collusion with private monitoring and durable goods

Upload: caae-comercial-uc

Post on 05-Apr-2018

220 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    1/72

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    Collusion with Private Monitoring andDurable Goods

    Enrique Ide Carvallo2012

    Ponticia Universidad Cat olica de ChileInstituto de Economa

    March 29, 2012

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    2/72

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    MotivationLiterature ReviewSketch of my Model

    Contents

    1 Introduction and MotivationMotivationLiterature ReviewSketch of my Model

    2 General ModelAssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    3 Consumers Patience: A Simple ExampleOutlineResults

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    3/72

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    MotivationLiterature ReviewSketch of my Model

    Motivation

    Two of the favorite hobbies of IO researchers...

    How durability affects market power and industry congurationStudy of Collusion. Lately, involving asymmetric information amongrms.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    4/72

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    MotivationLiterature ReviewSketch of my Model

    Motivation

    Two of the favorite hobbies of IO researchers...

    How durability affects market power and industry congurationStudy of Collusion. Lately, involving asymmetric information amongrms.

    Idea: connect those two strands of literature.

    Kind of asymmetric info we are studying today? Private Monitoring.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    I d i d M i i

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    5/72

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    MotivationLiterature ReviewSketch of my Model

    Motivation

    So...

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    I t d ti d M ti ti

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    6/72

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    MotivationLiterature ReviewSketch of my Model

    Literature Review - Durable Good Markets

    Coase Conjecture (Coase, 1972): Loss Monopoly Power in DurableGood Markets with Patient Consumers

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    7/72

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    MotivationLiterature ReviewSketch of my Model

    Literature Review - Durable Good Markets

    Coase Conjecture (Coase, 1972): Loss Monopoly Power in DurableGood Markets with Patient Consumers

    However Ausubel and Deneckere (1987); Gul (1987); Dutta et al.(2007) showed that a Duopoly can overcome the conjecture

    Presence of a rival, helps credibly commit to a price over time

    Moreover, same authors prove that more patient consumers helpscollusion

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    8/72

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    MotivationLiterature ReviewSketch of my Model

    Literature Review - Durable Good Markets

    Figure 1: Collusion with Durable Goods and Perfect Monitoring - Ausubel andDeneckere (1987); Gul (1987); Dutta et al. (2007)

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods Introduction and Motivation M i i

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    9/72

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    MotivationLiterature ReviewSketch of my Model

    Literature Review - Collusion with Asymetric Information

    Big area of research lately, although it has intrigued researchers atleast since Stigler (1964)

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation M ti ti

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    10/72

    t oduct o a d ot vat oGeneral Model

    Consumers Patience: A Simple ExampleReferences

    MotivationLiterature ReviewSketch of my Model

    Literature Review - Collusion with Asymetric Information

    Big area of research lately, although it has intrigued researchers atleast since Stigler (1964)

    Imperfect Public Monitoring: some variable is public knowledge(either price or quantity)

    Green and Porter (1984); Porter (1983a); Abreu et al. (1986);Harrington and Skrzypacz (2007)

    Private Monitoring: rms price AND quantities are privateinformationHarrington and Skrzypacz (2011)

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation Motivation

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    11/72

    General ModelConsumers Patience: A Simple Example

    References

    MotivationLiterature ReviewSketch of my Model

    Literature Review - Collusion with Asymetric Information

    Green and Porter (1984): Collusion when price is public knowledge ispossible by periodically switching to a punishment phasePresence of Price Wars in equilibrium path.

    Shortcomings...Empirically: we dont usually see price wars in cartelized industriesTheoretically: when monitoring is through sales Harrington andSkrzypacz (2007) showed that collusion is unsustainable withsymmetric punishments

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation Motivation

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    12/72

    General ModelConsumers Patience: A Simple Example

    References

    MotivationLiterature ReviewSketch of my Model

    Literature Review - Collusion with Asymetric Information

    Green and Porter (1984): Collusion when price is public knowledge ispossible by periodically switching to a punishment phasePresence of Price Wars in equilibrium path.

