Notes on Bias Uncertainty and Communication in Committees 1

Size: px
Start display at page:

Download "Notes on Bias Uncertainty and Communication in Committees 1"

Transcription

1 Notes on Bas Uncertanty and Communcaton n Commttees 1 Davd Austen-Smth MEDS, Kellogg Graduate School of Management Northwestern Unversty Evanston, IL Tm Feddersen MEDS, Kellogg Graduate School of Management Northwestern Unversty Evanston, IL March Prepared for presentaton to the Roy Semnar, Pars, March Part of these Notes rely on Austen-Smth and Feddersen (2006).

2 1 Introducton It s well understood that when commttee members vote under ncomplete nformaton, the resultng commttee decson need not re ect the decson that would have been made under fully shared nformaton (eg Austen-Smth and Banks 1996; Feddersen and Pesendorfer 1998). But people n commttees often talk before votng and so have an opportunty to share decson-relevant nformaton. Over the past few years there has been a growng strategc game-theoretc lterature concerned to understand better what mplcatons such communcaton mght have for the character and qualty of collectve decsons under ncomplete nformaton. 1 One ssue here concerns how d erent votng rules for reachng a nal collectve choce n uences the nformaton that mght be shared n any pror debate. Gerard and Yarv (2004) suggest that the votng rule (essentally) s rrelevant, but ther argument hnges on usng domnated (debate-condtonal) votng strateges. Assumng undomnated votng changes the pcture (Austen-Smth and Feddersen 2005). And n ths case, t turns out that whether or not ndvduals bases, that s, d erences n full nformaton preferences over avalable collectve choces, are common knowledge plays a crtcal role. 2 In these Notes, we revew some of the postve ssues arsng from bas uncertanty and suggest some ntutons regardng the welfare mplcatons of bases beng prvate nformaton. 2 An example Consder a commttee of three people, = 1; 2; 3, that has to choose between a xed par of alternatves fx; Y g. Each ndvdual has prvate nformaton (b ; s ) 2 fx; yg fx; yg, where b s a preference parameter, or bas, and s s a nosy but nformatve sgnal regardng the 1 For example, see: Austen-Smth 1990; Calvert and Johnson 1998; Coughlan 2000; Doraszelsk et al 2003; Gerard and Yarv 2004; Merowtz 2006, 2007; Hafer and Landa 2007; Austen-Smth and Feddersen 2005, 2006; Callaud and Trole An ntuton noted too n Merowtz (2006). 1

3 alternatves. Let (b; s) = ((b 1 ; b 2 ; b 3 ); (s 1 ; s 2 ; s 3 )) denote a pro le of realzed bases and sgnals. For ths example, assume sgnals are uncorrelated wth bases. Let fa 1 ; A 2 g be states such that, for all, Pr[s = xja 1 ] = Pr[s = yja 2 ] = p 2 (1=2; 1) wth the common pror probablty on A 1 beng 1=2; and, for all, assume Pr[b = x] = r 2 (0; 1). Wrte u(z; b; s) for an ndvdual s payo from the collectve choce Z 2 fx; Y g, gven that ndvdual s bas s b 2 fx; yg and the pro le of sgnals s s 2 fx; yg 3. Assume 8 < 0 f s 2 f (y; y; y)g u(x; x; s) = : 1 otherwse u(y; x; s) = 1 u(x; x; s) 8 < 1 f s 2 f (x; x; x)g u(x; y; s) = : 0 otherwse u(y; y; s) = 1 u(x; y; s) Suppose rst that the commttee s choce s determned by a majorty vote. In the absence of any pre-vote communcaton, there s no equlbrum n whch all ndvduals vote nformatvely, that s, wth ther sgnals (eg Austen-Smth and Banks, 1996). As a result, the collectve choce may favour a mnorty ex post. More formally, let v : fx; ygfx; yg! [0; 1] be a vote strategy, where v(b; s) s the probablty an ndvdual wth bas b and sgnal s votes for X (we assume strateges are anonymous throughout). equlbrum n undomnated strateges n whch v(b; x) = 1 Then under majorty rule, there s no (Bayesan) v(b; y) = 1 for all b. To see ths suppose not and, wthout loss of generalty, consder an ndvdual wth bas b = y. Gven the other commttee members are votng wth ther sgnals, s pvotal only n the event that one ndvdual has a sgnal x and the other has a sgnal y; but then s unque best response s to vote surely for Y rrespectve of hs sgnal. There s, however, a mxed strategy equlbrum when q s not too extreme: n ths equlbrum, v(x; x) = 1 2 v(y; y) = 1, v(x; y) 2 (0; 1) and

4 v(y; x) 2 (0; 1). And clearly, the nal decson may not re ect the full nformaton majorty preference. Now suppose that the commttee can delberate before votng; spec cally, suppose that votng follows one round of smultaneous cheap talk sgnalng. Note that n ths context, and n contrast to n-person sender-recever games (eg Ottavan and Sorenson 2001; Glazer and Rubnsten 2001), every commttee member s both a "sender" and a "recever" and no one ndvdual has the rght to dctate the nal decson. Let : fx; yg fx; yg! M be a message strategy for, where M s n general an arbtrarly large set of messages. For current purposes, t su ces to take M = fx; yg wth the nterpretaton that (b; s) s the probablty an ndvdual wth bas b and sgnal s declares "s = x". 3 Wth talk, ndvduals votng behavour depend on the realzed lst of messages from debate: wth an abuse of notaton, wrte v : fx; yg fx; yg fx; yg 3! fx; Y g be a vote strategy for, where v (b; s; m) s the probablty an ndvdual wth bas b and sgnal s who hears debate m = (m 1 ; m 2 ; m 3 ) 2 fx; yg 3 votes for X. A perfect Bayesan equlbrum n undomnated strateges for ths settng s a pro le of strateges (; v) such that, for every trple (b; s; m) and all : (1) v (b; s; m) s an undomnated best response to (; v ); (2) (b; s) s an undomnated best response votng strategy to ( ; v); and (3) belefs are consstent wth the strategy pro le and Bayes rule where de ned. Assume all ndvduals truthfully reveal ther sgnals: (b; x) = 1 (b; y) = 1 for all and bases b. Then there s fully shared nformaton at the votng stage and the unque undomnated votng equlbrum condtonal on these messages s for each ndvdual to vote sncerely for ther most preferred alternatve as de ned by (b ; (s 1 ; s 2 ; s 3 )). Wth ths fact n mnd, consder whether n fact truthful nformaton sharng n debate s ncentve compatble. Say that an ndvdual s message pvotal n debate f the nal colllectve choce s senstve to whch sgnal the ndvdual reveals he or she has observed. Then t s mmedate that any 3 Later, n a smpler verson of ths example, we explctly consder allowng bases to be revealed n debate. Dealng wth ths possblty here s not essental. 3

5 x-based ndvdual wth an x sgnal, or y-based ndvdual wth y sgnal, has a domnant strategy to reveal ther sgnal truthfully. So, wthout loss of generalty, consder an x-based ndvdual wth a y sgnal: (b ; s ) = (x; y). Then there are essentally two events n whch s message pvotal: (1) both of the other commttee members are x-based and both have observed a y sgnal, whch occurs wth (condtonal) probablty 1 2 r2 p 3 + (1 p) 3 ; (2) both of the other commttee members are y-based and both have observed an x sgnal, whch occurs wth (condtonal) probablty 1 2 (1 r)2 (1 p) 2 p + (1 p)p 2 = 1 2 (1 r)2 p(1 p): In event (1), ndvdual strctly prefers to tell the truth whle, n event (2), strctly prefers to dssemble. Makng the requste calculatons, tellng the truth n debate condtonal on all others fully revealng ther sgnals s ncentve compatble for an x-based ndvdual wth a y sgnal f and only f r 2 1 r p(1 p) (1 p) 3 + p 3 : Smlarly, ncentve compatblty of truthtellng for a y-based ndvdual wth an x sgnal s nsured f 1 r 2 r p(1 p) (1 p) 3 + p 3 : For every r 2 (0; 1), therefore, there exsts some p < 1 for whch there exsts an equlbrum n whch all ndvduals reveal ther sgnals truthfully n debate and subsequently vote under full nformaton to yeld the ex post rst best de facto majorty preferred outcome. It s mportant to note that the left sde of these nequaltes descrbes the relatve lkelhood of the speaker s bas-type beng the majorty bas n the commttee. Moreover, f r 2 f0; 1g then truthtellng n debate cannot be an equlbrum for any p 2 (1=2; 1): bas uncertanty 4

