Price Competition in Multi-sided Markets *

Size: px
Start display at page:

Download "Price Competition in Multi-sided Markets *"

Transcription

1 Prce Competton n Mult-sded Markets * Guofu Tan Junje Zhou July 27, 207 Abstract Ths paper studes a general model of prce competton among platforms offerng dfferentated servces n mult-sded markets. We ncorporate a general form of both wthn-sde and cross-sde externaltes nto a dscrete choce model of random utlty maxmzaton by consumers on each sde of the markets. We consder a two-stage game n whch the platforms choose prces (or user fees) smultaneously n the frst stage, followed by consumers on all sdes smultaneously decdng whch platform to jon (sngle-homng) n the second stage. We show that n a symmetrc settng wth full market coverage, there exsts a symmetrc equlbrum n prces and the equlbrum prce on each sde follows a smple rule: The prce equals the cost, plus a mark-up due to product dfferentaton, mnus a subsdy due to cross-sde externaltes. The subsdy to each sde accounts for the degree of the aggregate margnal externaltes of that sde mposed on all the other sdes. As competton among platforms ncreases, both the product dfferentaton effect and the crosssubsdy are shown to decrease. As such, the prce on one sde can decrease whle the prce on another sde may ncrease wth the number of platforms. We also dscuss the ncentves for platforms to merge and the extent of excessve free entry of platforms nto the markets as compared to the socal optmum. We further compare unform prcng rule wth dscrmnatory prcng across dfferent sdes of the markets and fnd that the average prce across sdes under the dscrmnatory prcng s hgher than the unform prce when the externaltes are small or when the number of platforms s large. The mpacts of consumers outsde optons on the equlbrum prces are also studed. *We are grateful for comments from Alessandro Pavan, Andrew Rhodes, Adam Szedl, Alex Whte, Adam Wong, Julan Wrght, Jdong Zhou and semnar partcpants at CUHK, HKU, HKUST, Melbourne, Monash, NUS, Shangha Jaotong Unversty, Shangha Mcroeconomcs Workshop, SHUFE, SoCal NEGT Symposum, Wuhan Unversty, and Zhejang IO Workshop. Department of Economcs, Unversty of Southern Calforna. guofutan@usc.edu Department of Economcs, Natonal Unversty of Sngapore. ecszjj@nus.edu.sg Electronc copy avalable at:

2 JEL classfcaton: L3, L4. Keywords: mult-sded markets, platform prce competton, cross subsdzaton, dscrete choce models, free entry, merger ncentves, dscrmnatory prcng, unform prcng, outsde opton. Introducton In ths paper, we study a general model of prce competton among platforms provdng dfferentated servces to customers n mult-sded markets. We ncorporate a general form of cross-sde externaltes nto a dscrete choce model of random utlty maxmzaton by customers on each sde of the markets and study the mpacts of platform competton on the equlbrum prces and customer partcpaton rates. There are numerous examples of platforms provdng servces to allow multple groups (or sdes) of customers to nteract, where one group s beneft from jonng a platform depends on the szes of the other groups that also jon the platform (.e., crosssde externaltes). Examples of ths nclude Internet companes such as Albaba, ebay, Facebook, Uber. Albaba and ebay facltate e-commerce transactons between buyers and sellers; Uber connects rders and drvers through ts rde-sharng platform; Facebook can be vewed as a three-sded platform wth users, advertsers, and publshers (or content provders). The presence of cross-group externaltes provdes a ratonale for the platforms to charge dfferental prces across dfferent sdes, often n the form of subsdzng some groups of customers. The huge success of these platform busness models nevtably nvte competton from entrants. Dd Chuxng and Uber Chna fercely competed for several years to nduce passengers and tax drvers to adopt ther own platforms; Uber and Lyft have been aggressvely competng wth local cab companes n the U.S. markets. The popularty of onlne advertsement exchange platforms (Ad Exchanges) has attracted major competng operators ncludng Google s DoubleClck, Mcrosoft s AdECN, Yahoo s RghtMeda and Facebook s FBX, through whch users, advertsers, and publshers (or content provders) nteract wth each other. How do platforms compete for multple groups or sdes of potental customers? Would an ncrease n platform competton reduce cross-subsdzaton and ncrease the prces (or user fees) and welfare? How would the presence of dstnct multple sdes of customers affect the equlbrum outcome n terms of both dscrmnatory and unform prcng across sdes? There s now a large lterature on platform provsons n mult- 2 Electronc copy avalable at:

3 sded markets. Wth some exceptons, most of the papers n ths lterature focus on two sdes, duopolstc platforms, and a Hotellng specfcaton of product dfferentaton. To answer the above questons and partcularly study the mpacts of platform competton on the prces and welfare n many applcatons, we need to allow an arbtrary number of platforms, a more general form of product dfferentaton, and possbly more than two sdes. The purpose of ths paper s to provde a general model ncorporatng both own-sde and cross-sde externaltes nto a dscrete choce model of random utlty maxmzaton by customers on each sde of the markets. We follow closely the framework of Armstrong (2006) and consder a two-stage game n whch n 2 platforms choose prces (or user fees) smultaneously n the frst stage, followed by s groups of customers smultaneously decdng whch platform to jon (sngle-homng or one-stop shoppng) n the second stage. The decson by a customer on each sde of whch platform to jon s formulated as a dscrete choce maxmzaton of random utltes. In the presence of own-sde and crosssde externaltes, the demand system n our settng s nterdependent across platforms and s mplctly determned by a system of nonlnear equatons. Whle the exstence of partcpaton equlbrum (PE) follows from Brouwer s fxed pont theorem, multple equlbra may exst due to the externaltes across sdes. We provde suffcent condtons to ensure the unqueness of PE for any prce profle, and such condtons hold as long as the extent of externaltes s not so large, relatve to the degree of product dfferentaton among platforms. To show the exstence of the prce equlbrum n the frst stage nvolves two techncal challenges. Frst, as mentoned above, the demand system does not have an explct expresson, whch mples that the changes n a platform s demand respect to hs own prces (.e., the Jacoban matrx) and the platform s frst-order condtons for proft maxmzaton cannot be explctly determned. Second, the condtons to deter global devatons n prces by each platform (.e., the second-order condtons) are not easy to specfy. 2 We provde an approach of transformng the proft maxmzaton problem n The semnar work on two-sded markets starts wth Rochet and Trole (2003, 2006), Callaud and Jullen (2003) and Armstrong (2006), whch provdes a basc framework for studyng prcng schemes under both monopoly and duopoly structures. See Rysman (2009) for a survey of the early lterature on the economcs of two-sded markets. Weyl (200) provdes a theory of mult-sded markets under monopoly usng nonlnear nsulatng tarffs. Jullen (20) studes Stackelberg prce competton n a mult-sded market. For examples of more recent studes and references, see Whte and Weyl (206) and Jullen and Pavan (206). 2 Wthout externaltes, each platform s proft maxmzaton problem n prces s separable across sdes 3

