A Resurrection of the Condorcet Jury Theorem
|
|
- Constance Jennings
- 6 years ago
- Views:
Transcription
1 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, () there s no conflct of nterest among ndvduals, and () nformaton acquston s costly. The commttee members smultaneously decde whether or not to acqure nformaton, and then, they mae the ex-post effcent decson. The optmal commttee sze,, s bounded. The man result of ths paper s that any arbtrarly large commttee can aggregate the decentralzed nformaton more effcently than the commttee of sze 2. Ths result suggests that havng larger commttees generates relatvely small neffcences. 1 Introducton The classcal Condorcet Jury Theorem (CJT) states that large commttees can aggregate decentralzed nformaton more effcently than small ones. Its orgn ascends to the dawn of the French Revoluton when Mare Jean Antone Ncolas Cartat le Marqus de Condorcet [1785, translaton 1994] nvestgated decson-mang processes n socetes. 1 Recently, a growng lterature on commttee desgn ponted out that f the nformaton acquston s costly, the CJT fals to hold. The reason s that f the sze of a commttee s large, a commttee member realzes that the probablty that she can nfluence the fnal decson s too small compared to the cost of nformaton acquston. As a result, she mght prefer to save ths cost and free-rde on the nformaton of others. Therefore, too large commttees mght generate lower socal welfare than smaller ones. These results suggest that, n the presence of costly nformaton acquston, optmally choosng the sze of a commttee s an mportant and delcate ssue. On the one hand, we also dentfy a welfare loss assocated to overszed commttees. On the other hand, we show that ths loss s surprsngly small n certan envronments E. 59th Sreet, Chcago, IL The authors are grateful to the semnar partcpants at worshop at Unversty of Chcago for helpful comments. 1 Summares of the hstory of the CJT can be found n, for example, Grofman and Owen [1986], Mller [1986], or Gerlng, Grüner, Kel, and Schulte [2003]. 1
2 The reason why commttee desgn receves a consderable attenton by economsts s that, n many stuatons, groups rather than ndvduals mae decsons. Informaton about the desrablty of the possble decsons s often decentralzed: Indvdual group members must separately acqure costly nformaton about the alternatves. A classcal example s a tral jury where a jury has to decde whether a defendant s gulty or nnocent. Each juror may ndvdually obtan some nformaton about the defendant at some effort cost (payng attenton to the tral, nvestgatng evdences etc.). Another example s a frm facng a decson whether to mplement a project. Each member of the executve commttee can collect nformaton about the proftablty of the project (by spendng tme and exertng effort). Yet another example s the hrng decsons of academc departments. Each member of the recrutng commttee s supposed to revew the applcatons before mang a collectve decson. What s common n these examples s that nformaton acquston s costly and often unobservable. The specfc setup analyzed n ths paper s the followng. A group of ndvduals has to mae a bnary decson. There s no conflct of nterest among the group members, but they have mperfect nformaton about whch decson s the best. Frst, ndvduals are ased to serve n a commttee. Then, the commttee members smultaneously decde whether or not to nvest nto an nformatve sgnal. Fnally, the optmal decson s made gven the acqured nformaton. We do not explctly model how the commttee members communcate and aggregate nformaton. Instead, we just assume that they end up mang the ex-post effcent decson. 2 The only strategc choce an ndvdual has to mae n our model s whether or not to acqure a sgnal upon beng selected to serve n the commttee. The central queston of our paper s: How does the commttee sze,, affect socal welfare? Frst, for each, we fully characterze the set of equlbra (ncludng asymmetrc and mxedstrategy ones). We show that there exsts a p such that whenever p, there s a unque equlbrum n whch each commttee member nvests nto nformaton wth probablty one. In addton, the socal welfare generated by these equlbra s an ncreasng functon of. If> p, then there are multple equlbra, and many of them nvolve randomzatons by the members. We also show that the socal welfare generated by the worst equlbrum n the game where the commttee sze s s decreasng n. The optmal commttee sze,,sdefned such that () f thecommtteeszes, then there exsts an equlbrum that maxmzes socal welfare, and () n ths equlbrum, each member nvests nto nformaton wth postve probablty. It turns out that the optmal commttee sze,,sether p or p +1. Ths mples that the CJT fals to hold: large commttees can generate smaller socal welfare than smaller commttees. Nevertheless, we show that f the sze of the commttee s larger than, even the worst equlbrum generates hgher socal welfare than the unque equlbrum n the commttee of sze 2. That s, the 2 Snce there s no conflct of nterest among the ndvduals, t s easy to desgn a mechansm whch s ncentve compatble and effcently aggregates the sgnals. Alternatvely, one can assume that the collected nformaton s hard. 2
3 welfare loss due to an overszed commttee s qute small. Lterature Revew The Condorcet s Jury Theorem provdes an mportant support for the bass of democracy. In hs orgnal wor, Condorcet assumed nave (sncere) votng, that s, each voter behaves as f he or she was the only decson maer n the socety. Ths s not compatble wth ratonal behavor n general, because nave votng mght be dfferent from the best response gven nave votng by the others. Austen-Smth and Bans [1996] showed that ratonal ndvduals do not necessarly tae the same acton n the collectve decson mang process as n the ndvdual one. Ths s because a voter s ratonal choce s based on the condton n whch her vote s pvotal. Feddersen and Pesendorfer [1998] showed that t mght be even wrong to follow the ntuton accompaned wth the nave votng. As the jury sze ncreases, the probablty of convctng an nnocent ncreases when the voters tae ratonal actons under the unanmty rule, because beng pvotal s a strong sgn of the suspect to be gulty. A strong mplcaton from these papers s that an approprate assumpton should be a strategc votng rather than nave one. Under the assumpton of strategc votng, a varety of papers addressed the queston of asymptotc effcency of votng as an nformaton aggregaton devce. Feddersen and Pesendorfer [1997] showed that prvate nformaton s effcently aggregated under strategc votng as the commttee sze goes to nfnty. Intuton s that the belef over states converges to the dstrbuton where the result of votes s almost always ted condtonal on beng pvotal, thus the wnner s elected by those who tae nformatve acton. Hence the each voter has an ncentve to vote nformatvely. Myerson [1998] showed that the asymptotc effcency holds n Posson Games where there s populaton uncertanty. However, Condorcet Jury Theorem mght fal f the nformaton acquston s costly 3. Muhopadhaya [2003] showed that a larger jury may mae a worse decson, because of the free rder problem. Ths result s farly ntutve. When the commttee sze becomes large, each commttee member becomes less responsble n the decson mang process, hence becomes less wllng to pay the postve cost to acqure nformaton. Therefore CJT fals n the Muhopadhaya s settng. He showed by numercal examples that mxed-strategy equlbra n large commttees may yeld a lower expected welfare than the optmal small commttee. The paper most related to ours n sprt s probably Martnell [2006]. The author also ntroduced cost of nformaton acquston and, n addton, he allows the precson of the sgnals contnuously depend on the amount of nvestment. Martnell [2006] proves that f the cost, and the margnal cost of the precson s zero at zero level of precson, then the decson s asymptotcally effcent. That s, f the sze of the commttee converges to nfnty, there s a sequence of symmetrc equlbra n whch each member nvests only a lttle, and the probablty of correct 3 The results becomes qute dfferent f we assume that the cost occurs when the voters go to vote, nstead of when they collect nformaton. For costly votng, see e.g. Borgers [2004] 3
