Mechanism Design in Hidden Action and Hidden Information: Richness and Pure Groves

Size: px
Start display at page:

Download "Mechanism Design in Hidden Action and Hidden Information: Richness and Pure Groves"

Transcription

1 CIRJE-F-1015 Mechansm Desgn n Hdden Acton and Hdden Informaton: Rchness and Pure Groves Htosh Matsushma Unversty of Tokyo Shunya Noda Stanford Unversty May 2016 CIRJE Dscusson Papers can be downloaded wthout charge from: Dscusson Papers are a seres of manuscrpts n ther draft form They are not ntended for crculaton or dstrbuton except as ndcated by the author For that reason Dscusson Papers may not be reproduced or dstrbuted wthout the wrtten consent of the author

2 1 Mechansm Desgn n Hdden Acton and Hdden Informaton: Rchness and Pure Groves 1 Htosh Matsushma 2 Unversty of Tokyo Shunya Noda 3 Stanford Unversty May 30, 2016 Abstract We nvestgate general collectve decson problems related to hdden acton and hdden nformaton We assume that each agent has a wde avalablty of acton choces at an early stage, whch provdes sgnfcant externalty effects on the other agent s valuatons n all drectons We characterze the class of all mechansms that solve the hdden acton problem, and demonstrate equvalence propertes n the ex-post term Importantly, we fnd that pure Groves mechansms, defned as the smplest form of canoncal Groves mechansms, are the only effcent mechansms that solve such hdden acton problems We argue that the resoluton of the hdden acton problem automatcally resolves the hdden nformaton problem JEL Classfcaton Numbers: C72, D44, D82, D86 Keywords: Hdden Acton, Hdden Informaton, Externaltes, Rchness, Pure Groves, Equvalence 1 Ths study s supported by a grant-n-ad for scentfc research (KAKENHI , 26J10076, 16H02009) from the Japan Socety for the Promoton of Scence (JSPS) and the Mnstry of Educaton, Culture, Sports, Scence and Technology (MEXT) of the Japanese government Ths paper s also supported by the Funa Overseas Scholarshp, the E K Potter Fellowshp, the JSPS research fellowshp (DC1), and the CARF research fund at Unversty of Tokyo We thank Gabrel Carroll, Fuhto Kojma, Takuo Sugaya, Yuch Yamamoto, and all the semnar partcpants n Htotsubash Unversty, Unversty of Tokyo, Osaka Unversty, Kyoto Unversty, and Stanford Unversty All errors are ours 2 Department of Economcs, Unversty of Tokyo, Hongo, Bunkyo-ku, Tokyo , Japan (htosh[at]eu-tokyoacjp) 3 Department of Economcs, Stanford Unversty, 579 Serra Mall, Stanford, CA , USA (shunyanoda[at]gmalcom)

3 2 1 Introducton Ths study nvestgates a general class of collectve decson problems that ncludes ssues of mperfect montorng (hdden acton) as well as ncomplete nformaton (hdden nformaton), such as prncpal-agent relatonshps, partnershps, and general resource allocatons (ncludng auctons and publc good provson), wth the assumptons of quas-lnearty and rsk-neutralty Multple agents ndependently make ther acton choces at an early stage before a state occurs, nfluencng ther valuaton functons through stochastc state determnaton The central planner then determnes an allocaton that s relevant to the welfare of all agents as well as the central planner n a state-contngent manner We assume hdden acton n that the central planner cannot observe the agents acton choces The central planner, therefore, desgns a state-contngent mechansm, accordng to whch the central planner makes sde payments to agents, as well as an allocaton decson, ncentvzng the agents to select the acton profle that the central planner desres We frst demonstrate a benchmark model that addresses only the hdden acton problem We then ncorporate hdden nformaton by assumng that the central planner can observe nether the state nor the agents acton choces, and therefore requres agents to announce ther prvate nformaton regardng the state (e, ther types) The purpose of ths study s to clarfy whether, and how, the central planner overcomes the ncentve problem n cases of both hdden acton and hdden nformaton Ths study substantally dffers from prevous research on mechansm desgn and contract theory n that each agent potentally has varous aspects of actvtes such as nformaton acquston, R&D nvestment, patent control, standardzaton, M&A, rent-seekng, postve/negatve campagns, envronmental concern, product dfferentaton, entry/ext decsons, preparaton of nfrastructure, and headhuntng The central planner generally lacks nformaton about the breadth of these potental aspects, because of, for example, the separaton of ownershp and control Accordngly, the

4 3 central planner cannot know whch aspects of agents actvtes are actually relevant to the current problem In such cases, the central planner must, nevtably, account for all of these aspects n mechansm desgn One example s a dynamc aspect of auctons for allocatng busness resources such as spectrum lcenses, where each partcpant undertakes varous proft-seekng actvtes to gan an advantage over rval companes n the upcomng aucton event 4 Whether, and to what degree, such actvtes dstort the welfare crucally depends on the aucton format desgn Specfcally, the central planner worres that each agent s hdden acton has sgnfcant externalty effects on the other agents valuaton functons (e, each agent can smoothly change the dstrbuton of the state n all drectons va a unlateral devaton from the desred acton profle) Ths broad potental for externaltes, whch ths study terms rchness, dramatcally restrcts the range of possble mechansms that can ncentvze agents to make the desred acton choces (e, n ths study s termnology, t restrcts the range of possble mechansms that can nduce the desred acton profle) The man contrbuton of ths study s to characterze mechansms that can nduce the desred acton profle, even n the presence of such rchness Based on ths characterzaton, we present equvalence propertes n the ex-post term under the constrants of nducblty That s, the ex-post payments, the ex-post revenue, and the ex-post payments are unque up to constants We further clarfy the possblty that the central planner acheves an effcent acton profle and effcent allocatons wthout any trouble n lablty Wth the assumpton of prvate values, we defne pure Groves mechansms as the smplest form of canoncal Groves mechansms, n whch the central planner gves each agent the welfare of the other agents and the central planner and mposes on her a fxed monetary fee We show that a mechansm nduces an effcent acton profle f and only f t s pure Groves In other words, pure Groves mechansms are the only effcent mechansms that solve the hdden acton problem 4 See Klemperer (2004) and Mlgrom (2004), for example

5 4 Ths theoretcal fndng has mportant mplcatons not only n cases of hdden acton but also n cases of hdden nformaton Suppose that the central planner cannot observe the state, and, therefore, requres agents to announce ther prvate nformaton regardng the state The above-mentoned dynamc aucton s an example of ths case Importantly, once the central planner desgns any mechansm that nduces the effcent acton profle (e, solves the hdden acton problem), then ths mechansm automatcally solves the hdden nformaton problem, because of ts nternalzaton feature The generally accepted vew n mechansm desgn s that n some envronments, Groves mechansms 5 are the only mechansms that solve the hdden nformaton problem In contrast to ths vew, ths study shows that pure Groves mechansms (e, the specal form of Groves mechansms), are the only mechansms that solve the hdden acton problem Moreover, the resoluton of the hdden acton problem n ths manner generally and automatcally can solve the hdden nformaton problem Because the class of pure Groves mechansms s a proper subclass of Groves mechansms, t mght be more dffcult for the central planner to earn non-negatve revenues wth hdden acton than wthout hdden acton In fact, the popular Vckrey Clarke Groves (VCG) mechansm, whch algns each agent s payoff wth her margnal contrbuton, generally guarantees non-negatve revenues, but t s not pure Groves (e, t fals to satsfy nducblty) To be more precse, we show an mpossblty result that wth rchness, there exsts no pure Groves mechansm (e, no well-behaved mechansm) that satsfes the requrements of non-negatve revenues and ex-post ndvdual ratonalty, whle the VCG mechansm satsfes both Accordngly, our result suggests that the central planner, who wants to defeat rchness n order to overcome the hdden acton problem, should collect nformaton n advance about whch aspects of agents actvtes are actually relevant An example wthout rchness s a case of partnershps wth fnte acton spaces, where the central planner can successfully talor the sde-payment rule to the detal of specfcatons, makng nducblty compatble wth non-negatve revenues, or even 5 See Vckrey (1961), Clarke (1971), and Groves (1973)

6 5 budget-balancng (Legros and Matsushma (1991), for example) In contrast, wth rchness, the central planner cannot remove agents evasons when she replaces a pure Groves mechansm wth any complcated mechansm The dffculty n lablty depends on the presence of each agent s externalty effect on the other agents valuaton functons If any agent s acton choce has no externalty and the central planner recognzes ths n advance, then we would obtan the equvalence result for nterm payoffs, rather than ex-post payoffs Specfcally, to acheve effcency, a wder class of mechansms than the class of Groves mechansms (expectaton-groves mechansms) can solve the ncentve problems of both hdden acton and hdden nformaton, wthout defcts, or even wth budget-balancng The remander of ths paper s organzed as follows Secton 2 revews the lterature Secton 3 presents the benchmark model, where we account for only hdden actons Secton 4 ncorporates hdden nformaton nto the model Secton 5 focuses on the achevement of effcency Secton 6 examnes the central planner s revenues Secton 7 studes the case of no externalty Secton 8 presents an alternatve defnton of rchness from the vewpont of full dmensonalty Secton 9 concludes 2 Related Lterature Ths study makes mportant contrbutons to the lterature regardng the hdden acton problem as follows When an agent has a wde varety of acton choces, a complcated contract desgn mght motvate the agent to devate from desred behavor In ths case, a smply desgned contract could functon better than a complcated one For example, Holmström and Mlgrom (1987) studed prncpal-agent relatonshps n a dynamc context, where randomly determned outputs are accumulated through tme, and the agent flexbly adjusts the effort level dependng on output-hstores They showed that the optmal ncentve contract that maxmzes the prncpal s revenue must be lnear wth respect to the output accumulated at the endng tme Carroll (2014) nvestgated optmal contract desgn n a statc prncpal-agent relatonshp, where the prncpal experences

7 6 ambguty about the range of actvtes that the agent can undertake Carroll showed that the optmal contract must be lnear wth respect to the resultant output, provded that the prncpal follows the maxmn expected utlty hypothess Ths study examnes the hdden acton problem by ntroducng multple agents, general state spaces, general allocaton rules, and varous crtera such as effcency, nstead of revenue optmzaton We account delberately for the wde range of externalty effects of each agent s acton choce on other agents valuaton functons by assumng rchness However, we do not assume ambguty or behavoral modes such as maxmn utlty We then demonstrate the characterzaton result for well-behaved ncentve mechansms, mplyng that only a smple form of mechansm desgn (e, pure Groves) functons, and we further show equvalence propertes n the ex-post term Any well-behaved effcent mechansm must be pure Groves In ths regard, Athey and Segal (2013) showed that wth prvate values, but regardless of whether rchness s present, pure Groves mechansms nduce effcency n hdden acton We extend ther study to show that, wth rchness, pure Groves mechansms are the only mechansms that can acheve effcency n hdden acton The lterature regardng the hdden acton problem has shown that wthout rchness (e, when the scope of acton spaces s suffcently lmted), we can desgn effcent ncentve mechansms by talorng the payment rule to detaled specfcatons We can even make the ncentve constrants compatble wth ether full surplus extracton or budget-balancng See Matsushma (1989), Legros and Matsushma (1991), and Wllams and Radner (1995), for example See also Obara (2008), whose model s close to ths study s model, but wthout rchness Wth rchness, however, a mechansm s dependence on detal mght encourage each agent to devate, because such dependence nevtably results n a loophole that puts the agent n a more advantageous poston The poor functonng that results from complcaton n mechansm desgn makes t dffcult to cope wth ncentves n hdden acton as well as non-negatvty of revenues