    Shortcomings...Empirically: we dont usually see price wars in cartelized industriesTheoretically: when monitoring is through sales Harrington andSkrzypacz (2007) showed that collusion is unsustainable withsymmetric punishments

    Private Monitoring: Harrington and Skrzypacz (2011) and Chan andZhang (2009) proved that with asymmetric punishments, collusion ispossible even when self-reported sales are used to monitorcompliance.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationG l M d l Motivation

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    13/72

    General ModelConsumers Patience: A Simple Example

    References

    MotivationLiterature ReviewSketch of my Model

    Sketch of my Model

    Overlapping Generation Model with Durable Goods in the fashion of Dutta et al. (2007), but stochastic environment.

    Private Monitoring between rms

    Consumers cannot see the effective price charged to anotherconsumer (realistic assumption)

    We assume rms use Green and Porter (1984)s trigger strategies

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationG l M d l Motivation

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    14/72

    General ModelConsumers Patience: A Simple Example

    ReferencesLiterature ReviewSketch of my Model

    Sketch of my Model - Results

    Collusion is sustainable

    But contrary to the previous ndings, more patient consumers donot unambiguously facilitate collusion

    Reason: new counteracting effect of consumers patience: the demand speculation effect

    Given price-wars in equilibrium path, more patient consumers wait

    until bad realization of demand forces rm to switch phase to buycheaper harms collusive prot difficults collusion.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model Motivation

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    15/72

    General ModelConsumers Patience: A Simple Example

    ReferencesLiterature ReviewSketch of my Model

    Sketch of my Model - Results

    More generally when consumers may delay purchases, if asymmetric,transfer-style schemes are available, symmetric punishment phasesare an inefficiently costly discipline device.

    Transfer-schemes leave the pro-collusive effect of consumerspatience while eliminating the demand speculation effect

    Adds up to the theoretical contribution explaining why we do not see

    periodical Price Wars in cartelized industries.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model Assumptions

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    16/72

    General ModelConsumers Patience: A Simple Example

    ReferencesCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    Contents

    1 Introduction and MotivationMotivationLiterature ReviewSketch of my Model

    2 General ModelAssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    3 Consumers Patience: A Simple ExampleOutlineResults

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model AssumptionsCh i i h C i Ph

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    17/72

    General ModelConsumers Patience: A Simple Example

    ReferencesCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    Assumptions

    Assumptions about PopulationContinuum of consumers of mass 1, indexed by k , distributed on[0, 1]. Fraction t isyoungand 1 t are old. t i .i .d . r.v. dist. cdf G : [0, 1] [0, 1], where E [ t ] = (0 , 1).

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model AssumptionsCh t i i g th C ti Ph

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    18/72

    Consumers Patience: A Simple ExampleReferences

    Characterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    Assumptions

    Assumptions about PopulationContinuum of consumers of mass 1, indexed by k , distributed on[0, 1]. Fraction t isyoungand 1 t are old. t i .i .d . r.v. dist. cdf G : [0, 1] [0, 1], where E [ t ] = (0 , 1).

    Assumptions regarding ConsumersThe good is homogeneous and perfectly durableConsumers differ on their valuation v [0, v ] for the good.The valuation of a newly born generation always dist. cdf F : [0, v ] [0, 1]Buyers have rational expectations and disc. future by factor [0, 1]But because they die with chance E [ t ] = effective discountfactor is (1 ) .Resale is not allowed, and in case of a price tie, we simply assumethat the demand splits by halves.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model AssumptionsCharacterizing the Cooperative Phase

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    19/72

    Consumers Patience: A Simple ExampleReferences

    Characterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    Assumptions

    Assumptions about FirmsIndustry composed by two identical rms with constant marginal cost= 0 and same discount factor

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model AssumptionsCharacterizing the Cooperative Phase

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    20/72

    Consumers Patience: A Simple ExampleReferences

    Characterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    Assumptions

    Assumptions about FirmsIndustry composed by two identical rms with constant marginal cost= 0 and same discount factor

    Informational AssumptionsFirms cannot observe neither current or past charged prices, norquantities sold by rivalConsumers cannot see the effective price charged to anotherconsumer rms may charged different prices to different buyers.

