The Political Economy of New Media: Challenges and Opportunities Across Democracies and Autocracies Free

Abstract

There are several open research questions in the political economy of new media. What role do they play in the spread of “fake news”? What are their effects on voters’ beliefs and behavior, and on the overall level of ideological polarization? What is their impact on citizens’ trust in experts and democratic institutions? Furthermore, we still need to fully understand to what extent new media and digital platforms represent tools to overthrow autocracies or, rather, whether they provide new propaganda and surveillance instruments for such regimes.

1 Introduction

News media are key institutions of accountability. They allow citizens to be informed on politicians’ behavior and on public policy outcomes. The importance of media as institutions is exemplified by the words of Thomas Jefferson:

The recent literature on the political economy of media provides evidence in this regard. Traditional news media such as radio and newspapers tend to be associated with a higher level of citizens’ information (Snyder and Strömberg, 2010), a higher degree of electoral participation (Drago et al. 2014; Gentzkow et al. 2011; Snyder and Strömberg 2010; Strömberg 2004b) and higher and more efficient public goods’ provision (Drago et al. 2014; Snyder and Strömberg 2010; Strömberg 2004b). At the same time, the rise of less informative media (e.g., television) may crowd out existing—and more informative—types of news media and may end up generating negative effects in terms of political participation (Gentzkow 2006; George and Waldfogel 2008).¹ Importantly, in weakly institutionalized environments, the role of media as an instrument of political accountability might be jeopardized as they are subject to capture by incumbent politicians (Besley and Prat 2006). Hence, they may end up self-censoring information unfavorable to the government (Di Tella and Franceschelli 2011; McMillan and Zoido 2004), or even used as instruments for autocratic propaganda (Adena et al. 2015). Yet, existing evidence has also shown that market discipline may restrain media capture (Gentzkow et al. 2015; Petrova 2011; Qin et al. 2018) and that independent/foreign media may effectively challenge goverments in weakly instituzionalized/authoritarian regimes (Enikolopov et al., 2011; Gagliarducci et al. 2018.

The creation of the Internet and the consequent dramatic expansion in the number and types of news sources has revived interest in the role of news media while posing novel research questions. Overall, different views on the Internet have emerged in the last 20 years. In his 1996 seminal book, Being Digital, Nicholas Negroponte put forward an optimistic vision of the upcoming digital age, suggesting that “we are bound to find new hope and dignity in places where very little existed before” (Negroponte 1996, 231). In later years, some scholars have instead argued the possible negative effects of the Internet on democracy, highlighting the risks of polarization or “cyberbalkanization” (Putnam 2001; Sunstein 2009). Others have proposed a more balanced view. In particular, Hindman (2009) suggests that, overall, “the Internet seems both good news and bad news for the political voice of the average citizen” (Hindman 2009, 142).² While the political economy literature has recently started to provide theoretical and empirical studies aimed at shedding light on this debate, there are still many open research questions regarding the rise and consequences of “new media” (e.g., websites, blogs, social media).

The article is structured as follows. Section 2 provides an overview of the existing literature on the effects of Internet and new media on political participation and political polarization. Section 3 presents some open research questions in the political economy of new media. Specifically, Section 3.1 discusses some key issues related to the role of new media in fostering the diffusion of “fake news,” their potential link with the declining trust in experts and institutions, and the effects of algorithmic gatekeepers. Section 3.2 looks at the role of new media in weakly institutionalized environments. Section 4 concludes.

2 The Road Behind

As was the case with the rise of other types of new media, such as radio (Strömberg 2004b), newspapers (Gentzkow et al., 2011), and television (Gentzkow 2006), one of the first questions scholars have investigated concerns the overall impact of the Internet on some of the founding pillars of democracies: citizens’ electoral participation (Campante et al. 2017; Czernich 2012; Falck et al. 2014; Gavazza et al. 2015), public goods allocation (Gavazza et al. 2015), and party organizations (Campante et al. 2017). The fundamental question posed by the Internet in terms of political engagement is similar to the one posed by pre-existent media (such as television): Does it foster political participation by making citizens more informed and more engaged in political activities? Or does it “distract” them by offering new and vast entertainment opportunities? The literature provides evidence of a negative effect on electoral participation, at least in the short run (Campante et al. 2017; Falck et al. 2014; Gavazza et al. 2015), with heterogeneity across groups of voters and, thus, with important redistributive consequences (Gavazza et al. 2015). Nevertheless, Campante et al. (2017) also show that this initial negative effect may be reversed in the long run once party organizations start exploiting the presence of new digital platforms to remobilize voters (as was the case for the populist “Five Star Movement” party in Italy’s 2013 elections). Importantly, Campante et al. (2017) also show that the initial negative effect of the Internet on electoral participation is not necessarily the result of the entertainment effect prevailing over the engagement one. Rather, it may reflect the presence of new forms of political engagement online (blogs, social media platforms, etc.) that may substitute for offline political participation.