4 prces nto one n quanttes, and ths allows us to crcumvent the two techncal challenges mentoned above. In the symmetrc settng of platforms wth full market coverage, n Theorem of our paper, we prove the exstence of a symmetrc equlbrum n whch the prce on each sde follows a smple, decomposable formula: The prce equals the cost (c), plus a mark-up (M (n)) due to market power assocated wth product dfferentaton, mnus a subsdy (η (n)) due to cross-sde externaltes,.e., p (n) = c + M (n) η (n), where both the producton dfferentaton effect and subsdy effect depend on the number of platforms n. To support ths symmetrc equlbrum, we check the ncentve of a platform, say platform, to devate from the proposed prces. Instead of wrtng the devatng proft as a functon of prces, whch does not have an explct expresson, we rewrte t as a functon of quanttes by consderng hs nverse demand system condtonng on other platforms prces. Intutvely, gven the other platforms prces at the symmetrc equlbrum, for platform to mplement demands q = (q, q s), the addtonal prce he could charge on each sde s equal to the nverse demand wthout externaltes, H ( q ), adjusted by the dfferences n the externaltes that a customer on each sde enjoys between platform and any other platform who shares the the remanng market wth the n platforms equally. The devatng proft as a functon of quanttes, s separable n terms of the mpacts of the dstrbutons and the externaltes, and separate assumptons on these terms can be easly made to guarantee ts concavty. The characterzaton of the equlbrum prces allows us to study the mpacts of platform competton on prces, profts, and welfare. We rase the followng three questons. The frst s how prces change wth respect to competton. Based on the explct prcng formulae, we show that the market power mark-up due to product dfferentaton, M, s monotoncally decreasng n n under the log-concavty assumpton on the dstrbutons, whch s satsfed for many commonly used dstrbutons. The subsdy to each sde, η, accounts for dfferent degrees of total externaltes of a group on all the other groups, whch s shown to be decreasng n the number of platforms under farly weak condtons. The net effect of competton on prces depends on the degree of product of the markets, and for each sde the demand functon, explctly derved from the dscrete choce model, can be shown to be log-concave n ts own prce under certan assumptons on the dstrbutons of the random utltes (e.g., log-concavty of the dstrbutons, see Capln and Nalebuff (99)), and hence the proft on each sde s log-concave n prce and the second-order condton s satsfed. 4

5 dfferentaton and the sze of externaltes. To llustrate the net effect we consder a class of dstrbutons (.e., Gumbel dstrbutons) and lnear forms of externaltes and fnd that as competton ncreases, the prce on one sde can decrease, but the prce on another sde may ncrease. In other words, due to cross subsdzaton, the prces charged to certan sdes can actually ncrease as competton among platforms goes up. Second, how would equlbrum profts and total surplus change wth respect to platform competton, and n partcular, would free entry lead to excessve entry as compared to the socal optmum when there s a fxed cost of entry? In the presence of cross-sde externaltes, the equlbrum proft per platform may not necessarly decrease wth n. Except for the prce effects, the customers prefer varetes, so the expected consumer surplus may ncrease or decrease wth n. The total surplus may not necessarly ncrease wth the number of platforms snce the magntudes of the externaltes decrease even though the prce effects do not affect the total surplus. We provde suffcent condtons for excessve entry, whch rely only on the dstrbutons, but not on the externalty functons. Under Gumbel dstrbutons, we show that there s always excessve entry nto the markets, regardless of the degree of externaltes, although both the socally optmal number of platforms and the equlbrum number of platforms under free entry are affected by cross-subsdzaton. A thrd queston we ask s whether platforms have ncentves to rase prces through a merger. Utlzng the equlbrum condtons for platform proft maxmzaton, we are able to demonstrate that the mergng platforms have ncentves to ncrease ther prces margnally on every sde of the markets. Furthermore, we fnd the magntudes of cost reductons on the mergng platforms that make the equlbrum prces unchanged before and after the merger are proportonal to the pre-merger prce-cost margns. Moreover, we compare unform prcng wth dscrmnatory prcng across sdes. The prce-cost mark-up under the symmetrc unform prce s determned by the famlar demand sem-elastcty, evaluated aggregately over all sdes. The product dfferentaton effect and subsdy effect under the unform prcng cannot be solated as n the case of dscrmnatory prcng. Snce the total welfare remans unchanged under the two prcng rules, platforms and customers have opposte preferences, dependng on the relatve magntudes of the average dscrmnatory prces across sdes and the unform prce. We show that the average dscrmnatory prces across sdes s greater than the unform prce and the platforms prefer dscrmnatory prcng over unform prcng rule when the number of platforms s large or when the degree of externaltes s small. Fnally, we explctly study the mpacts of outsde optons, manly focusng on two- 5

6 sded markets wth Gumbel dstrbutons. In the presence of explct outsde optons, a full characterzaton of the equlbrum s avalable, but the equlbrum prces do not have the decomposable three-term formulae as n the case of no outsde optons. Nevertheless, we show that the mpacts of outsde optons s small when there s suffcent competton among platforms. The remander of ths paper s organzed as follows. Secton 2 ntroduces the model and assumptons. Secton 3 characterzes the equlbrum outcome of the two-stage game. The effects of platform competton on prces and welfare are presented n Secton 4, and some nsghts nto merger analyss are dscussed n Secton 5. Secton 6 compares the equlbrum outcomes under dscrmnatory prcng and unform prcng rules. Secton 7 studes the mpacts of outsde optons, and Secton 8 concludes. All the proofs are presented n Appendx. 2 Model There are n platforms competng for customers from s sdes by chargng prces (or user fees), n 2 and s. Let p k = (p k,, pk s) denote the prces charged by platform k. On each sde, there s a contnuum of customers wth measure. The utlty functon of a customer on sde S := {, 2,, s} from jonng platform k N := {, 2,, n} s u k = v k + φ (x k ) p k + ɛk. () Here v k denotes the ntrnsc valuaton. Snce we focus on ex ante symmetrc platforms, we set v k = v for any k. Let x k = (x k,, xk s) denote the demand profle of platform k. The functon φ s a mappng from [0, ] s to R, and φ (x k ) captures the externaltes, enjoyed by customer on sde from all other sdes (possbly sde tself) n platform k. The term ɛ k s random utlty representng customer preference characterstcs, product characterstcs, functonal form msspecfcaton, and so on. 3 In our specfcaton (), we focus on the prce as user fee that s not condtonal on the partcpaton of customers on any sde. 4 3 See Perloff and Salop (985), Capln and Nalebuff (99) and Anderson et al. (992) for examples of nterpretatons of the random utlty term. 4 Armstrong (2006) consders a platform s two-part tarff on each sde condtonal on the partcpaton of the other sde on the same platform. Whte and Weyl (206) allow general nonlnear tarffs that are condtonal on the partcpatons of customers on all platforms. 6

