Chapter 5
Chapter 5
Chapter 5
Introduction
Dear colleague! The change in equilibrium states in response to a change in exogenous variables
leads to a type of analysis known as comparative static (equilibrium) analysis where as the
question of attainability and stability of equilibrium lies on the field of dynamic analysis. But this
unit will be emphasized on the nature of comparative static (equilibrium) and its application on
economic models (particularly on market and national income models) and its limitations.
Dear Colleague! First of all you should have good understanding about the meaning of
equilibrium in order to know the concept of comparative static analysis.
What is equilibrium? Discuss with your friends and try to answer this question. Write your
answer on rough paper.
---------------------------------------------------------------------------------------------------------------------
--------------------------------------------------------
Dear colleague! Have you answered this question? Ok, good. Try to relate your answer with the
following analysis.
1
Fritz Machlup, “ Equilibrium and Disequilibrium : Misplaced concreteness and Disguised politics,”
Economic Journal, March 1958, P.9 Quoted in Chaing A. C., Fundamental Methods of Mathematical
Economies, 3rd ed, p. 35.
120
Section 5.1 Nature of Comparative Static, Differentiation and its Application on
Comparative Static Analysis
---------------------------------------------------------------------------------------------------------------------
--------------------------------------------------------------------------------------------------
Dear colleague! Have you answered this question? Ok good try to relate your answer with the
following analysis.
As its name indicates the term comparative static is emphasized on the comparison of different
equilibrium states, which are related, with different set of values of parameters and exogenous
variables. First of all, it is usual to suppose a given initial equilibrium state so as to carry out this
comparison. For instance, this initial equilibrium will be indicated by a define price P and its
corresponding quantity Q in a closed–market model.
In the simple national income model a determinate income Y and a corresponding consumption
C will represent the initial equilibrium. This initial equilibrium will be disturbed provided that
there is a change in exogenous variable or certain parameters in the model. Consequently, several
endogenous variables have to be subjected to some adjustment. Supposing that some of the new
equilibrium state can be defined and attained, the problem faced by comparative static analysis is
comparing the new equilibrium with the old one.
We don't pay attention for the process of adjustments in comparative static analysis rather we
compare the old equilibrium with the new equilibrium state. In this analysis, we assume that the
new equilibrium has to be attainable as we did for the old one.
The nature of comparative static may be either qualitative or quantitative. The analysis will be
qualitative provided that our emphasis is only to determine the direction of the change. On the
other hand, quantitative analysis is concerned on determining the magnitude of the change in the
endogenous variables resulting from a given change in some parameters or exogenous variables
to the model. However, it is clear that we can get the direction of the change from the algebraic
sign of the quantitative solution. This means quantitative analysis often incorporate qualitative
analysis.
121
Dear colleague! How can we determine the rate by which the endogenous variable is changed due
to a change in the exogenous variable, i.e. the rate of change?
---------------------------------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------------------------
Have you answered this question? Ok, good. Try to relate your answer with the following
analysis.
Dear colleague! As we know the main problem in comparative static analysis is finding the rate
of change of the dependent variable with respect to a small change in the independent variables.
For instance, in a function
y f (x )
Where y is dependent variable
x is an independent variable
The rate of change of ' y' with respect to 'x' is denoted by
dy
dx
Dear colleague! I think you have learnt about the concept of differentiation and its rules in
quantitative method for economists I and in the first unit of this course. Therefore, go to these
sources and check what you have studied and discuss with your friends in order to avoid the
redundancy.
Have you checked and grasp the concept of differentiation and its rules? Ok. Good. By now,
therefore, we have the ability to solve certain problems we face in comparative static analysis
such as determining the equilibrium values of the endogenous variables when there is a change in
any of the exogenous variables or parameters. Let us apply the mathematical method of derivative
on the following economic models.
Market Model
Let us take a simple-one commodity market model to observe the application of techniques of
derivative to determine equilibrium quantity and equilibrium price. There are three variables in
this model as it is a one commodity model. These are:
quantity demanded of a good ( Qd )
quantity supplied of a good ( Qs )
price of a good ( P)
Dear colleague! It is clear that we should put assumptions considering the working of the market.
Assumptions
It is assumed that market equilibrium will be achieved when quantity demanded is equal
to quantity supplied, i.e. when the market is cleared.
Quantity demanded is supposed to be a decreasing linear function of price. This means,
as price decreases, quantity demanded of a good increase, citrus paribus and vice versa.
Quantity supplied is assumed to be an increasing function of price, other things being
equal. In this case no quantity will be supplied if the price is not greater than a certain
positive value.
The model includes:-
One equilibrium condition.
122
Two behavioral equations.
Algebraically, the model can be expressed as
Qd Q s
Qd 1 1 P
( 1, 1 0) ---------------------------- (1)
Qs 2 2 P ( 2 , 2 >0)
Equation 3 indicates the fact that the determinate value, i.e., equilibrium price, is expressed in
terms of the parameters. P is positive because we put the restriction that all the four parameters
are positive.
