On Tacit Collusion among Asymmetric Firms in Bertrand Competition

Size: px
Start display at page:

Download "On Tacit Collusion among Asymmetric Firms in Bertrand Competition"

Transcription

1 On Tact Colluson among Asymmetrc Frms n Bertrand Competton Ichro Obara Department of Economcs UCLA Federco Zncenko Department of Economcs UCLA November 11, 2011 Abstract Ths paper studes a model of repeated Bertrand competton among asymmetrc frms that produce a homogeneous product. The dscountng rates and margnal costs may vary across frms. We dentfy the crtcal level of dscount factor such that a collusve outcome can be sustaned f and only f the average dscount factor wthn the lowest cost frms s above the crtcal level. We also characterze the set of all effcent collusve equlbra when frms dffer only n ther dscountng rates. Due to dfferental dscountng, mpatent frms gan a larger share of the market at an earler stage of the game and patent frms gan a larger share at a later stage n effcent equlbrum. Although there are many effcent collusve equlbra, our model provdes a unque predcton n the long run n the sense that every effcent collusve equlbrum converges to the unque effcent statonary collusve equlbrum wthn fnte tme. JEL Classfcaton: C72, C73, D43. Keywords: Bertrand Competton, Colluson, Dfferental Dscountng, Repeated Game, Subgame Perfect Equlbrum. 1 Introducton The model of repeated Bertrand competton explans how frms may be able to collude and sustan a hgh prce even when they produce dentcal goods. Thus t resolves so called Bertrand paradox, whch would arse n one-shot nteracton, that frms lose any monopoly power and make no proft as soon as two frms are present n the market. (Trole [7]). 1 Snce t s a smple and very convenent model, t has been used n numerous appled works. However, we stll do not fully understand when and how colluson can be sustaned except for the very specal case where frms are symmetrc. Ths assumpton of symmetrc frms s of course very strong and unrealstc; frms n general dffer n varous dmensons. What 1 There are many other ways to resolve Bertrand paradox such as ntroducng capacty constrants or dfferentated demands etc. 1

2 we thnk s partcularly strong s the assumpton of equal dscountng. There are at least two reason to beleve that future proft s dscounted dfferently by dfferent frms. Frst, some frms may be subject to a less favorable nterest rate than others due to some knd of credt market mperfecton. Second, even f the tme preference s the same across frms, the tme preferences of the managers who run those frms can be dfferent. Some manager may dscount future heavly f she expects to retre or be fred soon. Some manager s preference may be more n lne wth the preference of the frm f she may own more stocks (and stock optons) of the frm. The goal of ths paper s to understand the nature of colluson n the repeated Bertrand competton model when frms are asymmetrc, especally when dfferent frms dscount future profts n dfferent way. 2 We have two man results. Frst we dentfy the crtcal level of dscount factor such that a collusve outcome can be sustaned f and only f the average dscount factor wthn the lowest cost frms s above the crtcal level. More generally, we show that the necessary and suffcent condton for sustanng a colluson at a certan prce (or more) s that the average dscount factor of all the frms whose margnal cost s below the prce must be larger than n 1 n, where n s the number of such frms. A more patent frm s wllng to gve up more market shares to more mpatent frms, whose ncentve constrants are then relaxed. So the dstrbuton of dscountng rates matters n general. In our smple settng wth homogeneous good, the mean of dscountng rates among colludng frms determnes the possblty of colluson. Our second result s a characterzaton of all effcent (proft-maxmzng) collusve equlbra when frms dffer only n ther dscountng rates. In effcent equlbra, more mpatent frms gan a larger share of the market at an earler stage and more patent frms gan a larger share at a later stage. Such an ntertemporal substtuton of the market share s subjectve to the ncentve constrant: we cannot assgn 0% share forever even to the most mpatent frm. Hence the equlbrum outcome s not the frst best. Our characterzaton provdes a totally new pcture of colluson, whch s radcally dfferent from the one among symmetrc frms. Frst, the equlbrum market share n any effcent collusve equlbrum changes over tme. More specfcally, the market share dynamcs of each frm can be descrbed by three phases. In the frst phase, a frm has no share of the market, leavng the market to more mpatent frms. In the second phase, the frm enters 2 We assume that heterogeneous dscountng rates are gven exogenously. Of course, t would be nterestng to thnk about a model n whch they are endogenously determned for a varety of reasons. We thnk that our model wth fxed heterogeneous dscountng rates would open a possblty of buldng such a model. 2

3 the market and gans all the rest after leavng more mpatent frms the mnmum amount of statonary market share, whch correspond to the worst statonary collusve equlbrum market share for them. The fnal phase starts when a more patent frm enters the market. In the fnal phase, the frm s marker share drops to the level that corresponds to ts worst statonary collusve equlbrum market share and stays there forever. Secondly, our results delver the unque predcton n the long run. As descrbed above, the equlbrum market share for each frm, except for the most patent frm, converges to ts worst statonary collusve equlbrum market share n any effcent collusve equlbrum. More precsely, every effcent collusve equlbrum converges to the unque statonary collusve equlbrum wthn fnte tme. 3 We know that, wth symmetrc frms, there are many effcent statonary equlbra wth dfferent market shares because how to share the market s rrelevant for effcency. Wth asymmetrc dscountng, however, effcency mposes a sharp restrcton on how the market should be allocated ntertemporally. As a consequence, even though there are many effcent equlbra, the long run market share must be the same across all effcent equlbra. From a more theoretcal perspectve, our results delver new nsghts nto the theory of repeated games wth dfferental dscountng. As revewed brefly next, the major results for repeated games wth dfferental dscountng are restrcted to asymptotc results (.e. frms are nfntely patent) and the two-player case. In our settng, we characterze all the effcent equlbra wth n players for a fxed dscount factor, possbly due to some specal structure of Bertrand competton game. Related Lterature It s not wthout reason that prevous works have focused on the symmetrc model. Frst, there s the ssue of equlbrum selecton as mentoned. There are always many equlbra - hence there s always the ssue of equlbrum selecton - n repeated games. The model of dynamc Bertrand competton s no excepton. For symmetrc models, t mght make sense to focus on the symmetrc (and effcent) equlbrum, possbly as a focal pont. However, t s not clear whch equlbrum would be a focal pont when frms are asymmetrc. Secondly, the theory of repeated games wth dfferental dscountng s stll at ts development stage. For these reasons, there are not many works that study colluson among heterogeneous frms. In our vew, ths fact lmts the scope of applcatons of the repeated Bertrand competton model. 3 The tme to reach the effcent statonary collusve equlbrum s bounded across all effcent collusve equlbra for a gven profle of dscountng rates. 3

