Problem Set #2 Solutions

Similar documents
EconS 501 Micro Theory I 1 Recitation #12 Imperfect Competition

Chapter 3: Oligopoly

Cournot Equilibrium with N firms

Preference and Demand Examples

Advanced Microeconomics (ES30025)

Fermi-Dirac statistics

Optimal Marketing Strategies for a Customer Data Intermediary. Technical Appendix

Endogenous timing in a mixed oligopoly consisting of a single public firm and foreign competitors. Abstract

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

CHAPTER 6 CONSTRAINED OPTIMIZATION 1: K-T CONDITIONS

Collusive Decisions and Punishment under Demand and Cost Uncertainty

COS 511: Theoretical Machine Learning

Imperfect Competition

In the figure below, the point d indicates the location of the consumer that is under competition. Transportation costs are given by td.

= z 20 z n. (k 20) + 4 z k = 4

3.2. Cournot Model Cournot Model

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

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

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

Economics 101. Lecture 4 - Equilibrium and Efficiency

Hila Etzion. Min-Seok Pang

Market structure and Innovation

Chapter 12 Lyes KADEM [Thermodynamics II] 2007

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

The oligopolistic markets

LECTURE :FACTOR ANALYSIS

The Second Anti-Mathima on Game Theory

Lecture 3: Probability Distributions

Solutions for Homework #9

System in Weibull Distribution

Multipoint Analysis for Sibling Pairs. Biostatistics 666 Lecture 18

Lecture-24. Enzyme kinetics and Enzyme inhibition-ii

Least Squares Fitting of Data

Chapter 3 Describing Data Using Numerical Measures

THE CHINESE REMAINDER THEOREM. We should thank the Chinese for their wonderful remainder theorem. Glenn Stevens

Structure and Drive Paul A. Jensen Copyright July 20, 2003

CS-433: Simulation and Modeling Modeling and Probability Review

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

Page 1. SPH4U: Lecture 7. New Topic: Friction. Today s Agenda. Surface Friction... Surface Friction...

Mathematics Intersection of Lines

HMMT February 2016 February 20, 2016

CS286r Assign One. Answer Key

Answers Problem Set 2 Chem 314A Williamsen Spring 2000

MAE140 - Linear Circuits - Fall 13 Midterm, October 31

On Pfaff s solution of the Pfaff problem

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

XII.3 The EM (Expectation-Maximization) Algorithm

Excess Error, Approximation Error, and Estimation Error

Scattering by a perfectly conducting infinite cylinder

Quantity Precommitment and Cournot and Bertrand Models with Complementary Goods

One-sided finite-difference approximations suitable for use with Richardson extrapolation

Assortment Optimization under MNL

1 Review From Last Time

Copyright (C) 2008 David K. Levine This document is an open textbook; you can redistribute it and/or modify it under the terms of the Creative

Recap: the SVM problem

Price competition with capacity constraints. Consumers are rationed at the low-price firm. But who are the rationed ones?

The Fundamental Theorem of Algebra. Objective To use the Fundamental Theorem of Algebra to solve polynomial equations with complex solutions

Kernel Methods and SVMs Extension

Weighted Voting Systems

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009

Case A. P k = Ni ( 2L i k 1 ) + (# big cells) 10d 2 P k.

PROBLEM SET 7 GENERAL EQUILIBRIUM

Temperature. Chapter Heat Engine

Department of Electrical & Electronic Engineeing Imperial College London. E4.20 Digital IC Design. Median Filter Project Specification

Lecture 14: Bandits with Budget Constraints

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

Physics 3A: Linear Momentum. Physics 3A: Linear Momentum. Physics 3A: Linear Momentum. Physics 3A: Linear Momentum

8.1 Arc Length. What is the length of a curve? How can we approximate it? We could do it following the pattern we ve used before

Chapter One Mixture of Ideal Gases

Least Squares Fitting of Data

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan

1 Definition of Rademacher Complexity

THE SUMMATION NOTATION Ʃ

Introduction to Vapor/Liquid Equilibrium, part 2. Raoult s Law:

G = G 1 + G 2 + G 3 G 2 +G 3 G1 G2 G3. Network (a) Network (b) Network (c) Network (d)

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

Equilibrium with Complete Markets. Instructor: Dmytro Hryshko

total If no external forces act, the total linear momentum of the system is conserved. This occurs in collisions and explosions.

Managing Capacity Through Reward Programs. on-line companion page. Byung-Do Kim Seoul National University College of Business Administration

Lecture 10 Support Vector Machines II

Copyright 2017 by Taylor Enterprises, Inc., All Rights Reserved. Adjusted Control Limits for P Charts. Dr. Wayne A. Taylor

Final Exam Solutions, 1998

CS : Algorithms and Uncertainty Lecture 17 Date: October 26, 2016

and problem sheet 2

Let p z be the price of z and p 1 and p 2 be the prices of the goods making up y. In general there is no problem in grouping goods.

