Topic II: Multivariate Calculus

Chapter IV: Calculus of functions of n variables

In general, calculus is the mathematics of change. One of the main applications

of calculus is determining how one variable changes in relation to another. The
purpose of the derivative is to express in a convenient way how a change in the
level of one variable (say X) determines a change in the level of another variable
(say Y). Much of economics is in fact concerned with this sort of analysis. For
example, we study how a change in a firm's output level affects its costs and how
a change in a country's money supply affects the rate of inflation.

We will start by recalling the definition and the techniques of the derivative of a

function in just one variable in the form y F x .

Definition of the derivative:

The derivative of a function y f x at the point x1 , y1 is the slope of the
tangent line at this point. It represents the rate of change in the variable y due to
a change in x .

The first derivative of the function y f x is denoted by f / x , y / and

dy 1 1
(in addition to many other notation such that f x , y , Dx y ) which all refer
dx
to the derivative of y with respect to x w.r.t x .

Using the initial definition, the derivative of y f x is

f x x f x
f/ x lim ,
x 0 x
where x refers to the change in the independent variable x .

82
 Recall that the first derivative of the function f x represents in economics the

marginal function. For example, c / x is the marginal cost function that

determines the change in the cost due to one-unit change in the production level

x . For example, c / 6 is the cost of the unit N 7 .

Higher order derivatives also have important applications in economics. For any
positive n , the nth derivative of the function f x is obtained by

differentiating successively n times. It is denoted by:

n n dny
y , f x or
dx n

d2y 2
For example, y y " , which means that we will differentiate twice.
dx 2
Rules and techniques of differentiation:
The initial definition was used to derive some rules of differentiation that greatly
simplify the process of computing the derivative of a function y f x .
Recall the following rules

The constant rule: If f x c (where c is a constant), then f / x 0

Ex: If - f x 3, f/ x 0,

- f x 2, f/ x 0

1
- f x , f/ x 0
2

83
 The power rule: If f x xn , f / x nx n 1 , for any real number n .
Ex: If - f ( x) x4 , f / ( x) 4 x3
2
- f ( x) x , f/ x 2x 3

1 2
/ 1
- f x x 3
, f x x 3
3
/
- f x x , f x 1

The constant multiple rule: If G x cf x , G/ x cf / x

Ex: If - f x 3x 6 , f / x 3(6 x 5 ) 18x 5

x2 1 2 2
- f ( x) , f/ x 2x 3
x 3

3 3 3 3x 3
1 3
7 / 7 7
-f x , f x 7x 2
, f x x 2
3
x 2
2x 2

The exponential rule:If f x ex , f / x e x for any real numbers x, x R

Ex: If - f x 4e x f/ x 4e x
1
The logarithmic rule: If f ( x ) ln x , f/ x , for any positive
x
number x, x R
1 4
Ex: If - f x 4ln x f/ x 4x
x x
The sum rule: If h x f x g x , h/ x f/ x g/ x

Ex: If - f ( x) x 2 4 x3 , f/ x 2x 3 12 x2

- f ( x) 6x3 15x 3 6ln x ,

2
f / x 18x2 10x
1
3 6
x

84
 The product rule:

If h x f x x g x , h/ x f x g/ x g x f/ x

Ex: If - f x x 2e x f/ x x 2e x e x 2 x ex x2 2x

1 8
-f x 4 x 2 ln x f / x 4 x 2 ln x 1 4 4ln x
x x
1
x
If - f x x 3 5e ln x x 2
3 5e x ln x ,

1 1
/ x 1 x 1
f x x 2
3 5e 5e ln x x 2
x 2

The quotient rule:

f x g x f/ x f x g/ x
If h x , h/ x 2
if g x 0
g x g x

x2 3 /
x 2x x2 3 1 2 x2 x2 3
Ex: If - f x , f x
x x2 x2
x2 3
. 1 3 2
x 2
x

x3 ex 2 3x 2 x3 e x
-f x x
, f/ x 2
e 2 ex 2

x 2 3e x xe x 6
2
ex 2

85
The chain rule:

If y f u and u g x , then

y f g x h x , where h x is a composite function.

Using the chain rule:

dy dy du
h/ x f / u .g / x ,
dx du dx

Ex: If y u 3 3u 2 1, where u x2 1

dy
y/ 3u 2 6u 2 x , substituting by u by
dx
2
3 x2 1 6 x2 1 2x

2
3 x2 1 2 x 6 x2 1 2 x

We may reach the same result by substituting first u by x 2 1 and then

3
getting the derivative of the outer function multiplied by the derivative of the
2
inner function x 1 . Considering the composite function

3 2
y x2 1 3 x2 1 1

dy 2
y/ 3 x 2 1 2 x 3x2 x 2 1 2 x
dx
2
3 x2 1 2 x 6 x2 1 2 x The same result.