8 7 Ths study makes mportant contrbutons to the lterature on the hdden nformaton problem Green and Laffont (1977, 1979) and Holmström (1979) showed that n hdden nformaton envronments wth dfferentable valuaton functons, smooth path-connectedness, and prvate values, Groves mechansms are the only effcent mechansms that satsfy ncentve compatblty n the domnant strategy Ths study reconsders Groves mechansms from the vewpont of hdden acton, and shows that only pure Groves mechansms resolve the hdden acton problem Whle the resoluton of the hdden acton problem automatcally resolves the hdden nformaton problem, the reverse s not true 6 Ths study also nvestgates the case n whch externalty effects are absent, showng that expectaton-groves mechansms, whch are more general than Groves and requre each agent to pay the same amount as Groves n expectaton, are the only mechansms that can acheve effcency Hatfeld, Kojma, and Komners (2015) s relevant to ths result, because they showed that when we confne our attenton to effcent mechansms that are detal-free (e, ndependent of detaled knowledge about the set of actons that the agents can take), the mechansms must be Groves We permt a mechansm to not be detal-free, and then show that expectaton-groves mechansms successfully acheve effcency Importantly, the class of expectaton-groves mechansms ncludes the AGV mechansm (see Arrow (1979) and D Aspremont and Gerard-Varet (1979)), whch satsfes budget-balancng Ths result contrasts wth that of Hatfeld et al because Groves mechansms generally fal to satsfy budget-balancng It s worth notng that our results are rrelevant to the fne detal of the state space and valuaton functons Consequently, ths study artculates the desrablty of pure Groves mechansms and expectaton-groves mechansms, even f the prerequstes of Green Laffont Holmström fals to hold (eg, even f type spaces are fnte) 6 Hausch and L (1993) and Persco (2000) are related They demonstrate that frst-prce and second-prce auctons provde dfferent ncentves n nformaton acquston that make the other agents valuaton more accurate We explan that ths dfference comes from the dfference n ex-post payoffs between these aucton formats

9 8 3 Hdden Acton Let us consder a settng wth one central planner and n agents ndexed by N {1,2,, n} As a benchmark model of ths study, we nvestgate an allocaton problem that conssts of the followng four stages Stage 1: The central planner commts to a mechansm defned as ( g, x ), where n g : A and x x ) : R Here, denotes the fnte set of all states and ( N A denotes the set of all allocatons We call g and x the allocaton rule and the payment rule, respectvely 7 Stage 2: Each agent N selects a hdden acton b B, where B denotes the set of all actons for agent The cost functon for the agent s acton choce s gven by c : B R We assume that there s a no-effort opton B B and b( b1,, bn ) B N b 0 B such that c b Let 0 ( ) 0 Stage 3: The state s randomly drawn accordng to a condtonal probablty functon f ( b) ( ), where b B s the acton profle selected at stage 2, ( ) denotes the set of all dstrbutons (e, lotteres) over states, and f ( b) denotes the probablty that the state occurs provded that the agents select the acton profle b 7 Ths study assumes that the central planner commts to the mechansm before agents take acton Wthout ths assumpton, the desred outcomes, such as effcency, may not be achevable Consder the central planner who commts to a mechansm after the agents acton choces In ths case, the central planner prefers the VCG mechansm because t yelds greater revenue than the mechansm that ths study dscusses The VCG mechansm, however, fals to nduce any effcent acton profle; by antcpatng that the central planner wll set a VCG mechansm, agents are wllng to select neffcent actons

10 9 Stage 4: The central planner determnes the allocaton g( ) A and the sde payment n vector pad to the central planner x( ) ( x( )) N R, where s the state that occurs at stage 3 The resultant payoff of each agent N s gven by (1) v ( g( ), ) x ( ) c ( b), where we assume that each agent's payoff functon s quas-lnear and rsk-neutral, and the cost of the agent s acton choce s addtvely separable Fgure 1 descrbes the tmelne of the benchmark model Ths secton ntensvely studes the ncentves n hdden acton at stage 2 Fgure 1: Tmelne wth Hdden Acton Defnton 1 (Inducblty): A mechansm ( gx, ) s sad to nduce an acton profle b B f b s a Nash equlbrum n the game mpled by the mechansm ( gx,, ) e, for every N, (2) Ev [ ( g( ), ) x( ) b] c( b) Ev [ ( g( ), ) x( ) b, b ] c( b ) for all b B, where E[ b] denotes the expectaton operator condtonal on b, e, for every functon : R, E[ ( ) b] ( ) f ( b)

11 10 Lemma 1: Consder an arbtrary combnaton of an acton profle and an allocaton rule ( bg, ) There exsts a payment rule x such that ( g, x ) nduces b f and only f there n exsts a functon w( w ) : R such that N (3) E[ w( ) b] c( b) E[ w( ) b, b ] c( b ) for all N and b B Proof: Consder an arbtrary ( b,( g, x )) We specfy w by w( ) v( g( ), ) x( ) for all N and It s clear from Defnton 1 that ( g, x ) nduces b f and only f (3) holds QED Let nt ( ) ( ) denote the set of all full-support dstrbutons over states We ntroduce a condton on an acton profle b B, namely rchness, as follows Defnton 2 (Rchness): An acton profle b B s sad to be rch f for every N and ( ), there exst 0 and a path on B, (, ):[, ] B, such that (,0), b f( (, ), b ) f( b) (4) lm ( ) f ( b), 0 and c ( (, )) s dfferentable n at 0 8 Rchness mples that each agent N has a varety of acton choces around b that can smoothly and locally change the dstrbuton over states n all drectons from f ( b) at a dfferentable cost 9 8 Ths study assumes that the state space s fnte We, however, can elmnate ths assumpton wthout substantal changes Instead of the fnte state space, we can assume that the state space s a subset of a mult-dmensonal Eucldan space and the dstrbuton over states s contnuous In fact, we can obtan the unqueness results such as Lemmas 2 and 3 almost everywhere 9 Secton 8 wll replace Defnton 2 wth an alternatve that concerns only fntely many drectons

12 11 Under the rchness, f ( g, x ) nduces b, e, b s a Nash equlbrum n the game mpled by ( g, x ), the followng frst order condton holds for every N and ( ): Ev [ ( g( ), ) x( ) (, ), b ] c( (, )) 0 Furthermore, the rchness assures that each agent can change the expected value of any non-constant payment rules by makng a unlateral devaton Consder an arbtrary non-constant functon : R Note that there exsts such that 0 ( ) E[ ( ) b] Let denote the degenerate dstrbuton where ( ) 1 Takng (, ):[, ] B as defned n Defnton 2, we have E[ ( ) (, ), b ] 0 0 f ( (, ), b ) f ( b) ( ) lm ( ){ ( ) f ( b)} ( ) E[ ( ) b] 0 Hence, each agent can make a unlateral devaton to take advantage of non-constant In the sngle-agent case, where we regard the set of all actons of agent 1 as the set of all full-support dstrbutons over states, e, B 1 nt ( ), the well-known drectonal dfferentablty for arbtrary drectons wll be a tractable suffcent condton for the rchness 1011 Proposton 1: In the sngle-agent case, an acton B1 nt ( ) s rch f for every ( ), the agent s cost functon c 1 :n t ( ) R has the drectonal dervatve along : 10 The exstence of drectonal dervatves of c 1 at for all drectons s mpled by the total dfferentablty of c 1 at 11 When s a contnuum, the drectonal dfferentablty s generalzed to the Gâteaux dfferentablty

13 12 1((1 ) ) ( ) 1( ) lm c c c 0 Proof: For every ( ), specfy (, ):[, ] 1 B by (, ) (1 ) 1 for all [, ], where 0 s selected suffcently small so that 1(, ) nt ( ) B1 holds for all [, ] (snce s a full-support dstrbuton, such 0 always exsts) Then, t follows from c ( (, ) ) c ( ) c ((1 ) ) c ( ) lm lm c1( ) 0 0 that the drectonal dfferentablty of c 1 along assures the dfferentablty of c ( (, )) n at 0 Hence, s rch 1 1 QED Note that the converse of Proposton 1 does not hold, because the drectonal dfferentablty does not account for the dfferentals along curves whose tangent at b satsfes (4) Note also that t follows from Proposton 1 that n ths sngle-agent case, f c 1 s drectonally dfferentable everywhere on B 1, then every acton b1 B1 s rch We further consder the followng class of mult-agent problems For every N, specfy where we denote B nt ( ) B, 2 b ( b, b ) nt ( ) B Each agent N makes a recommendaton about the state dstrbuton as the frst component of her acton, b nt ( ), and lobbes her recommendaton as the second component of her acton, 1 b B Denote b ( b) N and b 2 ( b 2 ) N Defne f b b b, 2 1 ( ) ( ) () N

14 13 where and 2 ( b ) (0,1) N 2 ( b ) 1 The state dstrbuton s determned as the compromse among the agents recommendatons, e, the average of ther recommendatons 1 b weghted by each agent s nfluence 2 ( ( b )) N, whch s the consequent of the agents lobbyng actvtes We assume that for every N and b B, 2 c (, b ) s drectonally dfferentable In the same manner as Proposton 1, we can prove that every acton profle b B s rch n ths case Regardless of the acton profle b B, each agent N can smoothly and locally change the dstrbuton n all drectons from f( b) by manpulatng her recommendaton (the frst component) b 1 nt ( ) Note that the class of problems dscussed here ncludes substantally general mult-agent problems In fact, t does not requre any restrcton on the lobbyng acton spaces 2 ( ) N B, the N nfluence functons ( ) c N, and the cost functons ( ), besdes the drectonal dfferentablty of 2 ( c(, b )) N The followng lemma mples that wth rchness, the functon w that satsfes (3) s unque up to constants Lemma 2: Suppose that an acton profle b s rch, and a functon w satsfes (3) For n every functon w ( w ) : R, w satsfes the propertes mpled by (3), e, for N every N, Ew [ ( ) b] c( b) Ew [ ( ) b, b ] c( b ) for all b B, n f and only f there exsts a vector z ( z ) R such that N w ( ) w( ) z for all N and Proof: The proof of the suffcency s straghtforward We present the proof of the necessty as follows Take an arbtrary w whch satsfy (3) and consder an arbtrary agent N Take an arbtrary w such that w w s a non-constant functon