To got the equilibrium quantity of the market which corresponds to P we should substitute
equation (3) in to one of the equations of equation (2). Substituting it in the supply equation gives
us
1 2
Q = 2 2 ( )
1 2
1 2 2 2
Q = 2 +
1 2
2 ( 1 2 ) ( 1 2 2 2 )
Q=
1 2
2 1 2 2 1 2 2 2
=
1 2
1 2 2 1
Q = , 1 2 0 ------------------------ (4)
1 2
Similarly in equation (4) we have observed that equilibrium quantity is expressed in terms of the
parameters. We know that quantity never be negative. Thus, this situation requires that the
numerator ( 1 2 2 1 ) should be positive. This means 1 2 2 1 .
123
Dear colleague! Now we can observe the effect of a small change in one of the parameters either
on P or Q . To do this, we must partially differentiate each variable with respect to each of
the parameters. Through observing the sign of the partial derivative of the variables, for example
P , with respect to the parameters, we can point out the direction in which P will move when
the parameter changes. If the magnitude is known, it will constitute quantitative analysis. Let us
see the effect of change in the parameters on P ,
P 1
=
1 ( 1 2 )
P 0( 1 2 ) 1( 1 2 )
=
1 ( 1 2 ) 2
( 1 2 )
= ----------------------Quotient rule)
(1 2 ) 2
P 1
2 1 2
P P ( 1 2 )
= =
2 1 (1 2 ) 2
We have put a restriction that all parameters are positive in this model. As a result, we have the
ability to conclude that
P P
= 0
1 2
It indicates that as the parameter 1 increases, there will be an upward shift in the demand curve
which leads to an increase in equilibrium price, i.e. p , given the supply curve. However, the
P
slope of the demand curve is the same. Similarly, 0 shows that an increase in the
2
parameter 2 leads to a parallel downward shift in the supply curve which in turn results in an
increase in equilibrium price, i.e. p given the demand curve. In this case the slope of the supply
curve is constant.
P P
In addition to this = < 0.
1 2
P
< 0 reflects the fact that as 1 increases, the slope of the demand curve will be steeper
1
and it will rotate to the left (inward) at the point of the intercept. Therefore, the equilibrium price
P
decreases given the supply curve. Similarly, <0 shows that as the slope of the supply curve
2
increases, the original supply curve will be rotated to the left about the point of it's intercept
which decreases the equilibrium prices given the original demand curve. It is in line with the
negative sign of the above derivative.
Show the effect of change in these parameters on equilibrium quantity and interpreter the
result. --------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------------------
124
Dear colleague! Have you done the above exercise? Ok that is good .Now try to relate your
answer with the following analysis. We know that
1 2 2 1
Q =
1 2
From this equation one can see the effect of change in each of the parameters using the technique
of partial differentiation as follows.
Q 2
>0
1 1 2
Q 2 ( 1 2 ) 1( 1 2 2 1 )
------------- (Quotient rule)
1 ( 1 2 ) 2
2 1 2 2 1 2 2 1
=
( 1 2 ) 2
Q ( 2 2 1 2 )
<0
1 (1 2 ) 2
Q 1
<0
2 1 2
Q 1 ( 1 2 ) 1( 1 2 2 1 )
2 (1 2 ) 2
1 1 1 2 1 2 2 1
=
(1 2 ) 2
Q 2 1
1 1 >0
2 ( 1 2 ) 2
125
P represents the equilibrium price (market price) of the commodity. Substituting the value of
equilibrium price either in the demand or the supply equation gives us the equilibrium quantity.
When we substitute it in the demand equation
29
Q = 24- 2( )
9
58
Q 24
9
216 58
Q
9
158
Q
9
What happens to the equilibrium price P when 1 increases from 24 to 25 other things being
equal?
It becomes 25-2p = -5+7p
25+5 = 9p
30= 9p
30
p =
9
29 30
The equilibrium price increases from to when the parameter 1 increases from 24 to
9 9
25 units.
What happens to the equilibrium price, P when the parameter 2 increases from 7 to 8?
Dear colleague! Please, try to observe the effect of change in the remaining parameters on
equilibrium price and equilibrium quantity of the commodity.
Given
Qd = 51 - 3 P
Qs 6P - 10
As it is shown above, the equilibrium condition of the national income is reflected in the first
equation. But the second and the third equations indicate the way to determine consumption (C)
and taxes (T) respectively.
126
In this model, a , b, c and d are parameters. Their values are restricted and they are described as
follows.
The parameter a represents autonomous consumption, i.e. consumption without income. It is
positive. It indicates that the consumer may consume from past saving, family remittance or
borrowing with zero income.
b shows marginal propensity to consume (MPC). It is the rate at which consumption changes
when disposable income changes by one.
C
Algebraically, MPC = b =
(Y T )
Its value lies in between zero and one. MPC = b= 0.6 means that as disposable income increases
(decreases) by one unit consumption increases (decreases) by 0.6 units.