4 One notable excepton s Harrngton (1989) [4]. It shows that a statonary collusve equlbrum can be sustaned wth dfferental dscountng f and only f the average dscount factor exceeds some crtcal level. Our frst result bulds on and mproves on ths result. We provde a more complete characterzaton regardng the possblty of colluson by consderng all equlbra ncludng nonstatonary ones. 4 Clearly t s mportant to consder nonstatonary equlbra because almost all statonary equlbra are not effcent wth dfferental dscountng as our second result shows. Another dfference between our paper and [4] s that we obtan the unque equlbrum n the long run. To cope wth the ssue of multple statonary equlbra, Harrngton [4] uses a barganng soluton to select one equlbrum. On the other hand, we show that the long run equlbrum behavor s the same across all effcent equlbra. Thus we do not need to rely on any equlbrum selecton crteron other than effcency as long as we are concerned wth the long-run outcome. The semnal contrbuton n the theory of repeated game wth dfferental dscountng s Lehrer and Pauzner (1999) [5]. It studes a general two-player repeated game wth dfferental dscountng and shows that the set of feasble payoffs s larger than the convex hull of the underlyng stage game payoffs because players can mutually beneft from tradng payoffs across tme. They also characterze the lmt equlbrum payoff set as dscount factors go to 1 keepng ther rato fxed. In partcular, they show that there s some ndvdually ratonal and feasble payoff that cannot be sustaned n equlbrum no matter how patent the players are. There are some recent contrbutons n the theory of repeated games wth dfferental dscountng. Chen [1] and Gueron et. al [3] study stage games wth one dmensonal payoffs. Sugaya [6] proves a folk theorem for repeated games wth mperfect montorng and wth dfferental dscountng. Fong and Surt [2] study repeated prsoner s dlemma games wth dfferental dscountng and wth sde payments. Ths paper seems to be partcularly related to our paper because we use market share as a way to transfer utlty. Ths paper s organzed as follows. We descrbe the model n detal n the next secton. In secton 3, we prove our frst result regardng the crtcal average dscount factor. secton 4, we characterze effcent equlbra. We conclude and dscuss potental extensons of our results n the last secton. Most of the proofs are relegated to the appendx. 4 Snce a collusve outcome can be sustaned by a statonary equlbrum when the average dscount factor exceeds the crtcal level, the crucal step for our result s to show that no nonstatonary collusve equlbrum exsts when the average dscount factor s below the same crtcal level In 4

5 2 Model of Repeated Bertrand Competton wth Dfferental Dscountng Ths secton descrbes the basc structure of our model, an nfntely repeated Bertrand game. In what follows, we frst defne the stage game, then construct the nfntely repeated game. The man features of the stage game are the followngs. The players are n 2 frms represented by the numbers I = {1, 2,..., n}. They offer a homogeneous product whose market demand s characterzed by contnuous functon D : R + R +. Each frm has a lnear cost functon C : R + R + gven by C (q ) = c q, where I, c 0 s the margnal cost, and q ndcates the quantty produced by frm. We suppose that c 1 c 2... c n wthout loss of generalty and denote I = { I : c = c 1 } and n = #(I ). We assume that n 2. Hence, n one-shot Bertrand competton, the market prce would be c 1 and no frm would make any proft. It s assumed that the demand functon satsfes the followng regularty condtons: D s decreasng on (0, ); there exsts the monopoly prce for each frm: p m > c for frm that maxmzes p (D(p) c ). We assume that the margnal costs are not very dfferent: even the hghest cost c n s less than p m 1. Ths mples that pm for any, j I. 5 At the begnnng of a stage game, frms make prce decsons and suggest how to allocate output quotas n case of a draw n prces. If a frm charges a prce that s hgher than a prce charged by another frm, then the frm s market share s 0. The frm that charges the lowest prce must produce enough output to satsfy the market demand. In case there are more than one frm that charges the lowest prce, the market s allocated among those frms accordng to ther suggestons. Formally, frm s pure acton s gven by a 2-tuple a = (p, r ) A, where p s the prce choce, r reflects frm s request of market share n case of te. Hence A = R + [0, 1] s the set of pure actons avalable for player. The set of pure acton profles s A = I A. Frm s proft functon π : A R can be wrtten as π [a] = D(p )(p c ) f p < p, r R D(p )(p c ) f p = p and R 0, 1 Î D(p )(p c ) f p = p and R = 0, 0 f p > p, where p = mn j p j, Î = { I : p = mn j I p j }, and R = j Î r j. 5 If the margnal of some frm s too hgh, t s lkely that the presence of such a frm s rrelevant for our analyss. > c j 5

6 Gven the stage game descrbed above, we now defne the nfntely repeated game. Bascally, we adopt a dscrete tme model n whch the prevous stage game s played n each of the perods t N. The dstngushng feature of our dynamc Bertrand competton model s that the players have dfferent dscount factors gven by δ (0, 1), I. The set of possble hstores n perod t s gven by H t = A t 1, where A 0 ndcates the empty set, and A t denotes the t-fold product of A. A perod t-hstory s thus a lst of t 1 acton profles. We suppose perfect montorng throughout,.e., at the end of each perod, all players observe the acton profle chosen n all the prevous perods. Settng H = t N H t, a pure strategy for frm s defned as a mappng σ : H A, and consequently, a strategy profle s gven by σ = (σ ) I. Each strategy profle σ nduces an nfnte sequence of acton profles a(σ) = (a t (σ)) t N A, where a t (σ) A denotes the acton profle nduced by σ n perod t. We call the sequence a(σ) outcome path (or more smply, outcome) generated by a strategy profle σ. Fnally, for a gven strategy profle σ, and ts correspondng outcome path a(σ) = (a t (σ)) t N, the tme-average repeated game payoff for frm at tme t s U,t [a(σ)] = (1 δ ) τ=t δ τ t π [a τ (σ)]. In the followng sectons, we wll just focus on subgame perfect equlbrum solutons, and we wll lmt our attenton to pure strategy equlbra. 3 Crtcal Average Dscount Factor for Colluson In ths secton, we derve a necessary and suffcent condton to sustan a collusve equlbrum outcome. We say that the frms are colludng when there s at least one perod n whch the equlbrum outcome s not a compettve one,.e. when there s at least one frm that makes postve proft n some perod. We formalze ths as follows. Defnton 1. An outcome a = (a t ) t N s consdered a collusve outcome f and only f there exsts t N such that π (a t ) > 0 for some I. A collusve equlbrum s a subgame perfect equlbrum that generates a collusve outcome. Then we can obtan the followng sharp characterzaton, whch says that a collusve outcome can be sustaned f and only f the average dscount factor among the lowest cost frms s above some threshold. 6