6 s essental for fully revealng debate. In partcular, wth bas uncertanty there are multple message pvotal events under majorty rule; whether or not any ndvdual tells the truth n debate thus depends on the relatve lkelhood of those events n whch truthtellng s a best response; that s, on those events n whch the ndvdual s a member of the full nformaton wnnng coalton. And because message pvotal events are de ned both n terms of a partcular pars of bas and sgnal pro les, the fact that sgnals are nformatve, coupled wth the event of beng message pvotal, provdes the ndvdual wth nformaton regardng the realzed bas pro le, even when bases and sgnals are statstcally ndependent. Now suppose commttee decson makng s by unanmty rule wth status quo X: that s, X s the outcome unless all commttee members vote for Y. Wthout debate, ndvduals condton ther vote n equlbrum on beng vote pvotal, whch occurs only f both of the other commttee members are votng for the alternatve, Y. Gven all others are votng wth ther sgnal, a y-based ndvdual s best response s therefore to vote for Y rrespectve of her sgnal. Hence, as wth majorty rule, the ex post commttee choce may not be the decson adopted under full nformaton. Suppose there s a debate stage n whch all ndvduals are supposed to report ther sgnals truthfully and consder a y-based ndvdual wth an x sgnal. Under unanmty rule, any ndvdual s best votng decson s ndependent of her own message n debate. Thus the only message pvotal event s when at least one of the other two ndvduals s x-based and both have y sgnals (any b-based ndvdual wth an s = b sgnal surely votes wth hs sgnal). But then the unque best response message s to le and report "y": bas uncertanty s nsu cent to promote fully revealng debate under unanmty rule. In other words, there can only be a sngle message pvotal event under unanmty rule and ths event nduces dssemblng as a best response for at least some types of commttee member. In sum, the example su ces to show that bas uncertanty can promote full nformaton sharng n debate under majorty rule, but does not demonstrate that the same s false under unanmty rule. Furthermore, whle the presence of bas uncertanty can a ect the extent of nformaton sharng n debate, ths n tself does not say anythng about the welfare propertes 5

7 of commttee decson makng under bas uncertanty. In the next secton we address the rst ssue regardng bas uncertanty and unanmty rule qute generally and, n the subsequent secton, use a consderably less general model to consder some welfare mplcatons of bas uncertanty. 3 Model and theorems The commttee N = f1; 2; :::; ng, n 2, has to choose an alternatve Z 2 fx; Y g; let X be the status quo polcy. Each ndvdual 2 N has prvate nformaton (b ; s ) 2 B S, where b s a preference parameter, or bas, and s s a sgnal regardng the alternatves. Assume the sets B and S are nte and common across ndvduals 2 N. Wrte B n B and S n S; a stuaton s any par (b; s) 2 B S, where b = (b 1 ; :::; b n ), s = (s 1 ; :::; s n ). And wth a convenent abuse of language, call any pro le of ndvdual sgnals s 2 S a state, as such a pro le exhausts all of the relevant nformaton for the collectve decson. Let p(b; s) be the probablty that stuaton (b; s) 2 Bl commttee member 2 N, s preferences over fx; Y g depend exclusvely on s own bas b 2 B and on the state s 2 S: gven a bas b and state s, an ndvdual s payo from a commttee decson Z 2 fx; Y g s wrtten u(z; b; s). We assume there are no dogmatc or partsan types; that s, for any bas b 2 B there s a nonempty subset of states S b S such that s 2 S b mples u(y; b; s) > u(x; b; s) and s =2 S b mples u(y; b; s) < u(x; b; s): n other words, all ndvduals preferences over the two alternatves are subject to change. To avod trvaltes we assume that every stuaton occurs wth postve probablty and, because the concern here s wth unanmty rule, that there always exst states at whch all members prefer alternatve Y. The rst two axoms, respectvely, formalze these two essentally techncal assumptons. Full Support For all (b; s) 2 B S, p(b; s) > 0. Consensus For all b = (b 1 ; :::; b n ) 2 B, S(b) \ 2N S b 6= ;. The nal, most substantve, axom mposes some structure on the set of sgnals. We rst 6

8 state the axom formally and then dscuss ts motvaton. Assume that the set of sgnals S s ordered by a bnary relaton,, such that the followng monotoncty condton obtans. Monotoncty For any s; s 0 2 S such that s s 0 and s 2 S n 1, let s = (s ; s) and s 0 = (s ; s 0 ) 2 S. Then u(y; b; s) > u(y; b; s 0 ) and u(x; b; s) < u(x; b; s 0 ) for any b 2 B. In words, suppose there s a par of states that d er only n that some member has observed s 2 S n the rst state and s 0 2 S; then s s 0 mples that s s stronger nformaton than s 0 n favor of Y and aganst X. And notce that ths axom also bulds n a degree of symmetry: any ndvdual s relatve evaluaton of the two alternatves s monotone n sgnals whatever the ndvdual s bas b and rrespectve of exactly whch commttee member receves what sgnal. Thus, the axom nssts that f one ndvdual consders a partcular sgnal to be better evdence n favor of choosng X over Y than some other sgnal, then all ndvduals share ths relatve evaluaton, but they may dsagree about how much better s the evdence. Although a lmtaton, the axom s conservatve for provng a negatve result, as the possbltes for nformaton sharng are clearly greater when all ndvduals share a common vew of what any partcular pece of nformaton mght mean. The commttee chooses an outcome by votng under unanmty rule. That s, X s the outcome unless every member of the commttee votes for y. Pror to votng we assume there s a delberaton phase n whch every member of the commttee can smultaneously send a message m to every other member of the commttee. The focus here s on delberaton that yelds nformaton relevant to the collectve choce beng shared pror to the votng stage, so there s no loss of generalty n assocatng messages drectly wth the nformaton they are presumed to report. Therefore we take the set of avalable messages to any ndvdual to be a set M such that S M. A message strategy for 2 N s a functon, : B S! M. A message pro le m = (m 1 ; m 2 ; :::; m n ) 2 M n M s a debate. De nton 1 A message strategy pro le s fully revealng f, for all 2 N, for all pars of dstnct sgnals s; s 0 2 S, [[ b2b (b; s)] \ [[ b2b (b; s 0 )] = ;. 7

9 As de ned here, fully revealng message strateges may or may not reveal nformaton about ndvdual bases. 4 Because ndvduals preferences depend only on the state and on ther own bas, f a debate fully reveals the state then addtonal nformaton about others bases s decson-rrelevant. Thus the key feature of a fully revealng message strategy s that t provdes full nformaton about the speaker s sgnal. That s, f s fully revealng then, for all ndvduals 2 N and all bas and sgnal pars (b; s) 2 B S, the message (b; s) 2 M unambguously reveals that s prvate sgnal s s. De nton 2 A commttee s mnmally dverse f and only f there exst b; b 0 2 B such that S b 6= S b 0. Note that under the full support assumpton, t s possble for all ndvduals to exhbt the same bas and, therefore, the only commttees that are not mnmally dverse are commttees n whch there s never any dsagreement about when alternatve y s the best choce (S b = S b 0 for all b; b 0 2 B). A votng strategy for member 2 N s a functon v : B S M! fx; yg that maps every debate nto a votng decson. A fully revealng debate equlbrum s a Perfect Bayesan Equlbrum (; v) = (( 1 ; : : : ; n ); (v 1 ; : : : ; v n )) such that s fully revealng and v s a pro le of weakly undomnated votng strateges. Before presentng the result on unanmty rule, t s useful to recall a (slght generalzaton of a) result due to Coughlan (2000). For any q 2 f1; 2; : : : ; ng, a q-rule s a votng rule such that f at least q 1 commttee members vote for Y aganst X, then Y s the commttee decson. Unanmty rule s a q-rule wth q = n. Theorem 1 (Coughlan 2000) Assume the realzed bas pro le s common knowledge and that every state s 2 S occurs wth postve probablty. Assume consensus and monotoncty. Then, 4 At rst glance, the de nton here mght seem unnecessarly awkward wth a smpler verson beng to requre only that f, for all, all b and any s 6= s 0, (b; s) 6= (b; s 0 ). But ths does not work, as t admts the possblty that (b 0 ; s) = (b; s 0_ ) for some b 0 6= b, n whch case s sgnal regardng the state s not revealed. 8

