In contrast, when the intensity of the contest among sites is beyond a certain threshold, competition tends to rapidly increase the fragmentation of published news: as the number of competing publishers increases, more and more a priori unlikely topics are reported resulting in a large diversity of published topics . This result is reminiscent of the emergence of “funny lists” and “heartwarming videos” in the news mentioned by the Financial Times . Our analysis extends to pure- and mixed-strategies and distinguishes between cases with small or large number of competing publishers.Next, in a model with firm asymmetries, we find that when some firms have better technology to forecast the popularity of topics, then, surprisingly, the overall diversity of news published by the remaining firms declines as these firms tend to take refuge in publishing ‘safer’ topics. When a subset of firms have extra revenue from a published ‘hit’ from loyal users then these ‘branded’ publishers tend to be conservative in their choice of topics as their loyal customer base represents ‘insurance’ against the contest. In contrast, the diversity of news published by unbranded outlets increases as unbranded publishers tend to avoid branded ones by putting more weight on a priori unlikely stories. These results are consistent with anecdotal evidence in the news industry where traditional news outlets are more conservative in their reporting whereas new entrants do not shy away from controversial stories. The findings are also conform to the broadly observed increase of diversity in the public agenda by communication theorists.In a final analysis,vertical planting tower we consider endogenous success probabilities. It is widely accepted that the media often ‘makes the news’ in the sense that a topic may become relevant simply because it got published. Interestingly, such a dynamic has an ambiguous effect on the diversity of published topics. If the contest is very strong then it results in a concentrated set of a priori likely topics. When the contest is moderate then the diversity of topics may be higher depending on the number of competing outlets.

The article is organized as follows. In the next section, we summarize the relevant literature. This is followed by the description of the basic model and its analysis where we first present a variety of results concerning symmetric competitors. Next, we extend the model to explore the impact of asymmetries across firms. Our last extension considers the case of endogenous success probabilities. The article ends with a discussion of the results and their applicability to other contexts. To facilitate reading, all proofs are relegated to the Appendix. The topic of this article is generally related to the literature on agenda setting for an excellent recent review that studies the role of media in focusing the public on certain topics instead of others. It is broadly believed that agenda setting has a greater influence on the public than published opinion whose explicit purpose is to influence the readers’ perspective. As the famous saying by Bernard Cohen goes: “The media may not be successful in telling people what to think but they are stunningly successful in telling their audiences what to think about”. The literature examines the mechanisms that lead to the emergence of topics and the diversity of topics across media outlets. In particular, McCombs and Zhu show that the general diversity of topics as well as their volatility has been steadily increasing over time. The general focus of our article is similar: we show that the nature of competition is an important mechanism affecting the diversity of public agenda. Agenda setting is also addressed in the literature studying the political economy of mass media for an excellent review.The standard theory states that media coverage is higher for topics that are of interest for larger groups, with larger advertising potential, and when the topic is journalistically more “newsworthy” and cheaper to distribute. Although there is little empirical evidence to support , the other hypotheses are generally supported and Snyder and Stromberg , among others. Hypotheses is particularly interesting from our standpoint. Eisensee and Stromberg show that the demand for topics can vary substantially over time. For example, sensational topics of general interest may crowd out other ‘important’ topics that would be covered otherwise. This supports the general notion that media needs to constantly forecast the likely success of topics and select among them accordingly.

Our main interest is different from this literature’s as we primarily focus on media competition as opposed to what causes variations in demand. Taking the demand as given, our goal is to understand how the competitive forces between media firms influences the selection and diversity of topics, which then has a major impact on the public agenda. As such, the article also relates to the literature on media competition where strategic behavior influences product variety. Early thoretical work by Steiner and Beebe on the “genre” selection of broadcasters explains cases of insufficient variety provision in an oligopoly. Interestingly, they show that although certain situations lead to the duplication of popular genres , other scenarios may lead to a “lowest common denominator” outcome where no consumer’s first choice of genre is ever served. A good discussion of these models and their extensions can be found in Anderson and Waldfogel . Our work is different from this literature in two important ways: we do not have consumer heterogeneity and, we do not rely on barriers to entry to explain limited variety. In fact, we study variety precisely when these factors’ importance is greatly diminished. On the empirical side, research on competition primarily focuses on how media concentration affects the diversity of news both in terms of the issues discussed in the media as well as the diversity of opinion on a particular issue. For example, George and Oberholzer-Gee show that in local broadcast news, “issue diversity” grows with increased competition even though political diversity tends to decrease. Franceschelli studies the impact of the Internet on news coverage, in particular the recent decrease in the lead-time for catching up with missed breaking news. He argues that missing the breaking news has less impact, as the news outlet can catch up with rivals in less time. This might lead to a free-riding effect among media outlets, where there is less incentive to identify the breaking news. Both of these articles have consistent empirical findings with our results/assumptions. In terms of the analytical model, we rely on the literature studying competitive contests among forecasters. For example, Ottaviani and Sorensen use a similar framework to model competition among financial analysts. Our model is different in that we explore in more detail the structure of the state space, we generalize the contest model by considering all possible prize-sharing structures and extend it in a variety of ways, most notably by analyzing asymmetries across players. This article studies competition among news providers who compete in a contest to publish on a relatively small number of topics from a large set when these topics’ prior success probabilities differ and when their success may be correlated.

