You are on page 1of 65

1.

Introduction

1.1 The Nature of Mathematical Economics


Since economics emerged as a subject of study, mathematics has played an important role in both the
exploration and exposition of economic ideas.
Mathematical economics refers to the application of mathematics to the purely theoretical aspects of
economic analysis. It enables us to express economic theory in mathematical form. Mathematical
economics is not a distinct branch of economics. Rather, it is an approach to economic analysis in which
economists use mathematical symbols and varieties of mathematical techniques (geometrical methods,
matrix algebra, differential and integral calculus, differential and difference equations, etc.) in the process
of reasoning.
Since mathematical economics is merely an approach to economic analysis, it is not fundamentally
different from the non-mathematical approach to economic analysis.
The purpose of any theoretical analysis, regardless of the approach, is always to derive a set of
conclusions or theorems from a given set of assumptions or postulates via a process of reasoning.
The major difference between mathematical economics and literary economics is:
 In the former, the assumptions and conclusions are stated in mathematical symbols and equations
rather than in words and sentences.
 In place of literary logic, mathematical economics uses mathematical theorems in the reasoning
process.
Mathematical approach has the following advantages:
 The language used is concise and precise.
 It forces us to describe or state explicitly all our assumptions as a prerequisite to use the
mathematical theorems. So, it keeps us from the pitfall of an unintentional adoption of unwanted
implicit assumptions.
 There exists a wealth of mathematical theorems at our service.
 It allows us to treat the general n-variable case. Mathematical approach allows us to analyze more
than three variables case economic relationships, while it is impossible to deal with more than three
variables cases using geometric approach. Similarly, the literary approach lacks preciseness.

1.2 Review of differential and integral calculus

1.2.1 Rules of Differentiation


Given the function f ( x) and other functions such as g( x ) and h(x ) which are differentiable, then the
following are rules of differentiation:

1
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
1. Constant-function rule
Given f ( x )=k , where k is any constant number
d f (x )
f ' ( x )= =0
dx
Example: f ( x )=5
'
f ( x )=0
2. Power -function rule
Given f ( x )=x n , x ≠ 0 and n is any real number

f ' ( x )=n x n−1


Example: f ( x )=x 7
' 6
f ( x )=7 x
3. The generalized power-function rule
When a multiplicative constantk appears in the power function, so that f ( x )=k x n .
' n−1
f ( x )=kn x
Example: f ( x )=4 x 3
' 2
f ( x )=12 x
f ( x )=(g( x ))n , f ' ( x )=n .( g(x))n−1 . g '(x )
4
E.g. f ( x )=(3 x 2+2 x+ 1)5 , f ' ( x ) =5∗( 3 x 2 +2 x +1 ) ∗(6 x+ 2)
4. Linear function rule
Given a function f ( x )=ax+ b, where a and b are constants, then
'
f ( x )=a
Example: f ( x )=3 x +7
'
f ( x )=3
5. Sum/difference rule
Given f ( x )=g ( x ) ± h( x), then

f ' ( x )=g' ( x ) ± h' ( x )


Example: f ( x )=( 2 x +3 ) +(5 x 2 +3 x)
'
f ( x )=2+10 x+ 3=10 x +5
6. Product rule
Given f ( x )=g ( x ) .h ( x), then

f ' ( x )=g' ( x ) . h ( x ) + g ( x ) . h' ( x )


2
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Example: f ( x )=( 7.5+ 0.2 x 2 ) (4+8 x−1 )

f ' ( x )=( 0.4 x ) ( 4 +8 x−1 ) + ( 7.5+0.2 x 2) (−8 x−2)


−2
¿ 1.6 x+ 3.2−60 x −1.6
−2
¿ 1.6+1.6 x−60 x
7. Quotient Rule
g(x)
Given f ( x )= , h(x )≠ 0
h( x )
' '
g ( x ) .h ( x )−g ( x ) . h ( x )
f ' ( x )= 2
[ h (x)]
2
4x
Example: f ( x )=
8+ 0.2 x

' 8 x ( 8+0.2 x )−[ 0.2(4 x 2) ]


f ( x )=
(8+0.2 x)2
2 2
64 x+1.6 x −0.8 x
¿ 2
( 8+0.2 x )
64 x−0.8 x 2
¿ 2
(8+ 0.2 x )
8. Chain rule
dy d y dz
Given a function y=f ( z )and z=f ( x) ,then = .
dx dz dx
Example: Given P=z 0.5 and z=150−0.2Q , then
dP dP dz −0.5 −0.5
= . =0.5 z .(−0.2)=−0.1 z =−¿0.1(150−0.2Q ¿ ¿−0.5
dQ dz dQ
9. Implicit-function rule
Functions of the form y=f (x ) express the dependent variable y in terms of independent variable x
explicitly. Such functions are called explicit functions. Whereas functions of the form f ( x , y )=0 do not
express y in terms of x , or x in terms of y . Such functions are referred to as implicit functions.If the
implicit function f ( x , y )=0exists, and the first-order partial derivative of the function with respect to y is
different from zero, then with the use of total differentiation,
f x . dx+ f y . dy=0
f x . dx=−f y . dy

3
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
dy −f x ∂ f (x , y ) ∂ f (x , y)
= , where f x = and f y =
dx fy ∂x ∂y
Example 1: f ( x , y )=x 2+ y 2−9=0

dy −f x −2 x −x
= = =
dx f y 2y y
Example 2: f ( x , y )=x 3−2 x2 y +3 x y 2−22=0

dy −f x
= =−¿)
dx fy
10. Inverse-function rule
According to the inverse function rule, the derivative of the inverse function is the reciprocal of the
dy
derivative of the original function. Given y=f ( x ), is the derivative of the original function. Derivative
dx
1
of the inverse function will be dy .
dx
Example: IfQ=P3 +2 P2 +7 P, then find the derivative of P with respect toQ .
dP 1 1
= = 2
dQ dQ 3 P + 4 P+7
dP
11. Logarithmic function rule
i. Natural logarithmic function
Given the function, f ( x )=ln g (x), where f ( x )and g( x ) are differentiable, then

' g ' (x)


f ( x )=
g( x )
Example: If f ( x )=ln(6 x ¿+7)¿
6
f ' ( x )=
6 x+7
ii. Logarithmic function with any base b
Given the function,( x )=log b g(x ) , where f ( x) and g( x ) are differentiable, then
'
' g ( x)
f ( x )=
g ( x)ln b
Example: If f ( x)=log 6 3 x , then
'
' (3 x) 1
f ( x )= =
3 x ln 6 x ln 6
4
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
12. Exponential function rule
i. Natural exponential function
Given f ( x )=e g(x ), where f ( x) and g( x ) are differentiable, then
' ' g (x)
f ( x )=g ( x). e
2
Example: If f ( x )=5 e (1− x ) , then

f ( x )=−2 x . ( 5 e ) =−10 e (1− x )


2 2
' (1− x )

ii. Exponential function with any base b


Given the function, f ( x )=bg (x), where f ( x) and g( x ) are differentiable, then
' ' g(x)
f ( x )=g ( x). b ln b
Example: If f ( x )=a3 x , then

f ' ( x )=3 a3 x . ln a
Exercise 1
1. A firm faces a non – linear demand function, P=(650−0.2Q)1.5. Determine the marginal revenue
function of the firm.
252 . Determine the marginal revenue function.
2. If a monopoly faces a non-linear demand function, P=
¿¿
3. A firm is a monopoly seller of good and faces the demand schedule, P=200−Q , and the short-run
production function is given as Q=4 L0.5.Determine the marginal revenue productivity of labour.
4. Find the price elasticity of demand in terms of quantity sold for the following inverse demand function,
10
P= 2
( Q+ 1)

1.2.2 Rules of Integration


The rules of integration are heavily dependent on the rules of differentiation.
Basic rules
1. Constant rule

∫ kdx=kx +c , where k is any constant number.


E.g.∫ dx=x+ c

2. The simple power rule


n+1
x
∫ x n dx= n+1 + c , n ≠−1

5
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
4
x
E.g.∫ x 3 dx= +c
4
3. The exponential rule

∫ e x dx=e x+ c
x
a
∫ a dx=
x
ln a
+c

akx
∫ a dx=kx
k ln a
+c

25 x
E.g.∫ 25 x dx= +c
5 ln 2
4. The logarithmic rule
1
∫ x dx=ln|x|+ c , x ≠ 0
1 dx
The integral ∫ dx is sometimes written as ∫ .
x x
General rules
n+1
5. ∫ [ f ( x ) ] f ( x ) dx=
n ' [ f (x)]
+c
n+1
8
E.g.∫ 2 [ 2 x +3 ]7 dx=
[ 2 x +3 ]
+c
8
6. ∫ f ( x ) e
' f (x) f (x)
dx=e +c

E.g.∫ 2 e
(2 x +3) (2 x +3)
dx=e +c

f '( x )
7. ∫ dx=ln|f (x )|+c , x ≠ 0
f ( x)
2
E.g.∫ dx=ln|2 x +3|+ c
(2 x +3)

Rules of operation
8. Constant multiple rule

∫ kf ( x ¿)dx=k ∫ f ( x ) dx ¿
3
5x
E.g.∫ 5 x 2 dx=5∫ x 2 dx= +c
3
9. Sum/difference rule

∫ [ f ( x ) ± g (x)] dx=∫ f ( x ) dx ± ∫ g ( x ) dx
6
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Examples: (1) ∫ ( x 3− x+1 ) dx=∫ x 3 dx−∫ x dx+∫ 1 dx

¿ ( x4
4
x2
)( )
+c 1 − +c 2 + ( x+ c3 )
2
x 4 x2
¿ − + x +c
4 2

(
(2) ∫ 2 e +
2x 14 x
2
7 x +5 ) 14 x
dx=∫ 2 e2 x dx +∫ 2 dx
7 x +5
¿ e +ln |7 x + 5|+c
2x 2

Note that the derivative of the integral must be equal to the integrand.
10. The substitution rule
du
∫ f (u) dx dx=∫ f (u ) du
du
E.g. ∫ 2 x ( x +1 ) dx .Let u=x +1 , dx=
2 2
. Therefore,
2x
2
∫ 2 x ( x 2 +1 ) dx=∫ 2 xu 2dux =∫ udu= u2 + c1
1 1 1 1
¿ ( x 4 +2 x2 +1 ) + c1 = x 4 + x 2 + + c , where c= + c1
2 2 2 2
11. Integration by parts

∫ vdu=uv−∫ udv
d ( uv )=v du+u dv (product rule of differentials)

∫ d ( uv )=∫ v du+∫ u dv (integrate both sides)


uv=∫ v du+∫ u dv → ∫ vdu=uv−∫ udv
1 2
E.g.
∫ x ( x+1 ) 2
dx ,let v=x , implying dv =dx and also let u= 3 ¿,
1
du=( x+ 1 ) dx . Therefore,
2

∫ x ( x+1 ) dx=∫ vdu=uv−∫ udv


2

2
¿ ¿
3
2
¿ ¿
3

7
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
12. Integration by partial fractions
P( X) P( X)
To integrate rational functions , functions that are ratio of two polynomials, we split
Q(X ) Q(X )
into easier pieces.
Steps:
a) Factorize the denominator
b) Set up
c) Solve for the unknowns
d) Integrate each partial fraction
4 x−1
E.g. 1.∫ dx [degree of P(x) < degree of Q(x), roots are not repeated]
x 2+ x −2
4 x−1 4 x−1
=
2
x + x−2 (x −1)(x +2)
4 x−1 A B
= +
( x−1)(x+ 2) ( x−1) ( x+2)
To solve for A multiply both sides by (x-1) (x+2)
4 x−1= A( x+2)+B ( x−1)
4 x±1=( A+ B ) x+(2 A−B)

{2 A−B=−1
A+ B=4
→ A=1∧B=3 , You can also use cover up method to find A and B.

