Preference and Demand Examples

Similar documents
Excess Error, Approximation Error, and Estimation Error

CHAPTER 6 CONSTRAINED OPTIMIZATION 1: K-T CONDITIONS

CHAPTER 7 CONSTRAINED OPTIMIZATION 1: THE KARUSH-KUHN-TUCKER CONDITIONS

Applied Mathematics Letters

Least Squares Fitting of Data

Least Squares Fitting of Data

Xiangwen Li. March 8th and March 13th, 2001

Economics 101. Lecture 4 - Equilibrium and Efficiency

PHYS 705: Classical Mechanics. Calculus of Variations II

On Pfaff s solution of the Pfaff problem

What is LP? LP is an optimization technique that allocates limited resources among competing activities in the best possible manner.

1. relation between exp. function and IUF

AN ANALYSIS OF A FRACTAL KINETICS CURVE OF SAVAGEAU

1 Review From Last Time

Lecture 10 Support Vector Machines II

,..., k N. , k 2. ,..., k i. The derivative with respect to temperature T is calculated by using the chain rule: & ( (5) dj j dt = "J j. k i.

On the number of regions in an m-dimensional space cut by n hyperplanes

Fermi-Dirac statistics

COS 511: Theoretical Machine Learning

Problem Set #2 Solutions

Computational and Statistical Learning theory Assignment 4

Optimal Marketing Strategies for a Customer Data Intermediary. Technical Appendix

PROBLEM SET 7 GENERAL EQUILIBRIUM

LECTURE :FACTOR ANALYSIS

Chapter 12 Lyes KADEM [Thermodynamics II] 2007

A NOTE ON CES FUNCTIONS Drago Bergholt, BI Norwegian Business School 2011

( ) 2 ( ) ( ) Problem Set 4 Suggested Solutions. Problem 1

Math1110 (Spring 2009) Prelim 3 - Solutions

Perfect Competition and the Nash Bargaining Solution

1 Definition of Rademacher Complexity

PHYS 705: Classical Mechanics. Canonical Transformation II

1 Matrix representations of canonical matrices

Special Relativity and Riemannian Geometry. Department of Mathematical Sciences

LK, represents the total amount of labor and capital available in the economy, P, P denote the prices

System in Weibull Distribution

Economics 8105 Macroeconomic Theory Recitation 1

Solutions for Homework #9

princeton univ. F 17 cos 521: Advanced Algorithm Design Lecture 7: LP Duality Lecturer: Matt Weinberg

A Proof of a Conjecture for the Number of Ramified Coverings of the Sphere by the Torus

Handling Overload (G. Buttazzo, Hard Real-Time Systems, Ch. 9) Causes for Overload

Solutions to selected problems from homework 1.

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

Solutions to exam in SF1811 Optimization, Jan 14, 2015

Recap: the SVM problem

Elastic Collisions. Definition: two point masses on which no external forces act collide without losing any energy.

Section 8.3 Polar Form of Complex Numbers

Irreversible Work of Separation and Heat-Driven Separation

12. The Hamilton-Jacobi Equation Michael Fowler

Chapter 1. Theory of Gravitation

The Second Anti-Mathima on Game Theory

k t+1 + c t A t k t, t=0

A FURTHER GENERALIZATION OF THE SOLOW GROWTH MODEL: THE ROLE OF THE PUBLIC SECTOR

Denote the function derivatives f(x) in given points. x a b. Using relationships (1.2), polynomials (1.1) are written in the form

f(x,y) = (4(x 2 4)x,2y) = 0 H(x,y) =

Bezier curves. Michael S. Floater. August 25, These notes provide an introduction to Bezier curves. i=0

14 Lagrange Multipliers

ON THE NUMBER OF PRIMITIVE PYTHAGOREAN QUINTUPLES

Our focus will be on linear systems. A system is linear if it obeys the principle of superposition and homogenity, i.e.