The Impact of the Earth s Movement through the Space on Measuring the Velocity of Light

EEL 6266 Power System Operation and Control. Chapter 3 Economic Dispatch Using Dynamic Programming

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

Lecture 12: Discrete Laplacian

Additional Codes using Finite Difference Method. 1 HJB Equation for Consumption-Saving Problem Without Uncertainty

A Cournot-Stackelberg Advertising Duopoly Derived From A Cobb-Douglas Utility Function

Physics 114 Exam 2 Fall 2014 Solutions. Name:

Eigenvalues of Random Graphs

Gravitational Acceleration: A case of constant acceleration (approx. 2 hr.) (6/7/11)

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

AS-Level Maths: Statistics 1 for Edexcel

Some modelling aspects for the Matlab implementation of MMA

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

Singular Value Decomposition: Theory and Applications

COS 521: Advanced Algorithms Game Theory and Linear Programming

Transcription:

4.0 Sprng 003 Page Proble Set # Solutons Proble : a) A onopolst solves the followng proble: ( Q ) Q C ( )= 00Q Q 0Q ax P Q wth frst-order condton (FOC) b) Gven the results fro part a, Q 90 Q = 0 Q P = 55 π = 4050 * = 45 * P = P = 55 * π = 05 On the other hand, a Cournot copettor, as n the punshent phases, solves the followng proble: ax 00 ( + j ) 0 wth FOC 90 j = 0 whch ples the followng best response functon: j C Due to syetry, we know that C = j ; hence, C C = j = 60 P C = 40 C π = 800 Last, suppose I decde to cheat; what uantty do I produce? I should produce the best response to what the other fr (who s not cheatng) s producng. Reeber, as long as I a cheatng, I ght as well cheat as well as possble. Hence, f I cheat, or devate fro the proposed eulbru, I should d ( 45)= 67. 5 P d = 43. 75 d π = 78.5 If I play along wth the cooperatve strategy, the PDV of y proft strea s

4.0 Sprng 003 Proble Set Solutons Page But, f I devate, the PDV s 05 05 ( + δ + δ +...) = δ 78.5 + 800( δ + δ +...) = 78. 5 + 800δ δ The cooperatve strategy s an eulbru f 05 800δ 78. 5 + δ δ 05 78.5 + 800δ ( δ ) = 78. 5 478. 5δ 478.5δ 53. 5 δ 0. 594 c) th a perod detecton lag, the devator can cheat for two perods. If I devate, the PDV s 3 800δ 78.5 + 78. 5δ + 800(δ + δ +...) = 78.5( + δ ) + δ The cooperatve strategy s an eulbru f 05 800δ 78.5( + δ ) + δ δ 05 78.5 δ )+ 800δ ( = 78. 5 478. 5δ δ 0. 594 δ 0.776 d) Frs copetng n prce n a hoogeneous goods arket ths s a Bertrand gae. Frs bd the prce down to argnal cost. Hence, the only eulbru s for both frs to charge a prce eual to the argnal cost of 0. The total arket uantty s 80. e) hen both frs cooperate by settng prce to be the onopoly prce P = 55, each fr akes a proft of 05 n each perod (See Part a). In the punshent phases, frs are Bertrand copettors and ake zero proft. hat about when one fr cheats? A fr s optal cheatng strategy s to set prce eual to P ε,.e., just undercut the other fr and capture the whole arket. In ths way, the cheatng fr essentally acts as the sole onopolst and akes the entre onopoly proft of 4050. The cooperatve strategy s an eulbru f

4.0 Sprng 003 Proble Set Solutons Page 3 05 4050 δ δ δ Prce copetton s ore lkely to sustan a onopoly outcoe because, due to the ferce copetton nspred by prce copetton, the punshent s uch harsher. f) Frs are capacty constraned no one can produce the full arket deand at a prce eual to the argnal cost. Frs keep undercuttng the rval s prce down to argnal cost, at whch pont no fr akes proft and can beneft fro rasng ts prce. Snce nether fr can eet the full arket deand, soe consuers are forced to ether not buy or buy fro the other fr. Frs then begn agan to undercut each other back down n prce. In ths way, prces cycle up and down. Ths phenoenon, known as Edgeworth cycles, actually happens. Proble : a) The jont proft-axzaton outcoe s the sae as n part a of Proble. In the cartel, the total producton uota, allotted soehow aong the fve frs, ust be Q = Q The cartel can actually allocate the 90 unts any way t wants; evenly dvdng t aong the fve frs sees lkely. The prce s the onopoly prce of 55; profts are dstrbuted aong the fve frs n the sae way that the uanttes are. b) In the fve fr cartel, each fr akes proft 80 (=4050/5). Suppose one fr decdes not to enter nto the cartel. Let Q 4 be the total uantty produced by the four frs n the cartel. Then the non-cartel fr solves the followng proble: ax 00 ( Q 4 + ) 0 wth FOC 90 Q 4 = 0 Q 4 Knowng the non-cartel fr wll behave n ths way, the cartel axzes ts jont proft as follows: ax 00 Q 4 Q 4 +90 Q 4 Q 4 0Q 4 = 90 The FOC s 90 Q 4 = 0 Hence, the eulbru s Q 4 Q Q 4 4