4 x−1 1 3
∫ x 2+ x −2 dx=∫ x−1 dx +¿ ∫ x+2 dx ¿
4 x−1
∫ x 2+ x −2 dx=ln ¿ x−1∨¿+3 ln ¿ x+ 2∨¿+ c ¿ ¿
x 2 +2
E.g. 2.∫ dx [degree of P(x) < degree of Q(x), roots are repeated]
( x−1 )2 ( x+ 2)
x 2 +2 A B C
∫ ( x−1 )2 (x+ 2) dx= ( x−1) +
(x−1)
2
+
( x +2)
2
x A BX +C
E.g. 3.∫ dx= + 2
2
( x −1)(x +1) (x−1) ( x +1)

x3 x3
E.g. 4. = 2 , by using long division
( x−1)(x+ 2) x + x−2
3
x 3 x−2
=( x−1 ) + 2
( x−1)(x+ 2) x + x−2
8
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Exercise 2
1. If the marginal cost of a firm is MC=2 e 0.2 Q, and the fixed cost is C F =90 , find the total cost
function of the firm C (Q) . (Ans )
−1
2. If the marginal propensity to save is the function of income, S' ( y )=0.3−0.1 y 2 , and the aggregate

saving S is nil when income y is 81, find the saving function S ( y ). (Ans
)

2. Derivatives in Use

Univariate Calculus

2.1 Elasticities: definition and as logarithmic derivatives


Given y=f ( x ), arc elasticityis the average elasticity between two points on the function. It is a measure
of the percentage change in the dependent variable ( y ) due to a given change in the independent variable
( x ).
Consider a simpler linear demand function, y=f ( p). Let ( p1 , y 1 ¿ and ( p2 , y 2 ¿ be two points on the
demand function, then we can define arc elasticity of demand as
%Δ y ∆ y /( y 2+ y 1)
=
%Δp ∆ p/( p2 + P1 )

9
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
One of the problems associated with arc elasticity is that its value depends on the size of price change, ∆ p
, and there is no obvious amount by which one should change price to determine the elasticity value.
Therefore, the concept of arc elasticity is troublesome.
By taking the limit of the arc elasticity formula as∆ p → 0, we get the two points converging and in so
doing we find elasticity at the point ( p1 , y 1 ¿, which is termed as point elasticity of demand.
In a continuous demand function, ∆ y → 0as ∆ p → 0, and we get the following expression for point
price elasticity of demand:
∆ y /( y 2+ y 1)
ε = lim
∆ p→0 ∆ p/( p2 + P1 )
Δ y p 1 dy p 1
¿ lim =
∆ p→0 Δ p y 1 dp y 1
Example: If demand function for a good is, y=10−0.02 p , the point elasticity of demand at p=100 is
dy
dy p dp marginal function
ε= = =
dp y y average function
p

ε =−0.02 ( 1008 )=−28=−0.25=0.25


Find εd if the demand function is Q = 200 − 4P.
Elasticity as logarithmic derivative
For a function, y=f ¿), the point elasticity of y with respect to x is
ε y , x =d ¿ ¿
Proof: d ¿ ¿
d ¿¿
d ¿¿
Example: Find the point elasticity of demand for the demand function, y=100 p−2.

dy p −3 p −3 p −200 p−2
Solution: Usingthe direct approach,ε = =−200 p =−200 p = =−2=2
dp y y 100 p−2 100 p−2
Using elasticity as logarithmic derivative approach, ln y=ln 100−2 ln p
ε y , p=d ¿ ¿

10
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
n
Exercise 3: Given the demand function, Q d =k / p , where k and n are positive constants, find the point
price elasticity of demand using the direct approach and using elasticity as logarithmic derivative
approach.

2.2 Higher-order derivatives


So far we have discussed the first-order derivative of y=f ( x ), which is denoted by f ' ( x ). Now let’s
introduce the concept of second-order derivative and high-order derivatives. Having this knowledge
enables us to develop alternative criteria for determining the relative maximum or minimum of a function.
Higher-order derivatives are found by applying the rules of differentiation to the lower-order derivatives.
Given y=f ( x ), first, second and high-order derivatives are denoted and defined as follows:
dy '
First-order derivative: =f (x )
dx
It measures the rate of change of the original function, f (x).

d2 y ' '
Second-order derivative: =f (x )
d x2
It is the measure of the rate of change of the first-order derivative, f ' (x ). In other words, it measures the
rate of change of the rate of change of the original function, f ( x).
The higher-order derivatives are denoted by f '' ' ( x ), f (4 ) ( x ) , .. . , f (n) ( x ) or
3 4 n
d y d y d y
3
, 4 ,. . . , n
dx dx dx
Example 1: Find the first through the fifth derivatives of the function
y=f ( x )=4 x 4−x 3 +17 x2 +3 x−1
Solution: f ' ( x )=16 x 3−3 x 2+34 x+3

f '' ( x ) =48 x 2−6 x +34


f ' ' ' ( x )=96 x −6
f (4 ) ( x )=96
(5)
f ( x )=0
(5)
f ( x )equal to zero does not mean that the fifth derivative does not exist, rather it has the value zero.
In this example, each successive derivative function emerges as a lower –order polynomial.
Example 2: Find the first four derivatives of the rational function.
x
y=g ( x )= (x ≠−1)
1+ x
Solution: y=g ( x )=x (1+ x)−1

11
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
' −2
g ( x )=(1+ x)
'' −3
g ( x )=−2(1+ x ) ( x ≠ 1)
' '' −4
g ( x )=6(1+ x )
(4) −5
g ( x )=−24 (1+ x)
In this case, repeated derivation does not tend to simplify the subsequent derivative expressions.

2.2.1 Concavity and Convexity


If the first derivative of a function f(x) at x=x o is f ' ( x )=0 , then the value of the function at x o, f(xo), will
be

a) A relative maximum if the derivative f ' ( x ) changes its sign from positive to negativefrom the
immediate left of the point xoto its immediate right
b) A relative minimum if the derivative f ' ( x ) changes its sign from negative to positive from the
immediate left of the point xo to its immediate right
c) Neither a relative maximum nor a relative minimum if f ' ( x ) has the same sign on both the
immediate left and right of point xo→ inflection point. A zero derivative value is not a required
condition for inflection point.

E.g. f ( x )=x 3−12 x 2+ 36 x +8 , x=2 (r. max), x=6 (r. min.)

If the second derivative f ' ' ( x ) is negative for allx, then the primitive function f ( x ) must be a strictly
concave function.E.g. f ( x )=−x 4

If the second derivative f ' ' ( x ) is positive for allx, then the primitive function f ( x ) must be a strictly
convex function.E.g. f ( x )=x 4

If the first derivative of a function f(x) at x=x o is f ' ( x )=0 , then the value of the function at x o, f(xo), will
be
''
a) A relative maximum if the derivative f ( x o ) ≤0
''
b) A relative minimum if the derivative f ( x o ) ≥0
''
c) Either a relative maximum or a relative minimum or inflection point if f ( x o )=0

, x=0,1,2

For unconstrained optimization problems, concavity (convexity) of the objective function guarantees the
existence of absolute maximum (absolute minimum).Absolute maximum is the highest of all points on a
graph.Absolute minimum is the lowest of all points on a graph.A function that gives rise to a hill (valley)
over the entire domain is said to be a concave (convex) function.

12
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Given (x 1 , x 2) , a function f is concave(convex) if and only if , for any pair of distinct points uand v in
the domain of f , and for any 0<θ <1,

θf (u )+(1−θ) f ( v ) ≤ f ( θu+ ( 1−θ ) v )


θf (u )+(1−θ) f ( v ) ≥ f ( θu+ ( 1−θ ) v ), respectively.

Further, if the weak inequality ≤(≥) is replaced by the strictly inequality < ( >), the function is said to be
strictly concave (strictly convex). In the non-strict case, the hill or valley is allowed contain one or more
flat (as against curved) portions, such as line segments on a curve. However, the presence of the word
“strict” rules out such line segments.
An extremum of a concave function must be a peak- a maximum. Moreover, that maximum must be an
absolute maximum (as against relative maximum). However, that absolute maximum may not be unique,
because multiple maxima may occur if the hill contains a flat horizontal top. Absolute maximum is
unique if the function is strictly concave.
Analogously, an extremum of a convex function must be a bottom-a minimum, and that minimum must
be an absolute minimum, which may not be unique. But an extremum of strictly convex function must be
a unique absolute minimum. Note that a strictly concave (strictly convex) function must be concave
(convex), but the converse is not true.
From the definitions of concavity and convexity, we have the following theorems:
Theorem I (linear function): If f ( x) is a linear function, then it is a concave function as well as a convex
function but not strictly so.
Theorem II (negative of a function): If f ( x) is a concave function, then −f (x) is a convex function, and
vice versa. Similarly, if f (x) is a strictly concave function, then −f (x) is a strictly convex function, and
vice versa.
Theorem III (sum of functions): If f ( x) and g( x ) are both concave (convex) functions, then f ( x ) + g(x )
is also a concave (convex) function. In addition, if either one or both of them are strictly concave (strictly
convex), then f ( x ) + g(x ) is strictly concave (strictly convex).

If the function is differentiable, concavity and convexity can also be defined in terms of its first
derivatives. In the one- variable case, the definition is:A differentiable function f ( x) is concave (convex)
iff, for any given point u and any other point v in the domain,
f ( v ) ≤ f ( u )+ f ' ( u ) (v−u)
f ( v ) ≥ f ( u )+ f ' ( u ) (v−u), respectively.

13
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Concavity and convexity will be strict if the weak inequalities are replaced by the strict inequalities, ¿and¿
, respectively. Geometrically, a concave (convex) curve is the one that lies on or below (above) all its
tangent lines. To be a strictly concave (strictly convex) curve, the curve must lie strictly below (above) all
the tangent lines except at the points of tangency.

Example: Check that f ( x )=x 2 is strictly convex.

f ( v )=v 2

f ( u )=u2

f ' ( u )=2u
' 2
f ( u )+ f (u )( v−u ) =u + 2u (v−u)

f ( v )−f ( u )+ f (u )( v−u ) =v −( u + 2u ( v−u ) ) =v −u −2 uv+2 u


' 2 2 2 2 2

2 2
¿ v −2uv +u

¿(v−u)2

Since v ≠ u, (v−u)2> 0 which implies f ( x ) is strictly convex.

When there are two or more independent variables, the definition needs a slight modification:

A differentiable function f ( x )=f ( x 1 , x 2 ,−−−, x n ) is concave (convex) iff , for any given point
u=(u1 , u2 ,−−−, un ) and any other point v=(v 1 , v 2 ,−−−, v n ) in the domain,
n
f ( v ) ≤ f ( u )+ ∑ f j ( u ) (v j −u j )
j=1
n
∂f
f ( v ) ≥ f ( u )+ ∑ f j ( u ) (v j −u j ), respectively. Where f j ( u )= is evaluated at
j=1
∂ xj
u=(u1 , u2 ,−−−, un ).

For strict concavity and convexity, the weak inequalities should be changed to strict inequalities.

A twice continuously differentiable function Z=f ( x1 , x2 ,−−−, xn ) is concave (convex) iff d 2 Z is


everywhere negative (positive) semi definite. The said function is strictly concave (convex) if (but not
only if) d 2 Z is everywhere negative (positive) definite. Here, it is easy to verify whether a function is
strictly concave (strictly convex) by checking whether the Hessian determinant is negative (positive)
definite.
2 2
Example: Check Z=x 1 + x 2 for concavity or convexity.

14
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
|H|=
| || |
f 11 f 12 2 0
=
f 21 f 22 0 2

|H 1|=2
|H 2|=4 . d 2 Z is positive definite. Thus, the function is strictly convex.

2.2.2 Linear Approximations


Linear function is a mathematical equation in which no independent-variable“x” is raised to a power
greater than one.

Polynomials are the nicest possible functions. They are easy to differentiate and integrate, and
polynomials can be evaluated at any point. So what one does in practice is to approximate any function f
of interest by polynomials. When the approximation is done by linear polynomials, then it is called a
linear approximation, which pictorially corresponds to linearizing the graph of f.

The basic idea of linear approximation is “local linearity:” This idea says that a tangent line to a function
really lies quite close to the function, at least near the point of tangency. Therefore, it can serve as a very
easily computed and conceptually simple approximation of the original function.

The idea here in ‘geometric’ terms is that in some vague sense a curved line can be approximated by a
straight line tangent to it. Of course, this approximation is only good at all ‘near’ the point of tangency,
and so on. So the only formula here is secretly the formula for the tangent line to the graph of a function.

If we know the function value at some point (say f (a)) and the value of the derivative at the same point
(f’(a)), we can use these to find the tangent line, and then use the tangent line to approximate f (x) for
other points x. Of course, this approximation will only be good when x is relatively near a. The tangent
line approximation of f (x) for x near a is called the first degree Taylor Polynomial of f(x).

The linear approximation formula arises from the definition of the derivative of a function y = f(x) at a
point x = a:

f ( x )−f ( a) '
lim =f (a)
x→ a x−a

f ( x )−f (a) '


It follows that for x near a (but x ≠a), ≅ f (a).
x −a

f ( x ) ≅ f ( a )+ f ' ( a ) ( x−a). This is the linear approximation formula.

f ( x ) ≅ f ( a )+ f ' ( a ) ( x−a)is the equation of a line with slope f’(a) and (x, y) = (a, f(a)) is one point on the
line. It is the equation of the tangent line to the graph y = f(x) at the point where x = a. Graphically, the
linear approximation formula says that the graph y = f(x) is close to the graph of its tangent line y = f(a) +
f’(a)(x − a) if we stay near to the point of tangency (x, y) = (a, f(a)).

15
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
The linear function, whose graph is the tangent line to the curve y = f(x) at x = a is called the linearization
of f at a.

Examples

1
1. √ 1+ x ≈ 1+ x if x ≈ 0
2
For x=0.01, √ 1+0.01=1.00498 and 1+0.5∗0.01=1.005 , d/ce=1.005−1.00498=0.00002
For x=0.09, √ 1+0.09=1.04403 and 1+0.5∗0.09=1.045 , d/ce = 1.045-1.04403=0.00097
2. (1+ x)r ≈ 1+rx if x ≈ 0

Find √ 9.02 approximately using linear approximation.

This means we use f ( x )= √ x and a=9 in the linear approximation formula.

1 1 1
√ x=√ 9+ ( x−9 )=3+ ( x−9 )=3+ ( 0.02 ) =3.00333
2 √9 6 6

The real significance of the linear approximation is the use of it to convert intractable (non-linear)
problems into linear ones (and linear problems are generally easy to solve).

It turns out that the more times one can differentiate f, the higher is the degree of the polynomial one can
approximate it with, and more importantly, the better the approximation becomes, as one sees it
intuitively.

f ( x )=x
2
√2 x +12at x=2 f ( x )=17 x−18

2.2.3
Polynomial Approximations (Quadratic and Higher Order
Approximations)
A polynomial is a function that can be written in the form p ( x ) =c 0 +c 1 x +…+ c x n for some coefficients
c0. . . cn. If cn ≠ 0, then the polynomial is said to be of order n. A first-order (linear) polynomial is just the
equation of a straight line, while a second-order (quadratic) polynomial describes a parabola.

