Flirting with the Enemy: Online Competitor Referral and Entry-Deterrence

Size: px
Start display at page:

Download "Flirting with the Enemy: Online Competitor Referral and Entry-Deterrence"

Transcription

1 Flirting with the Enemy: Online Cometitor Referral and Entry-Deterrence January 11, 2016 Abstract Internet retailers often comete fiercely for consumers through exensive efforts like search engine advertising, online couons and a variety of secial deals. Against this background, it is somewhat uzzling that many online retailers have recently begun referring their website visitors to their direct cometitors. In this aer, using an analytical model, we examine this counterintuitive ractice and osit that an entry deterrence motive can otentially exlain this marketlace uzzle. Secifically, we develo a model where two incumbents comete for consumers business while facing a otential entrant who is deciding whether to enter the market. In addition to setting rice, each incumbent firm could otentially dislay a referral link to its direct cometitor. We find that when confronted with a otential entry, an incumbent may refer consumers to its cometitor, intensifying the market cometition and resulting in shutting off the entrant. Only one incumbent may refer its cometitor (i.e., one-way referral) when no loyal consumers exist for either incumbent whereas both incumbents may refer each other (i.e, two-way referral) when each incumbent has a set of loyal consumers. As long as a one-way referral can deter entry, a two-way referral is always sub-otimal. Interestingly, we find that a relatively weak incumbent may refer the consumers to a strong incumbent even when the strong incumbent does not recirocate. Surrisingly, our analysis shows that the ractice of online cometitor referral which imroves the rice transarency could end u hurting aggregate consumer welfare. Overall, our results indicate that firms may be motivated by entry deterrence to voluntarily refer consumers to their direct cometitors even when they are aid nothing for the referral. Keywords: Online Referral, Cometitor Referral, Entry-Deterrence, Game Theory

2 1 Introduction Online retailers often devote considerable financial resources in acquiring new customers either organically or via oaching from cometition. Online brokerages like E*Trade and TD Ameritrade offer cash offers as high as USD 600 to customers to oen a new IRA account (Tergesen, 2015). To induce switching, cellular comanies like T-Mobile and Srint rominently advertise on their websites their willingness to ay early termination fees and the ayments remaining on the hone obtained through cometing service roviders (Bonnington, 2014). Such significant investments towards acquisition of new customers are not surrising given that marketers have often advocated firms to invest into ufront marketing activities with an eye on customer lifetime value. In other words, even when a firm might be seen losing money in the short term while acquiring a customer, long term future income streams from this customer might add u and be rofitable for the firm over a longer horizon (Guta et al., 2004; Shin and Sudhir, 2010). This logic rings esecially true when consumers might face switching costs when moving to cometition. Against this background, a curious new henomenon has emerged recently whereby many online retailers have started dislaying links for their customers and visitors to click to find the rice and other information for a similar roduct at a cometing retailer(s) s website. This henomenon which we call online cometitor referral refers to the ractice wherein firms refer consumers to their direct cometitors on their websites. For examle, RetailMeNot.com, the largest online couon site in the US, dislays referral links to its cometitors web-ages where similar couons might be found (see Macy s couon examle in Fig.1). Grouon.com and chameleonjohn.com also follow a similar ractice to kee consumers informed of couons offered by their cometitors. This ractice is observed in many other online contexts ranging from large national retail chains to small local retailers. As an examle, Greenville Ballet School (greenvilleballet.com) lists all its major cometitors online along with referral links. K9Caers (k9caersdaycenter.com), a day center for dogs, encourages consumers to check out its cometitors by roviding names, addresses, hone numbers and links to their websites. ProfitWell (home.rofitwell.com), a technology firm, dislays cometitors logos with referral links and allows consumers to confirm whether ProfitWell is the best. On Esurance.com (see Fig.2), an online insurance comany, when consumers request a quote, they can also view real-time comarison quotes, visit cometitors websites and buy a cometitor s olicy (Esurance.com, 2006). Zaos.com is well known for referring consumers to its cometitors for consumers best interest (Zaos.com, 2008). Online travel agencies such as Orbitz, Priceline and Exedia also actively 1

3 dislay links to a host of cometitors. Figure 1: Cometitor Referral Page at RetailMeNot.com Figure 2: Cometitor Referral Page at Esurance.com This ractice seems truly uzzling at a first glance because in industries like insurance, travel and online e-tailing, firms comete very fiercely for consumers business for roducts that are largely homogeneous and consumers are very likely to urchase from an e-tailer that offers the lowest rice. Consider a consumer intending to buy an airline ticket between Boston and Atlanta on Exedia.com. This customer has her credit card details and other ersonal information on Exedia and can seamlessly book a ticket there. By referring consumers to cometition, Exedia is lowering consumers search costs and ossibly roviding incentives for them to make their 2

4 booking elsewhere (if they can find a sufficiently attractive rice). While such referrals aear to be getting ubiquitous, neither ractitioners nor researchers seem to gras the underlying mechanism and tradeoffs involved in the ractice. For examle, in a ost on the MOZ Q&A forum (The Moz Q&A Forum, 2013) and in a follow-u ost on the SEO Chat forum (SEO Chat Forum, 2015), ractitioners were debating whether they should offer consumers links to their cometitors. While they desire to kee consumers informed, they worry more about the intensified cometition after dislaying referral links. As ointed out in the comments (SEO Chat Forum, 2015), no one is interested in sending consumers to their cometitors. However, this concern contradicts the fact that many firms across multile industries are doing so online. To the best of our knowledge, the revious research has not investigated the mechanisms for voluntarily referring cometitors online. We seek to address this ga in the resent aer. Secifically, we examine the following questions: (1) What is the rationale behind cometitor referrals? Why do firms refer consumers to their cometitors even though they are aarently worse off by doing so? (2) Under what conditions would firms refer each other and under what conditions would one firm refer other firms but not be referred? (3) When would stronger firms refer weaker firms and when would weaker firms refer stronger firms? These questions are esecially intriguing when we consider that the ractice of referring cometitors seems common in markets where firms comete intensively for business. On a broader level, online sales are steadily accounting for higher fraction of US retail sales, surassing a figure of USD 300 billion dollars for the first time in the year An enhanced understanding of the online retail ractices has clear managerial and ublic olicy imlications. Our study fits within the academic literature researching referral rograms. Consumer referral rograms in which firms reward existing customers for bringing in new customers is a secific ractice of word-of-mouth marketing. As an examle, Schmitt et al. (2011) examined the rofitability and loyalty of the customers acquired through consumer referral rograms and found that they are more rofitable, loyal and valuable than other customers. Research on firm referrals has mostly focused on the imlementation and imact of online referral services. For examle, Chen et al. (2002) studied the otimal contracts of online referral infomediaries that rovide consumers with rice quotes from enrolled brick-and-mortar retailers and direct consumer traffic to articiating retailers. Garicano and Santos (2004) treated referrals as matching oortunities with talent and roosed contracts that can lead to efficient referrals. Ghose et al. (2007) examined the imact of indeendent third-arty and manufacturer-owned internet re- 3

5 ferral services on a suly chain. On the other hand, little research has studied the underlying motivations for firms engaged in referring cometitors. One excetion is Arbatskaya and Konishi (2012) where a firm may refer to its cometitors only when the firm knows that a consumer will not urchase its roduct. Different from such ex-ost referrals, we focus on ex-ante referrals where firms refer to cometitors regardless whether consumers will urchase from them. The other related work is Cai and Chen (2011), who looked at cometitor referral, wherein one retailer could dislay links to a cometing retailer directly (e.g., direct referral) or indirectly through an advertising agency (e.g., third-arty referral). However, direct referral in their context is essentially latform selling: a large latform owner, such as Amazon.com, acts both as a retailer and as a host for many small retailers and makes money by charging commission for sales haening on the latform. Jiang et al. (2011) suggested an alternative exlanation for such a latform selling latform owner can learn from indeendent sellers which of the roducts are worthwhile for direct selling. Another strand of literature studies third-arty referral which refers to the henomenon wherein firms monetize their consumer traffic through a third-arty ad ublisher (Kenny and Marshall, 2000). Recall that both the direct and third-arty referrals include a revenue-sharing contract, which drives retailers referrals to their cometitors. In contrast, we consider the context where a seller directly refers its cometitor online comletely for free. For examle, RetailMeNot.com voluntarily refers Couon.com on its website. In this study, we show that even when a firm does not earn any revenue by referring a consumer to cometitor, it might willingly do so from the ersective of entry-deterrence. This rovides an alternative exlanation for online firm referral in industries where in site of fierce cometition, the entry deterrence might be an imortant force leading to the increasing emergence of this ractice. We develo a model where two incumbents comete for consumers business in the same market. Each incumbent makes a decision on whether to dislay a referral link to its rival while an entrant decides whether to enter the market. We osit that when confronted with a otential entry, an incumbent may refer consumers to its cometitor, intensifying the market cometition to scare off the entrant from entering the market a more severe threat to the incumbents. When no consumers are loyal to a articular incumbent, only one incumbent may refer its cometitor (i.e., one-way referral). In contrast, when a fraction of consumers are loyal to a articular incumbent, both incumbents may refer each other (i.e, two-way referral). As long as a one-way referral can deter entry, a two-way referral is always sub-otimal. In addition, a stronger incumbent may refer its weaker cometitor and, more interestingly, the weaker incumbent may also refer the strong incumbent even when the strong incumbent does 4

6 not recirocate. Surrisingly, our analysis shows that online cometitor referral which imroves the rice transarency in an existing market could end u hurting aggregate consumer welfare. Overall, our results indicate that firms may be motivated by entry-deterrence to voluntarily refer consumers to their direct cometitors comletely for free. To the best of our knowledge, we are among the first in the literature to rovide an unifying exlanation for online cometitor referral shedding light on the underlying motivations for firms seemingly irrational referral behavior. The rest of this aer is organized as follows. In Section 2, we develo a base model where two cometing incumbents make referral decisions while one entrant decides whether to enter the market. In Section 3, we extend the base model to the context where a fraction of consumers are loyal to a articular incumbent. In Section 4, we extend the base model to the case where one incumbent is stronger than the other. We examine the consumer welfare in the base model and two extensions in Section 5. In Section 6, we conclude with a discussion of our findings, alternative modeling ossibilities and future research directions. 2 The Base Model Consider a market for a homogeneous roduct served by two incumbent online sellers, 1 and 2. On the demand side, there is a unit mass of consumers who desire at most one unit of good with a common reservation rice v. Consumers may be aware of only one incumbent. Assume that each incumbent has a fraction α (0, 1 2 ) of customers who visit only its own website. These customers are artially informed and are aware of the existence of one of the incumbent sellers; hence they are loyal to the incumbent they are aware of (referred as loyalist henceforth) and urchase the roduct from this seller as long as the rice is below the reservation rice v. The remaining fraction 1 2α are fully informed consumers who are aware of both sellers. These customers are rice sensitive and always buy the roduct with the lowest rice (referred as switchers henceforth). Online cometitor referral can hel artially informed customers (loyalists) become fully informed (switchers). For examle, if seller 1 rovides a referral link to seller 2, customers who used to visit seller 1 only now are informed of both sellers. The third layer, seller 3 is considering entering the same market to comete with sellers 1 and 2. The entry cost (e.g., the cost of oening an online store) is G > 0. We assume that if seller 3 enters the market a fraction β (0, 1) of consumers will become aware of its existence. β and α are indeendent of each other. 5