    However consumers personal characteristic are private information.Price offers must be made simultaneously.Neither rms nor consumers are able to observe the currentrealization of t

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    C P i A Si l E lAssumptionsCharacterizing the Cooperative Phase

    http://goforward/http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    21/72

    Consumers Patience: A Simple ExampleReferences

    Characterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    Assumptions

    Denition

    The players action space consist of all p i , t (k ) functions satisfying p i , t : [0, 1] [0, + )

    A contingent strategy is therefore an innite sequence s i = {p i , t (k )}t = t =0 with p i ,0(k ) given, and where s i , t +1 : R

    ( t +1)+

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleAssumptionsCharacterizing the Cooperative Phase

    http://goforward/http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    22/72

    Consumers Patience: A Simple ExampleReferences

    g pOptimal Deviation and the Cartels Problem

    Assumptions

    Strategy considered:

    Start playing p i ,0(k ) = p k [0, 1] (cooperative phase). Then,If in cooperative phase in period t 1 and

    q i , t 1 q then p i , t (k ) = p k [0, 1] and remain in thecooperative phase.q i , t 1 < q then p i , t (k ) = 0 k [0, 1] and go to the punishmentphase

    If in the th period of the punishment phase in period t 1 and

    < T 1 then p i , t (k ) = 0 k [0, 1] and continue in thepunishment phase T 1 then p i , t (k ) = p k [0, 1] and go to the cooperativephase

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    AssumptionsCharacterizing the Cooperative Phase

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    23/72

    Consumer s Patience: A Simple ExampleReferences

    g pOptimal Deviation and the Cartels Problem

    Characterizing the Cooperative Phase

    Suppose P (Price War ) = G () if both rms always cooperateThen E t p t +1 = [1 P (Price War )] p = [1 G ()] p

    But

    t i .i .d

    Et p = [1 G ()] p

    >t .

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    AssumptionsCharacterizing the Cooperative Phase

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    24/72

    Consumer s Patience: A Simple ExampleReferences Optimal Deviation and the Cartels Problem

    Characterizing the Cooperative Phase

    Suppose P (Price War ) = G () if both rms always cooperateThen E t p t +1 = [1 P (Price War )] p = [1 G ()] p

    But

    t i .i .d

    Et p = [1 G ()] p

    >t .

    Now, we can separate population in three groups:1 Young consumers2 Old consumers who have not experience a price war

    3 Old consumers who have experience a price war

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    AssumptionsCharacterizing the Cooperative PhaseO i l D i i d h C l P bl

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    25/72

    Consumer s Patience: A Simple ExampleReferences Optimal Deviation and the Cartels Problem

    Characterizing the Cooperative Phase

    Young Consumers Problem

    max U = max {v p , (v E t p t +1 ), 2 (v E t p t +2 ), . . .} (1)

    However, because E t p = [1 G ()] p > t and < 1, we cansimplify it to

    max U = max {v p , (v E t p )} (2)

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    26/72

    p pReferences Optimal Deviation and the Cartels Problem

    Characterizing the Cooperative Phase

    Young Consumers Problem

    max U = max {v p , (v E t p t +1 ), 2 (v E t p t +2 ), . . .} (1)

    However, because E t p = [1 G ()] p > t and < 1, we cansimplify it to

    max U = max {v p , (v E t p )} (2)

    Thus he buys today if and only if:

    v p (v E

    t p ) v v C

    (p , q , ) =

    p E t p

    1 (3)

    Therefore total demand emanated from this group is

    2

    {1 F [v C (p , q , )]} (4)

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    27/72

    References Optimal Deviation and the Cartel s Problem

    Characterizing the Cooperative Phase

    Old consumers who have not experience a price war

    Have only received offersp = p . Hence they bought the good whenyoung if v v C (p , q , ), or have not otherwise.

