Essential Mathematics for Economics

Vectors and Matrices

Outline
 This section will study properties of linear
equations
 Consider a system of two equations
𝑎𝑥 + 𝑏𝑦 = 𝑐
𝑑𝑥 + 𝑒𝑦 = 𝑓
 The system can be written in matrix form
𝑎 𝑏 𝑥 𝑐
=
𝑑 𝑒 𝑦 𝑓

1
Outline
 We will cover topics in linear algebra
 Matrix operations
◼ Addition and subtraction
◼ Multiplication
◼ Finding inverse matrices
◼ Solutions of linear equations

 Applications

Solving an equation
 Let 𝑎𝑥 = 𝑏
𝑥 is a variable; a and b are constants
 There are three possible cases for the solution:
𝑏
 if a  0, there is a unique solution: 𝑥 = 𝑎 = 𝑎−1 𝑏
 if a = b = 0, any real number 𝑥 is a solution
 if a = 0 and b  0, there is no solution

2
Matrices
 A matrix with r rows and c columns is “an r  c
matrix”
 We also say it is “of order r  c”
 Matrix A is a 2  3 matrix
𝑎11 𝑎12 𝑎13
𝑨= 𝑎
21 𝑎22 𝑎23
 The first subscript of 𝑎𝑖𝑗 is the row; the second is the
column

Vectors
 A matrix with one row is a row vector:
𝒂 = 𝑎11 𝑎12 𝑎13
 A matrix with one column is a column vector:
𝑏11
𝒃 = 𝑏21
𝑏31

3
Matrix addition and subtraction
 Addition of matrices occurs element by element:
𝑎11 𝑎12 𝑏11 𝑏12
𝑨+𝑩= 𝑎 +
21 𝑎22 𝑏21 𝑏22
𝑎 + 𝑏11 𝑎12 + 𝑏12
= 11
𝑎21 + 𝑏21 𝑎22 + 𝑏22
 Scalar multiplication rescales each element:
𝑎11 𝑎12 𝑘𝑎11 𝑘𝑎12
𝑘 𝑎 =
21 𝑎22 𝑘𝑎21 𝑘𝑎22

Matrix addition and subtraction

 Subtraction of matrices occurs element by
element:
𝑨 − 𝑩 = A + (–1)B
𝑎11 𝑎12 −𝑏11 −𝑏12
= 𝑎 +
21 𝑎22 −𝑏21 −𝑏22
𝑎 − 𝑏11 𝑎12 − 𝑏12
= 11
𝑎21 − 𝑏21 𝑎22 − 𝑏22

4
Matrix multiplication
 Multiply matrices A and B to get C:
𝑎11 𝑎12 𝑏11 𝑏12
A×B= 𝑎 𝑎22 ×
21 𝑏21 𝑏22
𝑎11 𝑏11 + 𝑎12 𝑏21 𝑎11 𝑏12 + 𝑎12 𝑏22
= =𝑪
𝑎21 𝑏11 + 𝑎22 𝑏21 𝑎21 𝑏12 + 𝑎22 𝑏22
 To find element cij of C, multiply each element of
row i of A by the corresponding element of
column j of B, and add the products

Vector multiplication
 A special case is the product of row vector 𝒂 and
column vector 𝒃:
𝑏11
𝒂 × 𝒃 = 𝑎11 𝑎12 𝑎13 𝑏21
𝑏31

= [𝑎11 𝑏11 + 𝑎12 𝑏21 + 𝑎13 𝑏31 ]

5
Vector multiplication
 Now do the multiplication in the opposite order:
𝑏11
𝒃 × 𝒂 = 𝑏21 𝑎11 𝑎12 𝑎13
𝑏31
𝑏11 𝑎11 𝑏11 𝑎12 𝑏11 𝑎13
= 𝑏21 𝑎11 𝑏21 𝑎12 𝑏21 𝑎13
𝑏31 𝑎11 𝑏31 𝑎12 𝑏31 𝑎13

Example
 Consider the following matrix equation:
1 2 𝑥 4
× 𝑦 =
2 1 5
 Pre-multiply both sides by the same matrix:
1 2 1 2
− −
3 3 × 1 2 × 𝑥 = 3 3 × 4
2 1 2 1 𝑦 2 1 5
− −
3 3 3 3

6
Example
 The lefthand side is:
1 2
− 1 2 𝑥 1 0 𝑥 𝑥
3 3
2 1 × × 𝑦 = × 𝑦 = 𝑦
− 2 1 0 1
3 3

 The righthand side is:

1 2
− 4 2 𝑥 2
3 3
2 1 × = ⇒ 𝑦 =
− 5 1 1
3 3

Rules for matrix multiplication

 The matrix multiplication AB is feasible if and
only if the number of columns in A equals the
number of rows in B
A and B are conformable
3 5
 Let 𝑨 = 1 2 and 𝑩 =
4 2
 𝑨𝑩 = 11 9
 𝑩𝑨 does not exist

7
Rules for matrix multiplication
 Several properties hold for matrix multiplication
 Associative property: (AB)C = A(BC)
 Left distributive property: A(B + C) = AB + AC
 Right distributive property: (A + B)C = AC + BC
 Scalar multiplication: (kA)B = A(kB) = k(AB)

Examples
5 −1 2 0 1 2
 Calculate
1 4 3 1 2 3

5 −1 2 0 1 2
 Calculate
1 4 3 1 2 3

8
Matrix transpose

5 2
 Let 𝑨 =
3 4
5 3
 Its transpose is 𝑨𝑇 =
2 4
 Let 𝑩 = 1 3 5

1
 Its transpose is 𝑩𝑇 = 3
5

Matrix transpose

1 2 3
 Let 𝑪 =
4 5 6
1 4
 Its transpose is 𝑪𝑇 = 2 5
3 6
 Note that 𝑪𝑇 𝑇
=𝐶

9
Symmetric matrix
 Matrix 𝑨 is symmetric if 𝑨𝑇 = 𝑨
5 3
 For example, 𝑨 = is symmetric
3 4
 For two conformable matrices 𝑩 and 𝑪, the
transpose of the product is 𝑩𝑪 𝑇 = 𝑪𝑇 𝑩𝑇
 This means 𝑨𝑇 𝑨 𝑇 = 𝑨𝑇 (𝑨𝑇 )𝑇 = 𝑨𝑇 𝑨

Example
1 2
 Let 𝑨 =
3 4
 Find 𝑨 𝑨 and 𝑨𝑇 𝑨
𝑇 𝑇

10
Example
1 2
 Given 𝑨 = , we have
3 4
1 3 1 2 10 14
𝑨𝑇 𝑨 = =
2 4 3 4 14 20
 We also have
10 14
𝑨𝑇 𝑨 𝑇 =
14 20

Example
𝑥1 1
 Let 𝑨 = 𝑥2 and 𝑩 = 1
𝑥3 1
 Find 𝑨 𝑩 and 𝑨 𝑨
𝑇 𝑇

11
Transposes and multiplication
 Multiply matrices A and B:
𝑎11 𝑎12 𝑏11 𝑏12
A×B= 𝑎 𝑎22 ×
21 𝑏21 𝑏22
 Multiply matrices 𝑩𝑻 and 𝑨𝑻 :

𝑏11 𝑏21 𝑎11 𝑎21

𝑩𝑻 × 𝑨𝑻 = × 𝑎 𝑎22
𝑏12 𝑏22 12

Identity matrix
 An identity matrix is a square matrix with 1s
along the main diagonal and all other entries
equal to 0
 The 22 identity matrix is
1 0
𝑰2 =
0 1
 If C is an m  n matrix, then ImC = C and C = CIn

12
Example
 Calculate
1 0 0
1 0 𝑐11 𝑐12 𝑐13
0 1 0
0 1 𝑐21 𝑐22 𝑐23
0 0 1

Inverse matrix
𝑎11 𝑎12
 The inverse matrix of 𝑨 = 𝑎 is
21 𝑎22
1 𝑎22 −𝑎12
𝑨−1 =
𝑎11 𝑎22 −𝑎21 𝑎12 −𝑎21 𝑎11
 |A| = a11a22 – a21a12 is the determinant of A
 An inverse matrix does not exist if |A| ≠ 0
 The inverse matrix satisfies:
1 0
𝑨𝑨−1 = 𝑨−1 𝑨 =
0 1

13
Inverse matrix

 Check 𝑨−1 𝑨:

1 𝑎22 −𝑎12 𝑎11 𝑎12

|A| −𝑎21 𝑎11 𝑎21 𝑎22

1 𝑎22 𝑎11 − 𝑎12 𝑎21 𝑎22 𝑎12 − 𝑎12 𝑎22

=
|A| −𝑎21 𝑎11 − 𝑎11 𝑎21 −𝑎21 𝑎12 + 𝑎11 𝑎22

1 |A| 0
=
|A| 0 |A|

Systems of equations
 Consider a system of two equations:
𝑎11𝑥1 + 𝑎12𝑥2 = 𝑏1
𝑎21𝑥1 + 𝑎22𝑥2 = 𝑏2
 We can write the system in matrix notation:
𝑎11 𝑎12 𝑥1 𝑏1
𝑎21 𝑎22 𝑥2 = or 𝑨𝒙 = 𝒃
𝑏2