Scholars have also looked at whether the diversity of news sources provided by the Internet and the digital platforms that emerged from it (e.g., social media) have fostered polarization and ideological segregation. The literature is mixed. Gentzkow and Shapiro (2011) show that a large share of Internet traffic is captured by moderate news websites. Moreover, the extent of online segregation in news consumption is limited by the fact that individuals tend to visit multiple news outlets. In particular, individuals who visit ideologically extreme websites are also the ones more likely to visit multiple news sources. For example, the average reader of extreme conservative websites is more likely than a typical online news reader to visit the New York Times website. On the other hand, Bar-Gill and Gandal (2017), Bakshy et al. (2015), Bessi et al. (2015), Flaxman et al. (2016), and Halberstam and Knight (2016) find evidence of ideological segregation in online platforms, although with different degrees of polarization depending on the platform and the population under analysis.

While these articles look at the extent to which individuals are exposed to different viewpoints or are segregated into “echo-chambers,” the crucial underlying issue is the extent to which exposure translates into polarized beliefs. For example, as pointed out by Gentzkow and Shapiro (2011), even when individuals are exposed to the same information, they may interpret such information differently where they have different prior beliefs. For example, Allcott and Gentzkow (2017) show that both Democrats and Republicans are more likely to believe ideologically aligned articles than nonaligned articles. Therefore, minor differences in news exposure may be magnified by the belief updating process and may end up generating significant differences in posterior beliefs and hence in actual polarization. Indeed, exploiting a randomized deactivation of Facebook accounts, Allcott et al. (2019) show that a lower exposure to social media is associated with a decrease in political polarization. At the same time, Boxell et al. (2017) show that the overall increase in polarization in the United States between 1996 and 2012 is larger in demographic groups that are least likely to use the Internet and social media (i.e., older individuals), suggesting that the Internet and its digital platforms may not necessarily be the main drivers of the increase in political polarization observed in the recent past (Gentzkow 2016).

The extent to which the Internet and different digital platforms contribute to political polarization remains an issue that is worth exploring further, in particular outside the United States where the evidence is scarcer. The next section illustrates other open research questions concerning online information sources.

3 The Road Ahead: New Technologies, New Issues

There are three key features that make new media different from traditional media: (1) cost structure and incentives (which contribute to the sizable difference in “fake news” present in the two types of media); (2) ranking algorithms; and (3) media capture. All of them pose novel research questions.

3.1 The Rise and Consequences of “Fake News”

A first crucial difference between new and traditional media lies in the intentional bias prevalent in one compared to the other. Traditional media have a weaker incentive to bias their contents by reporting “alternative facts” than new media. Two factors jointly contribute to this difference: the reputational incentives and the cost structure present in the two types of media.

Reporting false information typically results in very high reputation costs for traditional media. Accordingly, it is rare to witness such an event.³ Rather than reporting false information, traditional media may bias its news reports by selectively omitting information (Anderson and McLaren 2012), giving more prominence to a particular issue (Larcinese et al. 2011), framing news by using specific language (Gentzkow and Shapiro 2010), or by hiring journalists with a given political ideology who acquire information in a biased way (Sobbrio 2014a). Instead, new media may be anonymous and thus, even if exposed as reporting false information, do not face long-run reputation costs (e.g., they may simply open another website with a different name).