    But, in this period they also received p = p , hence they only buy if v v C (p , q , ). Thus their expected demand is zero.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    28/72

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    R f

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    29/72

    References Optimal Deviation and the Cartel s Problem

    Characterizing the Cooperative Phase

    Hence

    P (q i , t q ) = P t 2

    {1 F [v C (p , q , )]} q = G 2 q

    1 F [v C (p , q , )]

    Which implies that

    (p , q , ) =2 q

    1 F [v C (p , q , )]

    Thus, the expected future price E t p is:

    E t p = p [1 G ()] = p 1 G 2 q

    1 F [v C (p , q , )]

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    30/72

    References p

    Characterizing the Cooperative Phase

    Which, given rational expectations means that v C must satisfy

    v C (p , q , ) =p

    1 1 1 G

    2 q 1 F [v C (p , q , )]

    (5)

    Proposition

    There exists a continuous increasing functionv C : [0, v ]x [0, 1] x [0, 1 ] [0, v ], convex in p with

    1 v C (p = 0 , q , ) = 0 (q , ) [0, 1]x [0, 1 ]2 v C (p , q = 0 , ) = p (p , ) [0, v ]x [0, 1 ]3 v C (p , q , = 0) = p (p , q ) [0, v ]x [0, 1]4 v C (p , q = 1 , ) = p 1 (p , ) [0, v ] x [0, 1 ]

    that satises condition ( 5 ).

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    31/72

    References

    Characterizing the Cooperative Phase

    Note that because v C (p , q , ) is increasing in (and therefore in ). Wehave

    C = p 2 {1 F [v C (p , q , )]} C = p 2 {1 F [v

    C (p , q , )]}

    Iff . Moreover, if we assume that v C () is differentiable, it isstraightforward to prove that the price p that maximizes this expression isdecreasing in patience of consumers.

    These facts will be the heart of our argument.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    32/72

    References

    Optimal Deviations and the Cartels Problem

    For the Cartel to be successful, must be sustainable for every possiblehistory of { }

    = t 1 =0

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    33/72

    References

    Optimal Deviations and the Cartels Problem

    For the Cartel to be successful, must be sustainable for every possiblehistory of { }

    = t 1 =0

    The incentives to deviate are increasing in residual demand fromoldconsumers without experience of punishment phase yet.

    more stringent history of s: set of old consumers who alreadybought the good equal to zero

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    34/72

    Optimal Deviations and the Cartels Problem

    DenitionA deviation for rm i consists on region U and a price p (k ) < p k U

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    35/72

    Optimal Deviations and the Cartels Problem

    DenitionA deviation for rm i consists on region U and a price p (k ) < p k U

    Suppose for example U = [0 , ], then

    q j , t (p , q ; ) = t [1 ]

    2{1 F [v C (p , q , )]}

    Thus the expected future price predicted by consumers living in the [0, ]interval will be:

    E t p = p 1 G 2q

    {1 }{1 F [v C (p , q , )]}= p 1 G

    1

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    36/72

    Optimal Deviations and the Cartels Problem

    DenitionA deviation for rm i consists on region U and a price p (k ) < p k U

    Suppose for example U = [0 , ], then

    q j , t (p , q ; ) = t [1 ]

    2{1 F [v C (p , q , )]}

    Thus the expected future price predicted by consumers living in the [0, ]interval will be:

    E t p = p 1 G 2q

    {1 }{1 F [v C (p , q , )]}= p 1 G

    1

    Notice that this price expectation is independent of the actual deviationprice offered p (k )