4 decson converges to one. Our paper emphaszes the fxed cost aspect of nformaton acquston and should be vewed as a complementary result to Martnell [2006]. In hs accompanyng paper Martnell [2007], he showed that there s no equlbrum wth nformaton acquston when the cost s heterogeneous and prvate nformaton. However n hs model, abstenton s not allowed 4. Persco [2004] dscusses the relatonshp between decson rules and the accuracy of the sgnal. A votng rule whch requres a strong consensus to upset the status quo (n the extreme, unanmty) can be optmal only f the sgnal s suffcently accurate. The ntuton s as follows. Under the unanmty rule, the probablty of beng pvotal s small and t becomes even smaller f the sgnal s not accurate. Hence the sgnal needs to be suffcently accurate to gve an ncentve for a voter to nvest n the costly nformaton. L [2001] showed that the optmal decson rule s ex post suboptmal, when the pror belef s based toward one of the two alternatves. Gerard, Yarv [2006] also found that an ex ante optmal devce s ex post neffcent n general. Ths s because there s a trade-off between lettng agents to acqure nformaton and extractng the maxmal amount of nformaton from them. They also found that the optmal sze s bounded 5, but does not necessarly concde wth the maxmum number of agents who can be nduced to purchase nformaton n equlbrum. Gershov, Szentes [2004] showed that the optmal mechansm s sequental and characterzed the ex ante optmal votng scheme among the ex post effcent ones. The socal planner follows the cut-off decson rule. Secton 2 descrbes the model. The man result of the paper s stated and proved n Secton 3. Secton 4 concludes. Some of the proofs are relegated to the Appendx. 2 The Model There s a populaton consstng of N(> 1) ndvduals. The state of the world, ω, can tae one of two values: 1 and 1. Furthermore,Pr[ω =1]=π (0, 1). The socety must mae a decson, d, whch s ether 1 or 1. There s no conflct of nterest among ndvduals. Each of them has a beneft ofu (d, ω) f decson d s made when the state of the world s ω. Inpartcular, 0 f d = ω, u (d, ω) = q f d = 1 and ω =1, (1 q) f d =1and ω = 1, where q (0, 1) s the parameter ndcatng the severty of type-i error 6. Each ndvdual can purchase a sgnal at a cost c (> 0) at most once. Sgnals are d across ndvduals condtonal 4 If abstenton s allowed, there may be mxed-strategy equlbra n large commttee wth nformaton acquston. 5 They have ths result even though the cost s not ncluded n the defnton of socal surplus functon. 6 In the jury context, where ω =1corresponds to the nnocence of the suspect, q ndcates how severe error t s to covct an nnocent. 4
5 on the realzaton of the state of the world. The ex-post payoff of an ndvdual who nvests nto nformaton s u c. Each ndvdual maxmzes her expected payoff. There are two stages of the decson-mang process. At Stage 1, ( N) members of the socety s desgnated to serve n the commttee at random. At Stage 2, the members decde smultaneously and ndependently whether or not to nvest nto nformaton. Then, the effcent decson s made gven the sgnals collected by the members. We do not model explctly how do commttee members delberate at Stage 2. Snce there s no conflct of nterest among the members, t s easy to desgn a communcaton protocol whch effcently aggregates nformaton. Alternatvely, one can assume that the acqured nformaton s hard and hence, no communcaton s necessary for mang the ex-post effcent decson. We focus merely on the ncentves of the commttee members to acqure nformaton. Next, we turn our attenton to the defnton of socal welfare. Frst, let µ denote the ex-post effcent decson rule. That s, µ s a mappng from sets of sgnals nto possble decsons and f the sgnal profle s (s 1,..., s ) then µ (s 1,..., s )=1 E ω [u (1,ω):s 1,..., s ] E ω [u (1,ω):s 1,..., s ]. The socal welfare s measured as the expected sum of the payoffs of the ndvduals, that s, NE s1,,,.s,ω [u (µ (s 1,..., s ),ω) c], (1) where the expectaton also taes nto account the possble randomzaton of the ndvduals, that s, can be a random varable. If the commttee s large then a member mght prefer to save the cost of nformaton acquston and choose to rely on the opnons of others. On the other hand, f s too small, there s too lttle nformaton to aggregate, and hence the fnal decson s lely to be neffcent. The queston s: What s the optmal that maxmzes ex-ante socal welfare? To be more specfc, the optmal sze of the commttee s, f() themosteffcent equlbrum, n the commttee wth members, maxmzes the socal surplus among all equlbra n any commttee, and () each member acqures nformaton wth postve probablty n ths equlbrum. Snce the sgnals are d condtonal on the state of the world, the expected beneft ofan ndvdual from the ex post effcent decson s a functon of the number of sgnals acqured. Defne ths functon as follows: η () =E s1,,s,ω [u (µ (s 1,,s ),ω)]. We assume that the sgnals are nformatve, but only mperfectly, about the state of the world. That s, as the number of sgnals goes to nfnty, the probablty of mang the correct decson s strctly ncreasng and converges to one. Formally, the functon η s strctly ncreasng and lm η () =0. An ndvdual s margnal beneft from collectng an addtonal sgnal, when sgnals are already obtaned, s g () =η ( +1) η (). 5
6 Note that lm g () =0. For our man theorem to hold we need the followng assumpton. Assumpton 1 The functon g s log-convex. Ths assumpton s equvalent to g ( +1)/g () beng ncreasng n ( N). Whether or not ths assumpton s satsfed depends on the dstrbuton of the sgnals and also on the parameters q and π. An mmedate consequence of ths assumpton s the followng. Remar 1 The functon g s decreasng. Proof. Suppose by contradcton that there exsts an nteger, n 0 N, suchthat,g (n 0 +1) > g (n 0 ). Snce g ( +1)/g () s ncreasng n, t follows that g (n +1)>g(n) whenever n n 0. Hence, g (n) >g(n 0 ) > 0 whenever n>n 0. Thsmplesthatlm g () 6= 0, a contradcton. In general, t s hard to chec whether ths assumpton holds because t s often dffcult (or mpossble) to express g () analytcally. The next secton provdes examples where Assumpton 1 s satsfed. 2.1 Examples for Assumpton 1 Frst, suppose that the sgnals are normally dstrbuted around the true state of the world. The log-convexty assumpton s satsfed for the model where π + q =1, that s the socety would be ndfferent between the two possble decsons f nformaton acquston was mpossble. The assumpton s also satsfed even f π + q 6= 1f the sgnals are suffcently precse. Formally: Proposton 1 Suppose s N (ω,σ). () If q + π =1then Assumpton 1 s satsfed. () If σ s small enough then Assumpton 1 s satsfed. Proof. See the Appendx. In our next example the sgnal s trnary, that s, ts possble values are { 1, 0, 1}. In addton Pr (s = ω ω) = pr, Pr (s =0 ω) = 1 r, Pr (s = ω ω) = (1 p) r. Notce that r ( (0, 1)) s the probablty that the realzaton of the sgnal s nformatve, and p s the precson of the sgnal condtonal on beng nformatve. Proposton 2 Suppose that the sgnal s trnary. Then, there exsts a p (r) (0, 1) such that, f p > p (r), Assumpton 1 s satsfed. 6