86
 1
Ex: If - f x x 3 3x 2 x 3 3x 2 2

1
1 3
f /
x x 3x 2 2 3x 2 3
2
3x 2 3
2 x 3 3x 2

1 3x
-f x ln 2 x3 4 , f/ x 6x
2 x3 4 x3 2

General rules of differentiation:

These rules facilitate the application of the chain rule

n n 1
1- General power rule: If y f x , y/ n f x f/ x

3 2
Ex: If - f x 2 x4 x , f/ x 3 2x4 x 8 x3 1

4 1
- f x 4 2x2 5
2 x2 5

2
f/ x 4 1 2 x2 5 4x

2 16 x
16 x 2 x 2 5 2
2 x2 5

87
 f x f x
2- General exponential rule: If y e , y/ e f/ x
4 4 4
2x 3 / 2x 3 3 2x 3 3
Ex: if - y e , y e 4 2x 3 2 8e 2x 3

1
x 3 ex
1
2 3
/ x
1
2 3 1
- y e y e x 2
2

1 x 3
e
2 x

1
3- General logarithmic rule: If y ln g x , y / g/ x
g x
1
2 2 3
Ex: If -y 3 ln 5x 2x 1 ln 5x 2x 1

2
/ 1 1
y ln 5 x 2 2 x 1 3
10 x 2
3 5x2 2x 1
3
- y ln 2 x 2 x 7 y 3ln 2 x 2 x 7 (properties of ln)

3
y/ 2
4x 1
2x x 7
1 2
or y/ 3
2x2 x 7 4x 1
2
2x x 7

3
2
4 x 1 The same result
2x x 7

88
The higher order derivatives:
For any positive integer n the nth derivative of a function is obtained by
differentiating successively n times.

Ex 1: Find the fifth derivative of the function: f x 4 x3 5 x 2 6 x 1

/
Sol: f x 12 x 2 10 x 6

f // x 24 x 10

f /// x 24

f //// x 0 (all possible derivatives)

f ///// x 0

Ex 2: Find the third derivative of the following function f ( x ) x 2ex

1
Sol: y x 2e x e x 2 x ex x2 2x
2
y ex 2x 2 x2 2x e x ex x2 4x 2
3
y ex 2x 4 x2 4 x 2 ex = ex x2 6x 6

Note: When computing high order derivatives, we have to simplify the function
after each derivative in order to calculate the next derivative easily.

//
Ex 3: If f x ln x 2 4
7 x , find f x

Sol: f x 4ln x 2 7 x (properties of ln) ,

1 8 x 28
f/ x 4 2x 7
x2 7 x x2 7 x

89
 //
x2 7 x 8 8 x 28 2 x 7
f x 2
x2 7 x

The second order derivatives play an important role in the optimization of a

function in n variables.

The implicit differentiation:

- For the function in the form y f x we have an explicit function, for
3 x
example y 2 x 4 x e but sometimes, certain equations in x and y
cannot be solved explicitly for y in terms of x, for example:
x 2 y 3 6 5x 2 2 xy, x 2 y 3 3xy 2 y 7.
- The implicit functions have important applications in economic. For example
if Q x, y gives the production corresponding to x units of one input and y
units of another, then the equation Q x, y c for constant c is called an
isoquant. Such equations are used to explore the different combinations of
inputs x and y that result in the same level of production. In this context, the
dy
rate is called the marginal rate of technical substitution MRTS . In
dx
general, it is used in the discussion of level curves of a function of two
variables studied in this chapter.
- To get the differentiation of an implicit function, we first differentiate both
sides of the defining equation with respect to x and then solving algebraically
/ dy
for y or .
dx
2
Ex: For x y y2 y2 x 3 , find y / . (we suppose y f x )

Sol: x2 y / y (2 x ) 2 yy / 3x 2

x2 y / 2 yy / 3x 2 2 xy, y / x2 2y 3x 2 2 xy

/ 3x 2 2 xy
y
x2 2 y

91
 Also, if we put the implicit function in the form f x, g c, y / is calculated
/ f x the derivative of f x, y w.r.t. x
using the following rule: y
f y the derivative of f x, y w.r.t. y
2
In our example: we put the function as x y y2 x3 0

/
2 xy 3x 2 3x 2 2 xy
Therefore y (the same result)
x2 2y x2 2 y
The total differential:
- If f / a is the derivative of the function f x at the point x a, the
total differential at the point x a is dy f / a dx .

- The total differential is used to approximate the change in y y for a

given change in x x within a neighborhood of the point x a,
which is: y f a x f a

Ex: Let y x 2 . Assume x changes from the value a 10 to a x 10.1 .

2
(i.e. x 0.1). The corresponding value y will change from y 10 100
2
to y 10.1 102.01. Thus y 2.01 (The exact value of the change in
y ). Using the total differential as an approximation of y we have
dy f / a dx, where f / x 2x 2 10 0.1 2
Notice that the total differential (2) is a good approximation for the change in
y, y 2.01. The approximation improves if x is taken to be smaller and
smaller. The opposite is also true, the approximation will be poorer for a large
value of x .

The total differential is used in the multivariate case to approximate changes in

the variable y (production) given simultaneous changes in the inputs xi ,
i 1, 2,.....n .

