Mathematics for Economics Anthony Tay

18. Partial Derivatives

The concept of “ceteris paribus” is one that is well known even to beginning students of economics. It

is a mental experiment where we consider the effect of a change in one variable on another variable

holding everything else fixed. The concept of partial derivatives delivers exactly these sorts of

statements.

Example 18.1 Let

= (k , l ) 2k 0.3l 0.5 be a production function with inputs k and l .
y f=

Suppose l is fixed at l = 4 . What is the effect on y of increasing k ?

0.3 0.5
If l = 4 , then
= (k , 4) 2k=
y f= 4 4k 0.3 . Thus fixing l = 4 reduces the problem to one with a

function of one variable, and we know how to proceed:

d
f (k , 4) = 1.2k −0.7 .
dk

If k = 1 , then df (1, 4) / dk = 1.2 , i.e. when k = 1 , a one-unit increase in k will increase output by

approximately 1.2 units. If k = 2 , then df (2, 4) / dk = 0.739 .

We can also consider higher derivatives. For instance,

d2
−0.84k −1.7 < 0 ,
f (k , 4) =
dk 2
so output increases at a decreasing rate as more capital is added to the process, if labor is held fixed at

l =4.

All these values would of course change if l were fixed at some other value:

If l = 9 , then
d d2
f (k ,9) = 6k 0.3 , f (k ,9) = 1.8k −0.7 and −1.26k −1.7 < 0
f (k ,9) =
dk dk 2
For any value of k , an increase in k will result in a larger increase in output than in the previous case

when l = 4 . Additional capital is more productive when l = 9 than when l = 4 , although this

productivity is diminishing at a faster rate. Clearly the effect on y of increasing k depends not only

on the value of k , but also on the value of l .

Mathematics for Economics 18-1

Partial Derivatives In practice, we do not need to repeat this process for specific values of l .

Instead, we simply treat l as fixed (as in the ceteris paribus mental experiment) and treat the function

as a function of one variable k . Then we can take derivatives in the usual way. We say we are taking

partial derivatives of y with respect to k , and write

∂y ∂f
= = fk′
= f1′ (k=
,l) 0.6k −0.7 l 0.5 .
∂k ∂k

(I have given four commonly used notations for partial derivatives -- use whichever notation seems

most appropriate or convenient in your given situation.)

Note the use of “ ∂ ” rather than " d " in the ∂y / ∂k notation. This is merely convention: we use ∂ when

the derivative is a partial derivative (some variable is being held constant) and “ d ” for derivatives of

functions of one variable.

Our answer to the question “what is the effect on y of increasing k when l is held at l = 4 ?” can then

be answered by putting l = 4 into the expression of the partial derivative

f1′ (k , 4)
= 0.6
= k −0.7 40.5 1.2k −0.7

In similar fashion, we can ask what the effect is on y of increasing l holding k fixed. We have

∂y ∂f
= = fl′
= f 2′ (k ,=
l) k 0.3l −0.5
∂l ∂l

∂z ∂z
Example 18.2 z = x 3 y 4 . Then = 3 x 2 y 4 , and = 4 x3 y 3 .
∂x ∂y

There is another notation for partial derivatives: given z = f ( x, y ) , I can also write f1′ for f x′ , and f 2′

for f y′ . The '1' in the subscript of f1′ means differentiate with respect to the first variable which appears

in the list of independent variables in f ( x, y ) , i.e., ' x ' . The '2' in the subscript of f 2′ means

differentiate with respect to the first variable which appears in the list of independent variables in

f ( x, y ) , i.e., ' y ' . This is especially useful when we apply the chain rule.

Mathematics for Economics 18-2

We can also take higher derivatives. In particular, for z = f ( x, y ) we define

∂2 f ∂  ∂f 
2
≡   ≡ f xx′′
∂x ∂x  ∂x 
The interpretation of these two higher
2
∂ f ∂  ∂f  partial derivatives should be obvious.
2
≡   ≡ f yy′′
∂y ∂y  ∂y 

∂2 f ∂  ∂f 
≡   ≡ f yx′′ How does the “rate at which f changes with y ”
∂x∂y ∂x  ∂y 
change when x changes?

∂2 f ∂  ∂f  How does the “rate at which f changes with x ”

≡   ≡ f xy′′ change when y changes?
∂y∂x ∂y  ∂x 

∂2 f ∂2 f
and are called cross-partials.
∂l ∂k ∂k ∂l

Example 18.3 For f (k , l ) = Ak a l b

∂f ∂f
= Aak a −1l b = Abk a l b −1
∂k ∂l

∂2 f ∂2 f
= Aa (a − 1)k a − 2l b = Ab(b − 1)k a l b − 2
∂k 2 ∂l 2
∂2 f ∂2 f
= Aabk a −1l b −1 = Aabk a −1l b −1
∂l ∂k ∂k ∂l

∂2 f ∂2 f
Notice that = .
∂l ∂k ∂k ∂l
This is not a coincidence. It turns out that the order with which you take cross partials is irrelevant. This

is a result known as Young’s Theorem, which we will not prove in this course.