7 Proof. See the Appendx. Next, we provde an example where the logconvexty assumpton s not satsfed. Suppose that the sgnal s bnary, that s, s { 1, 1} and Pr (s = ω ω) = p Pr (s = ω ω) = 1 p. Proposton 3 If the sgnal s bnary then Assumpton 1 s not satsfed. Proof. See the Appendx. 3 Results Ths secton s devoted to the proofs of our man theorems. To that end, we frst characterze the set of equlbra for all. The next subsecton shows that f s small, the equlbrum s unque and each member ncurs the cost of nformaton (Proposton 4). Secton 3.2 descrbes the set of mxed-strategy equlbra for large enough (Propostons 5). Fnally, Secton 3.3 proves the man theorems (Theorems 1 and 2). 3.1 Pure-strategy equlbrum Suppose that the sze of the commttee s. Ifthefrst 1 members acqure nformaton, the expected gan from collectng nformaton for the th member s g( 1). She s wllng to nvest f ths gan exceeds the cost of the sgnal, that s, f c g ( 1). (2) Ths nequalty s the ncentve compatblty constrant guaranteeng that a commttee member s wllng to nvest nto nformaton f the sze of the commttee s. Proposton 4 Let denote the sze of the commttee. There exsts a P N, such that, there exsts a unque equlbrum n whch each member nvests nto a sgnal wth probablty one f and only f mn { p,n}. Furthermore, the socal welfare generated by these equlbra s monotoncally ncreasng n mn P,N ª. Proof. Recall from Remar 1 that g s decreasng and lm g () =0. Therefore, for any postve cost c<g(0) 7, there exsts a unque P N such that g P <c<g P 1. (3) Frst we show that f < P then there s a unque equlbrum n whch each commttee member nvests nto nformaton. Suppose that n an equlbrum the frst 1 members randomze 7 If c>η(1) η(0), then nobody has ncentve to collect nformaton, hence P =0. 7
8 accordngtotheprofle (r 1,...,r 1 ), where r [0, 1] denotes the probablty that the th member nvests. Let I denote the number of sgnals collected by the frst ( 1) members. Snce the members randomze I s a random varable. Notce that I 1, and E r1,...,r 1 [g (I)] g ( 1) because g s decreasng. Also notce that from < p and (3) t follows that g ( 1) >c. Combnng the prevous two nequaltes, we get E r1,...,r 1 [g (I)] >c. Ths nequalty mples that no matter what the strateges of the frst ( 1) members are, the th member strctly prefers to nvest nto nformaton. From ths observaton, the exstence and unqueness of the pure-strategy equlbrum follow. It remans to show that f > P such a pure-strategy equlbrum does not exsts. But f p then g () <c.therefore the ncentve compatblty constrant (2) s volated, and there s no equlbrum where each member ncurs the cost of the sgnal. Fnally, we have to show that the socal welfare generated by these pure-strategy equlbra s ncreasng n ( mn { p,n}). Notce that snce N>1, c<g( 1) = η() η( 1) <N(η() η( 1)). After addng Nη( 1) c we get Nη( 1) c ( 1) <Nη() c. The left-hand sde s the socal welfare generated by the equlbrum n commttee of sze 1 whle the rght-hand sde s the socal welfare nduced by the commttee of sze. Ths s what Muhopadhaya [2003] showed n the case where sgnal s bnary. He also showed by numercal examples that mxed-strategy equlbra n large commttees would yeld lower expected welfare than small commttees. Our analyss goes further by analytcally comparng the expected welfare of all mxed-strategy equlbra ncludng asymmetrc ones. Fgure 1 s the graph of g () and c. The amount of purchased nformaton n equlbrum s neffcently small. The reason s that although nformaton s a publc good, when a commttee member decdes whether or not to nvest, she only consders her prvate beneft nstead of the socetes. Hence, the total number of sgnals acqured s smaller than the socally optmal amount, and the socal welfare s monotoncally ncreasng n the commttee sze, aslongas s smaller than the frst best level. 8
9 Fgure 1: Expected gan g () and the cost c 3.2 Mxed-strategy equlbrum Supposenowthattheszeofthecommtteeslargerthan p. We consder strategy profles n whch the commttee members can randomze when mang a decson about ncurrng the cost of nformaton acquston. The followng proposton characterzes the set of mxed-strategy equlbra (ncludng asymmetrc ones). We show that each equlbrum s characterzed by a par of ntegers (a, b). In the commttee, a members nvest nto a sgnal wth probablty one, and b members acqure nformaton wth postve but less than one probablty. The rest of the members, a (a + b) of them, do not ncur the cost. We call such an equlbrum a type-(a, b) equlbrum. Proposton 5 Letthecommtteeszebe> p. Then, for all equlbra there s a par (a, b) such that, a members nvests for sure, b members nvests wth probablty r (0, 1), and (a + b) members do not nvest. In addton a P a + b, (4) where the frst two nequaltes are strct whenever b>0. Proof. Frst, we explan that f, n an equlbrum a member nvests wth probablty r 1 (0, 1) and another one wth probablty r 2 (0, 1), then r 1 = r 2. Snce the margnal beneft from 9
10 an addtonal sgnal s decreasng, our games exhbt strategc substtuton. That s, the more nformaton the others acqure the less ncentve a member has to nvest. Hence, f r 1 <r 2 then the ndvdual who nvests wth probablty r 1 faces more nformaton n expectaton and has less ncentve to nvest than the ndvdual who nvests wth probablty r 2. On the other hand, snce r 1,r 2 (0, 1) both ndvduals must be exactly ndfferent between nvestng and not nvestng, a contradcton. Now, we formalze ths argument. Let r [0, 1] ( =1,,) be the probablty that the member collects nformaton n equlbrum. Suppose that r 1, r 2 (0, 1), andr 1 >r 2. Let I 1 and I 2 denote the number of sgnals collected by members 2, 3,..., and by members 1, 3,...,, respectvely. Notce, that snce r 1 >r 2,andg s decreasng E r2,r 3,...,r (g (I 1 )) >E r1,r 3,...,r (g (I 2 )). (5) On the other hand, a member who strctly randomzes must be ndfferent between nvestng and not nvestng, and hence, for j =1, 2 E rj,r 3,...,r (g (I j )) = c. (6) Ths equalty mples that (5) should hold wth equalty, a contradcton. Therefore, each equlbrum can be characterzed by a par (a, b) where a members collect nformaton for sure, and b members randomze, but collect nformaton wth the same probablty. It remans to show that there exsts a type-(a, b) equlbrum f and only f (a, b) satsfes (4). Suppose that n a commttee a members nvest nto nformaton for sure and b 1 nvests wth probablty r. LetG(r; a, b) denote the expected gan from acqurng nformaton for the (a + b)th member. That s, µ b 1 G(r; a, b) = r (1 r) b 1 g (a + ). =0 We clam that