Finance 30210: Managerial
Economics
Optimization
Optimization deals with functions. A function is simply a
mapping from one space to another. (that is, a set of instructions
describing how to get from one location to another)
B
f :A B
f
Is the range
is a function
A
f
A
B
Is the domain
For example
For
0 x5
Domain
f x 3x 5
Range
Function
A 0,5
3
x A
5 y 20
B [5,20]
f 3 14
14
y
y 3x 5
20
Y =14
0 x5
Range
For
5
0
Domain
X =3
x
5
max 3x 5
y
20
0 x5
Range
Here, the optimum occurs at
x = 5 (y = 20)
5
0
Domain
x
5
Optimization involves finding the maximum value for y over an allowable
domain.
What is the solution to this optimization problem?
1
max
5 x
y
0 x 10
5
10
x
There is no optimum because
f(x) is discontinuous at x = 5
What is the solution to this optimization problem?
max 2 x
y
0 x6
12
There is no optimum because
the domain is open (that is,
the maximum occurs at x = 6,
but x = 6 is NOT in the
domain!)
x
0
6
What is the solution to this optimization problem?
y
max
x
x0
12
There is no optimum because
the domain is unbounded (x is
allowed to become arbitrarily
large)
x
0
The Weierstrass theorem provides sufficient conditions for an optimum
to exist, the conditions are as follows:
y
f x is continuous
max f ( x)
over the domain of
x
The domain for
x
is closed and bounded
x
Finding maxima/minima involves taking derivatives….
Formally, the derivative of
f x
is defined as follows:
f ( x x) f ( x)
f ' x lim
x 0
x
All you need to remember is the derivative represents a slope (a
rate of change)
Actually, to be more accurate, the derivative represents a
trajectory
Graphically…
y
y
f x x
f x
f x
x
x x
f ( x x) f ( x)
slope
x
x
x
x
Now, let the change
in x get arbitrarily
small
f ( x x) f ( x)
f ' x lim
x 0
x
First Order Necessary Conditions
If
x* is a solution to the optimization problem
max f ( x)
min f ( x )
or
x
x
then
f ' ( x*) 0
f x
f ' x 0
f ' x 0
x*
f x
f ' x 0
x
f ' x 0
f ' x 0
x*
x
f ' x 0
Useful derivatives
Logarithms
Linear Functions
f ( x) Ax f ' ( x) A
Example:
f ( x) 4 x
f ' x 4
A
f ( x) A ln( x) f ' ( x)
x
Example:
Exponents
f ( x) Ax n f ' ( x) nAx n1
Example:
f ( x) 3x 5
f ' x 15 x 4
f ( x) 12 ln x
12
f ' x
x
An Example
Suppose that your company owns a corporate jet. Your
annual expenses are as follows:
You pay your flight crew (pilot, co-pilot, and navigator a
combined annual salary of $500,000.
Annual insurance costs on the jet are $250,000
Fuel/Supplies cost $1,500 per flight hour
Per hour maintenance costs on the jet are proportional to the
number of hours flown per year.
Maintenance costs (per flight hour) = 1.5(Annual Flight Hours)
If you would like to minimize the hourly cost of your jet, how
many hours should you use it per year?
Let x = Number of Flight Hours
$750,000
Hourly Cost
$1500 $1.5 x
x
$750,000
min
$1500 $1.5 x
x 0
x
First Order Necessary Condition
$750,000
750,000
$1.5 0 x
707hrs
2
x
1.5
Hourly Cost ($)
An Example
Annual Flight Hours
How can we be sure we are at a maximum/minimum?
For a maximization…
For a minimization…
f x
f x
f ' x 0
f ' x 0
f ' x 0
x*
x
f ' x 0
f ' x 0
x*
x
f ' x 0
f ' ' ( x*) 0
f ' ' ( x*) 0
Slope is decreasing
Slope is increasing
Let x = Number of Flight Hours
$750,000
min
$1500 $1.5 x
x
x
First Order Necessary Conditions
$750,000
750,000
$1.5 0 x
707hrs
2
x
1.5
Second Order Necessary Conditions
$1,500,000
0
3
x
For X>0
Suppose you know that demand for your product depends on the
price that you set and the level of advertising expenditures.