The parameter c represents tax revenue without income. It is clear that government can have
positive tax revenue from the tax base other than income. As a result c is positive.
The exogenous variables in the model namely investment (I 0 ) and government expenditure (
G0 ) are supposed to be nonnegative. Similar to the parameters, these exogenous variables are
assumed to be independent of one another.
In this model, it is possible to solve the equilibrium income in such a way that substituting the
third equation of the model in to the second one and then substituting the resulting equation in to
the first equation.
Thus, the solution value of Y (the equilibrium national income, Y ) in reduced form is
a bc I 0 G0
Y =
1 b bd --------------------- (6)
Dear colleague! From this equation we can get six comparative static derivatives. What
are they? Please, partially differentiate Y with respect to a, b, c ,d, I 0 and G 0 , and thereby
answer this
question .----------------------------------------------------------------------------------------------------------
--------------------------------------------------------------------------
Have you answered this question? Ok, good. Now try to relate your answer with the following
analysis and correct it if there you have done a mistake.
127
1
=
a 1 b bd
2(1 d ) c(1 b bd )
=
b (1 b bd ) 2
b
=
c 1 b bd
b(a bc I 0 G0 ) bY
= =
d 1 b bd ) 2
1 b bd
Y 1
(Investment multiplier)
I 0 1 b bd
Y 1
(Government expenditure multiplier)
G0 1 b bd
From these partial derivatives, the following three are very crucial from policy point of view. 2
b
= <0 -------------------------------------------- (7)
c 1 b bd
Y b(a bc I 0 G0 ) bY
0 ------------------- (8)
d (1 b bd ) 2
1 b bd
1
= > 0 ----------------------------------------------- (9)
G0 1 b bd
The partial derivative shown in equation (7) above indicates the non- income tax multiplier. This
equation explains how the change in government revenue from non- income tax sources affects
equilibrium income. It is negative in the model as the denominator is positive and the numerator
is negative as shown above.
.
The partial derivative indicated in equation (8) represents an income-tax rate multiplier which is
negative for any positive equilibrium income. As tax rate increases, then equilibrium income
decreases.
Finally, equation (9) shows the partial derivative of equilibrium income with respect to
government expenditure. It is referred to as government expenditure multiplier. The value is
positive which indicates that increase (decrease) in government expenditure increases (decreases)
equilibrium national income.
Example
Given the following national income mode,
Y = C+ I 0 + G 0
C = 50 + 0.6 (Y-T) --------------------------------- (10)
T = 30 + 0.3Y
I 0 = 20
G 0 = 15
Find , C and T
Solution
2
Chiang A.C., fundamental Methods of Mathematical Economics, 3rd ed. P.182
128
From the given model, substituting the third equation in to the second equation given
C = 50+ 0. 6 [Y- (30 + 0.3Y)]
= 50 + 0.6 (Y-30-0.3Y)
= 50+ 0.6 (0.7Y -30)
= 50-18+0.42Y
C = 32 + 0.42Y------------------------------------------------- (11)
Substituting equation (11) in to the first equation of equation 10, and the value of I 0 and G 0 in
this equation, we get
Y= 32 +0.42Y+20+15
Y = 67 + 0.42 Y
Y-0.42Y = 67
0.58Y = 67
67
Y
0.58
Y 115 .5 Equilibrium national income
Therefore,
C = 32 + 0.42 (115.5)
= 32+48.51
C = 80.51 or
We can get the value of C by using its reduced form
a bc b(1 d (1 d )( I 0 G0 )
C =
1 b bd
In our example a = 50, b = 0.6, c = 30, d = 0.3
I 0 = 20, G 0 = 15
50 0.6(30) 0.6(1 0.3)( 20 15)
Thus, C =
1 0.6 (0.6)(0.3)
50 18 (0.42)(35)
=
0.4 0.18
C 80.52
Similarly,
T = 30 + 0.3 Y
= 30+0.3(115.5)
= 30+34.65
T = 64.65
Or
(c bc) d (a I 0 G0 )
T =
1 b bd
30 (0.6)(30) 0.3(50 20 15)
=
1 0.6 (0.6)(0.3)
30 18 0.3(85)
=
0.58
12 25.5
=
0.58
T = 64.65
129
What happens to equilibrium national income when government expenditure increases by 5
units? What about consumption and taxes at this point?
1 5
Y G ( ) 8.62
0.58 0.58
Y 115 .5 8.62 124.12
It shows that an increase in government increases equilibrium national income.
Dear colleague! Try to determine C and T at this new level of equilibrium income.
Dear Colleague! I think you have studied about the concept of input - output model in your
Quantitative method for economists II course. Therefore, please try to revise what you studied
from these sources in order to understand the forthcoming discussion properly.
As you know the solution of the open input - output model is described in the form of matrix
equation as follows.
