You are on page 1of 63

Structure of this lecture

Basic info on myself and this module


Essential of matrix arithmetic
Multiple regression models
Statistical inference and hypothesis testing

Main learning outcomes:


- Basic understanding of matrix algebra
- Basic understanding of classical linear regression model
assumptions
- Basic understanding of how to construct and interpret a
hypothesis test

My contact details
Dr Enrico Onali

Division: Financial Studies


Location: Room 2.07, Alun Building
Telephone: 01248 38 3650
Email: e.onali@bangor.ac.uk
Office hours:
Monday 11.30-12.30
Monday 13.30-14.30
or by appointment
2

Who am I?

Senior Lecturer in Finance


Italian
Experience in Italian retail bank
Current research interests: Banking and financial regulation,
dividend policy, market efficiency, applied econometrics
Profile at Google Scholar:
http://scholar.google.co.uk/citations?user=cro4BU0AAAAJ&hl=it&oi=ao

Links to my publications (access from BU intranet needed):


http://www.sciencedirect.com/science/article/pii/S0278425414000696
http://onlinelibrary.wiley.com/doi/10.1111/jbfa.12057/abstract
http://0-www.sciencedirect.com.unicat.bangor.ac.uk/science/article/pii/S1057521912000579
http://0-www.sciencedirect.com.unicat.bangor.ac.uk/science/article/pii/S0165176512002844
http://0-www.sciencedirect.com.unicat.bangor.ac.uk/science/article/pii/S1057521911000123
http://0-www.sciencedirect.com.unicat.bangor.ac.uk/science/article/pii/S1057521909000337
http://0-onlinelibrary.wiley.com.unicat.bangor.ac.uk/doi/10.1111/j.1467-646X.2010.01037.x/full

Financial econometrics
What is it? Why is it useful?
Financial econometrics: Basically, a branch of applied
mathematics
Dissertation: This module will be very helpful for those of you
who decide to write a dissertation
Employability: employers like people with numeracy skills,
although returns to skills vary with the country
http://www.economist.com/blogs/freeexchange/2014/01/economic-value-skills

Those of you who plan to work in the academia, central banks,


think-tanks, will find this module extremely useful.

Short course overview

Technical background (today)


Regression analysis using time series data
Univariate modelling of stationary time series variables
Modelling non-stationary time series variables
Multivariate modelling of stationary time series variables
Cointegration and multivariate modelling of non-stationary
time series
Modelling volatility in financial time series data
Panel data econometrics
5

Textbook and readings


Main textbook:
C. Brooks. Introductory econometrics for finance, 2nd edition.
Cambridge University Press, 2008.
C. Brooks. Introductory econometrics for finance, 3rd edition.
Cambridge University Press, 2014.
Other references:
B.H. Baltagi. Econometric analysis of panel data, 4th ed., Wiley 2008.
W. Enders. Applied econometric time series, 2nd ed., Wiley, 2005.
W.H. Greene. Econometric analysis, 6th edition, Prentice Hall, 2002.
G. Koop. Analysis of financial data, Wiley, 2006.
M. Verbeek. A guide to modern econometrics, 3rd ed., Wiley, 2008.
C. Cameron and P. K. Trivedi. Microeconometrics Using Stata, Revised
Edition. Stata Press, 2010.

1.

TECHNICAL BACKGROUND

Essentials of matrix arithmetic


(Several of the examples used here are taken from Brooks pp.611-615,
2nd edition or pp. 41-48, 3rd edition).

1.

TECHNICAL BACKGROUND

Essentials of matrix arithmetic


(Several of the examples used here are taken from Brooks pp.611-615,
2nd edition or pp. 41-48, 3rd edition).
Notation and special types of matrix
A matrix is a collection or array of numbers. In econometrics, it is often
convenient to use matrices to represent data. Using matrices, we can write
down concise formulae for estimators or test statistics in terms of the data
from which they are computed.

1.