Polynomials are just about the simplest mathematical functions that exist, requiring only multiplications
and additions for their evaluation. Yet they also have the flexibility to represent very general nonlinear
relationships. Approximation of more complicated functions by polynomials is a basic building block for
a great many numerical techniques.

Quadratic Approximations

To get a better approximation of our function we try to approximate it using a quadratic polynomial.
Another thing we could try is to find a polynomial that has the same value as the function at some point a,
the same derivative at that point a and also the same second derivative there. We do both at once and
define the second degree Taylor Polynomial for f (x) near the point x = a.

16
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
''
f (a )
f ( x )= p2 ( x ) ≅ f ( a )+ f ' ( a ) ( x−a ) + ( x−a)2
2

1 1 2
E.g.√ 1+ x ≈ 1+ x− x if x ≈ 0
2 8
1 2
For x=0.01, √ 1+0.01=1.00498 and 1+0.5∗0.01− (0.01) =1.0049875
8

Higher Order Taylor Polynomials

We get better and better polynomial approximations by using more derivatives, and getting higher
degreepolynomials. The Taylor Polynomial of Degree n, for x near a is given by:
'' '' n
f ( a) f (a) f (a )
f ( x )= p n ( x ) ≅ f ( a )+ f ' ( a ) ( x−a ) + ( x−a )2+ ( x−a )3+ …+ ( x−a)n
2 3∗2 n!

2.2.4 Estimation of Functions (Maclaurin and Taylor Series)


Taylor Series

We start by supposing that f is any function that can be represented by a power series:

f ( x )=∑ c n ( x−a)n
n =0

f ( x )=c 0+ c1 ( x−a )+ c 2(x −a) + c 3( x−a) +c 4 (x−a) +…|x−a|< R (1)


2 3 4

Let’s try to determine what the coefficients c n must be in terms of f. To begin, notice that if we put x = a
in Equation 1, then all terms after the first one are 0 and we get f ( a )=c 0 .

We can differentiate the series in Equation 1 term by term:

f ' ( x )=c1 +2 c 2 (x−a)+3 c 3 (x−a) + 4 c 4 (x−a) + …| x−a|< R


2 3
(2)

and substitution of x = a in Equation 2 gives f ' ( a )=c1

Now we differentiate both sides of Equation 2 and obtain

f (x )=2 c 2+2∗3 c 3 ( x−a ) +3∗4 c 4 ( x−a) +…|x−a|< R


'' 2
(3)

Again we put x = a in Equation 3. The result is f ' ' ( a )=2 c 2

Let’s apply the procedure one more time. Differentiation of the series in Equation 3 gives

f (x)=2∗3 c 3 +2∗3∗4 c 4 (x−a)+ …| x−a|< R


'' '
(4)

'' '
and substitution of x = a in Equation 4 gives f (a)=2∗3 c 3=3 ! c 3

17
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
By now you can see the pattern. If we continue to differentiate and substitute x = a, we obtain
n
f ( a )=2∗3∗4∗…∗n c n=n ! cn

f n( a)
Solving this equation for the nth coefficient cn, we getc n=
n!

This formula remains valid even for n = 0 if we adopt the conventions that 0! = 1 and f (0) = f.

Substituting this formula for cn back into the series, we see that if f has a power series expansion at a, then
it must be of the following form.

f (n ) (a) f ' ( a) f '' ( a ) f ' '' ( a)
f ( x ) =∑
n
( x−a) =f ( a ) + ( x−a )+ ( x−a )2 + ( x−a )3 +…
n =0 n! 1! 2! 3!

The series is called the Taylor series of the function f at a (or about a or centered at a). For the special
case a = 0 the Taylor series becomes

f (n ) (0) n f ' (0 ) f ' ' ( 0 ) 2 f '' ' ( 0 ) 3
f ( x )=∑ (x ) =f ( 0 )+ ( x) + (x) + ( x) +…
n =0 n! 1! 2! 3!

This case arises frequently enough that it is given the special name Maclaurin series.
2 3
h h
The general form of the Taylor series is given by f ( x +h ) =f ( x ) +f ' ( x ) h+ f ' ' ( x ) + f ' ' ' (x) +…
2! 3!
provided that all derivatives of f(x) are continuous and exist in the interval [x,x+h].

We have shown that if f can be represented as a power series about a, then f is equal to the sum of its
Taylor series. But there exist functions that are not equal to the sum of their Taylor series. For example,
one can show that the function defined by

{
−1
x2
f ( x )= e if x ≠ 0 is not equal to its Maclaurin series.
0 if x=0

Under what circumstances is a function equal to the sum of its Taylor series? In other words, if f has
∞ ( n)
f (a)
derivatives of all orders, when is it true that ( x )=∑ (x−a)n ?
n=0 n!

As with any convergent series, this means that f(x) is the limit of the sequence of partial sums. In the case
of the Taylor series, the partial sums are
∞ (n )
f (a )
T n ( x )= ∑
n
( x−a )
n=0 n!
' '' ' '' n
f ( a) f (a) 2 f ( a) 3 f (a )
¿ f ( x−a ) + ( x−a )+ ( x −a ) + ( x−a ) +…+ ( x−a )n
1! 2! 3! n!
18
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Notice that Tn is a polynomial of degree n called the nth-degree Taylor polynomial of f at a.

In general, f(x) is the sum of its Taylor series if f ( x )= xlim


→∞
T n ( x ). If we let R ( x )=f ( x )−T ( x ) so that
n n

f ( x )=R n ( x ) +T n ( x ) then Rn ( x ) is called the remainder of the Taylor series. If we can somehow show that
lim Rn ( x ), then it follows that lim T n ( x )= lim [¿ f ( x )−Rn ( x )]=f ( x )− lim R n ( x )=f ( x ) ¿.
x→ ∞ x→ ∞ x →∞ x →∞

If f ( x )=R n ( x ) +T n ( x ), where Tn is the nth-degree Taylor polynomial of f at a and lim Rn ( x )=0 for | x −
x→ ∞
a | < R, then f is equal to the sum of its Taylor series on the interval | x − a | < R.

Taylor’s Remainder Formula

If f has n+1 derivatives in an interval I that contains the number a, then for x in I there is a number z
strictly between x and a such that the remainder term in the Taylor series can be expressed as

f ( n+1) ( z )
Rn ( x )= ( x−a)n+1
(n+1)!

It turns out that there is a simple formula which gives us a bound on the size of the error R. Again, this is
only a bound on the size of the error and does not tell us the exact error. This bound is called the
remainder formula.

For the special case n = 0, if we put x = b and z = c in Taylor’s Formula, we get


f ( b )=f ( a ) +f ' (c)( b−a) which is the Mean Value Theorem.

Notice that the remainder term is very similar to the terms in the Taylor series except that is f (n+ 1)
evaluated at z instead of at a. All we say about the number z is that it lies somewhere between x and a.
The expression for Rn ( x ) is known as Lagrange’s form of the remainder term. You are probably worrying
how on earth we can use this formula to get actual numbers if we don’t know what c is. Good question.
What we need to do is look at all the values of f(n+1)(c) (for all a < c < x) and use the largest of them. Or,
pick something that we know is surely larger than all of them.

xn
lim =0 for every real number x.
x→ ∞ n !

Example 1: Give a bound on the error for when e 0.5 is approximated by the fourth degree Taylor
Polynomial of ex centered at 0. The fifth derivative of e x is again ex. We don’t know what that is, but we
do know that that ex is an increasing function between 0 ≤ x ≤ .5. So,

|f (5 )( x)|≤|e0.5|=√ e < √ 3<2

| |
( 5)
f (c ) 2
Thus, the error can be bounded: |f ( 0.5 )− pn (0.5)|≤|R4|=
5
( x−0 ) = x 5
5! 5!

19
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
( .5 )2 ( .5 )3 (.5)4
So if we approximate e0.5 by P4 ( .5 )=1+ (.5 )+ + + our approximation will be within
2 3! 4!

2
(0.5)5 <0.0006 of the true value.
120

Example

Find the value of f(6) given that f(4)=125, f’(4)=74, f’’(4)=30, f’’’(4)=6 and all other higher derivatives
of f(x) at x=4 are zero.

Solution
2 3
( ) ( ) '
( ) ''
( ) h '' h
f x +h =f x +f x h+ f x + f ' (x) +…
2! 3!

x=4, x+h=6, h=2

As fourth and higher derivatives of f(x) are zero at x=4.

' '' 22 ' ' 23


f ( 4+2 ) =f ( 4 ) + f ( 4 ) 2+ f ( 4 ) + f ' ( 4) +…
2! 3!

( ) ( )
2 3
2 2
f ( 6 )=125+74 ( 2 ) +30 +6 =125+148+60+8=341
2! 3!

Note that to find f(6) exactly, we only needed the value of the function and all its derivatives at some
other point, in this case, x=4. f ( x )=x 3 +3 x 2+2 x +5

2.2.5 Intermediate Value Theorem and Newton’s Method


Rolle's Theorem: Let f be continuous on [a,b] and differentiable on (a,b). If f(a) = f(b), then there exists a
point c such that f ' (c) = 1.

We look into some methods for finding solutions of equations, something that is often required. We will
look at two fairly basic methods, the bisection method and Newton’s method.

(Intermediate Value Theorem) If y = f(x) is continuous for all x in the finite closed interval [a, b] and if
f(a) and f(b) have opposite signs (a short way to write that is f(a)f(b) < 0) then there must be some c ∈ (a,
b) with f(c) = 0.

The bisection method is an approximation method, which will never get the exact answer (except by
accident) but it can get the answer to any desired accuracy (or any number of decimal places) if used for
long enough.It is a method for locating roots of an equation f(x) = 0. To start it, we need a range a ≤ x ≤ b
where f(x) is continuous and where f(x) changes sign. So assume f(a)*f(b) < 0 (meaning that the signs of
y = f(a) and y = f(b) are opposite).

20
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
a+ b
Look at the mid-point x = c= . If f(a)*f(c) < 0 then there is a root in the interval (a, c). If not, then
2
f(c)*f(b) ≤ 0 and there is a root in [c, b). So the result of this basic step is to find that y = f(x) has a root in
either one of the two intervals [a, c] or [c, b]. Replace the original interval [a, b] by the half where there is
a sign change and repeat the basic step.

Graphically,

E.g. Find zero of the function f ( x )=x 2−2 in [1,2]

f(1) = -1 & f(2)=2,

Step 1: mid-point x=1.5, f(1) = -1, f(1.5)= 0.25 & f(2)=2

Step 2: mid-point x=1.25, f(1) = -1, f(1.25)= - 0.4375, f(1.5)= 0.25

Step 3: mid-point x= 1.375, f(1.25)= - 0.4375, f(1.375)= -0.109375, f(1.5)= 0.25

Step 4: mid-point x= 1.4375, f(1.375)= -0.109375, f(1.4375)=0.06640625, f(1.5)= 0.25

Step 5: mid-point x= 1.40625, f(1.375)= -0.109375, f(1.40625)= -0.0224609375, f(1.4375)=0.06640625

Step 6: mid-point x= 1.421875, f(1.40625)= -0.0224609375, f(1.421875)=0.021728515625,


f(1.4375)=0.06640625

……..stops when you get f(c)=0⇒c= √ 2=1.4142135623730950488016887242097

Newton's method is an example of how the first derivative is used to find zeros of functions and solve
equations numerically. Newton’s method for computing the zeros of functions is a good example of the
practical application of the Mean Value Theorem {Mean value theorem: Let f(x) be differentiable on the
open interval (a, b) and continuous on the closed interval [a, b]. Then there is at least one point c in (a, b)
such that f’(c) = (f(b)-f(a))/(b-a)}. It has a more sophisticated idea behind it and generally works very
well. However, it is not guaranteed to work (as the bisection method is). Newton’s method uses linear
approximation (and so requires f’(x) to exist).

To begin with we assume we have a rough idea where there is a solution. We call this our initial guess
and write x = xo for this value of x. (The effect of a bad or wild guess is often not serious. Typically a
better guess will save effort and get to the answer more quickly, but there are cases where a bad guess
will just not work. In those cases, you discover that the guess is bad as you go along. There are some
cases where the method won’t work, even with a pretty good guess.)

Wereplace the correct equation f(x) = 0 by its linear approximation centered at x = x o. That is, we solve

f ( x o ) + f ' ( x o )( x−x o )=0

f ( xo)
T ( x)=x =x o− ' (Newton’s map)
f ( xo)

21
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
This is not likely to be the correct solution of the real equation f(x) = 0 but the idea is that it should be a
f ( xo)
‘better guess’ than x = xo was. We put x 1=x o− ' as an improved guess.
f ( xo)

Then we repeat the idea starting with x = x 1 where we had x = x0 and we get a further ‘improved’ guess
f ( x1 )
x 2=x 1− . Then we go again, and again. We get a sequence of improved guesses x = x 1, x = x2, . . .
f ' ( x1 )
f ( xn)
where each new guess xn+1 is related to the previous one by x n+1=x n − (n=0 , 1, 2 , … .¿ . If they
f ' ( xn )
don’t stabilize, try starting again with another guess. X0↔X1

E.g. Find a root of f ( x )=x 2−2 using Newton’s method.

f (x)
Let’s start with x = 1. T ( x )=x− ' ,
f (x)

T(1) = 1 − (1 − 2)/2 = 3/2

T(3/2) = 3/2 − ((3/2)2 − 2)/3 = 17/12 = 1.4166666666666666666666666666667

T(17/12) = 577/408 = 1.414215686274509803921568627451

T(577/408) = 665857/470832= 1.4142135623746899106262955788901

The Newton method converges extremely fast to a root f(p) = 0 if f ′ (p) ≠ 0 if we start sufficiently close
to the root.The method uses a local linear approximation, which clearly breaks down near a turning point.