15 14 Then, there exsts such that ( ) E[ ( ) b] Let denote the degenerate dstrbuton where ( ) 1 Due to rchness, there exst 0 and (, ):[, ] B such that f( (, ), b ) f( b) lm ( ) f ( b) 0 Snce w satsfes (3), the frst order condton along (,) must hold, e, (5) Ew b c On the other hand, [ ( ) (, ), ] ( (, )) 0 Ew [ ( ) (, ), b ] c( (, )) 0 0 E[ ( ) w( ) (, ), b ] c( (, )) 0 0 E[ ( ) (, ), b ] ( ) E[ ( ) b] 0 Hence, f w w s non-constant, agent has ncentve to ncrease along (,) from 0 Accordngly, whenever w and w satsfy (3), w w s constant, e, there exsts z n R such that ( ) ( ) w w z for all QED Based on Lemmas 1 and 2, we show the followng theorem, whch states that the payment rule that guarantees nducblty s unque up to constants Theorem 1: Consder an arbtrary combnaton of an acton profle and a mechansm ( b,( g, x )) Suppose that b s rch and ( g, x ) nduces b Accordngly, for every payment rule x, the assocated mechansm ( g, x ) nduces b f and only f there exsts a vector z n R such that

16 15 x( ) x ( ) z for all Proof: The proof of the suffcency s straghtforward We present the proof of the necessty as follows Suppose that ( g, x ) nduces b Accordng to the proof of Lemma 1, we specfy w by (6) w( ) v( g( ), ) x( ) for all N and Suppose also that ( g, x ) nduces b Smlarly, we specfy w by (7) w ( ) v( g( ), ) x ( ) for all N and Lemma 2 mples that there exsts By subtractng (6) from (7), we have x( ) x ( ) z for all z n R such that ( ) ( ) w w z for all QED Theorem 1 mples the followng equvalence propertes n the ex-post term Consder an arbtrary combnaton of an acton profle and an allocaton rule (, bg ) Consder two arbtrary payment rules x and x such that both ( gx, ) and ( gx, ) nduce b Let U each agent N: and R and U R denote the respectve ex-ante expected payoff for U E[ v ( g( ), ) x ( ) b] c ( b), U E[ v ( g( ), ) x ( ) b] c ( b ) Accordngly, Theorem 1 exhbts that the ex-post payment for each agent s unque up to constants n that x ( ) x ( ) U U for all, the ex-post revenue for the central planner s unque up to constants n that x ( ) x ( ) ( U U ) for all, N N N

17 16 and the ex-post payoff for each agent s unque up to constants n that v ( g( ), ) x ( ) c ( b ) { v ( g( ), ) x ( ) c ( b )} U U for all 4 Hdden Informaton Let us specfy an nformaton structure where the state s decomposed as 0 1 n (,,, ) Here, we call 0 a publc sgnal and a type for each agent N Let 0 denote the set of all publc sgnals and N{0} denote the set of all types for each agent N Let We assume that the publc sgnal 0 0 becomes observable to all agents as well as the central planner just before the central planner determnes an allocaton and sde payments, and t s, therefore, contractble The central planner, however, cannot observe the profle of all agents types, whch s denoted by 0 ( ) N 0 Each agent N can observe her type, but cannot observe the profle of the other agents types, denoted by ( ) {0}\{ } j j N j jn{0}\{ } Because of the above-mentoned hdden nformaton structure, we replace stages 3 and 4 wth the followng stages (e, stages 3 and 4, respectvely) Importantly, the central planner requres each agent to reveal her type, whch the central planner cannot drectly observe 12 N 12 Because the central planner nduces a pure acton profle, we can safely focus on revelaton mechansms where each agent only reports her type If the central planner attempts to nduce a mxed acton profle, we need to consder mechansms where each agent reports not only her own type but also her selecton of pure acton See Obara (2008) For further dscussons, see Footnote 17

18 17 Stage 3 : The state ( 0, 1,, n ) s randomly drawn accordng to the condtonal probablty functon f( b) ( ), where b B s the acton profle selected at stage 2 Each agent N observes hs type, but cannot observe at ths stage Stage 4 : Each agent N announces about her type Afterward, all agents, as well as the central planner, observe the publc sgnal 0 0 Accordng to the profle of the agents announcements 0 ( ) N 0 and the observed publc sgnal 0 0, the central planner determnes the allocaton g 0 0 payment vector x( 0, 0) R n (, ) A and the sde The resultant payoff of each agent s gven by v( g(, ), ) x (, ) c( b) Fgure 2: Tmelne wth Hdden Acton and Hdden Informaton Fgure 2 descrbes the tmelne of the model wth hdden acton and hdden nformaton We consder the agents ncentves n hdden nformaton by ntroducng the concept of ex-post ncentve compatblty

19 18 Defnton 3 (Ex-Post Incentve Compatblty): A mechansm ( gx, ) s sad to be ex-post ncentve compatble (hereafter EPIC) f truth-tellng s an ex-post equlbrum; for every N and, v( g( ), ) x( ) v( g(, ), ) x(, ) for all 13 EPIC s ndependent of the acton profle, because of the addtve separablty of the cost of actons We further ntroduce a weaker noton, namely Bayesan Implementablty, as follows Defnton 4 (Bayesan Implementablty): A combnaton of an acton profle and a mechansm (,( b g, x )) s sad to be Bayesan mplementable (hereafter BI) f the selecton of the acton profle b at stage 2 and the truthful revelaton at stage 4 results n a perfect Bayesan equlbrum; for every N, every b B, and every functon :, E[ v ( g( ), ) x ( ) b] c( b) E[ v( g( ( ), ), ) x( ( ), ) b, b ] c( b) BI ncludes the nducblty of b n order to account for the possblty of contngent devatons; BI requres (,( b g, x )) to exclude the possblty that each agent benefts by devatng from both the acton choce b at stage 2 and the truthful revelaton at stage 4 The followng theorem states that f there s a mechansm that nduces an acton profle b but fals to be ncentve compatble, then t s generally mpossble to dscover mechansms that satsfy both nducblty and ncentve compatblty 13 Because of addtve separablty, the ncentve compatblty s rrelevant to the shape of the cost functons

20 19 Theorem 2: Consder an arbtrary combnaton of an allocaton rule and an acton profle (, bg ), where we assume that b s rch 1 Suppose that there exsts a payment rule x such that ( gx, ) nduces b and satsfes EPIC For every payment rule x, whenever ( gx, ) nduces b, t satsfes EPIC 2 Suppose that there exsts a payment rule x such that (,( b g, x )) satsfes BI For every payment rule x, whenever ( gx, ) nduces b,( b,(, g x )) satsfes BI Proof: Suppose that both ( gx, ) and ( gx, ) nduce b From Theorem 1, there exsts z n R such that x( ) x ( ) z for all, whch mples that ( gx, ) satsfes EPIC f and only f ( gx, ) satsfes EPIC We can smlarly prove that ( b,( g, x )) satsfes BI f and only f ( b,( g, x )) satsfes BI QED () () Theorem 2 mples that the followng two statements are equvalent There exsts a mechansm that satsfes nducblty but fals to satsfy ncentve compatblty There exsts no mechansm that satsfes both nducblty and ncentve compatblty 5 Effcency We denote by v : A R 0 the valuaton functon for the central planner Ths secton and the next ntensvely study allocaton rules and acton profles that are effcent (e, maxmze the welfare that ncludes the central planner s welfare as well as the agents welfare); an allocaton rule g s sad to be effcent f

21 20 v ( g( ), ) v ( a, ) N{0} N{0} for all a A and A combnaton of an acton profle and an allocaton rule (, bg ) s sad to be effcent f g s effcent and the selecton of b maxmzes the expected welfare: such that E[ v ( g( ), ) b] c ( b) N{0} N{0} E[ v ( g( ), ) b] c ( b ) for all b B N{0} N{0} A payment rule x s sad to be Groves f there exsts y : R for each N x ( ) v ( g( ), ) y ( ) j jn{0}\{ } for all N and Accordng to a Groves payment rule, the central planner gves each agent N the monetary amount that s equvalent to the other agents welfare plus the central planner s welfare (e, jn{0}\{ } y( ) that s ndependent of v ( g( ), ) ), and mposes on the agent the monetary payment j A payment rule x s sad to be pure Groves f t s Groves and y () s constant n for each N; there exsts a vector z ( z ) R such that N x ( ) v ( g( ), ) z j jn{0}\{ } for all N and Pure Grove payment rules are specal cases of Grove payment rules where the central planner mposes a fxed amount z on each agent as a non-ncentve term The followng theorem states that an effcent mechansm nduces an effcent acton profle f and only f the payment rule s pure Groves Accordngly, under effcency and nducblty, wthout loss of generalty, we can focus on pure Groves mechansms (e, combnatons of an effcent allocaton rule and a pure Groves payment rule) Ths study makes a slght extenson of the canoncal defnton of a Groves mechansm, by accountng for the valuatons of the central planner

22 21 Theorem 3: Suppose that ( b, g ) s effcent For every payment rule x, ( g, x ) nduces b f x s pure Groves Suppose that b s rch and ( bg, ) s effcent For every payment rule x, ( g, x ) nduces b f and only f x s pure Groves Proof: If (, bg ) s effcent, for every N and b B, E[ v ( g( ), ) x ( ) b] E[ v ( g( ), ) x ( ) b, b ] E[ v ( g( ), ) b] E[ v ( g( ), ) b, b ] j j jn{0} jn{0} 0, whch mples the nducblty of b If b s rch, t s clear from Theorem 2 and the defnton of the pure Groves payment rule that any payment rule that guarantees nducblty must be pure Groves QED Note that we can construct any pure Groves mechansm wthout detaled knowledge of ( f, Bc, ) Even f the central planner s allowed to utlze such knowledge, the central planner s best choce would be pure Groves The followng theorem demonstrates a necessary and suffcent condton for the satsfacton of both nducblty and ncentve compatblty on the assumpton of effcency and rchness Theorem 4: Suppose that b s rch and ( bg, ) s effcent There exsts a payment rule x such that ( gx, ) nduces b and satsfes EPIC f and only f for every N,, and, (8) v ( g( ), ) v ( g(, ), ) v ( g(, ),, ) j j jn {0} jn{0}\{ } There exsts a payment rule x such that (,( b g, x )) satsfes BI f and only f for every N, b B, and :,