91
Functions of n variables:
We have already discussed the basic principles of calculus for functions in one
variable, y f x with x R . Economic analysis, however, often demands
consideration of functions of more than one variable. For example, it is often
important to model how the level of output produced by a firm depends on
several inputs rather just one; labor, capital … etc. In this chapter we consider the
fundamental relationships of differential calculus for functions of more than one
variable. Much of what we have learned about a function in one variable is
extended to the multivariate case.

Let y f x f x1 , x2 ,......, xn denotes a multivariate function f of n

variables, where x Rn .
n
Sometimes, we use the notation f : R R, which reads as: the function f
n
maps from R into R .

Example:

Suppose the output y of a firm depends on two inputs: capital x1 and labor

x2 . So we have y f (X ) f x1, x2 , where f : R 2 R, that is the

function f maps points in R 2 (different capital and labor combinations) into a

single value (output) on the real line R . One of the most known functional form
of f x is the cobb-Douglas production function that takes the form:

y f x Ax1 x2 ,

where A, and are constants, x1 and x2 are the capital and labor

respectively and y is the output.

92
Partial differentiation:

For y f x , we defined the derivative as:

dy f x x f x
f/ x Lim ,
dx x 0 x

y
Which represents the rate at which y changes when x changes as x
x
becomes arbitrarily small x 0 . Suppose y depends on more than one
variable, for example x1 and x2 , and y f x1 , x2 , we can define the rate at
which y changes when x1 changes keeping x2 constant as the partial derivative
y
of y with respect to x1 and it is denoted by where:
x1

y f x1 , x2 f x1 x1 , x2 f x1 , x2
Lim
x1 x1 x1 0 x1
Similarly:

The rate at which y changes when x2 changes keeping x1 constant is the partial
y
derivative of y with respect to x2 , denoted by , where:
x2

y f x1 , x2 f x1, x2 x2 f x1, x2
Lim
x2 x2 x2 0 x2

The reason for calling these rates partial derivatives and using the notation is
that we are changing only one of x1 or x2 at a time, even though y depends on
both x1 and x2 .

f x1 , x2 f x1 , x2
Sometimes we may refer to as f1 x1 , x2 or f x1 and as
x1 x2
f 2 x1 , x2 or f x2 .

93
In general, the partial derivative of a function y f x1 , x2 ,........, xn with
respect to the variable xi is:

f f x1 ,......xi xi ,.....xn f x1 ,..., x1,....xn

fi Lim
xi xi 0 xi

Ex 1: Let y f x1 , x2 x13 5x2 . Find the partial derivative with respect to

. x1 and x2 .
f
Sol: f1 3x12
x1
f
f2 5
x2

In this example f is additively separable. This is why f i depends on xi only.

Ex 2: Find f1 and f 2 if f x1 , x2 x12 x2

Sol: f1 2 x1 x2

f2 x12

Ex3: If y f x1 , x2 10 x1 x22 5x13 4 x2 x1 x2 . Find the partial derivatives:

y y
, .
x1 x2
y
Sol: 10 x22 15 x12 x2
x1
y
20 x1 x2 4 x1
x2
Notice that in the last two examples the partial derivative depends, in general, on
all the arguments of f .
94
Ex 4: Find the partial derivatives if f x e xy ln x 2 y

f 1 2
Sol: e xy y 2
2 xy ye xy
x x y x
f 1 1
e xy x 2
x2 xe xy
y x y y

Ex 5: Find the partial derivatives of the following functions:

5
a) f x1 , x2 x12 x1 x2 x2
2 xy
b) Z xe
2
ln x 2 y
c) f x, y (Preparation by properties of ln)
y2
2x 3y
d) Z
y x
f 4
Sol: a) f1 5 x12 x1 x2 x2 2 x1 x2
x1
f 4
f2 5 x12 x1 x2 x2 x2 1
x2

Z 2 xy 2 xy
b) xe 2y e 1
x
2 xy 2 xy xy
2 xye e e 1 2 xy

Z 2 xy
xe 2x 2 x 2e 2 xy

2ln x 2 y
c) f x, y (preparation by properties of ln)
y2

95
 f 1 1 2
2 1
x y2 x 2y xy 2 2 y3

2
y2 2 2 ln x 2 y 2 y
f x 2y
4
y y

4y
4 ln x 2 y
x 2y
y2

Z y x 2 2x 3y 1 2 y 2x 2x 3y
d) Z1 2 2
x y x y x

5y
2
y x

Z y x 3 2x 3 y 1 3 y 3x 2 x 3 y
Z2 2 2
y y x y x

5x
2
y x

y
Recall that the partial derivative of the function y f x1 , x2 ,......., xn is the
xi
marginal function that represents the change in the function y due to one unit

change in xi assuming all the other variables x j j i are fixed. The following
applications illustrate the different economic meanings of the marginal functions and
how the partial derivatives serve in economic concepts and relationships.
96
1) The demand function:

Suppose the demand function of product a will depend on the price of a Pa ,

the price of a related good b Pb as well as income I as follows:

Qa f Pa , Pb , I
We have the following expected signs for the partial derivatives (marginal
functions) that satisfy economic concepts.