there exsts a type-(a, b) equlbrum f and only f there exsts an r (0, 1) such that G (r; a, b) =c. Suppose frstthatsuchanr exsts. We argue frst, that there exsts a type- (a, b) equlbrum, n whch b members nvest wth probablty r. Ths means that the b members, who are randomzng, are ndfferent between collectng nformaton and not collectng t. The a members, who nvest for sure, strctly prefer to nvest, because there s margnal gan from an addtonal sgnal exceeds G (r; a, b). Smlarly, those members who don t nvest, a (a + b) of them, are strctly better off not nvestng because ther margnal gans are strctly smaller than G (r; a, b). Next, we argue that f G (r; a, b) =c does not have a soluton n (0, 1) then there exsts no type-(a, b) equlbrum. But ths mmedately follows from the observaton that f b members are strctly randomzng, they must be ndfferent between nvestng and not nvestng and hence G (r; a, b) =c must hold. Therefore, t s suffcent to show that G (r; a, b) =c has a soluton n (0, 1) f and only f (4) holds. Notce that G(r; a, b) and strctly decreasng n r, becauseg s strctly decreasng. Also observe that G(0; a, b) =g (a) and G(1; a, b) =g (a + b 1). By the Intermedate Value Theorem, 10
11 Fgure 2: The set of mxed-strategy equlbra G(r; a, b) =c has a soluton n (0, 1) f and only f G(1; a, b) <c<g(0; a, b), whch s equvalent to g (a + b 1) <c<g(a). (7) Recall that P satsfes g P <c<g P 1. Snce g s decreasng, (1) holds f and only f a P 1 and a + b 1 P, that s, a +1< P +1 a + b. The last nequalty n (4), a + b, must hold because a + b cannot exceed the sze of the commttee,. Fgure 2 graphcally represents the set of pars (a, b) whch satsfy (4). Accordng to the prevous proposton, there are several equlbra n whch more than P members acqure nformaton wth postve probablty. A natural queston to as s: Can these mxed-strategy equlbra be compared from the pont of vew of socal welfare? The next proposton partally answers ths queston. We show that f one fxes a, the number of members who acqure nformaton for sure, then the larger s b, the number of members who randomze, the smaller s the socal welfare generated by the equlbrum. Ths proposton plays an mportant role n pnnng down the optmal sze of the commttee. Proposton 6 Suppose that N, s such that there are both type-(a, b) and type-(a, b +1) equlbra. Then the socal welfare s strctly smaller n the type-(a, b +1) equlbrum. 11
12 In order to prove ths proposton we need the followng results. Lemma 1 () G (r; a, b) >G(r; a, b +1) for r, and () r a,b >r a,b+1,wherer a,b and r a,b+1 are the solutons of G (r; a, b) =c and G (r; a, b +1)=c respectvely. Proof. See the Appendx. Proof of Proposton 6. Suppose that a members collect nformaton wth probablty one, and b members nvest wth probablty r. Letf (r; a, b) denote the beneft of an ndvdual, that s, Clearly f (r; a, b) r = bx =1 f(r; a, b) = bx =0 µ b r 1 (1 r) b η (a + ) µ b r (1 r) b η(a + ). =0 µ b r (b )(1 r) b 1 η (a + ). µ µ µ µ b b 1 b b 1 Notce that = b and (b ) =b. Therefore, the rght-hand sde of the 1 prevous equalty can be rewrtten as bx µ b 1 µ b 1 b r 1 (1 r) b η (a + ) b r (1 r) b 1 η (a + ). 1 =1 After changng the notaton n the frst expresson, ths can be further rewrtten: =0 µ b 1 µ b 1 b r (1 r) b 1 η (a + +1) b r (1 r) b 1 η (a + ) =0 =0 µ b 1 = b r (1 r) b 1 η (a + +1) η (a + ). =0 But notce that ths last expresson s just bg (r; a, b), and hence, we have Next, we show that Snce f(0 a, b) =f(0 a, b +1)=η(a) f (r; a, b) r = bg (r; a, b). f (r a,b a, b) f (r a,b+1 a, b +1)>b(r a,b r a,b+1 ) c. (8) f (r a,b a, b) f (r a,b+1 a, b +1) = f (r a,b a, b) f (0 a, b) f (r a,b+1 a, b +1) f (0 a, b +1) = b Z ra,b 0 G (r; a, b) dr b Z ra,b+1 0 G (r; a, b +1)dr. 12
13 By part () of Lemma 1, ths last dfference s smaller than b Z ra,b G (r; a, b) dr b 0 0 Z ra,b+1 G (r; a, b) dr = b Z ra,b r a,b+1 G (r; a, b) dr. By part () of Lemma we now that r a,b+1 <r a,b. In addton, snce G s decreasng n r ths last expresson s smaller than b (r a,b r a,b+1 ) G (r a,b ; a, b). Recall, that r a,b s defned as the soluton to G (r; a, b) =r, and hence, we can conclude (8). Let S(a, b) denote the socal welfare n the type-(a, b) equlbrum, that s: S(a, b) =Nf (r a,b a, b) c (a + br a,b ). Then, S (a, b) S (a, b +1) = Nf (r a,b a, b) c (a + br a,b ) [Nf (r a,b+1 a, b +1) c (a + br a,b+1 )] > Nb(r a,b r a,b+1 ) c cb (r a,b r a,b+1 ) = (N 1) cb (r a,b r a,b+1 ) > 0, where the frst nequalty follows from (8), and the last one follows from part () of Lemma The Proofs of the Theorems Frst, we show that the optmal commttee sze s ether P or P +1. Second, we prove that f > then even the worst possble equlbrum yelds hgher socal welfare than the unque equlbrum n the commttee of sze 2. Theorem 1 The optmal commttee sze,,sether P or P +1. We want to emphasze that for a certan set of parameter values the optmal sze s = p and for another set = p +1. Proof. Suppose that s the optmal sze of the commttee and the equlbrum that maxmzes socal welfare s of type-(a, b). By the defnton of optmal sze, a + b =. If b =0,thenall the commttee members nvest nto nformaton n ths equlbrum. From Proposton 4 p follows. In addton, Proposton 4 also states that the socal welfare s ncreasng n as long as p. Therefore, = p follows. Suppose now that b>0. If there exsts an equlbrum of type-(a, b 1), then, by Proposton 6, s not the optmal commttee sze. Hence, whenever the sze of the commttee s,thereexstsnotype-(a, b 1). By Proposton 5 t can only happen f the par (a, b 1) volates the nequalty chan (4) wth =.Sncethefrst and last nequaltes n (4) hold because there s a type-(a, b) equlbrum, t must be the case that the second nequalty s volated. That s, P a+b 1 = 1. Ths mples that p +1. Agan, from Proposton 4, t follows that = p or p +1. Next, we turn our attenton to the potental welfare loss due to overszed commttees. 13
14 Theorem 2 In any commttee of sze (> ), all equlbra nduce hgher socal welfare than the unque equlbrum n the commttee of sze 2. The followng lemma plays an mportant role n the proof. We pont out that the statement of ths lemma s a consequence of Assumpton 1. Lemma 2 For all N, g ( 1) {g () g ()} {g () g ( 1)}{η () η ()}, (9) and t holds wth equalty f and only f = or 1. ProofofTheorem2. Let S (a, b) denote the socal welfare generated by the equlbrum of type-(a.b). Thats,S (a, b) =S (a, µ b, r a,b ).Notcethatfexactlya+members collect nformaton, b whch happens wth probablty ra,b (1 r a,b) b, the socal welfare s Nη(a + ) c(a + ). Therefore, bx µ b h S (a, b) = r a,b (1 r a,b ) b Nη(a + ) c (a + ) =0 ( bx µ b h = r a,b (1 r a,b ) b Nη(a + ) c ) ca =0 ( bx µ ) b = N r a,b (1 r a,b ) b η (a + ) c (a + br a,b ). =0 In the last equaton, we used the fact P b b =0 ra,b (1 r a,b ) b = br. From Theorem 1, we now that = P or P +1. Also notce that by Proposton 4 S P 2, 0 <S P 1, 0. Therefore, n order to establsh S ( 2, 0) <S(a, b), tsenough to show that S P 1, 0 <S(a, b), for all pars of (a, b) whch satsfy (4). The prevous nequalty can be rewrtten as Nη P 1 c P 1 ( bx µ ) b <N ra,b (1 r a,b ) b η (a + ) c (a + br a,b ). Snce a P 1 by (4) and b N, the rght hand sde s larger than ( bx µ ) b N ra,b (1 r a,b ) b η (a + ) c ( p 1+Nr a,b ) =0 =0 =0 Hence t suffces to show that Nη P 1 c P 1 ( bx µ ) b <N ra,b (1 r a,b ) b η (a + ) c ( p 1+Nr a,b ). 14