Newton’s method does not always work. For instance, when f’(x n) = 0. Sometimes it does not converge.
1
E.g. f ( x )=x 3

2.3 Multivariate Calculus


Many functions in economics depend on several variables.

2.3.1 Partial Derivatives (First and Second Order)


In many problems involving functions of two variables, the goal is to find the rate of change of the
function with respect to one of its variables when the other is held constant. That is, the goal is to

22
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
differentiate the function with respect to the particular variable in question while keeping the other
variable fixed. This process is known as partial differentiation, and the resulting derivative is said to be a
partial derivative of the function.

No new rules are needed for the computation of partial derivatives. To compute f x, simply
differentiate f with respect to the single variable x, pretending that y is a constant. To compute f y,
differentiate f with respect to y, pretending that x is a constant.

E.g. z ( x , y )=x 3+ 4 xy +5 y 2
FOPD
∂z ∂z
=3 x 2+ 4 y , =4 x +10 y
∂x ∂y
SOPD

Direct partial derivatives:


∂2 Z ∂2 Z
=6 x , =10
∂ x2 ∂ y2
Cross (mixed) partial derivatives:

( ) ( )
2 2
∂ Z ∂ ∂z ∂ Z ∂ ∂z
= =4 , = =4
∂ y∂ x ∂ y ∂ x ∂x∂y ∂x ∂ y

Note that the two cross-partial derivatives are the same: it doesn’t matter whether we di fferentiate with
respect to x first and then with respect to y, or vice-versa. This happens with all “well-behaved” functions.

2.3.2 The Multivariate Chain Rule


If z is a function of two variables, x and y, and both x and y depend on another variable, t (time, for
example), then z also depends on t. If z = z(x, y), and x and y are functions of t, then:

∂z ∂z ∂x ∂ z ∂ z
= +
∂t ∂ x ∂t ∂ y ∂t

Example: Suppose the aggregate output of a country is given by: Y = A K α L β where K is the capital
stock and L is the labour force. If the capital stock and the labour force are each growing according to:
nt mt
L ( t )=L o e and K ( t ) =K o e

(i) What are the proportional growth rates of capital and labour?
dL dK
=n Lo ent =nL ( t ) , =m K o e mt =mK ( t )
dt dt

So labour is growing at a constant proportional rate n, and capital is growing at constant


proportional rate m.

(ii) What is the rate of growth of output?


dY ∂Y ∂ K ∂Y ∂ L
Applying the Chain Rule: = +
dt ∂ K ∂ t ∂ L ∂t

23
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
dY
=αA K α−1 L β∗mK + βA K α L β−1∗nL
dt
dY α β α β
=αmA K L + βnA K L
dt
dY α β
=( αm+ βn ) A K L =( αm+ βn ) Y
dt

So output grows at a constant proportional rate (αm + βn).

2.3.3 Homogeneous Functions and Euler’s Theorem


 A function of two variables is called homogeneous of degree n if f(λx, λy) = λ n f(x, y). If a
production function is homogeneous of degree 1, it has constant returns to scale.
 All Cobb-Douglas production functions are homogenous.
 Quick check of degree of homogeneity in the case of Cobb-Douglas production functions: add the
powers on K and L so if Y =K α L β, then degree of homogeneity¿ α + β
 Leonhard Euler (pronounced "oiler") was a Swiss mathematician who lived from 1707 to 1783.
Euler’s Theorem states that if the function f(x, y) is homogeneous of degree n:
∂f ∂f
x +y =nf ( x , y ) (special property of homogeneous functions)
∂x ∂y
E.g. f ( x , y )=2 x2 y 2+ x y 3
(i) Show that f is homogenous.
f ( λx , λy )=λ 4 f (x , y)
(ii) Verify that it satisfies Euler’s Theorem.
∂f 2 3 ∂f 2 2
=4 x y + y , =4 x y +3 x y
∂x ∂y
∂f ∂f
x +y =x ( 4 x y 2 + y 3 ) + y ( 4 x 2 y+3 x y 2 )
∂x ∂y
2 2 3 2 2 3
¿ 4 x y + x y +4 x y + 3 x y
2 2 3
¿ 8 x y +4 x y
¿ 4 f ( x , y)

2.3.4 Total Differentials and Implicit Differentiation


∂z ∂z
 The total differential of the function z(x, y) is dz= dx + dy
∂x ∂y
E.g. F ( x , y ) =x 4 + x y 3
dF=( 4 x + y ) dx+ ( 3 y ) dy
3 3 2

 We may or may not be able to solve the equation F(x, y) = 0 for y as an explicit function of x.
 Implicit differentiation is a method for determining the tangent lines to this curve, without
explicitly solving for y = y(x) in terms of x.
 The formula for implicit differentiation can be derived from if dz=0:

24
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
dy −z x
=
dx zy
E.g. F ( x , y ) =x 4 + x y 3

dy −z x −4 x + y
3 3
= =
dx zy 3x y
2

25
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
3. Unconstrained optimization
The term optimization may be used both for maximization and minimization. It implies thebestresult
under the available circumstances. Optimization can be generally classified into two: unconstrained (free)
optimization and constrained optimization.Unconstrained simply means that the choice variable can take
on any value-there are no restrictions. Constrained means that the choice variables can only take on
certain values within a larger range.

In formulating an optimization problem, the first order of business is to delineate an objective function.
The essence of the optimization process is simply to find the set of values of choice variables
(independent variables) that will yield the desired extreme of the objective function. Extreme value is the
collective term for maximum and minimum.

When the objective function is free of constraints, we call it unconstrained optimization. Within the
unconstrained optimization problem heading, we can have single-variable and multi-variable types of
problems. Single-variable problems involve only one choice variable; multi-variable problems involve
more than one choice variable.

3.1 Functions of One Independent Variable


A function y=f (x ) will have a maximum value at the point it changes from increasing to decreasing
function. A maximum value of a function is greater than any value of the function immediately preceding
or following the given value. The function will have a minimum value at the point it changes from
decreasing to increasing function. A minimum value of a function is less than any value of the function
immediately preceding or following that value.

We have to distinguish between necessary and sufficient conditions for a point to be the maximum (or
minimum) of some function f(x).

Necessary means the condition is required - necessary doesn't imply that the condition alone will
guarantee the result.They only give us candidate solutions.

Sufficient means presence of the condition is enough to guarantee the result, however, sufficient
conditions are not necessarily required to guarantee the result.

A function may have several interiormaxima or minima. If there are several points of maxima (or
minima) in various neighborhoods, the highest /lowest of them is called the global maximum/minimum.

Given y=f ( x ), this function can be optimized if the following conditions are fulfilled.

 For maximization,
a. f ' ( x )=0 -----------------necessary condition/first order condition
b. f '' ( x ) < 0 at the critical values------------sufficient condition/second order condition
 For minimization,
a. f ' ( x )=0 -----------------necessary condition
b. f '' ( x ) > 0 at the critical values------------sufficient condition

26
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
However, f ' ' ( x ) may be zero when evaluated at the critical value(s); in this case we should apply the
successive derivative of f ( x) until we get the non-zero value for the derivative.

If the non-zero value for the derivative occurs at odd numbered derivative, f ( x) is at its inflection point.
If the non-zero value for the derivative occurs at even numbered derivative and is negative, f ( x ) is at its
maximum point. If the non-zero value for the derivative occurs at even numbered derivative and is
positive, f ( x ) is at its minimum point.

Example: Find the extreme value(s) of the function,( x )=3 x 4−10 x 3 +6 x 2+5 .

Solution:

First order condition


'
f ( x )=0

12 x3 −30 x 2 +12 x =0

3 x ( 4 x 2−10 x+ 4 )=0

3 x=0 Or4 x2 −10 x + 4=0

10 ± √ 100−64 10 ±6 1
x=0 or x= = =2∨
8 8 2

1
Therefore, x=0 , x=2∧x= are critical values.
2

Second order condition

f '' ( x ) =36 x2 −60 x +12


''
f ( 0 )=12>0 ⇒ f ( x )has a minimum value at x=0 .

f ( x)min =3 ¿

f '' ( 2 )=36> 0⇒ f ( x )has a minimum value at x=2.

f ( x)min =3 ¿

f '' ( 12 )=−9<0 ⇒ f (x)has a maximum value at x= 12 .


f (x)max =3 ¿

Exercise: Find the extreme value(s) of the following functions.


a. f ( x )=x 3−12 x 2+ 36 x +8

27
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
b. f ( x )=(x 2−2 x+1)2+ 2

Economic Applications
a. Maximization of revenue functions
A firm can maximize its total revenue (TR ) when

dTR
i. =0 ⇒ MR=0 ---------------------------FOC
dQ
d 2 TR dMR
ii. 2
<0 ⇒ <0 ---------------------SOC
dQ dQ

−x
Example 1: If the demand function for commodity x is given by ¿ 12 e 4 , find the level of production
that can maximize total revenue.
Solution: TR=P . Q

TR=( 12 e ) . x=12 x e
−x −x
4 4

−x −x
dTR
=0 ⇒ 12 e 4 −3 x e 4 =0
dQ
−x
4
3e ( 4−x )=0
4−x =0
x=4

( ) ( )
2 −x −x −x
d TR −1 4 4 −1 4
SOC: =12 e −(3 e +3 x e )
dQ
2
4 4
−x −x −x
3
¿−3 e 4 −3 e 4 + x e 4
4
−x −x
3
¿−6 e 4 + x e 4
4
−x
4 3
¿ e ( x−6)
4

( )
2 −4
d TR 3 −3
At x=4, =e 4 ( 4 )−6 =e−1 ( 3−6 )=e−1 (−3 ) = <0
dQ
2
4 e
Therefore, x=4 is the level of production that maximizes the total revenue.
−4
48
The maximum revenue will be TR=12 ( 4 ) e 4
=48 e−1=
e
Example 2: A farmer wants to send his perishable product to a city market as soon as possible. He has
established that he can send now 1.5 tons of the product per day and can get a price of 2,500 Birr per
tones. If he waits, he can get a price increase of 20 Birr per tones per day but the quantity of the product
to be supplied will be reduced by 0.01 tons per day. For how many days should he wait so that his
revenue becomes maximum?

28
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Solution: Let x be the number of days
P=2,500+ 20 x
Q=1.5−0.01 x
TR=P . Q
TR=( 2,500+20 x ) (1.5−0.01 x)
2
TR=3,750−25 x +30 x −0.2 x
2
TR=3,750+5 x −0.2 x
d TR
FOC ⟹ =0⟹ 5−0.4 x=0 ⇒ x=12.5 days
dX
d2 TR
SOC⟹ =−0.4 <0
dx 2
Therefore, the farmer should wait 12.5 days to maximize his revenue.

Exercise: Suppose a 200 room hotel in Addis Ababa will rent all its rooms when it charges 125 Birr per
night per room. However, from past experience the manager knows that for each 5 Birr increase in rent
per night per room, 4 rooms remain unoccupied per night. What rent per room will maximize total
revenue per night?
b. Minimization of cost functions
Given C=f (Q) , firms can minimize total cost if and only if:

dTC
i. =0⇒ MC =0 ---------------------------FOC
dQ
d 2 TC d MC
ii. 2
>0 ⇒ >0 ---------------------SOC
dQ dQ
Example: Find the minimum average cost if the total cost function is given C=36 x−10 x 2+2 x 3 .

C 2
Solution: AC= =36−10 x+ 2 x
X

dAC
=−10+ 4 x=0 ⟹ x=2.5
dx
2
d AC
SOC: 2
=4>0
dx

Therefore, x=2.5 gives the minimum value of the average cost.


2
AC min =36−10 ( 2.5 )+ 2 ( 2.5 ) =23.5

Exercise: Given a cubic total cost function as C ( Q ) =a Q 3+ b Q 2+ cQ+d , a , c , d >0 , b< 0∧b 2< 3 ac
.Determine the amount of Q that minimizes marginal cost.

c. Maximization of profit functions


π (Q )=TR ( Q )−TC (Q)

29
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
The profit functions will be maximized when the following conditions are fulfilled.
dπ dTR dTC
FOC :
= − =0 ⇒ MR=MC=0
d Q dQ dQ
2 2 2
d π d TR d TC d MR d MC
SOC : 2 = 2
− 2
<0 ⇒ < .i.e., the slope of marginal revenue must be less than
dQ dQ dQ dQ dQ
the slope of marginal cost.
Example: Suppose a monopolist faces a demand curve, Q=600−0.5 P and total cost curve
C ( Q ) =Q3−61.25 Q2+ 1528.5Q+2000 . Determine the profit maximizing level of output and price.

Solution: P=1,200−2 Q

TR(Q)=1,200 Q−2Q 2

C ( Q ) =Q3−61.25 Q2+ 1528.5Q+2000

π (Q )=−Q 3 +59.25 Q 2−328.5 Q−2000

dπ 2
FOC ⇒ =−3Q +118.5 Q−328.5=0
dQ

⇒ Q=36.5∨3

d2 π
SOC ⇒ =−6 Q+118.5
dQ 2

d2 π
When Q=3 , 2
=−6 ( 3 )+118.5=−18+118.5=100.5>0
dQ
2
d π
When Q=36.5 , 2
=−6 ( 36.5 ) +118.5=−219+118.5=−100.5<0
dQ

Therefore, the profit maximizing level of output is ¿ 36.5 .