14
Systems of equations
 Let 𝑨𝒙 = 𝒃 and pre-multiply both sides by A–1:
𝑨−1 𝑨𝒙 = 𝑨−1 𝒃 ⇒ 𝑰𝒙 = 𝑨−1 𝒃 ⇒ 𝒙 = 𝑨−1 𝒃
 This yields the solution:
𝑥1 1 𝑎22 −𝑎12 𝑏1
𝑥2 =
𝑎11 𝑎22 −𝑎21 𝑎12 −𝑎21 𝑎11 𝑏2
1 𝑎22 𝑏1 − 𝑎12 𝑏2
=
𝑎11 𝑎22 −𝑎21 𝑎12 −𝑎21 𝑏1 + 𝑎11 𝑏2

Determinant equal to zero

 Consider the following system:
4𝑥1 + 6𝑥2 = 𝑏1
2𝑥1 + 3𝑥2 = 𝑏2
 We can write the system as 𝑨𝒙 = 𝒃:
4 6
𝑨= ⇒ 𝑨 =4 3 −2 6 =0
2 3

15
Determinant equal to zero
 The original system is equivalent to the
following:
0 = 𝑏1 − 2𝑏2
2𝑥1 + 3𝑥2 = 𝑏2
 There may be no solution or infinitely many
solutions to 𝑨𝒙 = 𝒃

Input-output analysis
 In 1949, Leontief divided data from the U.S.
economy into 500 sectors (e.g., coal industry,
automotive industry, etc.)
 A linear equation described how each sector
divided its output among the sectors
 Leontief aggregated the sectors so that there were
42 equations in 42 unknowns
 It took a computer 56 hours to solve the system of
equations

16
The general input-output model
 There are fixed proportions and constant returns
to scale in production
 Useaij units of good i when producing one unit of
good j
 If the output of good j is 𝑥𝑗, total usage of 𝑥1 as an
input is:
𝑎11𝑥1 + 𝑎12𝑥2 + ⋯ + 𝑎1𝑛𝑥𝑛
 Good k’s usage as an input is:
𝑎𝑘1𝑥1 + 𝑎𝑘2𝑥2 + ⋯ + 𝑎𝑘𝑛𝑥𝑛

The general input-output model

 Total input usage is
𝑎11 𝑎12 ... 𝑎1𝑛 𝑥1
𝑎21 𝑎22 ... 𝑎2𝑛 𝑥2
𝑨𝒙 = . . . ... ... ... ...
𝑎𝑛1 𝑎𝑛2 ... 𝑎𝑛𝑛 𝑥𝑛
𝑎11 𝑥1 + 𝑎12 𝑥2 +. . . +𝑎1𝑛 𝑥𝑛
𝑎 𝑥 + 𝑎22 𝑥2 +. . . +𝑎2𝑛 𝑥𝑛
= 21 1 ...
𝑎𝑛1 𝑥1 + 𝑎𝑛2 𝑥2 +. . . +𝑎𝑛𝑛 𝑥𝑛

17
The general input-output model
 Production of each good is divided between
input use and consumption:
𝒙 = 𝑨𝒙 + 𝒄 ⇒ (𝑰– 𝑨)𝒙 = 𝒄
𝒙 is the production vector
 𝑨𝒙 is the input requirements vector
c is the consumption vector
 Production is 𝒙 = (𝑰– 𝑨)−1 𝒄

Example
 Let desired consumption be c1 = 18 and c2 = 24,
0 0.5
and let 𝑨 =
0.5 0
 Production is 𝒙 = (𝑰– 𝑨)−1 𝒄, where:
1 −0.5
𝑰−𝑨=
−0.5 1
4 1 0.5
 (𝑰 − 𝑨)−1 =
3 0.5 1

18
Example
 We now have:
4 1 0.5 18 40
𝒙 = (𝑰 − 𝑨)−1 𝒄 = =
3 0.5 1 24 44
 Input requirements are:
0 0.5 40 22
𝑨𝒙 = =
0.5 0 44 20