7 Let P = (p,, p n ) denote the prce profle and p k denote all the prces except of platform k. Gven P and demand profle X = (x,, x n ), platform k s proft s Π k (p k, p k ; X) = (p k c )x k, (2) S where c s the margnal cost of servng a customer on sde, whch s assumed to be dentcal across platforms. We further make the followng assumptons: Frst, our model of customer behavor s based on a dscrete choce model wth random utlty. Each customer on sde has a unt demand, she jons at most one platform (sngle homng). Moreover, we focus on full market coverage by assumng away outsde optons. As such, each customer on sde partcpates n one and only one platform. An alternatve modelng assumpton s that there s ndeed an outsde opton whch yelds utlty level u 0 = v 0 + ɛ 0 on each sde, but the ntrnsc value v s suffcently large as compared wth v 0 so that customers always opt out the outsde opton. Later n Secton 7, we explctly model the mpact of non-trval outsde optons on equlbrum prces. Second, for our man result n the next secton, we make two assumptons on random utltes. We assume that across dfferent sdes = j S, random varables {ɛ, ɛ2,, ɛn } are ndependent from {ɛ j, ɛ2 j,, ɛn j }.5 Wthn each sde, we assume that ɛ, ɛ2,, ɛn are symmetrcally dstrbuted,.e., ther jont dstrbuton s nvarant under any permutaton of the order of these n random varables. 6 For certan comparatve statcs results n Secton 4, we consder the case of ndependent and dentcally dstrbuted (IID) shocks and condtonally ndependent and dentcally dstrbuted (CIID) shocks, respectvely. For the case of IID, we assume that ɛ k IID F ( ), k N (sde-specfc). Moreover, we assume that F s contnuous and dfferentable on the support [a, ā ] wth contnuous probablty densty functon (PDF) f. Beyond these, we do not mpose any functonal form restrctons on F, although the followng examples are frequently adopted for llustratons n our paper: Type I Extreme value dstrbuton (double exponental or Gumbel dstrbuton); Normal dstrbuton; 5 See Jullen and Pavan (206) for a model of platform competton n two-sded markets wth correlated customer preferences between the two sdes. Assumng full market coverage and symmetry between two platforms, they study the mpacts of correlaton between customers nformaton on the equlbrum prces and partcpaton rates. 6 Our assumpton of wthn-sde symmetry could accommodate some spatal models of prce dfferentaton such as the Hotellng specfcaton wth unform dstrbuton of consumer locaton (for n = 2) and the Spokes model (for any n 2, see Chen and Rordan (2007)). 7

8 Unform; Exponental dstrbuton. For the case of CIID, we assume that condtonal on τ, ɛ k IID F ( τ ), k N, where τ s dstrbuted accordng to G ( ). When G s degenerate, the model reduces to the IID case. Thrd, we assume that φ (x) s contnuously dfferentable n x wth φ (0) = 0 (normalzaton). The most common specfcaton of φ s lnear,.e., φ (x, x 2,, x s ) = j S γ j x j. The parameters γ j s not necessarly nonnegatve,.e., negatve externaltes (congeston effect) are also allowed. Our specfcaton could ncorporate wthn-sde φ x network externaltes: = 0, whch s natural n some applcatons. Moreover, we are not lmted to lnear functonal forms, and ncreasng (decreasng) return to scale of externaltes are also permtted, for example, φ (x, x 2,, x s ) = j S γ j ρ(x j ) wth ρ > 0, ρ 0(ρ < 0); or φ (x, x 2,, x s ) = µ( j S γ j x j ), wth µ 0, µ > 0(µ < 0). We study the subgame perfect equlbrum of the followng two-stage game: Platform provders smultaneously choose ther prces frst, followed by customers smultaneously decdng whch platform to partcpate. We use the followng notaton: Let Ψ j (x) = φ (x) x, and j Ψ(x) = (Ψ j (x)),j s denote the margnal externaltes matrx at x. When φ takes the lnear form,.e., φ (x) = j S γ j x j, we let Γ = (γ j ) s s be the lnear externalty matrx,.e., Ψ(x) = Γ, x. For gven vector w = (w,, w n ), we let Dag(w) = Dag(w,, w n ) denote the n by n dagonal matrx wth w on ts -entry. For a matrx M, ts transpose s denoted as M. The n by n dentty matrx s denoted as I n, whle 0 denotes the zero matrx wth sutable dmenson. The column vectors of s of length s s denoted as s = (,,, ). 3 Equlbrum analyss In ths secton we determne the subgame perfect equlbrum of the two-stage game. By backward nducton, we frst study the equlbrum n the second stage, characterzng customers smultaneously partcpaton decsons. For each fxed prce profle P, we call demand profle X(P) a partcpaton equlbrum (PE) assocated wth P f t s a pure strategy Nash Equlbrum n the second stage. Gven P, each customer on sde jons the platform that yelds the hghest utlty, as specfed n (). Clearly, X(P) s a PE assocated wth P f and only f the followng equatons smultaneously hold: x k (P) = Pr(v + φ (x k (P)) p k + ɛk (3) max t =k {v + φ (x t (P)) p t + ɛt }) 8

9 for any S, k N. In other words, the demand system X(P) s nterdependent across platforms and mplctly determned by the above system of equatons. Applyng Brouwer s fxed pont theorem, we see that the system (3) has at least one soluton. Proposton. For any prce profle P, there exsts at least one partcpaton equlbrum. Due to cross-group externaltes, the unqueness of PE s not guaranteed. 7 Nevertheless, suffcent condtons can be derved to ensure unqueness. Intutvely, when the degree of externaltes s small, relatvely to the dsperson of customer preferences for dfferentated products (or the nose dstrbutons), there exsts a unque PE. To llustrate, we focus on the lnear form of externaltes. For n = 2 and general dstrbutons, we apply the contracton-mappng theorem and obtan the followng suffcent condton for unqueness: 8 2{max h (θ ; θ 2)}{ γ j } <, S, j S where h s defned n the next paragraph. 9 When n 2 and ɛ t s IID accordng to Gumbel dstrbuton wth parameter β, the followng condton s suffcent: max { γ j } < 2 mn(β,, β s ). S j S Note that ths condton does not rely on n. We now characterze the prce equlbrum n the frst stage. Let H (θ; n) and h (θ; n) denote the CDF and PDF of the random varable ɛ max(ɛ 2,, ɛn ), respectvely.0 7 See Brock and Durlauf (200, 2002) for dscussons on possble multple equlbra n a sngle-sded market wth externaltes. Multple equlbra also arse n coordnaton games wth strong complementarty, see Carlsson and Van Damme (993); Morrs and Shn (998, 2003); Hellwg (2002); Vves (2005). 8 When γ j 0, alternatve condton s gven by: 2λ max (Γ) max S {max θ h (θ ; 2)} <, where λ max (Γ) s the largest egenvalue of nonnegatve matrx Γ, whch also equals the spectral radus of Γ by Perron-Frobenus Theorem. For n 2, smlar condtons can be derved but are more nvolved. 9 See Appendx B for detals. 0 The exact formula of H and h depends on the specfcaton of random shocks. For Gumbel dstrbuton, H s a logstcs functon, see (24). For Hotellng specfcaton wth unform dstrbuton of consumer locaton (n = 2), H s lnear (see Example on page ). For presentng our man result we mpose mld restrctons on H. For comparatve statc analyss n the next secton, we wll focus on the case of CIID and several examples of the IID case. 9