7 The sequence of events roceeds as follows. In the first stage, the two incumbents simultaneously decide whether to dislay a referral link to their cometitor. Denote seller i s referral and non-referral strategy by R i and R i (i = 1, 2), resectively. The cost of online referral is normalized to zero. In the second stage, the otential entrant decides whether or not to enter the market, denoted as strategy E and Ē resectively. In the third stage, the two firms (if no entry occurs) or the three firms (if entry occurs) set rices simultaneously. Denote seller i s rice as i, where i = 1, 2, 3. Given these rices, consumers decide whether and where to urchase, if they are aware of more than one seller. In this section, we analyze the case where only one incumbent is considering referring its cometitor (one-way referral). Although we allow both the incumbents to deloy referral strategy simultaneously (two-way referral), as will be shown shortly in subsection 2.5, a two-way referral never occurs under equilibrium in the base model. Without loss of generality, we assume that seller 1 decides whether to refer its cometitor whereas seller 2 does not do so. We examine the following 2 2 subgames: seller 1 alies or does not aly a referral while seller 3 enters or does not enter the market. 2.1 Benchmark: No Referral and No Entry First consider the benchmark case where there is no otential entry or referral ( R 1 Ē). The two incumbents comete for both the loyalists and switchers. Under this setu, analogous to the arguments made in Varian (1980) and Narasimhan (1988), it can easily be shown that no ure strategy equilibrium in rices exists, and only a mixed strategy Nash equilibrium is feasible (see Aendix A for a reca of these arguments). Denote the CDF of i as F i (), which is the robability that seller i charges a rice no higher than. Let F i () = 1 F i (), the robability that layer i charges a rice higher than. When seller i chooses rice while seller j (j i) uses a mixed strategy, the exected rofit of seller i is π i = [α + (1 2α) F j ()]. (1) Sellers 1 and 2 simultaneously set rices to maximize individual rofits. Solving the firms otimization roblem, we obtain the equilibrium rices as below (see Aendix A for the derivation of equilibria): 0 if < α ( ) v, F 1 () = F 2 () = 1 2 α v α if v v, 1 if > v. 6

8 The equilibrium rofit of each seller is equal to the rofit when the two sellers charge the maximum rice = v and serve only their monooly segment; that is, π R 1 Ē 1 = π R 1 Ē 2 = αv, (2) where the suerscrit denotes the corresonding subgame equilibrium. In other words, the rofits that each incumbent can realize deend on the the extent of consumer ignorance the higher the number of loyalists (or non-searchers), the higher the rofits realized by each incumbent in equilibrium. 2.2 Online Referral without Entry When seller 1 refers seller 2 and there is no otential entry (R 1 Ē), seller 1 s customers become aware of seller 2. Therefore, a fraction α of consumers remain loyal to seller 2 and the fraction 1 α are the remaining switchers in the market. The exected rofits of the two sellers (at rice ) are: π 1 = (1 α) F 2 (), π 2 = [α + (1 α) F 1 ()]. (3) Following Narasimhan (1988), we obtain the asymmetric mixed ricing strategies as follows (see Aendix A for derivation): 0 if < αv, ( ) F 1 () = 1 1 αv if αv v, 1 if > v, where seller 2 has a mass α at = v. 0 if < αv, F 2 () = 1 αv if αv v, 1 if > v, The equilibrium rofit of seller 2 (who has a larger base of loyal customers) equals to the rofit when it serves only its monooly segment by quoting the maximum rice = v. contrast, the equilibrium rofit of seller 1 (who has a smaller base of loyal customers) equals the rofit when it charges the minimum rice = αv, oaching all of the consumers who are aware of seller 1; that is, π R1Ē 1 = α(1 α)v, π R1Ē 2 = αv. (4) By comaring subgames R 1 Ē and R 1 Ē, we have the following roosition: In 7

9 Proosition 1 Online referral makes the referring firm worse off but does not affect the rofit of the referred firm. Without otential entry, seller 1 would never refer its cometitor. Proosition 1 shows that online referral makes the referring firm worse off; that is, π R1Ē 1 < π R 1 Ē 1, which confirms the intuition that no firm should be willing to undercut its monooly ower by roviding consumers easy access to the information about its cometitor. Somewhat surrisingly, online cometitor referral does not hel the referred firm; that is, π R1Ē 2 = π R 1 Ē 2. This result is due to the fact that seller 2 cannot benefit from a larger customer base without discounting its rice. As long as the size of its loyal consumer segment remains unchanged, seller 2 always has to trade off attracting switchers with a lower rice against serving loyalists at a higher rice. Figure 3: Equilibrium Prices under R 1 Ē and R 1 Ē (v = 1, α = 1 4 ) Fig.3 resents the mixed rices under R 1 Ē and R 1 Ē. It shows that seller 1 charges a lower rice on average when referring its cometitor, imlying that seller 1 loses its monooly ower because of referral. However, seller 2 may raise or cut its rice on average in resonse to seller 1 s referral. 1 1 As can be calculated, E( R 1 Ē 1 ) = E( R 1 Ē 2 ) = αv 1 2α ln, α E(R 1Ē 1 ) = αv ln 1, and α E(R 1Ē 2 ) = αv(1 + ln 1 ). α By algebra, E( R 1Ē 1 ) < E( R 1 Ē 1 ), E( R 1Ē 2 ) < E( R 1 Ē 2 ) α > 0.34, and E( R 1Ē 1 ) < E( R 1Ē 2 ). 8

10 Table 1: Market Segmentation for R 1 E Segment Fraction of consumers Notation Loyal to 1 α(1 β) n 1 Loyal to 2 α(1 β) n 2 Switch between 1 and 3 αβ s 13 Switch between 2 and 3 αβ s 23 Switch between 1 and 2 (1 2α)(1 β) s 12 Switch among 1, 2 and 3 (1 2α)β s Entry without Online Referral In this section, we analyze a scenario where a third entrant could otentially enter the market (and the layers cannot engage in cometitive referral). As noted before, if seller 3 enters the market a fraction β (0, 1) of consumers will be aware of its existence. β and α are indeendent of each other and the consumers who are aware of layer 3 are drawn from all the existing segments of loyalists and switchers. If the entry indeed haens then this subgame without online referral ( R 1 E) involves three layers and six market segments as shown in Table 1. Relying on the arguments made in Varian (1980), Vives (2001) and Xu et al. (2011), it can be shown that the ure ricing strategies do not exist. We follow oligooly ricing techniques to arrive at the equilibrium outcomes (see Aendix A for derivation). Denote the fraction of customers who are loyal to seller i as n i, the fraction of switchers between sellers i and j as s ij, and let s 123 be the fraction of customers who switch among the three sellers. The values of n i, s ij and s 123 are listed in Table 1 (n 3 = 0). The exected rofits of sellers 1, 2 and 3 (at rice ) are given by, resectively, π 1 = [n 1 + s 12 F2 () + s 13 F3 () + s 123 F2 () F 3 ()], (5) π 2 = [n 2 + s 12 F1 () + s 23 F3 () + s 123 F1 () F 3 ()], (6) π 3 = [s 13 F1 () + s 23 F2 () + s 123 F1 () F 2 ()] G, (7) where seller 3 has an entry cost of G. The three sellers simultaneously choose rices to maximize 9

11 their own rofits. The equilibrium rice strategies are 0 if < min, ( 1 2 α h() ) if min max, F 1 () = F 2 () = ( ) 1 2 α v if max v, 1 if > v, 0 if < min, F 3 () = 1 β 1 β v β h() if min max, 1 if > max, where we define min = α(1 β) v, (1 β)2 (1 2α) max = 2 +4α 2 () 2 (1 β)(1 2α) 2α() v, and h() = 2 + (1 β)(1 2α)v α(). The exressions of F 1 (), F 2 () and F 3 () indicate that sellers 1 and 2 set rices over the entire interval [ min, v] whereas seller 3 sets the uer bound max lower than v. The rationale is that seller 3 has the smallest (zero) loyal segment and the weakest ower in setting rice. The mixed ricing strategy of the three-layer game is similar to that in Koçaş and Bohlmann (2008) and Xu et al. (2011). We show in Aendix A that when β aroaches zero, F 1 () and F 2 () reduce to the exressions derived under R 1 Ē. The equilibrium rofits of the three sellers are π R 1E 1 = π R 1E 2 = α(1 β)v, π R 1E 3 = αβ(1 β) v G. (8) 1 α Similar to the equilibrium rices, the equilibrium rofits of sellers 1 and 2 also aroach those under R 1 Ē when β 0. Comaring subgames R 1 Ē and R 1 E, we have π R 1E 1 < π R 1 Ē 1 and π R 1E 2 < π R 1 Ē 2. This makes intuitive sense since the entry of a third layer now reduces the ricing ower of the two incumbents and makes them worse off. resented in the roosition below: This outcome is formally Proosition 2 Absent online referral, seller 3 enters the market as long as G αβ(1 β) v. The entry of the third layer makes the two incumbents worse off. Proosition 2 is very intuitive because no incumbent wants higher cometition in the market. Fig.4 illustrates the mixed rices under R 1 Ē and R 1 E. When seller 3 enters the market, both sellers 1 and 2 discount rices more heavily. In other words, the new entry intensifies the market cometition making the incumbents worse off. 2.4 Online Referral with Entry Now we consider the outcomes in a scenario where online referrals are a ossibility. In the subgame R 1 E where seller 3 enters the market and seller 1 refers seller 2, the market segments 10