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    37/72

    Optimal Deviations and the Cartels Problem

    Now, given this expectations, consumer belonging to the fraction willbuy the good today to rm i only if v v D ( p , , p , q , ). Wherev

    D

    ( p (k ), , p , q , ) = p E t p

    1 =

    1

    1 p (k ) p 1 G

    1

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    38/72

    Optimal Deviations and the Cartels Problem

    Now, given this expectations, consumer belonging to the fraction willbuy the good today to rm i only if v v D ( p , , p , q , ). Wherev

    D

    ( p (k ), , p , q , ) = p E t p

    1 =

    1

    1 p (k ) p 1 G

    1

    The optimal deviation price for consumer k [0, ]:

    maxp (k ) [0, p ]

    p (k ) 1 F v D + (1 ) max 0, F v C F v D

    But then is clear that optimal value of p (k ) is independent of k . Thus

    p (k ) = p for all k [0, ].Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    39/72

    Optimal Deviations and the Cartels Problem

    Then, this period deviation prots are given by D + (1 ) C t , where D comes from the maximization of the previous problem.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    40/72

    Optimal Deviations and the Cartels Problem

    Then, this period deviation prots are given by D + (1 ) C t , where D comes from the maximization of the previous problem.

    Notice the trade-off the deviator faces:The greater the deviation region, the greater are deviation prots.However, the greater the deviation region, the greater probability of switching to a punishment phase tomorrow

    decreases todays demand in the deviation region forcing thedeviator rm to offer lower deviation prices

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    41/72

    Optimal Deviations and the Cartels Problem

    We can now obtain our incentive compatibility constraint. DeneV (p , = 0) as the value function of rm i if he does not deviates.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    42/72

    Optimal Deviations and the Cartels Problem

    We can now obtain our incentive compatibility constraint. DeneV (p , = 0) as the value function of rm i if he does not deviates.Then,

    V (p , = 0) = C t (p , q ; )+ G 2 q 1 F [v C (p , q , )]T V (p , = 0)

    + 1 G 2 q

    1 F [v C (p , q , )]V (p , = 0) (6)

    V (p , 0) = C t (p , q ; )

    1 + G () { T +1 }(7)

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    43/72

    Optimal Deviations and the Cartels Problem

    Now the incentive compatibility constraint requires that

    ( = 0) arg

    max D

    (,

    ) + (1 ) C

    t (p , q ; )

    + G

    1 T V (p , 0) + 1 G

    1

    V (p , 0) (8)

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    AssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    44/72

    Optimal Deviations and the Cartels Problem

    Now the incentive compatibility constraint requires that

    (

    = 0) arg

    max D

    (,

    ) + (1

    ) C

    t (p ,

    q ;

    )

    + G

    1 T V (p , 0) + 1 G

    1

    V (p , 0) (8)

    The cartels problem is therefore maximizes (7) subject to ( 8).

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    OutlineResults

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    45/72

    Contents

    1 Introduction and MotivationMotivationLiterature ReviewSketch of my Model

    2 General ModelAssumptionsCharacterizing the Cooperative PhaseOptimal Deviation and the Cartels Problem

    3 Consumers Patience: A Simple ExampleOutlineResults

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    OutlineResults

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    46/72

    Consumers Patience: A Simple ExampleOutline - Assumptions

    Things get simpler if t is discrete-binary r.v.

