 3, 105-25.

Leontief model
Definition. Demand for outputs
emanating from the production
of more outputs among the
industries are called intermediate
demands. Output demand for
final consumption by economic
sectors outside these industries
(e.g., households, government,
external sector) are called final
demands.

Leontief model
Input-output table:
Final Total
Sector 1 Sector 2 … Sector n
demand output

Sector 1 x11 x12 … x1n d1 x1

Sector 2 x21 x22 … x2n d2 x2
…

…
…

Sector n xn1 xn2 … xnn dn xn

Total primary input
Primary
input 1 z11 z12 … z1n y1
…

…
…

Primary
input m zm1 zm2 … zmn ym
Total
input x1 x2 … xn

chapter 5: input-output models 5

LC Abueg: mathematical economics

Leontief model
where
xi = output level of sector i
xij = output of sector i supplied to
sector j
di = final demand of sector i
zkj = amount of primary input k in
sector j

Leontief model
Remark. An input-output table is
also called a transactions table.
Total output is also synonymous
to total demand. It also assumes
the following:
i. each industry produces only
one homogeneous output
ii. each industry uses a fixed
input ratio, and

Leontief model
iii. production in every industry
is subject to constant returns
to scale.
These assumptions are called
the economese conditions.

chapter 5: input-output models 6

LC Abueg: mathematical economics

Leontief model
Definition. In the input-output
model, the matrix A defined
below
x 
A = aij  =  ij 
 x j  i , j =1,...n
is called the technology matrix.
Every aij is called a technical
coefficient.

Leontief model
Remark. Every technical
coefficient from A gives the output
of sector i supplied to sector j to
produce one unit of output j, i.e.,
xij
aij =
xj
Re-expressing the above, we get
xij = aij x j

Leontief model
If each sector produces output
just enough to meet intermediate
demands and final demand, we
have, for a particular sector j,

x j = aj1x1 + aj 2 x2 + ⋯ + ajn xn + d j
Doing this for all sectors
j = 1,2,…,n, we have the system
of equations

chapter 5: input-output models 7

LC Abueg: mathematical economics

Leontief model
x1 = a11x1 + a12 x2 + ⋯ + a1n xn + d1
x2 = a21x1 + a22 x2 + ⋯ + a2 n xn + d2
⋮
xn = an1x1 + an2 x2 + ⋯ + ann xn + dn
In matrix notation, we have
x = Ax + d

Leontief model
or
x − Ax = d
(I − A)x = d
called the Leontief equation. The
matrix (I – A) is called the Leontief
matrix. Suppose (I – A) is
nonsingular. Applying Theorem
4.16, we obtain a unique solution
x * = (I − A)−1d

Leontief model
Note that the equation
x * = (I − A)−1d
gives the gross output vector that
must be produced in order to
satisfy a given set of final
demands.

chapter 5: input-output models 8

LC Abueg: mathematical economics

Leontief model
Example. Consider an input-output
table of Marlboro Country
Final Total
Sector1 Sector2 Sector3
demand output

Sector1 44.7 214.5 15.8 197.3 472.3

Sector2 42.8 210.1 174.2 861.9 1289.0
Sector3 28.6 149.9 290.8 1105.0 1574.3
Primary
input 356.2 714.5 1093.5
Total
input 472.3 1289.0 1574.3

Leontief model
We have the technology matrix
 44.7 214.5 15.8 
 472.3 1289.0 1574.3 
 
 42.8 210.1 174.2 
A=
 472.3 1289.0 1574.3 
 
 28.6 149.9 290.8 
 472.3 1289.0 1574.3 

Leontief model
Simplifying, we get
0.095 0.167 0.01 
 
A =  0.091 0.163 0.111
 0.06 0.116 0.185
 0.905 −0.167 −0.01 
 
⇒ (I − A) =  −0.091 0.837 −0.111
 −0.06 −0.116 0.815 

chapter 5: input-output models 9

LC Abueg: mathematical economics

Leontief model
Thus,
 1.131 0.232 0.045 
−1 
(I − A) =  0.137 1.246 0.171
 0.103 0.194 1.255 
 473.451 

−1 
x * = (I − A) d = 1290.000 
1574.291

Leontief model
Exercise. Consider an input-
output model with three sectors.
Sector 1 is heavy industry, sector
2 is light industry, and sector 3 is
agriculture. Suppose that the
requirements are given by the
following table:

Leontief model
Heavy Light Agriculture
industry industry
Units of heavy
0.1 0.2 0.1
industry goods
Units of light
0.3 0.2 0.2
industry goods
Units of
agricultural 0.2 0.2 0.1
goods

Suppose that final demands for

the three goods are 85, 95, and
20 units, respectively.

chapter 5: input-output models 10

LC Abueg: mathematical economics

Leontief model
Let x, y and z denote the number
of units that have to be
produced in the three sectors.
Verify that the solution for the
corresponding Leontief
economy matrix equation is
given by
T
x * = 150 200 100 

Leontief model
Example. Consider the following
technology matrix

0.365 0 0 
 
A= 0 0.897 0 
 0 0 0.331

What does this matrix tell about

the three sectors in the economy?

Leontief model
Example. Consider the following
technology matrix

0.05 0.1 0 
 
A =  0.1 0.25 0 
 0 0 0.8 

What does this matrix tell about

the three sectors in the economy?

chapter 5: input-output models 11

LC Abueg: mathematical economics

Hawkins-Simon condition

Hawkins-Simon condition
Definition. A Leontief economy with
technology matrix A is said to be
viable if and only if the Leontief
equation (I – A)x = d has a
nonnegative solution x* for every
nonnegative vector d.
Definition. A square matrix C
satisfies the Hawkins-Simon
condition if and only if all its leading
principal minors are positive.

Hawkins-Simon condition
Theorem 5.1. Let C be a square
matrix such that its off-diagonal
entries are nonpositive. Then the
following are equivalent:
i. Cx = d has a solution x* > θ
for some d > θ.
ii. Cx = d has a solution x* > θ
for every d > θ.
iii. C satisfies the Hawkins-Simon
condition.

chapter 5: input-output models 12

LC Abueg: mathematical economics

Hawkins-Simon condition
Example. Consider again the
input-output table of Marlboro
Country with the technology matrix
0.095 0.167 0.01 
 
A =  0.091 0.163 0.111
 0.06 0.116 0.185
We verify Hawkins-Simon condition
for I – A:

Hawkins-Simon condition
LPM1 = det  0.905
= 0.905 > 0
 0.905 −0.167 
LPM2 = det  
 −0.091 0.837 
= 0.742 > 0
LPM3 = det(I − A)
= 0.592 > 0

Hawkins-Simon condition
Example. Consider an input-output
table of the economy of Hogwarts
with two industries: Broomsticks
and Magicwands.

Broomsticks Magicwands Final demand Total output

Broomsticks 189.0 239.5 615.5 1044.0

Magicwands 168.8 412.8 416.3 997.9
Primary input 686.2 345.6
Total input 1044.0 997.9

chapter 5: input-output models 13

LC Abueg: mathematical economics

Hawkins-Simon condition
i. Calculate the technology
matrix.
ii. Obtain the Leontief matrix
and verify that it satisfies
Hawkins-Simon condition.
iii. Obtain the inverse of the
Leontief matrix.

Hawkins-Simon condition
iv. If final demand for
broomsticks decreased by
200 and final demand for
magicwands increased by
50, obtain the new levels of
output for the two sectors.

Hawkins-Simon condition
Solution. From the transactions
table, we obtain the technology
matrix
 189 239.5 
 1044 997.9 
A= 
168.8 412.8 
 1044 997.9 
 0.18 0.24 
= 
 0.16 0.41

chapter 5: input-output models 14

LC Abueg: mathematical economics

Hawkins-Simon condition
Thus, the Leontief matrix is given by
L = I−A
 1 0  0.18 0.24 
= − 
0 1  0.16 0.41
 0.82 −0.24 
= 
 −0.16 0.59 

Hawkins-Simon condition
Obtaining the inverse, we our
formula in obtaining inverses of
order 2 square matrices:

1  0.59 0.24 
(I − A)−1 =
0.44 0.16 0.82 

1.34 0.55 
= 
 0.36 1.86 

Hawkins-Simon condition
Example. Consider the following
transactions table of a certain
country (in million pesos):
Final Total
Agriculture Services
demand demand

Agriculture 20 30 30 80
Services 40 40 20 100

chapter 5: input-output models 15

LC Abueg: mathematical economics

Hawkins-Simon condition
Reconstruct the transactions
table with the same matrix of
technical coefficients if total
demands for agriculture and
services are 200 and 500,
respectively. Include the primary
input requirements in the table.