12 Figure 4: Equilibrium Prices under R 1 Ē and R 1 E (v = 1, α = 1 4, β = 1 2 ) Table 2: Market Segmentation for R 1 E Segment Fraction of consumers Notation Loyal to 2 α(1 β) n 2 Switch between 2 and 3 αβ s 23 Switch between 1 and 2 (1 α)(1 β) s 12 Switch among 1, 2 and 3 (1 α)β s 123 are shown in Table 2. Note that since the seller 1 refers seller 2 through an online referral link, this seller now does not retain any loyal customers. The exected rofits of sellers 1, 2 and 3 (at rice ) are given by, resectively, π 1 = F 2 ()[s 12 + s 123 F3 ()], (9) π 2 = [n 2 + s 12 F1 () + s 23 F3 () + s 123 F1 () F 3 ()], (10) π 3 = F 2 ()[s 23 + s 123 F1 ()] G. (11) The attern of mixed rices is similar to that under R 1 E: Two sellers set high rices on average and they set rices over the entire interval, while one seller may set the uer bound lower than other two. R 1 E differs from R 1 E in the following: Since n 2 > n 1 = n 3 = 0, whether seller 1 or seller 3 sets the lowest rice deends on whether s 23 exceeds s 12 (see Aendix A for derivation). 11

13 If s 12 > s 23, i.e., α + β < 1, seller 3 has the weakest ower in setting rice. The equilibrium rices of sellers 1, 2 and 3 are, resectively, 0 if < α(1 β)v, 1 F 1 () = 1 α(1 β)v if α(1 β)v α 1 β v, 1 α(v ) α () if 1 β v v, 1 if > v, 0 if < α(1 β)v, 1 F 2 () = 1 αv α if 1 β v v, α(1 β)v if α(1 β)v α 1 β v, 1 if > v, 0 if < α(1 β)v, F 3 () = 1 β 1 β 1 if > α 1 β v, where seller 2 has a mass α at = v. α(1 β)v if α(1 β)v α 1 β v, If s 12 s 23, i.e., α + β 1, seller 1 has the weakest ower in setting rice. In this case, the equilibrium rices of sellers 1, 2 and 3 are, resectively, 0 if < α(1 β)v, F 1 () = 1 1 α(1 β)v if α(1 β)v 1 β α v, 1 if > 1 β α v, 0 if < α(1 β)v, 1 α(1 β)v if α(1 β)v 1 β α F 2 () = v, 1 (1 β)v if 1 β α v v, 1 if > v, F 3 () = 0 if < α(1 β)v, 1 β 1 β α(1 β)v 1 (1 β)(v ) β where seller 2 has a mass 1 β at = v. if α(1 β)v 1 β α v, if 1 β α v v, 1 if > v, No matter which firm charges the lowest rice, seller 2, who has n 2 fraction of loyalists, always earns the rofit equal to n 2 v; while sellers 1 and 3, both of whom have zero loyalist, earn rofits equal to (s 12 + s 123 ) min and (s 23 + s 123 ) min G, resectively; that is, π R1E 1 = α(1 α)(1 β)v, π R1E 2 = α(1 β)v, π R1E 3 = αβ(1 β)v G. (12) 12

14 Note that when β 0, the equilibrium rices and rofits of R 1 E aroach the form under R 1 Ē. Next roosition is obtained by comaring results from R 1 E and R 1 E. Proosition 3 In the resence of online cometitor referral dislayed by seller 1, seller 3 enters the market as long as G αβ(1 β)v. The referral strategy makes seller 1 and seller 3 worse off without affecting seller 2 s rofit. Online referral is detrimental to the referring firm, i.e., π R1E 1 < π R 1E 1. Therefor, no incumbent has an incentive to refer its cometitor conditional on the entry of the third layer. In addition, online referral does not influence the rofit of the referred firm, i.e., π R1E 2 = π R 1E 2. These results are consistent with the case with no entry (see Proosition 1). It is noteworthy that π R1E 3 < π R 1E 3 ; that is, seller 3 is also hurt by seller 1 s referral strategy because of the intensified cometition. This finding suggests that seller 3, who could originally earn ositive rofits through entry, may be blocked out of the market when an incumbent refers his cometitor. This key insight exlains why at equilibrium seller 1 could refer its cometitor even when its rofit is attenuated because this is comensated via non-entry of the third firm, as will be shown later. Fig.5 illustrates the equilibrium rices under R 1 E and R 1 E, showing the influence of online referral on rices under the case with new entry. Figure 5: Equilibrium Prices under R 1 E and R 1 E (v = 1, α = 1 3, β = 1 4 ) 13

15 2.5 Equilibrium Referral Strategy Subsections have discussed one-way referral dislayed by seller 1. Considering that sellers 1 and 2 are symmetric with each other, the referral dislayed by seller 2 will result in similar subgame equilibrium rofits as shown in Eqs.(2, 4, 8, 12). In case of two-way referral where sellers 1 and 2 refer each other, all consumers will be fully informed. The market oerates under Bertrand rice cometition, which drives the equilibrium rices and rofits of both incumbents to zero. Therefore, two-way referral is never adoted in the equilibrium. This leads to our equilibrium referral strategy outcome stated as follows: Proosition 4 The equilibrium referral and entry strategies are as follows: (i) If G αβ(1 β)v, or αβ(1 β)v < G αβ(1 β) v and β α, there is no referral by incumbents and seller 3 enters the market; (ii) If αβ(1 β)v < G αβ(1 β) v and β > α, one incumbent seller refers its consumers to the other incumbent while the other does not recirocate, and seller 3 does not enter the market; (iii) If G > αβ(1 β) v, no incumbent ractices referral, and seller 3 does not enter the market. Proosition 4 shows that the equilibrium referral and entry strategies are (i) R 1 R2 E if the entry cost G or the fraction of customers who are aware of the entrant β is sufficiently small, (ii) R 1 R2 Ē or R 1 R 2 Ē if G is medium and β is sufficiently large, and (iii) R 1 R2 Ē if G is sufficiently large (see Fig.6). Figure 6: Equilibrium under Base Model Part (i) and art (iii) of Proosition 4 are consistent with our intuition: If the entry cost is 14

16 small, seller 3 always enters the market regardless whether seller 1 (2) refers seller 2 (1). In this case, online cometitor referral lays no other role besides romoting cometition. Thus, seller 1 (2) has no reason to refer seller 2 (1). On the other hand, if the fraction of customers who are aware of the entrant s roduct is small, seller 1 (2) can tolerate the influence of the new entry and therefore does not refer its cometitor. However, if the entry cost is too high, seller 3 never enters the market, which is indeendent of referral decisions of the two incumbents. In this case too, seller 1 (2) has no incentive to refer seller 2 (1). Part (ii) of Proosition 4 is the core interesting insight gleaned from our analysis. It shows that if G is medium and β is large, counter-intuitively, online referral occurs in equilibrium even though it makes the referring firm worse off (see Proosition 1 for π R1Ē 1 < π R 1 Ē 1 ). Why does this seller might self-cut its rofit by referring its cometitor? The rationale behind this counterintuitive behavior lies in the entry-deterrence motive. In a market with weak cometition, otential entrant has high incentives to enter for a share of the ie. In contrast, if firms comete intensely in the market, the entrant would rather stay outside of the market. Therefore, online cometitor referral can serve as an entry-deterrence strategy via intensifying the cometition. Although the referral hurts the referring firm, it is the lesser of two evils and can discourage the entrant from entry, a more severe threat to the incumbents (π R1Ē 1 > π R 1E 1 β > α). Part (ii) of the roosition also indicates that in equilibrium either seller 1 or seller 2 has to shoulder the cost of referral since the act hurts the referring firm. Note that sellers 1 and 2 make referral decisions simultaneously and non-cooeratively. Therefore, each incumbent is better off if the other incumbent does the referral. There exist two Nash equilibria: seller 1 refers seller 2, or seller 2 refers seller 1. Additionally, it can be easily shown that if the incumbents use mixed referral strategy, in equilibrium each seller will aly referral with robability τ and does not aly referral with robability 1 τ, where τ = β α 1+β α (see Aendix B3). Irresective of whether firms used ure or mixed strategies, the key insight is that the referral will emerge as a voluntary ractice under certain conditions even when firms are not being aid for rendering this service. 3 Imerfect Referral In the base model, we assume that all of the loyalists of the referring firm become switchers once it refers its customers to a cometitor. In this section, we relax this assumtion and allow a fraction of consumers to remain loyal to the referring firm even when it ractices online referral. In reality, some consumers may stick to a website like RetailMeNot or Esurance because of 15

17 switching costs, inattention or brand loyalty, even when they are rovided with easy access to cometition. For examle, consumers may already have a social network and be very familiar with how to find good couons on RetailMeNot, but they have to start from scratch and learn how to find couons on RetailMeNot s cometitors. In addition, sellers aly online cometitor referral with varying details and on varying ositions on their websites. For instance, Greenville Ballet School laces cometitors links at the center of the webage which facilitates customers clicking on the links. In contrast, cometitors links on RetailMeNot are located at the bottom of the webage where a large fraction of consumers may never reach. We assume that when one incumbent refers the other, only a fraction ϕ (0, 1) of consumers who visit the referring incumbent comare shoing at the two firms. The remaining fraction 1 ϕ stays loyal to the referring firm. 3.1 One-way Referral Suose that only seller 1 could otentially ractice referral whilst seller 2 could not. scenario is similar to the base model, with a minor difference where the referring firm owns a ositive roortion of loyalists. The equilibrium rofits of the 2 2 subgames are (see Aendix B1 for derivation): π R1E 1 = This π R 1 Ē 1 = π R 1 Ē 2 = αv, (13) π R1Ē α(1 α) 1 = v, πr1ē 2 = αv, (14) 1 α + αϕ π R 1E 1 = π R 1E 2 = α(1 β)v, π R 1E 3 = α(1 α)(1 β) v, πr1e 2 = α(1 β)v, π R1E 3 = 1 α + αϕ αβ(1 β) v G, (15) 1 α αβ(1 β) v G. (16) 1 α + αϕ By comaring, we find that π R1Ē 1 < π R 1 Ē 1 and π R1E 1 < π R 1E 1, imlying that seller 1 has no incentive to refer seller 2 on conditional that seller 3 enters or does not enter the market. Furthermore, π R1E 3 < π R 1E 3 ; that is, seller 3 is worse off resulting from the incumbent s online referral. αβ(1 β) This indicates that seller 1 is able to deloy referral to deter otential entry when +αϕ v < G αβ(1 β) v. Seller 1 is willing to embrace this strategy if it earns a higher rofit under R 1 Ē than under R 1 E which requires that β should be sufficiently large, i.e., π R1Ē 1 > π R 1E 1 β > αϕ +αϕ. The comarisons above yield the following roosition. Proosition 5 One-way imerfect referral can be used to deter entry and imrove rofit if αβ(1 β) +αϕ v < G αβ(1 β) αϕ v and β > +αϕ. 16