10 for all q rules, n=2 < q < n, there exsts a FRDE f and only f the commttee s not mnmally dverse. Therefore, bas uncertanty s typcally necessary for full nformaton sharng n debate under any rule shy of unanmty. And whle the example showed that bas uncertanty can support a FRDE, ths s by no means guaranteed. That bas uncertanty can never help wth unanmty, however, s the content of the next result. Theorem 2 Assume full support, consensus and monotoncty. There exsts a fully revealng debate equlbrum under unanmty rule f and only f the commttee s not mnmally dverse. Thus the crcumstances under whch unanmty rule promotes fully revealng delberaton are con ned to those n whch t s common knowledge that the commttee s homogenous wth respect to preferences over alternatves, whether or not there s bas uncertanty. A proof for ths result s n the Appendx (where we also con rm that the theorem extends to the case that the true bas pro le b 2 B s common knowledge). 4 Welfare An mportant problem s to dentfy the optmal rule for commttee choce under ncomplete nformaton wth debate. 5 Ths problem s partcularly challengng f we nsst that ndvduals votng strateges are undomnated condtonal on the realzed debate, a requrement that opens up the possblty, as llustrated by precedng theorems, that votng rules a ect the ncentves for delberaton. 6 A more lmted welfare queston concerns the extent to whch the de facto 5 Chwe (2006) derves the optmal votng rule when there s no debate. Interestngly, ths rule turns out to be nonmonotonc n votes and s thus not a q-rule. 6 Gerard and Yarv (2004) show, rst, that all non-unanmous q rules are equvalent n that the sets of sequental equlbrum outcomes nduced by use of any q rules wth q 6= 1; n are dentcal once votng s preceded by delberaton and, second, that those outcomes nduced by unanmty rule are a (not necessarly proper) 9

11 commttee decson under a gven votng rule (typcally, a q-rule) wth ncomplete nformaton and debate re ects the commttee s decson condtonal on all prvate nformaton beng common knowledge at the tme of the vote (eg McLennan 1998; Federsen and Pesendorfer 1997). Equvalently, gven the rule, the goal s to maxmze the aggregate expected payo of those n the full nformaton wnnng coalton. And the ntuton suggested by the results above s that, at least for nonunanmous q-rules, bas uncertanty can mprove the welfare of the full nformaton wnnng coalton by facltatng more nformaton sharng n debate: when ndvduals are not sure ex ante whether they are members of the full nformaton wnnng coalton, then they have an ncentve to reveal prvate nformaton n debate that s absent when they are con dent they are not members of ths coalton. Unfortunately, even restrctng attenton to q-rules, addressng the latter welfare ssue drectly s complcated by havng to descrbe the full equlbrum set to any more or less general ncomplete nformaton (debate and votng) game nduced by the rule (see for example Austen- Smth and Feddersen (2005) on ths ssue). At ths stage, no such characterzaton s avalable. 7 So rather than attempt a general result here, we nstead llustrate the ntuton above wth a smpler varant of the openng example. In the varant of the example, we assume two ndvduals n a commttee n whch one xed ndvdual has the rght to make the commttee decson. Indvduals preferences and subset of those of nduced by any non-unanmous rule. Thus any non-unanmous q rule can be chosen wthout a ectng what s possble n a pre-vote debate. However, Gerard and Yarv do not requre votng strateges to be undomnated condtonal on the realzed debate. Instead, ther result explots the fact that the rules governng debate are unconstraned and that (at least on the equlbrum path) all voters votng unanmously s consstent wth sequental ratonalty for non-unanmous q rules. Such consstency s obtaned n ther analyss ether by admttng weakly domnated strateges or, wth a mld doman restrcton, precludng domnated strateges de ned n terms of (ex ante or nterm) expectatons formed pror to any ndvdual hearng any debate. 7 Lkewse, the optmal mechansm here s as yet unavalable n general. Chwe s (2006) result for majorty votng wth ncomplete nformaton suggests such a mechansm wll be complex. 10

12 the nformatonal assumptons on sgnals s 2 fx; yg are exactly as spec ed above. However, we perturb the assumptons on the dstrbuton of bases: for each = 1; 2, assume that whle the pror probablty that b = x s 1=2, realzed bas types are correlated; that s, for dstnct ; j 2 f1; 2g Pr[b = b j jb ] = 2 [0; 1]: If = 0 they have con ctng preferences wth probablty 1 whereas f = 1 both have dentcal preferences. The dea here, s that the decson-maker s a proxy for the pvotal voter de nng the wnnng coalton under a q-rule n a larger commttee, and the other ndvdual (the advsor) s unsure whether she too s a member of that wnnng coalton. Hereafter, let ndvdual = 1 be the advsor and ndvdual = 2 be the decson-maker. We consder two communcaton protocols. In the rst protocol there s a sngle debate stage exactly as htherto, n whch both ndvduals make cheap talk speeches regardng ther sgnals, followng whch the decson-maker chooses X or Y. For the second protocol, we add an earler debate stage n whch both players can declare ther bases followng whch they both send messages regardng ther sgnals and, nally, the decson-maker makes a decson. Equlbra are agan perfect Bayesan wth undomnated strateges. Note that, because an ndvdual s payo s, gven her bas, depend exclusvely on the nal decson and the pro le of realzed sgnals s, there can be no value (n n sgnalng both bas and sgnal smultaneously or of havng a sgnal debate pror to makng any statements regardng bases. Thus the role of permttng bas revelaton early n any commttee dscusson s to mprove the possblty of coordnaton between commttee members. Begn wth assumng only a sngle debate stage n whch ndvduals make speeches about ther sgnals. Strateges are as before: : fx; yg fx; yg! fx; yg s a debate strategy wth (b; s) beng the probablty that ndvdual wth bas b and sgnal s declares "s = x"; and v : fx; yg fx; yg fx; yg 2! fx; Y g s the decson-maker s decson (vote) strategy wth v(b; s; m) beng the probablty that the decson-maker (ndvdual 2) wth bas b and sgnal 11

13 s who hears debate m = (m 1 ; m 2 ) 2 fx; yg 2 chooses X. Hereon, we economze on notaton by wrtng b;s for the probablty an advsor wth bas type b and sgnal s sends message x; and wrtng v b s(n) for the probablty the decson-maker wth bas b; sgnal s who hears message n chooses X. Proposton 1 There exsts a FRDE (1 p). If < (1 p) the advsor cannot reveal any nformaton n equlbrum and the decson-maker chooses wth hs sgnal. Proof: In a fully revealng equlbrum we have, for all, b;x = 1 b;y = 1 for every b 2 fx; yg and = 1; 2. In ths case the best response strategy for a decson-maker wth bas b who observes sgnal s 0 and hears message s from the advsor s 8 1 f (s; s >< 0 ) 2 fx; xg 8b vs b 0(s) 1 f (s; s 0 ) 2 ffx; yg; fy; xgg and b = x >: 0 otherwse The ncentve compatblty condtons for a FRDE are obvously sats ed for the decsonmaker snce he s correctly choosng as f he were fully nformed. So, we only need to check the IC constrants for the advsor. Wthout loss of generalty, assume the sender has bas b 1 = x. Then t s trvally the case (mod nversons of natural language) that she surely reveals a sgnal s 1 = x. So suppose she has observed s 1 = y. There are only two pvotal events n whch the sender s message a ects the outcome here: ether the recever shares the sender s bas (x) and has observed a sgnal s 2 = y; or the recever has the opposte bas (y) and has observed a sgnal s 2 = x. In the former event, the sender s best response s to tell the truth and reveal s 1 = y; n the latter event, the sender s best response s to le and clam s 1 = x. The probablty of these two events s, respectvely, p and (1 )(1 p). Therefore the sender s wllng to tell the truth when she has sgnal y p (1 )(1 p); 12