    Turns out in this case q independent of p and T

    solve problem sequentially: obtain rst q and then for p and T

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    47/72

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    OutlineResults

  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    48/72

    Consumers Patience: A Simple ExampleOutline - Cooperative Phase

    In this case, if no rm ever deviates P (Price War | No Dev ) = 1 E t p = p > t

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    OutlineResults

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    49/72

    Consumers Patience: A Simple ExampleOutline - Cooperative Phase

    In this case, if no rm ever deviates P (Price War | No Dev ) = 1 E t p = p > t

    Thus,

    v C (p , ) = p 1 1

    (10)

    E t q i = 1 F (v C )

    2 C =

    p 1 F (v C )

    2Again, it is self-evident that C is decreasing in and that the price thatmaximizes it, is also decreasing in that parameter.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    OutlineResults

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    50/72

    Consumers Patience: A Simple ExampleOutline - Cooperative Phase

    Figure 2: Optimal Prices - Linear Valuations

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    OutlineResults

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    51/72

    Consumers Patience: A Simple ExampleOutline - Optimal Deviation

    Suppose the rival deviates on U = [0 , ] of the market.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    OutlineResults

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    52/72

    Consumers Patience: A Simple ExampleOutline - Optimal Deviation

    Suppose the rival deviates on U = [0 , ] of the market.

    Because P (Price War | No Dev ), and therefore C , independent of q ,

    optimal q should maximizeP

    (Price War | Dev )Formally,

    q = arg maxq

    {P (Price War | Dev ) = (1 )+

    P {1

    } 1

    F (v C )

    2 q (11)

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    53/72

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    OutlineResults

    l l

  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    54/72

    Consumers Patience: A Simple ExampleOutline - Optimal Deviation

    Obvious solution:

    q = 1 F (v C )

    2 (Price War | Dev ) = (1 ) + = 1 (12)

    Independent of the deviation fraction .

    Therefore, if a rm is ever going to deviate = 1 v D = p D

    1 .Then per-unit deviation prots are:

    maxp D [0, p ] D

    = p D

    1 F (v

    D

    ) + 1 max 0, F (v

    C

    ) F (v D

    )+ (1 )max 0, F (v C ) F (v D )

    (13)

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral Model

    Consumers Patience: A Simple ExampleReferences

    OutlineResults

    C P i A Si l E l

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    55/72

    Consumers Patience: A Simple ExampleOutline - Cartels Problem

    Dene V as value function on date t given cooperative phase. Then

    V = C + (1 ) T V + V V = C

    1 { + (1 ) T }(14)

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    OutlineResults

    C P i A Si l E l

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    56/72

    Consumers Patience: A Simple ExampleOutline - Cartels Problem

    Dene V as value function on date t given cooperative phase. Then

    V = C + (1 ) T V + V V = C

    1 { + (1 ) T }(14)

    The incentive-compatibility constraint requires:

    D + T +1 V C + (1 ) T V + V

    D

    C 1 1 T

    1 { + (1 ) T } (15)

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    OutlineResults

    C P ti A Si l E l

    http://find/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    57/72

    Consumers Patience: A Simple ExampleOutline - Cartels Problem

    Dene V as value function on date t given cooperative phase. Then

    V = C + (1 ) T V + V V = C

    1 { + (1 ) T }(14)

    The incentive-compatibility constraint requires:

    D + T +1 V C + (1 ) T V + V

    D

    C 1 1 T

    1 { + (1 ) T } (15)

    Therefore the Cartels Problem is to maximizes (14) subject to ( 15).

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    OutlineResults

    C P ti A Si l E l

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    58/72

    Consumers Patience: A Simple ExampleResults - Simulations

    Two different though experiments:1 Assume that collusion is sustainable for a given parameter

    conguration and collusive price. How does the length of thepunishment period changes in response to changes in , to keep theagreement sustainable?

    2 Given a parameter conguration, how does the cartels value

    function changes in response to changes in

    , and how does thecollusion outcome evolves?