The price will be, P=1200−2(36.5)=1127

Exercise:

1. ABC Company is planning to market a new model of product x . The management collects
information from retailers that how many units of product x they would buy for various prices.
From this survey, it is determined that the unit demand function is; x=−1500 P+30000. The
fixed costs to the company for the production of product x are found to be 28000 Birr and the
cost for material and labour to produce each unit of product x is estimated to be 8 Birr per unit.
Determine the price that the company should charge to maximize its profit. Find the profit
maximizing level of x .

30
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
2. The total cost and demand functions of a monopolist are ¿ a x 2+ bx+ c , P=β −αx where
α , β , a , b , c are positive constants. Find the profit maximizing output level and show that the
output level is the same whether output or price is fixed.

3.2 Functions of Several Independent Variables


For a function of two variables f(x, y) we can try to find out which pair of values, x and y, give the largest
value for f. If we think of the function as a land surface in 3-dimensions, with f(x, y) representing the
height of the land at map-coordinates (x, y), then a maximum point is the top of a hill, and a minimum
point is the bottom of a valley. As for functions of one variable, the gradient (slope) is zero at maximum
and minimum points - whether you move in the x-direction or the y-direction:

The gradient may be zero at other points which are not maxima and minima - for example you can
imagine the three-dimensional equivalent of a point of inflexion, where on the side of a hill the land
momentarily flattens out but then the gradient increases again. There might also be a saddlepoint - like
the point in the middle of a horse’s back which if you move in the direction of the head or tail is a local
minimum, but if you move sideways is a maximum point.
Having found a point that satisfies the first-order conditions (a stationary point) we can check the second-
order conditions to see whether it is a maximum, a minimum, or something else.

Why are the second-order (sufficient) conditions so complicated? If you think about a maximum point
you can see that the gradient must be decreasing in the x- and y-directions, so we must have fxx< 0 and
fyy< 0. But this is not enough - it is possible to imagine a stationary point where if you walked North,
South, East or West you would go down-hill, but if you walked North-East you would go uphill. The
condition guarantees that the point is a maximum in all directions.
Note that, as for functions of one variable, we find local optima by solving the first- and second-order
conditions. We need to think about the shape of the whole function to decide whether we have found the
global optimum.
Note thatif f xx . f yy <( f ¿¿ xy)2 ¿ , f is at a saddle point or at an inflection point. Saddle (dual personality)
and inflection points are not extreme points.
f is at a saddle point when f xx and f yy have different signs.
f is at a inflection point when f xx and f yy have the same sign.
Example: Find the extreme value(s) of Z=x+ 2ey −e x −e2 y .
x 2y
Solution: FOC ⟹ Z x =1−e =0 and Z y =2e-2 e =0. This occurs when x=0∧ y=1/2.
SOC ⇒ Z xx =−e x .
Z yy =−4 e 2 y.
Z xy =0 .
When we evaluate the second order partial derivatives at x=0 and ¿ 1/2 , we find that

31
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Z xx =−1 , and Z yy =−4 e . Since Z xx and Z yy are both negative(−1 ) (−4 e ) >0 , we conclude that Z has
maximum value at x=0 and y=1 /2.
2( )
1
This maximum value is, Z max=0+ 2 e
1
2
0 2
()
−e −e =0+ e−1−e=−1.

Exercise: Find and classify the stationary point(s) of


a. Z=2 y 3−x 3+147 x−54 y
b. Z=8 x 3+ 2 xy−3 x 2 + y 2 +1
c. f ( x , y )=−x 2+ xy−2 y 2−3 x+12 y +50
d. g ( x , y )= y 3 −12 y + x 2 e y

Alternative method to the sufficient conditions


We can test the sufficient conditions using the concept of second order total differential of Z=f ( x , y )
which is denoted by d 2 Z .
dZ=Z x dx +Z y dy

2 ∂ ∂
d Z=d ( dZ )= ( dZ ) dx+ ( dZ ) dy
∂x ∂y

∂ ∂
¿
∂x
( Z x dx +Z y dy ) dx + ( Z dx + Z y dy ) dy
∂y x

¿ ( Z xx dx+ Z xy dy ) dx + ( Z yx dx+ Z yy dy ) dy

2 2
¿ Z xx dx + Z xy dydx + Z yx dxdy + Z yy dy
2 2
¿ Z xx dx +2 Z xy dydx+ Z yy dy

The second order total differential expressed in the above equation is a quadratic form.

Letdx=u ,dy =v , Z xx =a , Z yy =b , Z xy =Z yx=h . Then the above second order total differential can be
identified as a quadratic form q in two variables u and v : q=a u2 +2 huv +b v 2

The discriminant of a quadratic form q=a u2 +2 huv +b v 2 is the symmetric determinant, |ah hb| .

2 2 2
In the particular case of the quadratic formd Z=Z xx dx +2 Z xy dydx + Z yy dy , the discriminant is a
determinant with the second order partial derivatives as its elements. Such determinant is called a

Hessian determinant. In the two variable cases, the Hessian determinant is, |H|= |
Z xx Z xy
Z yx Z yy |
.

The successive principal minors may be denoted by |H 1|=Z xxand |H 2|= |


Z xx Z xy
Z yx Z yy |
=Z xx . Z yy −¿ .

32
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
d Z is positive definite iff |H 1|>0 and |H 2|>0 . In this case, Z is at its minimum.
2

d Z is negative definite iff |H 1|<0 and |H 2|>0 . In this case, Z is at its maximum.
2

When there are n-choice variables, the objective function may be expressed as
Z=f (x1 , x2 , x3 ,−−−, n)

The necessary condition for extremum is that all the first order partial derivatives be zero, the same as the
two-variable case. Z x 1=Z x2 =Z x3 =0.

Sufficient condition

| |
Z 11 Z 12−−−¿ Z 1 n
Z 21 Z 22−−−¿ Z 2 n ¦
The Hessian determinant is expressed as |H|= ¦ ¦ ¿.
¦ ¿ Zn 3
Zn 1

The sufficient condition for an extremum of Z is stated as follows;

 Z is at maximum if |H 1|<0 ,|H 2|> 0| H 3|<0 ,−−−,(−1)n ¿ (¿ d 2 Z is negative definite)


 Z is at minimum if |H 1|>0 ,|H 2|> 0| H 3|>0 ,−−−, ¿ (¿ d 2 Z is positive definite)
2 2 2
Example: Find the extreme value(s) of Z=2 x 1 + x 1 x 2+ 4 x 2 + x1 x 3+ x 3 +2
Solution: Z x 1=0 ⇒ 4 x 1+ x 2 + x 3=0 -------------------(1)
Z x 2=0 ⇒ x 1 +8 x2 =0--------------------------(2)
Z x 3=0 ⇒ x 1 +2 x 3=0--------------------------(3)
This is a homogeneous linear equation system. By matrix inversion method,

[ ][] [ ]
−1
x1 4 1 1 0 x 1=0 4 1 1
⌈ x2 ⌉= 1 8 0 0 ⇒ x 2=0|H |= 1 8 0
x3 1 0 2 0 x3 =0 1 0 2

[ ]
4 1 1
|H 1|=4 , |H 2|= 1| |
4 1
8
=31, |H 3|=|H|= 1 8 0 =54
1 0 2

Therefore, Z is at its minimum. Z min=2

Economic applications

1. A monopolist firm produces two products ( Q1and Q2) . Suppose that the demands facing the
monopolist firm are as follows;
Q1=40−2 P1 + P2
Q2=15+ P1 −P 2

33
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
2 2
Production costs are related and the firm faces the total cost function; TC =Q 1 + Q1 Q2 +Q2 .Determine
the profit maximizing level of output and price.
Solution: π=TR−TC
TR=TR1+ TR 2=P1 Q1 + P2 Q2

But, it is necessary to solve prices ( P1∧P2 ¿ in terms of quantities (Q 1∧Q2 ¿. + {


−2 P 1+ P2 =Q1−40
P1−P2=Q 2−15
−P1=Q1 +Q2−55
P1=55−Q1−Q2

P2=15+ P1 −Q2

P2=15+55−Q1−Q2−Q 2

P2=70−Q1−2 Q2

TR=( 55−Q1−Q2) Q1 + ( 70−Q1−2 Q2 ) Q2


2 2
TR=55Q1 −Q 1 −Q1 Q2 +70 Q2−Q1 Q 2−Q2

TR=55Q1 +70 Q2−2 Q 1 Q2 −Q12 −2Q22

TC =Q12+ Q1 Q2 +Q22

π=TR−TC

π=55 Q1 +70 Q2−2 Q1 Q2−Q12−2 Q22−Q12−Q1 Q2−Q22

π=55 Q1 +70 Q2−3 Q1 Q 2−2Q12−3 Q22

∂π
=55−3 Q 2−4 Q1=0
∂Q 1

⇒ 4 Q1 +3Q 2=55-----------------(1)

∂π
=70−3 Q 1−6 Q2=0
∂Q 2

⇒ 3 Q 1 +6 Q2=70-----------------(2)

2
Solving equations (1) and (2) simultaneously, we get Q 1=8 andQ 2=7 .
3

1 2
P1=39 and P1=46 .
3 3

34
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
SOC ⟹|H|=
| ||
π 11 π 12 −4 −3
=
π 21 π 22 −3 −6 |
|
|H 1|=−4 , |H 2|= −4
−3 |
−3 =15
−6

2
Therefore, the firm can maximize its profit when Q 1=8 and Q2=7 .
3

2. A multi-plant monopolist operates two plants whose total cost functions are given by;
2
TC 1 =8.5+0.03 Q1
2
TC 2=5.2+ 0.04 Q2
If the demand function is given by P=60−0.04 Q , where Q=Q 1 +Q2 . How much output should the
monopolist produce in each plant to maximize its profit.
Solution: π=TR−TC
TR=P . Q=( 60−0.04 Q ) Q=60 Q−0.04 Q , but Q=Q 1 +Q 2
2

¿ 60 ( Q 1 +Q 2 )−0.04 (Q 1 +Q2)2
2 2
¿ 60 Q 1+ 60Q 2−0.04 Q1 −0.08 Q1 Q 2−0.04 Q2
TC=TC 1 +TC 2
¿ 8.5+0.03 Q12 +5.2+0.04 Q22
2 2
¿ 13.7+0.03 Q1 +0.04 Q2
π=TR−TC
¿ 60 Q1+ 60Q 2−0.04 Q1 −0.08 Q1 Q 2−0.04 Q22−13.7−0.03 Q12−0.04 Q22
2

2 2
¿ 60 Q1+ 60Q 2−0.07 Q1 −0.08 Q1 Q2 −0.08 Q2 −13.7
∂π
FOC ⇒ =60−0.14 Q1−0.08 Q2=0
∂ Q1
⇒ 0.04 Q 1+0.08 Q2=60--------------------------------(1)
∂π
=60−0.08 Q1−0.16 Q2=0
∂Q 2
⇒ 0.08 Q1 +0.16 Q 2=60 --------------------------------(2)
Solving equations (1) and (2) simultaneously, we get Q 1=300 and Q 2=225
The total output,Q=Q 1 +Q 2=525

SOC ⇒ |H|=
| ||
π 11 π 12 −0.14 −0.08
=
π 21 π 22 −0.08 −0.16 |
|H 1|=−0.14

|
|H 2|= −0.14
−0.08
−0.08
−0.16 |
=0.016

Therefore, the monopolist can maximize its profit when it produces Q 1=300 and Q2=225 .

35
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Exercises

1. A discriminating monopolist has two distinct markets described by


P1=80−5 x 1
P2=180−20 x2
If the total cost function is TC=20 x+ 20, where x=x 1 + x2 , find the prices the monopolist will charge in
each market while maximizing his profits.

2. Suppose a firm faces a production function, Q=0.8 K 0.4 L0.3. It sells its output at a fixed price of 450
Birr per unit and can buy K and L at 15 Birr per unit and 8 Birr per unit, respectively. What input
combination will maximize the profit of the firm?

3.3 Implicit Function Theorem and Unconstrained Envelope Theorem


The previous sections deal with the way how the maximum or minimum value of functions depends on
the value of independent variables under consideration. However, in economic theory, economists are
usually interested in how the optimal value of functions depends on some parameters such as tax rates,
subsidy, etc.How do changes in the value of the parameter affect the optimal value(s) of the function?

Although these parameters are assumed to be constant during the process of optimization, they may vary
according to economic situation. Therefore, what happens to the optimal value of the objective function
when these parameters change?

The implicit function theorem states that if for some point (x0, y0)
on a level curve f (x, y) = c, then one can find a function y = g (x) defined near x = x 0 such that the level
curve is locally the graph of this function, i.e. one has f (x, g (x)) = c for x near x 0.

The implicit function theorem implies that the first order conditions to be used: to characterize the
solution (optimal value of the control variable(s)) as a function of the parameters of the optimization
problem. The implicit function differentiation rule allows the comparative statics of such functions to be
derived. That is to say, the "implicit function differentiation rule" can be used to describe how the "ideal
values" of the control variables change as parameters of the choice problem change.