23 22 (9) E[ v ( g( ), ) b] c ( b) Ev [ ( g ( ( ), ), ) jn{0} jn{0}\{ } j v ( g( ( ), ), ( ), ) b, b ] c ( b) j Proof: Consder the proof for the case of EPIC From Theorem 3, ths proof only requres us to clarfy a necessary and suffcent condton for a pure Grove payment rule to guarantee EPIC Let x denote the pure Groves payment rule gven by x ( ) v ( g( ), ) jn{0}\{ } j for all N and Accordngly, ( gx, ) satsfes EPIC f and only f for every N,, and, v ( g( ), ) x ( ) v ( g( ), ) j j jn{0} v ( g(, ), ) v ( g(, ),, ) j jn{0}\{ } v ( g(, ), ) x (, ), whch s equvalent to (8) We can prove the case of BI n a smlar manner QED Note that (8) s a necessary and suffcent condton for a Groves mechansm to satsfy EPIC However, f each agent s payoff functon has nterdependent values, any Groves mechansm does not satsfy EPIC, and, therefore, (8) generally fals 15 Based on ths outcome, we shall focus on the case of prvate values, where for every N, the valuaton v ( a, ) s ndependent of the profle of the other agents types, and v ( a, ) s ndependent of 0 0 Wth prvate 0 ( j ) jn \{ } 0 jn \{ } j values, any Groves mechansm satsfes (8) (e, EPIC) Wth prvate values, we wrte 15 Maskn (1992), Dasgupta and Maskn (2000), and Bergemann and Välmäk (2002) proposed the generalzed VCG mechansm and showed that t satsfes EPIC for some envronments, even wth nterdependent values The generalzed VCG mechansm, however, fals to nduce an effcent acton profle, because t s not pure Groves Mezzett (2004), however, showed that f the realzed valuaton v ( g( ), ) s observable as an ex-post publc sgnal and s contractble, we can acheve effcency wth EPIC even wth nterdependent values See Noda (2016) for more general ex-post sgnals

24 23 v ( a, 0, ) nstead of v( a, ) for each N, and v 0 ( a, 0 ) nstead of v 0 ( a, ) 16 Wth prvate values, EPIC s equvalent to ncentve compatblty n a domnant strategy (thereafter DIC), where for every N,, and, v ( g( ),, ) x ( ) v ( g(, ),, ) x (, ) 0 0 Theorem 5: Suppose that (, bg ) s effcent Wth prvate values, for every payment rule x, ( gx, ) nduces b and satsfes DIC f x s pure Groves Suppose that b s rch and (, bg ) s effcent Wth prvate values, for every payment rule x, ( gx, ) nduces b and satsfes DIC f and only f x s pure Groves Proof: From effcency, on the assumpton of prvate values, nequaltes (8) always hold Because, wth rchness, nequaltes (8) are necessary and suffcent for a Groves mechansm to satsfy DIC, Theorems 3 and 4 mply the latter part of Theorem 5 QED Wth prvate values, any DIC mechansm that satsfes nducblty and effcency must be pure Groves, although any Groves mechansm generally satsfes DIC and effcency For example, consder the VCG mechansm ( gx,, ) whch s defned as an effcent (Groves) mechansm specfed by x ( ) v ( g( ),, ) mn v ( g(, ),, ) 0 j 0 j jn{0} \{ } jn{0} for all N and Clearly, the VCG mechansm s not pure Groves; thus, t fals to satsfy the nducblty of effcent acton profle We permt the valuatons to depend on the publc sgnal 0 For convenence, we sometmes wrte v0( a, 0, 0) nstead of v 0 ( a, 0 ) 17 Obara (2008) showed that, when the acton space s fnte, full surplus extracton s achevable n approxmaton, under a weaker condton than exact full surplus extracton By carefully mxng

25 24 6 Revenues and Defcts By assumng prvate values, ths secton determnes whether the central planner can acheve effcency even wthout defcts We defne the central planner s ex-post revenue by (10) v ( ( ), ) ( ) 0 g 0 x, 18 N and the expected revenue n the ex-ante term by E[ v ( g( ), ) x ( ) b] 0 0 N We ntroduce three concepts of ndvdual ratonalty as follows The tmngs of ext opportuntes are depcted n Fgure 2 Defnton 5 (Ex-Ante Indvdual Ratonalty): A combnaton of an acton profle and a mechansm (,( b g, x )) s sad to satsfy ex-ante ndvdual ratonalty (hereafter EAIR) f Ev [ ( g( ),, ) x( ) b] c( b) 0 for all N 0 Defnton 6 (Interm Indvdual Ratonalty): A combnaton of an acton profle and a mechansm (,( b g, x )) s sad to satsfy nterm ndvdual ratonalty (hereafter IIR) f E [ v ( g(, ),, ) x (, ) b, ] 0 0 for all N and, where E [ (, ) b, ] (, ) f ( b, ) agents actons, Obara generated a correlaton between ( b, ) and ( b, ), and then appled the technque of Cremer and McLean (1985, 1988) Obara s argument, however, requred extremely large sde payments and very detaled knowledge about model specfcatons 18 Note that the central planner s revenue ncludes not only payments from agents, but also the valuaton of the central planner In the sngle agent case, where the allocaton space s degenerate, the value of (10) corresponds to the prncpal s payoff n a standard prncpal-agent model wth rsk-neutralty

26 25 denotes the expectaton of a functon (, ): condtonal on (, b ) R Defnton 7 (Ex-Post Indvdual Ratonalty): A mechansm ( gx, ) s sad to satsfy ex-post ndvdual ratonalty (hereafter EPIR) f v( g( ),, ) x( ) 0 for all N and 0 EPIR automatcally mples IIR IIR, however, does not necessarly mply EAIR, because the cost for the acton choce at stage 2 s a sunk cost The followng proposton shows that EPIR mples EAIR Proposton 2: Suppose that ( gx, ) nduces b Whenever ( gx, ) satsfes EPIR, (,( b g, x )) satsfes EAIR Proof: Because c b, t follows from EPIR and nducblty that 0 ( ) 0 E[ v ( g( ),, ) x( ) b] c( b) E[ v( g( ), 0, ) x( ) b, b ] c( b ) 0, whch mples EAIR QED EPIR s the strongest requrement for voluntary partcpaton among the above three concepts The purpose of ths secton s to show that EPIR s not compatble wth the non-negatvty of revenue n expectaton The followng proposton calculates the maxmal expected revenues (e, the least upper bounds of the central planner s expected revenues)

27 26 Proposton 3: Suppose that ( bg, ) s effcent, b s rch, and the prvate value assumpton s satsfed Then, the maxmal expected revenue from the mechansm ( gx, ) that nduces b and satsfes EPIR s gven by EPIR (11) R nmn vj( g( ), 0, j ) j ( n1) E vj( g( ), 0, j) b N{0} jn{0} The maxmal expected revenue from the mechansm ( gx, ) that nduces b and satsfes IIR s gven by IIR (12) R mn E v ( (, ), 0, ), j g j b N jn{0} ( n1) E vj( g( ), 0, j) b jn{0} The maxmal expected revenue from the mechansm ( gx, ) that nduces b and satsfes EAIR s gven by EAIR (13) R E vj( g( ), 0, j) bcj( bj ) jn{0} jn The maxmal expected revenue from the mechansm ( gx, ) that nduces b and satsfes IIR and EAIR s gven by (14) EAIR, IIR R Ev ( g( ), ) b mn Ev ( g( ),, ) bc ( b), N mn E v ( (, ), 0, ), j g j b E vj( g( ), 0, j) b jn{0} jn{0}\{ } Proof: See the Appendx EAIR Clearly, R s equal to the maxmzed expected socal welfare From the relatve strength of ncentve compatblty constrants, t s also clear that EPIR IIR, EAIR IIR R R R and R IIR, EAIR EAIR R

28 27 IIR However, whch s greater between R and R EAIR depends on specfcatons The followng proposton ndcates that wth the constrants of EPIR, t s generally dffcult for the central planner to acheve effcency wthout defcts Proposton 4: Suppose that ( bg, ) s effcent, b s rch, and the prvate value assumpton s satsfed Suppose also that c ( b ) 0 N and there exsts a null state 0 n (,, ) n the sense that and for every N, v0( a, 0) 0 for all a A, v ( a,, ) 0 for all a A and EPIR Wth EPIR, the central planner has a defct n expectaton: R 0 Proof: See the Appendx EAIR If R 0, the concluson s mmedate from R R EPIR EAIR EAIR R Suppose 0 It follows from c ( b ) 0 N that the second term of to the presence of the null state, the frst term of EPIR R n (11) s negatve Due R EPIR n (11) s non-postve EPIR Accordngly, R s negatve Wth prvate values, the VCG mechansm generally satsfes EPIR (and DIC) and guarantees that the central planner wll earn a non-negatve ex-post revenue at all tmes In contrast, once we requre nducblty, the VCG mechansm does not functon, and the central planner fals to earn non-negatve revenue, even wth an ex-ante expectaton By replacng EPIR wth weaker constrants, such as IIR and EAIR, and addng some restrctons, t becomes much easer for the central planner to acheve effcency wthout defcts The followng proposton states that the maxmal expected revenue of the central planner s rrelevant to whether nducblty s requred

29 28 Proposton 5: Suppose that ( bg, ) s effcent, b s rch, and the prvate value assumpton s satsfed Suppose also that we have: Condtonal Independence: For every b B and, f ( b) f( b ), N{0} where f ( b ) denotes the probablty of occurrng when the agents select the acton profle b The expected revenue acheved by any Groves mechansm that satsfes IIR and EAIR s less than or equal to R IIR, EAIR Furthermore, there exsts a pure Groves mechansm that satsfes IIR and EAIR and acheves IIR, EAIR R Proof: See the Appendx We mght expect that the central planner can receve larger expected revenue than IIR, EAIR R once we remove the requrement of nducblty However, wth condtonal ndependence, no Grove mechansm s able to make the expected revenue greater than IIR, EAIR IIR, EAIR R Ths mples that R s the upper bound of expected revenue regardless of whether we requre nducblty Proposton 5 s related to the observaton from rsk-sharng n a classcal prncpal-agent model When the prncpal s rsk-averse but the agent s rsk-neutral, the prncpal s best choce s to sell the company to the agent by gvng the entre outcome (externaltes to the prncpal) n exchange for a fxed constant fee 19 By regardng pure Groves mechansms as an extenson of such sellng-out contracts, Proposton 5 ndcates that wth rchness, sellng the company s the best choce even f the prncpal s rsk-neutral It s well-accepted that effcency s achevable by a Groves mechansm wthout runnng expected defcts f we do not requre nducblty Wth condtonal 19 See Harrs and Ravv (1979), for example

30 29 ndependence and some moderate restrctons, we can guarantee the non-negatvty of IIR, EAIR R as follows Proposton 6: Assume the suppostons n Proposton 4, condtonal ndependence, and the followng condtons Non-Negatve Valuaton: For every N {0} and, v ( g( ),, ) 0 0 Non-Negatve Expected Payoff: For every N, (15) Ev [ ( g( ), 0, j) b] c( b) 0 Wth IIR and EAIR, the central planner has non-negatve expected revenue: IIR, EAIR R 0 Proof: See the Appendx Non-negatve valuaton excludes the case of blateral barganng addressed by Myerson and Satterthwate (1983), where t s mpossble for the central planner to acheve effcency wthout defcts Non-negatve expected payoff excludes the case of opportunsm n the hold-up problem, where the sunk cost c ( b ) s so sgnfcant that t volates nequalty (15) By elmnatng these cases, and replacng EPIR wth IIR and EAIR, we can derve the possblty result n lablty mpled by Proposton 6 7 No Externalty So far, we have assumed rchness, e, that each agent s acton choce provdes a sgnfcant externalty effect on the other agents types and the publc sgnal Ths secton assumes that no such externalty exsts Specfcally, ths secton assumes ndependence of the nformaton structure, requrng condtonal ndependence and requrng that each agent s acton choce