X ( I A) 1 d
Where X represents the output vector in the model
I - is an identity Matrix
A- Technical coefficient Matrix
(I-A) – is Leontief Matrix (technology Matrix)
d- is the final demand vector
Representing the inverse of the technology Matrix (I - A) 1 by C= [ C ij ], the solution of the a
simple three industry economy can be described as
X =Cd
or in matrix form
Partially differentiating this system of three equations gives us a total of nine comparative- static
derivatives. These derivatives are
130
X 1 X 1 X 1
= C11 = C12 = C13
d1 d 2 d 3
X 2 X 2 X 2 C
= C 21 = C 22 = 23
d1 d 2 d 3
X 3 X 3 X 3
= C 31 = C 32 = C 33
d 1 d 2 d 3
These nine derivatives can be summarized in matrix form as shown below.
It is a compact way of representing all of the comparative static derivatives of our open input -
output model.
It is clear that comparative static derivatives of the input – output model are beneficial for
economic planning as they offer a means of solving the problem of changing the output target of
the industries due to a change in the final demand of consumers.
Numerical Example
Suppose a simple economy has three inter - industrial sector such as agriculture, Manufacturing,
and service, and one final demand sector, i.e., Household sector. It is given that a birr wroth of
agricultural product requires 20 cents worth of agriculture, manufacturing and service sectors
product each respectively as an input, a birr worth of manufacturing product needs 5 cents, 20
cents and 10 cents worth of agriculture, manufacturing and service sector's output respectively as
input and that of a birr worth of service sector's product requires 10 cents, 20 cents and 40 cents
worth of agricultural, manufacturing and service sector's output respectively as inputs.
Based on the given above determine the technical coefficient matrix in the order of agriculture,
manufacturing and service sector.
131
0.8 - 0.05 -0.1
(I - A) = -0.2 0.8 -0.2
-0.2 -0.1 0.6
With this technology matrix, the open input - out put system can be expressed in the form (I- A)
X = d as follows
By inverting the 3x3 technology matrix (I - A), the solution of the above system can be
determined as
X = (I - A) 1 d
To put it in matrix form, we ought to find the determinant of the technical coefficient matrix and
the inverse of the technology matrix, i.e., (I -A) 1 .
1
(I -A) 1 = adj.( I A)
/ I A/
C 11 C 12 C 13
C= C 21 C 22 C 23
C 31 C 32 C 33
Therefore,
C 11 = (-1) 2 0.8 - 0.2 , C 12 ( -1 ) 3 - 0.2 0.2
- 0.1 0.6 -0.2 0.6
132
C 21 = (- 1) 3 - 0.05 - 0.1 C 22 = (-1) 4 0.8 - 0.1 , C 23 = ( -1) 5 0.8 -
0.05
- 0.1 0.6 - 0.2 0.6 - 0.2 - 0.1
And the ad joint matrix is the transpose of the cofactor matrix. As a result, the ad joint matrix of
the technology matrix is
We know that
1
(I - A) 1 = .adj ( I A)
( I A)
1
Therefore, (I - A) 1 = 0.46 0.04 0.09
0.342
0.16 0.46 0.18
0.18 0.09 0.63
133
X 1 = 1.35d 1 + 0.12 d 2 + 0.26 d 3
X 2 = 0.42d 1 + 1.35 d 2 + 0.53 d 3
X 3 = 0.53 d 1 + 0.26d 2 + 1.83 d 3
Dear colleague! Determine the nine comparative static derivatives based on the above
system of equations
---------------------------------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------
If the final demand for agriculture, manufacturing and service sector's output of this simple
economy is valued 40 Birr, 80 Birr and 140 Birr respectively, find the gross output (equilibrium
output level) of each sector for the given final demand. Putting it in matrix form
134
---------------------------------------------------------------------------------------------------------------------
-------------------------------------------------------------------------
4. Given the total cost function C= Q 3 -5Q 2 + 14 Q+75, write out the variable cost function.
Find the derivative of the variable cost function and interpret the economic meaning of that
derivative-------------------------------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------------------------
5. Given the demand function Q d = 30 - 2 p
the supply function Q s = - 6+5P
Find equilibrium price P and quantity Q
---------------------------------------------------------------------------------------------------------------------
-----------------------------------------------------------
6. Given the national income model
Y= C+ I 0 + G 0
C 25 0.6Y
I 0 = 16
G 0 = 14
Find the equilibrium national income Y and consumption C
---------------------------------------------------------------------------------------------------------------------
---------------------------------------
Dear colleague! Have you answered these questions? If your answer is yes, go to the next
section. If no, please re-read this section and try to answer these questions.
Dear colleague! Do you remember what you have studied about the determinant of a
square matrix and its benefits in determining the existence of unique solution or linear
dependency of functions? How can we decide whether there exists a unique solution for a system
135
of linear equations or not?
---------------------------------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------------------------------
---------------------------------------------------------------------
Have you answered this question? If your answer is no, re-read what you have learnt in
Quantitative method for economists II and try to answer this question. If yes, very good. Go to
the next analysis and try to relate your answer with it.