7 Theorem 3.1. There exsts a collusve equlbrum f and only f I δ n 1. n n Proof. See the appendx. When the frms are symmetrc, there exsts a collusve equlbrum f and only f δ n 1 n. Thus our result s a substantal generalzaton of ths well-known result to the case wth heterogeneous dscountng and costs. It follows from the result n [4] that n 1 n s the crtcal threshold to support a collusve outcome by a statonary collusve equlbrum,.e. an equlbrum n whch each frm keeps a certan level of market share every perod and the prce s always the same. Take any prce p strctly between the mnmum cost c = mn I c and the next smallest cost. There exsts a statonary collusve equlbrum by the lowest cost frms n whch the market prce s always p and frm ( I ) gans share α [0, 1] of the jont proft π n every perod f the followng nequaltes are satsfed for all I. (1 δ )π α π By dvdng both sdes by π and summng up these nequaltes across the frms, t can be shown that such α, I exsts f and only f the average dscount factor among the lowest cost frms s larger than or equal to n 1 n. A more dffcult part of the proof s to show that colluson s mpossble when the average dscount factor s less than n 1 n, even f nonstatonary equlbra are consdered. In nonstatonary equlbrum, t s possble to transfer market shares over tme to generate larger contnuaton profts n the future, whch may enable the frms to sustan colluson. It turns out that such transfer does not work. To mprove effcency, t s necessary to let less patent frms to gan more market shares frst and let more patent frms to gan more shares later. Intutvely, such an arrangement s n conflct wth less patent frms ncentve constrants n later perods. Here s a sketch of our formal proof. We assume that the margnal cost s the same across all frms to smplfy our exposton. Frm s ncentve constrant n perod t s gven by the equalty U,t = (1 δ ) π ( a t ) + δ U,t+1 = (1 δ ) π ( a t) + η,t 7

8 where a t s the acton profle n perod t, π ( a t) = π ( a t ) s the jont proft n perod t, U,t+1 s frm s contnuaton proft from perod t + 1 on, and η,t 0 s a slack varable (frm s ncentve constrant s bndng n perod t f and only f η,t = 0). Note that each frm gans the same equlbrum jon proft by prce-cuttng because the cost s assumed to be the same. Snce ths equalty holds n every perod, we can replace U,t+1 wth (1 δ ) π ( a t+1) + η,t+1 and dvde both sdes by 1 δ to obtan π ( a t ) + δ π ( a t+1) = π ( a t) + η,t δ η,t+1 1 δ. Summng up these equaltes across the frms, we have the followng equaton regardng π ( a t) : π ( a t+1) = n 1 I δ π ( a t) + 1 I δ u,t, I where u,t = η,t δ η,t+1 1 δ. The coeffcent of π ( a t) s larger than 1 f and only f the average dscount factor s less than n 1 n. In fact, we can show that, when the jont proft s strctly postve n some perod, the sequence π ( a t), t = 1, 2,.. must dverge to nfnty, whch s a contradcton. To prove ths formally, however, we need to examne carefully the behavor of I u,t, t = 1, 2, 3,... A collusve equlbrum we construct uses a prce between the lowest cost and the second lowest cost, so t s not very proftable when ths dfference between them s small. such a case, the lowest cost frms would prefer to nclude the second lowest cost frm(s) n ther coalton to rase the equlbrum prce. In Our result can be easly generalzed to accommodate such possblty. Let p be any prce. Let I(p) be the set of frms such that c p f and only f I(p) and I(p) = n (p). Call a subgame perfect equlbrum p-collusve equlbrum f the equlbrum prce s always at least as large as p. We can prove the followng generalzaton of the above result. Theorem 3.2. For any 0 < p p m, there exsts a p-collusve equlbrum f and only f I(p) δ n (p) n (p) 1. n (p) The proof s almost the same, hence omtted. Comment 8

9 When I = 1,.e. there s the unque lowest cost frm, Theorem 2 stll holds. But we need to rely on a less natural punshment. The assumpton I 2 guarantees that any devaton from a collusve outcome s punshed by Nash reverson wth 0 proft forever. If I = 1, the 0 proft equlbrum requres that there are at least two frms chargng c 1, but frm 1 serves the whole market (r 1 = 1, r = 0 for all 1). 4 Characterzaton of Effcent Collusve Equlbra In ths secton, we characterze effcent collusve equlbra wth dfferental dscountng rates. We assume that the margnal cost s the same across frms and normalze t to 0. Then the monopoly prce can be determned wthout any ambguty. Let p m be the monopoly prce and π m be the monopoly proft. We also assume that 0 < δ 1 < δ 2 <... < δ n 1 < δ n < 1 for the sake of smplcty. The result can be easly extended to the case where the dscountng factors of some frms are the same. Let π,t, = 1,..., n, t N be a sequence of profts assocated wth any collusve equlbrum. By defnton, they satsfy the followng ncentve compatblty condton n every perod: (1 δ ) π t U,t where U,t s frm s equlbrum contnuaton proft n the begnnng of perod t and π t = π,t. On the other hand, t s clear that any sequence of proft profles that satsfy those condtons are generated by a collusve equlbrum. Hence we use such a sequence of proft profles to descrbe any collusve equlbrum. A collusve equlbrum s effcent f there s no subgame perfect equlbrum that makes every frm better off weakly and some strctly. Observe that π t s always n (0, π m ] for any effcent collusve equlbrum. π t cannot exceed the monopoly proft by defnton. If π t < 0, then we can construct a more effcent equlbrum by just droppng perod t. We know that there exsts a statonary collusve equlbrum wth monopoly prce f and only f n =1 δ n n 1 n. When the average dscount factor s strctly larger than n 1 there s a range of market shares that can be supported by statonary collusve equlbrum. Let π be frm s per perod proft n the worst statonary collusve equlbrum proft for frm. Note that π = (1 δ ) π m by the ncentve compatblty condton. We assume n =1 δ n > n 1 n for the rest of ths secton. We frst prove that, n any effcent collusve equlbrum, the jont proft must be strctly ncreasng untl t reaches the monopoly proft and stays there forever. 9 n,