4.0 Sprng 003 Proble Set Solutons Page 4 Q 4 = 45 P = 3. 5 It thus pays not to enter the cartel. non cartel π = ( P 0) = 0. 5 c) Suppose there are n frs. Let us check f an n-fr cartel s stable; does a fr have ncentve not to enter. Frs n the n-fr cartel ake proft 4050/n. Usng reasonng analogous to part b, a non-cartel fr solves the followng proble: ax 00 ( Q n + ) 0 wth FOC 90 Q n = 0 Q n Knowng the non-cartel fr wll behave n ths way, the cartel axzes ts jont proft as follows: ax 90 Q n Q n Q n The FOC s 90 Q n = 0 Hence, the eulbru s Q n Q n = 45 P = 3. 5 non cartel π = 0.5 Therefore, coparng the proft to frs nsde the n-fr cartel to the proft of the noncartel fr suggests n < 4 Stable cartel n = 4 Indfferent n > 4 Incentve to not jon Due to the syetry of the frs, there s no dfference between the (n-)-fr cartel and a sngle Stackelberg leader; look at the FOC to confr ths. th a large nuber of frs, collusve profts are heavly dluted; thus, despte the fact that a Stackelberg follower does coparatvely worse than the leader, t stll pays to hold out and gan the undsspated Stackelberg follower profts. Proble 3: a) Each fr solves the followng proble:

4.0 Sprng 003 Proble Set Solutons Page 5 wth FOC ax P Q C ( Q )= (P 0 )( 80 P + P j ) P 80 4P + P j + 0 = 0 00 + P j P = 4 Due to syetry, we know that P R = P ; hence, P R = P = 00 / 3 wth FOC π = 70 3 = 088. 8 b) ave knows Rah s best response functon s 00 + P = PR 4 because Rah s proft-axzaton proble has not changed; therefore, ave solves the followng proble: ax (P 0 00 + P ) 80 P + P 4 Therefore, 80 4P + 5 + P + 0. 5 = 0.5 3.5 P = 0 P = 35 P = 35 P = 33. 75 R π = 093. 75 π R = 8. 5 ave should take advantage of the cotent strategy because t (slghtly) ncreases hs profts. Proble 4: For splcty, label the ponts at whch the players ove as nodes one through seven, nuberng the up and down, left to rght; player one oves at nodes, 4, 5, and 6, whle player two oves at nodes and 3. The two nodes connected by the dotted lne are n fact a sngle node. The dotted lne sgnfes the fact that player one does not know at whch node he was. To fnd an SPNE, we eploy backwards nducton. At node 6, player one wll choose u, whether player two had chosen U or D at node three. At node fve, player one wll choose d ; at node four, he wll choose u. Steppng back to nodes two and three, we can decde what s optal for player two knowng what player one wll do at the ensung nodes. At node three, player two wll choose U; at node two, he wll also choose U. Last,

4.0 Sprng 003 Proble Set Solutons Page 6 player one wll begn the gae by choosng d. e can represent the eulbru strateges for the two players as follows: ( d, u, d, u ) ( U, U ) where the frst vector s the strateges for player one and the second vector s the strateges for player two, strateges are ordered by node. Proble 5: a) Ths s analogous to the GE/estnghouse case when GE began postng all ts transacton prces. Ths s lkely to facltate colluson because t ontors the actons of the frs; colludng frs leave ontorng and detectng cheatng to the Cosson. b) The envronental polces are lkely to nhbt colluson. Colluson s enforced by the threat of future punshent,.e., the attractveness of colludng s the collusve future revenue strea. The fact that the product wll be banned n fve years truncates the possble beneft to colludng; less ncentve to collude s ore ncentve to not collude. c) Uncertanty tends to hnder colluson prarly by akng detecton ore dffcult. For exaple, a low prce ay ndcate low deand or cheatng be a fr, producng less output. d) Multarket contact s lkely to facltate colluson. It akes punshent easer; as referred to n the lecture notes, May relax ncentve constrant; transfer slack n nocheatng condton fro one arket to another to expand scope of cooperatve behavor. Proble 6: a) The ajor reason the two prces are dfferent s probably prce dscrnaton. Suppose soe consuers have uch lower search costs than others; they enjoy surfng the web bargan-huntng or are ore techno-savvy. Slarly, t ay be dscrnaton between dfferent types of consuers; college students and yuppes pay less than older adults. Busness travelers ay not search around for low prces n the sae way that pleasure travelers do. Possbly, the two roos are not really dentcal; there ay be sall dfferences between the products that we do not observe. b) The hotels ay want to prevent prces fro beng posted because t helps attract ore consuers to the lower prces. Fewer consuers pay the hgh prces relshed by the hotels. Prcelne, on the other hand, ay not want prces posted because t akes ther servce less attractve n the future. th prces ade publc, people ay be able to bargan drectly wth the hotels, or prces ay becoe standardzed at soe lower prce, akng Prcelne obsolete.