Example 18.4 Let z = x / y . Then

∂z 1 ∂z x
= ; = − 2;
∂x y ∂y y

∂2 z ∂2 z 2x
= 0; = 3 ;
∂x 2 ∂y 2
y
∂2 z 1 ∂2 z 1
= − 2; = − 2.
∂y∂x y ∂x∂y y

Mathematics for Economics 18-3

Partial Derivatives of Functions of Many Variables If z = f ( x1 , x2 ,..., xn ) , we can ask how z

responds to a change in the variable xi , holding all other variables fixed. To answer such questions,

treat x1 , x2 ,..., xi −1 , xi +1 ,..., xn as constants, and differentiate z with respect to xi .

x4
Example 18.5 If f ( x, y, z ) = , then
yz
4 x3 x4 x4
f x′ = ; f y′ = − ; f z′ = − ;
yz y2 z yz 2

For each of the n partial derivatives of the function z = f ( x1 , x2 ,..., xn ) , we can compute n second-order

partial derivatives:
∂  ∂f  ∂2 f
=f ij′′ =  
∂x j  ∂xi  ∂x j ∂xi

In other words, for a function of n-variables, we have n 2 second-order partial derivatives. We often

organize these n 2 partial derivatives into the following square array, called the “Hessian Matrix”:

 f ′′ (x) f12′′ (x)  f1n′′ (x) 

 11 
 f ′′ (x) f 22′′ (x)  f 2 n′′ (x) 
f ′′(x) =  21 
     
 
 f n1′′ (x) f n 2′′ (x)  f nn′′ (x) 

x4
Example 18.6 f ( x, y , z ) =
yz
4 x3 x4 x4
f x′ = ; f y′ = − ; f z′ = − ;
yz y2 z yz 2

12 x 2 4 x3 4 x3
f xx′′ = f xy′′ = − f xz′′ = −
yz y2 z yz 2
4 x3 2 x4 x4
f yx′′ = − f yy′′ = f yz′′ =
y2 z y3 z y2 z2
4 x3 x4 2 x4
f zx′′ = − f zy′′ = f zz′′ =
yz 2 y2 z2 yz 3

We can compute even higher-ordered derivatives. It turns out that the cross-derivatives are the same

regardless of the order in which the differentiation was carried out. We have

f xy′′ = f yx′′ , f xz′′ = f zx′′ , f yz′′ = f zy′′

This is always true. For any function of n variables z = f ( x1 , x2 ,..., xn ) , we always have
∂2 f ∂2 f
= .
∂xi ∂x j ∂x j ∂xi

Mathematics for Economics 18-4

 Exercises

f ( x + h, y ) − f ( x , y ) f ( x, y + h ) − f ( x, y )
1. Given f ( x, y ) below, find and . Find
h h
f ( x + h, y ) − f ( x , y ) f ( x, y + h ) − f ( x, y )
(i) lim h→0 and (ii) lim h→0
h h
(Comment: as you do these drills, try to get the intuition that you are, in the case of (i) differentiating

the function with respect to x , treating y as though it were a constant, and in the case of (ii),

differentiating the function with respect to y , treating x as though it were a constant.)

) 3x + 2 y ;
(a) f ( x, y=

(b) f ( x, y ) = 5 xy ;

) x2 + y 2 ;
(c) f ( x, y=

(d) f ( x, y ) = 4 x 2 y ;

(e) f ( x, y ) = 3 x 2 ;

(f) f ( x, y ) = y 2 / x ;
f ( x + h, y ) − f ( x , y ) f ( x, y + h ) − f ( x, y )
In each case, find lim h→0 and lim h→0
h h

2. For all of the functions f ( x, y ) in question 1, find

(a) f1′ and f 2′

(b) f11′′ , f 22
′′ , f12′′ , and f 21
′′ . Verify in each case that f12′′ = f 21
′′ .

3. For all of the functions f ( x, y, z ) in question 2 and 4, find

(a) f1′ , f 2′ , f3′

(b) f11′′ , f 22
′′ , f33
′′ , f12′′ , f 21
′′ , f13′′ , f31
′′ , f 23 ′′ . Verify in each case that f ′′ij = f ′′ji .
′′ and f32

4 (a) Given f ( x, y ) =x 2 + 2 xy + y 2 , find

(i) f (−1, 2) (ii) f (a, a ) (iii) f (1/ x,1/ y ) (iv) g ( x) = f ( x,1 / x)

(b) Given F ( x, y, z ) = y e x y z , find

(i) F (1,1,1) (ii) F ( x, x 2 , x 3 ) (iii) F ( x,1,1)

Mathematics for Economics 18-5

5. Let f ( x=
, y ) ln( y − 2 x) .

(i) Find the largest possible domain,

(ii) Find the range of the function when the function is defined over the largest possible

domain;

(iii) Find the largest possible domain if it is desired that f ( x, y ) ≥ 0 ;

(iv) Sketch the level curves f ( x, y ) = k for k = −1, 0, 1, 2 .

6. Given f ( x, y ) below, find f x′ ( x, y ) and f y′ ( x, y ) .

(a) f ( x, y ) =x 2 + 2 xy + y 2 x, y ) x 3 e − y + y 3
(b) f (=
x2 y3
(c) f ( x, y ) = x 2 ye xy (d) f ( x, y ) =
x+ y

7. Given f ( x, y ) =x 2 + 2 xy + y 2 , find

(a) [ f ( x + h, y ) − f ( x, y )]/ h

(b) [ f ( x, y + h) − f ( x, y )]/ h

In both cases, evaluate the expression when h → 0 .

8. The volume of a cone of height h and radius (at its base) r is given by

= ( r , h) π r 2 h / 3 .
V f=

How would you interpret the equation ∂V / ∂r =2V / r . Show that the change in the volume

of the cone as you increase its height while holding the base radius constant is proportional to

the area of the base.

9. Given w = f ( x, y, z ) below, find ∂w / ∂x , ∂w / ∂y , and ∂w / ∂z

x2 − y 2
(a) f ( x, y, z ) = ye xyz (b) f ( x, y, z ) =
y2 + z2
(c) f ( x, y, z ) = xyz (d) f ( x, y, z ) = ye z log( xz )

Mathematics for Economics 18-6