15 After addng c ( p 1) to both sdes and dvdng through by N we have η( P 1) < ( bx =0 µ ) b r a,b(1 r a,b ) b η(a + ) cr a,b. (10) The left-hand sde s a payoff of an ndvdual f P 1 sgnals are acqured by others, whle the rght-hand sde s an ndvdual s who s randomzng n a type-(a, b) equlbrum wth probablty r a,b. Snce ths ndvdual s ndfferent between randomzng and not collectng nformaton, the rght-hand sde of (10) can be rewrtten as Hence (10) s equvalent to µ b 1 r a,b(1 r a,b ) b 1 η(a + ). =0 µ b 1 η( P 1) < r a,b(1 r a,b ) b 1 η(a + ). (11) =0 By Lemma 2 ( µ ) b 1 g ( p 1) r a,b(1 r a,b ) b 1 g(a + ) g ( p ) =0 ( µ ) b 1 > {g ( p ) g ( p 1)} r a,b(1 r a,b ) b 1 η(a + ) η ( p ). =0 (12) Notce that µ b 1 r a,b(1 r a,b ) b 1 g(a + ) =c<g( p 1), =0 where the equalty just guarantees that a member who s randomzng s ndfferent between nvestng and not nvestng, and the nequalty holds by (3). Hence, from (12), g ( p 1) {g ( p 1) g ( p )} ( µ ) b 1 > {g ( p ) g ( p 1)} r a,b(1 r a,b ) b 1 η(a + ) η ( p ). Snce g P 1 g P > 0, the prevous nequalty can be rewrtten as =0 µ b 1 g ( p 1) >η( p ) r a,b(1 r a,b ) b 1 η(a + ). =0 Fnally, snce η ( p ) g ( p 1) = η ( p 1), thsnequaltysjust(11). Now (11) follows and the proof s done. Fgure 3 s the graph of the expected socal welfare, wth parameters N = 100,σ = 1,c =
16 Fgure 3: Socal Welfare, as a functon of the commttee sze 16
17 4 Concluson In ths paper, we have dscussed the optmal commttee sze and the potental welfare losses assocated to overszed commttees. We have focused on envronments where there s no conflct of nterest among ndvduals, but nformaton acquston s costly. Frst, we have confrmed that the optmal commttee sze s bounded. In other words, the Condorcet Jury Theorem fals to hold, that s, larger commttees mght nduce smaller socal welfare. However, we showed that the welfare loss due to overszed commttees s surprsngly small. In an arbtrarly large commttee, even the worst equlbrum generates a hgher welfare than an equlbrum n a commttee where there are two less members than n the optmal one. 5 Appendx Proof of Proposton 1. Part() If q + π =1, then qπ =(1 q)(1 π). Hence η () qπ = Pr [µ (s 1,,s )= 1 ω =1]+ Pr [µ (s 1,,s )=1 ω = 1] s 1,,s s 1,,s Ex post effcent decson rule s µ (s 1,,s )=1 f s s > 0, µ (s 1,,s )= 1 f s s < 0. The sum of normal dstrbuton s also normal; P ³ =1 s N ω, σ. Hence ³ P Pr (s s ω =1) = 1 σ φ =1 s σ ³ P Pr (s s ω = 1) = 1 σ φ =1 s + σ ³ where φ s the pdf of the standard normal dstrbuton; φ(x) =(2π) 1 2 exp η () qπ µ µ = Φ σ +1 Φ σ Ã! = 2Φ. σ x2 2. Therefore Hence Moreover, η 0 () = qπ σ η 0 ( +1) η 0 () Ã! 1 φ σ = for >0 r µ +1 exp 1 2σ 2 17
18 s ncreasng n (> 0). By tang ε =0n the followng lemma, g ( +1)/g () s ncreasng n N. Lemma 3 Let ε 0 and suppose η 0 ( +1)/η 0 () s ncreasng for >ε.theng ( +1)/g () < g ( +2)/g ( +1) for ε. Proof. Fx ( ε). For any t (, +1), Therefore η 0 ( +2) η 0 ( +1) < η0 (t +2) η 0 (t +1) η0 ( +2)η 0 (t +1)<η 0 ( +1)η 0 (t +2) η 0 ( +2) Z +1 η 0 (t +1)dt < η 0 ( +1) Z +1 η 0 (t +2)dt η 0 ( +2)[η ( +2) η ( +1)]<η 0 ( +1)[η ( +3) η ( +2)] η0 ( +2) η ( +3) η ( +2) η 0 < ( +1) η ( +2) η ( +1) Smlarly, for any t (, +1), Therefore Hencewehave η 0 ( +2) η 0 ( +1) > η0 (t +1) η 0 η 0 ( +2)η 0 (t) >η 0 ( +1)η 0 (t +1) (t) η 0 ( +2) Z +1 η 0 (t) dt > η 0 ( +1) Z +1 η 0 (t +1)dt η 0 ( +2)[η ( +1) η ()] >η 0 ( +1)[η ( +2) η ( +1)] η0 ( +2) η ( +2) η ( +1) η 0 > ( +1) η ( +1) η () η ( +2) η ( +1) η ( +1) η () < η0 ( +2) η ( +3) η ( +2) g ( +1) g ( +2) η 0 < < ( +1) η ( +2) η ( +1) g () g ( +1) for any ε. Part() Now, let us consder the asymmetrc model. Ex post effcent decson s a cut-off rule for normal sgnal: µ (s 1,,s ) = 1 f s s >θ, µ (s 1,,s ) = 1 f s s <θ where the cut-off pont s θ = σ2 2 (1 q)(1 π) log. qπ 18
19 Hence η () = qπ Pr [µ (s 1,,s )= 1 ω =1] s 1,,s (1 q)(1 π) Pr [µ (s 1,,s )=1 ω = 1] s 1,,s µ µ θ = qπφ σ (1 q)(1 π) Φ θ + σ. We are gong to show the followng. For any ε>0, there exsts δ>0, such that σ<δmples g ( +1)/g () <g( +2)/g ( +1) for. For that purpose, we clam that for any ε>0, there s a suffcently small σ such that η 0 ( +1)/η 0 () s ncreasng for >ε. ³ r η 0 ( +1) φ θ (+1) σ +1 η 0 = ³ () +1 θ φ σ ³ ³ r exp 1 θ 2 2σ = θ + +1 ³ ³ +1 exp 1 θ 2 2σ 2 2θ + r µ µ 1 = +1 exp θ 2 2σ 2 ( +1) 1 Hence η 0 ( +1)/η 0 () s ncreasng f and only f ϕ () +1 µ exp θ 2 σ 2 ( +1) s decreasng n. µ ϕ 0 θ 2 ( µ +1 θ 2 Ã! ) 1 () =exp σ 2 ( +1) σ ( +1) Therefore ϕ s decreasng +1 µ θ 2 σ 2 θ2 σ 2 < 1 ³ +1 1 σ2 4 µ log Ã! ( +1) < (+1) 2 2 = µ 2 (1 q)(1 π) < qπ ( +1) 2 +1 ( +1) The rght-hand sde s strctly ncreasng n ( 0) and lm 0 { ( +1)/ (2 +1)} =0. Hence for suffcently small σ, ϕ s decreasng for >ε.therefore, η 0 ( +1)/η 0 () s ncreasng for >ε. Then by Lemma 3, g ( +1)/g () <g( +2)/g ( +1) for ε. By tang ε<1, we have shown that g ( +1)/g () <g( +2)/g ( +1) for =1, 2,. What s left to be shown s g (1) /g (0) <g(2) /g (1). As above, η 0 (2) /η 0 (1) <η 0 (t +1)/η 0 (t) for t (1, 2) mples 19
20 η 0 (2) /η 0 (1) <g(2) /g (1). Hence t suffces to show that g (1) /g (0) <η 0 (2) /η 0 (1) for suffcently small σ. Defne L =log{(1 q)(1 π) / (qπ)}. Then for >0, ( Ã! Ã Lσ η () = qπ Φ 2 + e L Φ Lσ!) σ 2 σ and η 0 (2) η 0 (1) ³ 1 2 φ θ 2 σ 2 = φ θ 1 σ Ã ( µθ = 1 exp 1 2 µ )! 2 2 θ σ 2 σ = 1 µ exp 1 µ θ2 2 2σ = 1 µ L 2 exp 2 16 σ2 1 2σ 2. We are gong to show that Then lm σ 0 η()! qπ exp L 2 16 σ2 1 2σ 2 lm σ 0 lm σ 0 g (1) η 0 (2) /η 0 (1) =0 g (1) /g (0) η 0 (2) /η 0 (1) =0 because lm σ 0 g (0) = mn {qπ, (1 q)(1 π)}, whch mples g (1) /g (0) <η 0 (2) /η 0 (1) for suffcently small σ. g (1) lm σ 0 η 0 (2) /η 0 (1) = Ã η(1) qπ 2qπ lm! qπ σ 0 exp. L 2 16 σ2 1 2σ 2 By l Hôptal s theorem, for =1and 2, Ã ³ η() σ qπ = lm σ 0 = lm σ 0 = lm σ 0 σ exp L 2 16 σ2 1 2σ 2 2 σ φ 2 L 2 8 σ + 1 ³ Lσ 2 σ σ exp L 2 3 2σ ³ Lσ φ 2 σ L 2 8 σ4 +1 exp L 2 16 σ σ2 1 2σ 2 2σ 2 20
21 Now ³ lm φ Lσ 2 σ σ 0 exp L 2 16 σ2 1 2σ 2 = = = 1 2π lm σ 0 exp 1 2π lm σ 0 exp à 1 2 µ ( 0 f =2 1 2π e L 2 f =1 Lσ 2 µ ! 2 µ L 2 16 σ2 1 2σ 2 σ L 2 σ 2 + L σ 2 In ether case, Hence lm σ 0 Ã! η() qπ exp L 2 16 σ2 1 2σ 2 lm σ 0 g (1) η 0 (2) /η 0 (1) =0 whch completes the proof. ProofofProposton2. Frst, we clam that the ex post effcent decson rule µ : { 1, 0, 1} { 1, 1} s the followng cut-off rule: where =0 X s > b θ µ (s 1,,s )=1 (13) =1 X s < b θ µ (s 1,,s )= 1 (14) =1 b θ = log ³ (1 q)(1 π) qπ ³ log p 1 p and te s broen by a far con toss. Suppose (s 1,,s ) s a permutaton of 1,, 1, 0,, 0, 1,, 1. {z } {z } {z } a a b b Then µ (s 1,,s )=1f and E ω [u (ω, 1) s 1,,s ] > E ω [u (ω, 1) s 1,,s ] (1 q)(1 π) Pr [s 1,,s ω = 1] > qπ Pr [s 1,,s ω =1] Pr [s 1,,s ] Pr [s 1,,s ] Pr [s 1,,s ω =1] = (pr) a (1 r) a b (r (1 p)) b Pr [s 1,,s ω = 1] = (pr) b (1 r) a b (r (1 p)) a. 21