Within most economic models, there are independent variables that affect the dependent variables, the
interactions of which are what drives our interest in our study. However, the interactions are moderated
by other parameters, whose changes will affect all the models. The examination of how independent
variables affect the independent variables at the critical points is known as comparative statics. A useful
theorem for us here is the Envelope Theorem. The envelope theorem is a general principle describing how
the value of an optimization problem changes as the parameters of the problem change.It simplifies
computation of change in value function (optimized value of objective function) as parameters change.It
implies that the total derivative of fwith respect to a parameter is equal to partial derivative of f
with respect to the parameter.
A maximum (or minimum) value function is an objective function where the choice variables have been
assigned their optimal values. These optimal values of the choice variables are, in turn, functions of the
exogenous variables and parameters of the problem. Once the optimal values of the choice variables have
been substituted into the original objective function, the function indirectly becomes a function of the

36
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
parameters (through the parameters’ influence on the optimal values of the choice variables). Thus, the
maximum value function is also referred to as the indirect objective function.
What is the significance of the indirect objective function? Consider that in any optimization problem the
direct objective function is maximized (or minimized) for a given set of parameters. The indirect
objective function gives all the maximum values of the objective function as these parameters vary.
Hence, the indirect objective function is an “envelope” of the set of optimized objective functions
generated by varying the parameters of the model.
To illustrate, consider the following maximization problem with two choice variables x and y, and one
parameter, α:
Maximize U = f(x, y, α) (1)
The first order necessary condition are
fx (x, y, α) = fy (x, y, α) = 0 (2)
if second-order conditions are met, these two equations implicitly define the solutions
x = x∗(α), y = y∗(α) (3)
If we substitute these solutions into the objective function, we obtain a new function
V (α) = f(x∗(α), y∗(α), α) (4)
where this function is the value of f when the values of x and y are those that maximize f(x, y, α).
Therefore, V (α) is the maximum value function (or indirect objective function). If we differentiate V with
respect to α
∂V ∂ x¿ ∂ y¿
=f x +f y +f (5)
∂α ∂α ∂α α
However, from the first order conditions we know f x = fy = 0. Therefore, the first two terms disappear and
the result becomes
∂V
=f α
∂α
This result says that, at the optimum, as α varies, with x∗ and y∗ allowed to adjust optimally gives the
same result as if x∗ and y∗ were held constant! Note that α enters maximum value function (equation 4) in
three places: one direct and two indirect (through x∗ and y∗). Equations 5 and 6 show that, at the
optimum, only the direct effect of α on the objective function matters. This is the essence of the envelope
theorem. The envelope theorem says only the direct effects of a change in an exogenous variable need be
considered, even though the exogenous variable may enter the maximum value function indirectly as part
of the solution to the endogenous choice variables.
Example
Let f ( x , a )=−x 2 +2 ax +4 a2 be a function in one variable x that depends on a parameter a. For a given
value of a, the stationary points of f is given by
∂f
=−2 x +2 a=0 ⇔ x=a and this is a (local and global) maximum point since f (x; a) is concave
∂x
considered as a function in x. We write x *(a) =a for the maximum point.The optimal value function
f ( x ¿ ( a ) , a )=−a2+ 2 a∗a+4 a2=5 a 2gives the corresponding maximum value.
∂f¿ ∂
= ( 5 a ) =10 a.
2
The derivative of the value function is given by
∂a ∂a

37
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
2 2
On the other hand, we see that f ( x , a )=−x +2 ax +4 a gives
∂f ∂f
=2 x+ 8 a⇒( ) =2 a+8 a=10 a since x ¿ ( a ) =a
∂a ∂ a x=x (a) ¿

The fact that these computations give the same result is not a coincidence, but a consequence of the
envelope theorem for unconstrained optimization problems.
Example: Suppose a firm is producing Q units and wants to maximize its profit. Assume a tax rate t is
imposed on a production of Q . What is the effect of change in the tax rate on total profit?
TR=R (Q)
TC =C ( Q ) +tQ
π=TR−TC
π=R ( Q )−C ( Q )−tQ
∂π ' '
FOC ⇒ =R (Q )−C ( Q )−t=0
∂Q
' '
⇒ R ( Q )=C ( Q ) +t
⇒ MR=MC +t gives the profit maximizing level of output,Q x (t ).
'' ''
SOC ⇒ R ( Q )−C (Q )< 0
dπ x
=−Q ( t)
dt
Example: Assume that the demand and the total cost functions of the monopolist are P=24−3 x and
2
¿ x + 8 x , respectively.
a. Find the maximum profit without any tax.
b. Find the maximum profit with a tax rate of 4 Birr per unit.
c. Determine the rate of change of the profit function with a tax rate of Birr 4 per unit of production
is imposed.
Solution:
a. π=24 x −3 x 2−x 2−8 x =16 x −4 x 2

=16−8 x=0 ⇒ x=2
dx
π max =16
b. π=24 x −3 x 2−( x2 +8 x +4 x ) =12 x −4 x 2

=12−8 x=0⇒ x¿ ( t )=1.5
dx
π max =9

38
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
dπ ¿
4. =−x ( t )=−1.5Constrained Optimization
dt

The last chapter presented a general method for finding the relative extreme of an objective function of
two or more choice variables. In that discussion all the independent variables are independent of one
another, in the sense that the decision made regarding one variable does not impinge upon the choices of
the remaining variables. In many cases, however, optimization problems are the constrained optimization
problem. The idea of constrained optimization is that the choice of one variable often affects the amount
of another variable that can be used. While agents try to maximize or minimize certain objectives, they
face constraints that limit how much they can achieve. Therefore, their optimization is constrained. For
instance, every consumer maximizes his/her utility subject to budget constraint. A firm minimizes the
cost of production with the constraint of production technology. Constrained optimization, therefore, is
the way in which we study the interaction between scarcity and rational choice.

There are two components to any constrained optimization problem: the objective and the constraint.
First, you have to figure out what you are optimizing: it’s a function of some variables known as choice
variables. Second, you have to figure out what are the limitations on these variables.

A constraint is binding if at the optimum the constraint function holds with equality (sometimes
called an equality constraint) giving a boundary solution somewhere on the constraint itself. Otherwise
the constraint is non-binding or slack or ineffective (sometimes called an inequality constraint).

4.1 Static Optimization with Equality Constraints


In this section, we will consider the problem of optimization with equality constraints. For these types of
optimization problems the number of constraints should be less than the number of choice variables. If x o
'
is a critical value of f ( x),i.e., f ( x o ) =0 , f ( x o ) is referred to as a stationary value of the function, and
( x o , f ( x o ) ) is called stationary point.

There are two methods of finding stationary values of multivariate constrained optimization problems:

a. Substitution method
b. Lagrange-multiplier method
a. Substitution Method
It is more elementary, and uses the constraint to reduce the number of variables and transform the
problem to an unconstrained optimization problem in fewer variables.But it is not always feasible.

Let us consider a consumer choice problem: maximizes his utility U ( x 1 , x 2 )=x 1 x 2 +2 x 1 subject to the
budget constraint: 4 x1 +2 x 2=60.

Solve for x2 from budget equation, x 2=30−2 x 1 and substitute in the utility function.

39
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
2
U =32 x 1−2 x1

dU
FOC: =32−4 x 1=0 → x 1=8 ,∧x 2=14
d x1

d2U
SOC: =−4 <0 → constrained maximum.
d x 12

b. Lagrange Multiplier Method

It is more sophisticated and actually introduces a new variable (Lagrange multiplier) to the problem. This
additional variable yields important information about the optimization problem that is more difficult to
derive using the substitution method. This method is preferred when there are several constraints.
The Lagrangean function for n choice variables and m constraints
m
L=f ( x 1 , x 2 ,… , x n)−∑ λi ¿ ¿ ¿
i=1

FOC:

∂L ∂L
=0 , =0
∂ xi ∂ λi

SOC for two choice variables (x and y) and one constraint

| |{
0 gx gy
{negative definite }
d 2 Lis positivedefinite s . t . dg ( ¿ g x dx+ g y dy ) =0 iff ∨H ∨¿ g x L xx
g y L yx
Lxy
L yy
¿0
¿0

|H|is known as borderedHessiandeterminant and it is the bordered discriminant for constrained


function. A positive |H | is to establish a relative maximum while a negative |H | is sufficient to establish
a relative minimum.

Above example revisited


L=x 1 x 2 +2 x1 −λ ( 4 x1 +2 x 2−60 )

FOC:

L x1=x 2 +2−4 λ=0


L x2=x 1−2 λ=0
L λ=− ( 4 x 1 +2 x 2−60 )=0
x 1=8 , x 2=14 , λ=4

40
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
λ measures sensitivity of L to changes in the constant of the constraint.
SOC:

| |
0 4 2
|H|= 4 0 1 =16>0
2 1 0

The critical value of the Lagrange multiplier, λ, is equal to the rate of change of the constrained optimum
with respect to the constraint, and is therefore (approximately) equal to the change in the optimal value
due to a one-unit change in the constraint.

Two of the most basic optimization problems in economics are the problems of minimizing the cost of
producing a certain number of units, and maximizing output, given a fixed production budget.If the cost
of producing q units is minimized at the point (K *, L*), then themarginal product of a dollar of capital
must be equal to the marginal product ofa dollar of labor at that point.

Remember that points obtained using this formula may or may not be a maximum or minimum, since the
first order conditions are only necessary conditions. They only give us candidate solutions.There is
another more subtle way that this process may fail, however. Consider the case where ∇h(x∗) = 0, or in
other words, the point which maximizes f(x) is also a critical point of h(x).

Remember our necessary condition for a maximum ∇f(x∗) = λ ∇ h(x∗), Since ∇ h(x∗) = 0, this implies
that ∇ f(x∗) = 0. However, this the necessary condition for an unconstrained optimization problem, not a
constrained one! In effect, when ∇h(x∗) = 0, the constraint is no longer taken into account in the problem,
and therefore we arrive at the wrong solution.

The generalization of the condition that ∇ h(x∗) ≠ 0 for the case when m = 1 is that the Jacobian matrix
must be of rank m. Otherwise, one of the constraints is not being taken into account, and the analysis
fails. We call this condition the non-degenerate constraint qualification (NDCQ).The point of the NDCQ
was to ensure that binding constraints do not drop out.Note that we only have to check whether the non-
degenerate constraint qualification (NDCQ) holds at points in the constraint set, since points outside the
constraint set are not solution candidates anyways. If we test for NDCQ and find that the constraint is
violated for some point within our constraint set, we have to add this point to our candidate solution set.
The Lagrangean technique simply does not give us any information about this point.

Example
2 2
Find local maxima/minima for f ( x 1 , x 2 ) =x1 +3 x 2, subject to the constraint g ( x 1 , x 2 )=x 1 + x 2 =10 .
[The twosolutions (x1, x2, λ) = (1, 3, 1/2) and (x1, x2, λ) = (−1, −3, −1/2).The bordered Hessian is |H | =
40 in x = (1, 3), and |H |= −40 in x = (−1, −3). So (1, 3) is a local maximum and that (−1, −3) is a local
minimum.]

4.2 Optimization with Inequality Constraints and the Theorem of Kuhn-Tucker


The Karush-Kuhn-Tucker (KKT) is a generalization of the method of Lagrange multipliers to inequality
constraints. This is an alternative, and slightly simpler method for dealing with the common case where

41
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
there are positivity constraints, that is constraints of the form xi ≥ 0 or − xi ≤ 0.The Lagrangean function
for function involving n variables and m constraints
m
L=f ( x 1 , x 2 ,… , x n)−∑ λi ¿ ¿ ¿
i=1

FOC (the Kuhn-Tucker conditions) for maximization

∂L
∂ xi ( )
≤ 0 , x i ≥ 0∧x i
∂L
∂ xi
=0

∂L
∂ λi
≥ 0 , λi ≥0∧x i( )
∂L
∂ λi
=0

FOC (the Kuhn-Tucker conditions) for minimization

∂L
∂ xi ( )
≥ 0 , x i ≥ 0∧x i
∂L
∂ xi
=0

∂L
∂ λi
≤ 0 , λi ≥0∧x i( )
∂L
∂ λi
=0

The other way to minimize f is maximizing – f and also reversing the constraint inequalities by
multiplying every constraint through by −1.

The complementary slackness conditions (CSCs- third column) says that either the inequality holds as
¿ ¿
equality or else the corresponding variable, ( x i or λ i ) is zero. Since we do not know a priori whether or not
∂L
the constraint will be binding at the optimum, we cannot use the condition =0 because this condition
∂λ
is equivalent to g(x , y)−b=0.

When we solve the Kuhn-Tucker conditions together with the inequality constraints, we obtain candidates
for maximum/minimum. Also, notice that the theorem says that satisfying the Karush-Kuhn-Tucker first
order conditions is a necessary condition for an optimum. Therefore, we would be taking a big risk
assuming that just because the KKT conditions are satisfied we have solved the problem.

If λ* = 0, then the constraint can be slack and the FOC turn into the (Unconstrained FOC) and we can
pretty much forget about the Kuhn-Tucker conditions. If λ *> 0, then the constraint must be binding and
the problem turns into the standard Lagrangean. Thus, the Kuhn-Tucker condition provides a neat
mathematical way of turning the problem into either an unconstrained problem or a constrained one.
Unfortunately you typically have to check both cases, unless you know for sure it’s one or the other.

The NDCQ with inequality constraints is the same as the equality constraints, except for the fact that we
only care about the Jacobian matrix of the binding constraints, or the Jacobian for the constraints with λi>
0. (Notice that we cannot tell in advancewhich constraints will be binding, so we need to check all of
them when we check the NDCQ before computing the solution candidates.)

42
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Example 1

Maximize U =xy subject to x + y ≤ 100, x ≤ 40∧x , y ≥ 0.

The Lagrangean is L=xy + λ 1 ( 100−x− y ) + λ2 (40−x )

And the Kuhn-Tucker conditions become

∂L
= y−λ1 −λ2 ≤0 , x ≥ 0∧x L x =0
∂x

∂L
=x −λ1 ≤0 , y ≥ 0∧ y L y =0
∂y

∂L
=100−x− y ≥ 0 , λ1 ≥0∧λ 1 L λ =0
∂ λ1 1

∂L
=40−x ≥ 0 , λ2 ≥ 0∧λ 2 L λ =0
∂ λ2 2

To find a solution we use the trial-and-error approach. We may start by assuming (i) zero/non-zero value
for a choice variable or (ii) one of the constraints to be nonbinding.