14 that s, (1 p). If < (1 p), then clearly the advsor has a strct best response to announce "s = b" rrespectve of her sgnal; hence no nformaton s credbly conveyed n debate (the decson-maker can always reveal hs sgnal n debate, but ths s of no consequence for the nal outcome). Hence the decson-maker must choose on the bass of hs own nformaton. Suppose, wthout loss of generalty, that b 2 = x. Then player 2 has a domnant strategy to choose X condtonal on s 2 = x. If s 2 = y, however, hs payo from choosng X s (1 p)=2 whereas that from choosng Y s p=2. Snce p > 1=2, the best response s to choose Y. It s worth emphassng that a FRDE can exst here for very small values of, dependng on just how nformatve s the sgnal, p: the more nformaton that any gven sgnal provdes, the less concern the advsor exhbts about the probablty of not beng a member of the de facto wnnng coalton. Ths result s clearly a drect analogue of the condtons de nng exstence of a FRDE for the majorty rule example wth whch we began. The analytcal advantage here, however, s that the proposton completely descrbes the equlbrum set up to the mxed equlbrum for the nongenerc event = (1 here are therefore easy to descrbe. p). The welfare mplcatons of bas uncertanty Proposton 2 Assume the most nformatve equlbrum s played for every parameterzaton (p; ). (a) If > (1 p) then, ex ante, the decson-maker (respectvely, advsor) strctly prefers bases to reman secret (respectvely, revealed) before the game s played. (b) If < (1 p) then, ex ante, both players strctly prefer bases to be revealed pror to the game beng played. Proof: Suppose that both players bases are made common knowledge pror to the game beng played. Then ether they share the same bas, n whch case the most nformatve equlbrum s a FRDE, or they have opposng bases n whch case (as s easy to con rm) no nformaton s revealed by the advsor, player 1. The payo s to the two players are therefore: 1 + p EU1 rev = EU2 rev = + (1 ) : 2 13

15 To see ths, rst note that f bases are the same, all nformaton s revealed by player 1 and both players are assured a payo of one. Next, suppose the bases are revealed to be dstnct. Then no nformaton s revealed and the decson-maker can do no better than choose accordng to hs sgnal; that s, choose X s 2 = x. In ths case the sender (player 1) receves one ether s 1 = s 2, whch occurs wth probablty p, or her bas s x (respectvely, y) and s 2 = x (respectvely, y), whch occurs wth probablty (1 p)=2. These facts yeld the expresson for EU rev 1. Smlarly, gven the bases are dstnct, the decson-maker obtans a payo equal to one ether s 1 = s 2 or her bas s x (respectvely, y) and s 2 = x (respectvely, y). Ths just es EU rev 2. Now suppose both players bases reman secret. If > (1 p) then there exsts a FRDE and the decson-maker obtans a payo EU pvt 2 = 1 surely. The advsor, however, receves payo one surely only f the bases are the same; f bases are d erent, then the advsor reveals her sgnal and receves payo one s 1 = s 2. Thus EU pvt 1 = +(1 )p. On the other hand, f < (1 p) then the only equlbrum nvolves no nformaton revelaton and the decson-maker chooses accordng to hs sgnal. Hence, followng the reasonng for the case n whch bases are revealed, we obtan EU pvt 1 = EU pvt 2 = 1 + p 2 : Therefore, f > (1 p) the decson-maker gans (1 )(1 + p)=2 when bases are secret but the advsor loses (1 )(1 p)=2. And when < (1 p), both players strctly prefer bases to be revealed. If the lkelhood that the advsor s n fact a member of the wnnng coalton n that she shares a common bas wth the decson-maker, then t the presence of bas uncertanty strctly mproves the welfare of that coalton (.e. the decson-maker) snce all decsonrelevant nformaton can be shared n equlbrum. Because the two players agree about the desrablty of bas-revelaton when s su cently low relatve to p, however, t s nterestng to ask whether provdng an opportunty to coordnate drectly by revealng bases pror to 14

16 debatng sgnals can mprove commttee performance. To address ths ssue, we turn to the second protocol descrbed earler, whereby there s a pror debate stage n whch (e ectvely) ndvduals can reveal ther bases. 8 Strateges for ths protocol are as follows. Let b;s 2 [0; 1] denote the probablty that ndvdual wth bas b and sgnal s announces her bas s x; gven bas-stage messages (m; m 0 ) 2 fx; yg 2 ; let b;s (m; m 0 ) 2 [0; 1] s the probablty ndvdual wth bas b and sgnal s who has announced her bas s m and heard a message that the other player s bas s m 0, announces that her sgnal s x; and let v b s((m; m 0 ); (n; n 0 )) be the probablty the decson-maker wth bas b and sgnal s who announces bas m and sgnal n, hears the advsor s bas message m 0 and sgnal message n 0, chooses outcome X. Proposton 3 Assume each player has observed hs or her partcular bas and sgnal. If 1=2, there exsts an equlbrum n whch each player truthfully reveals ther bas n the rst message round and (1) f the bases are the same, players truthfully reveal ther sgnal n the second round and the recever chooses on the bass of full nformaton; (2) f the bases are d erent, players smply announce ther true bas ndependently of ther sgnal and the decson-maker chooses wth hs sgnal. Proof: Consder the followng strateges (spec ed only for an x-based ndvdual; those for a y-based ndvdual are symmetrc): x;x = x;y = 1; = 1; 2 x;x (m; m 0 ) = 1; 8(m; m 0 ); = 1; 2 x;y (m; x) = 1 x;y (m; y) = 0; 8m; = 1; 2 8 Of course ndvduals can talk about anythng at any stage. However, gven messages are cheap talk, t s not hard to see that assumng that bases are the subject of the rst communcaton round and sgnals the subject of second s, at least n equlbrum, wthout loss of generalty. 15

17 v x x( m; m 0 ; n; n 0 ) = 1; 8( m; m 0 ; n; n 0 ) v x y ((m; m) ; (; x)) = v x y ((x; y) ; (x; x)) = 1; 8m v x y ( m; m 0 ; n; n 0 ) = 0 otherwse To con rm that ths strategy pro le consttutes an equlbrum f 1=2, we calculate where E[U b;s E[U x;x j x;x = 1; ] = E[U x;x E[U x;y j x;y (1 + p) = 1; ] = + (1 ) 2 E[U x;y j x;y (1 + p) = 0; ] = + (1 ) 2 j b;s j x;x (1 + p) = 0; ] = + (1 ) 2 ; ] denotes the equlbrum expected payo for ndvdual wth bas b and sgnal s from adoptng bas-debate strategy b;s. Hence tellng the truth at the bas revelaton stage s ncentve compatble E[U x;y j x;y = 1; ] E[U x;y j x;y = 0; ], whch obtans f 1=2. And gven bases are revealed truthfully, the subsequent debate and decson strateges are easly checked to be best responses. Thus allowng an opportunty to coordnate through sharng bas nformaton before revealng anythng about decson-relevant sgnals, cannot mprove the welfare propertes for the decsonmaker when < 1 p. 9 Note that n the dent ed equlbrum (say, the bas revelaton equlbrum), so long as 1=2, both ndvduals reveal ther bases n the rst round of talk; f ther bases are 9 For completeness, we note that there s an essentally dentcal equlbrum wth the three person majorty rule commttee ntroduced at the start of these Notes. Suppose n that settng too we ntroduced two rounds of debate, n whch ndvduals smultaneously send messages regardng ther bases n the rst round of talk. Then there s an equlbrum n whch all bases are revealed n the rst stage; those ndvduals n the majorty then reveal ther sgnals truthfully n the second round of debate, and the mnorty player smply announces hs bas agan; and nally, the majorty votes together for the alternatve that maxmzes ts expected welfare gven the nformaton revealed n debate. The mnorty player votes sncerely under full nformaton. Clearly, just as wth the current two player example, the de facto decson does not always re ect the full nformaton welfare maxmzng decson for the majorty coalton. 16