At the same time, the cost structure of traditional media is typically characterized by increasing returns to scale, making them “natural oligopolies” (Belleflamme and Peitz 2015). Accordingly, as shown by Strömberg (2004a), they have an incentive to provide news that is valuable to large groups. Hence, reporting “alternative facts” is unlikely to be economically profitable for traditional media since such content would be of interest only to a minority of individuals. In contrast, since fixed costs are close to zero for many online media (such as personal websites, blogs, Facebook pages, Twitter accounts), their cost structure is not characterized by (significant) increasing returns to scale. Hence, apart from a few big players catering to large audiences (e.g., Wikipedia), new media operate in a market characterized by horizontal product differentiation and almost no entry costs. As a consequence, new media typically cater to niche audiences. In turn, the characteristics of such niche markets limit the typical discipline effect—in terms of accurate news reporting—induced by competition. In short, the size of the audience of each of these outlets provide weak incentives for potential competitors to expose falsehood.⁴

All in all, the cost structure and the weak (reputational and competitive) incentives to report correct information characterizing new media have contributed to the presence of a sizable number of online media, each catering to a profitable niche (i.e., providing “alternative facts” to like-minded individuals).⁵

Allcott and Gentzkow (2017) provide the first study on this issue by measuring the extent of fake news shared in the social media during the 2016 U.S. presidential elections. The authors document the presence of 41 pro-Clinton (or anti-Trump) and 115 pro-Trump (or anti-Clinton) fake news articles that were shared on Facebook 7.6 and 30.3 million times, respectively. Despite this massive diffusion of fake news, Allcott and Gentzkow (2017) show that, once controlling for false-recalls by using a set of placebo fake news articles as a benchmark, an average individual saw, remembered, and believed only around 1.14 fake news articles. This implies that the effect of fake news on actual voting choices might be limited. Barrera et al. (2017), Nyhan et al. (2017), and Nyhan and Reifler (2016) look at a complementary research question by analyzing whether and to what extent “debunking” (i.e., exposure to fact-checking) is able to counteract the effects of fake news. Interestingly, these studies show that, while exposure to fact-checking may help push voters’ beliefs closer to the truth, it does not substantially change their voting choice. Moreover, fact-checking may even backfire in some instances by raising the salience of the issue at stake (Barrera et al., 2017). Importantly, Alcott et al. (2019 document that while the fake news exposure and interaction has been steadily increasing in Twitter in recent years, this trend has reversed in Facebook at the end of 2016. This suggests that platforms may effectively limit the diffusion of misinformation. Assessing the impact of fake news and of fact-checking on voters’ beliefs and behavior remains an issue that is worth exploring further.⁶

3.2.1 Trust in Experts and Institutions

Beyond their direct and immediate effects on political outcomes, are digital platforms affecting citizens’ trust in experts and institutions? During the campaign for the “Brexit” referendum in the United Kingdom, Michael Gove dismissed the advice of many experts and institutions to remain in the EU by stating that, “People in this country have had enough of experts.” Many observers have seen this statement as marking an era of distrust in experts.⁷ Most importantly, the Internet has been accused of being a key driver of a general negative trend in trust in experts (Nichols, 2017).⁸

Aggregate data provide a composite picture of the supposed connection between the Internet and institutional trust. Figures 1 and 2 show, for a selected sample of countries, the evolution over time in the percentage of people who have a high level of confidence in two key democratic institutions: the parliament and the press.

It is difficult to spot a general trend in institutional trust from these graphs. It is also worth noting that such trends do seem to differ greatly when comparing advanced democracies with countries characterized by weaker institutions (e.g., Russia and Turkey). Most importantly, even in the case of the United States where individuals seem to have been losing confidence in both the press and the parliamentary institutions over time, such trends seem to precede the arrival of the Internet and of the Web 2.0 revolution (with the associated rise in social media platforms). Yet, more recent data provide suggestive evidence of an increasing gap in institutional trust between the informed public and the general population (Edelman, 2016). Moreover, the Pew Research Center (2016) reports that, in the United States, younger generations are both more likely to get their news online and less likely to trust the national media with respect to older ones. These conflicting data call for research exploring whether there is any causal link between the rise of new media and the attitude of citizens toward experts and, more generally, democratic institutions.