TECHNICAL BACKGROUND

Essentials of matrix arithmetic


(Several of the examples used here are taken from Brooks pp.611-615,
2nd edition or pp. 41-48, 3rd edition).
Notation and special types of matrix
A matrix is a collection or array of numbers. In econometrics, it is often
convenient to use matrices to represent data. Using matrices, we can write
down concise formulae for estimators or test statistics in terms of the data
from which they are computed.
The dimensions of a matrix are quoted as (RC), where R = number of
rows, C = number of columns.

1.

TECHNICAL BACKGROUND

Essentials of matrix arithmetic


(Several of the examples used here are taken from Brooks pp.611-615,
2nd edition or pp. 41-48, 3rd edition).
Notation and special types of matrix
A matrix is a collection or array of numbers. In econometrics, it is often
convenient to use matrices to represent data. Using matrices, we can write
down concise formulae for estimators or test statistics in terms of the data
from which they are computed.
The dimensions of a matrix are quoted as (RC), where R = number of
rows, C = number of columns.

Each element of a matrix is referred to using subscripts. For the matrix A,


ai,j refers to the element in the ith row and the jth column of A. If the
dimensions of A are (23), we can write:
A = a 1,1 a 1, 2 a 1,3
a

2,1 a 2, 2 a 2,3

10

Essentials of matrix arithmetic (cont'd)


A matrix with only one row, or dimensions of (1C), is known as a row
vector.

A matrix with only one column, or dimensions of (R1), is known as a


column vector.

11

Essentials of matrix arithmetic (cont'd)


A matrix with only one row, or dimensions of (1C), is known as a row
vector.

A matrix with only one column, or dimensions of (R1), is known as a


column vector.
A square matrix has the same number of rows and columns.

12

Essentials of matrix arithmetic (cont'd)


A matrix with only one row, or dimensions of (1C), is known as a row
vector.

A matrix with only one column, or dimensions of (R1), is known as a


column vector.
A square matrix has the same number of rows and columns.
A diagonal matrix is a square matrix with non-zero values on the main
diagonal, and 0s in all other elements. For example:
3 0 0 0

0 1 0 0
A=
0 0 2 0

0 0 0 1

13

Essentials of matrix arithmetic (cont'd)


An identity matrix, denoted I or Im (where m denotes the number of rows
and columns), is a square matrix with 1s in all elements on the main
diagonal, and 0s in all other elements.
I2 = 1 0

0 1

14

Essentials of matrix arithmetic (cont'd)


An identity matrix, denoted I or Im (where m denotes the number of rows
and columns), is a square matrix with 1s in all elements on the main
diagonal, and 0s in all other elements.
I2 = 1 0

0 1

I3 = 1 0 0

0 1 0
0 0 1

15

Essentials of matrix arithmetic (cont'd)


An identity matrix, denoted I or Im (where m denotes the number of rows
and columns), is a square matrix with 1s in all elements on the main
diagonal, and 0s in all other elements.
I2 = 1 0

0 1

I3 = 1 0 0

0 1 0
0 0 1

I4 = 1

0
0

0
1
0
0

0
0
1
0

0
0

16

Essentials of matrix arithmetic (cont'd)


An identity matrix, denoted I or Im (where m denotes the number of rows
and columns), is a square matrix with 1s in all elements on the main
diagonal, and 0s in all other elements.
I2 = 1 0

0 1

I3 = 1 0 0

0 1 0
0 0 1

I4 = 1

0
0

0
1
0
0

0
0
1
0

0
0

A symmetric matrix is a square matrix that is symmetric around the main


diagonal, so that ai,j=aj,i for all i and j. For example:

4
7
1 2
A=

9
2 3 6
4 6
2 8

7 9 8 0

17

Essentials of matrix arithmetic (cont'd)


Addition and subtraction
Two matrices can be added or subtracted if (and only if) both matrices
have the same dimensions. If so, the corresponding elements of the two
matrices are added or subtracted one by one.