Machine Learning. What is a good Decision Boundary? Support Vector Machines

Slobodan Lakić. Communicated by R. Van Keer

Lecture 21: Numerical methods for pricing American type derivatives

Finite Vector Space Representations Ross Bannister Data Assimilation Research Centre, Reading, UK Last updated: 2nd August 2003

An Optimal Bound for Sum of Square Roots of Special Type of Integers

Quantum Particle Motion in Physical Space

APPENDIX 2 FITTING A STRAIGHT LINE TO OBSERVATIONS

How Strong Are Weak Patents? Joseph Farrell and Carl Shapiro. Supplementary Material Licensing Probabilistic Patents to Cournot Oligopolists *

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U)

Revision: December 13, E Main Suite D Pullman, WA (509) Voice and Fax

Final Exam Solutions, 1998

The Parity of the Number of Irreducible Factors for Some Pentanomials

SELECTED PROOFS. DeMorgan s formulas: The first one is clear from Venn diagram, or the following truth table:

Applied Mathematics and Computation

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography

Modelli Clamfim Equazione del Calore Lezione ottobre 2014

Two Conjectures About Recency Rank Encoding

Fall 2012 Analysis of Experimental Measurements B. Eisenstein/rev. S. Errede. ) with a symmetric Pcovariance matrix of the y( x ) measurements V

,, MRTS is the marginal rate of technical substitution

COS 521: Advanced Algorithms Game Theory and Linear Programming

Some Notes on Consumer Theory

Lecture 12: Discrete Laplacian

Problem Set 9 Solutions

Multiplicative Functions and Möbius Inversion Formula

and problem sheet 2

Modelli Clamfim Equazioni differenziali 7 ottobre 2013

MAE140 - Linear Circuits - Fall 13 Midterm, October 31

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal

Complex Numbers. x = B B 2 4AC 2A. or x = x = 2 ± 4 4 (1) (5) 2 (1)

Chapter 3: Oligopoly

University of California, Davis Date: June 22, 2009 Department of Agricultural and Resource Economics. PRELIMINARY EXAMINATION FOR THE Ph.D.

Some modelling aspects for the Matlab implementation of MMA

Online Appendix. t=1 (p t w)q t. Then the first order condition shows that

Solutions HW #2. minimize. Ax = b. Give the dual problem, and make the implicit equality constraints explicit. Solution.

By M. O'Neill,* I. G. Sinclairf and Francis J. Smith

INDEX NUMBER THEORY AND MEASUREMENT ECONOMICS. By W.E. Diewert. February CHAPTER 9: Two Stage Aggregation and Homogeneous Weak Separability

Chapter 12. Ordinary Differential Equation Boundary Value (BV) Problems

Lecture 11. minimize. c j x j. j=1. 1 x j 0 j. +, b R m + and c R n +

THE VIBRATIONS OF MOLECULES II THE CARBON DIOXIDE MOLECULE Student Instructions

Modified parallel multisplitting iterative methods for non-hermitian positive definite systems

Appendix for Causal Interaction in Factorial Experiments: Application to Conjoint Analysis

XII.3 The EM (Expectation-Maximization) Algorithm

Transcription:

Dvson of the Huantes and Socal Scences Preference and Deand Exaples KC Border October, 2002 Revsed Noveber 206 These notes show how to use the Lagrange Karush Kuhn Tucker ultpler theores to solve the proble of axzng utlty subject to a budget constrant: axze ux subject to the constrants p x 0, x 0, x where p 0 and > 0 The basc steps are: If possble, dentfy whch constrants wll bnd, so the Karush Kuhn Tucker frst-order condtons can be replaced by sple Lagrange condtons 2 Use the frst order condtons to express the optal expendtures x n ters of the Lagrange ultpler λ 3 Substtute these expendtures nto the budget constrant to solve for λ 4 Substtute λ back nto the expressons for x to obtan the deand functon Next we evaluate the utlty at the deand functon x p, to obtan the ndrect utlty functon vp, = u x p, Then we use drect coputaton to verfy Roy s Law: p, p = x p, Next I typcally nvert the ndrect utlty to get an expresson for the Hcksan expendture functon ep, υ, whch s the optal value functon for the constraned nzaton proble nze p x subject to the constrants ux υ, x 0 x The ndrect utlty and the expendture functon satsfy v p, ep, υ = υ, so we can copute the expendture functon by solvng for n ters of v, and changng the sybol for to e and the sybol for v to υ Note the dstncton between the Roan letter vee, v, and the Greek letter ypslon, υ [Ok, aybe ths s not the best set of fonts to show the dfference] Then we use the envelope theore to calculate the Hcksan copensated deands ˆxp, υ va p, υ = ˆx p, υ

