Zsolt Katona - Research |
|
| Home | | | Research | | | Teaching | | | Dancing | | |
|
Work in progress / under review
Social Media Marketing: How Much Are Influentials Worth? Abstract This paper studies the competition between firms for social media influencers. Firms spend effort to convince influencers in a social network to recommend their products. The results show that the value of an influencer depends only on the structure of the influence network and not the level of competition in the product market. Influencers who exclusively cover a high number of consumers are more valuable to firms than those who mostly cover consumers also covered by other influencers. The total effort expended by firms is completely determined by the in-degree distribution of the network. Firm profits are highest when there are many consumers with a very low in-degree or with very high in-degree. Consumers with an intermediate level of in-degree contribute negatively to profits and high in-degree consumers increase profits when market competition is not intense. Prices are generally lower when consumers are covered by many influencers, however, firms are not always worse off with lower prices. The nature of consumer response to recommendations makes an important difference. When first impressions dominate, firm profits for dense networks are higher, but when recommendations have a cumulative influence profits are reduced as the network becomes dense. An extension models the incentives firms provide to influencers and how these affect influencers' network formation behavior. Direct monetary incentives lead to dense networks and consumers covered by many influencers. Such networks are only beneficial to firms when market competition is not intense. When the market is more competitive, firms are better off providing smaller direct incentives just enough to ensure a sparse coverage of consumers by influencers. Democracy in Product Design: Consumer Participation and Differentiation Strategies Abstract An increasing number of firms use social media to allow their customers to vote on new product design. This paper studies the implications of employing such a democratic product design (DPD) process on firm strategies. We use a linear city model with random locations to capture uncertainty about consumer preferences and to study the implication in monopoly and duopoly. We find that a monopolist will use a DPD as long as there is sufficient uncertainty about consumers. In a duopoly we find an asymmetric equilibrium with only one firm using DPD. The corresponding unique symmetric equilibrium in mixed strategies reveals that firms are more likely to use DPD when uncertainty is higher. We separate the informational and commitment effects of using DPD and show that the informational effect can lead to increased price competition and lower profits, whereas the commitment effect is positive for the firm using DPD. Furthermore, we show that preselecting candidates for the voting process can give substantial control to firms over the final winner, but this could lead to lower profits. Surprisingly, consumer welfare is higher when firms select candidates or when they do not commit to the winning consumer candidate. Finally, we study the level of consumer engagement in a DPD and find that a monopolist always benefits from a higher positive engagement and is hurt by negative engagement, although to a lesser extent. The results are reversed for a duopolist as negative sentiments can serve as an additional differentiator.
Quality Score that Makes You Invest Abstract This paper examines the role of quality score in ad auctions. Many search engines and other ad auction use a quality score to favor certain advertisers in the auction. The typical justification for quality score is the need to favor advertisers with higher click-through rates. This work examines the impact of different forms of quality scores on advertiser's quality investments. The results suggests that quality score plays an important role in providing incentives to advertisers to invest. Often a type of quality score that is stronger than one that only estimates click-through rates is optimal. Such a quality score not only increases advertisers qualities, but in turn raises the auctioneer's revenue. These results are stronger when there are (i) more advertising slots, (ii) when these slots are similar, and (iii) when quality investments are expensive.
Contextual Advertising Abstract This paper studies the strategic aspects of contextual advertising. Such advertising entails the display of relevant ads based on the topic of the content a consumer views and takes advantage of the possibility that consumers' content browsing preferences are indicative of their product preferences. The results show that contextual targeting impacts advertiser profit in two ways: first, advertising through relevant content topics helps advertisers reach consumers who have strong preferences for their products. Second, heterogeneity in consumers' content preferences can be leveraged to reduce product market competition, even when consumers are homogeneous in their product preferences. The contextual advertising intermediary's incentives to strategically design its content structure and the targeting precision are governed by the following forces. When product market competition is high, the intermediary offers homogeneous content and increases its targeting precision. This encourages each advertiser to bid for multiple keywords to preventing its competitors from advertising to the consumers. This may lead to an asymmetric equilibrium where one advertiser monopolizes all the advertising spaces to completely preempt competition. When product market competition is low, the intermediary offers heterogeneous content but intentionally decrease its targeting precision. This encourages each advertiser to bid for multiple advertising spaces in order to reach consumers who prefer its product.