3.1.2 Algorithmic Gatekeepers

A second crucial difference between new and traditional media lies in the unintentional bias that might be present in online information. Digital platforms such as search engines and social media rely on automated algorithms to establish the ranking of information provided by their users (e.g., the ranking of websites obtained for a given search query in the case of search engines and the ranking of posts/tweets in the case of social media such as Facebook or Twitter). The presence of such automated algorithms is a direct consequence of the amount of data fostered by the presence of digital platforms themselves. In 2010, Eric Schmidt, the CEO of Google, stated that every two days “we now create as much information as we did from the dawn of civilization up until 2003.” In this world of “data deluge,” the ranking algorithms used by search engines and social media represent a necessary tool to “pluck the diamond from the waste” (The Economist, “The Data Deluge,” February 25, 2010). At the same time, these algorithms make search engines and social media very different from any offline counterpart (traditional news media or offline social networks). As pointed out by Rieder and Sire (2013) with regard to search engines, “While humans are certainly responsible for editorial decisions, these [search engines] decisions are mainly expressed in the form of software which thoroughly transforms the ways in which procedures are imagined, discussed, implemented, and managed. In a sense, we are closer to statistics than to journalism when it comes to bias in Web search” (Rieder and Sire 2013, 2). That is, in the case of offline media, the gate-keeping decisions on what information to collect and disclose is the result of case-by-case decisions, while search engines and social media are “algorithmic gate-keepers” (Granka, 2010; Tufekci, 2015). Therefore, whatever unintentional bias might originate from the presence of ranking algorithms, it would be intrinsically different from the one present in traditional media. Accordingly, studying the effects of information platforms relying on ranking algorithms on the accuracy of individuals’ beliefs requires a rather different approach compared to that used by theoretical models of media bias (e.g., Gentzkow and Shapiro, 2006; Mullainathan and Shleifer, 2005; Sobbrio, 2014a; Strömberg, 2004a). Germano and Sobbrio (2018) provide a first attempt to address this issue by proposing a theoretical model capturing the interaction of individual search behavior with some of the key components of automated ranking algorithms, namely popularity and personalized rankings. The insights of the model show that popularity-driven rankings have an overall positive effect on the probability of individuals reading a website reporting correct information. Popularity-driven rankings enhance the aggregation of the private information dispersed among individuals. At the same time, ranking algorithms can also contribute to the diffusion of misinformation. In particular, there is an advantage of the fewer effect at play. While it is always better to have a majority of websites reporting correct information, a larger majority might decrease the probability of individuals reading correct information. A smaller minority of websites reporting incorrect information (e.g., conspiracy websites) might each get a larger audience (by like-minded individuals). Remarkably, this static effect (clearly present also in traditional media), is amplified over time by the dynamic induced by popularity-driven rankings. Accordingly, the advantage of the fewer provides a further rationale to explain the large diffusion of fake news present in the current information environment, dominated by algorithmic gatekeepers, such as search engines and social media. Personalization is another important component embedded in the ranking algorithms used by search engines and social media. Different individuals typically observe a different ranking of websites for the same search query or a different order of news posts by their Facebooks friends. Germano and Sobbrio (2018) point out that, while personalized rankings might be useful on private value issues (e.g., attributes of a commercial product), they may end up decreasing the probability of individuals reading correct information when it comes to common value issues (e.g., side effects of a vaccine).

Further studying—theoretically and empirically—the effects of ranking algorithms is an important area of future research as the relevance of information platforms using algorithmic rankings is increasing. The Pew Research Center (2016) reports that 38% of U.S. adults often get their news online (social media, websites, apps). In the 18–29 age segment, this percentage goes up to 50%, becoming the most frequently used news platform.

3.2 New Media Capture in Weakly Institutionalized Environments

There is a third crucial difference between traditional and new media. In weakly institutionalized environments (e.g., authoritarian regimes), traditional media such as newspapers, radio, and television might be easily captured by powerful political elites. New media are less subject to capture, and hence potentially represent independent sources of information and, ultimately, of political accountability. For example, Enikolopov et al. (2018) show that blog posts on corruption episodes connected to Russian state-controlled companies have a negative impact on market returns also leading to higher management turnover. Accordingly, social media may improve political accountability beyond its effect on political information and elections.⁹

More generally, to understand whether new media may pose a serious threat to authoritarian regimes (and thus favor the transition to democracy) two main questions must be addressed: (1) Are new media playing any role in mobilizing people against autocratic regimes? And if so, do they foster participation in mass protests by informing and motivating individuals or by coordinating their actions? (2) How do autocratic governments respond to the presence (and potential threat) of online news and digital platforms? Accordingly, do individuals living in authoritarian countries adapt their online behavior by taking into account the instrumental use and manipulation of digital platforms by their governments?