10 For a gven vector a R s, defne the followng mult-varable functon R : [0, ] s R { R(z; a) := z H ( z ; n) + z φ (z) φ ( } s z n ) + z a. (4) S S We assume the followng regularty condton on R. Assumpton. For any a R s, R(z; a) s concave n z [0, ] s. Notce that the concavty of R(z; a) n z s clearly not affected by a, as the thrd term z a n (4) s lnear n z. An equvalent statement of Assumpton s that for some a, R(z; a) s concave n z [0, ] s. Later we show that the R functon, for approprately chosen a, becomes the proft functon of a platform, f the platform uses quanttes {z,, z s } as hs strategy varables, nstead of prces. Essentally, Assumpton requres that platform s revenue functon, hence proft functon, be concave n quanttes. Ths assumpton can be easly verfed once the dstrbutons of random utltes and the functons of externaltes are gven. To state our man result on the equlbrum prces, we defne M (n) := H (0; n), and η h (0; n) (n) := n Ψ j (x) x= n s, j S S where M represents a sem-elastcty of prce and measures the degree of product dfferentaton among platforms on sde, whch s completely determned by the dstrbutons of random utltes, and η measures the mpact of externaltes that sde generates to all other sdes when the markets are equally dvded among platforms. Theorem. Under full market coverage and Assumpton, there exsts a subgame perfect Nash equlbrum wth the outcome that all the platforms charge the same prces p = (p,, p s ) n the frst stage and the equlbrum market demand s x = n s for each platform n the second stage, where p (n) = c + M (n) η (n), S (5) Theorem shows that n our general settng, the equlbrum prces follow a smple formula. The prce charged to each sde conssts of three addtvely separable terms: the cost, a mark-up due to product dfferentaton, and a subsdy due to externaltes. Each term n the prcng formula represents a separate economc factor and s explctly Under the suffcent condton for the unqueness of PE, the symmetrc prcng equlbrum s unque, and takes the form gven by (5). 0

11 determned by the prmtves of the model. 2 The term M (n) s the standard measure of market power of the olgopolstc frms offerng dfferentated products and s determned by the dstrbutons of random utltes (see Perloff and Salop (985)), whch depends explctly on the number of platforms n the case of symmetry. What s nterestng here s the determnaton of the subsdy term. Note that the partcpaton by sde customer generates externaltes to all sdes (ncludng her own sde). When a platform, say platform, attracts addtonal users on sde, the externaltes enjoyed by any other sde j from jonng platform s enhanced by Ψ j, whle the ex- ternaltes from jonng any other competng platform s reduced by Ψ j n, snce each of the remanng n platforms would lose n on sde due to a busness stealng effect. The dfference n wllngness to pay by each sde j customer then equals Ψ j ( + n ). The addtonal proft that the platform could capture from attractng addtonal user s thus equal to the sum of the margnal externaltes multpled by the equlbrum market share x j that sde mposed on all sdes. In other words, to encourage customer from sde to partcpate, n equlbrum platform needs to lower hs prce by an amount equal to the aggregate margnal profts: Ψ j ( + n )x j, j S whch s the subsdy n Theorem, where x j platform. = /n s the equlbrum market share of Example. To llustrate Theorem, we consder a specal case of our model. In hs model of platform competton (n = 2) n two-sded markets (s = 2), Armstrong (2006) uses a Hotellng specfcaton wth unform dstrbuton of consumer locaton. The equlbrum prces n hs settng are gven by p = c + t α 2, p2 = c 2 + t 2 α, where t and t 2 are the product dfferentaton (or transport cost) parameters, and α and α 2 are the degrees of cross-group externaltes enjoyed by sde and 2, respectvely (see equaton (2) n Proposton 2 n hs paper). Ths prcng formula s the same as 2 The smplcty of the prcng formula s a jont product of symmetry, one-stop shoppng, full market coverage, and the addtve separablty of the consumer payoff, and does not rely on the functonal forms of externaltes and the dstrbutons of random utltes. We wll show n Secton 7 that n the presence of explct outsde optons, the markets are not fully covered, and the equlbrum prces cannot easly decomposed nto the three separate effects.

12 ours n (5). Indeed, n ths Hotellng specfcaton of product dfferentaton, the random utltes can be represented by ɛ = t ɛ, ɛ 2 = t ( ɛ ) wth ɛ beng unformly dstrbuted on the unt nterval [0, ]. It follows that H (θ; 2) = /2 + 2t θ, θ [ t, t ], and hence ( H (0; 2))/h (0; 2) = t [, =, 2. Moreover, ] n ths settng, R s a quadratc 2t α + α 2 functon of z wth Hessan matrx 2, and Assumpton means that α + α 2 2t 2 ths Hessan matrx s negatve defnte, or equvalently 4t t 2 > (α + α 2 ) 2, whch s what Armstrong (2006) assumes (see equaton (8) n hs paper). What s mplctly assumed n the Hotellng duopoly model n Armstrong (2006) s that the ntrnsc value v s relatvely large so that customers always jon at least one of the two platforms, and hence the market are always fully covered. Ths argument s applcable to our general settng wth an outsde opton mposed on each sde, and hence the symmetrc prcng n Theorem remans to be an equlbrum f () ntrnsc values are suffcently large, relatvely to the values of the outsde optons, and () the supports of the dstrbutons of random utltes are bounded. Before further analyzng the above prcng formula, we provde a sketch of the proof of Theorem. We need to show that there s no ncentve for any platform to devate unlaterally from the proposed equlbrum prce vector p n (5). To check the ncentve of a platform, say platform, to devate from p to p, we need to pn down hs devatng proft as a functon of p. One challenge s that due to the cross-sde externaltes, the demand system for platform s servces s mplctly defned by (3), and hence the devatng proft functon does not have an explct expresson. Instead, we rewrte platform s devatng proft as a functon of quanttes q by consderng hs nverse demand system. Intutvely, gven other platforms choces of p, platform behaves as a monopolst, and choosng prces p by the monopolst s equvalent to choosng quanttes q, subject to the demand system condtonal on other platforms prces. Assume that the n platforms have the same demand when chargng the same prce p. Ths together wth full market coverage assumpton mples that whatever quanttes q platform offers, the remanng market s shared equally among the n platforms, wth each takng s q n. It follows that the (condtonal) nverse demand system that platform faces has an explct expresson: The extra wllngness to pay by each sde, p p, equals H ( q ) (the nverse demand wthout externalty), adjusted by φ (q ) φ ( s q n ) (the dfference n the externaltes that customer enjoys between platform and any other platform). We can then reformulate the devatng proft as R(q ; p c) n (4). Devatng s not proftable f R(q ; p c) s maxmzed 2

13 at the equlbrum demand profle, q = n s. The frst-order condtons for ths proft maxmzaton problem hold when p satsfes (5). Moreover, gven that R s concave n q by Assumpton, the frst-order condtons are necessary and suffcent for proft maxmzaton. Therefore, there s no proftable devaton by platform. Moreover, to support the equlbrum outcome gven n Theorem, we need to specfy partcpaton equlbrum (PE) for each prce profle P. As dscussed prevously, there are suffcent condtons under whch for each prce profle P, there s a unque PE. However, n general there mght be multple partcpaton equlbra. We have made the followng reasonable selecton: on the equlbrum path as every platform offers p, we pck PE such that markets are shared equally among these n platforms; off the equlbrum path when only one platform devates to a dfferent prce p whle remanng n platforms charge p, we focus on the sem-symmetrc PE such that the remanng n platforms charge the same prce receve the same demand. Such sem-symmetrc PE always exsts and can be characterzed by fxed-pont condtons, smlar to, but smpler than (3). When more than 2 platforms devate from p, smply pck any PE. One advantage wth transformng a platform s proft maxmzaton n prces nto the one n quanttes s that suffcent condtons for maxmzaton can be easly specfed and checked. 3 Indeed, snce R functon n (4) s separable n terms of the mpacts of the dstrbutons and externaltes, separate assumptons on these functons can be made to guarantee Assumpton. Ths s not possble when consderng proft maxmzaton n prces, snce there s no explct demand system n ths general settng of mult-sded markets. Assumpton can be replaced by smpler but stronger assumptons. Suppose H (θ ; n) s log-concave n θ. Then z H ( z ; n) s concave n z (as shown n Appendx C). As a consequence, the frst term n R(z, a), z H ( z ; n), s concave n z [0, ] s. In other words, wthout cross-sde externaltes, the log-concavty of each H s suffcent to mply Assumpton. In general, the cross-sde externalty term φ non-trvally affects the Hessan matrx of R, and Assumpton roughly means that the degree of externaltes s relatvely small, as compared wth the product dfferentaton effect. Under the lnear form of externaltes, we can nstead mpose the followng alternatve assumptons: 3 See Nocke and Schutz (206) for a model of mult-product olgopolstc competton usng technques from aggregate games. Duet to the mult-sded features and cross-group externaltes, the demand system n our model s not explct and the products wthn each platform are complements nstead of substtutes. Therefore, ther approach does not apply to our model. 3