We show that the competitive dynamic generated by a strong enough contest causes firms to publish ‘isolated’ topics with relatively small prior success probabilities. The stronger the competition , the more diverse the published news is likely to be. Applied to the context of today’s news markets characterized by increased competition between firms,vertical hydroponic farming new entrants and reduced customer loyalty, we expect a more diverse set of topics covered by the news industry. Although direct evidence is scarce, there seems to be strong empirical support for the general notion that the public agenda has become more diverse over time while also exhibiting more volatility McCombs . This general finding is consistent with our results. Although diversity of news may generally be considered a good thing, agenda setting, i.e. focusing the public on a few, worthy topics maybe impaired by increased competition. In a next step, we explore differences across news providers and find that branded outlets with a loyal customer base are likely to be conservative with their choice of reporting in the sense that they report news that is a priori agreed to be important. Facing new competitors with better forecasting ability also makes traditional media more conservative. In sum, if the public considers traditional media and not the new entrants as the key players in agenda setting, then increased competition may actually make for a more concentrated set of a priori important topics on the agenda. It is not clear however, that traditional news outlets can maintain forever their privileged status in this regard. Some new entrants have managed to build a relatively strong ‘voice’ over the last few years. We also explore what happens when the success of news is endogenous, i.e. if the act of publishing a topic ends up increasing its likely success. Interestingly, we find that an excessively strong contest tends to concentrate reporting on topics with the highest a priori success probabilities. We also find that the number of competitors has a somewhat ambiguous effect on the outcome. If there are too few or too many competing firms then, again agenda setting tends to remain conservative in the sense of focusing on the a priori likely topics. These results also resonate to anecdotal evidence concerning today’s industry dynamics. Our analysis did not consider social welfare. This is hard to do as it is not clear how one measures consumer surplus in the context of news. Indeed, the model is silent as to what is consumers’ utility when it comes to the diversity of news. Although policy makers generally consider the diversity of news as a desirable outcome, a view that often guides policy and regulatory choices, it is not entirely clear that, beyond a certain threshold, more diversity is always good for consumers. As mentioned in the introduction, the media does have an agenda setting role and it is hard to argue that every topic equally represented in the news is a useful agenda to coordinate collective social decisions . Nevertheless, our goal was to identify the competitive forces that may play a role in determining the diversity of news in today’s environment increasingly dominated by social media. Our analysis indicates that these forces do not necessarily have a straightforward impact on diversity. The generalized contest model presented has implications for other economic situations that may be well-described by contests. In this sense, our most relevant results are those that describe the outcome as a function of the reward-sharing patterns across winners. Indeed, we characterize all such patterns with a simple parameter, r, and show that depending on r there are only three qualitatively different outcomes leading to vastly different firm behaviors. Different r’s may characterise different contexts. For our case a finite, albeit varying r seemed appropriate and r = ∞ is less likely. In the case of a contest describing R&D competition, r = ∞ is quite plausible. Conversely, the case of r = 0 may well apply to contests among forecasters whose reward might be linked more closely to actually forecasting the event and less to how many other forecasters managed to do so. Our analysis of the case with a small number of firms may also be useful in particular situations ; we show that this case is tractable and shares many characteristics with the case involving many players. An important insight from our analysis is that contest models need to be carefully adjusted to the particular situations studied. Our framework can be extended in a number of directions. So far, we assumed a static model, one where repeated contests are entirely independent. One could also study the industry with repeated contests between media firms where an assumption is made on how success in a period may influence the reward or the predictive power of a medium in the next period. A similar, setup is studied with a Markovian model by Ofek and Sarvary to describeindustry evolution for the hi-tech product categories. Finally, our article generated a number of hypotheses that would be interesting to verify in future empirical research.