KC Border Preference and Deand Exaples 2 On wrtng Lagrangeans Soe of y students have asked whether to wrte Lagrangeans wth a plus sgn or a nus sgn n front of the Lagrange ultplers, that s, whether to wrte fx + λgx or fx λgx In one sense, t doesn t atter, snce the only dfference s the sgn of the Lagrange ultpler, but n econoc probles because of the way we use the envelope theore, the Lagrange ultpler usually has an nterpretaton as a rate of exchange, or prce, and I usually want those nubers to be postve To do ths I use the followng rule of thub: Wrte the constrant functon g so that the constran s g 0 even f you know that t ust bnd, and could be replaced by g = 0 For a axzaton proble use a plus sgn, and for a nzaton proble use a nus sgn In all the exaples I can thnk of off the top of y head ths wll result n the quantty of nterest beng λ 0 nstead havng the quantty of nterest be λ where λ 0 Cobb Douglas preferences I: Logarthc for where α > 0, =,, n, and n α = u I x,, x n = α ln x Reark By conventon ln 0 =, a coon practce n convex analyss It s clear then that any optal consupton ust satsfy x 0, so we ay gnore the nonnegatvty constrants, and treat the frst order condtons as equaltes It s also clear that u s onotonc, so the budget constrant wll bnd 6 5 4 3 2 0 0 2 3 4 5 6 Representatve contours of 2 5 ln x + 3 5 ln x 2

KC Border Preference and Deand Exaples 3 Lagrangean: α ln x + λ x Frst order condtons, usng the bndng constrant = n x : So Sung over yelds as n α =, so becoes L = α x x λ = 0 =,, n α = λ x =,, n = λ x = α, that s, α s the fracton of ncoe spent on good, so the deand functon s x p, = α Thus the ndrect utlty functon s vp, = α α ln = ln α ln + α ln α 2 You ght be tepted to wrte ths as ln + n α ln α, whch s ore copact, but t akes t harder to read the dervatves The envelope theore assures us that that the partal dervatves of v are just the partal dervatves of the Lagrangean, so t ust be that λ = /, the argnal utlty of oney, whch dfferentaton shops us s /, whch we derved n the lne after Verfy Roy s Law usng the partals coputed fro 2: = α = α = x p, Recall that the expendture functon e gves the level of ncoe needed to acheve a gven level of utlty υ It therefore satsfes v p, ep, υ = υ, so we can copute the expendture functon by solvng 2 for n ters of v, and changng the sybol for to e and the sybol for v to υ So rewrte 2 to get υ = ln e α ln + α ln α,

KC Border Preference and Deand Exaples 4 rearrangng gves so exponentatng gves ln e = υ + α ln α ln α, n p α ep, υ = expυ n α α The Hcksan copensated deands are the dervatves of the expendture functon, so ˆx j p, υ = expυ n p α n α α 2 Cobb Douglas preferences II: Multplcatve for u II x,, x n = where α > 0, =,, n, and n α = Reark 2 Ths functonal for s gotten by transforng the prevous utlty by the ncreasng transforaton u II = expu I, so the deand should be the sae, but the ndrect utlty and expendture functons wll be transfored Note that n ths forulaton u s zero f any x s zero, so at any optu we jst have x 0, and as before we ay gnore the nonnegatvty constrants, and treat the frst order condtons as equaltes It s also clear that u s onotonc, so the budget constrant wll bnd Lagrangean: x α x α + λ x Frst order condtons, usng the bndng constrant = n x : L n x α = α x x λ = 0 =,, n So lettng u = n x α, Sung over yelds as n α =, so 3 becoes α u = λ x =,, n 3 u = λ α λ = λ x,