18 the same then all nformaton s revealed n the second round whereas, f bases are d erent, no further decson-relevant nformaton sharng takes place. It s mmedate, therefore, that the expected payo to any (b; s)-type ndvdual from playng the bas revelaton equlbrum dent ed n Proposton 3 equals the expected payo they acheve f bases could be revealed ex ante by some external agent. Spec cally, for all ; b; s, E[U b;s j bases revealed] = + (1 ) (1 + p) : 2 Furthermore, because p > 1=2, Propostons 1 and 3 together mply there s also a FRDE wth a bas debate stage when 1 p: both ndvduals babble durng the rst round of talk and, subsequently, all decson-relevant nformaton s revealed durng the second round of communcaton exactly as descrbed n Proposton 1; that s, for all ; b; s, b;s = 1=2 and b;x = 1 b;y = 1. Hence we have the same welfare comparsons as descrbed n Proposton 2, wth the further observaton that, when 1=2 > 1 p, the advsor strctly prefers to play the bas revelaton equlbrum to playng the FRDE wth no bas revelaton (of course, the decson-maker has the opposte preferences). In sum, bas uncertanty can mprove the welfare of the de facto full nformaton wnnng coalton. And even n those crcumstances n whch bas revelaton s avalable n equlbrum, there exsts a superor (from the decson-maker s perspectve) equlbrum n whch bases reman prvate nformaton. 5 Concluson There s clearly a great deal left to be learned regardng communcaton n commttees. In partcular, dentfyng the optmal votng rule n the presence of debate s an mportant and open ssue. 17

19 Appendx For all b 2 B, let T 0 (b) S(b) and, for any k = 1; 2; : : :, recursvely de ne the sets T k (b) = fs =2 [ l=k l=1 Tk l (b)j9s; s 0 2 S : s 0 s, (s ; s) = s; (s ; s 0 ) = s 0 and s 0 2 T k 1 (b)g: Thus T 1 (b) s the set of states not n T 0 (b) such that, gven the realzed bas pro le b, changng any one person s nformaton from s to s 0 results n a state n T 0 (b) S(b); T 2 (b) s the set of states not n T 1 (b) such that changng any one person s nformaton from s to s 0 results n a state n ; and so on. Informally, the set T k (b) s the set of states such that there s a path of k sngle coordnate changes of nformaton that lead to a state at whch y s preferred unanmously. Snce S and N are nte t follows that [ k=0;1;:::;n T k (b) = S: For example, suppose n = 3, S = f0; 1g 3 and S(b) = f(1; 1; 1)g. Then T 0 (b) = f(1; 1; 1)g T 1 (b) = f(0; 1; 1); (1; 0; 1); (1; 1; 0)g T 2 (b) = f(1; 0; 0); (0; 1; 0); (0; 0; 1)g T 3 (b) = f(0; 0; 0)g: The followng property of mnmally dverse commttees n envronments satsfyng the three axoms s useful for provng the man theorem. The lemma nsures that n mnmally dverse commttees, there must exst two bas types and a state such that, rst, the two bas types have strctly opposng preferences at the state and, second, that the state d ers n only one component from another state at whch all bas types strctly prefer y to x. Lemma Assume full support, consensus and monotoncty. In a mnmally dverse commttee there exsts a bas pro le b = (b ; b; b 0 ) 2 B and a state s 2 T 1 (b) such that s =2 S b but s 2 S b 0: 18

20 Proof Let b = (b ; b; b 0 ) 2 B (where, by an abuse of notaton, t s understood that b 2 B n 2 ); by consensus, S(b) 6= ;. Frst assume there s a state s 2 S b \T k+1 (b). By full support and de nton of T k (b), there exsts a sgnal s 0 s such that (s ; s 0 ) = s 0 2 T k (b); moreover, by monotoncty, s 0 2 S b. Hence, s 2 S b \ T k+1 (b) mples there exsts a state s 0 2 S b \ T k (b). Now suppose b s such that, for any s 2 T k (b), s =2 S b. Then by the prevous argument, there can be no s 2 T k+1 (b) such that s 2 S b. Hence, S b \ T 1 (b) = ; mples S b \ T k (b) = ; for all k > 1 n whch case, because [ k=0;1;:::;n T k (b) = S, t must be that S b = S(b). It follows that f, contrary to the lemma, for all b 2 B there exsts no s 2 T 1 (b) and components b; b 0 of b such that s =2 S b but s 2 S b 0, then S b = S(b) for all components of b, volatng mnmal dversty. Proof of Theorem 2 (Necessty) In any fully revealng debate equlbrum, the restrcton to weakly undomnated votng strateges mples v (b; s; m) = y f and only f (s ; s) 2 S b, where s = m for every 2 N and b 2 B. It follows that a member s votng strategy does not depend on the message she sends n debate. Consder the delberaton stage and, by way of contradcton, suppose s fully revealng yet the commttee s mnmally dverse. Then, gven the behavor at the votng stage, fully revealng message strateges consttute an equlbrum f and only f, for every 2 N and every (b ; s ) 2 B S, t s the case that EU(m = s ; b ; s ) EU (m = s 0 ; b ; s ) 0 for any s 0 2 Mnfs g (1) where EU(m ; b ; s ) = X X b 2B n 1 s 2S n 1 p(b ; s jb ; s ) [Pr(xjb; s; m )u(x; b ; s) + Pr(yjb; s; m )u(y; b ; s)] and Pr(zjb; s; m ) s the probablty that z 2 fx; yg s the commttee decson gven bas pro le b = (b ; b ), state s = (s ; s ) and debate (m ; m ) = (s ; m ). (b ; s ) = (b; s); for any s 0 2 Mnfsg, de ne the functon Fx 2 N and let ' (b;s) (s; s 0 ; b ; s ) Pr(xjb; s; s) Pr(xjb; s; s 0 ) [u(x; b; s) u(y; b; s)] 19

21 wth b = (b ; b) and s = (s ; s). Then we can rewrte (1) equvalently as requrng that for all (b; s) 2 B S and all s 0 2 Mnfsg, X X b 2B n 1 s 2S n 1 p(b ; s jb; s)' (b;s) (s; s 0 ; b ; s ) 0: (2) By assumpton, s fully revealng of all others sgnals and, by the precedng argument on v, for all messages m 2 M and all bas pro les (b ; b) 2 B, (s ; s) 2 SnS b mples Pr(xj(b ; b); (s ; s); m ) = 1. Smlarly, for any state (s ; s) 2 Sn(S(b) [ T 1 (b)) t must be that Pr(xj(b ; b); (s ; s); m ) = 1. Gven (b ; s ) = (b; s), therefore, for all s 0 2 Mnfsg and all b 2 B n 1, (s ; s) 2 Sn S(b) [ T 1 (b) [ S b ) '(b;s) (s; s 0 ; b ; s ) = 0: (3) The precedng argument mples that an ndvdual wth bas b can change the outcome by swtchng from message s to some s 0 6= s only n stuatons (b; s) such that (s ; s) 2 S(b) [ T 1 (b) \ S b. For all b 2 B, de ne x (b; s; s 0 ) = f(s ; s) 2 S(b)j(s ; s 0 ) =2 S(b)g to be the set of states such that f an ndvdual who s supposed to report s nstead reports s 0 then, condtonal on b, the outcome changes from y to x. Smlarly, de ne y (b; s; s 0 ) = f(s ; s) 2 T 1 (b) \ S b j(s ; s 0 ) 2 S(b)g to be the set of states n whch prefers y and, f s supposed to report s but nstead reports s 0 at b, the outcome changes from x to y. Note that, by monotoncty, f y (b; s; s 0 ) 6= ; for some b 2 B, then x (b; s; s 0 ) = ; for all b 2 B and, f x (b; s; s 0 ) 6= ; for some b 2 B, then y (b; s; s 0 ) = ; for all b 2 B: That s, y (b; s; s 0 ) 6= ; for some b 2 B mples that s 0 s stronger evdence for y than s, whereas x (b; s; s 0 ) 6= ; for some b 2 B mples s 0 s weaker evdence for y than s. By monotoncty both statements cannot be true. For any b 2 B and s; s 0 2 S, let Z (b; s; s 0 ) fs 2 S n 1 j(s ; s) 2 y (b; s; s 0 ) [ x (b; s; s 0 ) g: 20