Hawkins-Simon condition
Exercise. Recall the coefficient
matrix in the model of
international trade:

c1 − m1 m12 m13 

 
A =  m21 c2 − m2 m23 
 m31 m32 c3 − m3 

Hawkins-Simon condition
By forming the corresponding
Leontief matrix, use the Hawkins-
Simon condition to show that the
Leontief equation
(I − A)x = d
has a solution vector x* > θ for
every d > θ.

chapter 5: input-output models 16

LC Abueg: mathematical economics

Brauer-Solow condition

Brauer-Solow condition
Remark. In the Leontief
equation (I – A)x = d, it may be
laborious to calculate LPMs as
the order of (I – A) becomes
sufficiently large. We then go to
another condition which only
involves row [or column] sums,
called the Brauer-Solow
condition.

Brauer-Solow condition
Theorem 5.2. [Brauer-Solow] Let A
be a nonnegative matrix. If at least
one of the two conditions below is
satisfied by A, then (I – A) satisfies
the Hawkins-Simon condition:
n
(i) rowi = ∑ aij < 1, ∀i = 1,2,..., n
j =1
n
( ii ) col j = ∑ aij < 1, ∀j = 1,2,..., n
i =1

chapter 5: input-output models 17

LC Abueg: mathematical economics

Brauer-Solow condition
Example. Consider again the
technology matrix of Marlboro
Country,
0.095 0.167 0.01 
 
A =  0.091 0.163 0.111
 0.06 0.116 0.185

Brauer-Solow condition
Remark. Brauer-Solow condition is
a sufficient condition, i.e., not all
matrices satisfying the Hawkins-
Simon condition satisfies the
Brauer-Solow condition. Example:
 0.1 0.3 
A= 
2.0 0.2 
 0.9 −0.3 
I−A =  
 −2.0 0.8 

Interpretation of (I – A)–1

chapter 5: input-output models 18

LC Abueg: mathematical economics

Interpretation of (I – A)–1
Definition. Let sector j increase in
final demand by 1 unit while the
other sectors remain the same.
The jth column sum
n
L j = ∑ ℓ ij
i =1

is called the output multiplier

corresponding to a unit increase
in the final demand of sector j.

Interpretation of (I – A)–1
Example. Consider
 1.131 0.232 0.045 
 −1 
(I − A) =  0.137 1.246 0.171
 0.103 0.194 1.255 

Interpretation of (I – A)–1
Example. Consider an input-output
table of the economy of Hogwarts
with two industries: Broomsticks
and Magicwands.

Broomsticks Magicwands Final demand Total output

Broomsticks 189.0 239.5 615.5 1044.0

Magicwands 168.8 412.8 416.3 997.9
Primary input 686.2 345.6
Total input 1044.0 997.9

chapter 5: input-output models 19

LC Abueg: mathematical economics

Interpretation of (I – A)–1
Example. The following inverse of
a Leontief matrix was obtained
from a transactions table whose
entries are in pesos:

1.59 0.65 0.29

 
0.68 1.50 0.39
 0.65 0.69 1.40 

Interpretation of (I – A)–1
i. Interpret the columns.
ii. Derive the technology matrix
A from which this matrix was
obtained.

Further results on (I – A)–1

chapter 5: input-output models 20

LC Abueg: mathematical economics

Further results on (I – A)–1

Theorem 5.3. Let A be a
nonnegative square matrix. Then
the Leontief equation
(I − A)x = d
has a nonnegative solution x* for
every nonnegative vector d if and
only if

1 > λ = max {α : α ∈ σ (A)}

Further results on (I – A)–1

Theorem 5.4. Let A be a
technology matrix of a Leontief
economy. Then the Leontief inverse
matrix L–1 = (I – A)–1 satisfies the
following properties

(i) (I − A)−1 ≥ Θ

( ii ) ℓ ii > 1, ∀i = 1,2,..., n

Further results on (I – A)–1

Example. Recall the inverse of a
Leontief matrix which was
obtained from a transactions
table whose entries are in pesos:

1.59 0.65 0.29


−1 
(I − A) = 0.68 1.50 0.39
 0.65 0.69 1.40 

chapter 5: input-output models 21

LC Abueg: mathematical economics

Demand for primary inputs

Demand for primary inputs

Example. Consider again the
input-output table of Marlboro
Country:
Final Total
Sector1 Sector2 Sector3
demand output

Sector1 44.7 214.5 15.8 197.3 472.3

Sector2 42.8 210.1 174.2 861.9 1289.0
Sector3 28.6 149.9 290.8 1105.0 1574.3
Primary
input 356.2 714.5 1093.5
Total
input 472.3 1289.0 1574.3

Demand for primary inputs

Note that we obtained the
corresponding technology matrix A:

0.095 0.167 0.01 

 
A =  0.091 0.163 0.111
 0.06 0.116 0.185
We do the same in the case of
primary inputs, given in the next
definition.

chapter 5: input-output models 22

LC Abueg: mathematical economics

Demand for primary inputs

Definition. In the input-output
model, the matrix B defined by

 zkj 
B = bkj  =  
 x j  j =1,... n
k =1,... m

is called the primary input matrix.