KC Border Preference and Deand Exaples 5 or x = α, that s, α s the fracton of ncoe spent on good, so the deand functon s Thus the ndrect utlty functon s x p, = α vp, = α α 4 Thus = Use 4 to verfy Roy s Law: n α α α j p 2 j = α α p = n α α n α α = = x jp, We can copute the expendture functon by solvng 4 for n ters of v Changng the sybol for to e and the sybol for v to υ, rewrte 4 as α α υ = e Rearrangng gves If we wsh, we can rewrte ths as ep, υ = υ α p α ep, υ = υ up uα, where u = u II s the Cobb Douglas utlty n ultplcatve for, and α = α,, α n So the Hcksan deands are gven by ˆx p, υ = ep, υ = α expυ nj= p j nj= α j

KC Border Preference and Deand Exaples 6 3 Logarthc quas-lnear preferences uy, x,, x n = y + β where β, α > 0, =,, n, and n α = α ln x 0 5 0 5 20 25 30-2 Representatve contours of y + ln x Note that each ndfference curve s a vertcal translate of every other curve, and that each ntersects the x-axs For reasons that wll becoe clear, let us ake y the nuérare p y = Then the Lagrangean s y + β α ln x + λ y x Frst order condtons, usng the bndng constrant = y + n x : wth λ = f y > 0, and So assung y > 0, ths gves β α x λ 0 λ = 0 =,, n x p, = α β In other words, the aount spent on good s ndependent of prces and ncoe Thus y p, = β Note that ths only works for β, whch corresponds to y 0 If < β, then λ >, and the reanng frst order condtons becoe β α x λ = 0 =,, n, 5 4 3 2 0 -

KC Border Preference and Deand Exaples 7 whch by the sae reasonng as n proble a gves Puttng ths all together yelds x p, = α y p, = x p, = β β 0 β β α β p α β Ths leads to the ndrect utlty vp, = Or, settng we have { β + β ln β n α ln + n α ln α β β ln n α ln + n α ln α β n hp = β α ln α α ln vp, = { β + β ln β + hp β hp + β ln β 5 Roy s Law: p = β α α β β = α for β for β = x p, Solvng 5 for = e n ters of υ gves

KC Border Preference and Deand Exaples 8 ep, υ = υ + β β ln β α ln + α ln α = υ + β β ln β hp for υ β ln β + hp, and otherwse n p α ep, υ = expυ/β n α α υ hp = exp β So the Hcksan deands are gven by ˆx j p, υ = ep, υ = β / exp υ hp β for υ υ β ln β + hp otherwse 4 Lnear preferences ux,, x n = α x where α 0, =,, n, and n α = Hnt: Reeber Kuhn Tucker The Lagrangean s α x + λ x The frst-order condtons are so λ α = ax, and of α Then α λ 0 =,, n, < λ ples x j = 0 So for now assue that s the unque axzer j = x jp, = 0 otherwse

KC Border Preference and Deand Exaples 9 When s not unque, there s no unque soluton, but convex cobnatons of the above are all vald deands That s, x p, = convex hull of { e j : α }, =,, n, where e j s the j th unt coordnate vector The ndrect utlty s thus vp, = α x = α = ax α = n α Roy s Law: p x jp, = And the expendture functon satsfes α p = 2 α = = 0 α = x p, = 0 j ep, υ = υ n α So the Hcksan deands are gven by ˆx j p, υ = { ep, υ υ/αj j = = 0 otherwse 5 Leontef fxed-proporton preferences ux,, x n = n{α x,, α n x n } where α 0, =,, n Hnt: Calculus s not very useful here

KC Border Preference and Deand Exaples 0 2 5 05 0 05 5 2 0 Representatve contours of n{x, x 2 } It s easy to see that α x p, = = α nx np,, denote ths coon value by c Then x p, = c α, and sung over gves = c n α, so x p, = α nj= The ndrect utlty s then vp, = nj= 6 Roy s Law: p = α n j= nj= 2 = α nj= = x p, And by 6 the expendture functon satsfes

KC Border Preference and Deand Exaples ep, υ = υ j= So the Hcksan copensated deands are: ˆx p, υ = υ α