22 Collectng terms and usng (3), we can rewrte the ncentve compatblty constrant (2) as requrng, for all 2 N, (b; s) 2 B S and s 0 2 Mnfsg, X b 2B X s 2Z (b;s;s 0 ) p(b ; s ; jb; s)' (b;s) (s; s 0 ; b ; s ) 0: By the Lemma and full support, mnmal dversty mples there s a (b ; b) 2 B and a par of sgnals s; s 0 2 S such that y ((b ; b); s; s 0 ) 6= ; and x ((b ; b); s; s 0 ) = ;. By de nton, (s ; s) 2 y ((b ; b); s; s 0 ) mples u(x; b; (s ; s)) < u(y; b; (s ; s)) and Pr(xj(b ; b); s; s) Pr(xj(b ; b); s; s 0 ) = 1. Hence, for all (b ; b) 2 B, s 2 Z (b; s; s 0 ) ) ' (b;s) (s; s 0 ; b ; s ) < 0: But then the ncentve compatblty condtons are surely volated, contradctng the exstence of a fully revealng debate equlbrum n any mnmally dverse commttee. Ths proves necessty. (Su cency) Assume the commttee s not mnmally dverse. Then for all b 2 B and all b = (b ; b) 2 B, S b = S(b). In ths case there s no b 2 B and par of sgnals s; s 0 2 S such that y (b; s; s 0 ) 6= ; for any 2 N. Snce ncentve compatblty s assured for any 2 N, b 2 B and par of sgnals s; s 0 2 S such that x (b; s; s 0 ) 6= ; and y (b; s; s 0 ) = ;, full revelaton s an equlbrum strategy. Ths completes the proof. Fnally, to see that the theorem goes through under complete nformaton regardng ndvduals bases, x a bas pro le b = (b 1 ; : : : ; b n ), suppose b s common knowledge and let B = fbg. Then the de ntons and the argument drectly apply on replacng references to bases b; b 0 2 B wth references to ndvduals ; j 2 N wth bases b ; b j, and so on. 21

23 References Austen-Smth, Davd. 1990a. Informaton transmsson n debate. Amercan Journal of Poltcal Scence 34 (February): Austen-Smth, Davd and Je rey S. Banks Informaton aggregaton, ratonalty, and the Condorcet Jury Theorem. Amercan Poltcal Scence Revew 90 (March): Austen-Smth, Davd and Tmothy J. Feddersen Delberaton and votng rules. In Socal Choce and Strategc Decsons: Essays n Honor of Je rey S. Banks, eds. Davd Austen-Smth and John Duggan. Hedelberg:Sprnger. Austen-Smth, Davd and Tmothy J. Feddersen Delberaton, preference uncertanty and votng rules. Amercan Poltcal Scence Revew 100 (May): Callaud, Bernard and Jean Trole "Consensus buldng: how to persuade a group" CNRS Workng Paper. Calvert, Randall L. and James Johnson Ratonal argument, poltcal argument and democratc delberaton. Presented at the Annual Meetng of the Amercan Poltcal Scence Assocaton. Chwe, Mchael "A robust and optmal anonymous procedure for Condorcet s model." UCLA, Workng Paper. Coughlan, Peter J In defense of unanmous jury verdcts: mstrals, communcaton and strategc votng. Amercan Poltcal Scence Revew 94 (June): Doraszelsk, Ulrch, Dno Gerard and Francesco Squntan Communcaton and votng wth double-sded nformaton. Contrbutons to Theoretcal Economcs 3(1), Artcle 6. Feddersen, Tmothy J. and Wolfgang Pesendorfer Votng behavor and nformaton aggregaton n large electons wth prvate nformaton. Econometrca 65(5):

24 Feddersen, Tmothy J. and Wolfgang Pesendorfer Convctng the nnoncent: the nferorty of unanmous jury verdcts. Amercan Poltcal Scence Revew 92 (March): Gerard, Dno and Leeat Yarv Puttng your ballot where your mouth s: an analyss of collectve choce wth communcaton. Yale Unversty. Typescrpt. Glazer, Jacob and Arel Rubnsten Debates and decsons: on a ratonale of argumentaton rules. Games and Economc Behavor 36(2): Hafer, Catherne and Dmtr Landa Delberaton as self-dscovery and the nsttutons for poltcal speech. forthcomng n Journal of Theoretcal Poltcs. McLennan, Andrew Consequences of the Condorcet Jury Theorem for bene cal nformaton aggregaton by ratonal players. Amercan Poltcal Scence Revew. 92(2): Merowtz, Adam In defense of exclusonary delberaton: communcaton and votng wth prvate belefs and values forthcomng n Journal of Theoretcal Poltcs. Merowtz, Adam Delberatve democracy or market democracy: desgnng nsttutons to aggregate preferences and nformaton. Quarterly Journal of Poltcal Scence 1(4): Ottavan, Marco and Peter Sorensen Informaton aggregaton n debate: who should speak rst? Journal of Publc Economcs 81(3):

A Note on Preference Uncertainty and Communication in Committees 1

A Note on Preference Uncertainty and Communication in Committees 1 A Note on Preference Uncertanty and Communcaton n Commttees 1 Davd Austen-Smth MEDS, Kellogg Graduate School of Management Northwestern Unversty Evanston, IL 60208 Tm Feddersen MEDS, Kellogg Graduate School

More information

A Note on Preference Uncertainty and Communication in Committees 1

A Note on Preference Uncertainty and Communication in Committees 1 A Note on Preference Uncertanty and Communcaton n Commttees 1 Davd Austen-Smth MEDS, Kellogg Graduate School of Management Northwestern Unversty Evanston, IL 60208 Tm Feddersen MEDS, Kellogg Graduate School

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

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

Correlated Equilibrium in Games with Incomplete Information

Correlated Equilibrium in Games with Incomplete Information Correlated Equlbrum n Games wth Incomplete Informaton Drk Bergemann y Stephen Morrs z Frst Verson: October Current Verson: May 4, Abstract We de ne a noton of correlated equlbrum for games wth ncomplete

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

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

The basic point with mechanism design is that it allows a distinction between the underlying

The basic point with mechanism design is that it allows a distinction between the underlying 14 Mechansm Desgn The basc pont wth mechansm desgn s that t allows a dstncton between the underlyng economc envronment and the rules of the game. We wll take as gven some set of possble outcomes (alternatves,

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

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

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

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

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

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

Cowles Foundation for Research in Economics at Yale University

Cowles Foundation for Research in Economics at Yale University Cowles Foundaton for Research n Economcs at Yale Unversty Cowles Foundaton Dscusson Paper No. 666 ROBUST IMPLEMENTATION IN GENERAL MECHANISMS Drk Bergemann and Stephen Morrs Month 28 An author ndex to

More information

Internet Appendix for "Nonbinding Voting for Shareholder Proposals"

Internet Appendix for Nonbinding Voting for Shareholder Proposals Internet Appendx for "Nonbndng Votng for Shareholder Proposals" DORON LEVIT and NADYA MALENKO The Internet Appendx contans the omtted proofs from the man appendx (Secton I) and the analyss correspondng

More information

RATIONALIZABLE IMPLEMENTATION. Dirk Bergemann and Stephen Morris. May 2009 COWLES FOUNDATION DISCUSSION PAPER NO. 1697

RATIONALIZABLE IMPLEMENTATION. Dirk Bergemann and Stephen Morris. May 2009 COWLES FOUNDATION DISCUSSION PAPER NO. 1697 RATIONALIZABLE IMPLEMENTATION By Drk Bergemann and Stephen Morrs May 2009 COWLES FOUNDATION DISCUSSION PAPER NO. 1697 COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY Box 208281 New Haven, Connectcut

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 23 Ths Verson: June 25 Abstract A socal choce functon s robustly mplemented f every equlbrum on every

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

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

Implementation in Mixed Nash Equilibrium

Implementation in Mixed Nash Equilibrium Department of Economcs Workng Paper Seres Implementaton n Mxed Nash Equlbrum Claudo Mezzett & Ludovc Renou May 2012 Research Paper Number 1146 ISSN: 0819-2642 ISBN: 978 0 7340 4496 9 Department of Economcs