18 Table 3: Market Segmentation for R 1 R 2 E with Imerfect Referral Segment Fraction of Consumers Notation Loyal to 1 α(1 ϕ)(1 β) n 1 Loyal to 2 α(1 ϕ)(1 β) n 2 Switch between 1 and 3 α(1 ϕ)β s 13 Switch between 2 and 3 α(1 ϕ)β s 23 Switch between 1 and 2 [1 2α(1 ϕ)](1 β) s 12 Switch among 1, 2 and 3 [1 2α(1 ϕ)]β s 123 The intuition for Proosition 5 is similar to that for the base model: Although online referrals intensify market cometition and thereby hurt the referring firm, the referring firm may still rovide one to its cometitor. This is because the fiercer cometition induced by online referral could be a credible threat to the entrant scaring it out of the market. If entry is more detrimental than losing loyal customers, the incumbent seller will refer its cometitor. 3.2 Two-way Referral In this subsection, we discuss two-way referral where both sellers 1 and 2 simultaneously decide whether to refer their cometitor. imerfect two-way referral may benefit the incumbents. As will be shown shortly, unlike under the base model, The subgames R 1 R2 Ē and R 1 R2 E are the same as the subgames R 1 Ē and R 1 E in the base model, and thus π R 1 R2 Ē 1 = π R 1 R2 Ē 2 = αv, (17) π R 1 R2E 1 = π R 1 R2E 2 = α(1 β)v, π R 1 R2E 3 = αβ(1 β) v G. (18) 1 α When the two incumbents refer each other and seller 3 does not enter the market (R 1 R 2 Ē), the size of each market segment is n 1 = n 2 = α(1 ϕ). This subgame is similar to R 1 Ē in the base model. We easily obtain the equilibrium rofits as follows (see Aendix B1 for derivation): π R1R2Ē 1 = π R1R2Ē 2 = α(1 ϕ)v. (19) When the two incumbents refer each other and seller 3 enters the market (R 1 R 2 E), the size of each market segment is resented in Table 3. 17

19 Table 3 is very similar to Table 1, and the equilibrium rices and rofits can be obtained directly from the subgame R 1 E of the base model. The equilibrium rofits are as follows (the equilibrium rices are listed in the Aendix B1): π R1R2E 1 = π R1R2E 2 = α(1 ϕ)(1 β)v, π R1R2E 3 = αβ(1 β)(1 ϕ) v G. (20) 1 α(1 ϕ) Based on the 2 2 subgames analyzed above, we can investigate whether two-way referral still has an effect of entry deterrence, as shown in the next roosition. Proosition 6 If the two incumbents simultaneously decide whether to aly a referral strategy and the referral is imerfect in affecting the loyalists, then a two-way referral will be used to deter entry if αβ(1 β)(1 ϕ) (1 ϕ) v < G αβ(1 β) v and β > ϕ. Proosition 6 shows that two-way referral can be an effective entry-deterrence tool. Although the incumbents may be hurt by two-way referral because of more intense cometition, in equilibrium they still deloy it to avoid more serious consequences from entry (the market will not oerate under Bertrand cometition when ϕ < 1). The rationale here is similar to one-way referral where the referring firm is always worse off when referring a cometitor. However, two-way referral differs from one-way referral in the following two ways: First, two-way referral imlies more intense cometition relative to one-way referral, so as long as one-way referral can deter entry, two-way referral will be sub-otimal. Second, two-way referral may be referred over one-way referral. In articular, by comaring Proositions 5 and 6 we see that when αβ(1 β)(1 ϕ) (1 ϕ) v < G αβ(1 β) (1 ϕ) v and β > ϕ, two-way referral can discourage entry while one-way referral is no longer effective. This rovides an intuition of the situations in which two-way referral might be more effective than one-way referral A two-way referral is more effective under a lower range of fixed cost (G) and under the resence of a stronger entrant (higher β) relative to the the region where one-way referral works. In other words, when the threat of an entrant is higher for incumbents (due to lower fixed cost of entry and a higher otential of the entrant luring existing customers), a one-way referral might not suffice and incumbents will need to resort to two-way referrals to intensify the cometition and shut off the entrant. 3.3 Equilibrium Referral Strategy Based on the subgames of one-way and two-way referrals, we summarize equilibrium outcome of the three-stage game next in roosition 7, where we define G 1 = αβ(1 β) v, G 2 = αβ(1 β) (1 ϕ) v, G 3 = αβ(1 β)(1 ϕ) (1 ϕ) v, β 1 = ϕ, and β 2 = αϕ +αϕ. 18

20 Proosition 7 With imerfect referral, the equilibrium referral and entry strategies are: (i) If G G 3, or G 3 < G G 2 and β β 1, or G 2 < G G 1 and β β 2, there is no referral by incumbents and seller 3 enters the market; (ii) If G 3 < G G 2 and β > β 1, there is either no referral or two-way referral by incumbents, and seller 3 enters the market if there is no referral while does not enter if there is two-way referral; (iii) If G 2 < G G 1 and β > β 2, one incumbent seller refers its consumers to the other incumbent while the other does not recirocate, and seller 3 does not enter the market; (iv) If G > G 1, no incumbent ractices referral, and seller 3 does not enter the market. Figure 7: Equilibrium under Imerfect Referral Model Fig.7 deicts the equilibrium referral and entry strategies shown in Proosition 7, which carries the intuition similar to the base model but also brings forth additional nuances. First, within the region G 2 < G G 1 and β > β 2, both one-way and two-way referrals can deter entry. In this case, two-way referral is sub-otimal and could never be an equilibrium. Second, within the region G 3 < G G 2 and β > β 1, one-way referral can no longer deter entry. In this case, only when the two incumbents refer each other can the otential seller be scared off. However, no incumbent knows its cometitor s referral choice and any incumbent will lose more rofits if it refers its cometitor while the cometitor does not do the same. Therefore, the equilibrium in this region might be R 1 R 2 Ē or R 1 R2 E. In addition to this ure strategy equilibrium, a mixed equilibrium also exists wherein each incumbent refers the cometitor with robability τ and does not refer the cometitor with robability 1 τ, where τ = α(1 β)ϕ ()β ϕ(1 2α+αϕ) (see 19

21 Aendix B3). 4 Asymmetric Incumbents In this section we extend the base model to examine the case where sellers 1 and 2 differ in their number of loyalists. This is a natural extension since we do often see online retailers vary in their reach and loyalty due to differential brand strengths. We assume that seller i has a fraction α i (i = 1, 2) of loyal segment, where > α 2. Without loss of generality, we label seller 1 as the strong brand whilst labeling seller 2 as the weak brand. For simlicity, we kee ϕ = Strong-Brand Referral In this subsection, we analyze the strong-brand referral; that is, only seller 1 decides whether to aly referral whereas seller 2 does not. When there is no otential entry and no referral is adoted ( R 1 Ē), the sizes of the loyal consumer segment and the switcher segment are n 1 =, n 2 = α 2 and s 12 = 1 α 2. Following the same logic as the base model, we obtain equilibrium rofits of the incumbents as follows (see Aendix B2 for the equilibrium rices): π R 1 Ē 1 = v, π R 1 Ē 2 = (1 ) 1 α 2 v. (21) When there is no otential entry and seller 1 refers seller 2 (R 1 Ē), seller 1 ends u with zero loyalist and seller 2 has a fraction α 2 of loyalists. The remaining fraction 1 α 2 are switchers. Analogue to Eq.(4), we obtain the equilibrium rofits as follows: π R1Ē 1 = α 2 (1 α 2 )v, π R1Ē 2 = α 2 v. (22) In the case that seller 3 enters the market and seller 1 does not aly referral ( R 1 E), the market will be segmented as shown in Table 4. The exected rofit of seller i is π i = [n i + s i,i+1 Fi+1 () + s i,i+2 Fi+2 () + s 123 Fi+1 () F i+2 ()], (23) where n 1 > n 2 > n 3 = 0, and seller i and seller i + 3 denote the same entity (i = 1, 2, 3). Analogue to the base model, we can obtain the equilibrium rofits of the three sellers as follows (see Aendix B2 for derivation): π R 1E 1 = (1 β)v, π R 1E 2 = (1 )(1 β) 1 α 2 v, π R 1E 3 = β(1 β) 1 α 2 v G. (24) 20

22 Table 4: Market Segmentation for R 1 E under Strong-Brand Referral Segment Fraction of Consumers Notation Loyal to 1 (1 β) n 1 Loyal to 2 α 2 (1 β) n 2 Switch between 1 and 3 β s 13 Switch between 2 and 3 α 2 β s 23 Switch between 1 and 2 (1 α 2 )(1 β) s 12 Switch among 1, 2 and 3 (1 α 2 )β s 123 Table 5: Market Segmentation for R 1 E under Strong-Brand Referral Segment Fraction of Consumers Notation Loyal to 2 α 2 (1 β) n 2 Switch between 2 and 3 α 2 β s 23 Switch between 1 and 2 (1 α 2 )(1 β) s 12 Switch among 1, 2 and 3 (1 α 2 )β s 123 Finally, when seller 3 enters the market and seller 1 refers seller 2 (R 1 E), the market is segmented as shown in Table 5. Table 5 is very similar to Table 2. We just need to relace α with α 2 to get the equilibrium. Therefore, the equilibrium rofits of sellers 1, 2 and 3 are, resectively, π R1E 1 = α 2 (1 α 2 )(1 β)v, π R1E 2 = α 2 (1 β)v, π R1E 3 = α 2 β(1 β)v G. (25) Based on the four subgames analyzed above, we have the following roosition. Proosition 8 Under the strong-brand referral format, (i) online referral is always harmful to the referring firm as well as to the referred firm regardless of whether seller 3 enters the market or not; that is, π R1Ē 1 < π R 1 Ē 1, π R1Ē 2 < π R 1 Ē 2, π R1E 1 < π R 1E 1 and π R1E 2 < π R 1E 2 ; and (ii) seller 1 will deloy referral to deter entry if α 2 β(1 β)v < G α1β(1 β) 2 v and β > 1 α2(2). Proosition 8 shows that strong-brand referral can be used to deter entry when G is medium and β is large enough. This result is consistent with the base model. However, strong-brand 21