Q
0 (law of demand)
Pa
Q
0 (goods a and b are substitutes)
Pb
Q
0 (goods a and b are compliments)
Pb

Q
0 (good a is a normal good)
I
Q
0 (good a is an inferior good)
I
Ex: Let the demand function of product a takes the following form:

Qa f Pa , Pb , I 50Pa 10Pb 2I ,

where Pa and Pb are the prices of goods a and b respectively and I is the
income. Find all the partial derivatives and interpret your answers.
Sol:

Q
50 0, which means that if the price of good a will increase by
Pa
one unit, the demand on good a will decrease by 50 units. The negative
Q
sign of shows the downward sloping demand curve.
Pa

97
 Q
10 0, which means that if the price of good b increases by one
Pb
unit the demand on good a will increase, which means that goods a and b
are substitutes.

Q
2, which means that if the income increases by one unit, the
I
demand on good a will decrease, which means that good a is an inferior
good.
2) The utility function:

Let the utility function be given by U x1 , x2 x1 x12 , where 0 1 and

x1 and x2 are the quantities consumed from two goods (1), (2).
The partial derivatives take the following forms:

U 1 1 x21
x 1 x2
x1 x11

U 1 x1
x1 1 x2
x2 x2

Notice that for 0 1, the marginal utility functions (the partial derivatives)
satisfy the law of diminishing marginal utility as:

U U
0, however, decreases as x1 increases (if the quantity consumed x1
dx1 x1
increases, the utility increases but by a decreasing rate).
Also:

U U
0, however, decreases as x2 increases
x2 x2

This is because in both marginal utility functions x1 and x2 are in the

denominator and 0 1.
98
 1 3
Ex: If the utility function takes the form: U x1 , x2 x ,x
1
4 4
2

Check that the marginal utility functions satisfy the law of diminishing marginal
utility.
3
3 3
U 1 x4
Sol: The marginal utility function of good (1) is x1 x4 4
2
2
3
x1 4
4x1
4

Notice that the marginal utility is always positive but as x1 increases the
marginal utility decreases.
1
1 1
U 3 3x4
Similarly x x x2
4
x1
4 1
1
x2 4
4x
2
4

3) The production function:

Let Q f k , L be a production function for a firm, where Q, k and L denote

output, capital and labor respectively.
Q
The partial derivative is known as the marginal product of Labor (MPL)
L
Q
and the partial derivative is known as the marginal product of capital (MPK)
k
For Cobb-Douglas production function in the form Q AL k , where A 0,
0 , 1 are technology parameters, the marginal product functions are:
Q AK
A L 1k
L L1
and

Q 1 AL
AL k
k k1
Both marginal product functions satisfy the law of diminishing marginal
Q Q
productivity as we have 0; however as L
L L
99
 Q Q
Also 0; however as k
k k
Suppose the cobb-Douglas production function with three inputs in the form:

Q Ax1 x2 x3T ,

Q 1 T Ax2 x3T
The marginal function of x1 is A x1 x2 x 3
x1 x11
Q Q
and similarly both and include x2 and x3 in the denominator, hence
x2 x3
they all satisfy the law of diminishing marginal productivity.
The marginal product of each input is a function in the three inputs.
1 1
Ex: Suppose the production function in the form Q 10 L k , where Q, L and
3 2

k are the output, labor and capital respectively. Find and interpret the partial
derivatives of this function assuming the total labor force is (25) million workers
and the capital is (15) million LE.
1
2 1
Q 10 10k 2
Sol: L k 3 2
2
, for L 25 and k 15
L 3
3L 3

1
Q 10 15 2

2
1.51 ,which means that if labor increases by (1) unit (1
L 3 25 3

million workers) and capital stays fixed at (15), the output will increase by 1.51
units.
1
Q 5L 3
Q
Also, 1
and at L 25 and k 15 , 3.77 .
k k
k 2

Which means that if capital is (15) and it increases by one unit, the output will
increase by 3.77 units assuming the labor is fixed at (25) million workers.

111
Second order partial derivatives:
For a function y f x we can drive the first derivative, f / x , and if
f / x is differentiable we can find the second derivative, the third derivative
n
and so on. We can continue the process to find the nth derivative f x . The
same can be done for functions of more than one variable y f x1 , x2 ,....., xn
, as we can derive the first order partial derivatives f1 , f 2 ,...., f n and for each
partial derivative, we can get the second-order partial derivatives, with respect to
each of the n variables
f i x1 , x2 ,...., xn
Therefore, we have f ij , i, j 1,2,......, n
xj

f ij is found by first differentiating the function f x1 , x2 ,......, xn with respect

to xi , and then by differentiating the result with respect to x j . There will be n 2
second order partial derivatives.
Example: for y f x1 , x2 we have 4 second-order partial derivatives
f11 f12
f 21 f 22
And for y f x1 , x2 , x3 we have 9 second-order partial derivatives
f11 f12 f13
f 21 f 22 f 23
f31 f32 f33
It is important to have a simple technique to deal with all the possible derivatives,
we can neatly present the second-order partial derivatives using matrices.
Ex 1: Find the first partial derivatives and the second order partial derivatives of
the function: f x1 , x2 x12 x2
Sol:
f1 2 x1 x2 f2 x12
f11 2 x2 f 21 2 x1
f12 2 x1 f 22 0
111
 So we have the following matrices form:

f1 2 x1 x2 f11 f12 2 x2 2 x1
, and
f2 x12 f 21 f 22 2 x1 0

The gradient vector:

The gradient vector, denoted by f , is the vector that collects all the first-order
partial derivatives of the n-variable function f x1 , x2 ,...., xn . The gradient

vector is in the form:

f1
f2
f

fn

The Hessian matrix:

The Hessian matrix, denoted by 2 f , is the matrix that collects the second-
order derivatives of a function in n variables. It takes the form:

f11 f12 f1n

f 21 f 22 f2n
2 f

f n1 fn2 f nn

For example, if y f x1, x2 , n 2, we have;

f1 f11 f12
f 2 f
f 21 f 21 f 22

and for y f x1 , x2 , x3 we have

112
 f1 f11 f12 f13
f f2 , 2 f f 21 f 22 f 23 ,
f3 f31 f32 f33

and so on.

Ex 1: Find the gradient rector and the Hessian matrix for

f x1 , x2 x12 x23

f1 2 x1 x23
Sol: f
f2 3x12 x22

f11 f12 2 x23 6 x1 x22

2 f
f 21 f 22 6 x1 x22 6 x12 x2

Ex 2: Find the gradient vector and the Hessian matrix for

f x1, x2 , x3 x12 x2 2 x2 x32 3x1 x3

f1 2 x1 x2 3x3
Sol: f f2 x12 2 x32
f3 4 x2 x3 3x1

2 x2 2 x1 3
2 f 2 x1 0 4 x3
3 4 x3 4 x2

We notice that the Hessian matrix is always "symmetric" i.e., fij f ji i, j .

This is indicated in the following theorem:

113
Young's theorem:
For a function y f x1 , x2 ,......, xn with continuous first and second-order
partial derivatives, the order of differentiation in computing the cross-partials
does not matter. That is fij f ji i, j 1, 2,..., n .

Ex 3: Find the Hessian matrix for the function f x x12e3 x2 x3

f1 2 x1e3 x2 x3 f 2 3x12e3 x2 x3
f3 x12 e3 x2 x3
Sol:
f11 2e3 x2 x3 f 22 9 x12 e3 x2 x3
f31 2 x1e3 x2 x3
f12 6 x1e3 x2 x3
f 21 6 x1e3 x2 x3
f32 3 x12 e3 x2 x3

f13 2 x1e3 x2 x3
f 23 3x12 e x2 x3
f33 x12 e3 x2 x3

2e3 x2 x3 6 x1e3 x2 x3
2 x1e3 x2 x3
The Hessian matrix 6 x1e3 x2 x3 9 x12 e3 x2 x3
3x12e3 x2 x3
2 x1e3 x2 x3 3x12 e3 x2 x3
x12e3 x2 x3
Notice that the matrix is symmetric and Young's theorem is illustrated as
f12 f 21 , f13 f31 and f 23 f32

Ex 4: Illustrate Young's theorem for the following function: f x, y ln x 2 xy

1 2x y
Sol: f1 fx 2x y
x2 xy x 2 xy
1 x
f2 fy x
x2 xy x2 xy

x2 xy 1 2x y x x2 xy 2 x 2 xy x2
f12 f xy 2 2 2
x2 xy x2 xy x2 xy

x2 xy 1 x 2x y x2 xy 2 x 2 xy x2
f 21 f yx 2 2 2
x2 xy x2 xy x2 xy
Therefore f12 f 21 (Young's theorem)
Ex 5: Find the gradient vector and the hessian matrix for the following functions:

114
 f x1 , x2 , x3 2 x12 3x23 x34 1
Comment on your answers.
Sol: The gradient vector is:

4 x1
f 9 x22
4 x33
The Hessian matrix is

4 0 0
2 f 0 18 x2 0
0 0 12 x32
Notice that this function is an additively separable function. This implies that
f1 4 x1 is a function in x1 only and f 2 9 x22 is a function in x2 only and f 3 is
a function in x3 only.

That immediately implies that all the cross partials will be zeros, and the Hessian
matrix in this case is diagonal.
It is obvious that this feature holds in general for any additively separable
function.

115
The total differential:
The total differential is used to approximate the change in y y for a given
change in x x within a neighborhood of the point x a . That is:

y f a x f a ,
can be approximated by the total differential in the form:

dy f / a dx,
/
where f a is the derivative of the function y f x at the point x a.