10 We start wth the followng lemma. Lemma 4.1. Consder any effcent collusve equlbrum where, for some t 1, π t+1 < π m and there s a frm such that U,t+1 > (1 δ )π t+1 and π,t+1 > 0. Then π t+1 πt δ n. Proof. Defne Ĩt+1 = { I : U,t+1 = (1 δ )π t+1 }, whch s not empty (otherwse the jont proft can be ncreased to mprove effcency). Suppose that π t+1 < πt δ n. Then π,t π t δ π t+1 π t δ n π t+1 > 0 for all Ĩt+1. Consequently, the profts can be perturbed as follows: π,t = π,t δ ε and π,t+1 = π,t+1 + ε, for Ĩt+1; whereas π,t = π,t+ Ĩt+1 δ ε and π,t+1 = π,t+1 ( Ĩt+1 1)ε. Snce π t+1 < π m and π,t+1 > 0, ths new allocaton s feasble and ncentve compatble for ε > 0 small enough. Moreover, as Ĩt+1 δ > Ĩt+1 1, t also Pareto-domnates the ntal one. Ths s a contradcton. The next theorem proves a strong monotoncty property for effcent collusve equlbra. Theorem 4.1. For any effcent collusve equlbrum, there exsts T such that π t < π t+1 for t = 1,..., T 1 and π t = π m for any t T. Furthermore, ths T s bounded across all effcent collusve equlbra. Proof. Take any effcent collusve equlbrum. Let π t (0, π m ] be a jont proft n any perod t. We assume that π t > δ n π t+1 and π t+1 < π m, and derve a contradcton. If those two condtons are satsfed, then t must be the case that π t+1 = π,t+1 by Lemma Ĩt Therefore, there s j Ĩt+1 such that π j,t+1 > (1 δ j )π t+1, otherwse, π t+1 = π,t+1 π t+1 (1 δ ) = π t+1 ( Ĩt+1 δ ), Ĩt+1 Ĩt+1 Ĩt+1 but Ĩt+1 δ > Ĩt+1 1. As a result, (1 δ j )π t+2 U j,t+2 < (1 δ j )π t+1. The frst nequalty s derved from the ncentve constrant n perod t + 2, whereas the second one from the fact that π j,t+1 > (1 δ j )π t+1 and U j,t+1 = (1 δ j )π t+1. Then, π t+1 > π t+2. We can proceed n a smlar manner to obtan π t+k > π t+k+1 for every k 1, whch contradcts the effcency assumpton. Hence t must be the case that ether π t δ n π t+1 or and π t+1 = π m. Clearly ths mples that there s T such that π t < π t+1 for t = 1,..., T 1 and π t = π m for any t T. Fnally we prove that ths T s bounded across all effcent equlbra. For any gven T, each frm s proft per perod s at most δ T 1 n π m for the frst T T perods for any T T. If T s large, then frm s payoff s less than π. Such payoff profle s Pareto-domnated by any statonary collusve equlbrum. 10

11 Next we provde an (almost) complete characterzaton of effcent collusve equlbra. Consder any effcent collusve equlbrum where frm s equlbrum proft exceeds π. Then every frm s ncentve constrant s not bndng n the frst perod, hence the equlbrum jont proft must be π m n the frst perod. Gven our monotoncty result, ths mples that the equlbrum prce s always p m for ths class of effcent collusve equlbra. We call such collusve equlbrum p m -effcent collusve equlbrum. The next theorem characterzes the structure of p m -effcent collusve equlbrum. Observe that ths characterzaton s a complete characterzaton of the asymptotc behavor of all effcent collusve equlbra, because every effcent collusve equlbrum converges to some p m -effcent collusve equlbrum eventually wthn fnte tme by our prevous result. In p m -effcent collusve equlbrum, more patent frms lend the market share ntally to more mpatent frms. However, the ablty of mpatent frms to pay back the market share s lmted by the requrement that each frm s proft cannot be lower than ts worst statonary equlbrum proft π. Theorem 4.2. Every p m -effcent collusve equlbrum has the followng structure: there exsts t 1 t 2..., t n 1 such that, for every, 1. π,t = 0 for every t < t 1 2. π,t [0, π m 1 3. π,t = π m 1 4. π,t [ π, π m 1 π h ] for t = t 1 π h for t = t 1 + 1,..., t 1 5. π,t = π for t > t π h ] for t = t 6. Incentve Constrants n the frst perod [ ( ) { }] δ t 1 1 (1 δ ) π,t 1 + δ δ t 1 t 1 π m π h [ ] +δ t 1 1 (1 δ ) δ t t 1 π,t + δ t t 1 +1 π (1 δ ) π m Furthermore, f there exst (t 1, t 2,..., t n 1 ) and a sequence of proft profles π,t that satsfy the above condtons, then there exsts a correspondng p m -effcent collusve equlbrum that generates them. 11

12 Proof. See the appendx. In words, every p m -effcent collusve equlbrum has the followng propertes. From perod 1 to perod t 1 1, frm 1 gets the whole share. In perod t 1, frm 1 and 2 shares the market where π,t1 π 1. After ths perod, frm 1 s share s gong to be always π 1. From perod t to perod t 2 1, frm 2 gets π m π 1. In perod t 2, frm 2 and 3 shares the market where π,t2 π 2. After ths perod, frm 2 s share s gong to be always π 2. From perod t to perod t 3 1, frm 3 gets π m π 1 π After perod t n 1, frm n gets π m n 1 π h and frm h < n gets π h forever. There are two crtcal perods for frm : t 1 and t. Up to t 1, frm s market share s 0. The perods between t 1 and t s the pay back tme when frm gets all the market share subject to the constrant that each less patent frm h < gans π h. After t, frm s proft s reduced to π and stay there forever. It may be the case that there s some overlap: t k = t k+1 =,..., = t m = t for some m > k. Note that π,t such a case. π for = t k, t k+1,..., t m 1 n Comment One mplcaton of our theorem s that there exsts the unque effcent statonary collusve equlbrum, to whch every effcent collusve equlbrum converges. Ths s the statonary collusve equlbrum where the prce s p m, frm s market share s π for = 1,..., n 1 and frm n s market share s π m =1,...,n 1 π n every perod, whch corresponds to the worst statonary collusve equlbrum for frm = 1,..., n 1 (and the best one for frm n). All the other effcent collusve equlbra must be nonstatonary. Our result delvers the unque predcton n the long run wthout any equlbrum selecton. Ths s not the case f we focus on statonary collusve equlbra. 12