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

A Resurrection of the Condorcet Jury Theorem

A Resurrection of the Condorcet Jury Theorem A Resurrecton of the Condorcet Jury Theorem Yuo Koryama and Balázs Szentes yz Septemer 19, 2008 Astract hal-00391197, verson 1-3 Jun 2009 Ths paper analyzes the optmal sze of a deleratng commttee where,

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

A resurrection of the Condorcet Jury Theorem

A resurrection of the Condorcet Jury Theorem Theoretcal Economcs 4 (2009), 227 252 1555-7561/20090227 A resurrecton of the Condorcet Jury Theorem YUKIO KORIYAMA Département d Économe, École Polytechnque, Palaseau BALÁZS SZENTES Department of Economcs,

More information

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal Inner Product Defnton 1 () A Eucldean space s a fnte-dmensonal vector space over the reals R, wth an nner product,. Defnton 2 (Inner Product) An nner product, on a real vector space X s a symmetrc, blnear,

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

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

(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

A Resurrection of the Condorcet Jury Theorem

A Resurrection of the Condorcet Jury Theorem A Resurrecton of the Condorcet Jury Theorem Yuo Koryama Balázs Szentes Department of Economcs, Unversty of Chcago June 13, 2007 Abstract Ths paper analyzes the optmal sze of a delberatng commttee where,

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

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

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

and problem sheet 2

and problem sheet 2 -8 and 5-5 problem sheet Solutons to the followng seven exercses and optonal bonus problem are to be submtted through gradescope by :0PM on Wednesday th September 08. There are also some practce problems,

More information

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

THE CHINESE REMAINDER THEOREM. We should thank the Chinese for their wonderful remainder theorem. Glenn Stevens THE CHINESE REMAINDER THEOREM KEITH CONRAD We should thank the Chnese for ther wonderful remander theorem. Glenn Stevens 1. Introducton The Chnese remander theorem says we can unquely solve any par of

More information

Microeconomics: Auctions

Microeconomics: Auctions Mcroeconomcs: Auctons Frédérc Robert-coud ovember 8, Abstract We rst characterze the PBE n a smple rst prce and second prce sealed bd aucton wth prvate values. The key result s that the expected revenue

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

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

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

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

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

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

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

Folk Theorem in Stotchastic Games with Private State and Private Monitoring Preliminary: Please do not circulate without permission

Folk Theorem in Stotchastic Games with Private State and Private Monitoring Preliminary: Please do not circulate without permission Folk Theorem n Stotchastc Games wth Prvate State and Prvate Montorng Prelmnary: Please do not crculate wthout permsson Takuo Sugaya Stanford Graduate School of Busness December 9, 202 Abstract We show

More information

Mechanisms with Evidence: Commitment and Robustness 1

Mechanisms with Evidence: Commitment and Robustness 1 Mechansms wth Evdence: Commtment and Robustness 1 Elchanan Ben-Porath 2 Edde Dekel 3 Barton L. Lpman 4 Frst Draft January 2017 1 We thank the Natonal Scence Foundaton, grant SES 0820333 (Dekel), and the

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

Commitment and Robustness in Mechanisms with Evidence 1

Commitment and Robustness in Mechanisms with Evidence 1 Commtment and Robustness n Mechansms wth Evdence 1 Elchanan Ben-Porath 2 Edde Dekel 3 Barton L. Lpman 4 Frst Draft June 2016 1 We thank the Natonal Scence Foundaton, grant SES 0820333 (Dekel), and the

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

DIFFERENTIAL SCHEMES

DIFFERENTIAL SCHEMES DIFFERENTIAL SCHEMES RAYMOND T. HOOBLER Dedcated to the memory o Jerry Kovacc 1. schemes All rngs contan Q and are commutatve. We x a d erental rng A throughout ths secton. 1.1. The topologcal space. Let

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

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

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

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

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

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U) Econ 413 Exam 13 H ANSWERS Settet er nndelt 9 deloppgaver, A,B,C, som alle anbefales å telle lkt for å gøre det ltt lettere å stå. Svar er gtt . Unfortunately, there s a prntng error n the hnt of

More information

Linear Approximation with Regularization and Moving Least Squares

Linear Approximation with Regularization and Moving Least Squares Lnear Approxmaton wth Regularzaton and Movng Least Squares Igor Grešovn May 007 Revson 4.6 (Revson : March 004). 5 4 3 0.5 3 3.5 4 Contents: Lnear Fttng...4. Weghted Least Squares n Functon Approxmaton...

More information

Information Acquisition in Global Games of Regime Change

Information Acquisition in Global Games of Regime Change Informaton Acquston n Global Games of Regme Change Mchal Szkup and Isabel Trevno y Abstract We study costly nformaton acquston n global games of regme change (that s, coordnaton games where payo s are

More information

The Folk Theorem for Games with Private Almost-Perfect Monitoring

The Folk Theorem for Games with Private Almost-Perfect Monitoring The Folk Theorem for Games wth Prvate Almost-Perfect Montorng Johannes Hörner y Wojcech Olszewsk z October 2005 Abstract We prove the folk theorem for dscounted repeated games under prvate, almost-perfect

More information

Communication and Cooperation in Repeated Games

Communication and Cooperation in Repeated Games Communcaton and Cooperaton n Repeated Games Yu Awaya y and Vjay Krshna z May 4, 07 Abstract We study the role of communcaton n repeated games wth prvate montorng. We rst show that wthout communcaton, the

More information

Lecture 4. Instructor: Haipeng Luo

Lecture 4. Instructor: Haipeng Luo Lecture 4 Instructor: Hapeng Luo In the followng lectures, we focus on the expert problem and study more adaptve algorthms. Although Hedge s proven to be worst-case optmal, one may wonder how well t would

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

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

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

Basically, if you have a dummy dependent variable you will be estimating a probability.

Basically, if you have a dummy dependent variable you will be estimating a probability. ECON 497: Lecture Notes 13 Page 1 of 1 Metropoltan State Unversty ECON 497: Research and Forecastng Lecture Notes 13 Dummy Dependent Varable Technques Studenmund Chapter 13 Bascally, f you have a dummy

More information

Introductory Cardinality Theory Alan Kaylor Cline

Introductory Cardinality Theory Alan Kaylor Cline Introductory Cardnalty Theory lan Kaylor Clne lthough by name the theory of set cardnalty may seem to be an offshoot of combnatorcs, the central nterest s actually nfnte sets. Combnatorcs deals wth fnte

More information

Bayesian predictive Configural Frequency Analysis

Bayesian predictive Configural Frequency Analysis Psychologcal Test and Assessment Modelng, Volume 54, 2012 (3), 285-292 Bayesan predctve Confgural Frequency Analyss Eduardo Gutérrez-Peña 1 Abstract Confgural Frequency Analyss s a method for cell-wse

More information

The Mixed Strategy Nash Equilibrium of the Television News Scheduling Game

The Mixed Strategy Nash Equilibrium of the Television News Scheduling Game The Mxed Strategy Nash Equlbrum of the Televson News Schedulng Game Jean Gabszewcz Dder Laussel Mchel Le Breton July 007 Abstract We characterze the unque mxed-strategy equlbrum of an extenson of the "televson

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

Continuous Implementation

Continuous Implementation Contnuous Implementaton Maron Oury HEC Pars Olver Terceux Pars School of Economcs and CNRS Abstract It s well-known that mechansm desgn lterature makes many smplfyng nformatonal assumptons n partcular

More information

Optimal Allocation with Costly Verification 1

Optimal Allocation with Costly Verification 1 Optmal Allocaton wth Costly Verfcaton 1 Elchanan Ben-Porath 2 Edde Dekel 3 Barton L. Lpman 4 Frst Draft August 2012 1 We thank Rcky Vohra and numerous semnar audences for helpful comments. We also thank

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

Mechanisms with Evidence: Commitment and Robustness 1

Mechanisms with Evidence: Commitment and Robustness 1 Mechansms wth Evdence: Commtment and Robustness 1 Elchanan Ben-Porath 2 Edde Dekel 3 Barton L. Lpman 4 Frst Draft January 2017 Current Draft July 2018 1 We thank numerous semnar audences and Joel Sobel