For the present example, we assume both x and y to be non-zero and deduce L x =L y =0 from
complementary slackness. This means

y− λ1−λ 2=x−λ 1=0

y− λ2=x

Now assume the ration constraint to be nonbinding in the solution, which implies that λ 2=0. Then we
have x= y , and the given budget B=100 yields the trial solution x= y =50. But this solution violates
the ration constraint x ≤ 40. Hence we must adopt the alternative assumption that the ration constraint is
binding with x ¿=40. The budget constraint then allows the consumer to have y ¿ =40 . Moreover since
¿ ¿
complementary slackness dictates that L x =L y =0, we can readily calculate λ 1=40 , and λ 2=20.

Example 2

Minimize C=( x −4)2 +( y−4)2

subject to 2 x+3 y ≥ 6

−3 x−2 y ≥−12and x , y ≥ 0

The Lagrangean function for this problem is


2 2
L=(x−4 ) +( y−4) + λ1 ( 6−2 x−3 y ) + λ 2 (−12+3 x+2 y )

43
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
The Kuhn-Tucker conditions become

∂L
=2 ( x−4 ) −2 λ 1+ 3 λ2 ≥ 0 , x ≥ 0∧x Lx =0
∂x

∂L
=2 ( y −4 )−3 λ1 +2 λ2 ≥ 0 , y ≥ 0∧ y L y =0
∂y

∂L
=6−2 x−3 y ≤ 0 , λ1 ≥ 0∧λ1 Lλ =0
∂ λ1 1

∂L
=−12+3 x+ 2 y ≤ 0 , λ 2 ≥ 0∧λ2 Lλ =0
∂ λ2 2

∂L ∂ L
If we first try λ 1> 0and λ 2> 0, we must have = =0⇒2 x+3 y =6 and −3 x−2 y=−12. These
∂ λ1 ∂ λ2
24 −6
two equations yield the trial solution x= and y= , which violates the non-negativity restriction on
5 5
y.

∂L ∂ L
Let us next try x >0and y >0 , which would imply = =0 ⇒2 ( x−4 )−2 λ1 +3 λ 2=0 and
∂x ∂y
2 ( y−4 )−3 λ 1+ 2 λ 2=0 . Solve the equations simultaneously and eliminate λ 2 and the result is
4 x−6 y +5 λ1 +8=0. By further assuming λ 1=0, 4 x−6 y=−8. To solve for the two variables,
however, we need another relationship between x and y. For this purpose let’s assume that λ 2 ≠ 0, so that
∂L
=0⇒3 x+ 2 y =12.
∂ λ2

28 36 16
The two equations together yield x= , y= , λ1=0∧λ 2= . Since the solution values for the four
13 13 13
variables are all non-negative and satisfy both constraints, they are acceptable as the final solution.

Maximize f =x− y 2 subject to x 2+ y 2 ≤ 4 , x , y ≥ 0. Ans x=2, y=0 &λ=1/4

Minimize f =2 y−x 2 subject to x 2+ y 2 ≤ 1, x , y ≥ 0. Ans x=1, y=0 &λ=1

Maximize f =xy subject to x 2+ y 2 ≤ 1, x , y ≥ 0. Ans(a) x= -1/√2, y= -1/√2&λ= ½ (maxima) (b) x = 1/√2,


y = 1/√2 &λ = ½ (maxima) (c) x=y=λ=0 (saddle). Since the constraint is binding, we can use the bordered
Hessian method to establish that these solutions really are maxima.

4.3 The General Case: Mixed Constraint


Example

Maximize x− y 2 on the constraint set x 2+ y 2=4 , x ≥ 0 , y ≥0 .

The Lagrangean function for this problem is

44
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
L=x− y 2+ μ ( x 2 + y 2−4 ) + λ 1 x + λ 2 y

The first order conditions become

∂L
=1−2 μx + λ 1=0 , x ≥ 0∧ λ1 x=0
∂x

∂L
=−2 y −2 μy + λ 2=0 , y ≥ 0∧λ2 y=0
∂y

∂L 2 2
=x + y −4=0
∂μ

∂L
=x ≥ 0∧λ1 ≥0
∂ λ1

∂L
= y ≥0∧λ 2 ≥ 0
∂ λ2

∂L ∂L
From , 1+ λ 1=2 μx . Since λ 1 ≥ 0, 1+ λ 1> 0. Therefore, μ>0 and x >0 (and λ 1=0). From ,
∂x ∂y
2 y ( 1+ μ ) =λ2 . Since 1+ μ>0 , either both y and λ 2 are zero or both are positive. From λ 2 y=0, both
∂L ∂L 1
cannot be positive. Therefore, λ 2= y=0 . Now x=2 from and , λ 1=0 by λ 1 x=0 , and μ= by
∂μ ∂ λ 1 4
∂L 1
. This leads to a solution( x , y , μ , λ1 , λ 2 )=(2,0 , ,0,0).
∂x 4

45
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
5. Comparative Static Analysis
Comparative statics is the determination of the changes in the endogenous variables of a model that will
result from a change in the exogenous variables or parameters of that model. Comparative statics is
concerned with sensitivity of a solution to changes in parameters and exogenous variables.

5.1. The Nature of Comparative Static Analysis


Comparative statics is concerned with the comparison of different equilibrium states that are associated
with different sets of values of parameters and exogenous variables. The process of using derivatives to
find out how equilibrium depends on the parameters is called comparative statics. It is a common method
of economic analysis used to compare various points of equilibrium when certain factors change.

In comparative statics:

 We start by assuming a given initial equilibrium state.


 The process of adjustment of variables will be disregarded.
 The possibility of instability of equilibrium is precluded.

The comparative static analysis can be qualitative or quantitative in nature. The quantitative analysis
(change in magnitude) always embraces the qualitative (change in direction).

Most of economic theory consists of comparative statics analysis. A crucial bit of information is the sign
of the changes in the endogenous variables. There is very limited opportunity to establish the signs of the
impacts of changes in macroeconomics or any field that does not have an explicit maximization or
minimization operation involved. But in microeconomics comparative statics is a powerful tool for
establishing important deductions of theories.

Comparative statics are rates of changes in equilibrium values of endogenous variables in response to
changes in exogenous variables. For this reason, the mathematical concept of derivative takes on
preponderant significance in comparative statics.

5.2. Differentiation and its Applications to Comparative Static Analysis


Solving an economic model, we can obtain equilibrium quantities of endogenous variables, which are
functions of parameters and exogenous variables. Once the solutions are obtained, it is easy to find
comparative statics. But it is also possible to obtain comparative statics without solving the model
explicitly.

Market Model

Example 1

This model is the familiar supply and demand framework in a market producing 1 good. We will now use
calculus to look at how changes in the intercept and slopes of demand and supply functions a ffect
equilibrium price and quantity. The emphasis on the word equilibrium is important here as it indicates
that what we interested in is how the solution to the demand and supply functions changes when you
change one of the parameters of the model.

46
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
We’ve solved for the equilibrium price and quantity

We are now interested in analysing the comparative statics of the model.

This tells us P*will increase (decrease) if a increases (decreases).

This tells us P*will decrease (increase) if b increases (decreases).

This tells us P*will increase (decrease) if c increases (decreases).

This tells us P∗will decrease (increase) if d increases (decreases).

Find the partial derivatives of Q∗and check your results using graphs

National Income Model

Let the national income model be:

where Y is national income, C is (planned) consumption expenditure, I is investment expenditure, G is


government expenditure and T is taxes.

We’ve solved for the equilibrium

47
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Government multiplier:

An increase (decrease) in government expenditure will increase (decrease) equilibrium national income.

The effect of a change in non-income tax on equilibrium national income:

An increase (decrease) in non-income tax will decrease (increase) equilibrium national income.

The effect of a change in income tax on equilibrium national income:

An increase (decrease) in income tax will decrease (increase) equilibrium national income.

5.3. Comparative Static of General Function Models


In all the models we’ve dealt with so far, we’ve been able to solve for the reduced form equations (i.e.,
the endogenous variables expressed as a function of exogenous variables and parameters). This has
allowed us to use partial differentiation to figure out the comparative statics of our model. But to be able
to use partial differentiation in the first place, a requirement is that there must be functional
independence among explanatory variables.

When a model contains functions expressed in the general form we use techniques such as total
differentials, total derivatives, implicit function theorem and implicit function rule. In this case we can no
longer use partial differentiation and we cannot actually explicitly solve for the reduced form.

Market Model (a normal good)

The comparative static analysis of this model will be concerned exclusively with how a change in Y 0(=
income) will affect the equilibrium position of the model.

The equilibrium position of the market

48
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
⇒ ⇒

(Single equation)

Thus, the qualitative conclusion is that an increase (decrease) in the income level will always result in an
increase (decrease) in the equilibrium price. If the values which the derivatives of the demand and supply
functions take at the initial equilibrium are known, the above expression can yield a quantitative
conclusion also.

At equilibrium, and . Thus, the effect on the equilibrium quantity can be


determined using the chain rule

Thus, the equilibrium quantity is also positively related to Y 0 in this model. The expression can provide
quantitative conclusion if the values which the various derivatives take at the equilibrium are known.

The above expressions (dP/dY0, dQ/dY0) convey no more than the proposition that an upward shift of the
demand curve will result in a higher equilibrium price and a higher equilibrium quantity which can be
captured by graphic analysis. But the analytical procedure used here can handle many problems of greater
complexity that would prove to be beyond the capacity of the graphic approach.

5.4. Limitations of Comparative Static Analysis


 Comparative statics answers the question: how does the equilibrium change when there is a
change in a parameter or exogenous variable.
 Comparative statics ignores the process of adjustment from the old equilibrium to the new and
also neglects the length of time required in that adjustment process.
 It also disregards that the possibility that new equilibrium may be unstable (unattainable).
 The study of the process of adjustment per se belongs to the field of economic dynamics.

49
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
6. Differential Equations

6.1. Definition and Concepts


This is continuous time analysis. The fundamental idea underlying the methods of solution for differential
equation is the technique of integral calculus.

A differential equation expresses the change in some variable as a result of an infinitesimal change in
another variable. Differential equations can sometimes be solved analytically to yield an equation for the
dependent variable as a function of the independent variable(s) that does not involve derivatives. An
alternative is to approximate the solution by numerical integration.

Boundary conditions can be divided into two categories

 Initial value problems occur when all dependent variables are known at some starting value of the
independent variable.
 Two-point boundary problems occur when some dependent variables are known only at one value
of the independent variable and the rest are known only at some other value of the independent
variable.

6.2. First-Order Linear Differential Equations


The word order refers to the highest order of the derivatives (or differentials) appearing in the differential
equation. A first order linear differential equation will generally take the form

a) Constant Coefficient and Constant Term

The Homogeneous Case

If u and w are constant functions and if w happens to be identically zero and we have
where a is some constant. This differential equation is said to be homogeneous on account of the zero
constant term. . The solution can be written as:

or

The definite solution is established by using initial condition(s) of the time path.

Note that (a) the solution is not a numerical value (b) the solution is free of any derivative (differential)
dy
expressions (c) the derivative term has a unit coefficient.
dt

50
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
The Nonhomogeneous Case

When a nonzero constant takes the place of the zero, we have a nonhomogeneous linear differential
equation . The solution of this equation will consist of the sum of two terms, one of which
is called the complementary solution (y c–is the general solution of the reduced equation), and the other
known as the particular integral (yp-is any particular solution of the complete equation). Since the
particular integral is any particular solution of the complete equation, we can first try the simplest
possible type of solution, namely, y being some constant ( y=k ¿. If that fails to work we should try non-
constant solution ( y=kt , y=k t 2 , y =k e t , … ¿.

or

or

The use of the initial condition to definitize the arbitrary constant is – and should be – undertaken as the
final step, after we have found the general solution to the complete equation.

dy
Example 1 Solve the equation +2 y=6, with the initial condition y ( 0 )=10 . The solution is
dt
y ( t ) =7 e−2 t +3.

dy
Example 2 Solve the equation + 4 y=0, with the initial condition y ( 0 )=1. The solution is
dt

y ( t ) =e−4 t

dy
Example 3 Solve the equation =2, with the initial condition y ( 0 )=5 . The solution is
dt

y ( t ) =5+2 t

51
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
To check the validity of solutions to differential equations make sure that (1) differentiation of the time
path y (t ) is consistent with the given differential equation and (2) that the definite solution satisfies the
initial condition.

Exercise

b) Variable Coefficient and Variable Term

In the more general case of a first-order linear differential equation u(t) and w(t)
represent a variable coefficient and a variable term, respectively. How do we find the time path y(t) in this
case?

The Homogeneous Case

When w(t)=0, the general solution:

dy
Example 1 Find the general solution of the equation +3 t 2 y =0. The
dt

The Nonhomogeneous Case

When w(t) ≠ 0 the solution is found via the concept of exact differential equations.

The general solution is .

dy
Example 2 Find the general solution of the equation +2 ty=t . The general solution is
dt

y(t) .

52
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
dy
Example 3 Solve the equation + 4 ty=4 t . The general solution is y(t)
dt

Exercise

6.3. Exact Differential Equations


In general, a differential equation Mdy + Ndt =0 is exact if and only if there exists a function F(y, t) such
∂F ∂F
that M = and N= . By Young’s theorem, we can state that the equation is exact if and only if
∂y ∂t
∂M ∂ N
= .
∂t ∂y

For instance, F ( y , t ) = y 2 t +k , the total differential is dF=2 ytdy+ y 2 dt . Thus, the differential equation
2 dy y 2 ∂M
=2 y=
∂N
2 ytdy+ y dt=0or + =0. Since , the exactness of the equation is duly verified.
dt 2 yt ∂t ∂y

The general solution for exact differential equation [ dF ( y , t )=0 ] should clearly be in the form
F ( y , t ) =c .