Every bkj is called a primary input
coefficient.

Demand for primary inputs

Remark. Every primary input
coefficient from B gives the amount
of primary input k needed to
produce one unit of output in
sector j, i.e.,
zkj
bkj =
xj
Re-expressing the above, we get
zkj = bkj x j

Demand for primary inputs

Input-output table:
Final Total
Sector 1 Sector 2 … Sector n
demand output

Sector 1 x11 x12 … x1n d1 x1

Sector 2 x21 x22 … x2n d2 x2
…

…
…

Sector n xn1 xn2 … xnn dn xn

Total primary input
Primary
input 1 z11 z12 … z1n y1
…

…
…

Primary
input m zm1 zm2 … zmn ym
Total
input x1 x2 … xn

chapter 5: input-output models 23

LC Abueg: mathematical economics

Demand for primary inputs

Note that the coefficients bkj are
the primary input coefficients, and
bkjxj is the amount of primary input
k needed to produce xj. Thus, the
total amount of primary input k in
production is
y k = bk1x1 + bk 2 x2 + ⋯ + bkn xn
Doing this for all primary inputs
k = 1,2,…,m, we have the system
of equations

Demand for primary inputs

y1 = b11x1 + b12 x2 + ⋯ + b1n xn
y2 = b21x1 + b22 x2 + ⋯ + b2 n xn
⋮
ym = bm1x1 + bm2 x2 + ⋯ + bmn xn

In matrix notation, we have

y = Bx

Demand for primary inputs

Note that if I – A is nonsingular, the
Leontief equation
(I − A)x = d
has a [unique] solution
x * = (I − A)−1d
Thus,
y* = Bx *
y* = B(I − A)−1d

chapter 5: input-output models 24

LC Abueg: mathematical economics

Demand for primary inputs

Example. Consider again the
input-output table of Marlboro
Country
Final Total
Sector1 Sector2 Sector3
demand output

Sector1 44.7 214.5 15.8 197.3 472.3

Sector2 42.8 210.1 174.2 861.9 1289.0
Sector3 28.6 149.9 290.8 1105.0 1574.3
Primary
input 356.2 714.5 1093.5
Total
input 472.3 1289.0 1574.3

Demand for primary inputs

We have the primary input matrix

 356.2 714.5 1093.5 

B= 
 472.3 1289.0 1574.3 
= 0.754 0.554 0.695
Suppose that the total amount of
inputs available is 2250 units. Then,
the total amount of inputs needed
for production is

Demand for primary inputs

y * = Bx *
T
 0.754   473.451 
   
=  0.554  1290.000 
 0.695  1574.291
= 2165.774
Hence, there is enough inputs
available to meet production
requirements.

chapter 5: input-output models 25

LC Abueg: mathematical economics

Demand for primary inputs

Exercise. If there are 2000 units of
input for Hogwarts, are the input
requirements sufficient, given the
transactions table below?

Broomsticks Magicwands Final demand Total output

Broomsticks 189.0 239.5 615.5 1044.0

Magicwands 168.8 412.8 416.3 997.9
Primary input 686.2 345.6
Total input 1044.0 997.9

Duality

Duality
Definition. Consider the matrix
equations (or equivalently, the
systems of equations)
Bx = b, Cy = c
We say that these matrix equations
are dual to each other iff
BT = C

chapter 5: input-output models 26

LC Abueg: mathematical economics

Duality
Theorem 5.5. Consider the dual
matrix equations
Bx = b, Cy = c
If x* and y* are the respective
solutions of the systems, then
b T y * = cT x *

Duality
Remark. Consider the Leontief
equation
(I − A)x = d [4]

A closed Leontief model

chapter 5: input-output models 27

LC Abueg: mathematical economics

Closed model
Recall the “usual” Leontief
model and the economese
conditions: these are embodied
in the corresponding
transactions table [or the input-
output table]of a particular
economy, industry, or system.