31 30 b B only nfluences the margnal dstrbuton of the agent s type and b B, f( ( ) ( ) b) f0 0 f b N ; for every Here, f( b) denotes the dstrbuton of each agent s type that s assumed to depend only on b, and f () denotes the dstrbuton of the publc sgnal 0 0 that s assumed to be ndependent of the acton profle b B When we have such ndependence, agent s acton space s equvalent to the set of avalable margnal dstrbutons on agent s type Accordngly, ths restrcton expresses no externalty Wth ndependence, for every : R and :, we can smply wrte E R [ (, ) b ] and E[ ( ) b ] nstead of E [ (, ) b, ] and E[ ( ) b], respectvely On the acton profle b, we ntroduce a condton of prvate rchness, the no-externalty verson of rchness, as follows Defnton 8 (Prvate Rchness): An acton profle b B s sad to be prvately rch f we have the ndependence of the nformaton structure, and for every N and ( ), there exst 0 and a path on B, (, ):[, ] B, such that (16) (,0) b, 0 f( (, )) f( b) lm ( ) f( b), and c( (, )) s dfferentable n at 0 Prvate rchness mples that each agent N has a wde varety of acton choces that can smoothly and locally change the dstrbuton of the agent s type (not of the entre state ) n all drectons from f ( b ) Snce the acton choce problem of each agent s separated by the ndependence of the nformaton structure, Proposton 1

32 31 guarantees that an acton profle a A satsfes prvate rchness f the cost functon for each agent N, c, has drectonal dervatves at b Prvate rchness s a much weaker restrcton than the orgnal verson of rchness Accordngly, we can expect a much wder class of payment rules than pure Groves to satsfy nducblty The man purpose of ths secton s to characterze payment rules that guarantee nducblty under prvate rchness Lemma 3: Suppose that an acton profle b s prvately rch, and a functon w n satsfes (3) For every functon w ( w ) : R, w satsfes the propertes mpled by (3), e, for every N, E[ w( ) b] c( b) E[ w ( ) b, b ] c( b ) for all b B, f and only f E [ w (, ) w (, ) b ] s ndependent of N Proof: The proof of ths lemma parallels that of Lemma 2 See the Appendx Proposton 7: Consder an arbtrary combnaton of an acton profle and a mechansm ( b,( g, x )) Suppose that b s prvately rch and ( g, x ) nduces b For every payment rule x, the assocated mechansm ( g, x ) nduces b f and only f E [ x (, ) x (, ) b ] s ndependent of Proof: The proof of ths proposton parallels that of Theorem 1 See the Appendx Fx an arbtrary combnaton of an acton profle and an allocaton rule (, bg ) Consder two arbtrary payment rules x and x such that both ( gx, ) and ( gx, ) nduce b Let U each agent N: R and U R denote the respectve ex-ante expected payoff for U E[ v ( g( ), ) x ( ) b] c ( b),

33 32 and U E[ v ( g( ), ) x ( ) b] c ( b ) Proposton 7 mples that the nterm (rather than ex-post) payment for each agent s unque up to constants, n that E [ x (, ) b ] E [ x (, ) b ] U U for all Proposton 8: Consder an arbtrary combnaton of an allocaton rule and an acton profle (, bg ), where b s prvately rch If there exsts a payment rule x such that (,( b g, x )) satsfes BI, then, for every payment rule x, whenever ( gx, ) nduces b, (,(, b g x )) satsfes BI Proof: The proof of ths proposton parallels that of Theorem 2 See the Appendx Let us consder an arbtrary combnaton of acton profle and allocaton rule (, bg ) that are effcent A payment rule x s sad to be expectaton-groves f for each N, there exst r : R such that for every N, x ( ) v ( g( ), ) r( ) j jn{0}\{ } for all, and E [ r(, ) ] b s ndependent of Note that, wth effcency, any expectaton-groves payment rule guarantees nducblty Note that any Groves payment rule s expectaton-groves Whenever x s expectaton-groves, then any payment rule x s expectaton-groves f and only f E [ x (, ) x (, ) b ] s ndependent of Proposton 9: Suppose that (, bg ) s effcent Wth ndependence, ( gx, ) nduces b f x s expectaton-groves Suppose that b s prvately rch and (, bg ) s effcent Wth ndependence, ( gx, ) nduces b f and only f x s expectaton-groves

34 33 Proof: See the Appendx The replacement of rchness wth prvate rchness dramatcally reduces the dffculty of achevng effcency wthout defcts In fact, the VCG payment rule s not pure Groves, but s expectaton-groves and satsfes EPIC Wth prvate values, t generally guarantees the non-negatvty of the ex-post payment from each agent to the central planner Wth prvate values, the mechansm ( gx, ) explored by Arrow (1979) and D Aspremont and Gerard-Varet (1979) (e, the AGV mechansm) s expectaton-groves, where r ( ) ( (17) x ( ) v ( g ( ), ) ) s specfed as jn {0}\{ } j r( ) v ( g( ), ) E [ v ( g(, ), ) b ] j j j j jn{0}\{ } jn{0}\{ } 1 n 1 E [ v ( g(, ), ) b ] j h j j h j jn\{ } hn{0}\{ j} Ths mechansm clearly satsfes budget-balancng: x ( ) 0 for all N Accordngly, wth ndependence, the central planner can acheve effcency under the constrants of BI and budget-balancng Hatfeld, Kojma, and Komners (2015) studed a relevant problem by focusng on mechansms that are detal-free (e, ndependent of specfcatons such as acton spaces and cost functons) They showed that effcent and nducble mechansms must be Groves In contrast, we permt mechansms to not be detal-free, and then show that expectaton-groves mechansms, whch ncludes Groves as a proper subclass, are necessary and suffcent 20 8 Alternatve Defnton of Rchness: Full Dmensonalty 20 Note that the constructon of an expectaton-groves mechansm does not depend much on fne detals of the specfcaton of ( f, Bc, ) In fact, the only knowledge needed for ths constructon s the shape of f( b) at the effcent acton profle b

35 34 So far, we have defned rchness as a condton mplyng that each agent can change the dstrbuton n all drectons by selectng pure actons Ths secton ntroduces an alternatve defnton of rchness, n whch each agent can change the dstrbuton n fntely many drectons by selectng pure actons Denote K dm( ( )) 1 Takng advantage of the fnteness of the state space, we assume the followng condton as an alternatve to rchness: For each N, there exst 0, { k }, and k K 1 { :[, ] B } 1 such that the K vectors K k 1 K k k k { ( ) f ( b) } are lnearly ndependent, and for every k {1,, K}, k (0), b k f( ( ), b ) f( b) k lm ( ) f ( b), 0 k and c ( ( )) s dfferentable n at 0 In contrast to the orgnal defnton of rchness, we only assume that each agent can change the dstrbuton n fntely many k K drectons, e, { ( ) f ( b) } k1, by selectng pure actons However, accordng to the k K lnear ndependence of { ( ) f ( b) } 1, t follows that for every ( ), there k exsts a mxed acton for agent N, by selectng whch ths agent can change the dstrbuton n the drecton of ( ) f( b) at a dfferentable cost Hence, wth the alternatve rchness, each agent can change the dstrbuton n all drectons by selectng not pure but mxed actons Ths fndng has mportant mplcatons Even f we replace the orgnal defnton of rchness wth the above-mentoned alternatve, we can prove all results of ths study due to rchness For the satsfacton of the alternatve rchness, we do not requre each agent to have a huge breadth of acton space All we have to do for the satsfacton of the alternatve rchness s to clarfy whether each agent can change the dstrbuton by selectng pure actons n fntely many drectons In the same manner as the above argument, we can also replace the defnton of prvate rchness wth an alternatve concernng only fntely many drectons

Tit-For-Tat Equilibria in Discounted Repeated Games with. Private Monitoring

Tit-For-Tat Equilibria in Discounted Repeated Games with. Private Monitoring 1 Tt-For-Tat Equlbra n Dscounted Repeated Games wth Prvate Montorng Htosh Matsushma 1 Department of Economcs, Unversty of Tokyo 2 Aprl 24, 2007 Abstract We nvestgate nfntely repeated games wth mperfect

More information

(1 ) (1 ) 0 (1 ) (1 ) 0

(1 ) (1 ) 0 (1 ) (1 ) 0 Appendx A Appendx A contans proofs for resubmsson "Contractng Informaton Securty n the Presence of Double oral Hazard" Proof of Lemma 1: Assume that, to the contrary, BS efforts are achevable under a blateral

More information

Abstract Single Crossing and the Value Dimension

Abstract Single Crossing and the Value Dimension Abstract Sngle Crossng and the Value Dmenson Davd Rahman September 24, 2007 Abstract When auctonng an ndvsble good wthout consumpton externaltes, abstract sngle crossng s necessary and suffcent to mplement

More information

Role of Honesty in Full Implementation +

Role of Honesty in Full Implementation + 1 Role of Honesty n Full Implementaton + Htosh Matsushma * Faculty of Economcs Unversty of Tokyo June 4 2007 (Frst Verson: March 4 2002) + Ths paper s a revsed verson of the manuscrpt enttled Non-Consequental

More information

Implementation and Detection

Implementation and Detection 1 December 18 2014 Implementaton and Detecton Htosh Matsushma Department of Economcs Unversty of Tokyo 2 Ths paper consders mplementaton of scf: Mechansm Desgn wth Unqueness CP attempts to mplement scf

More information

Vickrey Auction VCG Combinatorial Auctions. Mechanism Design. Algorithms and Data Structures. Winter 2016

Vickrey Auction VCG Combinatorial Auctions. Mechanism Design. Algorithms and Data Structures. Winter 2016 Mechansm Desgn Algorthms and Data Structures Wnter 2016 1 / 39 Vckrey Aucton Vckrey-Clarke-Groves Mechansms Sngle-Mnded Combnatoral Auctons 2 / 39 Mechansm Desgn (wth Money) Set A of outcomes to choose

More information

CS286r Assign One. Answer Key

CS286r Assign One. Answer Key CS286r Assgn One Answer Key 1 Game theory 1.1 1.1.1 Let off-equlbrum strateges also be that people contnue to play n Nash equlbrum. Devatng from any Nash equlbrum s a weakly domnated strategy. That s,

More information

Perfect Competition and the Nash Bargaining Solution

Perfect Competition and the Nash Bargaining Solution Perfect Competton and the Nash Barganng Soluton Renhard John Department of Economcs Unversty of Bonn Adenauerallee 24-42 53113 Bonn, Germany emal: rohn@un-bonn.de May 2005 Abstract For a lnear exchange

More information

Module 17: Mechanism Design & Optimal Auctions

Module 17: Mechanism Design & Optimal Auctions Module 7: Mechansm Desgn & Optmal Auctons Informaton Economcs (Ec 55) George Georgads Examples: Auctons Blateral trade Producton and dstrbuton n socety General Setup N agents Each agent has prvate nformaton

More information

Welfare Properties of General Equilibrium. What can be said about optimality properties of resource allocation implied by general equilibrium?