Comparative statics
 If desired consumption rises by 𝒄, we must have:
𝒙 + 𝒙 = 𝑨 𝒙 + 𝒙 + 𝒄 + 𝒄
⇒ 𝒙 = 𝑨𝒙 + 𝒄
4 1 0.5 Δ𝑐1
⇒ 𝒙 = (𝑰 − 𝑨)−1 𝒄 =
3 0.5 1 Δ𝑐2
 This gives:
Δ𝑥1 4 Δ𝑐1 + 0.5Δ𝑐2 1 4Δ𝑐1 + 2Δ𝑐2
= =
Δ𝑥2 3 0.5Δ𝑐1 + Δ𝑐2 3 2Δ𝑐1 + 4Δ𝑐2

19
Comparative statics
Δ𝑐1 1
 If = , the change in production is:
Δ𝑐2 0
4
4 1 0.5 1 3
𝒙 = (𝑰 − 𝑨)−1 𝒄 = = 2
3 0.5 1 0
3
1
3
⇒ 𝒙 − ∆𝒄 = 2
3

Example
 Suppose that 0.8 units of good 1 are needed to
produce one unit of good 2 and 0.75 units of
good 2 are needed to produce one unit of good 1
 Find the consumption matrix, 𝑨

20
Transition matrices
 In many situations, the “state of the world”
tomorrow depends on the state of the world
today
 Consider the case of two states
 Initially, there are 𝑥𝑗0 people in state j
 The number of people in state j in period t is 𝑥𝑗𝑡

Transition matrices
 Let aij denote the fraction of people currently in
state j who will be in state i next period
 The number in state 1 in period 1 is the number
in state 1 in period 0 who remain, plus the
number from state 2 who move in:
𝑥11 = 𝑎11 𝑥10 + 𝑎12 𝑥20
𝑥21 = 𝑎21 𝑥10 + 𝑎22 𝑥20

21
Transition matrices
 The transition matrix is:
𝑎11 𝑎12
𝑨= 𝑎
21 𝑎22
 Populations in period 1 are:
𝑎11 𝑎12 𝑥10 𝑥11
𝑎21 𝑎22 𝑥 0 = 𝑥1
2 2

Transition matrices
 Suppose that the populations do not change
from period to period:
𝑥1𝑡+1 = 𝑥1𝑡 ⇒ 𝑥1𝑡 = 𝑎11 𝑥1𝑡 + 𝑎12 𝑥2𝑡
𝑥2𝑡+1 = 𝑥2𝑡 ⇒ 𝑥2𝑡 = 𝑎21 𝑥1𝑡 + 𝑎22 𝑥2𝑡
𝑥1
 In general, 𝑥 is a steady state if 𝑨𝒙 = 𝒙 or:
2
𝑥1 = 𝑎11 𝑥1 + 𝑎12 𝑥2
𝑥2 = 𝑎21 𝑥1 + 𝑎22 𝑥2

22
Example
 Let the transition matrix be:
0.5 0.5
𝑨=
0.5 0.5
 The populations in period 0 are 𝑥10 and 𝑥20 , with
𝑥10 + 𝑥20 = 𝑥ҧ
 Find the populations in period 1

Example

0.5 0.5
 Given 𝑨 = , period-1 populations are:
0.5 0.5

0.5 0.5 𝑥1
0
0.5(𝑥10 + 𝑥20 ) 0.5𝑥ҧ
0 = 0 0 =
0.5 0.5 𝑥2 0.5(𝑥1 + 𝑥2 ) 0.5𝑥ҧ
 Find the steady state

23
Example
 The steady state requires:
0.5 0.5 𝑥1 𝑥1
𝑨𝒙 = 𝒙 ⇒ =
0.5 0.5 𝑥2 𝑥2
0.5(𝑥1 + 𝑥2 ) 𝑥1
⇒ = 𝑥
0.5(𝑥1 + 𝑥2 ) 2

⇒ 𝑥1 = 𝑥2 = 0.5𝑥ҧ

Example

0 1
 Let the transition matrix be 𝑨 =
1 0
 Solving 𝑨𝒙 = 𝒙 gives:
𝑥1 𝑥1 0 1 𝑥1 𝑥1
𝑨 𝑥 = 𝑥 ⇒ =
2 2 1 0 𝑥2 𝑥2
𝑥2 𝑥1
⇒ 𝑥 = 𝑥
1 2

24
Example
 There are p people in Eatstown
 If someone eats dinner at home today, there is
probability 0.85 that the individual will eat
dinner at home tomorrow
 If someone eats dinner at a restaurant today,
there is probability 0.90 that the individual will
eat dinner at home tomorrow