The total differential for a function of two variables:

For y f x1 , x2 , where f : R 2 R, the total differential is:

f f
dy dx1 dx2 or dy f1 a1, a2 dx1 f 2 a1 , a2 dx2
x1 x2

For a function y f x1 , x2 at a specific point x1 a1 and x2 a2 , if x1 only

changes and x2 remains constant we have.

y f a1 x1 , a2 f a1 , a2
That can be approximated by:

dy f1 a1 , a2 dx1

Similarly, if x2 changes only and x1 remains constant, we have

y f a1 , a2 x2 f a1 , a2

approximated by dy f2 a1 , a2 dx2

Allowing for both x1 and x2 to change leads to the above definition of the total
differentiation.

116
The total differential for a function of n variables:

For y f x1, x2 ,..., xn , where f : R n R, the total differential is:

f f f
dy dx1 dx2 ... dxn
x1 x2 xn

or dy f1dx1 f 2dx2 ... f ndxn

which is used to approximate changes in y given simultaneous changes in the

inputs xi , i 1, 2,......, n

1
Ex1: Find the total differential for the function f x1 , x2 2 x12 x1 x2
2
Sol: The total differential is:

dy f1 x1 , x2 dx1 f 2 x1 , x2 dx2

1 1
4 x1 x2 dx1 x1 dx2
2 2

Ex2: Find the total differential for:

f x1 , x2 , x3 x1 x22 3x1 x3 3x22 4 x3

Sol: The total differential is given by

dy f1dx1 f 2 dx2 f3dx3

x22 3x3 dx1 2 x1 x2 6 x2 dx2 3x1 4 dx3

117
The implicit differentiation:
The implicit function for a function in one variable:
For an implicit function f x, y c, the first derivative of y with respect to x
dy fx
may be calculated by y/
dx fy

2 3 dy
Ex: For the implicit function x y 3xy 2 y 22, find the value of at
dx
the point 1 , 2 .

dy fx 2 xy 3 3 y 2
Sol:
dx fy 3x 2 y 2 6 xy 1

At the point 1 , 2 , we have:

3 2
dy 21 2 3 2 28
2 2
dx 31 2 61 2 1 25

The implicit function for a function in two variables:

For an implicit function in the form:

f x1 , x2 , y c , f y 0, we have

y f x1 y f x2
and
x1 fy x2 fy
The implicit function for a function in (n) variables:

Given f x1 , x2 ,..., xn , y c, f y 0 the partial derivatives are given by:

y f xi
xi fy

118
Ex1: Compute the partial derivatives of the function:

f x1 , x2 , y 3x1 x2 x2 y 2 x12 x2 y 10
Sol: The implicit function theorem is used to compute the partial derivatives,
we have:

y f x1 3x2 2 x1 x2 y
x1 fy 2 x2 y x12 x2

y f x2 3x1 y 2 x12 y
x2 fy 2 x2 y x12 x2
Ex2: Find the partial derivatives of the function:

f x1 , x2 , y 2 x1 x2 y 3e x1x2 x12 y
Sol: Using the implicit function theorem

y f x1 2 x2 y 3e x1x2 x2 2 x1 y
x1 fy 2 x1 x2 x12

2 x2 y 3x2e x1x2 2 x1 y
2 x1 x2 x12

y f x2 2 x1 y 3e x1x2 ( x1 )
x2 f y 2 x1 x2 x12

2 x1 y 3x1e x1x2
2 x1 x2 x12

119
Level Curves:
- For a function y f x1 , x2 , the set of x1 , x2 pairs that will generate some
specific level for y is called a level set. If, in the two-variable case, we can solve
the equation explicitly for x2 in terms of x1 and a fixed value of y, we have the
dx2
equation for a level curve in x1 , x2 - space. The derivative is then the
dx1
slope of the level curve.

- For complicated functions that it is difficult to solve explicitly for x2 as a

dx2
function of x1 , we can use the total differential to calculate as follows:
dx1

Suppose an explicit function y f x1 , x2 , then y f x1 , x2 c, (where c

is a constant) is the level curve of y. Using the total differential we have

dy f1dx1 f 2dx2 0
f 2 dx2 f1dx1 ,
dx2 f1
Thus
dx1 f2
For a certain level curve y , we consider y fixed along the level curve which
leads to dy 0.
Ex : Use the total differential to compute the slope of the level curve for the
function: f x1 , x2 2 x1 3x2

Sol.: The level curve for this function is y 2 x1 3x2 c, where c is a

constant and y is fixed along the level curve. The slope of the level curve is:

dx2 2
dx1 3

111
1) The indifference curve:

Let U x1 , x2 be the utility function from consumption of goods (1) and (2),
say x1 and x2 . The level curves are given by U x1 , x2 C,
that represent the goods combinations which will generate the same level of
consumer's utility (C).
Using the total differential, we have

U U
dU dx1 dx2 0
x1 x2
U
Define MU1 (Marginal utility w.r.t x1 ),
x1
U
and MU 2 (Marginal utility w.r.t x2 )
x2

We have: MU1dx1 MU 2dx2 0

MU 2dx2 MU1dx1
dx2 MU1
dx1 MU 2

dx2 MU1
Therefore MRS
dx1 MU 2
Slope of the Marginal rate
indifference curve of substitution

Which means that the derivative of x2 w.r.t x1 represents the marginal rate of
substitution (MRS) at which the consumer can substitute one good for the other
and continue to remain on the same indifference curve. Formally, it is the
negative of the slope of the indifference curve.