A system of equations can have unique solution if there is the same number of equations as
unknowns. However, it is not a sufficient condition .In addition to this; the equations are to be
consistent and functionally independent. There exists a systematic method for testing whether
there is a unique solution or not for a system.
For simultaneous linear equation models, we can use the concept of determinant of a matrix.
Given a linear equation system BX d where B is an n x n coefficient matrix the
determinant of the coefficient matrix B, i.e. B 0 , implies that
there is row ( column ) independence in matrix B.
B 1 exists and
a unique solution x = B 1 d exists
This means, the determinant of the matrix enables us to decide whether there is unique solution in
a system or not. If the determinant of a matrix vanishes, i.e. zero, a system of the equation does
not have a unique solution. If the determinant of a matrix is different from zero, these systems of
the equations can have a unique solution. However, the determinant criterion does not give the
algebraic sign of the solution value although we know that there is unique solution.
Dear colleague! Let us see the following example in order to make this the analysis clear.
Example
Given the systems of sum ' tenuous linear equations
7x 1 - 3x 2 -3x 3 = 7
2x 1 + 4x 2 + x 3 = 0
0 x 1 -2x 2 -x 3 = 2
Determine whether there is a unique solution or not. To do this, we should use the determinant of
the coefficient matrix of the system.
7 -3 - 3
A= 2 4 1 =-80
0 -2 1
Therefore, the equation system possesses unique solution because the determinate does not
vanish.
However, the technique of partial derivative and a special type of determinant known as Jacobian
determinant is needed in order to identify the existence of a unique solution provided that the
system of equations are non - linear models. This means, this type of determinant enables us to
136
determine whether there is functional dependency among a set of n - functions in n- variables or
not.
If the value of the Jacobian determinant is zero, for all values of x 1 ..... x n , then the functions
are (linearly or non - linearly) dependent of one another. In this case, there is no unique solution.
On the other hand, the functions are said to independent and there exists a unique solution if the
value of the Jacobian determinant is different from zero.
Dear colleague, for the sake of understanding this concept, let us consider the following
examples.
Example
y1 y1
x x 2
J= 1
y 2 y 2
x1 x 2
6x 1 4x2
J = 5 0
= 6x 1 (0) - 5 (4x 2 )
= 0 - 20 x 2
J 20 x 2 0
Therefore, the above two functions are functionally independent of each others. Consequently,
there is unique solution for this system of equations.
137
2. Given
y1 3 x12 x 2
y 2 9 x14 6 x12 ( x 2 4) x 2 ( x 2 8) 12
If we partially differentiate these two functions, we get
y1 y1
6x 1 , 1
x1 x 2
y 2 y 2
36 x13 12 x1 x 2 48 x1 , 6 x12 2 x 2 8
x1 x 2
Putting these partial derivatives in the form of Jacobian matrix
6 x1 1
J=
36 x1 12 x1 x 2 48 x1
3
6 x 2 x 2 8
2
1
6x 1 1
J = 36x 13 + 12x 1 x 2 +48x 1 6x 12 + 2x 2 + 8
Therefore, these two functions are dependent of each other. As a result, there is no unique
solution for the system of equations.
Test the existence of functional dependence between these two functions and the
existence of unique solution using the Jacobian determinant.
y 1 = 2x 1 + 3x 2
y 2 = 4x 12 +12x 1 x 2 + 9x 22
Dear colleague! Do you remember the necessary and sufficient conditions of a function
with two or more independent variables to achieve its optimum value?
What are these conditions for a function? Z = ƒ (x, y)?
---------------------------------------------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------------------------------
Have you answered this question? Ok, Good. Try to relate your answer with the following
analysis.
Given the objective function with two independent variables Z = ƒ (x, y), to get a maximum or
minimum of Z, it is necessary that the total differential of this function to be zero ( dZ 0 ) for
an arbitrary values of dx and dy where both are different from zero
dZ f x dx f y dy
138
Since dx and dy are a very small change in the independent variables ( dx 0 and dy 0 ), the
only situation which leads to a zero dZ is that
ƒx =ƒ y =0
This means, the necessary condition for a function to achieve its minimum or maximum value is
that both of the first order partial derivatives of the function must be equal to zero.
The sufficient condition to have the extremum value of the function is related with the positive
and negative definiteness of the second - order differential of the given function, d 2 z.
Example
139
2x+0=0
x=0
Thus, the critical points are at x = 0 and y = 0 .Now let us see the second order partial derivates at
the critical point. ƒ xx = 2, ƒ yy = 4, ƒ xy = ƒ yx = 1. They are independent of the critical
points. Arranging the second order partial derivates in the Hessian matrix
ƒ xx ƒ xy 2 1
H= =
ƒ yx ƒ yy 1 4
H = 2 1
1 4 = 2 x 4 - (1) 2 =7>0
( f 1 dx1 f 2 dx 2 f 3 dx3 )dx1 ( f1 dx1 f 2 dx 2 f 3 dx3 )dx 2 ( f 1 dx1 f 2 dx 2 f 3 dx3 )dx3
x1 x 2 x3
2
= ƒ 11 dx1 f 12 dx1 dx2 f 13 dx1 dx3 f 21dx1 dx2 f 22 dx22 f 23 dx2 dx3
f 31 dx3 dx1 f 32 dx3 dx 2 f 33 dx32
In order to determine the positive and / or negative definiteness of d 2 Z we should construct the
Hessian matrix and then its determinant based on the second order partial derivatives of the
function as follows.
