Foreign media slant, foreign investors, and informativeness of earnings

A Pew Research Center report shows that, in 2008, 796 foreign media outlets (from 113 countries and territories) were represented by at least one correspondent in Washington, D.C., an increase of 57% from the 507 media outlets (from 59 countries and territories) represented by at least one Washington correspondent in 1994 (Pew Research Center 2009).

Many corporate executives believe that earnings are the most important financial information they provide to the capital market (Graham et al. 2005). Studies show that the media play an important role in stock price reactions to corporate news announcements (e.g., Bonsall et al. 2020; Drake et al. 2017; Griffin et al. 2011; Huang et al. 2018). Research suggests that managers are very concerned about the market response to earnings announcements released through the media (e.g., Huang et al. 2018; Michaely et al. 2016).

For example, Gurun and Butler (2012) and Reuter and Zitzewitz (2006) find that media outlets that have advertising relationships with firms, compared with their counterparts without such relationships, use significantly fewer negative words in their articles. Piotroski et al. (2017) show that government-controlled newspapers with political aims use a significantly more positive tone when covering politically connected (versus apolitical) firms.

For example, Engelberg and Parsons (2011) find that local media coverage of the earnings announcements of S&P 500 firms predicts local trading, and Miller and Shanthikumar (2015) find that local media coverage elicits greater reactions from local traders.

In a similar vein, Mullainathan and Shleifer (2005) suggest that while news media can segment the market and slant toward extreme positions when readers have divergent beliefs, a reader with access to all news sources can obtain an unbiased perspective.

For example, Tykvova and Schertler (2014) document that while foreign investors can overcome the obstacles associated with geographical distance from local firms through syndication with local investors, the involvement of local investors does not enable foreign investors to overcome the obstacles associated with institutional distance. Research (e.g., Bottazzi et al. 2009) also shows that investors have difficulties transferring and enforcing the corporate governance structure and contract design that they use in their home countries to institutionally distant foreign countries.

Supporting this view, research suggests that investors face corporate governance disadvantages because of high information asymmetry when they seek out more profitable investment opportunities beyond their national borders (Chen et al. 2019; Coval and Moskowitz 1999; Kang and Stulz 1997).

Similar to this argument, various country-level institutional differences are shown to increase foreign entities’ costs of information acquisition and production in the context of institutional investors and financial analysts (e.g., Bae et al. 2008; Baik et al. 2013; Giannetti and Yafeh 2012; Jung et al. 2019; Mian 2006). For example, Jung et al. (2019) find that forecast revisions by analysts with surnames more favored by Americans elicit stronger market reactions and that surname favorability mitigates the underreaction to analyst forecast revisions.

Similarly, the literature shows that sharing an institutional logic complements international business relationships and encourages networking between parties across the globe (Abdi and Aulakh 2012).

To ensure that our results are robust to the choice of test windows, we repeat our article-level tests using two alternative windows (i.e., four and six days following the earnings announcement). We also repeat our earnings announcement-level tests in Sections 5.1‒5.3 using these two alternative measurement windows. Our findings endure in all of these tests.

Following previous studies (e.g., Bushman et al. 2017; Chapman et al. 2019), the tone of a news article is measured by RavenPack’s composite sentiment score, which represents the news sentiment of a given story. The score combines various textual analysis techniques to assign each article a sentiment score ranging from 0 to 100, where a score above 50 indicates a positive tone, a score of 50 indicates a neutral tone, and a score below 50 indicates a negative tone. Our inference endures when we exclude neutral articles (i.e., CSS = 50) from our sample.

We obtain the information on export volumes, bilateral trust, and countries’ primary religions from the World Trade Organization, World Values Survey Database, and Stulz and Williamson (2003), respectively.

However, we cannot perform this test when examining the likelihood of bad news coverage, as there is no within-firm-day variation in the reporting of good or bad news.

In this analysis, the benchmark cases consist of articles written by foreign media outlets with high economic proximity, political proximity, and cultural proximity to the United States, respectively, when FMC_LowProximity_Econ, FMC_LowProximity_Poli, and FMC_LowProximity_Cul are used as the test variable.

We collect the data on geographical distance from Mayer and Zignago’s (2011) GeoDist database. The data and a full description are given at http://​www.​cepii.​fr/​anglaisgraph/​bdd/​distances.​htm.

We define a media outlet’s home country based on its headquarters, which is consistent with the literature that examines foreign and domestic information intermediaries, albeit in a different setting (e.g., Bae et al. 2008 and Lai and Teo 2008 in the context of financial analysts). However, a change in the location of the ownership of a media outlet could lead to a change in its geographical or institutional distance from the United States, which may bias our results. To examine this possibility, we search the ownership location of the 98 foreign media outlets provided in Appendix Table 10 as a sample and find that only two media outlets experienced a cross-national change in ownership during our sample period: the London-headquartered Financial Times, whose ownership changed from a British company, Pearson plc, to a Japanese holding company, Nikkei Inc., in 2015; and the Netherlands-headquartered and -owned NRC Handelsblad, which was acquired by the Belgian company Mediahuis in 2015. While the sample sizes for both media outlets are too small for us to perform any meaningful tests, these two rare cases do not show a systematic change of ownership from distant to proximate institutions or vice versa: the Financial Times’s ownership was changed from a country with low institutional distance to a country with high institutional distance from the United States, and NRC Handelsblad’s ownership did not experience a change in institutional distance from the United States In both cases, the headquarters of the media outlets remained the same after the acquisitions. (We also do not find any change in headquarters of any of the 98 media outlets listed in Appendix Table 10 during our sample period.) Given the rarity of the cross-national media acquisition events and the lack of evidence that the geographical and institutional distances of the media outlets are systematically changed after the cross-national media acquisition events, the possible impact of the cross-national change of media ownership is unlikely to change our inferences.

Our inferences endure when we use alternative measures for benchmark returns, including CRSP value-weighted market returns and CRSP equally weighted market returns.

The results remain quantitatively and qualitatively similar when unexpected earnings are scaled by the stock price at the end of the quarter.

As the FERC model requires information on three-year-ahead returns and earnings, the data used to calculated E_t+1,t+3 and RET_{t+1, t+3} are extended to 2020.

RavenPack classifies media outlets with a credibility ranking of 1 as fully accountable, reputable, and impartial, those with a ranking of 2 as official, reliable, and honest, and those with a ranking of 3 as acknowledged, formal, and credible.