Public Media Do Serve the State: a Field Experiment

Abstract : We have conducted a randomized field experiment in collaboration with the nation's major public service broadcasting in which the capacity for viewing its programs has been randomly increased. We find the positive effect on the viewing time of the programs. Moreover, we find that the treated group has more positively evaluated the government's foreign policies than the control group after the treatment. To further study the mechanism, we use unsupervised machine learning to measure semantic similarity between the contents of TV programs and the official statements made by domestic and foreign governments. We find that programs in the public media are closer to the domestic government's statements than to the foreign government's statements. Programs in the private media, in contrast, show the opposite pattern. We find suggesting evidence that the positive evaluation on foreign policies is likely to be explained by individuals' being exposed to slanted information in favor of the domestic government. In contrast to previous studies showing media slant in the right-left political spectrum, this study adds a new empirical evidence that media slant in the domestic-foreign spectrum persuades viewers in support of the domestic government.

---

To Russia with Love? the Impact of Sanctions on Elections

Abstract : Do economic sanctions weaken the support for incumbent governments? To answer this question, we focus on the sanctions imposed on Russia after 2014 and estimate their effect on voting behavior in both presidential and parliamentary elections. For identification, we use cross-regional variation in (pre-determined) trade exposure to sanctioning and non-sanctioning countries and before-after voting data at both regional and district level. The sanctions caused an increase in support for the incumbent. This result is robust to alternative measures of sanction exposure, including a measure of trade loss, i.e., the difference between observed trade flows and counter-factual trade flows computed via a full-general-equilibrium gravity model. Absence of pre-trends, as well as several placebo estimations, supports the validity of the identification assumption. We then explore several potential mechanisms, including propaganda, electoral fraud as well as standard demand-supply effects. Overall, while it is hard to evaluate all the potential motives that sanctioning countries might have had, our results suggest that economic sanctions are not an effective tool for reaching one of their primary goals and can actually backfire.

---

Does Chinese Social Media Provide Less Biased Information to the Market Than Traditional Media?

Abstract : This paper examines whether privately owned social media in China supplies less positively biased information to the market than traditional state-run media. Using a comprehensive sample from 2009 to 2016 of corporate news of newspapers and the opinions shared by investors in Guba, an online stock forum, we find that the tone of the traditional and social media of the same firm on the same day is positively correlated. However, this positive correlation is significantly reduced when the tone of the newspapers is positive. Consistent with the conjecture that Guba serves as a check against the bias of state-owned media, our result suggests that the tone of Guba’s postings is less optimistic than that of the newspapers when the latter are more likely to be positively biased. We find that political factors such as state ownership of the firms being covered and the political sensitivity of the time periods during which the articles are posted shape Guba’s monitoring role against the newspapers’ bias. Finally, the positive stock return response to the tone of the traditional media’s articles is significantly attenuated when it deviates positively from that of the social media, but we do not find any significant change in stock return response to the tone of social media when it deviates from that of the traditional media.

---