18

Essentials of matrix arithmetic (cont'd)


Addition and subtraction
Two matrices can be added or subtracted if (and only if) both matrices
have the same dimensions. If so, the corresponding elements of the two
matrices are added or subtracted one by one.
If A = 0.3 0.6
0.1 0.7

and B = 0.2 0.1

0.3
0

19

Essentials of matrix arithmetic (cont'd)


Addition and subtraction
Two matrices can be added or subtracted if (and only if) both matrices
have the same dimensions. If so, the corresponding elements of the two
matrices are added or subtracted one by one.
If A = 0.3 0.6
0.1 0.7

and B = 0.2 0.1

0.3
0

0.5 0.5
0.3 0.2 0.6 0.1
A+B=

=
0.1 1
0.1 0 0.7 0.3

20

Essentials of matrix arithmetic (cont'd)


Addition and subtraction
Two matrices can be added or subtracted if (and only if) both matrices
have the same dimensions. If so, the corresponding elements of the two
matrices are added or subtracted one by one.
If A = 0.3 0.6
0.1 0.7

and B = 0.2 0.1

0.3
0

0.5 0.5
0.3 0.2 0.6 0.1
A+B=

=
0.1 1
0.1 0 0.7 0.3
0.3 0.2 0.6 0.1
0.1 0.7
A B =
=

0.1 0.4
0.1 0 0.7 0.3

21

Essentials of matrix arithmetic (cont'd)


Multiplication

(i)

Scalar multiplication

To multiply a matrix by a scalar (a single number), every element of the


matrix is multiplied by that number.
If

A = 0.3 0.6
0.1 0.7

then

2A = 2 0.3 0.6 = 0.6 1.2


0.1 0.7 0.2 1.4

22

Essentials of matrix arithmetic (cont'd)


Multiplication

(i)

Scalar multiplication

To multiply a matrix by a scalar (a single number), every element of the


matrix is multiplied by that number.
If

(ii)

A = 0.3 0.6
0.1 0.7

then

2A = 2 0.3 0.6 = 0.6 1.2


0.1 0.7 0.2 1.4

Matrix multiplication

Two matrices can be multiplied together if (and only if) the number of
columns in the 1st matrix equals the number of rows in the 2nd matrix.
The number of rows in the 1st matrix determines the number of rows in
the product matrix, and the number of columns in the 2nd matrix
determines the number of columns in the product matrix.

23

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

In other words, A and B can be multiplied together if A is (mn) and B is


(np). The dimensions of C=AB are (mp).

24

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

In other words, A and B can be multiplied together if A is (mn) and B is


(np). The dimensions of C=AB are (mp).
The element in the ith row and jth column of the product matrix is
obtained by summing the products of the corresponding elements in the
ith row of the 1st matrix and the jth column of the 2nd matrix.

25

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

In other words, A and B can be multiplied together if A is (mn) and B is


(np). The dimensions of C=AB are (mp).
The element in the ith row and jth column of the product matrix is
obtained by summing the products of the corresponding elements in the
ith row of the 1st matrix and the jth column of the 2nd matrix.

1 2
Let A =

7 3
1 6

and B = 0 2 4 9
6 3 0 2

26

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

In other words, A and B can be multiplied together if A is (mn) and B is


(np). The dimensions of C=AB are (mp).
The element in the ith row and jth column of the product matrix is
obtained by summing the products of the corresponding elements in the
ith row of the 1st matrix and the jth column of the 2nd matrix.

0 2 4 9
1 2
Let A =
and B =

6 3 0 2
7 3
1 6

The dimensions of A are (32) and the dimensions of B are (24).