22 Hence µ (s 1,,s )=1f (1 q)(1 π) p b (1 p) a > qπp a (1 p) b ³ log (1 q)(1 π) qπ a b> = b θ. log ³ p 1 p By defnton, P =1 s = a b, whch gves (13). The opposte sgn of nequalty gves (14). Ex post effcent rule s a cut-off rule. Remember that expected utlty η () s gven by η () = qπ Pr [µ (s 1,,s )= 1 ω =1] s 1,,s (1 q)(1 π) Pr [µ (s 1,,s )=1 ω = 1] s 1,,s Now we consder the case where p converges to 1. Notcethat bθ < 1 when p s close enough to one. There are 3 dstnct cases accordng to the parameter ³ values of q and π. () Frst, suppose that q + π>1. Then log (1 q)(1 π) qπ < 0, whch gves b θ<0. When p s close enough to one, 1 < b θ<0. Then η () = Ã! Pr [a b = 1 ω =1] qπ +Pr[a b< 1 ω =1] Pr [a b =0 ω = 1] (1 q)(1 π) +Pr[a b =1 ω = 1] +Pr[a b>1 ω = 1] and Let ε =1 p. Then Smlarly, and Pr [a b =0 ω = 1] = [ 2 X ]! j!( 2j)!j! (pr)j (1 r) 2j ((1 p) r) j j=0 Pr [a b =0 ω = 1] = (1 r) + ( 1) r 2 (1 r) 2 ε + o ε 2 Pr [a b =0 ω =1] = (1 r) + ( 1) r 2 (1 r) 2 ε + o ε 2 Pr [a b = 1 ω =1] = r (1 r) 1 ε + o ε 2 Pr [a b =1 ω = 1] = r (1 r) 1 ε + o ε 2 Pr [a b< 1 ω =1] = o ε 2 Pr [a b>1 ω = 1] = o ε 2 22
23 Therefore, η () = qπr (1 r) 1 ε Ã! (1 r) + ( 1) r 2 (1 r) 2 ε (1 q)(1 π) +r (1 r) 1 + o ε 2 ε = (1 q)(1 π)(1 r) qπr (1 r)+(1 q)(1 π) ( 1) r 2 + r (1 r) ª (1 r) 2 ε + o ε 2 ³ () Smlarly, f q +π <1, then log (1 q)(1 π) qπ > 0, whch gves b θ>0. When p s close enough to one, 0 < b θ<1. Then, η () = qπ (1 r) (1 q)(1 π) r (1 r)+qπ ( 1) r 2 + r (1 r) ª (1 r) 2 ε + o ε 2 () In the symmetrc case q + π =1, b θ = Pr [a b =0 ω =1] η () = qπ +Pr[a b = 1 ω =1] +Pr[a b< 1 ω =1] 1 2 Pr [a b =0 ω = 1] (1 q)(1 π) +Pr[a b =1 ω = 1] +Pr[a b>1 ω = 1] Snce qπ =(1 p)(1 π), Pr [a b =0 ω =1] η () = qπ +2 Pr [a b = 1 ω =1] +2 Pr [a b< 1 ω =1] Ã! (1 r) + ( 1) r 2 (1 r) 2 ε = qπ +2r (1 r) 1 + o ε 2 ε In the frst 2 cases, η ( +1) η () = qπr (1 r r)(1 r) 1 ε Ã (1 r) + ( 1) r 2 (1 r) 2 ε (1 q)(1 π) +r (1 r) 1 ε = A ()+B() ε + o ε 2! + o ε 2 where A () = (1 q)(1 π)(1 r) B () = qπr (1 r r)(1 r) 1 (1 q)(1 π) (1 2r + r) r (1 r) 2 23
24 Then g ( +1) g () η ( +2) η ( +1) = η ( +1) η () = A ( +1)+B ( +1)ε + o ε 2 A ()+B() ε + o (ε 2 ) ³ A ( +1) 1+ B(+1) A(+1) ε + o 2 ε = ³ A () 1+ B() A() ε + o (ε2 ) Snce g ( +1) g () 1+ B(+1) A(+1) ε + o ε 2 1+ B() A() ε + o (ε2 ) µ B ( +1) =1+ A ( +1) B () ε + o ε 2, A () µ A ( +1) = 1+ A () Ã Ã qπr = (1 r) 1+ (1 q)(1 π)(1 r) µ B ( +1) A ( +1) B () ε + o ε 2 A () + r (1 + 2r r) (1 r) 2!! ε + o ε 2 Therefore, g ( +1)/g () s ncreasng n for suffcently small ε, as long as r s strctly smaller than 1 (whch s what we wanted to show). In the symmetrc case, Ã! r (1 r) + (2 r r) r 2 (1 r) 2 ε η ( +1) η () = qπ +2r (1 r) 1 + o ε 2 (1 r r) ε Hence g ( +1) g () Ã =(1 r) 1+! 2r ( 1+r + r)+2(1 r)( 1+2r +2r) (1 r) 2 ε + o ε 2 Therefore, g ( +1)/g () s ncreasng n for suffcently small ε, as long as r s strctly smaller than 1. ProofofProposton3. Remember that η () = E ω,s [u (ω,µ (s))] = qπ Pr [µ (s) = 1 ω =1] (1 q)(1 π)pr[µ (s) =1 ω = 1] (15) s s Ex post effcent rule s agan the cutoff rule; µ (s) =1 E ω [u (ω, 1) s] E ω [u (ω, 1) s] (1 q)pr ω [ω = 1 s] qpr [ω =1 s] ω (1 q)(1 π)pr[s ω = 1] qπ Pr [s ω =1] s s Pr s [s ω =1] (1 q)(1 π) Pr s [s ω = 1] qπ 24
25 For the bnary sgnal, for both s =1and 1. Hence µ s Pr s [s ω =1] p Pr s [s ω = 1] = 1 p µ (s) =1 µ P s p X 1 p s θ (1 q)(1 π) qπ where log θ = log ³ (1 q)(1 π) qπ ³ p 1 p By symmetry, assume θ 0 wthout loss of generalty. (that s, d = 1 s the status quo.) () Suppose θ>1. Then for any <θ,no sgnal realzaton (s 1,,s ) can upset the status quo, thus η () = qπ. Then the gan g () s zero for <θ 1. Then g ( +1)/g () s not well-defned. () Suppose 0 θ<1. Then we clam that mx µ η (2m) = {qπ +(1 q)(1 π)} 2m p j (1 p) 2m j (16) j j=0 µ 2m +(1 q)(1 π) p m (1 p) m m mx µ η (2m +1) = {qπ +(1 q)(1 π)} 2m +1 p j (1 p) 2m+1 j (17) j Suppose (= 2m) s even. Then µ (s) = 1 ff P s { 2m, 2m +2,, 2, 0}. It happens when there are j ( {0,,m}) sgnals of s =1and the other sgnals are s = 1. Hence mx µ 2m Pr [µ (s) = 1 ω =1]= p j (1 p) 2m j s j j=0 Smlarly, µ (s) =1ff P s {2, 4,, 2m}. It happens when there are j ( {0,,m 1}) sgnals of s = 1 and the other sgnals are s =1. Hence Pr s [µ (s) =1 ω = 1] = m 1 X j=0 j=0 µ 2m j p j (1 p) 2m j Then by (15), (16) follows. Suppose (= 2m +1) s odd. Then µ (s) = 1 ff P s { 2m 1, 2m +1,, 3, 1}. It happens when there are j ( {0,,m}) sgnals of s =1and the other sgnals are s = 1. Hence mx µ 2m Pr [µ (s) = 1 ω =1]= p j (1 p) 2m+1 j s j j=0 25
26 Smlarly, µ (s) =1ff P s {1, 3,, 2m +1}. It happens when there are j ( {0,,m}) sgnals of s = 1 and the other sgnals are s =1. Hence mx µ 2m Pr [µ (s) =1 ω = 1] = p j (1 p) 2m+1 j s j Then by (15), (17) follows. Now, usng (16) and (17), expected gan s µ 2m g (2m) = {pqπ (1 p)(1 q)(1 π)} m µ 2m +1 g (2m +1) = {(1 q)(1 π) qπ} m j=0 p m (1 p) m p m (1 p) m Recall that we assumed 0 θ<1, whch s equvalent to pqπ (1 p)(1 q)(1 π) > 0 and (1 q)(1 π) qπ 0. Now, f θ>0, g (2m +2) pqπ (1 p)(1 q)(1 π) = 2p (1 p) g (2m +1) (1 q)(1 π) qπ s a constant, Assumpton 1 cannot be satsfed. If θ =0,g(2m +2)/g (2m +1)s not well-defned. ProofofLemma1. Part (). Notce that µ b 1 G (r; a, b) = r (1 r) b 1 g (a + ). =0 Snce r (1 r) b 1 = r (1 r) b + r +1 (1 r) b 1, µ b 1 r G (r; a, b) = (1 r) b + r +1 (1 r) b 1 g (a + ). =0 Snce g s decreasng µ b 1 µ G (r; a, b) > r (1 r) b b 1 g (a + )+ r +1 (1 r) b 1 g (a + +1) =0 =0 µ b 1 bx µ = r (1 r) b b 1 g (a + )+ r (1 r) b g (a + ) 1 =0 =1 bx µ µ b 1 b 1 = + r (1 r) b g (a + ), 1 =0 where the frstequaltyholdsbecausewehavejustredefned µ theµ notaton n the second summaton, n n and the second equalty holds because, by conventon, = =0for all n N. Fnally, µ µ µ 1 n +1 b 1 b 1 b usng + =,wehave 1 bx µ b G (r; a, b) > r (1 r) b g (a + ) =G (r; a, b +1). =0 26
27 Part (). By the defntons of r a,b and r a,b+1,wehave c = G (r a,b ; a, b) =G (r a,b+1 ; a, b +1), and by part () of ths lemma, G (r a,b+1 ; a, b +1)<G(r a,b+1 ; a, b). Therefore G (r a,b ; a, b) <G(r a,b+1 ; a, b) Snce G (r; a, b) s strctly decreasng n r, r a,b >r a,b+1 follows. Proof of Lemma 2. Frst, notce that f a j > 0 for j {0,..., } and a j+1 /a j <a j+2 /a j+1 for all j {0,..., 2}, then a a 1 > P j= a j P 1 j= 1 a j for all {1,..., 1}. Assumpton 1 allows us to apply ths result for the sequence a j = g (j), and hence, g () g ( 1) > P j= P g (j) η ( +1) η ( +1) 1 = j= g (j) η () η () for all {1,..., 1}.!!!(equalty f and only f = 1.) Smlarly, f a j > 0 and a j+1 /a j <a j+2 /a j+1 for all j N, then for all >0, a a 1 < P j= g (j) P 1 j= 1 g (j) for all >. Applyng ths result to the sequence a j = g (j), weget g () g ( 1) < P j= P g (j) η ( +1) η ( +1) 1 = j= 1 g (j) η () η () for all >. So far, we have shown that g ( 1) [η ( +1) η ( +1)] g ()[η() η ()], and t holds wth equalty f and only f = 1 or. By subtractng g ( 1) [η () η ()], g ( 1) [g () g ()] [g () g ( 1)] [η () η ()]. 27