More information

Voting Games with Positive Weights and. Dummy Players: Facts and Theory

Voting Games with Positive Weights and. Dummy Players: Facts and Theory Appled Mathematcal Scences, Vol 10, 2016, no 53, 2637-2646 HIKARI Ltd, wwwm-hkarcom http://dxdoorg/1012988/ams201667209 Votng Games wth Postve Weghts and Dummy Players: Facts and Theory Zdravko Dmtrov

More information

Foundations of Arithmetic

Foundations of Arithmetic Foundatons of Arthmetc Notaton We shall denote the sum and product of numbers n the usual notaton as a 2 + a 2 + a 3 + + a = a, a 1 a 2 a 3 a = a The notaton a b means a dvdes b,.e. ac = b where c s an

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

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

Understanding Reasoning Using Utility Proportional Beliefs

Understanding Reasoning Using Utility Proportional Beliefs Understandng Reasonng Usng Utlty Proportonal Belefs Chrstan Nauerz EpCenter, Maastrcht Unversty c.nauerz@maastrchtunversty.nl Abstract. Tradtonally very lttle attenton has been pad to the reasonng process

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

A new construction of 3-separable matrices via an improved decoding of Macula s construction

A new construction of 3-separable matrices via an improved decoding of Macula s construction Dscrete Optmzaton 5 008 700 704 Contents lsts avalable at ScenceDrect Dscrete Optmzaton journal homepage: wwwelsevercom/locate/dsopt A new constructon of 3-separable matrces va an mproved decodng of Macula

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

An Introduction to Mechanism Design

An Introduction to Mechanism Design An Introducton to Mechansm Desgn Felx Munoz-Garca School of Economc Scences Washngton State Unversty 1 1 Introducton In ths chapter, we consder stuatons n whch some central authorty wshes to mplement a

More information

Genericity of Critical Types

Genericity of Critical Types Genercty of Crtcal Types Y-Chun Chen Alfredo D Tllo Eduardo Fangold Syang Xong September 2008 Abstract Ely and Pesk 2008 offers an nsghtful characterzaton of crtcal types: a type s crtcal f and only f

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

Ph 219a/CS 219a. Exercises Due: Wednesday 23 October 2013

Ph 219a/CS 219a. Exercises Due: Wednesday 23 October 2013 1 Ph 219a/CS 219a Exercses Due: Wednesday 23 October 2013 1.1 How far apart are two quantum states? Consder two quantum states descrbed by densty operators ρ and ρ n an N-dmensonal Hlbert space, and consder

More information

COS 511: Theoretical Machine Learning. Lecturer: Rob Schapire Lecture # 15 Scribe: Jieming Mao April 1, 2013

COS 511: Theoretical Machine Learning. Lecturer: Rob Schapire Lecture # 15 Scribe: Jieming Mao April 1, 2013 COS 511: heoretcal Machne Learnng Lecturer: Rob Schapre Lecture # 15 Scrbe: Jemng Mao Aprl 1, 013 1 Bref revew 1.1 Learnng wth expert advce Last tme, we started to talk about learnng wth expert advce.

More information

Stat260: Bayesian Modeling and Inference Lecture Date: February 22, Reference Priors

Stat260: Bayesian Modeling and Inference Lecture Date: February 22, Reference Priors Stat60: Bayesan Modelng and Inference Lecture Date: February, 00 Reference Prors Lecturer: Mchael I. Jordan Scrbe: Steven Troxler and Wayne Lee In ths lecture, we assume that θ R; n hgher-dmensons, reference

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

C/CS/Phy191 Problem Set 3 Solutions Out: Oct 1, 2008., where ( 00. ), so the overall state of the system is ) ( ( ( ( 00 ± 11 ), Φ ± = 1

C/CS/Phy191 Problem Set 3 Solutions Out: Oct 1, 2008., where ( 00. ), so the overall state of the system is ) ( ( ( ( 00 ± 11 ), Φ ± = 1 C/CS/Phy9 Problem Set 3 Solutons Out: Oct, 8 Suppose you have two qubts n some arbtrary entangled state ψ You apply the teleportaton protocol to each of the qubts separately What s the resultng state obtaned

More information

A note on the one-deviation property in extensive form games

A note on the one-deviation property in extensive form games Games and Economc Behavor 40 (2002) 322 338 www.academcpress.com Note A note on the one-devaton property n extensve form games Andrés Perea Departamento de Economía, Unversdad Carlos III de Madrd, Calle

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

Lecture Space-Bounded Derandomization

Lecture Space-Bounded Derandomization Notes on Complexty Theory Last updated: October, 2008 Jonathan Katz Lecture Space-Bounded Derandomzaton 1 Space-Bounded Derandomzaton We now dscuss derandomzaton of space-bounded algorthms. Here non-trval

More information

Graph Reconstruction by Permutations

Graph Reconstruction by Permutations Graph Reconstructon by Permutatons Perre Ille and Wllam Kocay* Insttut de Mathémathques de Lumny CNRS UMR 6206 163 avenue de Lumny, Case 907 13288 Marselle Cedex 9, France e-mal: lle@ml.unv-mrs.fr Computer

More information

Interim Correlated Rationalizability 1

Interim Correlated Rationalizability 1 Interm Correlated Ratonalzablty Edde Dekel Northwestern Unversty and Tel Avv Unversty Drew Fudenberg Harvard Unversty Stephen Morrs Prnceton Unversty Frst Draft: May 2003. Ths Draft: November 2006 Ths

More information

Outline. Bayesian Networks: Maximum Likelihood Estimation and Tree Structure Learning. Our Model and Data. Outline

Outline. Bayesian Networks: Maximum Likelihood Estimation and Tree Structure Learning. Our Model and Data. Outline Outlne Bayesan Networks: Maxmum Lkelhood Estmaton and Tree Structure Learnng Huzhen Yu janey.yu@cs.helsnk.f Dept. Computer Scence, Unv. of Helsnk Probablstc Models, Sprng, 200 Notces: I corrected a number

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 12 10/21/2013. Martingale Concentration Inequalities and Applications

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 12 10/21/2013. Martingale Concentration Inequalities and Applications MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.65/15.070J Fall 013 Lecture 1 10/1/013 Martngale Concentraton Inequaltes and Applcatons Content. 1. Exponental concentraton for martngales wth bounded ncrements.

More information

Appendix B. Criterion of Riemann-Stieltjes Integrability

Appendix B. Criterion of Riemann-Stieltjes Integrability Appendx B. Crteron of Remann-Steltes Integrablty Ths note s complementary to [R, Ch. 6] and [T, Sec. 3.5]. The man result of ths note s Theorem B.3, whch provdes the necessary and suffcent condtons for

More information

20. Mon, Oct. 13 What we have done so far corresponds roughly to Chapters 2 & 3 of Lee. Now we turn to Chapter 4. The first idea is connectedness.

20. Mon, Oct. 13 What we have done so far corresponds roughly to Chapters 2 & 3 of Lee. Now we turn to Chapter 4. The first idea is connectedness. 20. Mon, Oct. 13 What we have done so far corresponds roughly to Chapters 2 & 3 of Lee. Now we turn to Chapter 4. The frst dea s connectedness. Essentally, we want to say that a space cannot be decomposed

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

Explaining the Stein Paradox

Explaining the Stein Paradox Explanng the Sten Paradox Kwong Hu Yung 1999/06/10 Abstract Ths report offers several ratonale for the Sten paradox. Sectons 1 and defnes the multvarate normal mean estmaton problem and ntroduces Sten

More information

EPR Paradox and the Physical Meaning of an Experiment in Quantum Mechanics. Vesselin C. Noninski

EPR Paradox and the Physical Meaning of an Experiment in Quantum Mechanics. Vesselin C. Noninski EPR Paradox and the Physcal Meanng of an Experment n Quantum Mechancs Vesseln C Nonnsk vesselnnonnsk@verzonnet Abstract It s shown that there s one purely determnstc outcome when measurement s made on

More information

Polynomials. 1 More properties of polynomials

Polynomials. 1 More properties of polynomials Polynomals 1 More propertes of polynomals Recall that, for R a commutatve rng wth unty (as wth all rngs n ths course unless otherwse noted), we defne R[x] to be the set of expressons n =0 a x, where a

More information