Enikolopov et al. (2016) provide a positive answer to the first question by showing that social media penetration was associated with a higher level of participation in political protests in Russia in 2011.¹⁰ Moreover, their evidence suggests that social media affected protest participation mainly by improving coordination rather than by providing an alternative to traditional media. Importantly, the mechanism linking social media participation and mass protest might vary depending on social norms and on how the government responds (Enikolopov et al., 2016). As pointed out by Qin et al. (2017) the interaction of an authoritarian regime with social media goes beyond simple censorship. From the perspective of a nondemocratic government, social media is not only “(1) unattractive as a potential outlet for organized social protest but is also (2) useful as a method of monitoring local officials and (3) gauging public sentiments, as well as (4) a method for disseminating propaganda” (Qin et al. 2017, 119). Indeed, existing research provides evidence of such sophisticated exploitation of social media by authoritarian governments. By analyzing the contents removed from Chinese social media, King et al. (2013, 2014) provide evidence showing that the Chinese government censors social media contents related to real-world events with a collective action potential. At the same time, it seems to tolerate (i.e., not censor) criticism of the Chinese government as long as such criticism is not related to events with a collective action potential. In a related study, Qin et al. (2017) focus on the overall social media contents—rather than on the subset that is censored—and provide an even more composite picture of the interaction between the Chinese authoritarian regime and social media. In particular, they show that Chinese social media contain very few posts criticizing top Chinese national leaders or describing large-scale conflicts (due to censorship or self-censorship). Yet there is a sizable number of sensitive posts about local officials. Accordingly, the Chinese government seems to have strategically limited its censorship since the information provided by social media users may help it monitor local officials. Overall, Qin et al. (2017) argue that social media in China might have benefited the central government at the expense of local ones.

At the same time, from the political demand side (i.e., from the citizens perspective), “any perceived benefits of social media need to be evaluated in a context of (5) possible pervasive policing, punishment, and (6) censorship of such media” (Qin et al. 2017, 119). Survey evidence seem also supportive of this sophistication on the side of the citizens. On average, less than 50% of nationals surveyed in 2018 in several middle-eastern countries reported that it is safe to say on the Internet whatever one thinks about politics (53% in Saudi Arabia, 42% in Lebanon, 32% in Tunisia, 44% in the UAE).¹¹ Moreover, many individuals living in autocratic regimes in the middle-east use a VPN (54% in Saudi Arabia and 39% in Qatar, around 10% in other countries). Notably, younger and more educated nationals are more likely to use a VPN, as well as Internet users who are worried about government surveillance online (29% vs. 19% among the unconcerned ones).

4 Conclusions

The political economy of new media raises several open research questions. What role do they play in the spread of “fake news”? What are their consequences on voters’ beliefs, behavior, and on the overall level of ideological polarization? What is their impact on citizens’ trust in experts and democratic institutions? To what extent do new media and digital platforms represent tools to overthrow autocracies, or instead provide them with new propaganda and surveillance instruments? Do individuals living in authoritarian countries adapt their online behavior by taking into account the instrumental use and manipulation of digital platforms by their governments?

A final word on methodology. The Internet has not just introduced a vast array of new types of news sources and platforms. It also provides an impressive archive of digital text (high-dimensional data). As pointed out by Gentzkow et al. (2017), the presence of these digital archives opens up a new world of possibilities for social scientists interested in analyzing how institutions shape the incentives and behavior of individuals within an organization.¹²

Notes

References

Part of this article was published as Chapter 25 in Claude Menard, Mary Shirley, (eds.). A Research Agenda for New Institutional Economics, Edward Elgar Publishing, Ltd, 2018. It is here reproduced with permission of Edward Elgar Publishing Limited through PLSclear (Licence Date: 18/04/2019; PLSclear Ref No: 13323). I am very grateful to Claude Menard, Francesco Privitera, and Mary Shirley for insightful comments. The usual disclaimers apply.