28 References [1] D. Austen-Smth, J.S. Bans Informaton Aggregaton, Ratonalty, and the Condorcet Jury Theorem Amercan Poltcal Scence Revew 90(1) (1996) [2] R. Ben-Yashar, S. Ntzan The nvaldty of the Condorcet Jury Theorem under endogenous decsonal slls Economcs of Governance 2(3) (2001) [3] T. Börgers, Costly Votng Amercan Economc Revew 94 (2004) [4] Le Marqus de Condorcet Essa sur l applcaton de l analyse à la probablté des décsons rendues à la pluralté des vox Les Archves de la Revoluton Françase, Pergamon Press (1785) [5] P. Coughlan In Defense of Unanmous Jury Verdcts: Mstrals, Communcaton, and Strategc Votng Amercan Poltcal Scence Revew 94(2) (2000) [6] T. Feddersen, W. Pesendorfer Swng Voter s Curse Amercan Economc Revew 86 (1996) [7] T. Feddersen, W. Pesendorfer Votng Behavor and Informaton Aggregaton n Electons wth Prvate Informaton Econometrca 65(5) (1997) [8] T. Feddersen, W. Pesendorfer Convctng the Innocent: The Inferorty of Unanmous Jury Verdcts under Strategc Votng Amercan Poltcal Scence Revew 92(1) (1998) [9] D.Gerard,L.Yarv, Informaton Acquston n Commttees mmeo (2006) [10] K. Gerlng, H.P. Grüner, A. Kel, E. Schulte, Informaton Acquston and Decson Mang n Commttees: A Survey European Central Ban, Worng Paper No. 256 (2003) [11] A. Gershov, B. Szentes, Optmal Votng Scheme wth Costly Informaton Acquston mmeo (2006) [12] B. Grofman, G. Owen, Condorcet Models, Avenues for Future Research Informaton Poolng and Group Decson Mang, Ja Press (1986) [13] H. L, "A Theory of Conservatsm" Journal of Poltcal Economy 109(3) (2001) [14] C. Martnell, Would Ratonal Voters Acqure Costly Informaton? Journal of Economc Theory 129 (2006) [15] C. Martnell, Ratonal Ignorance and Votng Behavor Internatonal Journal of Game Theory 35 (2007) [16] A. McLennan, Consequences of the Condorcet Jury Theorem for Benefcal Informaton Aggregaton by Ratonal Agents Amercan Poltcal Scence Revew 92(2) (1998)
29 [17] N. Mller, Informaton, Electorates, and Democracy: Some Extensons and Interpretatons of the Condorcet Jury Theorem Informaton Poolng and Group Decson Mang, Ja Press (1986) [18] K. Muhopadhaya, JurySzeandtheFreeRderProblem JournalofLaw,Economcs,and Organzaton 19(1) (2003) [19] R.B. Myerson, Game Theory: Analyss of Conflct Cambrdge, MA: Harvard Unversty Press (1991) [20] R.B. Myerson, Extended Posson Games and the Condorcet Jury Theorem Games and Economc Behavor 25 (1998) [21] N. Persco Commttee Desgn wth Endogenous Informaton Revew of Economc Studes 71(1) (2004)
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 informationA 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 informationEconomics 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 informationEdge 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 informationGames 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 informationON 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 informationPerfect 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 informationWelfare 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 information3.1 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 informationMarket 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 informationCS286r 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 informationThe 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 informationOnline 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 informationTit-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 informationImplementation 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 informationEndogenous 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(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 informationPricing 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 informationprinceton 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 information4 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 information2E 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 informationHila 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 informationThe 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 informationAssortment 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 informationInfinitely Split Nash Equilibrium Problems in Repeated Games
Infntely Splt ash Equlbrum Problems n Repeated Games Jnlu L Department of Mathematcs Shawnee State Unversty Portsmouth, Oho 4566 USA Abstract In ths paper, we ntroduce the concept of nfntely splt ash equlbrum
More informationModule 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 informationCollege of Computer & Information Science Fall 2009 Northeastern University 20 October 2009
College of Computer & Informaton Scence Fall 2009 Northeastern Unversty 20 October 2009 CS7880: Algorthmc Power Tools Scrbe: Jan Wen and Laura Poplawsk Lecture Outlne: Prmal-dual schema Network Desgn:
More informationGame 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 informationConjectures 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 informationCredit 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 informationModule 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 informationLecture 10 Support Vector Machines II
Lecture 10 Support Vector Machnes II 22 February 2016 Taylor B. Arnold Yale Statstcs STAT 365/665 1/28 Notes: Problem 3 s posted and due ths upcomng Frday There was an early bug n the fake-test data; fxed
More informationSubjective 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 informationTHE 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 informationk 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 informationAbstract 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 informationFoundations 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 informationMaximizing 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 informationA 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 informationGraph 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 informationUniqueness 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 informationLinear 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 informationCompetitive Experimentation and Private Information
Compettve Expermentaton an Prvate Informaton Guseppe Moscarn an Francesco Squntan Omtte Analyss not Submtte for Publcaton Dervatons for te Gamma-Exponental Moel Dervaton of expecte azar rates. By Bayes
More informationIntroduction. 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 informationInformation 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 informationKernel 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 informationOnline 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 informationInvestment Secrecy and Competitive R&D
BE J. Econ. nal. Polcy 2016; aop Letter dt Sengupta* Investment Secrecy and Compettve R&D DOI 10.1515/beeap-2016-0047 bstract: Secrecy about nvestment n research and development (R&D) can promote greater
More informationStanford 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 informationAppendix 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 informationA 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 informationECE559VV 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 informationPower law and dimension of the maximum value for belief distribution with the max Deng entropy
Power law and dmenson of the maxmum value for belef dstrbuton wth the max Deng entropy Bngy Kang a, a College of Informaton Engneerng, Northwest A&F Unversty, Yanglng, Shaanx, 712100, Chna. Abstract Deng
More informationDeterministic 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 informationAppendix for Causal Interaction in Factorial Experiments: Application to Conjoint Analysis
A Appendx for Causal Interacton n Factoral Experments: Applcaton to Conjont Analyss Mathematcal Appendx: Proofs of Theorems A. Lemmas Below, we descrbe all the lemmas, whch are used to prove the man theorems
More informationANSWERS. 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 informationCOS 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 informationCS : 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 informationMMA 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 informationEcon107 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 informationFinding Dense Subgraphs in G(n, 1/2)