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and MotivationGeneral ModelConsumers Patience: A Simple Example

    References

    OutlineResults

    Consumers Patience: A Simple Example

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    59/72

    Consumers Patience: A Simple ExampleResults: Simulations - Though Experiment 1

    Figure 3: = 0 .5 / = 0 .5 Figure 4: = 0 .75 / = 0 .5

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    OutlineResults

    Consumers Patience: A Simple Example

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    60/72

    Consumer s Patience: A Simple ExampleResults: Simulations - Though Experiment 2

    Figure 5: = 0 .9 Figure 6: = 0 , 9

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    OutlineResults

    Consumers Patience: A Simple Example

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    61/72

    Consumer s Patience: A Simple ExampleResults: Simulations - Collusive Outcome

    Collusive Outcome for: = 0 .9 = 0 , 5 = 0 , 75

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    OutlineResults

    Consumers Patience: A Simple Example

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    62/72

    Consumer s Patience: A Simple ExampleResults - Intuition

    More patient consumers do not unambiguously facilitate collusion.

    This contrasts with previous results found in the literature.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    OutlineResults

    Consumers Patience: A Simple Example

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    63/72

    Consumer s Patience: A Simple ExampleResults - Intuition

    More patient consumers do not unambiguously facilitate collusion.

    This contrasts with previous results found in the literature.

    Intuition: New, counteracting effect: the demand speculation effect Given price-wars in equilibrium path, more patient consumers waituntil bad realization of demand forces rm to switch phase to buycheaper harms collusive prot difficults collusion.

    Thus the overall effect: consumer-induced discipline + the demand speculation effect , can go either way.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    OutlineResults

    Consumers Patience: A Simple Example

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    64/72

    Consumer s Patience: A Simple ExampleResults - Intuition

    More generally, when consumers may delay purchases, symmetricpunishment phases are an inefficiently costly discipline mechanism.

    Firms would be better-off with asymmetric, transfer-style, schemes. Eliminates the anti-collusive effect of active consumers, leavingthe pro-collusive effect only

    Adds up to the theoretical contribution explaining why we do not seeperiodical Price Wars in cartelized industries.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    OutlineResults

    Consumers Patience: A Simple Example

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    65/72

    Consumer s Patience: A Simple ExampleResults - Intuition

    More generally, when consumers may delay purchases, symmetricpunishment phases are an inefficiently costly discipline mechanism.

    Firms would be better-off with asymmetric, transfer-style, schemes. Eliminates the anti-collusive effect of active consumers, leavingthe pro-collusive effect only

    Adds up to the theoretical contribution explaining why we do not see

    periodical Price Wars in cartelized industries.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    OutlineResults

    Consumers Patience: A Simple Example

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    66/72

    Consumer s Patience: A Simple ExampleResults - Robustness

    Results do not change if we allow other distributions for t .

    In fact, for non-binary distributions, it is more likely that theanti-collusive effect of consumers patience dominates:

    Reason: More complex distributions difficult cartels inference

    should weaken the pro-collusive effects of active consumers andshould not affect the anti-collusive effect.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    OutlineResults

    Consumers Patience: A Simple Example

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    67/72

    Consumer s Patience: A Simple ExampleResults - Robustness

    Result should be robust to more complex punishment-phasesstrategies (see Abreu et al. (1986))

    Results driven by the existence , and not the form of the punishmentphase.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    OutlineResults

    Some Ideas

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    68/72

    The model is well-suited for analyzing whether the Cartel shouldagree on a price or in a geographical/buyer allocation.

    I have not prove it, but intuitively in this setting the Cartel shouldnot use buyer allocation because it looses the discipline coming fromactive consumers,

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    Abreu, D., D. Pearce, and E. Stacchetti ,Optimal Cartel Equilibriawith Imperfect Monitoring, Journal of Economic Theory, 1986, 39,

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    69/72

    with Imperfect Monitoring, Journal of Economic Theory , 1986, 39 ,251269.