Welfare Properties of General Equilibrium. What can be said about optimality properties of resource allocation implied by general equilibrium? APPLIED WELFARE ECONOMICS AND POLICY ANALYSIS Welfare Propertes of General Equlbrum What can be sad about optmalty propertes of resource allocaton mpled by general equlbrum? Any crteron used to compare

More information

The Second Anti-Mathima on Game Theory

The Second Anti-Mathima on Game Theory The Second Ant-Mathma on Game Theory Ath. Kehagas December 1 2006 1 Introducton In ths note we wll examne the noton of game equlbrum for three types of games 1. 2-player 2-acton zero-sum games 2. 2-player

More information

Market structure and Innovation

Market structure and Innovation Market structure and Innovaton Ths presentaton s based on the paper Market structure and Innovaton authored by Glenn C. Loury, publshed n The Quarterly Journal of Economcs, Vol. 93, No.3 ( Aug 1979) I.

More information

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification E395 - Pattern Recognton Solutons to Introducton to Pattern Recognton, Chapter : Bayesan pattern classfcaton Preface Ths document s a soluton manual for selected exercses from Introducton to Pattern Recognton

More information

Games of Threats. Elon Kohlberg Abraham Neyman. Working Paper

Games of Threats. Elon Kohlberg Abraham Neyman. Working Paper Games of Threats Elon Kohlberg Abraham Neyman Workng Paper 18-023 Games of Threats Elon Kohlberg Harvard Busness School Abraham Neyman The Hebrew Unversty of Jerusalem Workng Paper 18-023 Copyrght 2017

More information

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4)

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4) I. Classcal Assumptons Econ7 Appled Econometrcs Topc 3: Classcal Model (Studenmund, Chapter 4) We have defned OLS and studed some algebrac propertes of OLS. In ths topc we wll study statstcal propertes

More information

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

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan Wnter 2008 CS567 Stochastc Lnear/Integer Programmng Guest Lecturer: Xu, Huan Class 2: More Modelng Examples 1 Capacty Expanson Capacty expanson models optmal choces of the tmng and levels of nvestments

More information

Introduction. 1. The Model

Introduction. 1. The Model H23, Q5 Introducton In the feld of polluton regulaton the problems stemmng from the asymmetry of nformaton between the regulator and the pollutng frms have been thoroughly studed. The semnal works by Wetzman

More information

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

Endogenous timing in a mixed oligopoly consisting of a single public firm and foreign competitors. Abstract Endogenous tmng n a mxed olgopoly consstng o a sngle publc rm and oregn compettors Yuanzhu Lu Chna Economcs and Management Academy, Central Unversty o Fnance and Economcs Abstract We nvestgate endogenous

More information

Online Appendix: Reciprocity with Many Goods

Online Appendix: Reciprocity with Many Goods T D T A : O A Kyle Bagwell Stanford Unversty and NBER Robert W. Stager Dartmouth College and NBER March 2016 Abstract Ths onlne Appendx extends to a many-good settng the man features of recprocty emphaszed

More information

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

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 Statstcs 1: Probablty Theory II 37 3 EPECTATION OF SEVERAL RANDOM VARIABLES As n Probablty Theory I, the nterest n most stuatons les not on the actual dstrbuton of a random vector, but rather on a number

More information

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

Online Appendix. t=1 (p t w)q t. Then the first order condition shows that Artcle forthcomng to ; manuscrpt no (Please, provde the manuscrpt number!) 1 Onlne Appendx Appendx E: Proofs Proof of Proposton 1 Frst we derve the equlbrum when the manufacturer does not vertcally ntegrate

More information

Assortment Optimization under MNL

Assortment Optimization under MNL Assortment Optmzaton under MNL Haotan Song Aprl 30, 2017 1 Introducton The assortment optmzaton problem ams to fnd the revenue-maxmzng assortment of products to offer when the prces of products are fxed.

More information

The Order Relation and Trace Inequalities for. Hermitian Operators

The Order Relation and Trace Inequalities for. Hermitian Operators Internatonal Mathematcal Forum, Vol 3, 08, no, 507-57 HIKARI Ltd, wwwm-hkarcom https://doorg/0988/mf088055 The Order Relaton and Trace Inequaltes for Hermtan Operators Y Huang School of Informaton Scence

More information

Game Theory Course: Jackson, Leyton-Brown & Shoham. Vickrey-Clarke-Groves Mechanisms: Definitions

Game Theory Course: Jackson, Leyton-Brown & Shoham. Vickrey-Clarke-Groves Mechanisms: Definitions Vckrey-Clarke-Groves Mechansms: Defntons Game Theory Course: Jackson, Leyton-Brown & Shoham A postve result Recall that n the quaslnear utlty settng, a drect mechansm conssts of a choce rule and a payment

More information

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction ECONOMICS 5* -- NOTE (Summary) ECON 5* -- NOTE The Multple Classcal Lnear Regresson Model (CLRM): Specfcaton and Assumptons. Introducton CLRM stands for the Classcal Lnear Regresson Model. The CLRM s also

More information

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

CS : Algorithms and Uncertainty Lecture 17 Date: October 26, 2016 CS 29-128: Algorthms and Uncertanty Lecture 17 Date: October 26, 2016 Instructor: Nkhl Bansal Scrbe: Mchael Denns 1 Introducton In ths lecture we wll be lookng nto the secretary problem, and an nterestng

More information

Edge Isoperimetric Inequalities

Edge Isoperimetric Inequalities November 7, 2005 Ross M. Rchardson Edge Isopermetrc Inequaltes 1 Four Questons Recall that n the last lecture we looked at the problem of sopermetrc nequaltes n the hypercube, Q n. Our noton of boundary

More information

Hila Etzion. Min-Seok Pang

Hila Etzion. Min-Seok Pang RESERCH RTICLE COPLEENTRY ONLINE SERVICES IN COPETITIVE RKETS: INTINING PROFITILITY IN THE PRESENCE OF NETWORK EFFECTS Hla Etzon Department of Technology and Operatons, Stephen. Ross School of usness,

More information

Incentive Compatible Market Design with an Application to. Matching with Wages

Incentive Compatible Market Design with an Application to. Matching with Wages Incentve Compatble Market Desgn wth an Applcaton to Matchng wth Wages M. Bumn Yenmez Stanford Graduate School of Busness Job Market Paper November 1, 2009 Abstract: Ths paper studes markets for heterogeneous

More information

ON THE EQUIVALENCE OF ORDINAL BAYESIAN INCENTIVE COMPATIBILITY AND DOMINANT STRATEGY INCENTIVE COMPATIBILITY FOR RANDOM RULES

ON THE EQUIVALENCE OF ORDINAL BAYESIAN INCENTIVE COMPATIBILITY AND DOMINANT STRATEGY INCENTIVE COMPATIBILITY FOR RANDOM RULES ON THE EQUIVALENCE OF ORDINAL BAYESIAN INCENTIVE COMPATIBILITY AND DOMINANT STRATEGY INCENTIVE COMPATIBILITY FOR RANDOM RULES Madhuparna Karmokar 1 and Souvk Roy 1 1 Economc Research Unt, Indan Statstcal

More information

Economics 101. Lecture 4 - Equilibrium and Efficiency

Economics 101. Lecture 4 - Equilibrium and Efficiency Economcs 0 Lecture 4 - Equlbrum and Effcency Intro As dscussed n the prevous lecture, we wll now move from an envronment where we looed at consumers mang decsons n solaton to analyzng economes full of

More information

Incentive Compatible Transfers in Linear Environments

Incentive Compatible Transfers in Linear Environments Incentve Compatble Transfers n Lnear Envronments Nenad Kos Dept. of Economcs, IGIER Boccon Unversty Matthas Messner Dept. of Economcs, IGIER Boccon Unversty August 5, 2010 Abstract We study the mechansm

More information

Ex post implementation in environments with private goods

Ex post implementation in environments with private goods Theoretcal Economcs 1 (2006), 369 393 1555-7561/20060369 Ex post mplementaton n envronments wth prvate goods SUSHIL BIKHCHANDANI Anderson School of Management, Unversty of Calforna, Los Angeles We prove

More information

Game Theory. Lecture Notes By Y. Narahari. Department of Computer Science and Automation Indian Institute of Science Bangalore, India February 2008

Game Theory. Lecture Notes By Y. Narahari. Department of Computer Science and Automation Indian Institute of Science Bangalore, India February 2008 Game Theory Lecture Notes By Y. Narahar Department of Computer Scence and Automaton Indan Insttute of Scence Bangalore, Inda February 2008 Chapter 10: Two Person Zero Sum Games Note: Ths s a only a draft

More information

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

princeton univ. F 17 cos 521: Advanced Algorithm Design Lecture 7: LP Duality Lecturer: Matt Weinberg prnceton unv. F 17 cos 521: Advanced Algorthm Desgn Lecture 7: LP Dualty Lecturer: Matt Wenberg Scrbe: LP Dualty s an extremely useful tool for analyzng structural propertes of lnear programs. Whle there

More information

Economics 2450A: Public Economics Section 10: Education Policies and Simpler Theory of Capital Taxation

Economics 2450A: Public Economics Section 10: Education Policies and Simpler Theory of Capital Taxation Economcs 2450A: Publc Economcs Secton 10: Educaton Polces and Smpler Theory of Captal Taxaton Matteo Parads November 14, 2016 In ths secton we study educaton polces n a smplfed verson of framework analyzed

More information

COS 521: Advanced Algorithms Game Theory and Linear Programming

COS 521: Advanced Algorithms Game Theory and Linear Programming COS 521: Advanced Algorthms Game Theory and Lnear Programmng Moses Charkar February 27, 2013 In these notes, we ntroduce some basc concepts n game theory and lnear programmng (LP). We show a connecton

More information

Pricing Network Services by Jun Shu, Pravin Varaiya

Pricing Network Services by Jun Shu, Pravin Varaiya Prcng Network Servces by Jun Shu, Pravn Varaya Presented by Hayden So September 25, 2003 Introducton: Two Network Problems Engneerng: A game theoretcal sound congeston control mechansm that s ncentve compatble

More information

Optimal Dynamic Mechanism Design and the Virtual Pivot Mechanism

Optimal Dynamic Mechanism Design and the Virtual Pivot Mechanism Optmal Dynamc Mechansm Desgn and the Vrtual Pvot Mechansm Sham M. Kakade Ilan Lobel Hamd Nazerzadeh March 25, 2011 Abstract We consder the problem of desgnng optmal mechansms for settngs where agents have