Fndng Dense Subgraphs n Gn, 1/ Atsh Das Sarma 1, Amt Deshpande, and Rav Kannan 1 Georga Insttute of Technology,atsh@cc.gatech.edu Mcrosoft Research-Bangalore,amtdesh,annan@mcrosoft.com Abstract. Fndng
More informationMASSACHUSETTS 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 informationREAL ANALYSIS I HOMEWORK 1
REAL ANALYSIS I HOMEWORK CİHAN BAHRAN The questons are from Tao s text. Exercse 0.0.. If (x α ) α A s a collecton of numbers x α [0, + ] such that x α
More informationPROBLEM SET 7 GENERAL EQUILIBRIUM
PROBLEM SET 7 GENERAL EQUILIBRIUM Queston a Defnton: An Arrow-Debreu Compettve Equlbrum s a vector of prces {p t } and allocatons {c t, c 2 t } whch satsfes ( Gven {p t }, c t maxmzes βt ln c t subject
More informationThe 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 informationWeek3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity
Week3, Chapter 4 Moton n Two Dmensons Lecture Quz A partcle confned to moton along the x axs moves wth constant acceleraton from x =.0 m to x = 8.0 m durng a 1-s tme nterval. The velocty of the partcle
More informationOn 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 informationAPPROXIMATE PRICES OF BASKET AND ASIAN OPTIONS DUPONT OLIVIER. Premia 14
APPROXIMAE PRICES OF BASKE AND ASIAN OPIONS DUPON OLIVIER Prema 14 Contents Introducton 1 1. Framewor 1 1.1. Baset optons 1.. Asan optons. Computng the prce 3. Lower bound 3.1. Closed formula for the prce
More informationMarginal Effects in Probit Models: Interpretation and Testing. 1. Interpreting Probit Coefficients
ECON 5 -- NOE 15 Margnal Effects n Probt Models: Interpretaton and estng hs note ntroduces you to the two types of margnal effects n probt models: margnal ndex effects, and margnal probablty effects. It
More informationModule 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 informationStat260: 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 informationUnderstanding 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 informationAPPENDIX 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 informationU.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017
U.C. Berkeley CS94: Beyond Worst-Case Analyss Handout 4s Luca Trevsan September 5, 07 Summary of Lecture 4 In whch we ntroduce semdefnte programmng and apply t to Max Cut. Semdefnte Programmng Recall that
More informationAffine 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 informationMore 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 informationComplete subgraphs in multipartite graphs
Complete subgraphs n multpartte graphs FLORIAN PFENDER Unverstät Rostock, Insttut für Mathematk D-18057 Rostock, Germany Floran.Pfender@un-rostock.de Abstract Turán s Theorem states that every graph G
More informationNUMERICAL DIFFERENTIATION
NUMERICAL DIFFERENTIATION 1 Introducton Dfferentaton s a method to compute the rate at whch a dependent output y changes wth respect to the change n the ndependent nput x. Ths rate of change s called the
More informationProblem Set 9 Solutions
Desgn and Analyss of Algorthms May 4, 2015 Massachusetts Insttute of Technology 6.046J/18.410J Profs. Erk Demane, Srn Devadas, and Nancy Lynch Problem Set 9 Solutons Problem Set 9 Solutons Ths problem
More informationDifference 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 informationEquilibrium 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 informationCopyright (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 informationLecture 14: Bandits with Budget Constraints
IEOR 8100-001: Learnng and Optmzaton for Sequental Decson Makng 03/07/16 Lecture 14: andts wth udget Constrants Instructor: Shpra Agrawal Scrbed by: Zhpeng Lu 1 Problem defnton In the regular Mult-armed
More informationLectures - 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 informationLecture 12: Discrete Laplacian
Lecture 12: Dscrete Laplacan Scrbe: Tanye Lu Our goal s to come up wth a dscrete verson of Laplacan operator for trangulated surfaces, so that we can use t n practce to solve related problems We are mostly
More informationPerron Vectors of an Irreducible Nonnegative Interval Matrix
Perron Vectors of an Irreducble Nonnegatve Interval Matrx Jr Rohn August 4 2005 Abstract As s well known an rreducble nonnegatve matrx possesses a unquely determned Perron vector. As the man result of
More informationMixed 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 informationECONOMICS 351*-A Mid-Term Exam -- Fall Term 2000 Page 1 of 13 pages. QUEEN'S UNIVERSITY AT KINGSTON Department of Economics
ECOOMICS 35*-A Md-Term Exam -- Fall Term 000 Page of 3 pages QUEE'S UIVERSITY AT KIGSTO Department of Economcs ECOOMICS 35* - Secton A Introductory Econometrcs Fall Term 000 MID-TERM EAM ASWERS MG Abbott
More informationLecture 3: Shannon s Theorem
CSE 533: Error-Correctng Codes (Autumn 006 Lecture 3: Shannon s Theorem October 9, 006 Lecturer: Venkatesan Guruswam Scrbe: Wdad Machmouch 1 Communcaton Model The communcaton model we are usng conssts
More informationGeneral viscosity iterative method for a sequence of quasi-nonexpansive mappings
Avalable onlne at www.tjnsa.com J. Nonlnear Sc. Appl. 9 (2016), 5672 5682 Research Artcle General vscosty teratve method for a sequence of quas-nonexpansve mappngs Cuje Zhang, Ynan Wang College of Scence,
More informationIn the figure below, the point d indicates the location of the consumer that is under competition. Transportation costs are given by td.
UC Berkeley Economcs 11 Sprng 006 Prof. Joseph Farrell / GSI: Jenny Shanefelter Problem Set # - Suggested Solutons. 1.. In ths problem, we are extendng the usual Hotellng lne so that now t runs from [-a,
More informationInner 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 informationVickrey 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 informationRobust 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 information10-801: Advanced Optimization and Randomized Methods Lecture 2: Convex functions (Jan 15, 2014)
0-80: Advanced Optmzaton and Randomzed Methods Lecture : Convex functons (Jan 5, 04) Lecturer: Suvrt Sra Addr: Carnege Mellon Unversty, Sprng 04 Scrbes: Avnava Dubey, Ahmed Hefny Dsclamer: These notes
More informationa 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 informationEstimation: Part 2. Chapter GREG estimation
Chapter 9 Estmaton: Part 2 9. GREG estmaton In Chapter 8, we have seen that the regresson estmator s an effcent estmator when there s a lnear relatonshp between y and x. In ths chapter, we generalzed the
More informationSociété de Calcul Mathématique SA
Socété de Calcul Mathématque SA Outls d'ade à la décson Tools for decson help Probablstc Studes: Normalzng the Hstograms Bernard Beauzamy December, 202 I. General constructon of the hstogram Any probablstc
More informationCHAPTER 14 GENERAL PERTURBATION THEORY
CHAPTER 4 GENERAL PERTURBATION THEORY 4 Introducton A partcle n orbt around a pont mass or a sphercally symmetrc mass dstrbuton s movng n a gravtatonal potental of the form GM / r In ths potental t moves
More informationComputing 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