140
f 11 f 12 f 13
H f 21 f 22 f 23
f 31 f 32 f 33
The successive principal minors of this determinant are
H 1 f 11 H2 f 11
f 12 f 11 f 12
f 13
f 21 f 22 , and H 3 f 21 f 22 f 23 H
f 31 f 32 f 33
The second - order sufficient condition for an extremum of the given function can be described as
Z is at its maximum point if and only if H 1 0, H 2 0, H 3 0 (in this case d 2 Z is
negative definite).
Z is at its minimum if and only if H 1 0, H 2 0, H 3 0 ( d 2 Z is positive definite).
Example
Find the extremum values of
Z = x 1 3 x 2 3 x1 x 2 4 x 2 x3 6 x3
2 2 2
2 -3 0
-3 6 4 = 2(36-16) + 3(-36) = 68 0
0 4 12
It is different from zero which implies that all of the equations are independent to each other. As a
result, there exists a single solution
x1 x 2 x3 0
From equation (14), we got
2x 1 = 3x 2
3
x1 x2
2
141
4x 2 = - 12x 3
1
x3 x2
3
27 x 2 36 x 2 8 x 2
0
6
35 x 2 36 x 2 0
6
35 x 2 36 x 2 0
x2 0
3
x1 ( 0) 0
2
This means,
1
x3 ( 0) 0
3
Thus the critical values are x1 x 2 x3 0 .
Dear colleague! As you know the sufficient condition is determined by using the Hessian
determinant.
ƒ 11 f 12 f 13 2 -3 0
H = ƒ 21 ƒ 22 ƒ 23 = -3 6 4
ƒ 31 ƒ 32 ƒ 33 0 4 12
H 1 = 2, H2 = 2 -3 = 12 - 9 = 3
-3 6
Dear colleague! As we have seen all of the principal minors are greater than zero. This implies
that d 2 Z is positive definite. Thus, the function achieves its minimum value at x 1 = x 2 = x 3 =
0 and the minimum value is
Z = 0 2 3(0) 2 3(0)(0) 4(0)(0) 6(0) 2
142
Z=0
For functions of n-choice variables such as
Z f ( x1 , x 2 , x3 , x 4 ,..x n )
The necessary condition for the extremum of this function is that all of the n-first order partial
derivatives must be equal to zero.
The second order condition can be described by using the sign of Hessian determinant. as we
have discussed in unit three.
Dear colleague! We have discussed the use of the Hessian determinant in order to expresses the
second order sufficient condition of achieving free extremum. We can also use a determinant
form of expressing this condition for the problem of constrained optimization. In this case, we
will apply what is known as the bordered Hessian determinant, H rather than the Hessian
determinant, H .The bar on the top of the bordered Hessian indicates the border.
Given the function
Z f ( x, y ) Subject to g ( x, y ) c , the bordered Hessian determinant is
0 gx gy
H = gx Z xx Z xy
gy Z yx Z yy
Example
Suppose a consumer uses two goods with utility function U xy x subject to the budget
constraint 6x +2y =110. Determine the amount of x and y with maximize utility and check the
second order condition using the bordered Hessian determinant.
Solution
L xy x 110 6 x 2 y .....................................................................(1)
143
0 1 6 x 1
1
6
0
2
2
0
y = 0
110
0 1 1
A3 1 0 0
=112
6 2 110
A1 224 A2 648
Therefore, x , y 27
A 24 A 24
A3 112
A 24
Taking the second partial of L with respect to x and y and the first order partial derivative of the
constraint function with respect to x and y, we can form Hessian bordered as follows.
L xx 0, L yy 0, L xy L yx 1 (Young's theorem)
g x 6, gy 2
0 6 2
Thus, H 6
2
0
1
1 6( 2) 2(6) 24 0
0
b) Y1 5 x1 3 x 2
Y2 2 x1 4 x 2
144
---------------------------------------------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------------------------------
3. If the total cost function of the firm is C 5Q12 2Q1Q2 3Q22 800 and a production
quota represented by Q1 Q2 39 is imposed on a firm.
Minimize C 5Q12 2Q1Q2 3Q22 800
Subject to Q1 Q2 39
Check the second order sufficient condition using the bordered Hessian determinant.
---------------------------------------------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------------------------------
4. Suppose the demand functions of the product sold in two separate markets are given as
Q1 24 0.2 P1 ; Q2 10 0.05P2
C 35 40Q, where , Q Q1 Q2
Find the out put levels which maximize profit and check the second order sufficient condition
using the Hessian
determinant-------------------------------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------
Dear colleague! Have you done these questions? If your answer in no reread this section and try
to do them. If yes, go to the next section..