23 Table 6: Market Segmentation for R 2 E under Week-Brand Referral Segment Fraction of Consumers Notation Loyal to 1 (1 β) n 1 Switch between 1 and 3 β s 13 Switch between 1 and 2 (1 )(1 β) s 12 Switch among 1, 2 and 3 (1 )β s 123 referral differs from the base model in that it not only hurts the referring firm but also makes the referred firm worse off. The main reason lies in that seller 2 is a small firm (with fewer loyalists) without seller 1 s referral and can earn rofits from its loyalists as well as switchers since seller 1 is interested largely (with a larger share of loyalists) on earning higher rofits from its loyalists and kees its rice relatively higher. After referral, seller 1 (strong brand) has access only to switchers and hence rices aggressively to ta into this ool leading to aggressive rice cometition and seller 2 rofiting mainly from its (smaller) share of loyalists. 4.2 Weak-Brand Referral In this subsection, we analyze the weak-brand referral format; that is, only seller 2 decides whether to refer seller 1. The results of subgames R 2 Ē and R 2 E stay the same as those under strong-brand referral. So we discuss only the subgames R 2 Ē and R 2 E. When there is no otential entry and seller 2 refers seller 1 (R 2 Ē), the sizes of loyal segments of sellers 1 and 2 are and 0, resectively. The equilibrium rofits of the two sellers are as follows (see Aendix B2 for derivation): π R2Ē 1 = v, π R2Ē 2 = (1 )v. (26) When otential entry exists and seller 2 refers seller 1 (R 2 E), the size of each segment will be that in Table 6. The equilibrium rofits of the three sellers are as follows (see Aendix B2 for derivation): π R2E 1 = (1 β)v, π R2E 2 = (1 )(1 β)v, π R2E 3 = β(1 β)v G. (27) Based on the above analysis, we have the following roosition. Proosition 9 Under the weak-brand referral format, (i) online referral is harmful to the referring firm but has no influence on the referred firm regardless of whether seller 3 enters the 22

24 market; that is, π R2Ē 1 = π R 2 Ē 1, π R2Ē 2 < π R 2 Ē 2, π R2E 1 = π R 2E 1 and π R2E 2 < π R 2E 2 ; and (ii) seller 2 will deloy referral to deter entry if β(1 β)v < G α1β(1 β) 2 v and β > α 2. Proosition 9 shows that under the weak-brand referral format, the referring firm is always worse off whereas the referred firm is not affected. This is different from the strong-brand format because here the referring firm is always the relatively weak brand either before or after referral; hence the strong seller always receives the rofit that is equal to that when it serves only the loyalists by charging the monooly rice. In addition, the conditions for strong-brand referral to be effective differ from those for weak-brand referral. As shown in Proositions 8 and 9, the strong-brand referral format is used to deter entry when α 2 β(1 β)v < G α1β(1 β) 2 v and β > 1 α2(2), whereas the weak-brand referral is adoted when β(1 β)v < G α1β(1 β) 2 v and β > α 2. Obviously, strong-brand referral may be effective while weak-brand referral is not (when α 2 β(1 β)v < G β(1 β)v), and weak-brand referral is effective while strong-brand referral is not (when α 2 < β 1 α2(2) ). It conforms to our intuition that the strong-brand referral format is more effective in entrydeterrence than the weak-brand referral because the stronger seller earns more rofits without entry and is hurt more when an entry occurs. Therefore, the stronger seller has a stronger incentive to block otential entry. It seems uzzling that weak-brand referral may become more effective under certain conditions. In fact, the stronger seller has to ay a higher rice for referral than the weaker seller (referral hurts the stronger seller more since it loses more loyal customers when referring cometitor; on the contrary, the weaker seller loses less since it has fewer loyal consumers). Therefore, under certain circumstances (α 2 < β 1 α2(2) ), the stronger seller may find it unrofitable to deter entry via referral whereas the weaker seller finds it worthwhile to do so. 4.3 Equilibrium Referral Strategy Analogous to the base model, two-way referral results in zero rofit for sellers 1 and 2. Based on the strong-brand referral, weak-brand referral and two-way referral formats analyzed above, we have the following roosition that summarizes our analysis, where we define G a = α1β(1 β) 2 v, G b = β(1 β)v, G c = α 2 β(1 β)v, β a = 1 α2(2), and β b = α 2. Proosition 10 When the incumbents are asymmetric, the equilibrium referral and entry strategies are as follows: 23

25 (i) If G G c, or G c < G G b and β β a, or G b < G G a and β β b, there is no referral by incumbents and seller 3 enters the market; (ii) If G c < G G b and β > β a, the strong brand refers the weak brand whilst the weak brand does not refer the strong brand, and seller 3 does not enter the market; (iii) If G b < G G a and β b < β β a, the weak brand refers the strong brand whilst the strong brand does not refer the weak brand, and seller 3 does not enter the market; (iv) If G b < G G a and β > β a, no incumbent refers its cometitor or both incumbents ractice referrals, and seller 3 enters the market if there is no referral while does not enter if two-way referral is resented; (v) If G > G a, no incumbent ractices referral, and seller 3 does not enter the market. Figure 8: Equilibrium under Asymmetric Incumbent Model Proosition 10 is roughly sketched by Fig.8. The main finding from Proosition 10 is that the weaker seller may refer the stronger seller even when the stronger seller does not do so, as shown in arts (iii) and (iv) of the roosition. As exlained after Proosition 9 earlier, the stronger seller has a higher incentive to rotect its share but under certain conditions referral strategy may be not efficient because the seller might lose more loyal customers resulting in loss of its monooly ower. On the other hand, the weaker seller has less incentive to rotect its market share but under certain conditions may be efficient in using referral to deter entry. Therefore, Proosition 10 can exlain why a relatively weak incumbent like chameleonjohn.com refers the relatively strong cometitor like RetailMeNot. Note that when G b < G G a and β > β a, as shown in art (iv) of Proosition 10, 24

26 either strong or weak-brand referral could shut off the otential entrant. In this case, each incumbent may choose to shoulder the cost of referral or just wait for the rival s referral and thus take a free ride. In this case, there also exists a mixed referral strategy equilibrium wherein seller 1 refers seller 2 with robability τ 1 and seller 2 refers seller 1 with robability τ 2, where τ 1 = ( 1)(β α 2) α and τ 1( 1)(β α 2)+α 2( 2) 2 = α2(2) (1 β)α1 α 2( 2)+β (see Aendix B3). 5 Consumer Welfare Analysis In this section, we investigate how the use of online cometitor referral influences consumer welfare. From the analysis in the revious sections, if online referral is used, then in equilibrium, seller 3 is always blocked out of the market. It is not clear a riori what the effects of referral would be on consumer welfare. This is because, on the one hand, the referral is eroding the monooly ower of a layer since it brings higher rice transarency in the marketlace whereby more consumers are able to comare rices and buy the good available at a lower rate. On the other hand, consumers may be worse off because referring cometitors imlies less market cometition since an additional layer is blocked out of the market. Denote consumer welfare by w. If we treat the sellers and customers as an integrated system, the market rices cannot influence the total welfare (i.e., firm rofits lus consumer welfare) but only affect the surlus division among different entities. Obviously, if seller 3 does not enter the market, the total welfare is equal to consumers aggregate utilities obtained from consuming roducts, v. If seller 3 enters the market with a cost of G, the total welfare will be v G. Thus, the consumer welfare is given by w = v G 3 i=1 π i if seller 3 enters the market, or w = v 2 i=1 π i if seller 3 does not. 5.1 Consumer Welfare under Base Model We first discuss the base model. As shown in Proosition 4, online referral (one-way) will be used at the equilibrium to deter entry only if αβ(1 β)v < G αβ(1 β) v and β > α. In this case, the referring firm earns a rofit of α(1 α)v and the referred firm s rofit equals to αv. Therefore, consumer welfare under referral, denoted by w R, is equal to v α(1 α)v αv; that is, w R = (1 α) 2 v. (28) However, if online referral is not alied, seller 3 will enter the market and the rofits of the three sellers are given by Eq.(8); that is, π 1 = π 2 = α(1 β)v and π 3 = αβ(1 β) v G. 25

Econ 101A Midterm 2 Th 8 April 2009.

Econ 101A Midterm 2 Th 8 April 2009. Econ A Midterm Th 8 Aril 9. You have aroximately hour and minutes to answer the questions in the midterm. I will collect the exams at. shar. Show your work, and good luck! Problem. Production (38 oints).

More information

Handout #3: Peak Load Pricing

Handout #3: Peak Load Pricing andout #3: Peak Load Pricing Consider a firm that exeriences two kinds of costs a caacity cost and a marginal cost ow should caacity be riced? This issue is alicable to a wide variety of industries, including

More information

Economics 101. Lecture 7 - Monopoly and Oligopoly

Economics 101. Lecture 7 - Monopoly and Oligopoly Economics 0 Lecture 7 - Monooly and Oligooly Production Equilibrium After having exlored Walrasian equilibria with roduction in the Robinson Crusoe economy, we will now ste in to a more general setting.

More information

COMMUNICATION BETWEEN SHAREHOLDERS 1

COMMUNICATION BETWEEN SHAREHOLDERS 1 COMMUNICATION BTWN SHARHOLDRS 1 A B. O A : A D Lemma B.1. U to µ Z r 2 σ2 Z + σ2 X 2r ω 2 an additive constant that does not deend on a or θ, the agents ayoffs can be written as: 2r rθa ω2 + θ µ Y rcov

More information

Online Appendix to Accompany AComparisonof Traditional and Open-Access Appointment Scheduling Policies

Online Appendix to Accompany AComparisonof Traditional and Open-Access Appointment Scheduling Policies Online Aendix to Accomany AComarisonof Traditional and Oen-Access Aointment Scheduling Policies Lawrence W. Robinson Johnson Graduate School of Management Cornell University Ithaca, NY 14853-6201 lwr2@cornell.edu

More information

A search cost model of obfuscation

A search cost model of obfuscation RAND Journal of Economics Vol. 43, No. 3, Fall 2012. 417 441 A search cost model of obfuscation Glenn Ellison and Alexander Wolitzky This article develos models in which obfuscation is individually rational

More information

Online Appendix for The Timing and Method of Payment in Mergers when Acquirers Are Financially Constrained

Online Appendix for The Timing and Method of Payment in Mergers when Acquirers Are Financially Constrained Online Aendix for The Timing and Method of Payment in Mergers when Acquirers Are Financially Constrained Alexander S. Gorbenko USC Marshall School of Business Andrey Malenko MIT Sloan School of Management

More information

Oligopolistic Pricing with Online Search

Oligopolistic Pricing with Online Search Oligoolistic Pricing with Online Search Lizhen Xu, Jianqing Chen, and Andrew Whinston Lizhen Xu is a Ph.D. candidate in the Deartment of Information, Risk, and Oerations Management, Red McCombs School

More information

MANAGEMENT SCIENCE doi /mnsc ec

MANAGEMENT SCIENCE doi /mnsc ec MANAGEMENT SCIENCE doi 0287/mnsc0800993ec e-comanion ONLY AVAILABLE IN ELECTRONIC FORM informs 2009 INFORMS Electronic Comanion Otimal Entry Timing in Markets with Social Influence by Yogesh V Joshi, David

More information

4. CONTINUOUS VARIABLES AND ECONOMIC APPLICATIONS

4. CONTINUOUS VARIABLES AND ECONOMIC APPLICATIONS STATIC GAMES 4. CONTINUOUS VARIABLES AND ECONOMIC APPLICATIONS Universidad Carlos III de Madrid CONTINUOUS VARIABLES In many games, ure strategies that layers can choose are not only, 3 or any other finite

More information

PROFIT MAXIMIZATION. π = p y Σ n i=1 w i x i (2)

PROFIT MAXIMIZATION. π = p y Σ n i=1 w i x i (2) PROFIT MAXIMIZATION DEFINITION OF A NEOCLASSICAL FIRM A neoclassical firm is an organization that controls the transformation of inuts (resources it owns or urchases into oututs or roducts (valued roducts

More information

Exercise 2: Equivalence of the first two definitions for a differentiable function. is a convex combination of

Exercise 2: Equivalence of the first two definitions for a differentiable function. is a convex combination of March 07 Mathematical Foundations John Riley Module Marginal analysis and single variable calculus 6 Eercises Eercise : Alternative definitions of a concave function (a) For and that 0, and conve combination

More information

Solutions to exercises on delays. P (x = 0 θ = 1)P (θ = 1) P (x = 0) We can replace z in the first equation by its value in the second equation.

