8/10/2019 Partial Differentation
1/32
EC115 - Methods of Economic AnalysisLecture 3
Partial Differentiation
Renshaw - Chapter 14
University of Essex - Department of Economics
Week 18
Domenico Tabasso (Universityof Essex - Department of Economics)
Lecture 3 Week 18 1 / 35
http://find/8/10/2019 Partial Differentation
2/32
Topics for this week
Graphical and economic interpretation: the slope of thelevel curve
Second order partial derivatives
Economic Application: Utility Functions
Cross partials and Youngs Theorem
Domenico Tabasso (Universityof Essex - Department of Economics)
Lecture 3 Week 18 2 / 35
http://find/8/10/2019 Partial Differentation
3/32
Introduction
Recall that a function of several variables is a relation
between some independent variables x1, x2, x3,..., xnand some dependent variable zsuch that:
z=f(x1, x2, x3,...xn)
specifies the value ofzgiven the values ofx1, x2, x3,...xn.
A typical function we encounter in economics is the
Cobb-Douglas function:
u=U(x,y) =Axy1
where A andare positive constants and
[0, 1].
Domenico Tabasso (Universityof Essex - Department of Economics)
Lecture 3 Week 18 3 / 35
http://find/8/10/2019 Partial Differentation
4/32
Today we will analyze these type of functions further byasking the following questions:
How does zchange when x1 changes assumingx2, x3, ..., xn are constant?
How does x1 changes when x2 change assuming
z, x3,..., xn are constant?The answers to these questions:
will help us graph functions of the form z=f(x,y);
andhave interesting economic interpretations.
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 4 / 35
http://find/http://goback/8/10/2019 Partial Differentation
5/32
What is a Partial Derivative?
Let us start with a Macro example:
One of the most important questions inmacroeconomics is what determines the growth rate ofa countrys real income.
Suppose that the growth rate, GR, is determined by the following linear
relation:
GRt= 0 + 1It+ 2HCt+ 3INSTt+ 4Techt+ ...+
We would like to know the impact of each of these variables.
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 5 / 35
http://find/http://goback/8/10/2019 Partial Differentation
6/32
That is, by how much could we improve the growthperformance of an economy if we only
increasing the capital stock (investment), It?
improved the quality of education, HC?
improved our institutions (provide property rights,stable governments. etc.), INST?
improved our technology through innovations, Tech?
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 6 / 35
http://find/8/10/2019 Partial Differentation
7/32
To do this, first we have to analyze the effects of eachcomponent independently of the others, and then, secondwe have to analyze the effects of the components taken asa whole.
Partial derivativesallow us to perform this task.
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 7 / 35
http://find/http://goback/8/10/2019 Partial Differentation
8/32
Slope of a surface
Consider the utility function:
z=U(x,y) =x1/2y1/3.
and fix the consumption ofy=27.
Recall that this implies we are analyzing the set ofcoordinates that have the property(x, 27, z). That is,we are looking at the graph of the function:
z=U(x, 27) =3x1/2.
The partial derivative ofzwith respect to xwheny=27 is the slope of a straight line tangent to thisfunction at any point (x, z) on its graph.
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 8 / 35
http://find/8/10/2019 Partial Differentation
9/32
14
12
10
8
4
6
2
0
1 2 3 4 5 6 7 8 9 10 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 2 0 2 1 2 2 2 3 2 4 2 5 2 6 2 7 2 8 2 9 3 0 3 1
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 9 / 35
http://find/8/10/2019 Partial Differentation
10/32
Original function: z=U(x,y) =x1/2y1/3
Now fix the consumption ofx=16.
That is, we are looking at the graph of the function:
z=U(16,y) =4y1/
3
.
The partial derivative ofzwith respect to y whenx=16,is the slope of a straight line tangent to this
function at any point (y, z) on its graph.
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 10 / 35
http://find/8/10/2019 Partial Differentation
11/32
What does it mean?
Why do we say that it is a measure of the slope? A derivative is a measure of change. For example, how much does zchanges when y changes?
As the change in ygets smaller and smaller we findthat the change in zcan be approximated better andbetter by the change in ymultiplied by the slope of thetangent.
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 11 / 35
http://find/8/10/2019 Partial Differentation
12/32
Suppose that we wanted to know how much does zchanges when we increase y, from y0 to y1,when holding xconstant at 25:
z0 = U(16,y0) =4y1/30
z1 = U(16,y1) =4y1/31
Define the change in yto be y=y1 y0, so that:z1 z0 = U(16,y0+ y) U(16,y0)
z
y =
U(16,y0+ y)U(16,y0)
y
limy0
z
y =
U(16,y)
y .
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 12 / 35
http://find/8/10/2019 Partial Differentation
13/32
A graphical representation
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 13 / 35
S d b bl fi d h d f h
http://find/8/10/2019 Partial Differentation
14/32
So in order to be able to find the derivative ofz withrespect to xwe have to assume that y is just a constant(i.e. any number) and then apply the standard derivation
rules. Vice versa in case we take the derivative with respectto y. So for example:
Find the partial derivatives zx
and zy
of:
z=3x3 +2x2y+ y2 + y
z
x=9x2 +4xy
Why? Because y is a constant! So when x is multiplied by y, its
derivative will be multiplied by y; when y is just added, then itdisappears from the derivative.
z
y=2x2 +2y+1
As before: Now x is a constant!