Dear colleague! You remember that we have expressed the equilibrium values of the endogenous
variables of the model explicitly in terms of exogenous variables and parameters when we discuss
section 1 of this unit. In that case, we have applied a simple method of partial derivative to get
comparative static information. However, when the model incorporates functions which are
expressed in general form; explicit reduced form solution cannot be obtained. In this case, we
must find the comparative static derivatives directly from the general function model using the
concepts of total differentials, total derivatives and implicit function rule. Now let us first discuss
the market model.
Market Model
145
Let us take a one commodity market in which the quantity demanded for a product dependent on
the price of a product and consumer’s income which is determined exogenously. But quantity
supplied of a product depends only on price of a product. Therefore, at the point of market
equilibrium,
Qd Q s
Qd Qd
Where Qd f d P, y 0 0, 0 ........................(1)
P y 0
Qs
Qs f s ( P) 0
p
Assuming that the demand function as well as the supply function is differentiable, the restriction
Qd P 0 shows that the demand function is a decreasing function of price which meets the
law of demand and Qd Y0 0 indicates that demand is an increasing function of consumer’s
income. In this case the good is normal good. Similarly, Qs P 0 shows that the supply
function is an increasing function of price which satisfies the law of supply.
Dear colleague! As you know market equilibrium is determined by the interaction of market
demand and market supply, other things being equal. But a change in income shifts the market
demand which affects the equilibrium condition. The same will happen in our case. This means,
change in Y0 will disturb the equilibrium through the demand function. Therefore, let us see
how a change in Y0 will affect the equilibrium condition of the model as Y0 is the only
exogenous variable.
146
Now an equilibrium situation explained in equation (2) can be considered as an identity near the
equilibrium solution. As a result the equilibrium identity is expressed as
f d P, Y0 f s P 0.......................................................(4)
Now we can determine the comparative static derivative of d P dY using implicit function rule
0
as follows.
F Q
dp d
Y0 Y0
dY F
Qd
0................................(5) ,
0 dQs / d p
p p
Where Qd / p represents Qd p evaluated at the initial equilibrium. The same
interpretation is applied on Qs p .
Qd y 0 has also to be evaluated at the point of equilibrium. d p dy 0 is positive shows
that an increase in consumers income increases equilibrium price and a decrease in income
decreases equilibrium price.
Dear Colleague! Until now we have discussed the effect of change in consumer’s income, Y0 on
equilibrium price. Do you think that we can determine its effect on equilibrium quantity Q ?
---------------------------------------------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------------------------------
The answer is yes because at the point of equilibrium Q f s p and p f y 0 , therefore,
using the chain rule of differentiation,
dQ dQs d P dQs
dY 0 as is positive ………………………(6)
0 d P dY0 dP
Thus, change in consumer’s income affects equilibrium quantity Q, directly.
Equations (5) and (6) show the fact that a right ward shift of the demand curve (an increase in the
demand for a product) increases both equilibrium price and quantity of a product.
Simultaneous equation Approach
Dear colleague! In the above discussion we have determined the effect of change in consumer’s
income on equilibrium price and equilibrium quantity separately. Can we find them at the same
time?
---------------------------------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------------------
The answer is yes. To see it, let us formulate a two equation system since the endogenous
variables are two. At the point of equilibrium, we know that
Q Qd Qs Our market model can be formulated as
F 1 P, Q, y 0 f d P, Y0 Q 0
F 2 P, Q, y 0 f s P Q 0.......................................................(7)
1) F 1 and F 2 Should be differentiable as both the demand and supply functions are
differentiable.
147
2) The endogenous variable Jacobian should be different from zero, i.e,
F 1 F 1 Qd
1
P Q P dQs Qd
J 0.......... .......... .......... .(8)
F 2
F 2
dQs dP P
1
P Q dP
P f Y0 Q f Y0 ...................................................................(9)
and
These two functions are differentiable with respect to Y0 . We will have the following identities
near the equilibrium state.
f d P, Y0 Q 0..................................................................................(10)
fs P Q 0
dQ
Now we can determine d P dY and dY at the same time. Applying the implicit
0 0
Now we can apply the Cramer’s rule and there by can determine the solution as follows.
Qd
1
dP Y0 Qd
dY 0 1
y 0
........................................................(11)
0
J J
Qd Qd
dQ P Y0 dS Qd
dY dS
d P Y0
0
0
dP J
J
Where all the derivatives of demand and supply functions are evaluated at the initial equilibrium.
This result is similar to that of equation (5) and (6).
Dear Colleague! As you remember we have discussed a specific national income model which
includes consumption, government expenditure and investment in the first section of this unit.
Now let us reduce government expenditures and taxes for the purpose of variety and add the
foreign sector in to the model. In addition to this, let us incorporate the money market with the
goods market.