Solutions to exercises on delays. P (x = 0 θ = 1)P (θ = 1) P (x = 0) We can replace z in the first equation by its value in the second equation. Ec 517 Christohe Chamley Solutions to exercises on delays Ex 1: P (θ = 1 x = 0) = P (x = 0 θ = 1)P (θ = 1) P (x = 0) = 1 z)µ (1 z)µ + 1 µ. The value of z is solution of µ c = δµz(1 c). We can relace z

More information

1 Gambler s Ruin Problem

1 Gambler s Ruin Problem Coyright c 2017 by Karl Sigman 1 Gambler s Ruin Problem Let N 2 be an integer and let 1 i N 1. Consider a gambler who starts with an initial fortune of $i and then on each successive gamble either wins

More information

International Trade with a Public Intermediate Good and the Gains from Trade

International Trade with a Public Intermediate Good and the Gains from Trade International Trade with a Public Intermediate Good and the Gains from Trade Nobuhito Suga Graduate School of Economics, Nagoya University Makoto Tawada Graduate School of Economics, Nagoya University

More information

Advance Selling in the Presence of Experienced Consumers

Advance Selling in the Presence of Experienced Consumers Advance Selling in the Presence of Eerienced Consumers Oksana Loginova X. Henry Wang Chenhang Zeng June 30, 011 Abstract The advance selling strategy is imlemented when a firm offers consumers the oortunity

More information

Transmission charging and market distortion

Transmission charging and market distortion Transmission charging and market distortion Andy Philott Tony Downward Keith Ruddell s Electric Power Otimization Centre University of Auckland www.eoc.org.nz IPAM worksho, UCLA January 13, 2016 1/56 Outline

More information

Optimism, Delay and (In)Efficiency in a Stochastic Model of Bargaining

Optimism, Delay and (In)Efficiency in a Stochastic Model of Bargaining Otimism, Delay and In)Efficiency in a Stochastic Model of Bargaining Juan Ortner Boston University Setember 10, 2012 Abstract I study a bilateral bargaining game in which the size of the surlus follows

More information

Theory of Externalities Partial Equilibrium Analysis

Theory of Externalities Partial Equilibrium Analysis Theory of Externalities Partial Equilibrium Analysis Definition: An externality is resent whenever the well being of a consumer or the roduction ossibilities of a firm are directly affected by the actions

More information

The Impact of Demand Uncertainty on Consumer Subsidies for Green Technology Adoption

The Impact of Demand Uncertainty on Consumer Subsidies for Green Technology Adoption "The Imact of Consumer Subsidies for Green Technology Adotion." Cohen, Maxine C., Ruben Lobel and Georgia Perakis. Management Science Vol. 62, No. 5 (2016: 1235-1258. htt://dx.doi.org/10.1287/mnsc.2015.2173

More information

Combining Logistic Regression with Kriging for Mapping the Risk of Occurrence of Unexploded Ordnance (UXO)

Combining Logistic Regression with Kriging for Mapping the Risk of Occurrence of Unexploded Ordnance (UXO) Combining Logistic Regression with Kriging for Maing the Risk of Occurrence of Unexloded Ordnance (UXO) H. Saito (), P. Goovaerts (), S. A. McKenna (2) Environmental and Water Resources Engineering, Deartment

More information

Voting and Lobbying - 3 Models

Voting and Lobbying - 3 Models Voting and obbying - 3 Models Series of 3 aers eloring the effects of olitical actions on market outcomes. Current theories of regulation unsatisfying (to me!: Toulouse School: Agency Model regulators

More information

Feedback-error control

Feedback-error control Chater 4 Feedback-error control 4.1 Introduction This chater exlains the feedback-error (FBE) control scheme originally described by Kawato [, 87, 8]. FBE is a widely used neural network based controller

More information

8 STOCHASTIC PROCESSES

8 STOCHASTIC PROCESSES 8 STOCHASTIC PROCESSES The word stochastic is derived from the Greek στoχαστικoς, meaning to aim at a target. Stochastic rocesses involve state which changes in a random way. A Markov rocess is a articular

More information

Sequential Choice of Sharing Rules in Collective Contests

Sequential Choice of Sharing Rules in Collective Contests Sequential Choice of Sharing Rules in Collective Contests Pau Balart Sabine Flamand Oliver Gürtler Orestis Troumounis February 26, 2017 Abstract Grous cometing for a rize need to determine how to distribute

More information

Monopolist s mark-up and the elasticity of substitution

Monopolist s mark-up and the elasticity of substitution Croatian Oerational Research Review 377 CRORR 8(7), 377 39 Monoolist s mark-u and the elasticity of substitution Ilko Vrankić, Mira Kran, and Tomislav Herceg Deartment of Economic Theory, Faculty of Economics

More information

Shadow Computing: An Energy-Aware Fault Tolerant Computing Model

Shadow Computing: An Energy-Aware Fault Tolerant Computing Model Shadow Comuting: An Energy-Aware Fault Tolerant Comuting Model Bryan Mills, Taieb Znati, Rami Melhem Deartment of Comuter Science University of Pittsburgh (bmills, znati, melhem)@cs.itt.edu Index Terms

More information

2x2x2 Heckscher-Ohlin-Samuelson (H-O-S) model with factor substitution

2x2x2 Heckscher-Ohlin-Samuelson (H-O-S) model with factor substitution 2x2x2 Heckscher-Ohlin-amuelson (H-O- model with factor substitution The HAT ALGEBRA of the Heckscher-Ohlin model with factor substitution o far we were dealing with the easiest ossible version of the H-O-

More information

Approximate Dynamic Programming for Dynamic Capacity Allocation with Multiple Priority Levels

Approximate Dynamic Programming for Dynamic Capacity Allocation with Multiple Priority Levels Aroximate Dynamic Programming for Dynamic Caacity Allocation with Multile Priority Levels Alexander Erdelyi School of Oerations Research and Information Engineering, Cornell University, Ithaca, NY 14853,

More information

Trading OTC and Incentives to Clear Centrally

Trading OTC and Incentives to Clear Centrally Trading OTC and Incentives to Clear Centrally Gaetano Antinolfi Francesca Caraella Francesco Carli March 1, 2013 Abstract Central counterparties CCPs have been art of the modern financial system since

More information

Net Neutrality with Competing Internet Service Providers

Net Neutrality with Competing Internet Service Providers Net Neutrality with Cometing Internet Service Providers Marc Bourreau, Frago Kourandi y, Tommaso Valletti z December 1, 011 Abstract We roose a two-sided model with two cometing Internet Service Providers

More information

AP Physics C: Electricity and Magnetism 2004 Scoring Guidelines

AP Physics C: Electricity and Magnetism 2004 Scoring Guidelines AP Physics C: Electricity and Magnetism 4 Scoring Guidelines The materials included in these files are intended for noncommercial use by AP teachers for course and exam rearation; ermission for any other

More information

Sets of Real Numbers

Sets of Real Numbers Chater 4 Sets of Real Numbers 4. The Integers Z and their Proerties In our revious discussions about sets and functions the set of integers Z served as a key examle. Its ubiquitousness comes from the fact

More information

How Often Should You Reward Your Salesforce? Multi-Period. Incentives and Effort Dynamics

How Often Should You Reward Your Salesforce? Multi-Period. Incentives and Effort Dynamics How Often Should You Reward Your Salesforce? Multi-Period Incentives and Effort Dynamics Kinshuk Jerath Fei Long kj2323@gsb.columbia.edu FeiLong18@gsb.columbia.edu Columbia Business School Columbia Business

More information

John Weatherwax. Analysis of Parallel Depth First Search Algorithms

John Weatherwax. Analysis of Parallel Depth First Search Algorithms Sulementary Discussions and Solutions to Selected Problems in: Introduction to Parallel Comuting by Viin Kumar, Ananth Grama, Anshul Guta, & George Karyis John Weatherwax Chater 8 Analysis of Parallel

More information

Lower Confidence Bound for Process-Yield Index S pk with Autocorrelated Process Data

Lower Confidence Bound for Process-Yield Index S pk with Autocorrelated Process Data Quality Technology & Quantitative Management Vol. 1, No.,. 51-65, 15 QTQM IAQM 15 Lower onfidence Bound for Process-Yield Index with Autocorrelated Process Data Fu-Kwun Wang * and Yeneneh Tamirat Deartment

More information

Voting with Behavioral Heterogeneity

Voting with Behavioral Heterogeneity Voting with Behavioral Heterogeneity Youzong Xu Setember 22, 2016 Abstract This aer studies collective decisions made by behaviorally heterogeneous voters with asymmetric information. Here behavioral heterogeneity

More information

Competition among networks highlights the unexpected power of the weak

Competition among networks highlights the unexpected power of the weak 17 12 19 34 29 15 7 9 1 10 28 25 36 4 8 2 1 3 22 23 13 3 24 30 27 35 20 32 30 32 25 23 2 26 6 27 37 11 31 16 11 38 9 26 5 8 19 5 18 7 33 16 13 20 28 12 18 21 39 17 21 24 15 40 29 31 14 6 4 22 10 41 14

More information

Symmetric and Asymmetric Equilibria in a Spatial Duopoly

Symmetric and Asymmetric Equilibria in a Spatial Duopoly This version: February 003 Symmetric and Asymmetric Equilibria in a Satial Duooly Marcella Scrimitore Deartment of Economics, University of Lecce, Italy Jel Classification: L3, R39 Abstract We develo a