111
Ex 1: Find the derivative of x2 w.r.t. x1 and interpret your answer if
U x1 , x2 x1 x2 , where U is the utility function.
Sol.: The total differential for U x1 , x2 is:

U U
dU dx1 dx2 0 or du MU1dx1 MU 2dx2
x1 x2
which leads to

dx2 MU1 1
MRS , thus MRS 1
dx1 MU 2 1
Therefore, the consumer remains on the same indifference curve if he substitutes
one unit of one good for one unit of the other. Since the consumer is always
indifferent in one-for-one trade-off between these two goods, we say that the
goods represent Perfect substitutes for this consumer. In this case the
indifference curves take the following form:

Ex 2: Consider the Utility function: U x1 , x2 x12 x2 .,

Find the MRS and interpret

dx2 MU1
Sol.: MRS
dx1 MU 2
2 x1 x2 2 x2
Thus MRS
x12 x1
That means that the consumer needs to substitute one unit of x1 per two units of
x2 to keep his utility of these two goods constant.

112
 1 3
Ex 3: For the Utility function U x1 , x2 ln x1 ln x2 , where x1 and
4 4
x2 are he quantities of good (1) and good (2) respectively, find the marginal rate
of substitution (MRS) between the two goods.

1
dx2 MU1 4 x1 x2
Sol: MRS ,
dx1 MU 2 3 3x1
4 x2

which means that the consumer needs to substitute 3 units of x1 for every unit

of x2 in order to have the same utility. The level curves in this example take the
following form:

Notice that in Ex2 and Ex3 the MRS, which is the absolute value of the slope of
the indifference curve, is increasing in x2 and decreasing in x1 . The curves are
convex towards the origin. This because as one moves on the indifference curve
with more than x2 and less than x1 (i.e., good (1) becomes relatively scarcer) a
reduction in a unit of x1 requires a greater increase in x2 to compensate the
consumer and remain on the same level of the indifference curve.

113
2) The Isoquants:

Let Q f L, K be the production function using labor L and capital k

inputs. The level curves are given by Q f L, K C
Using the total differential, we have:

f f
dQ dL dk 0
L K
or dQ MPL dL MPk dk 0

Thus MPL dL MPk dk 0

MPk dk MPLdL

dk MPL
dL MPk

dk MPL
Therefore, MRTS
dL MPk
The slope of Marginal rate of
the isoquant technical substitution
The graph of Q f L, k C is called an isoquant (as iso means equal)
because all the input combinations lead to the same quantity of output produced.
Along an isoquant, Q is fixed and dQ 0 .

The marginal rate of technical substitution is the rate at which the firm can
substitute one input for the other and continue to produce the same level of output
(say C).
The isoquants have strictly convex to the origin curves.
1 2
Ex 1: For Q f L, k L k , find the rate of technical substitution between
3 3

the labor L and capital k .

114
 dk MPL
Sol : MRTS , where MPL is the marginal product function of
dL MPk
labor and MPk is the marginal product function of capital.
2
2 2 1 K 3
1 L k 3 3
3 3 L k
Thus MRTS 1 1 1
2 L3 k 3 2 L 3
2L
3
3 k
which means that the producer needs to substitute 2 units of labor per unit of
capital to keep output constant.
The level curves of output (isoquants) in this example take the following form:

Ex 2: Find the marginal rate of technical substitution for Q ,

1 1
where Q f L, k L 3k 2

2 1
1 L k 3 2
MPL 3 2k
Sol.: MRTS 1 1
MPk 1 L3 k 2
3L
2
The firm needs to substitute 3 units of labor per 2 units of capital in order
to produce the same level of production.

115
Homogeneous functions:
A multivariate function y f x1 , x2 ,...., xn is homogenous of degree k ,
k k
denoted by H , if and only if f x1 , x2 ,....., xn f x1 , x2 ,......, xn

Ex 1: Let y f x1 , x2 3x1 x2 . Prove that the function is homogenous.

Sol : f x1 , x2 3 x1 x2 3 2 x1 x2
2 2
3x1 x2 f x1 , x2

Therefore f x1 , x2 3x1 x2 is homogenous of the degree 2, H 2 .

x12
Ex 2: Show that the function y f x1 , x2 3
is homogenous
3x2
2 2
x1 x12 1 x12
Sol : f x1 , x2 3
3 x2 3 3 x2 3 3x23
1
f x1 , x2
1
The function is homogeneous of degree -1, H
1 1 3
Ex 3: Show that the function y f x1 , x2 x1 2 x2 3 x2 2 is not
homogeneous.
1 1 3
Sol : f x1 , x2 x1 2 x2 3 x2 2

1 1 1 1 3 3
2
x
1
2 3
x
3
2
2
x
2
2

5 1 1 3 3
6
x x 2 3
1 2
2
x
2
2

k
f x1 , x2
Therefore, this function is not homogeneous

116
If the function f x1 , x2 ,..., xn is a production function homogeneous of degree
k , H k , then multiplying the level of all inputs by the same factor will
increase output by k . The concept of homogeneous functions simplifies the
analysis of returns to scale.