Example
 Let h0 be the number who eat at home in period
0 and let r0 be the number who eat at a
restaurant
 Determine how many people eat at home in
period 1 and how many eat at a restaurant
 Express the answer as two linear equations and
in matrix form:
ℎ1 = 𝑨 ℎ0
𝑟1 𝑟0

25
Eigenvectors and Eigenvalues
 When a square matrix multiplies a vector, the
vector is rotated and scaled
 Vectors called eigenvectors (or characteristic
vectors) are only scaled
 The scaling factors are the eigenvalues (or
characteristic values)

Eigenvectors and Eigenvalues

1 4 2
 Let 𝒙 = and 𝑨 =
0 2 4
 The product is:
4 2 1 4
𝑨𝒙 = =
2 4 0 2
1
 Now let 𝒙 = :
−1
4 2 1 2
𝑨𝒙 = =
2 4 −1 −2

26
Eigenvectors and Eigenvalues

1
 Let 𝒙 =
1
 The product is:

4 2 1 6
𝑨𝒙 = =
2 4 1 6

Eigenvalues
 If 𝑨𝒙 = 𝒙 for some , then
𝑎 −𝜆 𝑎12 𝑥1
𝑨 − 𝑰 𝒙 = 11
𝑎21 𝑎22 −  𝑥2
𝑎11 − 𝜆 𝑥1 + 𝑎12 𝑥2 0
= =
𝑎21 𝑥1 + 𝑎22 − 𝜆 𝑥2 0
 |A–I| = 0 yields an equation called the
characteristic polynomial or the characteristic
equation
 The roots of the equation are the eigenvalues

27
Example

4 2
 Return to 𝑨 =
2 4
4−𝜆 2
 𝑨 − 𝜆𝑰 = and its determinant is
2 4−𝜆
(4 − 𝜆)2 − 4 = 𝜆2 − 8𝜆 + 12 = (𝜆 − 6)(𝜆 − 2)
 The roots are 𝜆 = 6 and 𝜆 = 2

Eigenvectors and Eigenvalues

 When  = 6, we have 𝑨𝒙 = 6𝒙 for particular
vectors
 𝑨𝒙 = 6𝒙 gives two equations in two unknowns:
4𝑥1 + 2𝑥2 = 6𝑥1
2𝑥1 + 4𝑥2 = 6𝑥2
1
 Both equations imply 𝑥1 = 𝑥2 , so 𝑥 = is an
1
eigenvector

28
Eigenvectors and Eigenvalues
 𝑨𝒙 = 2𝒙 gives two equations:
4𝑥1 + 2𝑥2 = 2𝑥1
2𝑥1 + 4𝑥2 = 2𝑥2
1
 Both imply 𝑥1 = – 𝑥2, so 𝑥 = is an
−1
eigenvector

Transition matrices and eigenvalues

𝑎11 𝑎12
 Let 𝑨 = 𝑎 𝑎22
21
 A steady state has 𝑨𝒙 = 𝒙, so  = 1 is one
eigenvalue
 We can also rewrite the transition matrix as
𝑎11 𝑎12
𝑨= 1−𝑎
11 1 − 𝑎12

29
Steady state
 Solving 𝑨𝒙 = 𝒙 yields:
𝑥1 = 𝑎11 𝑥1 + 𝑎12 𝑥2
𝑥2 = (1 − 𝑎11 )𝑥1 + (1 − 𝑎12 )𝑥2
 Both equations amount to
(1 − 𝑎11 )𝑥1 − 𝑎12 𝑥2 = 0
 We also have the aggregate population
condition:
𝑥1 + 𝑥2 = 𝑥

Steady state
 The equations are:
(1 − 𝑎11 )𝑥1 − 𝑎12 𝑥2 = 0
𝑥1 + 𝑥2 = 𝑥
 Substituting the second into the first yields:
(1 − 𝑎11 )𝑥1 − 𝑎12 ( 𝑥−𝑥1 ) = 0
⇒ 1 − 𝑎11 +𝑎12 𝑥1 = 𝑥

30
Steady state
 There are two cases for 1 − 𝑎11 + 𝑎12 𝑥1 = 𝑥
 If 1 − 𝑎11 +𝑎12 ≠ 0, we have:
𝑎12 𝑥
𝑥1 =
1−𝑎11 +𝑎12

 If 1 − 𝑎11 + 𝑎12 = 0, we must have 𝑎11 = 1 and

𝑎12 = 0

Summary
 We have introduced concepts from linear
algebra
 Matrices

 Vectors

 Matrix operations
 We have seen applications
 Transition matrices and steady states
 Eigenvectors and eigenvalues

31