Method of solution

∂F
How do we pin down the exact form of thisψ (t) term? By utilizing the fact that N= .
∂t

Example 1 Solve the exact differential equation 2 ytdy+ y 2 dt=0

53
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Example 2 Solve the equation( t+ 2 y ) dy + ( y +3 t ) dt=0. The solution is
2

Integrating Factor

Sometimes an inexact differential equation can be made exact by multiplying every term of the equation
by a particular common factor. Such a factor is called an integrating factor.

Example 3 The differential equation2 tdy+ ydt=0 . yis an integrating factor for the differential equation.

Solution of First Order Linear Differential Equation

dy
The general first order linear differential equation +uy =w , can be expressed as dy + ( uy−w ) dt =0
dt
has the integrating factor . Let I be the (yet unknown) integrating factor.

which can then be solved by the four step procedure.

54
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Exercise

6.4. Second-Order Linear Differential Equations


a) Constant Coefficient and Constant Term

The relevant differential equation is where a1, a2 and b are all constants. If b
is identically zero, we have a homogeneous equation, but if b is a non-zero constant, the equation is non
homogeneous.

Let’s determine the solution for the non homogeneous version and the solution of the homogeneous
version will emerge automatically as a by-product. The general solution of the complete equation is given
as a sum of complementary function and particular integral.

The Particular Integral

b
If y=k , y ' ( t )= y ' ' ( t )=0 , then the desired particular solution is y p= (where a2≠0). What if a2=0, so
2 a
b
that the expression y p= is not defined? In such a situation, since the constant solution for y p fails to
a2
work, we must try some non-constant ( y=kt , k t 2) form of solution.

Example 1 Find the particular integral of the equation y ' ' ( t )+ y ' ( t )−2 y=−10. By assuming that y=k ,
the particular integral is y p=5 .

55
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Example 2 Find the particular integral of the equation y ' ' ( t )+ y ' ( t )=−10. By assuming that y=kt , the
particular integral y p=−10t . y pis a moving equilibrium because it is a function of time.

Example 3 Find the particular integral of the equation y ' ' ( t )=−10 . By assuming that y=k t 2, the
particular integral y p=−5t 2. y p is a moving equilibrium because it is a function of time.

As a general rule, if the RHS of the equation has a function which is already in the kernel (i.e., one of the
homogeneous solutions), we multiply by t until the resulting function is no longer a solution to the
homogeneous equation.

The Complementary Function

The complementary function of is defined to be the general solution of its


reduced (homogeneous) equation .

If we adopt the trial solution y= A ert , we must also accept y ' ( t )=rA e rt , y ' ' ( t )=r 2 A e rt as the
derivatives of y.

rt
Since e can never be zero, we must either let A=0 or =0. Because the value of the
(arbitrary) constant A is to be definitized by use of initial conditions, however, we cannot simply set A=0
at will.

The equation is known as characteristic (auxiliary) equation. If the equation is in a


normalized form (i.e., the coefficient of r 2 is 1),

r 1 +r 2=−a1∧r 1 r 2=a2
,

r1 t r2 t
In effect, there are two solutions which will work, namely, y 1= A 1 e and y 2= A 2 e . But we want only
one general solution. We can simply take their sum, y 1 + y 2, as a general solution for the homogeneous
equation. Thus, the complementary function for the complete equation can, in general, be written as
y c = y 1 + y 2.

As far as values of r 1∧r 2 are concerned, three possible cases can arise,

Case 1 (distinct real roots)


2
When a 1 > 4 a2 is a real number, the two roots r1 and r2 will take distinct real values.

Because the two roots are distinct, the two exponential expressions must be linearly independent;
consequently, A1 and A2 will always remain as separate entities and provide us with two constants.

56
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Example 4 Solve the differential equation y ' ' ( t )+ y ' ( t )−2 y=−10, where y ( 0 )=12 and

y ' ( 0 )=−2.

The particular integral of this equation has already been found to be y p=5 in example 1. Let us find the
complementary function. The characteristic roots are r 1=1∧r 2=−2 . The complementary function is
y c = A 1 e t + A2 e−2 t . Therefore, the general solution can be written as y ( t ) =A 1 et + A 2 e−2 t +5 .

y ( 0 )= A 1+ A 2+5=12 ⇒ A 1 + A 2=7 A1=4∧ A2=3


'
y ( 0 )= A1−2 A2 =−2

Thus, the definite solution of the differential equation is y ( t ) =4 et + 3 e−2 t + 5

Case 2 (repeated real roots)

2 −a1
When a 1 =4 a2, the characteristic roots take an identical value: r 1=r 2= .
2

If we attempt to write the complementary function as y c = y 1 + y 2, the sum in this case collapses into a
rt rt rt rt
single expression ( y c = A 1 e + A2 e = ( A1 + A2 ) e = A 3 e ) leaving us with only one constant. This is not
sufficient to lead us from a second order differential equation back to its primitive function. The only way
out is to find another eligible component term for the sum – a term which satisfies the homogeneous
rt
equation and yet which is linearly independent of the term A3 e , so as to preclude such “collapsing”. In
2 a1 rt
as much as we have a 1 =4 a2and ¿− , A 4 t e qualifies as a solution.
2
rt rt
yc= A 3 e + A 4 t e

Example 5 Solve the differential equation

y ' ' ( t )+ 6 y ' ( t )+ 9 y=27 where y ( 0 )=5 and y ' ( 0 )=−5 .

The root of the characteristic equation is r =−3. Thus, the complementary function may be written as
y c = A 3 e + A 4 t e . The particular integral is y p=3 . The general solution of the complete equation is
−3 t −3 t

y ( t ) =A 3 e + A 4 t e + 3. By using the initial conditions A3 =2and A 4=1. So the definite solution can
−3t −3 t

be written as y ( t ) =2 e−3 t + t e−3 t +3 .

Case 3 (complex roots)

This occurs when a 1 < 4 a2 and involves square root of negative numbers. If the roots are r 1=α +iβ and
2

r 1=α −iβ , the general solution is y ( t ) =e αt ( A1 cos βt+ A 2 sin βt ).

Exercise

57
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
b) Variable Term

The complementary function is not affected by the presence of a variable term but we must modify our
particular integral.

Method of Undetermined Coefficients

This method is applicable to constant-coefficient variable-term differential equations as long as the


variable term and its successive derivatives together contain only a finite number of distinct types of
expression (apart from multiplicative constants).

Example 1 Find the particular integral of y ' ' ( t )+ 5 y ' ( t ) +3 y=6 t 2−t−1

By definition, the particular integral is a value of y satisfying the given equation, i.e., a value of y that will
make the left side identically equal to the right side regardless of the value of t.
2 '' ''
If we let y ( t ) =B 1 t + B2 t+ B3, then y ( t )=2 B1 t + B2 and y ( t )=2 B1 . Substituting these into the
original equation and collecting like terms, we get
2
¿=( 3 B 1) t + ( 10 B1 +3 B2 ) t+(2 B1 +5 B 2+3 B3 )

58
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
And when this is equated term by term to the right side,

3 B 1=6 B1=2

2 B 1+3 B2=−1 ⟹ B2=−7

2 B 1+5 B2 +3 B3 =−1 B3=10


2
Thus, the desired particular integral can be written as y p=2t −7 t+ 10.

2
Find the particular integral of y ' ' ( t )+ 5 y ' ( t ) =6 t 2 −t −1. y p=t(B1 t + B2 t + B3)

−2 −4 t
Find the particular integral of y ' ' ( t )+ 3 y ' ( t ) −4 y =2 e−4 t , y p=Bt e−4 t and y p= te
5

Exercise Solve the differential equation y ' ' ( t )−5 y ' ( t )+ 6 y=2 t+3 . The particular solution is assumed to
be of the form y p=ct+ d . The general solution of the differential equation is
3t 2t 1 7
y ( t ) =A 1 e + A 2 e + t + .
3 9

Exercise

6.5. Economic Applications of Differential Equations


Dynamics of Market Price

59
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
7. Difference Equations

7.1. Definitions and Concepts

A difference equation expresses the change in some variable as a result of a finite change in another
variable. Difference equations allow direct, exact integration to calculate the values of dependent
variables at all values of the independent variable. Difference equations imply a “synchronicity” to
changes in variables.

When time is taken to be a discrete variable so that the variable t is allowed to take integer values only,
the concept of the derivative obviously will no longer be appropriate. In this light, it becomes more
convenient to interpret the values of t as referring to periods – rather than points of time.In the analytical
sense, a period is merely a length of time that elapses before variable y undergoes a change. The pattern
Δy
of change is now represented by the difference quotient . The symbol Δ, meaning difference can be
Δt
interpreted as a directive to take the first difference of (y). As such, it constitutes the discrete-time
d
counterpart of the operator symbol .
dt

orΔyt= yt- yt-1or Δyt= yt+2 - yt+1

Even though difference equations have their name from difference expressions such as Δyt, there are
alternate equivalent forms of such equations which are completely free of Δ expressions and which are
more convenient to use. E.g. , . Analogous to the differential equations,
difference equations can be linear or non-linear, homogeneous or non-homogeneous, and of the first or
second (or higher) orders.

7.2. First–Order Linear Difference Equations

How do we solve a difference equation?

The first order difference equation describes the pattern of change of y between two consecutive periods
only.

Iterative Method

This is a method of determining the time path from repeated application of the pattern of change specified
in the difference equation.

Example 1 Find the solution of the difference equation ∆ y t =2, assuming an initial value of y 0=15 . The
solution is y t = y 0 +2 ( t )=15+ 2t .

The process of iteration is crude – it corresponds roughly to solving simple differential equations by
straight integration – but it serves to point out clearly the manner in which a time path is generated.

60
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Example 2 Solve the difference equation ∆ y t =−0.1 y t when an initial value is y 0. The solution is
t
y t =( 0.9) y 0 .
t
n
Example 3 Solve the homogeneous difference equation m y t+1 −n y t=0. The solution is y t =( ) y 0
m
t
which can be rewritten as y t =A b .

General Method

Suppose that we are seeking the solution to the first – order difference equation y t +1 +a y t =c where a
andc are two constants. The general solution will consist of the sum of two components: a particular
solution yp, which is any solution of the complete nonhomogeneous equation and a complementary
function yc, which is the general solution of the reduced equation, y t +1 +a y t =0.

The yp component again represents the intertemporal equilibrium level of y, and the y c component, the
deviations of the time path from that equilibrium. The sum of y cand yp constitutes the general solution,
because of the presence of an arbitrary constant. As before, in order to definitize the solution, an initial
condition is needed.
t
For the trial solution y c = A b to work for the homogeneous equation, b=−a and then the
t t
complementary function can be written as y c = A b = A (−a) .

c
If the trial solution y p=k can work out, then the particular solution can be written as y p= (a≠−1) .
1+a
The particular solution will be y p=ct if a=−1.

Note that the yt+1 term has a unit coefficient. If a given difference equation has a non-unit coefficient for
this term, it must be normalized before using the solution formula.

7
Example 4Solve the first order difference equation y t +1−5 y t =1if y 0= . The definite solution is
4
t 1
y t =2 (5 ) − .
4

Exercise

61
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
The Dynamic Stability of Equilibrium

7.3. Second-Order Linear Difference Equations

A second order difference equation is one that involves an expression ∆ 2 y t , called the second difference
of yt.

A second order difference equation is an equation involving a two-period time lag in the variable.

a) Constant Coefficients and Constant Term

Particular Solution

The particular integral is

Example 1 Find the particular integral of y t +2−3 y t +1 +4 y t =6 . The solution is y p=3 .

Example 2 Find the particular integral of y t +2 +3 y t+ 1−2 y t =12. The solution is y p=4 t .

Complementary Function

62
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
Let us try the solution of the form y t =A bt

, ,

Three possible situations may be encountered in regard to characteristic roots depending on the square
root expressions.

Case 1 (distinct real roots)

Example 3 Find the solution of y t +2 + y t +1−2 y t=12 when y 0=4 and y 1=15 . The definite solution is
t
y t =3+(−2) + 4 t .

Case 2 (repeated real roots)

Example 4 Find the solution of y t +2 +6 y t +1+ 9 y t=4 . The general solution is


t t 1
y t =A 3 (−3) + A 4 t(−3) + .
4

Case 3 (complex roots)

Exercise

b) Variable Term

t t 1 t
Example 1 Find the particular solution of y t +2− y t +1−3 y t =7 . Let y p=B(7) , y p= (7)
53

63
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
2 2
Example 2Find the particular solution of y t +2 +5 y t+1 +2 y t =t . Let y p=B1 t + B2 t+ B3,
1 2 7 13
y p= t − t + . This can be generalized for higher order polynomial
8 32 256
2 n
y t =B 0 +B 1 t + B 2 t + …+B n t .

7.4. Economic Applications

7.4.1. The Cobweb Model

7.4.2. The Dynamic Stability of Equilibrium

7.4.3. Market Model with Inventory

7.4.4. Dynamic Multiplier

64
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S
8. Dynamic optimization

8.1. Discrete time optimization

8.2. Continuous time optimization

8.3. Optimal control theory

8.4. Economic applications

65
MATHEMATHICAL ECONOMICS BY Ahmed A. and Mandefro. S

You might also like