Suppose for example f x1 , x2 x1 x22 is a homogeneous function of degree 3,

3 3
H , as f x1 , x2 x1 x22 . Assuming that f x1 , x2 is a production
function, that means if we multiply the level of all inputs by 2, we would
3
increase the output by 2 8 which is an example of increasing returns to
1 1
scale. But for the production function f x1 , x2 x x which is homogeneous
4 2
1 2

3
of degree , if we want to double all input levels, the output would be increased
4
3
by 2 4
1.68, this is the case of decreasing returns to scale. If doubling input in
1 1
1
the H function, for example f x1 , x2 x x , the output doubles. This is the
1
2 2
x

case of constant returns to scale. The returns to scale is important in order to take
an efficient decision for firms to merge or increase in size than to remain small.

Ex 4: Determine the degree of the homogenous function:

Q f L, k Lk , 0 , 1 and interpret.

Sol : f L, k L k Lk
a
f L, k

Therefore, it is homogeneous of degree and the interpretation is

summarized in the following table:

117
 Homo. of Returns to
Interpretation
degree scale (RTS)

1 1 (H1) Con. RTS Inputs by , output by

>1 >1 (H>1) Inc. RTS Inputs by , output by more than

<1 <1 (H<1) Dec. RTS Inputs by , output by less than

Euler's theorem:
If a multivariate function y f x1 , x2 ,..., xn is homogeneous of degree
k H k then:

f1 x1 f 2 x2 ... f n xn k f x1 , x2 ,..., xn
n
or fi xi ky
i 1

It means that if we multiply the marginal product of each input i by the level of
that input and summing up, it is equal to k times the value of the output.
Ex : Suppose the Cobb-Douglas production function:

Q f L, k L k . Prove Euler's theorem for this function.

Sol : This function is homogeneous of degree to prove Euler's

theorem, we get first the partial derivatives.

f1 L 1k f2 Lk 1

n
and f i xi f1L f 2k L 1k L Lk 1
k
i 1

( )L k ( ) f ( L, K )
Thus, Euler's theorem relates the marginal productivities, input, output and
returns to scale in a single equation.

118
Homothetic functions:
If a multivariate function y f x1 , x2 ,..., xn is homogeneous of degree

k H k , then g y is homothetic if g / y 0.

Homothetic functions are monotonic transformations of homogeneous functions:

which means that for any function g . , g / . 0
Examples of monotonic transformations:

g y y 2 g/ y 1

g y 2y g/ y 2

y 1
g y g/ y
2 2
g y y2 g/ y 2y

1
g y ln y g/ y
y

g y ey g/ y ey
Ex : Find two homothetic functions for

y f x1 , x2 x1 x2 .
Sol : To prove first that y is homogeneous
2
f x1, x2 x1 x2 x1 x2 .
2
Therefore, y is homogeneous of degree 2 H .

The function g y y 10 x1 x2 10 is a homothetic function because it is a

/
monotonic transformation of a homogeneous function as, g y 0

119
 1
Also, the function g y ln y ln x1 x2 is homothetic as g / y 0,
y
as ln y is defined only for positive values

Homothetic functions have important properties that discussed in details in

intermediate and advanced microeconomics.
Exercise:

1 1
Suppose the production function Q f k, L Lk 2 4

Prove Euler's theorem for this function.

a) We have to prove first that it is a homogeneous function and we have to
determine its degree of homogeneity (k).
1 1 1 1 1 1
f k, L L 2 k 4 2
L
2 4
k 4

3 1 1 3
4
Lk
2 4 4
f k,l

3
This function is homogeneous with degree R
4
To prove Euler's theory, we have to prove:

f1 x1 f 2 x2 kf x1 , x2 and in this example

fk k fLL Rf k , L

1 43 12 1 12 14 1 14 12 1 12 14
k L k L k L k L Lk
4 2 4 2

1 1 14 12 3 12 14
k L Lk Rf k1L
4 2 4
The theorem is proved.

121
 1 1
b) Find a homothetic function for y f k, L Lk2 4

Homothetic functions are monotonic transformations of homogenous functions.

3
Since this function is homogenous with degree . So we can find homothetic
4
function by one of the following transformations g . that satisfy g / . 0

g y 3y 7 g/ y 3 0

2y 2
g y g/ y 0
5 5

g y 3y2 g/ y 6 y ( y is always positive, as it’s a prod. fun.)

g y ey g/ y e y (always positive)

1
g y ln y g/ y ( y is always positive)
y

Note: We have to verify that the derivative is positive.

121