14 Assumpton 2. () For each S, H (θ ; n) s log-concave n θ ; () the matrx Dag(w,, w s ) s postve defnte, where /w = max θ h (θ ; n) > 0. n n (Γ + Γ ) In Appendx C, we formally show that Assumpton 2 mples Assumpton. Note that for IID random varables, Assumpton (2-) holds f for each, the PDF f ( ) s log-concave, whch holds for many commonly used dstrbutons ncludng normal, unform, exponental, and Gumbel dstrbutons. Assumpton (2-) only requres postve defnteness of a sngle matrx, whch s smpler to check. For Gumbel dstrbutons, (2-) s equvalent to 4Dag(β,, β s ) n n (Γ + Γ ) beng postve defnte, whch s true f the maxmal egenvalue of matrx (Γ + Γ ) s less than 4 mn{β }(n )/n. 4 The effect of platform competton on prces and welfare Theorem reveals decomposable prcng formula for symmetrc equlbrum prces: cost, product dfferentaton effect and cross subsdy, where the product dfferentaton effect s M (Perloff and Salop 985) and the cross subsdy equals η, whch summarzes the externaltes that group offers to all other groups. Both product dfferentaton effect and cross subsdy vary wth the degree of competton (n), and the net effect depends on the degree of product dfferentaton and the sze of externaltes. To study the mpact of platform competton on the equlbrum prces and welfare, we frst consder the case of CIID random varables. In ths case, we have ( ) H (θ; n) = Pr(ɛ max {t=2,,n} ɛt θ) = F (θ + ξ τ )df n (ξ τ ) dg (τ ) and h (θ; n) = ( ) f (θ + ξ τ )df n (ξ τ ) dg (τ ). Therefore, H (0; n) = (n )/n and h (0; n) = ( ) f (ξ τ )df n (ξ τ ) dg (τ ). Proposton 2. Suppose under the CIID, for each τ, F (θ τ ) s log-concave n θ. Then the product dfferentaton effect M (n) s decreasng n n. 4

15 For the case of IID random varables, the effect of competton on market power s known n the lterature. For nstance, Zhou (207) shows that under the log-concavty of F, M (n) s monotoncally decreasng n n, and approaches zero as n goes to nfnty, f addtonally lm θ f (θ )/( F (θ )) = +. 4 Proposton 2 extends the monotoncty of product dfferentaton effect to the case of CIID. On the other hand, n the lterature on mult-sded markets, not much attenton has been pad to the mpact of platform competton on the extent of cross-subsdes. The followng Proposton provdes some nsghts. Proposton 3. Suppose φ (x) = S γ j ρ(x j ) wth ρ > 0. 5 Then the equlbrum crosssubsdy equals η (n) = ρ (/n) n γ, where γ := j S γ j. Assume xρ (x) + ρ (x) > 0 and lm x 0 xρ (x) = 0. 6 Then η decreases wth n f and only f γ > 0. Moreover, η converges to 0 as n. Proposton 3 shows that for a general class of externaltes wth weakly ncreasng returns to scale, the magntude of subsdy to each sde decreases wth n. As competton ncreases, the equlbrum number of users on each sde decrease, and the loop effect s weaker, moreover the margnal externaltes s also weaker gven weakly ncreasng returns to scale, thus the cross-subsdy term decreases wth n. Here the loop effect s equal to + n as each addtonal user on platform mples /(n ) loss for each of the other platforms. Moreover, note that for any two sdes, and j, η η j = ρ (/n) n ( γ γ j ). The relatve degree of subsdes to the two sdes depends on the relatve margnal externaltes that each sde mposes on all the other sdes. Suppose c = c j and M = M j, then γ < γ j mples p > p j. That s, each platform subsdzes customers on sde j by chargng more on sde. As competton ncreases, the extent of cross-subsdes decreases snce ρ (/n) n decreases wth n. In other words, an ncreases n competton erodes the relatve degree of cross subsdes. Propostons 2 and 3 show the monotoncty of the product dfferentaton effect and cross-subsdy effect wth respect to the degree of competton. To llustrate the net effect 4 When the support of θ has a upper bound θ + and f (θ + ) > 0, lm θ θ + f (θ )/( F (θ )) s nfnte. When θ s unbounded from above, the lmt can be nfnte (e.g, normal dstrbutons) or fnte (e.g., Gumbel and exponental dstrbutons). See Proposton 2 and 3 and related dscussons n Perloff and Salop (985). 5 The case wth φ (x) = µ( S γ j x j ) wth ncreasng µ s smlar. 6 The condtons xρ (x) + ρ (x) > 0 and lm x 0 xρ (x) = 0 are satsfed when ρ s convex, or when ρ s a power functon (ρ(x) = x r, 0 < r < ) or log functon (ρ(x) = log( + tx), t > 0). 5

16 of competton on prces, we consder the followng example. Example 2. Suppose ɛ t s IID wth exponental dstrbuton F (θ) = e λ θ and λ > 0. It follows that the product dfferentaton effect M (n) = /λ, whch s constant n n (see Perloff and Salop 985). Wth lnear form of externaltes, the equlbrum prces are gven by p = c + λ n γ, S. Therefore, we have an ntrgung observaton that the equlbrum prce on each sde s ncreasng n the number of platforms n whenever there are postve aggregate externaltes ( γ > 0). For welfare analyss, we defne δ (n) = E[max k=,2, n ɛ k ] as the expected value of the maxmum of n random varables. The followng proposton presents the equlbrum proft for each platform, the expected consumer surplus for each sde, and the total surplus (the sum of total consumer surplus and total profts). Proposton 4. In the symmetrc equlbrum characterzed by Theorem, the followng hold. Each platform k N earns the same proft The expected consumer surplus for sde S equals Π(n) = n (M η ) ; (6) S CS (n) = v c M + δ + η + φ ( n s); (7) The total surplus equals ( TS(n) = v c + δ + φ ( ) n s). (8) S Note that δ reflects customer s preference over varety and ncreases wth n for the case of CIID. Wthout externaltes, Propostons 2 and 4 ndcate standard results that the equlbrum proft for each frm decreases wth n whle both the expected consumer surplus and total surplus ncrease wth n. In the presence of cross-group externaltes, as shown n Proposton 3, platforms have ncentves to subsdze across dfferent sdes, whch may cause the equlbrum prces on some sdes to ncrease wth n. Moreover, for monotoncally ncreasng functon of externaltes, φ ( n s) decreases wth n. As a result 6