Media, Aggregators and the Link Economy: Strategic Hyperlink Formation in Content Networks (download pdf) Abstract A key property of the World Wide Web is the possibility for firms to place virtually costless links to third-party content as a substitute or complement to their own content. This ability to hyperlink has enabled new types of players, such as search engines and content aggregators, to successfully enter content ecosystems, attracting traffic and revenues by hosting links to the content of others. This, in turn, has sparked a heated controversy between content producers and aggregators regarding the legitimacy and social costs/benefits of uninhibited free linking. This work is the first to model the implications of interrelated and strategic hyper-linking and content investments. Our results provide a nuanced view of the much-touted "link economy", highlighting both the beneficial consequences and the drawbacks of free hyperlinks for content producers and consumers. We show that content sites can reduce competition and improve profits by forming links to each other; in such networks one site makes high investments in content and other sites link to it. Interestingly, competitive dynamics often preclude the formation of link networks, even in settings where they would improve everyone's profits. Furthermore, such networks improve economic efficiency only when all members have similar abilities to produce content; otherwise the less capable nodes can free-ride on the content of the more capable nodes, reducing profits for the capable nodes as well as the average content quality available to consumers. Within these networks, aggregators have both positive and negative effects. By making it easier for consumers to access good quality content they increase the appeal of the entire content ecosystem relative to the alternatives. To the extent that this increases the total traffic flowing into the content ecosystem, aggregators can help increase the profits of the highest quality content sites. At the same time, however, the market entry of aggregators takes away some of the revenue that would otherwise go to pure content sites. Finally, by placing links to only a subset of available content, aggregators further increase competitive pressure on content sites. Interestingly, this can increase the likelihood that such sites will then attempt to alleviate the competitive pressure by forming link networks.
The Race for Sponsored Links: Bidding Patterns for Search Advertising (download pdf) Abstract Web sites invest significant resources in trying to influence their visibility among online search results. In addition to paying for sponsored links, they invest in methods known as search engine optimization (SEO) that improve the ranking of a site among the search results without improving its quality. We study the economic incentives of Web sites to invest in SEO and its implications on search engine and advertiser payoffs. We find that the process is equivalent to an all-pay auction with noise and headstarts. Our results show that, under certain conditions, a positive level of search engine optimization improves the search engine's ranking quality and thus the satisfaction of its visitors. In particular, if the quality of sites coincides with their valuation for visitors then search engine optimization serves as a mechanism that improves the ranking by correcting measurement errors. While this benefits consumers and increases traffic to the search engine, sites participating in search engine optimization could be worse off due to wasteful spending unless their valuation for traffic is very high. We also investigate how search engine optimization affects the revenues from sponsored links. Surprisingly, we find that in many cases search engine revenues are increased by SEO.