    Athey, S. and K. Bagwell ,Optimal Collusion with PrivateInformation, RAND Journal of Economics , 2001, 32 (3), 428465.Ausubel, L. and R. Deneckere ,One is Almost Enough for Monopoly,

    RAND Journal of Economics , 1987, 18 (2), 255274.Bulow, Jeremy ,Durable-Goods Monopolist, Journal of Political

    Economy , 1982, 90 (2), 314332.Carter, Michael , Foundations of Mathematical Economics , MIT Press,

    2001.Chan, J. and W. Zhang ,Price Wars, Side-payments and Collusion

    Enforcement, June 2009. Unpublished Manuscript.Coase, Ronald , Durability and Monopoly, Journal of Law and

    Economics , 1972, 15 (1), 143149.Compte, Olivier ,Secret Price Cutting and Strategic Buyers: An

    Illustration of the Divide and Conquer Tactic,April 2000. UnpublishedManuscript.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    de la Fuente, Angel , Mathematical Methods and Models for Economist ,Cambridge University Press, 2000.

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    70/72

    Ca b dge U ve s ty ess, 000.Dutta, P., A. Matros, and J. Weibull ,Long-run Price Competition,

    RAND Journal of Economics , 2007, 38 (2), 291313.Ellison, Glenn ,Theories of Cartel Stability and the Joint ExecutiveCommittee, RAND Journal of Economics , 1994, 25 (1), 3757.

    Green, E. and R. Porter ,Noncooperative Collusion Under ImperfectPrice Information, Econometrica , 1984, 52 (1), 87100.

    Gul, F., H. Sonnenschein, and R. Wilson ,Foundation of DynamicMonopoly and the Coase Conjecture, Journal of Economic Theory ,1986, 39 , 155190.

    Gul, Faruk ,Noncooperative Collusion in Durable Goods Oligopoly,RAND Journal of Economics , 1987, 18 (2), 248254.

    Harrington, J. and A. Skrzypacz ,Collusion Under Monitoring of Sales, RAND Journal of Economics , 2007, 38 (2), 314331.and ,Private Monitoring and Communication in Cartels:

    Explaining Recent Collusive Practices,American Economic Review ,2011, 101, 24252449.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    Harrington, Joseph E. ,How Do Cartels Operate?, Foundations and Trends in Microeconomics , 2006, 2 (1), 1105.

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    71/72

    , , ( ),Levenstein, M. and V. Suslow ,What Determines Cartel Success?,

    Journal of Economic Literature , 2006, 44 , 4395.Liski, M. and JP. Montero , Forward Trading and Collusion inOligopoly, Journal of Economic Theory , 2006, 131, 212230.

    Milgrom, P. and I. Segal ,Envelope Theorems for Arbitrary ChoiceSets, Econometrica , 2002, 70 (2), 583601.

    Porter, Robert ,Optimal Cartel Trigger Price Strategies, Journal of Economic Theory , 1983, 29 , 313338.,A Study of Cartel Stability: The Joint Executive Committe,1880-1886, Bell Journal of Economics , 1983, 14 (2), 301314.

    Sobel, Joel ,Durable Goods Monopoly with Entry of New Consumers,Econometrica, 1991, 59 (5), 14551485.

    Stigler, George ,A Theory of Oligopoly,Journal of Political Economy ,1964, 72 , 4461.

    Stokey, N., R. Lucas, and E. Prescott , Recursive Methods inEconomic Dynamics , Harvard University Press, 1989.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    Introduction and Motivation

    General ModelConsumers Patience: A Simple ExampleReferences

    Stokey, Nancy ,Intertemporal Price Discrimination, Quarterly Journal of Economics , 1979, 93 , 355371.

    http://find/http://goback/
  • 8/2/2019 Collusion With Private Monitoring and Durable Goods

    72/72

    , , ,,Rational Expectations and Durable Goods Pricing, Bell Journal of

    Economics , 1981, 12 , 112128.Tirole, Jean , The Theory of Industrial Organization, The MIT Press,1988.

    Vives, Xavier , Oligopoly Pricing: Old Ideas and New Tools , The MITPress, 2001.

    Enrique Ide Carvallo Collusion with Private Monitoring and Durable Goods

    http://find/http://goback/