Therefore the dimensions of C=AB are (34).

c1,1

Let C = c 2,1
c
3,1

c1, 2
c 2, 2
c 3, 2

c1,3
c 2,3
c 3, 3

c1, 4

c 2, 4
c 3, 4

27

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

1 2
A=

7 3
1 6

B = 0 2 4 9

6 3 0 2

c1,1 is the sum of the products of the 1st row of A and the 1st column of B
c1,1 = 10 + 26 = 12

28

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

A=

1 2

7 3
1 6

B = 0 2 4 9

6 3 0 2

c1,1 is the sum of the products of the 1st row of A and the 1st column of B
c1,1 = 10 + 26 = 12
c1,2 is the sum of the products of the 1st row of A and the 2nd column of B
c1,2 = 12 + 23 = 8

29

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

1 2
A=

7 3
1 6

B = 0 2 4 9

6 3 0 2

c1,1 is the sum of the products of the 1st row of A and the 1st column of B
c1,1 = 10 + 26 = 12
c1,2 is the sum of the products of the 1st row of A and the 2nd column of B
c1,2 = 12 + 23 = 8
c1,3 is the sum of the products of the 1st row of A and the 3rd column of B
c1,3 = 14 + 20 = 4

30

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

1 2
A=

7 3
1 6

B = 0 2 4 9

6 3 0 2

c1,1 is the sum of the products of the 1st row of A and the 1st column of B:
c1,1 = 10 + 26 = 12
c1,2 is the sum of the products of the 1st row of A and the 2nd column of B:
c1,2 = 12 + 23 = 8
c1,3 is the sum of the products of the 1st row of A and the 3rd column of B:
c1,3 = 14 + 20 = 4
c1,4 is the sum of the products of the 1st row of A and the 4th column of B
c1,4 = 19 + 22 = 13

31

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

1 2
A=

7 3
1 6

B = 0 2 4 9

6 3 0 2

c1,1 is the sum of the products of the 1st row of A and the 1st column of B:
c1,1 = 10 + 26 = 12
c1,2 is the sum of the products of the 1st row of A and the 2nd column of B:
c1,2 = 12 + 23 = 8
c1,3 is the sum of the products of the 1st row of A and the 3rd column of B:
c1,3 = 14 + 20 = 4
c1,4 is the sum of the products of the 1st row of A and the 4th column of B:
c1,4 = 19 + 22 = 13
c2,1 is the sum of the products of the 2nd row of A and the 1st column of B
c2,1 = 70 + 36 = 18
32

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

1 2
A=

7 3
1 6

B = 0 2 4 9

6 3 0 2

c1,1 is the sum of the products of the 1st row of A and the 1st column of B:
c1,1 = 10 + 26 = 12
c1,2 is the sum of the products of the 1st row of A and the 2nd column of B:
c1,2 = 12 + 23 = 8
c1,3 is the sum of the products of the 1st row of A and the 3rd column of B:
c1,3 = 14 + 20 = 4
c1,4 is the sum of the products of the 1st row of A and the 4th column of B:
c1,4 = 19 + 22 = 13
c2,1 is the sum of the products of the 2nd row of A and the 1st column of B:
c2,1 = 70 + 36 = 18
c2,2 is the sum of the products of the 2nd row of A and the 2nd column of B
33
c2,2 = 72 + 33 = 23
and so on.

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

The full product matrix C is as follows:


12 8 4 13
1 20 2 4 9
C = AB = 7 3
= 18 23 28 69

6 3 0 2
36 20 4 21
1 6

34

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

The full product matrix C is as follows:


12 8 4 13
1 20 2 4 9
C = AB = 7 3
= 18 23 28 69

6 3 0 2
36 20 4 21
1 6

Note that in this case the product matrix AB exists, but the product matrix
BA does not exist, because the number of columns in B the number of
rows in A.