Closed model
Input-output table:
Final Total
Sector 1 Sector 2 … Sector n
demand output

Sector 1 x11 x12 … x1n d1 x1

Sector 2 x21 x22 … x2n d2 x2
…

…
…

Sector n xn1 xn2 … xnn dn xn

Total primary input
Primary
input 1 z11 z12 … z1n y1
…

…
…

Primary
input m zm1 zm2 … zmn ym
Total
input x1 x2 … xn

Closed model
Observe that final demand
emanates from an external
entity (say, households,
government) and primary inputs
originate from outside the
producing sectors (or industries).
For this reason, we call this
“usual” input-output table an
open Leontief model.

chapter 5: input-output models 28

LC Abueg: mathematical economics

Closed model
Without loss of generality,
consider a Leontief model with
three sectors (1, 2, and 3).
Final Total
Sector 1 Sector 2 Sector 3
demand output

Sector 1 x11 x12 x13 d1 x1

Sector 2 x21 x22 x23 d2 x2
Sector 3 x31 x32 x33 d3 x3
Primary
input z11 z12 z13

Total
input x1 x2 x3

Closed model
Now suppose that final demand
and primary inputs are within the
model, say another sector (4).
Total
Sector 1 Sector 2 Sector 3 Sector 4
output

Sector 1 x11 x12 x13 x14 x1

Sector 2 x21 x22 x23 x24 x2
Sector 3 x31 x32 x33 x34 x3
Sector 4 x41 x42 x43 x44 x4
Total
input x1 x2 x3 x4

Closed model
At first glance, this conversion
would not seem to create any
significant change in the analysis.
However, under the economese
conditions, industries (or sectors)
must follow a fixed-proportion
ratio of production of
intermediate goods

chapter 5: input-output models 29

LC Abueg: mathematical economics

Closed model
Thus, primary inputs become
intermediate inputs, and final
demand follows some determined
ratio (similar to that of the three
producing sectors in the model).
For this reason, this modification
makes this Leontief model a
closed Leontief model.

Closed model
Theorem 5.6. A closed Leontief
model
(I − A)x = θ
has a nonnegative solution x* if
and only if I – A is singular, i.e.,
det(I − A) = 0

Productive matrices

chapter 5: input-output models 30

LC Abueg: mathematical economics

Productive matrices
Definition. A technology matrix A
is said to be productive iff given
the Leontief matrix equation

(I − A)x = d, d = 1n
we have
x > θn

Productive matrices
Example. Determine which of
the following matrices are
productive:
1/ 3 1/ 2 
A2 =  
1/ 9 1/ 3 
 1/ 2 1/ 4 1/10 
 
A3 =  1/ 4 1/ 4 2 / 5 
1/10 2 / 5 1/ 4 

Productive matrices
Theorem 5.7. If a matrix A is
productive, then for every
nonnegative vector d, the
equation
(I − A)x = d
has a unique solution that
satisfies
x≥d

chapter 5: input-output models 31

LC Abueg: mathematical economics

Productive matrices
Theorem 5.8. Let A be a
nonnegative matrix. Then the
following are equivalent:
(i) A is productive.
( ii ) lim Ak = Θ
k →∞
∞
( iii ) (I − A)−1 = ∑ Ak , A0 = I
k =0

( iv ) −1
(I − A) ≥ Θ

Productive matrices
Theorem 5.9. [Perron-Frobenius]
Let A be a nonnegative matrix
and define
 1 
λA = inf  µ > 0 : A productive 
 µ 
Then λA > 0 is the largest [real]
eigenvalue of A and has an
associated eigenvector x > θ
(note: “inf” = infimum).

Productive matrices
Definition. Let A be a
nonnegative matrix and define
λA as in Theorem 5.9. We call λA
the Perron-Frobenius root of A.

chapter 5: input-output models 32

LC Abueg: mathematical economics

Productive matrices
Exercise. Find the Perron-Frobenius
root and an associated
eigenvector for each of the
following matrices:

1/ 3 1/ 2   1 2
A= , B =  
1/ 9 1/ 3   2 1

To end...
We could of course, dismiss the rigorous
proof as being superfluous: if a theorem is
geometrically obvious why prove it? This
was exactly the attitude taken in the
eighteenth century. The result, in the
nineteenth century, was chaos and
confusion: for intuition, unsupported by
logic, habitually assumes that everything is
much nicer behaved than it really is.
I. Stewart [1975]

chapter 5: input-output models 33