13 When δ = δ +1 for some, ther market share s characterzed by smlar condtons: ther market share s 0 before t 1, π,t = π 1,t = π (= π +1 ) after t, and can be somewhat arbtrary between t and t 1 (but we can assume that ther market shares are constant wthn these perods wthout loss of generalty). 5 Concluson and Dscusson In the context of Bertrand prce competton n an nfntely repeated game, ths paper studes collusve behavor among n frms wth asymmetrc dscount factors and asymmetrc margnal costs. We fnd that t s possble to sustan a colluson f and only f the average dscount factor of the lowest cost frm s at least as large as (n 1)/n, where n s the number of the lowest cost frms. Ths paper also studes effcent collusve equlbra among n frms wth dfferental dscountng when the margnal cost s the same across frms. We succeed n characterzng the structure of effcent collusve equlbra. More specfcally, we show the followngs. In any effcent collusve equlbrum, the jont proft must be strctly ncreasng over tme untl t reaches the monopoly proft level wthn fnte tme and stay there forever. Every effcent collusve equlbrum where no frm s payoff s not too low must be a collusve equlbrum wth the monopoly prce n every perod ( p m -effcent collusve equlbrum ). In every p m -effcent collusve equlbrum, a frm s market share s 0 ntally, reaches a peak, then converges to the market share that corresponds to the worst statonary collusve equlbrum wth the monopoly prce (except for the most patent frm). Every effcent collusve equlbrum converges to the unque effcent statonary collusve equlbrum n the long run, where the equlbrum prce s p m, frm s proft per perod s π for = 1,..., n 1 and 1 =1,...,n 1 π for = n n every perod. 13

14 References [1] B. Chen (2008). On Effectve Mnmax Payoffs and Unequal Dscountng, Economcs Letters, 100, 1: [2] Y-F. Fong and J. Surt (2009). The Optmal Degree of Cooperaton n the Repeated Prsoners Dlemma wth Sde Payments, Games and Economc Behavor, 67,1: [3] Y. Guéron, T. Lamadon and C. Thomas (2011). On the Folk Theorem wth Onedmensonal Payoffs and Dfferent Dscount Factors, Games and Economc Behavor, 73, 1: [4] J. Harrngton (1989). Colluson among Asymmetrc Frms: The case of Dfferent Dscount Factors, Internatonal Journal of Industral Organzaton, 7: [5] Lehrer, E. and A. Pauzner (1999). Repeated Games wth Dfferental Tme Preferences, Econometrca, 67, 2: [6] T. Sugaya (2010), Characterzng the Lmt Set of PPE Payoffs wth Unequal Dscountng, mmeo. [7] J. Trole (1988). The Theory of Industral Organzaton, MIT press. 14

15 Appendx Proof of Theorem 3.1 Proof. We already dscussed that there exsts a collusve statonary subgame perfect equlbrum when the nequalty s satsfed. Thus we just need to show that there s no collusve I δ n subgame perfect equlbrum when < n 1 n. By contradcton, begn by assumng that ã = (ã t ) t N s a collusve equlbrum outcome, and wthout loss of generalty, assume that π (ã 1 ) = I π (ã 1 ) > 0. Note frst that for each I, there exsts a bounded nonnegatve sequence {η (t) : t N} defned by η (t) = U,t (ã) (1 δ )π (ã t ). Moreover, snce U,t (ã) = (1 δ )π (ã t ) + δ U,t (ã), we have that (1 δ )π (ã t ) + δ U,t (ã) = (1 δ )π (ã t ) + η (t), and therefore (1 δ )π (ã t ) + δ [(1 δ )π (ã t+1 ) + η (t+1) ] = (1 δ )π (ã t ) + η (t), or equvalently, [ ] [ ] π (ã t ) = π (ã t ) + η(t) δ π (ã t+1 ) + η(t+1). (1 δ ) (1 δ ) Addng up ths nequalty across I and denotng s = I δ, we obtan or more shortly, π (ã t ) = n π (ã t ) s π (ã t+1 ) + η (t) δ η (t+1), (1 δ ) I π (ã t+1 ) = γπ (ã t ) + 1 s where γ = (n 1)/s and u,t = (η (t) I u,t, δ η (t+1) )/(1 δ ). Before proceedng, t s useful to note that γ > 1 and therefore π (ã t+1 ) π (ã t ) + 1 s I u,t for every t N. π (ã 1 ) + 1 s I j=1 t u,j, 15

16 Now consder the seres t j=1 u,j for I, and observe that t maybe wrtten as t j=1 t u,j = η(1) (1 δ ) + j=2 η (j) δ η (t+1) (1 δ ). Snce the equlbrum proft s bounded above for each frm by assumpton, the equlbrum proft s bounded below as well for each frm; otherwse the average dscounted proft s negatve. Ths mples that there exsts M such that 0 η (j) M for all j N and I. Observe that ths mples that the seres t j=2 η(j) must be ether unbounded above or convergng to a fnte (nonnegatve) real number. Suppose frst that j=2 η(j) s unbounded above for some I. On the one hand, we know that I ( t j=1 u,t) s unbounded above, too. On the other hand, we have that π (ã t+1 ) π (ã 1 ) + (1/s ) t I j=1 u,t, whch s a contradcton because the sequence {π (ã t ) : N} s bounded above. Suppose now that j=2 η(j) s fnte for all I. Then we have that η (t) as well as u,t converge to zero for all I. If η (j) = 0 for all I and j N, t follows mmedately that I ( t j=1 u,t) 0. On the other hand, f η (t ) = c > 0 for some I and t N, there exsts T > t such that η (t) < c (1 δ )/(2δ ) for all t T. As a result, we have that t u,j j=1 η(1) (1 δ ) + c c 2, when t > T. As I s a fnte set, there s T N (ndependent of ) such that I ( t j=1 u,t) 0 for all t T, and consequently, π (ã t ) π (ã 1 ) as long as t T. Before proceedng, observe frst that there s t N such that γ t π (a 1 ) > 2M. Secondly, snce I s a fnte set and u,t converge to zero for each I, there exsts T N (ndependent of ) such that T > T and u,t < (s M )/(n tγ t ) for all I and t T. The followng nequalty s a straghtforward mplcaton: π (ã T +t ) = γπ (ã T +t 1 ) + 1 s I u, T +t 1 γπ (ã T +t 1 ) M tγ t, and by nducton, we can prove that π (ã T +t ) γ t π (ã T t 1 M ), tγ t j j=0 16