35

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

The full product matrix C is as follows:


12 8 4 13
1 20 2 4 9
C = AB = 7 3
= 18 23 28 69

6 3 0 2
36 20 4 21
1 6

Note that in this case the product matrix AB exists, but the product matrix
BA does not exist, because the number of columns in B the number of
rows in A.
Transposition
The transpose of an (RC) matrix A, written A', is the (CR) matrix
obtained by transposing (switching) the rows and columns of the matrix
A.
36

Essentials of matrix arithmetic (cont'd)


Multiplication (cont'd)

The full product matrix C is as follows:


12 8 4 13
1 20 2 4 9
C = AB = 7 3
= 18 23 28 69

6 3 0 2
36 20 4 21
1 6

Note that in this case the product matrix AB exists, but the product matrix
BA does not exist, because the number of columns in B the number of
rows in A.
Transposition
The transpose of an (RC) matrix A, written A', is the (CR) matrix
obtained by transposing (switching) the rows and columns of the matrix
A.
1 2
1 7 1

If A =
A' =

7
3

2 3 6
1 6

37

Essentials of matrix arithmetic (cont'd)


Rank

The rank of any matrix is the number of linearly independent rows or


columns. Loosely speaking, two rows are linearly independent if it is not
possible to express one of the rows as a multiple of another.

38

Essentials of matrix arithmetic (cont'd)


Rank (cont'd)
4 3
Suppose A =

1 2

3 6
and B =

2 4

0 0
and C =

0 0

rank(A) = 2, because row 2 cannot be expressed as a multiple of row 1 (or


vice versa). Therefore A has two linearly independent rows.

39

Essentials of matrix arithmetic (cont'd)


Rank (cont'd)
4 3
Suppose A =

1 2

3 6
and B =

2 4

0 0
and C =

0 0

rank(A) = 2, because row 2 cannot be expressed as a multiple of row 1 (or


vice versa). Therefore A has two linearly independent rows.
rank(B) = 1, because row 1 is 1.5 times row 2. Therefore rows 1 and 2 are
not linearly independent: B has only one linearly independent row.

40

Essentials of matrix arithmetic (cont'd)


Rank (cont'd)
4 3
Suppose A =

1 2

3 6
and B =

2 4

0 0
and C =

0 0

rank(A) = 2, because row 2 cannot be expressed as a multiple of row 1 (or


vice versa). Therefore A has two linearly independent rows.
rank(B) = 1, because row 1 is 1.5 times row 2. Therefore rows 1 and 2 are
not linearly independent: B has only one linearly independent row.
rank(C) = 0, because a square matrix containing zeros only has no
independent rows.

41

Essentials of matrix arithmetic (cont'd)


Rank (cont'd)
4 3
Suppose A =

1 2

3 6
and B =

2 4

0 0
and C =

0 0

rank(A) = 2, because row 2 cannot be expressed as a multiple of row 1 (or


vice versa). Therefore A has two linearly independent rows.
rank(B) = 1, because row 1 is 1.5 times row 2. Therefore rows 1 and 2 are
not linearly independent: B has only one linearly independent row.
rank(C) = 0, because a square matrix containing zeros only has no
independent rows.
A square matrix A with dimensions (mm) is said to have full rank if
rank(A) = m.

42

Essentials of matrix arithmetic (cont'd)


Matrix inversion

For any square matrix that has full rank, an inverse matrix can be defined
such that the product of the original matrix and the inverse matrix is an
identity matrix. In this case, the order in which the two matrices are
multiplied does not matter.

43

Essentials of matrix arithmetic (cont'd)


Matrix inversion

For any square matrix that has full rank, an inverse matrix can be defined
such that the product of the original matrix and the inverse matrix is an
identity matrix. In this case, the order in which the two matrices are
multiplied does not matter.
If the original matrix is A, the inverse matrix is denoted A1.

44

Essentials of matrix arithmetic (cont'd)


Matrix inversion

For any square matrix that has full rank, an inverse matrix can be defined
such that the product of the original matrix and the inverse matrix is an
identity matrix. In this case, the order in which the two matrices are
multiplied does not matter.
If the original matrix is A, the inverse matrix is denoted A1.
If A is (mm), AA1 = A1A = Im

45

Essentials of matrix arithmetic (cont'd)


Matrix inversion

For any square matrix that has full rank, an inverse matrix can be defined
such that the product of the original matrix and the inverse matrix is an
identity matrix. In this case, the order in which the two matrices are
multiplied does not matter.
If the original matrix is A, the inverse matrix is denoted A1.
If A is (mm), AA1 = A1A = Im
If A is (22), the formula for A1 is relatively simple.