Network Formation and the Structure of the Commercial World Wide Web (download pdf) Abstract We model the commercial World Wide Web (WWW) as a directed graph emerging as the equilibrium of a game in which utility maximizing Web sites purchase (advertising) in-links from each other, while also setting the price of these links. A key feature of our model is that we consider sites to be heterogeneous in terms of their ``content", i.e. their inherent value to consumers. In a world where consumers `surf' on the WWW, sites' revenues/profits originate from two sources: (i) the sales of content (products) to consumers, and (ii) the sales of links (traffic) to other sites. We find that in equilibrium, higher content sites tend to purchase more advertising links mirroring the Dorfman-Steiner rule. Sites with higher content sell less advertising links and offer such links at higher prices. As such, there seems to be specialization across sites in terms of revenue models: high content sites tend to earns revenue from the sales of content while low content ones from the sales of traffic (advertising), a tendency that is increasing with the size of the browsing population. In an extension, we also allow sites to establish (reference) out-links to each other beyond the sales of advertising links and find that there is a general tendency to establish reference link to sites with higher content. Overall, there is a strong positive correlation between a site's content and the number of its in-links. We also explore network formation in the presence of search engines and find that the higher the proportion of people using these, the more sites have an incentive to specialize in certain ``content areas". Our results have interesting practical implications for `search-engine optimization', the pricing of Internet advertising as well as the choice of Internet business models. They also shed light on why successful search engines (e.g. Google) can use simple heuristics based on in-links to rank sites with respect to their content. Keywords: random graphs, game theory, heterogenous agents
Network Effects and Personal Influences: Diffusion of an Online Social Network Abstract We study word-of-mouth effects on the growth of social networks. Aggregate diffusion models ignore the possibility that an individual who is connected to many others in a social network may have a higher adoption probability (degree effect). Furthermore, the density of connections in a group of already adopted consumers may also affect the adoption of individuals connected to this group (clustering effect). We analyze data from one social network for which we know the individual connections between every pair of members. The results support the existence of both degree and clustering effects. We also show how a linear degree effect at the individual level can be the underlying determinant of the Bass model. Furthermore, we present a new methodology to determine the influential power of every individual. Surprisingly, we find that highly connected individuals have a lower average influential power (average influence on a particular person). However, the results suggest that the large number of connections counterbalances the small average influential power, such that highly connected individuals have a higher total influence than their less connected counterparts. We also find evidence that gender and age affect influential power. Keywords: Social Networks, New Product Diffusion
The Role of Search Engine Optimization in Search Rankings Abstract Web sites invest significant resources in trying to influence their visibility in online search results. We study the economic incentives of Web sites to invest in this process known as search engine optimization. We focus on methods that improve sites' ranking among the search results without affecting their quality. We find that the process is equivalent to an all-pay auction with noise and headstarts. Our results show that in equilibrium, under certain conditions, some positive level of search engine optimization improves the search engine's ranking and thus the satisfaction of its visitors. In particular, if the quality of sites coincides with their valuation for visitors then search engine optimization serves as a mechanism that improves the ranking by correcting measurement errors. While this benefits consumers and search engines, sites participating in search engine optimization could be worse off unless their valuation for traffic is very high. We also investigate how search engine optimization affects sites' investment in content and find that it can lead to underinvestment as a result of wasteful spending on search engine optimization. Keywords: Search engine marketing, All-pay auctions
Competing for Attention in Social Communication Markets Abstract We investigate agents' incentives to communicate with their peers in the context of the emergence of new social media technologies. In particular, we examine the effects of different types of communication cost structures through which new technologies change the way agents send and receive messages. These new social networking technologies make it possible to send the same message to many receivers for the same fixed cost and for negligible (often zero) marginal cost. This is distinct from traditional word-of-mouth which involves additional marginal cost for each intended recipient. Our results show that the new social networking technologies may reduce incentives of some individuals to send messages leading to less diverse communication. As the marginal cost of sending messages to additional people goes down, senders target more receivers leading to higher levels of competition for attention between receivers. The intensified competition requires higher effort on senders' part leading to lower payoffs and, eventually, fewer senders in the population. Although receivers are targeted by more senders and costs go down and they have access to a higher variety of messages, the variety of messages sent by the population decreases. We also study the role of location and horizontal preferences and examine sender differentiation.