17 for every t N. Fnally, after replacng t = t n the prevous nequalty, we obtan the desred result: t 1 π (ã T + t ) γ t π (ã T ) > γ t π (ã 1 ) > 2M M. M tγ t j j=0 t 1 j=0 M t The second nequalty follows by π (ã T ) π (ã 1 ) and γ > 1, whereas the last one by γ t π (ã 1 ) > 2M. And obvously, ths s a contradcton because π (ã T + t ) M. Proof of Theorem 4.2 We prove the theorem through a seres of lemmata. Lemma 5.1. For any effcent subgame perfect equlbrum, f frm s ncentve constrant s not bndng n perod t > 1, then π j,t 1 = 0 for every j >. Proof. Suppose not,.e. there exsts some monopoly-prce effcent SPE where frm s ncentve constrant s not bndng n perod t > 1 and π j,t 1 > 0 for some j >. Then there s a perod t > t such that frm s ncentve constrant s not bndng for t, t + 1,..., t and π,t > 0. We can fnd such t, otherwse π,t+1 = π,t+2 =... = 0 (f π,t+1 = 0, then U,t+2 > π hence s ncentve constrant n perod t + 2 s not bndng. If π,t+2 = 0, then U,t+2...). Such a path s not sustanable. Now perturb the proft of frm and j as follows. π,t = π,t + ε, π j,t = π j,t ε, π,t = π,t ε, π j,t = π j,t + ε, We are bascally exchangng frm j s market share n perod t wth frm s market share n perod t,keepng every other frm s proft at the same level. Snce δ < δ j, π j,t > 0 and π,t > 0, we can pck ε, ε > 0 so that frm j s contnuaton payoff n every perod from t to t ncreases and frm s contnuaton payoff n perod t ncreases. So ths allocaton Paretodomnates the orgnal SPE allocaton. Frm j s ncentve constrants are not affected at all. Frm s ncentve constrants n perod t s satsfed by constructon. Fnally, we can 17

18 take ε, ε > 0 small enough so that frm s ncentve constrant from perod t + 1 to t s stll not bndng. So we can construct a more effcent SPE n ths case, a contradcton. Lemma 5.2. For any monopoly-prce effcent subgame perfect equlbrum, f π,t < π, then π j,t = 0 for every j >. Proof. If π,t < π, then U,t+1 > π. Hence frm s ncentve constrant s not bndng n perod t + 1. Then π j,t = 0 for every j > by Lemma 1. Lemma 5.3. For any effcent subgame perfect equlbrum wth π > π, (1)π 1,t π 1 for every t 1 and (2) π 1,t +k = π 1 for any k = 0, 1,... f frm 1 s ncentve constrant s bndng n perod t. Proof. If π 1,t < π 1 for any t, then π j,t = 0 for every j = 2, 3,..., n by Lemma 2. Ths contradcts to π,t = π m. Frm 1 s ncentve constrant s bndng n perod t f and only f U 1,t = π 1. Clearly ths holds f and only f π 1,t +k = π 1 for k = 0, 1, 2... By nducton, a smlar property holds for every frm. Lemma 5.4. For any effcent subgame perfect equlbrum wth π > π, suppose that π h,t+k = π h for every k = 0, 1, 2,... and every h = 1, 2,..., for some t and some I. Then (1) π +1,t+k π +1 for every k = 0, 1, 2,... and (2) π +1,t +k = π +1 for every k = 0, 1, 2... f frm + 1 s ncentve constrant s bndng n perod t t. Proof. Suppose that π +1,t+k < π +1 for any k. Then U +1,t+k+1 > π +1. Hence frm + 1 s ncentve constrant s not bndng n perod t + k + 1. Then π j,t+k+1 = 0 for every j > + 1 by Lemma 1. However, π h,t+k+1 = +1 π h,t+k+1 = π h + π +1,t+k+1 < +1 π h π, h whch s a contradcton. Ths proves (1). As for (2), frm + 1 s ncentve constrant s bndng n perod t t f and only f U,t = π. By Lemma 4, ths holds f and only f π +1,t +k = π +1 for every k = 0, 1, 2... Now we can prove Theorem

19 n Proof. In perod 1, we have π 1 such that (1) π h,1 = π m and (2) π h,1 π h for h = [ 1, 2,..., h 1 1, π h1,1 0, π m h ] 1 1 π h, and π h,1 = 0 for h > h 1 for some h 1 1 by.lemma 2. By Lemma 1, the ncentve constrant must be bndng for h = 1, 2,..., h 1 1 n perod 2. By Lemma 3 and Lemma 4, π h,1+k = π h for h = 1, 2,..., h 1 1 for the rest of the game (k = 1, 2,...). n In perod 2, we have π 2 such that (1) π h,2 = π m, (2) π h,2 = π h for h = 1, 2,..., h 1 1 [ (by the prevous step), (3) π h,2 π h for h = h 1, h 1 + 1,..., h 2 1, π h2,2 0, π m h ] 2 1 π h, and π h,2 = 0 for h > h 2 for h 2 for some h 2 h 1 by.lemma 4. By Lemma 1, the ncentve constrant must be bndng for h = h 1,..., h 2 1 n perod 3. By Lemma 3 and Lemma 4, π h,2+k = π h for h = h 1,..., h 2 1 for the rest of the game (k = 1, 2,...) and so on... Ths proves 1-6 n the statement of the theorem. On the other hand, suppose that there exst (t 1, t 2,..., t n 1 ) and a sequence of proft profles π,t that satsfy 1-6. It s clear that ths corresponds to some monopoly-prce SPE. So we just show that t s an effcent equlbrum. Suppose not. Then there exsts a Paretosuperor monopoly-prce effcent SPE, whch of course satsfes 1-6. Let ( t 1, t 2,..., t n 1 ) be the correspondng crtcal perods and π,t be the assocated sequence of proft profles. Snce ths equlbrum s more effcent than the former one, t must be the case that ether (1) t 1 < t 1 or (2) t 1 = t 1 and π 1,t1 π 1,t1. In ether case, t must be the case that, for frm 2, ether (1) t 2 < t 2 or (2) t 2 = t 2 and π 1,t2 π 1,t2. By nducton, ether (1) or (2) holds up to frm n 1. Then frm n s average proft gven π n,t s hgher than frm n s average proft gven π n,t. Ths s a contradcton to the assumpton that the latter equlbrum s more effcent. 19

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

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

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

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

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

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

More information

Assortment Optimization under MNL

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

More information

Pricing and Resource Allocation Game Theoretic Models

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

More information

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

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

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

More information

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

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

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

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

Module 9. Lecture 6. Duality in Assignment Problems

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

More information

(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

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

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

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

More information

3.2. Cournot Model Cournot Model

3.2. Cournot Model Cournot Model Matlde Machado Assumptons: All frms produce an homogenous product The market prce s therefore the result of the total supply (same prce for all frms) Frms decde smultaneously how much to produce Quantty

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

Finite State Equilibria in Dynamic Games

Finite State Equilibria in Dynamic Games Fnte State Equlbra n Dynamc Games Mchhro Kandor Faculty of Economcs Unversty of Tokyo Ichro Obara Department of Economcs UCLA June 21, 2007 Abstract An equlbrum n an nfnte horzon game s called a fnte state