More information

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix Lectures - Week 4 Matrx norms, Condtonng, Vector Spaces, Lnear Independence, Spannng sets and Bass, Null space and Range of a Matrx Matrx Norms Now we turn to assocatng a number to each matrx. We could

More information

More metrics on cartesian products

More metrics on cartesian products More metrcs on cartesan products If (X, d ) are metrc spaces for 1 n, then n Secton II4 of the lecture notes we defned three metrcs on X whose underlyng topologes are the product topology The purpose of

More information

Mixed Taxation and Production Efficiency

Mixed Taxation and Production Efficiency Floran Scheuer 2/23/2016 Mxed Taxaton and Producton Effcency 1 Overvew 1. Unform commodty taxaton under non-lnear ncome taxaton Atknson-Stgltz (JPubE 1976) Theorem Applcaton to captal taxaton 2. Unform

More information

Uniqueness of Nash Equilibrium in Private Provision of Public Goods: Extension. Nobuo Akai *

Uniqueness of Nash Equilibrium in Private Provision of Public Goods: Extension. Nobuo Akai * Unqueness of Nash Equlbrum n Prvate Provson of Publc Goods: Extenson Nobuo Aka * nsttute of Economc Research Kobe Unversty of Commerce Abstract Ths note proves unqueness of Nash equlbrum n prvate provson

More information

Maximizing the number of nonnegative subsets

Maximizing the number of nonnegative subsets Maxmzng the number of nonnegatve subsets Noga Alon Hao Huang December 1, 213 Abstract Gven a set of n real numbers, f the sum of elements of every subset of sze larger than k s negatve, what s the maxmum

More information

Smooth Games, Price of Anarchy and Composability of Auctions - a Quick Tutorial

Smooth Games, Price of Anarchy and Composability of Auctions - a Quick Tutorial Smooth Games, Prce of Anarchy and Composablty of Auctons - a Quck Tutoral Abhshek Snha Laboratory for Informaton and Decson Systems, Massachusetts Insttute of Technology, Cambrdge, MA 02139 Emal: snhaa@mt.edu

More information

Conjectures in Cournot Duopoly under Cost Uncertainty

Conjectures in Cournot Duopoly under Cost Uncertainty Conjectures n Cournot Duopoly under Cost Uncertanty Suyeol Ryu and Iltae Km * Ths paper presents a Cournot duopoly model based on a condton when frms are facng cost uncertanty under rsk neutralty and rsk

More information

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity LINEAR REGRESSION ANALYSIS MODULE IX Lecture - 30 Multcollnearty Dr. Shalabh Department of Mathematcs and Statstcs Indan Insttute of Technology Kanpur 2 Remedes for multcollnearty Varous technques have

More information

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

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 Copyrght (C) 008 Davd K. Levne Ths document s an open textbook; you can redstrbute t and/or modfy t under the terms of the Creatve Commons Attrbuton Lcense. Compettve Equlbrum wth Pure Exchange n traders

More information

APPENDIX A Some Linear Algebra

APPENDIX A Some Linear Algebra APPENDIX A Some Lnear Algebra The collecton of m, n matrces A.1 Matrces a 1,1,..., a 1,n A = a m,1,..., a m,n wth real elements a,j s denoted by R m,n. If n = 1 then A s called a column vector. Smlarly,

More information

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module 3 LOSSY IMAGE COMPRESSION SYSTEMS Verson ECE IIT, Kharagpur Lesson 6 Theory of Quantzaton Verson ECE IIT, Kharagpur Instructonal Objectves At the end of ths lesson, the students should be able to:

More information

Lecture Notes on Linear Regression

Lecture Notes on Linear Regression Lecture Notes on Lnear Regresson Feng L fl@sdueducn Shandong Unversty, Chna Lnear Regresson Problem In regresson problem, we am at predct a contnuous target value gven an nput feature vector We assume

More information

Externalities in wireless communication: A public goods solution approach to power allocation. by Shrutivandana Sharma

Externalities in wireless communication: A public goods solution approach to power allocation. by Shrutivandana Sharma Externaltes n wreless communcaton: A publc goods soluton approach to power allocaton by Shrutvandana Sharma SI 786 Tuesday, Feb 2, 2006 Outlne Externaltes: Introducton Plannng wth externaltes Power allocaton:

More information

Notes on Frequency Estimation in Data Streams

Notes on Frequency Estimation in Data Streams Notes on Frequency Estmaton n Data Streams In (one of) the data streamng model(s), the data s a sequence of arrvals a 1, a 2,..., a m of the form a j = (, v) where s the dentty of the tem and belongs to

More information

Learning from Private Information in Noisy Repeated Games

Learning from Private Information in Noisy Repeated Games Learnng from Prvate Informaton n Nosy Repeated Games Drew Fudenberg and Yuch Yamamoto Frst verson: February 18, 2009 Ths verson: November 3, 2010 Abstract We study the perfect type-contngently publc ex-post

More information

Mechanism Design with Multidimensional, Continuous Types and Interdependent Valuations

Mechanism Design with Multidimensional, Continuous Types and Interdependent Valuations Mechansm Desgn wth Multdmensonal, Contnuous Types and Interdependent Valuatons Nolan H. Mller, John W. Pratt, Rchard J. Zeckhauser and Scott Johnson, May 23, 2006 Abstract We consder the mechansm desgn

More information

Behavioral Aspects of Implementation Theory

Behavioral Aspects of Implementation Theory CIRJE-F-523 Behavoral Aspects of Ipleentaton Theory Htosh Matsusha Unversty of Tokyo October 2007 CIRJE Dscusson Papers can be downloaded wthout charge fro: http://www.e.u-tokyo.ac.jp/crje/research/03research02dp.htl

More information

Pricing and Resource Allocation Game Theoretic Models

Pricing and Resource Allocation Game Theoretic Models Prcng and Resource Allocaton Game Theoretc Models Zhy Huang Changbn Lu Q Zhang Computer and Informaton Scence December 8, 2009 Z. Huang, C. Lu, and Q. Zhang (CIS) Game Theoretc Models December 8, 2009

More information

a b a In case b 0, a being divisible by b is the same as to say that

a b a In case b 0, a being divisible by b is the same as to say that Secton 6.2 Dvsblty among the ntegers An nteger a ε s dvsble by b ε f there s an nteger c ε such that a = bc. Note that s dvsble by any nteger b, snce = b. On the other hand, a s dvsble by only f a = :

More information

Lecture 3: Probability Distributions

Lecture 3: Probability Distributions Lecture 3: Probablty Dstrbutons Random Varables Let us begn by defnng a sample space as a set of outcomes from an experment. We denote ths by S. A random varable s a functon whch maps outcomes nto the

More information

THE CHVÁTAL-ERDŐS CONDITION AND 2-FACTORS WITH A SPECIFIED NUMBER OF COMPONENTS

THE CHVÁTAL-ERDŐS CONDITION AND 2-FACTORS WITH A SPECIFIED NUMBER OF COMPONENTS Dscussones Mathematcae Graph Theory 27 (2007) 401 407 THE CHVÁTAL-ERDŐS CONDITION AND 2-FACTORS WITH A SPECIFIED NUMBER OF COMPONENTS Guantao Chen Department of Mathematcs and Statstcs Georga State Unversty,

More information

Composite Hypotheses testing

Composite Hypotheses testing Composte ypotheses testng In many hypothess testng problems there are many possble dstrbutons that can occur under each of the hypotheses. The output of the source s a set of parameters (ponts n a parameter

More information

Axiomatizations of Pareto Equilibria in Multicriteria Games

Axiomatizations of Pareto Equilibria in Multicriteria Games ames and Economc Behavor 28, 146154 1999. Artcle ID game.1998.0680, avalable onlne at http:www.dealbrary.com on Axomatzatons of Pareto Equlbra n Multcrtera ames Mark Voorneveld,* Dres Vermeulen, and Peter

More information

COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY

COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY ROBUST MECHANISM DESIGN By Drk Bergemann and Stephen Morrs Aprl 2004 COWLES FOUNDATION DISCUSSION PAPER NO. 1421R COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY Box 208281 New Haven, Connectcut

More information

Constant Best-Response Functions: Interpreting Cournot

Constant Best-Response Functions: Interpreting Cournot Internatonal Journal of Busness and Economcs, 009, Vol. 8, No., -6 Constant Best-Response Functons: Interpretng Cournot Zvan Forshner Department of Economcs, Unversty of Hafa, Israel Oz Shy * Research

More information

Deterministic versus Stochastic Mechanisms in Principal Agent Models

Deterministic versus Stochastic Mechanisms in Principal Agent Models Dscusson Paper No. 26 Determnstc versus Stochastc Mechansms n Prncpal Agent Models Roland Strausz* September 2004 *Roland Strausz, Free Unversty Berln, Department of Economcs, Boltzmannstr. 20, D-14195

More information

Equilibrium with Complete Markets. Instructor: Dmytro Hryshko

Equilibrium with Complete Markets. Instructor: Dmytro Hryshko Equlbrum wth Complete Markets Instructor: Dmytro Hryshko 1 / 33 Readngs Ljungqvst and Sargent. Recursve Macroeconomc Theory. MIT Press. Chapter 8. 2 / 33 Equlbrum n pure exchange, nfnte horzon economes,

More information

On the correction of the h-index for career length

On the correction of the h-index for career length 1 On the correcton of the h-ndex for career length by L. Egghe Unverstet Hasselt (UHasselt), Campus Depenbeek, Agoralaan, B-3590 Depenbeek, Belgum 1 and Unverstet Antwerpen (UA), IBW, Stadscampus, Venusstraat

More information

Subjective Uncertainty Over Behavior Strategies: A Correction

Subjective Uncertainty Over Behavior Strategies: A Correction Subjectve Uncertanty Over Behavor Strateges: A Correcton The Harvard communty has made ths artcle openly avalable. Please share how ths access benefts you. Your story matters. Ctaton Publshed Verson Accessed

More information

THE DYNAMIC PIVOT MECHANISM

THE DYNAMIC PIVOT MECHANISM Econometrca, Vol. 78, No. 2 (March, 2010), 771 789 THE DYNAMIC PIVOT MECHANISM BY DIRK BERGEMANN AND JUUSO VÄLIMÄKI 1 We consder truthful mplementaton of the socally effcent allocaton n an ndependent prvate-value

More information

EFFICIENT DYNAMIC AUCTIONS. Dirk Bergemann and Juuso Välimäki. October 2006 COWLES FOUNDATION DISCUSSION PAPER NO. 1584

EFFICIENT DYNAMIC AUCTIONS. Dirk Bergemann and Juuso Välimäki. October 2006 COWLES FOUNDATION DISCUSSION PAPER NO. 1584 EFFICIENT DYNAMIC AUCTIONS By Drk Bergemann and Juuso Välmäk October 2006 COWLES FOUNDATION DISCUSSION PAPER NO. 1584 COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY Box 208281 New Haven, Connectcut