More information

Analysis of Renewable Energy Policy: Feed-in Tariffs with Minimum Price Guarantees

Analysis of Renewable Energy Policy: Feed-in Tariffs with Minimum Price Guarantees Analysis of Renewable Energy Policy: Feed-in ariffs with Minimum Price Guarantees Luciana Barbosa a, Paulo Ferrão a, Artur Rodrigues b, and Alberto Sardinha c a MI Portugal Program and Instituto Suerior

More information

Statics and dynamics: some elementary concepts

Statics and dynamics: some elementary concepts 1 Statics and dynamics: some elementary concets Dynamics is the study of the movement through time of variables such as heartbeat, temerature, secies oulation, voltage, roduction, emloyment, rices and

More information

DYNAMIC COSTS AND MORAL HAZARD: A DUALITY BASED APPROACH. Guy Arie 1

DYNAMIC COSTS AND MORAL HAZARD: A DUALITY BASED APPROACH. Guy Arie 1 DYNAMIC COSTS AND MORAL HAZARD: A DUALITY BASED APPROACH Guy Arie 1 Abstract The marginal cost of effort often increases as effort is exerted. In a dynamic moral hazard setting, dynamically increasing

More information

Net Neutrality with Competing Internet Platforms

Net Neutrality with Competing Internet Platforms Net Neutrality with Cometing Internet Platforms Marc Bourreau y, Frago Kourandi z, Tommaso Valletti x March 21, 2013 Abstract We roose a two-sided model with two cometing Internet latforms, and a continuum

More information

arxiv: v1 [cs.gt] 2 Nov 2018

arxiv: v1 [cs.gt] 2 Nov 2018 Tight Aroximation Ratio of Anonymous Pricing Yaonan Jin Pinyan Lu Qi Qi Zhihao Gavin Tang Tao Xiao arxiv:8.763v [cs.gt] 2 Nov 28 Abstract We consider two canonical Bayesian mechanism design settings. In

More information

Information collection on a graph

Information collection on a graph Information collection on a grah Ilya O. Ryzhov Warren Powell February 10, 2010 Abstract We derive a knowledge gradient olicy for an otimal learning roblem on a grah, in which we use sequential measurements

More information

MATHEMATICAL MODELLING OF THE WIRELESS COMMUNICATION NETWORK

MATHEMATICAL MODELLING OF THE WIRELESS COMMUNICATION NETWORK Comuter Modelling and ew Technologies, 5, Vol.9, o., 3-39 Transort and Telecommunication Institute, Lomonosov, LV-9, Riga, Latvia MATHEMATICAL MODELLIG OF THE WIRELESS COMMUICATIO ETWORK M. KOPEETSK Deartment

More information

Stationary Monetary Equilibria with Strictly Increasing Value Functions and Non-Discrete Money Holdings Distributions: An Indeterminacy Result

Stationary Monetary Equilibria with Strictly Increasing Value Functions and Non-Discrete Money Holdings Distributions: An Indeterminacy Result CIRJE-F-615 Stationary Monetary Equilibria with Strictly Increasing Value Functions and Non-Discrete Money Holdings Distributions: An Indeterminacy Result Kazuya Kamiya University of Toyo Taashi Shimizu

More information

The Mathematics of Winning Streaks

The Mathematics of Winning Streaks The Mathematics of Winning Streaks Erik Leffler lefflererik@gmail.com under the direction of Prof. Henrik Eriksson Deartment of Comuter Science and Communications Royal Institute of Technology Research

More information

4. Score normalization technical details We now discuss the technical details of the score normalization method.

4. Score normalization technical details We now discuss the technical details of the score normalization method. SMT SCORING SYSTEM This document describes the scoring system for the Stanford Math Tournament We begin by giving an overview of the changes to scoring and a non-technical descrition of the scoring rules

More information

A note on the preferred hedge instrument

A note on the preferred hedge instrument ingnan University Digital Commons @ ingnan University ong ong Institute o Business tudies Working aer eries ong ong Institute o Business tudies 香港商學研究所 6-5 A note on the reerred hedge instrument Arthur

More information

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK Towards understanding the Lorenz curve using the Uniform distribution Chris J. Stehens Newcastle City Council, Newcastle uon Tyne, UK (For the Gini-Lorenz Conference, University of Siena, Italy, May 2005)

More information

Coordination Motives and Competition for Attention in Information Markets *

Coordination Motives and Competition for Attention in Information Markets * Coordination Motives and Cometition for Attention in Information Markets * Simone Galerti and Isabel Trevino UCSD May 20, 207 Abstract We examine markets for information where consumers want to coordinate

More information

Positive decomposition of transfer functions with multiple poles

Positive decomposition of transfer functions with multiple poles Positive decomosition of transfer functions with multile oles Béla Nagy 1, Máté Matolcsi 2, and Márta Szilvási 1 Deartment of Analysis, Technical University of Budaest (BME), H-1111, Budaest, Egry J. u.

More information

Structuring M&A Offers: Auctions, Negotiations and Go-Shop. Provisions

Structuring M&A Offers: Auctions, Negotiations and Go-Shop. Provisions Structuring M&A Offers: Auctions, Negotiations and Go-Sho Provisions Zhe Wang November 4, 01 Job Market Paer Abstract An imortant yet understudied asect of mergers and acquisitions is the selling rocedure.

More information

16.2. Infinite Series. Introduction. Prerequisites. Learning Outcomes

16.2. Infinite Series. Introduction. Prerequisites. Learning Outcomes Infinite Series 6.2 Introduction We extend the concet of a finite series, met in Section 6., to the situation in which the number of terms increase without bound. We define what is meant by an infinite

More information

16.2. Infinite Series. Introduction. Prerequisites. Learning Outcomes

16.2. Infinite Series. Introduction. Prerequisites. Learning Outcomes Infinite Series 6. Introduction We extend the concet of a finite series, met in section, to the situation in which the number of terms increase without bound. We define what is meant by an infinite series

More information

Prospect Theory Explains Newsvendor Behavior: The Role of Reference Points

Prospect Theory Explains Newsvendor Behavior: The Role of Reference Points Submitted to Management Science manuscrit (Please, rovide the mansucrit number! Authors are encouraged to submit new aers to INFORMS journals by means of a style file temlate, which includes the journal

More information

arxiv: v1 [physics.data-an] 26 Oct 2012

arxiv: v1 [physics.data-an] 26 Oct 2012 Constraints on Yield Parameters in Extended Maximum Likelihood Fits Till Moritz Karbach a, Maximilian Schlu b a TU Dortmund, Germany, moritz.karbach@cern.ch b TU Dortmund, Germany, maximilian.schlu@cern.ch

More information

CHAPTER-II Control Charts for Fraction Nonconforming using m-of-m Runs Rules

CHAPTER-II Control Charts for Fraction Nonconforming using m-of-m Runs Rules CHAPTER-II Control Charts for Fraction Nonconforming using m-of-m Runs Rules. Introduction: The is widely used in industry to monitor the number of fraction nonconforming units. A nonconforming unit is

More information

Homework Solution 4 for APPM4/5560 Markov Processes

Homework Solution 4 for APPM4/5560 Markov Processes Homework Solution 4 for APPM4/556 Markov Processes 9.Reflecting random walk on the line. Consider the oints,,, 4 to be marked on a straight line. Let X n be a Markov chain that moves to the right with

More information

An Analysis of Reliable Classifiers through ROC Isometrics

An Analysis of Reliable Classifiers through ROC Isometrics An Analysis of Reliable Classifiers through ROC Isometrics Stijn Vanderlooy s.vanderlooy@cs.unimaas.nl Ida G. Srinkhuizen-Kuyer kuyer@cs.unimaas.nl Evgueni N. Smirnov smirnov@cs.unimaas.nl MICC-IKAT, Universiteit

More information

MATH 2710: NOTES FOR ANALYSIS

MATH 2710: NOTES FOR ANALYSIS MATH 270: NOTES FOR ANALYSIS The main ideas we will learn from analysis center around the idea of a limit. Limits occurs in several settings. We will start with finite limits of sequences, then cover infinite

More information

Elementary Analysis in Q p

Elementary Analysis in Q p Elementary Analysis in Q Hannah Hutter, May Szedlák, Phili Wirth November 17, 2011 This reort follows very closely the book of Svetlana Katok 1. 1 Sequences and Series In this section we will see some

More information

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Technical Sciences and Alied Mathematics MODELING THE RELIABILITY OF CISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Cezar VASILESCU Regional Deartment of Defense Resources Management

More information

A Social Welfare Optimal Sequential Allocation Procedure

A Social Welfare Optimal Sequential Allocation Procedure A Social Welfare Otimal Sequential Allocation Procedure Thomas Kalinowsi Universität Rostoc, Germany Nina Narodytsa and Toby Walsh NICTA and UNSW, Australia May 2, 201 Abstract We consider a simle sequential

More information

Proof: We follow thearoach develoed in [4]. We adot a useful but non-intuitive notion of time; a bin with z balls at time t receives its next ball at

Proof: We follow thearoach develoed in [4]. We adot a useful but non-intuitive notion of time; a bin with z balls at time t receives its next ball at A Scaling Result for Exlosive Processes M. Mitzenmacher Λ J. Sencer We consider the following balls and bins model, as described in [, 4]. Balls are sequentially thrown into bins so that the robability

More information

A CONCRETE EXAMPLE OF PRIME BEHAVIOR IN QUADRATIC FIELDS. 1. Abstract

A CONCRETE EXAMPLE OF PRIME BEHAVIOR IN QUADRATIC FIELDS. 1. Abstract A CONCRETE EXAMPLE OF PRIME BEHAVIOR IN QUADRATIC FIELDS CASEY BRUCK 1. Abstract The goal of this aer is to rovide a concise way for undergraduate mathematics students to learn about how rime numbers behave

More information

Pro-Consumer Price Ceilings under Uncertainty

Pro-Consumer Price Ceilings under Uncertainty Pro-Consumer Price Ceilings under Uncertainty John Bennett y Ioana Chioveanu z March 11, 2015 Abstract We examine ro-consumer rice ceilings under regulatory uncertainty about demand and suly. In a erfectly

More information

Estimation of the large covariance matrix with two-step monotone missing data

Estimation of the large covariance matrix with two-step monotone missing data Estimation of the large covariance matrix with two-ste monotone missing data Masashi Hyodo, Nobumichi Shutoh 2, Takashi Seo, and Tatjana Pavlenko 3 Deartment of Mathematical Information Science, Tokyo

More information

MATH 361: NUMBER THEORY EIGHTH LECTURE

MATH 361: NUMBER THEORY EIGHTH LECTURE MATH 361: NUMBER THEORY EIGHTH LECTURE 1. Quadratic Recirocity: Introduction Quadratic recirocity is the first result of modern number theory. Lagrange conjectured it in the late 1700 s, but it was first