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

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

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

Games of Threats. Elon Kohlberg Abraham Neyman. Working Paper

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

More information

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

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

More information

Equilibrium with Complete Markets. Instructor: Dmytro Hryshko

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

More information

e - c o m p a n i o n

e - c o m p a n i o n OPERATIONS RESEARCH http://dxdoorg/0287/opre007ec e - c o m p a n o n ONLY AVAILABLE IN ELECTRONIC FORM 202 INFORMS Electronc Companon Generalzed Quantty Competton for Multple Products and Loss of Effcency

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

Credit Card Pricing and Impact of Adverse Selection

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

More information

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

University of California, Davis Date: June 22, 2009 Department of Agricultural and Resource Economics. PRELIMINARY EXAMINATION FOR THE Ph.D. Unversty of Calforna, Davs Date: June 22, 29 Department of Agrcultural and Resource Economcs Department of Economcs Tme: 5 hours Mcroeconomcs Readng Tme: 2 mnutes PRELIMIARY EXAMIATIO FOR THE Ph.D. DEGREE

More information

Mixed Taxation and Production Efficiency

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

More information

Infinitely Split Nash Equilibrium Problems in Repeated Games

Infinitely 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 information

PROBLEM SET 7 GENERAL EQUILIBRIUM

PROBLEM 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 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

Appendix for Causal Interaction in Factorial Experiments: Application to Conjoint Analysis

Appendix 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 information

Problem Set 9 Solutions

Problem 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 information

CS294 Topics in Algorithmic Game Theory October 11, Lecture 7

CS294 Topics in Algorithmic Game Theory October 11, Lecture 7 CS294 Topcs n Algorthmc Game Theory October 11, 2011 Lecture 7 Lecturer: Chrstos Papadmtrou Scrbe: Wald Krchene, Vjay Kamble 1 Exchange economy We consder an exchange market wth m agents and n goods. Agent

More information

MMA and GCMMA two methods for nonlinear optimization

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

More information

Cournot Equilibrium with N firms

Cournot Equilibrium with N firms Recap Last class (September 8, Thursday) Examples of games wth contnuous acton sets Tragedy of the commons Duopoly models: ournot o class on Sept. 13 due to areer Far Today (September 15, Thursday) Duopoly

More information

n α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0

n α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0 MODULE 2 Topcs: Lnear ndependence, bass and dmenson We have seen that f n a set of vectors one vector s a lnear combnaton of the remanng vectors n the set then the span of the set s unchanged f that vector

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

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

Volume 29, Issue 4. Incomplete third-degree price discrimination, and market partition problem. Yann Braouezec ESILV

Volume 29, Issue 4. Incomplete third-degree price discrimination, and market partition problem. Yann Braouezec ESILV Volume 29, Issue 4 Incomplete thrd-degree prce dscrmnaton, and market partton problem Yann Braouezec ESILV Abstract We ntroduce n ths paper the "ncomplete" thrd-degree prce dscrmnaton, whch s the stuaton

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

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

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

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

Complete subgraphs in multipartite graphs

Complete 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 information

Maximizing the number of nonnegative subsets

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

More information

U.C. Berkeley CS294: Spectral Methods and Expanders Handout 8 Luca Trevisan February 17, 2016

U.C. Berkeley CS294: Spectral Methods and Expanders Handout 8 Luca Trevisan February 17, 2016 U.C. Berkeley CS94: Spectral Methods and Expanders Handout 8 Luca Trevsan February 7, 06 Lecture 8: Spectral Algorthms Wrap-up In whch we talk about even more generalzatons of Cheeger s nequaltes, and

More information

A SURVEY OF PROPERTIES OF FINITE HORIZON DIFFERENTIAL GAMES UNDER ISAACS CONDITION. Contents

A SURVEY OF PROPERTIES OF FINITE HORIZON DIFFERENTIAL GAMES UNDER ISAACS CONDITION. Contents A SURVEY OF PROPERTIES OF FINITE HORIZON DIFFERENTIAL GAMES UNDER ISAACS CONDITION BOTAO WU Abstract. In ths paper, we attempt to answer the followng questons about dfferental games: 1) when does a two-player,

More information

Kernel Methods and SVMs Extension

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

More information

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

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

Price competition with capacity constraints. Consumers are rationed at the low-price firm. But who are the rationed ones? Prce competton wth capacty constrants Consumers are ratoned at the low-prce frm. But who are the ratoned ones? As before: two frms; homogeneous goods. Effcent ratonng If p < p and q < D(p ), then the resdual

More information

On the Multicriteria Integer Network Flow Problem

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

More information

Lecture 17 : Stochastic Processes II

Lecture 17 : Stochastic Processes II : Stochastc Processes II 1 Contnuous-tme stochastc process So far we have studed dscrete-tme stochastc processes. We studed the concept of Makov chans and martngales, tme seres analyss, and regresson analyss

More information

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

How Strong Are Weak Patents? Joseph Farrell and Carl Shapiro. Supplementary Material Licensing Probabilistic Patents to Cournot Oligopolists * How Strong Are Weak Patents? Joseph Farrell and Carl Shapro Supplementary Materal Lcensng Probablstc Patents to Cournot Olgopolsts * September 007 We study here the specal case n whch downstream competton

More information

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

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

More information

ECE559VV Project Report

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

More information

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

Computing Correlated Equilibria in Multi-Player Games

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

More information

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

Mergers among leaders and mergers among followers. Abstract

Mergers among leaders and mergers among followers. Abstract Mergers among leaders and mergers among followers John S. Heywood Unversty of Wsconsn - Mlwaukee Matthew McGnty Unversty of Wsconsn-Mlwaukee Abstract We are the frst to confrm that suffcent cost convexty

More information

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

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

More information

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

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

More information

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

DISCRETE TIME ATTACKER-DEFENDER GAME

DISCRETE TIME ATTACKER-DEFENDER GAME JP Journal of Appled Mathematcs Volume 5, Issue, 7, Pages 35-6 7 Ishaan Publshng House Ths paper s avalable onlne at http://www.phsc.com DISCRETE TIME ATTACKER-DEFENDER GAME Faculty of Socal and Economc

More information

9 Derivation of Rate Equations from Single-Cell Conductance (Hodgkin-Huxley-like) Equations

9 Derivation of Rate Equations from Single-Cell Conductance (Hodgkin-Huxley-like) Equations Physcs 171/271 - Chapter 9R -Davd Klenfeld - Fall 2005 9 Dervaton of Rate Equatons from Sngle-Cell Conductance (Hodgkn-Huxley-lke) Equatons We consder a network of many neurons, each of whch obeys a set