More information

Computing Correlated Equilibria in Multi-Player Games

Computing Correlated Equilibria in Multi-Player Games Computng Correlated Equlbra n Mult-Player Games Chrstos H. Papadmtrou Presented by Zhanxang Huang December 7th, 2005 1 The Author Dr. Chrstos H. Papadmtrou CS professor at UC Berkley (taught at Harvard,

More information

Robust Implementation: The Role of Large Type Spaces

Robust Implementation: The Role of Large Type Spaces Robust Implementaton: The Role of Large Type Spaces Drk Bergemann y Stephen Morrs z Frst Verson: March 2003 Ths Verson: Aprl 2004 Abstract We analyze the problem of fully mplementng a socal choce functon

More information

ECE559VV Project Report

ECE559VV Project Report ECE559VV Project Report (Supplementary Notes Loc Xuan Bu I. MAX SUM-RATE SCHEDULING: THE UPLINK CASE We have seen (n the presentaton that, for downlnk (broadcast channels, the strategy maxmzng the sum-rate

More information

2.3 Nilpotent endomorphisms

2.3 Nilpotent endomorphisms s a block dagonal matrx, wth A Mat dm U (C) In fact, we can assume that B = B 1 B k, wth B an ordered bass of U, and that A = [f U ] B, where f U : U U s the restrcton of f to U 40 23 Nlpotent endomorphsms

More information

On the Multicriteria Integer Network Flow Problem

On the Multicriteria Integer Network Flow Problem BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 5, No 2 Sofa 2005 On the Multcrtera Integer Network Flow Problem Vassl Vasslev, Marana Nkolova, Maryana Vassleva Insttute of

More information

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers Psychology 282 Lecture #24 Outlne Regresson Dagnostcs: Outlers In an earler lecture we studed the statstcal assumptons underlyng the regresson model, ncludng the followng ponts: Formal statement of assumptons.

More information

4 Analysis of Variance (ANOVA) 5 ANOVA. 5.1 Introduction. 5.2 Fixed Effects ANOVA

4 Analysis of Variance (ANOVA) 5 ANOVA. 5.1 Introduction. 5.2 Fixed Effects ANOVA 4 Analyss of Varance (ANOVA) 5 ANOVA 51 Introducton ANOVA ANOVA s a way to estmate and test the means of multple populatons We wll start wth one-way ANOVA If the populatons ncluded n the study are selected

More information

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

Introduction to Vapor/Liquid Equilibrium, part 2. Raoult s Law: CE304, Sprng 2004 Lecture 4 Introducton to Vapor/Lqud Equlbrum, part 2 Raoult s Law: The smplest model that allows us do VLE calculatons s obtaned when we assume that the vapor phase s an deal gas, and

More information

FREQUENCY DISTRIBUTIONS Page 1 of The idea of a frequency distribution for sets of observations will be introduced,

FREQUENCY DISTRIBUTIONS Page 1 of The idea of a frequency distribution for sets of observations will be introduced, FREQUENCY DISTRIBUTIONS Page 1 of 6 I. Introducton 1. The dea of a frequency dstrbuton for sets of observatons wll be ntroduced, together wth some of the mechancs for constructng dstrbutons of data. Then

More information

Dynamic Efcient Auctions

Dynamic Efcient Auctions Dynamc fcent Auctons Drk Bergemann y and Juuso Valmak z ABSTRACT (xtended Abstract) We consder the truthful mplementaton of the socally e cent allocaton n a dynamc prvate value envronment n whch agents

More information

j) = 1 (note sigma notation) ii. Continuous random variable (e.g. Normal distribution) 1. density function: f ( x) 0 and f ( x) dx = 1

j) = 1 (note sigma notation) ii. Continuous random variable (e.g. Normal distribution) 1. density function: f ( x) 0 and f ( x) dx = 1 Random varables Measure of central tendences and varablty (means and varances) Jont densty functons and ndependence Measures of assocaton (covarance and correlaton) Interestng result Condtonal dstrbutons

More information

On the Generic (Im)possibility of Full Surplus Extraction in Mechanism Design

On the Generic (Im)possibility of Full Surplus Extraction in Mechanism Design On the Generc (Im)possblty of Full Surplus Extracton n Mechansm Desgn Avad Hefetz vka Neeman Prelmnary Verson: February 27, 2004 Abstract A number of studes, most notably Crémer and McLean (985, 988),

More information

Numerical Heat and Mass Transfer

Numerical Heat and Mass Transfer Master degree n Mechancal Engneerng Numercal Heat and Mass Transfer 06-Fnte-Dfference Method (One-dmensonal, steady state heat conducton) Fausto Arpno f.arpno@uncas.t Introducton Why we use models and

More information

Informational Size and Efficient Auctions

Informational Size and Efficient Auctions Revew of Economc Studes (2004) 71, 809 827 0034-6527/04/00330809$02.00 c 2004 The Revew of Economc Studes Lmted Informatonal Sze and Effcent Auctons RICHARD MCLEAN Rutgers Unversty and ANDREW POSTLEWAITE

More information

Revenue Maximization in a Spectrum Auction for Dynamic Spectrum Access

Revenue Maximization in a Spectrum Auction for Dynamic Spectrum Access Revenue Maxmzaton n a Spectrum Aucton for Dynamc Spectrum Access Al Kakhbod, Ashutosh Nayyar and Demosthens Teneketzs Department of Electrcal Engneerng and Computer Scence Unversty of Mchgan, Ann Arbor,

More information

Module 9. Lecture 6. Duality in Assignment Problems

Module 9. Lecture 6. Duality in Assignment Problems Module 9 1 Lecture 6 Dualty n Assgnment Problems In ths lecture we attempt to answer few other mportant questons posed n earler lecture for (AP) and see how some of them can be explaned through the concept

More information

MECHANISM DESIGN BY AN INFORMED PRINCIPAL: THE QUASI-LINEAR PRIVATE-VALUES CASE

MECHANISM DESIGN BY AN INFORMED PRINCIPAL: THE QUASI-LINEAR PRIVATE-VALUES CASE USC FBE APPLIED ECONOMICS WORKSHOP presented by Tymofy Mylovanov FRIDAY, Oct. 5, 2012 1:30 pm - 3:00 pm, Room: HOH-706 MECHANISM DESIGN BY AN INFORMED PRINCIPAL: THE QUASI-LINEAR PRIVATE-VALUES CASE TYMOFIY

More information

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE Analytcal soluton s usually not possble when exctaton vares arbtrarly wth tme or f the system s nonlnear. Such problems can be solved by numercal tmesteppng

More information

Credit Card Pricing and Impact of Adverse Selection

Credit Card Pricing and Impact of Adverse Selection Credt Card Prcng and Impact of Adverse Selecton Bo Huang and Lyn C. Thomas Unversty of Southampton Contents Background Aucton model of credt card solctaton - Errors n probablty of beng Good - Errors n

More information

Managing Snow Risks: The Case of City Governments and Ski Resorts

Managing Snow Risks: The Case of City Governments and Ski Resorts Managng Snow Rsks: The Case of Cty Governments and Sk Resorts Haruyosh Ito* Assstant Professor of Fnance Graduate School of Internatonal Management Internatonal Unversty of Japan 777 Kokusacho, Mnam Uonuma-sh

More information

Feature Selection: Part 1

Feature Selection: Part 1 CSE 546: Machne Learnng Lecture 5 Feature Selecton: Part 1 Instructor: Sham Kakade 1 Regresson n the hgh dmensonal settng How do we learn when the number of features d s greater than the sample sze n?

More information

1. Inference on Regression Parameters a. Finding Mean, s.d and covariance amongst estimates. 2. Confidence Intervals and Working Hotelling Bands

1. Inference on Regression Parameters a. Finding Mean, s.d and covariance amongst estimates. 2. Confidence Intervals and Working Hotelling Bands Content. Inference on Regresson Parameters a. Fndng Mean, s.d and covarance amongst estmates.. Confdence Intervals and Workng Hotellng Bands 3. Cochran s Theorem 4. General Lnear Testng 5. Measures of

More information

MMA and GCMMA two methods for nonlinear optimization

MMA and GCMMA two methods for nonlinear optimization MMA and GCMMA two methods for nonlnear optmzaton Krster Svanberg Optmzaton and Systems Theory, KTH, Stockholm, Sweden. krlle@math.kth.se Ths note descrbes the algorthms used n the author s 2007 mplementatons

More information

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

k t+1 + c t A t k t, t=0 Macro II (UC3M, MA/PhD Econ) Professor: Matthas Kredler Fnal Exam 6 May 208 You have 50 mnutes to complete the exam There are 80 ponts n total The exam has 4 pages If somethng n the queston s unclear,

More information

Affine transformations and convexity

Affine transformations and convexity Affne transformatons and convexty The purpose of ths document s to prove some basc propertes of affne transformatons nvolvng convex sets. Here are a few onlne references for background nformaton: http://math.ucr.edu/

More information

Kernel Methods and SVMs Extension

Kernel Methods and SVMs Extension Kernel Methods and SVMs Extenson The purpose of ths document s to revew materal covered n Machne Learnng 1 Supervsed Learnng regardng support vector machnes (SVMs). Ths document also provdes a general

More information

MECHANISM DESIGN BY AN INFORMED PRINCIPAL: PRIVATE-VALUES WITH TRANSFERABLE UTILITY

MECHANISM DESIGN BY AN INFORMED PRINCIPAL: PRIVATE-VALUES WITH TRANSFERABLE UTILITY MECHANISM DESIGN BY AN INFORMED PRINCIPAL: PRIVATE-VALUES WITH TRANSFERABLE UTILITY TYMOFIY MYLOVANOV AND THOMAS TRÖGER Abstract. We provde a soluton to the nformed-prncpal problem n the ndependent-prvate-values

More information

UNIVERSITY OF ILLINOIS LIBRARY AT URBANA CHAMPAIGN BOOKSTACKS

UNIVERSITY OF ILLINOIS LIBRARY AT URBANA CHAMPAIGN BOOKSTACKS UNIVERSITY OF ILLINOIS LIBRARY AT URBANA CHAMPAIGN BOOKSTACKS Dgtzed by the Internet Archve n 2011 wth fundng from Unversty of Illnos Urbana-Champagn http://www.archve.org/detals/effcencymechan1641chak

More information

Difference Equations

Difference Equations Dfference Equatons c Jan Vrbk 1 Bascs Suppose a sequence of numbers, say a 0,a 1,a,a 3,... s defned by a certan general relatonshp between, say, three consecutve values of the sequence, e.g. a + +3a +1

More information

Stanford University CS359G: Graph Partitioning and Expanders Handout 4 Luca Trevisan January 13, 2011

Stanford University CS359G: Graph Partitioning and Expanders Handout 4 Luca Trevisan January 13, 2011 Stanford Unversty CS359G: Graph Parttonng and Expanders Handout 4 Luca Trevsan January 3, 0 Lecture 4 In whch we prove the dffcult drecton of Cheeger s nequalty. As n the past lectures, consder an undrected

More information