46

Essentials of matrix arithmetic (cont'd)


Matrix inversion
For any square matrix that has full rank, an inverse matrix can be defined
such that the product of the original matrix and the inverse matrix is an
identity matrix. In this case, the order in which the two matrices are
multiplied does not matter.
If the original matrix is A, the inverse matrix is denoted A1.
If A is (mm), AA1 = A1A = Im
If A is (22), the formula for A1 is relatively simple.

a b

If A =
c d

1 d b

then A1 =
ad bc c a

47

Essentials of matrix arithmetic (cont'd)


Matrix inversion (cont'd)
4 3
Suppose A =

1 2

48

Essentials of matrix arithmetic (cont'd)


Matrix inversion (cont'd)
4 3
Suppose A =

1 2

A1

1
2 3
=

4 2 3 1 1 4

49

Essentials of matrix arithmetic (cont'd)


Matrix inversion (cont'd)
4 3
Suppose A =

1 2

A1

1
2 3
=

4 2 3 1 1 4
1 2 3
0.4 0.6
=

=
5 1 4
0.2 0.8

50

Essentials of matrix arithmetic (cont'd)


Matrix inversion (cont'd)
4 3
Suppose A =

1 2

A1

1
2 3
=

4 2 3 1 1 4
1 2 3
0.4 0.6
=

=
5 1 4
0.2 0.8

4 3 0.4 0.6
1 0 = I
Check: AA1 =
=

2
1 2 0.2 0.8
0 1

51

Essentials of matrix arithmetic (cont'd)


Matrix inversion (cont'd)
4 3
Suppose A =

1 2

A1

1
2 3
=

4 2 3 1 1 4
1 2 3
0.4 0.6
=

=
5 1 4
0.2 0.8

4 3 0.4 0.6
1 0 = I
Check: AA1 =
=

2
1 2 0.2 0.8
0 1

If A is (22) as above, the quantity (ad bc) is known as the determinant


of matrix A, also denoted det(A) or |A|.

52

Essentials of matrix arithmetic (cont'd)


Matrix inversion (cont'd)
The determinant can be interpreted as a summary measure of the
information contained in the matrix, in the form of a single numerical
value.

53

Essentials of matrix arithmetic (cont'd)


Matrix inversion (cont'd)
The determinant can be interpreted as a summary measure of the
information contained in the matrix, in the form of a single numerical
value.

If A is (mm) where m>2, the formulae for det(A) and A1 is more


complex. det(A) and A1 would not usually be calculated manually.

54

Essentials of matrix arithmetic (cont'd)


Matrix inversion (cont'd)
The determinant can be interpreted as a summary measure of the
information contained in the matrix, in the form of a single numerical
value.

If A is (mm) where m>2, the formulae for det(A) and A1 is more


complex. det(A) and A1 would not usually be calculated manually.
The inverse of an (mm) matrix A only exists if rank(A)=m.

55

Essentials of matrix arithmetic (cont'd)


Matrix inversion (cont'd)
The determinant can be interpreted as a summary measure of the
information contained in the matrix, in the form of a single numerical
value.

If A is (mm) where m>2, the formulae for det(A) and A1 is more


complex. det(A) and A1 would not usually be calculated manually.
The inverse of an (mm) matrix A only exists if rank(A)=m.
Trace and eigenvalues of a square matrix
Two other useful summary measures of the information contained in a
square matrix are the trace and the eigenvalues. The trace of any square
matrix is the sum of the main-diagonal elements (from top-left to bottomright).