Goods market
In particular; it is supposed that the goods market has the following functions.
148
Investment is a function of interest rate,
I f (i ) , dI di 0
Saving s is a function of national income Y, and interest rate i,
S f (Y , i) (0 S y 1, S i 0)
Where S y S and S i S
Y i
Import (M) is also a function of national income Y,
dM
M f (Y ), 0 1
dY
Export X 0 is an exogenous variable.
X X0
Money market
Dear colleague! All of the above functions are supposed to be differentiable. In this model
equilibrium is achieved when the goods market and the money market are in equilibrium at the
same time as shown below.
f (i ) X 0 f S Y , i f Y (12)
f L Y , i M so
In this system the conditions for the implicitly function theorem are fulfilled. These are
1. F1 and F2 are differentiable as all ingredients functions are differentiable
2. The endogenous - variable Jacobian is different from zero, evaluated every where.
F 1 F 1 S dM dI S
Y i
Y dY di i
J
F 2 F 2 = M d M d
0. (13)
Y i Y i
149
Although it is difficult to solve Y and i explicitly, equation 121 can be expressed as a pair
of identities near the equilibrium state as
f (i ) X 0 f s (Y , i ) f (Y ) 0..........................15
f L (Y , i ) M so 0
We can obtain four comparative static derivatives from equation (15). Two of them are resulted
from change in X 0 and the remaining is emanated from change in money supply ( M S 0 ) .Now
let us determine the comparative static derivatives resulted from change in export holding money
supply constant. In this case dX o is a disequilibrating factor. Taking the total differential of
each identity in equation 15 with respect to X o , and dividing through by dX o , we get the
matrix equation
S dM dI S
1
Y dY di i
Y
smd md
X 0
=
0
y i
i
X
o
…………………..(16)
Using the Cramer's rule the solutions are
dI S
1
Y di i
X
o M d
0
i M d
0
i
J J ----------------17)
S dM
i Y
dY
1 M d
X
M d 0
o
Y
0 i
J J
Where all of the above derivatives are evaluated at Y Y and i i .
Equation (17) shows that an increase in export increases both equilibrium interest rate and
equilibrium income.
Dear colleague! By now you should determine the comparative static derivatives resulted from
change in money supply, i.e. dM so , using the same procedure.
We have discussed that it is concerned with the analysis of different equilibrium points.
Comparative static's is helpful in finding how the disequilibrating change in a parameter or
exogenous variable will affect the equilibrium state of the model. However, comparative static's
has the following shortcomings.
It neglects the process of adjustment from the old equilibrium to the new equilibrium.
It ignores the time element contained in the adjustment process.
It assumes that the new equilibrium can be defined and attained after a disequilibrating
change in the exogenous variable or a parameter.
150
Dear collogue! By now you have completed that last section of this Unit. Therefore, try to do the
following question to evaluate how you have understood the main point in this section.
5. What are the conditions that should be satisfied to differentiate the given function
f x, y 0, in the form of?
dy
dx,
Check List
Write '' in side the box which corresponds to the question that you can solve easily
Dear colleague! Is there any box in which you don't tick? If your answer is yes, please, reread this
unit and try to answer that question. If your answer is no, go to the next unit.
Unit summary
151
The change in equilibrium state in response to a change in exogenous variables or parameters
leads us to a type of analysis which is refereed to as comparative static analysis. Comparative
static analysis is emphasized on comparison of different equilibrium states. Comparative static
analysis may be either quantitative or qualitative in nature.
Jacobian determinant is a special type of determinate which enables us to determine whether there
is functional dependence among a set of n-functions in n-variables or not. We can also identify
the existence of unique solutions of the system of linear equations based on this determinant. The
zero value of this determinant shows that there is functional dependence among functions in a
system and there is no unique solution.
A Hessian determinant is a type of determinant composed of all the second order partial
derivatives with the second order direct partials lie on the principal diagonal. Observing the sign
of this determinant, we can check the second -order sufficient conditions of a function to achieve
it's free optimum value. However, the bordered Hessian determinant is used to determine second
order sufficient condition of the problem of constrained optimization.
When the equilibrium value of the endogenous variable is expressed in terms of the exogenous
variable and a parameter, we can apply a simple method of partial derivative to get comparative
static information where as if the functions are presented in a general form, we have to find the
comparative static derivatives directly from the general function models using the method total
differential, total derivatives and implicit function rule.
Important Points
Equilibrium
Comparative Static analysis
Parameters
Market model
National income model
Input out put model
Jacobian determinant
Hessian determinant
Total differential
Implicit differentiation
Bordered Hessian
dQ
36 138
5) P ,Q
7 7
6) Y 137.5 C 107. 5
152
0 1 1
3) Q1 13, Q2 26, 182 H 1 10 2 , H 2 56 0; H 3 H 12 0
1 2 6
10 0
4) Q1 8, Q2 4 H
0 40
400 0, H 1 10 0 The second order condition is
satisfied.
153