More information

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

Managing Capacity Through Reward Programs. on-line companion page. Byung-Do Kim Seoul National University College of Business Administration Managng Caacty Through eward Programs on-lne comanon age Byung-Do Km Seoul Natonal Unversty College of Busness Admnstraton Mengze Sh Unversty of Toronto otman School of Management Toronto ON M5S E6 Canada

More information

1 Derivation of Rate Equations from Single-Cell Conductance (Hodgkin-Huxley-like) Equations

1 Derivation of Rate Equations from Single-Cell Conductance (Hodgkin-Huxley-like) Equations Physcs 171/271 -Davd Klenfeld - Fall 2005 (revsed Wnter 2011) 1 Dervaton of Rate Equatons from Sngle-Cell Conductance (Hodgkn-Huxley-lke) Equatons We consder a network of many neurons, each of whch obeys

More information

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

College 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 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

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

Case A. P k = Ni ( 2L i k 1 ) + (# big cells) 10d 2 P k. THE CELLULAR METHOD In ths lecture, we ntroduce the cellular method as an approach to ncdence geometry theorems lke the Szemeréd-Trotter theorem. The method was ntroduced n the paper Combnatoral complexty

More information

BOUNDEDNESS OF THE RIESZ TRANSFORM WITH MATRIX A 2 WEIGHTS

BOUNDEDNESS OF THE RIESZ TRANSFORM WITH MATRIX A 2 WEIGHTS BOUNDEDNESS OF THE IESZ TANSFOM WITH MATIX A WEIGHTS Introducton Let L = L ( n, be the functon space wth norm (ˆ f L = f(x C dx d < For a d d matrx valued functon W : wth W (x postve sem-defnte for all

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

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

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

ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EQUATION

ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EQUATION Advanced Mathematcal Models & Applcatons Vol.3, No.3, 2018, pp.215-222 ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EUATION

More information

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity

Week3, 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 information

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

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

More information

Expected Value and Variance

Expected Value and Variance MATH 38 Expected Value and Varance Dr. Neal, WKU We now shall dscuss how to fnd the average and standard devaton of a random varable X. Expected Value Defnton. The expected value (or average value, or

More information

Lecture Notes, January 11, 2010

Lecture Notes, January 11, 2010 Economcs 200B UCSD Wnter 2010 Lecture otes, January 11, 2010 Partal equlbrum comparatve statcs Partal equlbrum: Market for one good only wth supply and demand as a functon of prce. Prce s defned as the

More information

2.3 Nilpotent endomorphisms

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

More information

Collusion in repeated auctions: a simple dynamic mechanism

Collusion in repeated auctions: a simple dynamic mechanism Colluson n repeated auctons: a smple dynamc mechansm Wouter Vergote a,b a CEREC, Facultés unverstares Sant-Lous, Boulevard du Jardn Botanque 43, B-1 Brussels, Belgum. b CORE, Unversté catholque de Louvan,

More information

An (almost) unbiased estimator for the S-Gini index

An (almost) unbiased estimator for the S-Gini index An (almost unbased estmator for the S-Gn ndex Thomas Demuynck February 25, 2009 Abstract Ths note provdes an unbased estmator for the absolute S-Gn and an almost unbased estmator for the relatve S-Gn for

More information

Lecture Notes on Linear Regression

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

More information

Supporting Information for: Two Monetary Models with Alternating Markets

Supporting Information for: Two Monetary Models with Alternating Markets Supportng Informaton for: Two Monetary Models wth Alternatng Markets Gabrele Camera Chapman Unversty & Unversty of Basel YL Chen St. Lous Fed November 2015 1 Optmal choces n the CIA model On date t, gven

More information

6.854J / J Advanced Algorithms Fall 2008

6.854J / J Advanced Algorithms Fall 2008 MIT OpenCourseWare http://ocw.mt.edu 6.854J / 18.415J Advanced Algorthms Fall 2008 For nformaton about ctng these materals or our Terms of Use, vst: http://ocw.mt.edu/terms. 18.415/6.854 Advanced Algorthms

More information

Supporting Materials for: Two Monetary Models with Alternating Markets

Supporting Materials for: Two Monetary Models with Alternating Markets Supportng Materals for: Two Monetary Models wth Alternatng Markets Gabrele Camera Chapman Unversty Unversty of Basel YL Chen Federal Reserve Bank of St. Lous 1 Optmal choces n the CIA model On date t,

More information

Random Walks on Digraphs

Random Walks on Digraphs Random Walks on Dgraphs J. J. P. Veerman October 23, 27 Introducton Let V = {, n} be a vertex set and S a non-negatve row-stochastc matrx (.e. rows sum to ). V and S defne a dgraph G = G(V, S) and a drected

More information

A note on almost sure behavior of randomly weighted sums of φ-mixing random variables with φ-mixing weights

A note on almost sure behavior of randomly weighted sums of φ-mixing random variables with φ-mixing weights ACTA ET COMMENTATIONES UNIVERSITATIS TARTUENSIS DE MATHEMATICA Volume 7, Number 2, December 203 Avalable onlne at http://acutm.math.ut.ee A note on almost sure behavor of randomly weghted sums of φ-mxng

More information

Feature Selection: Part 1

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

More information

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

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

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

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

EEL 6266 Power System Operation and Control. Chapter 3 Economic Dispatch Using Dynamic Programming EEL 6266 Power System Operaton and Control Chapter 3 Economc Dspatch Usng Dynamc Programmng Pecewse Lnear Cost Functons Common practce many utltes prefer to represent ther generator cost functons as sngle-

More information

Yong Joon Ryang. 1. Introduction Consider the multicommodity transportation problem with convex quadratic cost function. 1 2 (x x0 ) T Q(x x 0 )

Yong Joon Ryang. 1. Introduction Consider the multicommodity transportation problem with convex quadratic cost function. 1 2 (x x0 ) T Q(x x 0 ) Kangweon-Kyungk Math. Jour. 4 1996), No. 1, pp. 7 16 AN ITERATIVE ROW-ACTION METHOD FOR MULTICOMMODITY TRANSPORTATION PROBLEMS Yong Joon Ryang Abstract. The optmzaton problems wth quadratc constrants often

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

Portfolios with Trading Constraints and Payout Restrictions

Portfolios with Trading Constraints and Payout Restrictions Portfolos wth Tradng Constrants and Payout Restrctons John R. Brge Northwestern Unversty (ont wor wth Chrs Donohue Xaodong Xu and Gongyun Zhao) 1 General Problem (Very) long-term nvestor (eample: unversty

More information

Perron Vectors of an Irreducible Nonnegative Interval Matrix

Perron 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 information