56

Essentials of matrix arithmetic (cont'd)


Trace and eigenvalues of a square matrix (cont'd)

The eigenvalues of an (mm) square matrix A are the m solutions


(1,2, ... ,m) to the mth order polynomial in that is formed by the
expression det(AIm)=0 or |AIm|=0, listed in descending order of
absolute magnitude.

57

Essentials of matrix arithmetic (cont'd)


Trace and eigenvalues of a square matrix (cont'd)

The eigenvalues of an (mm) square matrix A are the m solutions


(1,2, ... ,m) to the mth order polynomial in that is formed by the
expression det(AIm)=0 or |AIm|=0, listed in descending order of
absolute magnitude.
4 3

For A =
1 2
3
A Im = 4 3 1 0 = 4 3 0 = 4

2
1 2
0 1 1 2 0 1

58

Essentials of matrix arithmetic (cont'd)


Trace and eigenvalues of a square matrix (cont'd)

The eigenvalues of an (mm) square matrix A are the m solutions


(1,2, ... ,m) to the mth order polynomial in that is formed by the
expression det(AIm)=0 or |AIm|=0, listed in descending order of
absolute magnitude.
4 3

For A =
1 2
3
A Im = 4 3 1 0 = 4 3 0 = 4

2
1 2
0 1 1 2 0 1
det(A Im) = (4 )(2 ) 3 1 = 8 6 + 2 3 = 2 6 + 5
59

Essentials of matrix arithmetic (cont'd)


Trace and eigenvalues of a square matrix (cont'd)

The eigenvalues of an (mm) square matrix A are the m solutions


(1,2, ... ,m) to the mth order polynomial in that is formed by the
expression det(AIm)=0 or |AIm|=0, listed in descending order of
absolute magnitude.
4 3

For A =
1 2
3
A Im = 4 3 1 0 = 4 3 0 = 4

2
1 2
0 1 1 2 0 1
det(A Im) = (4 )(2 ) 3 1 = 8 6 + 2 3 = 2 6 + 5
det(A Im) = 0

2 6 + 5 = 0
60

Essentials of matrix arithmetic (cont'd)


Trace and eigenvalues of a square matrix (cont'd)

The eigenvalues of an (mm) square matrix A are the m solutions


(1,2, ... ,m) to the mth order polynomial in that is formed by the
expression det(AIm)=0 or |AIm|=0, listed in descending order of
absolute magnitude.
4 3

For A =
1 2
3
A Im = 4 3 1 0 = 4 3 0 = 4

2
1 2
0 1 1 2 0 1
det(A Im) = (4 )(2 ) 3 1 = 8 6 + 2 3 = 2 6 + 5
det(A Im) = 0

2 6 + 5 = 0
( 5)( 1) = 0

61

Essentials of matrix arithmetic (cont'd)


Trace and eigenvalues of a square matrix (cont'd)

The eigenvalues of an (mm) square matrix A are the m solutions


(1,2, ... ,m) to the mth order polynomial in that is formed by the
expression det(AIm)=0 or |AIm|=0, listed in descending order of
absolute magnitude.
4 3

For A =
1 2
3
A Im = 4 3 1 0 = 4 3 0 = 4

2
1 2
0 1 1 2 0 1
det(A Im) = (4 )(2 ) 3 1 = 8 6 + 2 3 = 2 6 + 5
det(A Im) = 0

2 6 + 5 = 0
( 5)( 1) = 0

1 = 5 and 2 = 1 are the eigenvalues of the matrix A.

62

Essentials of matrix arithmetic (cont'd)


Why are eigenvalues useful?

There are important properties of eigenvalues, in particular:


1. The sum of the eigenvalues is the trace of the matrix
2. The product of the eigenvalues is the determinant
3. The number of non-zero eigenvalues is the rank
Thus, eigenvalues are important to understand whether a matrix
is of full rank, and whether all rows are linearly independent
(this is important, in particular, for the concept of cointegration)

63

You might also like