Q p, A 5,000 10 p 40 A pA .8 A2 .5 p 2
Choose the level of advertising AND price to
maximize sales
When you have functions of multiple variables, a partial derivative is the
derivative with respect to one variable, holding everything else constant
max 5,000 10 p 40 A pA .8 A2 .5 p 2
p 0 , A 0
First Order Necessary Conditions
Q p ( p, A) 10 A p 0
QA ( p, A) 40 p 1.6 A 0
With our two first order
conditions, we have two
variables and two unknowns
10 A p 0
A p 10
Q p ( p, A) 10 A p 0
QA ( p, A) 40 p 1.6 A 0
40 p 1.6 A 0
40 p 1.6 p 10 0
24 .6 p 0
p 40
A 50
How can we be sure we are at a maximum?
f ' ' ( x*) 0
its generally sufficient to see if all the second derivatives are negative…
Q p ( p, A) 10 A p
QA ( p, A) 40 p 1.6 A
Q pp ( p, A) 1 0
QAA ( p, A) 1.6 0
Practice Questions
1) Suppose that profits are a function of quantity produced and can be
written as
10 20Q 2Q
2
Find the quantity that maximizes profits
2) Suppose that costs are a function of two inputs and can be written as
C 4 3 X1 4 X 2 2 X 3 X X1 X 2
2
1
2
2
Find the quantities of the two inputs to minimize costs
Constrained optimizations attempt to maximize/minimize a function subject
to a series of restrictions on the allowable domain
max f x, y
x, y
subject to
g x,y 0
To solve these types of problems, we set up the lagrangian
( x, y) f x, y g x, y
Function to be
maximized
Constraint(s)
Multiplier
To solve these types of problems, we set up the lagrangian
( x) f x g x
0
(x ) f x
We know that at the
maximum…
(x )
f x
x*
x
x ( x) 0
g x 0
Once you have set up the lagrangian, take the derivatives and set
them equal to zero
( x, y) f x, y g x, y
First Order Necessary Conditions
x ( x, y ) f x x, y g x x, y 0
y ( x , y ) f y x , y g y x, y 0
Now, we have the “Multiplier” conditions…
0
g x, y 0
g x, y 0
Example: Suppose you sell two products ( X and Y ). Your profits as a
function of sales of X and Y are as follows:
( x, y) 10 x 20 y .1( x 2 y 2 )
Your production capacity is equal to 100 total units. Choose X and Y
to maximize profits subject to your capacity constraints.
x y 100
The key is to get
the problem in the
right format
f x, y
max 10 x 20 y .1( x 2 y 2 )
x 0, y 0
subject to
100 x y 0
g x, y
Multiplier
The first step is to create a Lagrangian
( x, y) 10 x 20 y .1( x y ) (100 x y)
2
Objective Function
2
Constraint
Now, take the derivative with respect to x and y
( x, y) 10 x 20 y .1( x y ) (100 x y)
2
2
First Order Necessary Conditions
x ( x, y ) 10 .2 x 0
y ( x, y ) 20 .2 y 0
“Multiplier” conditions
0
100 x y 0
(100 x y ) 0
First, lets consider the possibility that
lambda equals zero
10 .2 x 0
20 .2 y 0
10 .2 x 0
x 50
0
100 x y 0
100 x y 0
20 .2 y 0
y 100
100 50 100 50 0
Nope! This can’t work!
The other possibility is that lambda is
positive
10 .2 x 0
20 .2 y 0
10 .2x
20 .2 y
10 .2 x 20 .2 y
y x 50
y x 50
0
100 x y 0
100 x y 0
100 x y 0
100 x x 50 0
50 2 x 0
x 25
y 75
Lambda indicates the marginal value of relaxing the constraint. In this
case, suppose that our capacity increased to 101 units of total
production.
10 .2 x 20 .2 y
y 75
x 25
50
Assuming we respond optimally, our profits increase by $5
Example: Postal regulations require that a package whose length plus girth
exceeds 108 inches must be mailed at an oversize rate. What size package
will maximize the volume while staying within the 108 inch limit?
Z
Girth = 2x +2y
Volume = x*y*z
X
Y
xyz
x 0, y 0, z 0
max
subject to
2 x 2 y z 108
xyz
x 0, y 0, z 0
max
subject to
2 x 2 y z 108
First set up the lagrangian…
( x, y, z ) xyz (108 2 x 2 y z )
Now, take derivatives…
x ( x, y, z ) yz 2 0
y ( x, y, z ) xz 2 0
z ( x, y, z ) xy 0
Lets assume lambda is positive
yz 2 0
xz 2 0
xy 0
108 2 x 2 y z 0
xy
yz 2 0
yz 2 xy 0
z 2x
xz 2 0
xz 2 xy 0
z 2y
0
108 2x 2 y z 0
z 2x
z 2y
2 x 2 y z 108
z z z 108
z 108
y 18
xy 324
x 18
z 36
Suppose that you are able to produce output using capital (k) and labor (L) according
to the following process:
yk l
.5 .5
Labor costs $10 per hour and capital costs $40 per unit. You want to minimize the
cost of producing 100 units of output.
TC 10l 40k
min 10l 40k
l 0,k 0
subject to
k .5l .5 100
Minimizations need a minor adjustment…
( x, y) f x, y g x, y
A negative sign instead of a positive sign!!