More information

Topic: Lower Bounds on Randomized Algorithms Date: September 22, 2004 Scribe: Srinath Sridhar

Topic: Lower Bounds on Randomized Algorithms Date: September 22, 2004 Scribe: Srinath Sridhar 15-859(M): Randomized Algorithms Lecturer: Anuam Guta Toic: Lower Bounds on Randomized Algorithms Date: Setember 22, 2004 Scribe: Srinath Sridhar 4.1 Introduction In this lecture, we will first consider

More information

Information collection on a graph

Information collection on a graph Information collection on a grah Ilya O. Ryzhov Warren Powell October 25, 2009 Abstract We derive a knowledge gradient olicy for an otimal learning roblem on a grah, in which we use sequential measurements

More information

1 C 6, 3 4, 4. Player S 5, 2 3, 1

1 C 6, 3 4, 4. Player S 5, 2 3, 1 University of alifornia, Davis - Deartment of Economics PRING EN / ARE : MIROEONOMI TEORY Professor Giacomo Bonanno ====================================================================== MIDTERM EXAM ANWER

More information

Limiting Price Discrimination when Selling Products with Positive Network Externalities

Limiting Price Discrimination when Selling Products with Positive Network Externalities Limiting Price Discrimination when Selling Products with Positive Network Externalities Luděk Cigler, Wolfgang Dvořák, Monika Henzinger, Martin Starnberger University of Vienna, Faculty of Comuter Science,

More information

Bertrand Model of Price Competition. Advanced Microeconomic Theory 1

Bertrand Model of Price Competition. Advanced Microeconomic Theory 1 Bertrand Model of Price Competition Advanced Microeconomic Theory 1 ҧ Bertrand Model of Price Competition Consider: An industry with two firms, 1 and 2, selling a homogeneous product Firms face market

More information

Optimal Organization of Financial Intermediaries

Optimal Organization of Financial Intermediaries Otimal Organization of Financial Intermediaries Siros Bougheas Tianxi Wang Setember 2014 Abstract This aer rovides a unified framework for endogenizing two distinct organizational structures for financial

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY Introduction to Optimization (Spring 2004) Midterm Solutions

MASSACHUSETTS INSTITUTE OF TECHNOLOGY Introduction to Optimization (Spring 2004) Midterm Solutions MASSAHUSTTS INSTITUT OF THNOLOGY 15.053 Introduction to Otimization (Sring 2004) Midterm Solutions Please note that these solutions are much more detailed that what was required on the midterm. Aggregate

More information

School of Economics and Management

School of Economics and Management School of Economics and Management TECHNICAL UNIVERSITY OF LISBON Deartment of Economics Carlos Pestana Barros & Nicolas Peyoch José Pedro Pontes A Comarative Analysis of Productivity Change in Italian

More information

Standards Battles and Public Policy

Standards Battles and Public Policy Standards Battles and Public Policy Luís M B Cabral New York University and CEPR Tobias Kretschmer London School of Economics Aril 004 Abstract We examine the effectiveness of ublic olicy in a context

More information

Centralized decision making against informed lobbying

Centralized decision making against informed lobbying Centralized decision making against informed lobbying Rafael Costa Lima Humberto Moreira Thierry Verdier USP FGV PSE March 9, 01 Abstract We re-address the trade-off between centralized and decentralized

More information

Elliptic Curves and Cryptography

Elliptic Curves and Cryptography Ellitic Curves and Crytograhy Background in Ellitic Curves We'll now turn to the fascinating theory of ellitic curves. For simlicity, we'll restrict our discussion to ellitic curves over Z, where is a

More information

Brownian Motion and Random Prime Factorization

Brownian Motion and Random Prime Factorization Brownian Motion and Random Prime Factorization Kendrick Tang June 4, 202 Contents Introduction 2 2 Brownian Motion 2 2. Develoing Brownian Motion.................... 2 2.. Measure Saces and Borel Sigma-Algebras.........

More information

Oligopoly. Oligopoly. Xiang Sun. Wuhan University. March 23 April 6, /149

Oligopoly. Oligopoly. Xiang Sun. Wuhan University. March 23 April 6, /149 Oligopoly Xiang Sun Wuhan University March 23 April 6, 2016 1/149 Outline 1 Introduction 2 Game theory 3 Oligopoly models 4 Cournot competition Two symmetric firms Two asymmetric firms Many symmetric firms

More information

Pell's Equation and Fundamental Units Pell's equation was first introduced to me in the number theory class at Caltech that I never comleted. It was r

Pell's Equation and Fundamental Units Pell's equation was first introduced to me in the number theory class at Caltech that I never comleted. It was r Pell's Equation and Fundamental Units Kaisa Taiale University of Minnesota Summer 000 1 Pell's Equation and Fundamental Units Pell's equation was first introduced to me in the number theory class at Caltech

More information

Slides Prepared by JOHN S. LOUCKS St. Edward s s University Thomson/South-Western. Slide

Slides Prepared by JOHN S. LOUCKS St. Edward s s University Thomson/South-Western. Slide s Preared by JOHN S. LOUCKS St. Edward s s University 1 Chater 11 Comarisons Involving Proortions and a Test of Indeendence Inferences About the Difference Between Two Poulation Proortions Hyothesis Test

More information

GSOE9210 Engineering Decisions

GSOE9210 Engineering Decisions GSOE9 Engineering Decisions Problem Set 5. Consider the river roblem described in lectures: f f V B A B + (a) For =, what is the sloe of the Bayes indifference line through A? (b) Draw the Bayes indifference

More information

Partial Identification in Triangular Systems of Equations with Binary Dependent Variables

Partial Identification in Triangular Systems of Equations with Binary Dependent Variables Partial Identification in Triangular Systems of Equations with Binary Deendent Variables Azeem M. Shaikh Deartment of Economics University of Chicago amshaikh@uchicago.edu Edward J. Vytlacil Deartment

More information

Deceptive Advertising with Rational Buyers

Deceptive Advertising with Rational Buyers Deceptive Advertising with Rational Buyers September 6, 016 ONLINE APPENDIX In this Appendix we present in full additional results and extensions which are only mentioned in the paper. In the exposition

More information

INTRODUCTION. Please write to us at if you have any comments or ideas. We love to hear from you.

INTRODUCTION. Please write to us at if you have any comments or ideas. We love to hear from you. Casio FX-570ES One-Page Wonder INTRODUCTION Welcome to the world of Casio s Natural Dislay scientific calculators. Our exeriences of working with eole have us understand more about obstacles eole face

More information

Recovering preferences in the household production framework: The case of averting behavior

Recovering preferences in the household production framework: The case of averting behavior Udo Ebert Recovering references in the household roduction framework: The case of averting behavior February 2002 * Address: Deartment of Economics, University of Oldenburg, D-26 Oldenburg, ermany Tel.:

More information

CMSC 425: Lecture 4 Geometry and Geometric Programming

CMSC 425: Lecture 4 Geometry and Geometric Programming CMSC 425: Lecture 4 Geometry and Geometric Programming Geometry for Game Programming and Grahics: For the next few lectures, we will discuss some of the basic elements of geometry. There are many areas

More information

Entrepreneurship and new ventures finance. Designing a new business (3): Revenues and costs. Prof. Antonio Renzi

Entrepreneurship and new ventures finance. Designing a new business (3): Revenues and costs. Prof. Antonio Renzi Entrereneurshi and new ventures finance Designing a new business (3): Revenues and costs Prof. Antonio Renzi Agenda 1. Revenues analysis 2. Costs analysis 3. Break even analysis Revenue Model Primary Demand

More information

Tests for Two Proportions in a Stratified Design (Cochran/Mantel-Haenszel Test)

Tests for Two Proportions in a Stratified Design (Cochran/Mantel-Haenszel Test) Chater 225 Tests for Two Proortions in a Stratified Design (Cochran/Mantel-Haenszel Test) Introduction In a stratified design, the subects are selected from two or more strata which are formed from imortant

More information

Measuring center and spread for density curves. Calculating probabilities using the standard Normal Table (CIS Chapter 8, p 105 mainly p114)

Measuring center and spread for density curves. Calculating probabilities using the standard Normal Table (CIS Chapter 8, p 105 mainly p114) Objectives Density curves Measuring center and sread for density curves Normal distributions The 68-95-99.7 (Emirical) rule Standardizing observations Calculating robabilities using the standard Normal

More information

arxiv:cond-mat/ v2 25 Sep 2002

arxiv:cond-mat/ v2 25 Sep 2002 Energy fluctuations at the multicritical oint in two-dimensional sin glasses arxiv:cond-mat/0207694 v2 25 Se 2002 1. Introduction Hidetoshi Nishimori, Cyril Falvo and Yukiyasu Ozeki Deartment of Physics,

More information

School of Economic Sciences

School of Economic Sciences School of Economic Sciences Working Paer Series WP 2017-1 Information Transmission during the Trial: The ole of Punitive Damages and egal osts Ana Esinola-Arredondo Felix Munoz-Garcia January, 2017 Information

More information

SIGNALING IN CONTESTS. Tomer Ifergane and Aner Sela. Discussion Paper No November 2017

SIGNALING IN CONTESTS. Tomer Ifergane and Aner Sela. Discussion Paper No November 2017 SIGNALING IN CONTESTS Tomer Ifergane and Aner Sela Discussion Paer No. 17-08 November 017 Monaster Center for Economic Research Ben-Gurion University of the Negev P.O. Box 653 Beer Sheva, Israel Fax: 97-8-647941

More information

Optimal Learning Policies for the Newsvendor Problem with Censored Demand and Unobservable Lost Sales

Optimal Learning Policies for the Newsvendor Problem with Censored Demand and Unobservable Lost Sales Otimal Learning Policies for the Newsvendor Problem with Censored Demand and Unobservable Lost Sales Diana Negoescu Peter Frazier Warren Powell Abstract In this aer, we consider a version of the newsvendor

More information

Priority pricing by a durable goods monopolist

Priority pricing by a durable goods monopolist Priority ricing by a durable goods monoolist João Correia-da-Silva February 15 th, 17. Abstract. A durable goods monoolist faces buyers with rivately observed valuations in two eriods, being unable to

More information

Agent Failures in Totally Balanced Games and Convex Games

Agent Failures in Totally Balanced Games and Convex Games Agent Failures in Totally Balanced Games and Convex Games Yoram Bachrach 1, Ian Kash 1, and Nisarg Shah 2 1 Microsoft Research Ltd, Cambridge, UK. {yobach,iankash}@microsoft.com 2 Comuter Science Deartment,

More information