17 of the mpact of cross-group externaltes, the equlbrum proft per platform may not necessarly decrease wth n, and the expected consumer surplus and total surplus may not ncrease wth n ether. One mplcaton of Proposton 4 s that there mght be excessve entry nto the multsded markets even though there are cross-group externaltes. To llustrate we frst note that the margnal beneft of an addtonal platform on the total externaltes enjoyed by all sdes s negatvely related to the total cross-subsdes: { S φ ( n s)} n =,j S n 2 Ψ j( n s) = n n ( 2 η ), S whch holds for any functonal form of externaltes. When the aggregate externaltes are postve, the total subsdes are also postve, whch mples that the total benefts from externaltes decreases wth the number of platforms. Second, by Proposton 4, as n ncreases, the change n the equlbrum total surplus can be related to the per-platform proft as follows: TS n n n Π = ( δ n n S n 2 M ), where the rght-hand sde s ndependent of externaltes, and depends only on the dstrbutons of random utltes. 7 Suppose there s a fxed cost K > 0 assocated wth buldng up a platform. There s excessve entry f the equlbrum free-entry number of platforms determned by Π(n) K = 0 exceeds the socally optmal number of platforms that maxmzes the total socal surplus TS(n) nk. The above dscussons mply a suffcent condton for excessve entry as stated n the followng Corollary. Corollary. Suppose that TS(n) nk s quas-concave n n, and that for every sde, the dstrbuton of random utltes satsfes the condton that δ n n M n 2 0. Then there s excessve entry. The condtons n Corollary are mposed only on the dstrbutons of random utltes, not on externalty functons. To see how competton affects the prces and welfare through the trade-off between the product dfferentaton effect and cross-sde subsdy, we consder the followng class of dstrbutons and the lnear form of externaltes. 7 The rght-hand sde s always 0 for Gumbel dstrbutons, and negatve for unform dstrbutons. 7

18 Proposton 5. Suppose ɛ t s IID accordng to the Gumbel dstrbuton wth parameter β, and lnear forms of externaltes. Let β = S β and γ = S γ. Then the followng hold. () For each S, the equlbrum prce s p (n) = c + whch decreases wth n f and only f β > γ. () Each platform s equlbrum proft s whch decreases wth n f and only f γ/ β < n 2 /(2n ); () The total surplus s n n β n γ, (9) Π(n) = n β γ n(n ), (0) TS(n) = (v c ) + (ln(n) + κ) β + γ, () n S whch ncreases wth n. Here κ s the Euler-Marcheron constant. (v) There s always excessve entry. Part () In Proposton 5 provdes a smple formula for the equlbrum prces, whch mples that t s feasble to have the prce on one sde (say ) decrease wth n (when β > γ ) whle the prce on another sde (say ) ncrease wth n (when β < γ ). The latter nequaltes can hold for a number of sdes wthout contradctng wth the condtons for unqueness of partcpaton equlbrum and suffcency for prce equlbrum. In equlbrum, each platform balances out all sdes and the proft per platform and total surplus depend only on the aggregate parameter values, β and γ, as ndcated n parts () and () of Proposton 5. As competton ncreases, platform proft decreases when γ/ β < n 2 /(2n ), and total surplus ncreases when γ/ β < n. The latter condton s equvalent to postve proft for each platform. Both condtons are easly satsfed gven the requrement for suffcency of the prce equlbrum. Therefore, n ths settng the presence of cross-group externaltes does not affect the qualtatve mpact of competton on the aggregate welfare. 8 8 The platform proft ncreases wth the product dfferentaton parameter β, but decreases wth the degree of ndvdual externaltes γ j. The total surplus ncreases wth both β and γ j. It can be easly checked that the expected consumer surplus ncreases wth γ j, but decreases wth β when n = 2, and ncreases wth β when n 3. 8

19 The strength of externaltes and the curvature of φ, whether φ s concave, convex or lnear, clearly affect both the equlbrum number of platforms wth free entry and the socally optmal one, but do not affect the comparson between these two numbers under Gumbel assumptons. 9 5 Merger analyss One applcaton of our equlbrum analyss s to study the mpact of a merger between platforms n mult-sded markets. For nstance, would a merged entty have ncentves to rase the prces gven that all platforms engage n cross-subsdzaton both before and after the merger? Whle our analyss n the prevous secton reles on symmetry, t stll offers some nsghts nto ths queston. In general, a key to the merger analyss s to examne the changes n demands wth respect to prces both wthn and across platforms, whch determne the own-prce and cross-prce elastctes as well as the dverson matrces. Gven that platform k charges prces p k whle other platforms choose symmetrc prces p, assume the demands for other platforms servces are symmetrc. Then, the full market coverage mples a smple relaton between the demand of platform x k and that of other platforms,.e., x k = s x k n, k = k. Ths further mples that xk = p k n xk. As a consequence, the dverson p k matrx from k to k s smply ( xk p k ) ( xk p k ) = n I s, (2) whch s ndependent of p k, p, and x k. The standard measure of upward prcng pressure (UPP) at the equlbrum prces s then n (p c), whch s not nformatve for llustratng ncentves to rase or reduce prces, snce some of the margns can be negatve n ths settng. To evaluate the ncentves for the merged entty to rase prces or not, we need to compute precsely the matrx of the cross-prce partal dervatves. The followng Lemma presents these partal dervatves when all the prces are symmetrc across platforms. 9 When γ = 0, the number of frms under free entry mnus the socally optmal number equals exactly (see Anderson et al. (992), Table 7. on page 226, for ths observaton). In the presence of cross-sde externaltes wth γ > 0, the dfference s strctly larger than. 9

20 Lemma. Wth full market coverage, at any symmetrc prce p wth symmetrc allocaton x = s n, the followng hold: For any k, k N and k = k, x k p k = E and x k p k = E, (3) n where E = Dag{ h (0; n),, h s (0; n) } n n Ψ(x) x= n. (4) s Note that these partal dervatves n Lemma are surprsngly smple and do not depend on the symmetrc p. In Appendx A, we prove a general verson of Lemma, whch presents the matrces of partal dervatves when platform k charges prces p k whle other platforms choose symmetrc prces p. As expected, these matrces depend on (p k, p, x k ). When p k = p, x k = s n and the matrces greatly smplfy to (3) and (4) n Lemma. We now use Lemma to llustrate the margnal ncentve for the merged platforms to rase or reduce prces n mult-sded markets. For notatonal smplcty we consder platforms and 2 to merge and maxmze the jont profts by coordnatng on ther prces. The followng Proposton shows that the merged entty has ncentves to rase locally the prces across all sdes. Proposton 6. At the symmetrc equlbrum prce p, both mergng platform and 2 have strct ncentve to ncrease prce on every sde of the markets n the sense that for l =, 2, {Π + Π 2 } p l p k =p, k N = n(n ) s 0. Wthout cross-group externaltes, the above result s well known for prce competton wth substtute goods, as the mergng frms would nternalze some of the beneft from ncreasng prces. However, n our settng t s possble that the prce-cost margn p c s negatve on some sde before the merger, snce platforms engage n cross subsdzaton. It s not clear at the frst glance that the margnal beneft to platform 2 s postve or not. However, snce both platforms have the same pre-merger margns whch satsfes platform s frst-order condtons p c = ( x p ) n I s. 20