So, we set up the lagrangian again…now with a negative sign
(k , l ) 10l 40 (k .5l .5 100)
Take derivatives…
l (k , l ) 10 .5k l
.5 .5
0
k (k , l ) 40 .5k .5l .5 0
Lets again assume lambda is positive
0
k l 100
.5 .5
10 .5k l
.5 .5
0
40 .5k l 0
.5 .5
40 .5k .5l .5 0
80k .5l .5
10 .5k .5l .5 0
20l .5 k .5
k .5l .5 100
.5
20l k
l 4k
.5
.5 .5
80k l
k .5 4k 100
2k 100
.5
k 50
l 200
Suppose that you are choosing purchases of apples and bananas. Your total
satisfaction as a function of your consumption of apples and bananas can be written
as
UA B
.4
.6
Apples cost $4 each and bananas cost $5 each. You want to maximize your
satisfaction given that you have $100 to spend
4 A 5B 100
.4
max A B
.6
A 0 , B 0
subject to 100 4 A 5B 0
max A.4 B.6
A 0 , B 0
subject to 100 4 A 5B 0
First set up the lagrangian…
( A, B) A B (100 4 A 5B)
.4
.6
Now, take derivatives…
A ( A, B) .4 A.6 B.6 4 0
B ( A, B) .6 A.4 B .4 5 0
Lets again assume lambda is positive
.4 A B 4 0
.6
.4
.6 A B
.6
.4
0
4 A 5B 100
5 0
.6 A.4 B .4 5 0
.12 A.4 B .4
.4 A.6 B.6 4 0
.1A.6 B.6
.1A.6 B .6 .12 A.4 B .4
B 1.2 A
4 A 5B 100
4 A 51.2 A 100
A 10
B 12
Suppose that you are able to produce output using capital (k) and
labor (l) according to the following process:
yk l
.5 .5
The prices of capital and labor are p and w respectively.
Union agreements obligate you to use at least one unit
of labor.
Assuming you need to produce
y
units of output, how would
you choose capital and labor to minimize costs?
Non-Binding Constraints
min pk wl
k ,l
subject to
y k .5l .5
l 1
Just as in the previous problem, we
set up the lagrangian. This time we
have two constraints.
Will hold with equality
(k , l ) pk wl (k .5l .5 y ) (l 1)
Doesn’t necessarily hold with equality
(k , l ) pk wl (k .5l .5 y ) (l 1)
First Order Necessary Conditions
.5 .5
(
k
,
l
)
p
.5
k
l 0
k
(k , l ) w .5k l
.5 .5
l
0
y k .5l .5 0
l 1 0
(l 1) 0
0
l -1 0
Case #1:
0
l 1
Constraint is non-binding
First Order Necessary Conditions
w .5k
0
p . 5 k .5 l . 5 0
.5 . 5
l
w k
p l
w
k l
p
y k .5 l . 5 0
w 2
y kl l l
p
w
p
p
l y 1
w
w py 2
Case #2:
l 1
0
Constraint is binding
First Order Necessary Conditions
w .5k 0
p .5k .5 0
.5
yk 0
.5
2 p k 2 py
ky
2
w py 2 py 2
w
Constraint is Binding
w py 2
k y2
l 1
Constraint is Non-Binding
p
ly
w
w
k y
p
p
Try this one…
You have the choice between buying apples and bananas. You utility
(enjoyment) from eating apples and bananas can be written as:
U A, B A 2 B
.5
The prices of Apples and Bananas are given by
PA
and
PB
Maximize your utility assuming that you have $100 available to
spend
max A 2 B
A, B
.5
(Objective)
subject to
PA A PB B $100
A0 B0
(Income Constraint)
(You can’t eat negative apples/bananas!!)
( A, B) A.5 2B ($100 PA A PB B) 1 A 2 B
Objective
Income
Constraint
Non-Negative
Consumption
Constraint
( A, B) A.5 2B ($100 PA A PB B) 1 A 2 B
First Order Necessary Conditions
A ( A, B ) .5 A.5 PA 1 0
B ( A, B ) 2 PB 2 0
$100 PA A PB B 0
We can eliminate some of the multiplier conditions with a little reasoning…
1. You will always spend all your income
0
2. You will always consume a positive amount of apples
2 0
B0
2 B 0
1 0
Case #1: Constraint is non-binding
B 0 2 0
First Order Necessary Conditions
A ( A, B ) .5 A.5 PA 0
B ( A, B ) 2 PB 0
$100 PA A PB B 0
.5
.5 A
PA
2
PB
$100 PA A
B
PB
B 0 PB 40 PA
Case #1: Constraint is binding
B 0 2 0
First Order Necessary Conditions
A ( A, B) .5 A.5 PA 0
B ( A, B) 2 PB 2 0
$100 PA A 0
.5 A.5
PA
$100
A
PA
2 PB 2
2 0 PB 40 PA
PB
Constraint is Binding
B0
PB 40 PA
$100
A
PA
Constraint is Non-Binding
PB
A .25
PA
2
$100 PA A
B
PB
PA