Scribd
Upload
48 views

Economics 101A (Lecture 2) : Stefano Dellavigna

This document summarizes a lecture on optimization techniques in economics: 1. The professor discusses using derivatives to find maxima and minima of functions with one and multiple variables. Necessary conditions for an extremum are that the derivative is zero, and sufficient conditions involve the second derivative test. 2. Comparative statics examines how optimal choices change with parameters. Examples include production and consumption responses to costs and income. 3. The implicit function theorem allows solving implicit functions explicitly, and finding how implicit variables change with parameters.

Uploaded by

BenRB27
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
48 views

Economics 101A (Lecture 2) : Stefano Dellavigna

This document summarizes a lecture on optimization techniques in economics: 1. The professor discusses using derivatives to find maxima and minima of functions with one and multiple variables. Necessary conditions for an extremum are that the derivative is zero, and sufficient conditions involve the second derivative test. 2. Comparative statics examines how optimal choices change with parameters. Examples include production and consumption responses to costs and income. 3. The implicit function theorem allows solving implicit functions explicitly, and finding how implicit variables change with parameters.

Uploaded by

BenRB27
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 21

Economics 101A

(Lecture 2)

Stefano DellaVigna

January 17, 2011

Outline

1. Optimization with 1 variable

2. Multivariate optimization

3. Comparative Statics

4. Implicit function theorem

Optimization with 1 variable


Nicholson, Ch.2, pp. 20-23 (20-24, 9th Ed)
Example. Function = 2
What is the maximum?

Maximum is at 0
General method?

Sure! Use derivatives


Derivative is slope of the function at a point:
()
( + ) ()
= lim
0

Necessary condition for maximum is


()
=0

Try with = 2

()
=

= 0 = =

(1)

Does this guarantee a maximum? No!


Consider the function = 3

()
=

Plot = 3.

= 0 = =

Sucient condition for a (local) maximum:


()

2
()
= 0 and
0
2

(2)

Proof: At a maximum, ( + ) () 0 for


all .
()
2 () 2
1

Taylor Rule: ( +) ( ) = + 2 2 +

higher order terms.

()
Notice: = 0
2 () 2

( + ) ( ) 0 for all = 2
2 ()
0 = 2 0

Careful: Maximum may not exist: = exp()

Tricky examples:
Minimum. = 2

No maximum. = exp() for ( +)

Corner solution. = for [0 1]

Multivariate optimization
Nicholson, Ch.2, pp. 23-30 (24-32, 9th Ed)
Function from to : = (1 2 )
Partial derivative with respect to :
(1 )

(1 + ) (1 )
= lim
0

Slope along dimension


Total dierential:

()
()
()
=
1 +
2 + +

1
2

One important economic example

Example 1: Partial derivatives of = ( ) =


5 5
0 =
(marginal productivity of labor)
0 =

(marginal productivity of capital)

00

=

Maximization over an open set (like )


Necessary condition for maximum is
()
= 0

or in vectorial form

(3)

() = 0
These are commonly referred to as first order conditions (f.o.c.)

Sucient conditions? Define Hessian matrix :

0011

0012

001 002

00
1


00

Subdeterminant || of Matrix is defined as the


determinant of submatrix formed by first rows and
first columns of matrix

Examples.
||1 is determinant of 0011 that is, 0011
||2 is determinant of
=

0011
0021

0012
0022

Sucient condition for maximum


1. must satisy first order conditions;
2. Subdeterminants of matrix must have alternating signs, with subdeterminant of 1 negative.

Case with = 2
Condition 2 reduces to 0011 0 and 0011 0022
(0012 )2 0

Example 2: (1 2) = 1 21 +2 22 21 52
First order condition w/ respect to 1?
First order condition w/ respect to 2?
1 2 =
For which 1 2 is it a maximum?
For which 1 2 is it a minimum?

Comparative statics
Economics is all about comparative statics
What happens to optimal economic choices if we
change one parameter?
Example: Car production. Consumer:
1. Car purchase and increase in oil price
2. Car purchase and increase in income
Producer:
1. Car production and minimum wage increase
2. Car production and decrease in tari on Japanese
cars
Next two sections

Implicit function theorem


Implicit function: Ch. 2, pp. 31-32 (3233, 9th Ed)
Consider function 2 = (1 )
Can rewrite as 2 (1 ) = 0
Implicit function has form: (2 1 ) = 0
Often we need to go from implicit to explicit function

Example 3: 1 1 2 2 = 0
Write 1 as function of 2:
Write 2 as function of 1:

Univariate implicit function theorem (Dini): Consider an equation ( ) = 0 and a point (0 0)


solution of the equation. Assume:
1. continuously dierentiable in a neighbourhood
of (0 0);
2. 0 (0 0) 6= 0
Then:
1. There is one and only function = () defined
in a neighbourhood of 0 that satisfies ( ()) =
0 and (0) = 0;
2. The derivative of () is
0 ( ())

0() = 0
( ())

Example 3 (continued): 1 1 2 2 = 0
Find derivative of 2 = (1) implicitely defined for
(1 2) = (1 0)

Assumptions:
1. Satisfied?
2. Satisfied?

Compute derivative

Multivariate implicit function theorem (Dini):


Consider a set of equations (1(1 ; 1 ) =
0; ; (1 ; 1 ) = 0), and a point
(00) solution of the equation. Assume:
1. 1 continuously dierentiable in a neighbourhood of (00);
2. The following Jakobian matrix xf evaluated at
(00) has determinant dierent from 0:
f
x

1
1

Then:
1. There is one and only set of functions = g()
defined in a neighbourhood of 0 that satisfy
f ( g()) = 0 and g(0) =0;
2. The partial derivative of with respect to is
det

(1)
(11 +1)

det xf

Example 2 (continued): Max (1 2) = 1 21 +


2 22 21 52
f.o.c. 1 : 21 1 2 = 0 = 1()
f.o.c. 2 : 22 2 5 = 0 = 2()
Comparative statics of 1 with respect to 1?
First compute

1
1
2
1

f
det x

1
2
=
2
2

Then compute det

1
1
2
1

(1)
(11 +1)

1
2 =
2
2

1 =
Finally,
1

Why did you compute det xf

already?

Next Class
Next class:
Envelope Theorem
Convexity and Concavity
Constrained Maximization
Envelope Theorem II

Going toward:
Preferences
Utility Maximization (where we get to apply maximization techniques the first time)

You might also like