Endogenous Homophily in Social Networks Abstract It is well known that in most social networks neighbors tend to be similar to each other. Previous research attributed this phenomenon to homophily, that is, that individuals like to connect to others that are similar to them. This paper proposes a network formation model in which individuals can change their location to become more similar before linking to each other to benefit more from the connection. An equilibrium is the outcome describing both the locations and the network resulting in an endogenously formed network with endogenous locations. The results show that the equilibrium network exhibits more homophily than a benchmark with fixed node locations and the number of links increases as link formation costs go down. As costs of changing one's location decrease the network becomes more clustered, less connected and denser. The results explain how certain individuals become opinion leaders throughout this process. The paper also provides a framework that generalizes homophily to examine vertical spaces in addition to horizontal ones. In a horizontal space the distance between two nodes determines how much they benefit from the link, whereas in a vertical space linking to a node with a higher vertical attribute provides more benefits. Keywords: Social networks, Homphily, Network formation
Quality and Advertising in a Vertically Differentiated Market (download pdf) Abstract We examine firms' quality positions when consumers can only consider purchasing products that they are informed about through advertising. Consumers compare the alternatives in their consideration set and choose the product that maximizes their utility net of price. Firms choose product quality in a first stage, advertising strategy in a second stage, and prices in the last stage. We study two forms of advertising - blanket and targeted . Under blanket advertising, firms communicate indiscriminately and a consumer's probability of seeing an ad depends on the level of ad expenditure. We find that when blanket advertising is relatively ineffective, i.e., it is costly to ensure that all consumers are informed, both firms choose a light ad spending. This allows the firms to be relatively undifferentiated in qualities, without concern of intense price competition. When blanket advertising is very effective, the high quality firm expends heavily on advertising, while its rival differentiates with a low quality product and expends lightly on advertising. Interestingly, in a mid range of advertising effectiveness one firm chooses a high quality product and its rival positions close by. In this way, the lower quality firm induces its rival to advertise only lightly to avoid fierce price competition. Under targeted advertising, firms choose the specific segment(s) they wish to inform. We identify conditions such that both firms choose equally high quality products, but advertise to different segments. This can result in a middle pocket of unserved consumers, even though consumers with lower willingness to pay are served. Keywords: Product Quality, Advertising Strategy, Differentiation, Competition agents
Communication Networks: Communities and Network Closure Revisited (download pdf) Abstract This project seeks to give a generalized, mathematical definition of communities in the context of communication networks, an important and well-studied category of social networks. Our work links the sociological literature on communities and recent work in the area of applied graph theory in the sense that it justifies mathematically the claim, made by sociologists such as Wellman and others, that sparse network ties, as opposed to tightly structured solidarity groups, can form a foundation for social communities. To this end, we develop and test a model of communities and explore the organizational dynamics embedded in two sets of communication networks: (1) a twoyear collection of email data from the now-defunct Enron Corporation of Houston, Texas, USA; and (2) a larger dataset of approximately 70,000 subscribers of a fixed-line telephone provider in Eastern Europe. We show that this model is more useful for identifying groups of individuals that have strong internal relationships in closed networks than those defined by more general models of network closure. Further, our model extends the scope of network closure effects proposed by other researchers working with communication networks. The method of identifying communities we detail provides greater methodological flexibility to network researchers, who can specify parameters based on substantive criteria pertaining to the specific data they are working with. The model thus allows communities to become more loosely defined (that is, they have fewer common members) or tightly defined (more common members) as required by the question under study. Keywords: Communities, Social networks; Communication; Network closure
Bricks and Clicks: The Impact of Product Returns on the Strategies of Multi-Channel Retailers (download pdf) Abstract The Internet has increased the flexibility of retailers allowing them to operate an online arm in addition to their physical stores. While the online channel offers retailers potential advantages in selling to customer segments that value the conveniences of online shopping, it also raises new challenges. These include the higher likelihood of costly product returns when customers' ability to 'touch and feel' the product is important in determining fit. We study competing retailers operating dual channels ("Bricks and Clicks") and examine how pricing strategies and the level of physical store assistance change as a result of the additional Internet outlet. On the supply-side, firms endogenously determine prices and how much to invest in store characteristics that assist customers in finding matching products (e.g., greater shelf display capacity, more qualified staff, floor samples). On the demand-side, we capture two relevant sources of customer heterogeneity: (i) retailer preference, and (ii) shopping trip costs. A central result we obtain is that when differentiation among retailers is not too high, having an online channel actually increases (costly) investment in store assistance levels while prices are set higher. We also examine how firms' range of product categories can vary between the two channels. Our main finding here is that even though it is costless to offer products online, retailers will sell only a limited assortment over the Internet. In particular, only "safe" products with a low chance of being returned will be sold online, while retailers' full range of products will be sold in physical stores. Keywords: Channels of Distribution, Retailing, Internet Marketing, Pricing, Reverse Logistics, Product Returns
CAM: A Spreading Activation Network Model of Subcategory Construction (download pdf) Abstract A large body of research suggests that people process the entities that they encounter by placing them into mental categories (Barsalou 1992). Although previous research examines how people access information in hierarchical category structures, it does not examine how people construct individual new categories and, in particular, how the locus of these new categories may depend on the structure of the entire hierarchy. We describe this latter process with a spreading activation model of hierarchical category structures that we call the Category Activation Model (CAM). In an experiment and an empirical study, we show that the CAM reliably predicts the probability that a person will construct a new category at a specific location within a category structure, and we provide evidence that accessibility is the mechanism that underlies category construction. Keywords: random graphs, game theory, heterogenous agents
A new class of scale-free graphs (download pdf) Abstract Consider the following modification of the Barab\'asi--Albert random graph. At every step a new vertex is added to the graph. It is connected to the old vertices randomly, with probabilities proportional to the degree of the other vertex, and independently of each other. We show that the proportion of vertices of degree $k$ decreases at the rate of $k^{-3}$. Furthermore, we prove a strong law of large numbers for the maximum degree. Keywords: random graph, scale-free disribution
Levels of a scale-free tree (download pdf) Abstract Consider the random graph model of Barabási and Albert, where we add a new vertex in every step and connect it to some old vertices with probabilities proportional to their degrees. If we connect it to only one of the old vertices the graph will be a tree. These graphs have been shown to have power law degree distributions, the same as observed in some large real-world networks. We show that the degree distribution is the same on every sufficiently high level of the tree.AMS classification: 05C80, 60C05 Keywords: random graphs, scale-free disribution, width of trees
Width of a scale-free tree (download pdf) Abstract Consider the random graph model of Barabási and Albert, where we add a new vertex in every step and connect it to some old vertices with probabilities proportional to their degrees. If we connect it to only one of the old vertices this will be a tree. These graphs have been shown to have a power law degree distribution, the same as obsereved in some large real-world networks. We are interested in the width of the tree and show that it is W(n) ~ n/sqrt(Pi log n) and this also holds for a slight generalization of the model with another constant. Then we see how this theoretical result can be applied to directory trees.AMS classification: 05C80, 60C05 Keywords: random graphs, scale-free disribution, width of trees
3-wise exactly 1-intersecting families of sets (download pdf) Abstract Let f(l,t,n) be the maximal size of a family F a subset of 2[n] such that any l at least 2 sets of F have an exactly t-element intersection. If l is at least 3, it trivially comes from a paper of Füredi that the optimal families are trivially intersecting (there is a t-element core contained by all the members of the family). Hence it is easy to determine f(l,t,n)=[(t/2)(n-1)]+1. Let g(l,t,n) be the maximal size of an l-wise exaclty t-intersecting family that is not trivially t-intersecting. We give upper and lower bounds which only meet in the following case: g(3,1,n)=n2/3(1+o(1)).AMS classification: 05D05; 05B25 Keywords: extremal problems for families of finite sets; finite projective geometries
Multiply intersecting families of sets (with Zoltán Füredi) (download pdf) Abstract Let [n] denote the set {1,2,...,n}, 2[n] the collection of all subsets of [n] and F a subset of 2[n] be a family. The maximum of |F| is studied if any r subsets have an at least s-element intersection and there are no l subsets containing t+1 common elements. We show that |F| is at most \sum_{i=0}^{t-s} \binom{n-s}{i}+ \frac{t+\ell-s}{t+2-s}\binom{n-s}{t+1-s}+\min\{s-1, l-2\} and this bound is asymptotically the best possible as n goes to infinty and t is at least 2s is at least 2, and r,l that are at least 2 are fixed.AMS classification: 05D05; 05B25 Keywords: extremal problems for families of finite sets
Intersecting families of sets, no l containing two common elements (download pdf) Abstract Let H denote the set {f1,f2,...,fn}, 2[n] the collection of all subsets of H and F a subset of 2[n] be a family. The maximum of |F| is studied if any k subsets have a non-empty intersection and the intersection of any l distinct subsets (1<=k<l) is empty. This problem is reduced to a covering problem.AMS classification: 05D05; 05B25 Keywords: extremal problems for families of finite sets; finite projective geometries |
||
| more >>> | ||