21 Elmnatng the margns, we have a smple expresson for the margnal beneft of rasng prces at the pre-merger symmetrc equlbrum prces as follows: { x2 p } (p c) = { x2 p } ( x p ) n s, whch s equal to n(n ) s by Lemma. Ths mples postve ncentves to ncrease prces on every sde locally. 20 Second, we dentty the magntudes of the cost reductons, due to post-merger synerges, to make the equlbrum prces unaffected. Ths result can be useful when all the pre-merger prces are postve. Proposton 7. Suppose n > 2. There exsts a new cost vector ĉ = (ĉ,, ĉ s ) for the mergng platforms and 2 such that the equlbrum prces after the merger under the new cost structure (ĉ, ĉ, c,, c) stay the same as n Theorem wth orgnal symmetrc cost structure (c, c, c,, c). More precsely, we have ĉ = c n 2 (p c) where the symmetrc prcng p s gven by Theorem. The percentage of proft ncrease by each mergng platform from the above cost savng equals /(n 2). Proposton 7 suggests that to keep the equlbrum prces unchanged, the cost reducton on each sde by the merger needs to be proportonal to the pre-merger margn (when the margn s postve), and ths proporton depends on n but s not affected by sde. Moreover, post-merger equlbrum prces and equlbrum market allocatons stay the same as before, each customer obtans the same expected surplus. the beneft to each mergng platform from the above cost savng s proportonal to equlbrum proft as the cost reducton needed s proportonal to the pre-merger margn. 6 Dscrmnatory vs unform prcng The equlbrum prces n Theorem n general exhbt some degree of prce dscrmnaton across dfferent sdes. Now suppose government regulaton prohbts ths type ( ) ( ) 20 Note that f UPP vector s defned by dg x x 2 p p (p 2 c) as n Affeldt et al. (203), where dga denotes the dagonal matrx whch conssts of the dagonal elements n A, then the above argument suggests that n our settng UPP= n(n ) (dge) s > 0, and hence ths modfed measure of UPP provdes an ndcaton of the merged platforms to rase ther prces locally. 2

22 of prcng behavor by forcng each platform to charge a unform prce on all sdes. How does ths polcy affect the equlbrum prces, profts of platforms, and consumer welfare? The followng theorem determnes the symmetrc equlbrum unform prce: Theorem 2. Suppose c = c, S and all the platforms adopt the unform prcng rule. Then there exsts a symmetrc equlbrum unform prce p u wth where E s gven by (4). p u = c + s/n,j E j, (5) Here E s the sem-elastcty matrx x / p at the equlbrum prces. The prcng formula n (5) follows from standard optmal prcng rule, where s/n s the aggregate demand n equlbrum, and,j E j s the aggregate margnal demand ncrement when a platform rases hs prces unformly on all sdes. Notce that there should be stll product dfferentaton effect and subsdy effect, but the two effects cannot be solated as n the case of dscrmnatory prcng. Nevertheless, the mark-up n (5) s explctly determned by the prmtves of the model. Under both prcng rules, the demand profles are dentcal, hence the total surplus remans unchanged. However, a ban on prce dscrmnaton makes some group(s) better off and other group(s) worse off, and t also affects the platform profts. To compare platform profts and consumer surplus between the two prcng rules, t suffces to compare the average prces across sdes, as shown n the followng Proposton. Proposton 8. The equlbrum outcomes under dscrmnatory prcng and unform prcng characterzed n Theorem and 2 have the followng propertes: () Π d > Π u f and only f p /s > pu ; () For S, CS d > CS u f and only f p < p u ; () TS d = TS u. Snce the equlbrum prces n Theorem and Theorem 2 have explct expressons, the sgn of p /s pu can be determned. Let us frst consder the benchmark case wthout externaltes. Proposton 9. Suppose that at least two of M, S dffer. Wthout externaltes, p /s > p u and the platforms strctly prefer dscrmnatory prcng over unform prcng. 22

23 The logc for the superorty of dscrmnatory prcng over unform prcng s as follows: Wthout externaltes, we have p u c = [ ] s/n S h (0; n) = s M, S whch s the harmonc mean of the product dfferentaton effects M, S. Meanwhle, (p c)/s = s M S S whch s the arthmetc mean. Snce the arthmetc mean s greater than or equal to the harmonc mean, p u S p /s, where the nequalty must be strct by the assumpton on M, S. By Proposton 8, Π d > Π u,.e., platforms obtan strctly hgher profts under dscrmnatory prcng. By contnuty, the above results hold when the degree of externaltes s small. The followng Proposton provdes a comparson between the average prce under dscrmnaton and the unform prce n the presence of cross-sde externaltes when n s large. Let M ( ) = lm n + M (n) denote the product dfferentaton effect n the lmt. 2 Proposton 0. Suppose that the assumptons n Proposton 2 and 3 hold, and that at least two M ( ), S, dffer. Then lm n + S p /s > lm n + p u. In the presence of externaltes across sdes, f the externalty functons satsfy the assumptons n Proposton 3 then the mpact of cross subsdes η, S decreases as competton ncreases and vanshes n the lmt. The prce comparson s then largely determned by the product dfferentaton effects, and hence n the lmt we have smlar results to that n absence of externaltes. Proposton 0 suggests that there exsts a cutoff n such that platforms prefer prce dscrmnaton over unform prcng as long as n > n. Moreover, f at least one M ( ) s zero, lm n p u equals c; f at least one M ( ) s nonzero, lm n S p /s s strctly above c. The above result, although applcable only for large n, s vald for any s. For twosded markets, we have the followng smpler characterzaton for any fnte n. 2 Under the assumptons specfed by Proposton 2, the product dfferentaton term M (n) s monotoncally decreasng n n, and hence lm n + M must exst. 23

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

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

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

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

(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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

The Order Relation and Trace Inequalities for. Hermitian Operators

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

More information

NUMERICAL DIFFERENTIATION

NUMERICAL 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 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

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

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

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

Lecture 10 Support Vector Machines II

Lecture 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 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

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

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

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

The Geometry of Logit and Probit

The Geometry of Logit and Probit The Geometry of Logt and Probt Ths short note s meant as a supplement to Chapters and 3 of Spatal Models of Parlamentary Votng and the notaton and reference to fgures n the text below s to those two chapters.

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

Salmon: Lectures on partial differential equations. Consider the general linear, second-order PDE in the form. ,x 2

Salmon: Lectures on partial differential equations. Consider the general linear, second-order PDE in the form. ,x 2 Salmon: Lectures on partal dfferental equatons 5. Classfcaton of second-order equatons There are general methods for classfyng hgher-order partal dfferental equatons. One s very general (applyng even to

More information

f(x,y) = (4(x 2 4)x,2y) = 0 H(x,y) =

f(x,y) = (4(x 2 4)x,2y) = 0 H(x,y) = Problem Set 3: Unconstraned mzaton n R N. () Fnd all crtcal ponts of f(x,y) (x 4) +y and show whch are ma and whch are mnma. () Fnd all crtcal ponts of f(x,y) (y x ) x and show whch are ma and whch are

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

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

Structure and Drive Paul A. Jensen Copyright July 20, 2003

Structure and Drive Paul A. Jensen Copyright July 20, 2003 Structure and Drve Paul A. Jensen Copyrght July 20, 2003 A system s made up of several operatons wth flow passng between them. The structure of the system descrbes the flow paths from nputs to outputs.