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 14 / 35
http://find/8/10/2019 Partial Differentation
15/32
Another Example
z= (x3 +y2)1/2
This is a more complicated case: We have to apply theChain Rule. So the derivative will be the product of twofunctions: the derivative of theoutside function (i.e.
the power of 1/2) times the derivative of the insideone. So:
z
x = 1
2(x3 +y2)
1/2(3x2) = 3x
2
2x3+y2
zy
= 2y2
x3+y2
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 15 / 35
http://find/8/10/2019 Partial Differentation
16/32
Second order partial derivatives
Now it is easy to see that the first order partial
derivatives tells us if a function is increasing ordecreasing given that we hold constant the rest ofindependent variables.
For example, iff(x0,y)/y>0 for all y
[0, 10],thisimplies that given x=x0,the function f is increasingin yfor values ofythat range from 0 to 10 inclusive.Using the previous exercises we obtained that if
f(x,y) = (x3
+y2
)1/2
then the partial with respect toy is:
f(x0, y)
y = y(x30 + y
2)1/2 >0 for all values ofy>0 given thatx0 0.
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 16 / 35
http://find/http://goback/8/10/2019 Partial Differentation
17/32
Summary
If we have a have a function oftwovariables, we cancalculatetwopartial derivatives
U(x,y) =xy (1)
=
Ux
=x1y (2)
U
y =xy1 (3)
But each of these derivatives can itself be a function ofxand y. So for each of them we can calculate the derivatives
with respect to x and y!Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 17 / 35
http://find/8/10/2019 Partial Differentation
18/32
8/10/2019 Partial Differentation
19/32
Fi d i i f i
8/10/2019 Partial Differentation
20/32
First derivative function
Going back to a previous example:
z= (x3
+y2
)1/2
(12)Fixing x=x0 and performing the partial derivative weobtain a new function in y:
zy
=h(x0,y) =y(x30 +y2)1/2.
By differentiating again with respect to ywe can get
information about how this first partial changes when ychanges holding constant x=x0.
That is, how does h change when ychanges (i.e., whatis the slope ofh(x0,y))?
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 20 / 35
Th f i h
http://find/8/10/2019 Partial Differentation
21/32
The function h
Note that hdefines the rate of change ofz when y changes
and x=x0. In this case, we obtain:
h(x0,y)
y =
2z
y2 = (x30 +y
2)1/2 y2(x30 +y2)3/2.
Consider the case ofx=0; then:
h(0,y)
y = (y2)1/2 y2(y2)3/2
= y1 y2y3= y1 y1 =0.
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 21 / 35
http://find/8/10/2019 Partial Differentation
22/32
The interpretation is as follows whenx
=x
0=0 then:first partial: an increase in y increases z.z is anincreasingfunction in y;
second partial: this increase isconstantfor all y.
Note that we could have obtained the same results byputtingx=x0=0 into f(x,y) and then differentiating(f(0,y) = (03 +y2)1/2 =y).
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 22 / 35
A hi l t ti
http://find/8/10/2019 Partial Differentation
23/32
A graphical representation
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 23 / 35
http://find/8/10/2019 Partial Differentation
24/32
Economic Application: Utility Functions
8/10/2019 Partial Differentation
25/32
Economic Application: Utility Functions
Consumer Theory =Marginal Utility andDiminishing Marginal Utility
U(x,y)
x is called theMU
xwhen we hold constant theconsumption ofy.U(x,y)
y is called theMUywhen we hold constant theconsumption ofx.
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 25 / 35
http://find/http://goback/8/10/2019 Partial Differentation
26/32
Cross Partial Derivatives
8/10/2019 Partial Differentation
27/32
Cross Partial Derivatives
Recall that second order partial derivatives measurehow the slope ofz=f(x,y0) changes when xchanges.
Cross partials are a natural extension as they measurehow the slope ofz=f(x,y0) changes when y changeskeeping xconstant.
Formally cross partials are denoted by:
2f(x,y
)xy or
2f(x,y
)yx .
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 27 / 35
Youngs Theorem
http://find/8/10/2019 Partial Differentation
28/32
Young s Theorem
Definition (Youngs Theorem)
For a function y=f(x1, x2, . . . , xn), with continuous firstand second-order partial derivatives, theorderof
differentiation in computing cross-partials isirrelevant.
2f(x,y)
xy
=2f(x,y)
yx
.
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 28 / 35
Graphical Representation
http://find/8/10/2019 Partial Differentation
29/32
Graphical Representation
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 29 / 35
The other cross partial derivative
http://find/8/10/2019 Partial Differentation
30/32
The other cross partial derivative
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 30 / 35
Example: Cobb-Douglas Function
http://find/8/10/2019 Partial Differentation
31/32
Example: Cobb Douglas Function
For the Cobb-Douglas utility function we have
U(x,y) =x0.5y0.5
U(x,y)
x =
1
2yx0.5 and U(x,y)
y =
1
2xy
0.5
2U(x,y)
x2 = 1
4y0.5x1.5 and
2U(x,y)
y2 = 1
4x0.5y1.5
2U(x,y)xy
= 14x0.5y0.5 =
2U(x,y)yx
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 31 / 35
More exercises
http://find/8/10/2019 Partial Differentation
32/32
More exercises
Find the second and cross partial derivatives of
1 z=3
x3
+2x2y
+y2
+y
z
x =9x2 +4xy
z
y =2x2 +2y+1
2 z= (x3
+y2
)0.5
z
x =
3
2x2(x3 +y2)0.5
z
y =y(x3 +y2)0.5
3 z=100e2x+3y
z
x =200e2x+3y
z
y =300e2x+3y
Domenico Tabasso (Universityof Essex - Department of Economics)Lecture 3 Week 18 32 / 35
http://find/