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

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

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

More information

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

6) Derivatives, gradients and Hessian matrices

6) Derivatives, gradients and Hessian matrices 30C00300 Mathematcal Methods for Economsts (6 cr) 6) Dervatves, gradents and Hessan matrces Smon & Blume chapters: 14, 15 Sldes by: Tmo Kuosmanen 1 Outlne Defnton of dervatve functon Dervatve notatons

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

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

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

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

Notes on Frequency Estimation in Data Streams

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

More information

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems Numercal Analyss by Dr. Anta Pal Assstant Professor Department of Mathematcs Natonal Insttute of Technology Durgapur Durgapur-713209 emal: anta.bue@gmal.com 1 . Chapter 5 Soluton of System of Lnear Equatons

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

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

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

More information

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

Constant Best-Response Functions: Interpreting Cournot

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

More information

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

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

More information

Composite Hypotheses testing

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

More information

Environmental taxation: Privatization with Different Public Firm s Objective Functions

Environmental taxation: Privatization with Different Public Firm s Objective Functions Appl. Math. Inf. Sc. 0 No. 5 657-66 (06) 657 Appled Mathematcs & Informaton Scences An Internatonal Journal http://dx.do.org/0.8576/ams/00503 Envronmental taxaton: Prvatzaton wth Dfferent Publc Frm s Objectve

More information

Linear Approximation with Regularization and Moving Least Squares

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

More information

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

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

Canonical transformations

Canonical transformations Canoncal transformatons November 23, 2014 Recall that we have defned a symplectc transformaton to be any lnear transformaton M A B leavng the symplectc form nvarant, Ω AB M A CM B DΩ CD Coordnate transformatons,

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

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

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

Lecture 3: Probability Distributions

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

More information

Lecture 12: Discrete Laplacian

Lecture 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 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

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

General viscosity iterative method for a sequence of quasi-nonexpansive mappings

General 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 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

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

10-801: Advanced Optimization and Randomized Methods Lecture 2: Convex functions (Jan 15, 2014)

10-801: 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 information

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

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

More information

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

Lecture 21: Numerical methods for pricing American type derivatives

Lecture 21: Numerical methods for pricing American type derivatives Lecture 21: Numercal methods for prcng Amercan type dervatves Xaoguang Wang STAT 598W Aprl 10th, 2014 (STAT 598W) Lecture 21 1 / 26 Outlne 1 Fnte Dfference Method Explct Method Penalty Method (STAT 598W)

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

Norm Bounds for a Transformed Activity Level. Vector in Sraffian Systems: A Dual Exercise

Norm Bounds for a Transformed Activity Level. Vector in Sraffian Systems: A Dual Exercise ppled Mathematcal Scences, Vol. 4, 200, no. 60, 2955-296 Norm Bounds for a ransformed ctvty Level Vector n Sraffan Systems: Dual Exercse Nkolaos Rodousaks Department of Publc dmnstraton, Panteon Unversty

More information

1 Binary Response Models

1 Binary Response Models Bnary and Ordered Multnomal Response Models Dscrete qualtatve response models deal wth dscrete dependent varables. bnary: yes/no, partcpaton/non-partcpaton lnear probablty model LPM, probt or logt models

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

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

Additional Codes using Finite Difference Method. 1 HJB Equation for Consumption-Saving Problem Without Uncertainty

Additional Codes using Finite Difference Method. 1 HJB Equation for Consumption-Saving Problem Without Uncertainty Addtonal Codes usng Fnte Dfference Method Benamn Moll 1 HJB Equaton for Consumpton-Savng Problem Wthout Uncertanty Before consderng the case wth stochastc ncome n http://www.prnceton.edu/~moll/ HACTproect/HACT_Numercal_Appendx.pdf,

More information

Vapnik-Chervonenkis theory

Vapnik-Chervonenkis theory Vapnk-Chervonenks theory Rs Kondor June 13, 2008 For the purposes of ths lecture, we restrct ourselves to the bnary supervsed batch learnng settng. We assume that we have an nput space X, and an unknown

More information

Information Management and Pricing in Platform Markets Supplementary Material

Information Management and Pricing in Platform Markets Supplementary Material Informaton Management and Prcng n Platform Markets Supplementary Materal Bruno Jullen Toulouse School of Economcs Alessandro Pavan Northwestern Unversty May 4, 208 Ths document contans addtonal results

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

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

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

Welfare Analysis of Cournot and Bertrand Competition With(out) Investment in R & D

Welfare Analysis of Cournot and Bertrand Competition With(out) Investment in R & D MPRA Munch Personal RePEc Archve Welfare Analyss of Cournot and Bertrand Competton Wth(out) Investment n R & D Jean-Baptste Tondj Unversty of Ottawa 25 March 2016 Onlne at https://mpra.ub.un-muenchen.de/75806/

More information

The equation of motion of a dynamical system is given by a set of differential equations. That is (1)

The equation of motion of a dynamical system is given by a set of differential equations. That is (1) Dynamcal Systems Many engneerng and natural systems are dynamcal systems. For example a pendulum s a dynamcal system. State l The state of the dynamcal system specfes t condtons. For a pendulum n the absence

More information

MATH 241B FUNCTIONAL ANALYSIS - NOTES EXAMPLES OF C ALGEBRAS

MATH 241B FUNCTIONAL ANALYSIS - NOTES EXAMPLES OF C ALGEBRAS MATH 241B FUNCTIONAL ANALYSIS - NOTES EXAMPLES OF C ALGEBRAS These are nformal notes whch cover some of the materal whch s not n the course book. The man purpose s to gve a number of nontrval examples

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

Convexity preserving interpolation by splines of arbitrary degree

Convexity preserving interpolation by splines of arbitrary degree Computer Scence Journal of Moldova, vol.18, no.1(52), 2010 Convexty preservng nterpolaton by splnes of arbtrary degree Igor Verlan Abstract In the present paper an algorthm of C 2 nterpolaton of dscrete

More information

Exercise Solutions to Real Analysis

Exercise Solutions to Real Analysis xercse Solutons to Real Analyss Note: References refer to H. L. Royden, Real Analyss xersze 1. Gven any set A any ɛ > 0, there s an open set O such that A O m O m A + ɛ. Soluton 1. If m A =, then there

More information

A be a probability space. A random vector

A be a probability space. A random vector Statstcs 1: Probablty Theory II 8 1 JOINT AND MARGINAL DISTRIBUTIONS In Probablty Theory I we formulate the concept of a (real) random varable and descrbe the probablstc behavor of ths random varable by

More information

Distribution of subgraphs of random regular graphs

Distribution of subgraphs of random regular graphs Dstrbuton of subgraphs of random regular graphs Zhcheng Gao Faculty of Busness Admnstraton Unversty of Macau Macau Chna zcgao@umac.mo N. C. Wormald Department of Combnatorcs and Optmzaton Unversty of Waterloo

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

Abstract Single Crossing and the Value Dimension

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

More information

Quantity Precommitment and Cournot and Bertrand Models with Complementary Goods

Quantity Precommitment and Cournot and Bertrand Models with Complementary Goods Quantty Precommtment and Cournot and Bertrand Models wth Complementary Goods Kazuhro Ohnsh 1 Insttute for Basc Economc Scence, Osaka, Japan Abstract Ths paper nestgates Cournot and Bertrand duopoly models

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

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

CSCE 790S Background Results

CSCE 790S Background Results CSCE 790S Background Results Stephen A. Fenner September 8, 011 Abstract These results are background to the course CSCE 790S/CSCE 790B, Quantum Computaton and Informaton (Sprng 007 and Fall 011). Each

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