A central question about the media’s role in capital markets is whether media coverage influences investors’ reactions to firms’ disclosures, such as earnings releases. Multiple studies provide evidence consistent with the media disseminating, but not creating, information that influences the market’s earnings response. Specifically, they find that brief articles without, but not lengthier articles with, substantive editorial content are associated with the earnings response. In contrast to these prior studies, I provide evidence that the market does respond to earnings-related editorial content.
The earlier finding is somewhat puzzling because editorial content appears to have a significant impact in other financial settings, including identifying fraud and monitoring executive pay. Moreover, intuition suggests that editorial content could inform readers’ trading decisions around earnings announcements in several ways, such as by contextualizing or simplifying the firm’s disclosure. To be more specific, reporters’ earnings articles could disseminate (i.e., repeat without revision) firms’ and other market participants’ information more broadly, aggregate or repackage publicly available information from multiple sources, and/or contribute original insights. These activities could inform readers’ trading decisions through several channels, including (but not limited to) identifying which announcements merit attention, gathering and synthesizing diffuse information, corroborating or contradicting public information, revealing others’ private information through exclusive quotes, simplifying opaque earnings releases, noting industry- or economy-wide trends, and emphasizing issues that firms (or perhaps analysts) lack incentives to address. For example, the Wall Street Journal (WSJ) article covering General Dynamics’ fiscal 2013 earnings release contrasts perceived good news (share buyback) and bad news (lower profit and sales forecasts), compares the firm’s and competitors’ buyback programs, and discusses the effect of government spending cuts on the firm’s industry.
However, it is possible that WSJ articles have little effect on investors’ earnings response. Perhaps journalists lack sufficient training in finance and accounting to provide new insights about technical disclosures. Or high-frequency and other sophisticated traders may already impound earnings news into price before journalists can digest the news and publish an article. Alternatively, articles could even impede market reactions by inducing investors to focus on trivial or sensational ideas and ignore value-relevant information. If any of these is true, then journalists’ earnings articles might simply entertain investors (e.g., day traders) but not improve, or even influence, their trading. In other words, this entertainment role could fully explain why the media produces in-depth earnings coverage. While my study does not rule out the entertainment role playing some part, it suggests earnings-related editorial content also plays an information role.
My research methodology differs from the prior studies in several ways, which I introduce briefly here and describe further below. First, I study WSJ earnings articles, which feature more editorial content than many other earnings-related media. Second, I focus on highly visible S&P 500 firms whose news attracts many readers and is widely disseminated. Thus, dissemination is likely not the main issue for these firms. Third, I exploit a series of restructuring events at the WSJ that produced variation in earnings coverage that arose for reasons outside the firm. Fourth, I measure the extent to which, and specific ways in which, each media article differs from the firm’s press release to provide evidence that the media creates, not merely disseminates, value-relevant earnings news.
WSJ coverage of S&P 500 earnings is an ideal setting for testing the role of editorial content for multiple reasons. First, many earnings-related media articles, such as those published by the Associated Press (AP) and Dow Jones (DJ) Newswires, do little more than supplement the earnings number with analyst forecasts and recent stock returns. My tests instead focus on WSJ articles, which are substantially longer and more in-depth on average. Second, the benefits to the media of creating information increase relative to firms size and visibility, which reflect the number of stakeholders (i.e., potential readers). Accordingly, my tests examine only S&P 500 firms. As the largest and most visible firms in the economy, they are most likely to be the subject of original analysis (or in other words, “information creation”) by the media. These research choices help me determine whether the media merely disseminates and entertains, or also informs.
Identifying any causal effects of media coverage on investor decisions is challenging because the coverage is in part determined by the firm and news being covered (or not covered). For example, it is unclear whether investors react more when the media covers earnings due to the coverage or whether they would react more anyway, perhaps because the media covers firms with extreme news. I use an instrumental variables approach in some analyses to address this issue. I also exploit three distinct restructuring events at the WSJ that changed some firms’ earnings coverage for reasons independent of changes in the firms’ economic characteristics. Crucially, earnings coverage by other financial news outlets did not change at the same time as the WSJ restructuring, consistent with the WSJ changes occurring due to idiosyncratic strategic decisions rather than industry-wide trends.
Of course, reporters likely rank firms by newsworthiness and cover the highest-ranking firms. As a result, typical coverage changes occur due to changes in the newsworthiness ranking, which is closely related to changes in firms’ earnings reactions and other fundamentals. However, due to the restructuring events, the WSJ likely started (or stopped) covering the next most (or least) newsworthy firms they had not (or had) been covering, even though many of these firms’ newsworthiness ranking had likely not changed much, if at all. In other words, coverage could change because the firm becomes more or less newsworthy relative to other firms, and/or because the WSJ covers more or fewer firms from the newsworthiness ranking. Consistent with the latter, I find that the WSJ coverage changes are not explained by changes in current or anticipated newsworthiness. In contrast, coverage changes during normal times are significantly explained by changes in newsworthiness, suggesting they are endogenous.
In testing the hypothesis that the market responds to editorial content, the expected direction of the response is initially unclear. On the one hand, editorial content that corroborates the earnings reported by the firm likely speeds up price discovery and increases trading. On the other hand, editorial content could also contradict earnings, such as by pointing out transitory items, which could slow down price discovery and decrease trading. While my main results are consistent with editorial content corroborating earnings on average, subsequent analyses lead to more nuanced conclusions, as described below.
To test these possibilities, I follow the vast literature that examines the earnings response coefficient (ERC) to understand the effects of earnings releases on investor beliefs. The ERC captures the magnitude of the short-window price response to earnings. I compare the change in ERC of treatment firms, whose WSJ coverage changed as predicted at the respective event, and control firms, whose WSJ coverage did not change. I find WSJ coverage increases the ERC by about 39%, which is comparable to other recent studies on various ERC treatment effects. As an additional price discovery test, I consider intra-period timeliness (IPT), which captures the speed at which prices incorporate earnings during the few days following the announcement. The results suggest that WSJ articles also significantly increase IPT. Both the ERC and IPT findings are consistent with faster price discovery due to WSJ content.
Several studies suggest considering trading volume in conjunction with returns at disclosure events is likely to yield more refined insights than considering either in isolation. Accordingly, my next tests examine the effect of WSJ articles on the amount of earnings-announcement trading volume. I expect that WSJ articles influence investors’ beliefs about the implications of the earnings release for the firm’s value, leading to more trading. My estimates are consistent with this prediction. While more trading might be driven by investors who are entertained but not informed by the news, such “entertainment” trading is unlikely to increase market efficiency. Hence, my collective findings on price discovery and trading volume seem more consistent with editorial content filling an information role.
Another challenge facing the media literature is distinguishing between information dissemination and creation. As discussed above, earnings-related editorial content could fill either (or neither) role. My next analyses examine this issue using textual analysis to measure differences between the firm’s earnings press release and the associated WSJ article. I then test whether low and/or high difference articles are informative. Crucially, low (or high) difference articles seem more likely candidates to play an information dissemination (or creation) role. I calculate two measures: word and tone difference. Word difference is used by prior studies on media articles and firms’ disclosures to distinguish between dissemination of “stale” information and the creation of economically meaningful information. Tone difference, calculated using words with a negative connotation in the finance context, allows me to assess the extent to which the firm and media agree on whether the firms’ earnings and other fundamentals reflect positive or negative news.
For price discovery, the positive effects of WSJ coverage are concentrated among articles with high word difference, but only when these articles have low tone difference. For trading volume, both word and tone differences have a positive effect. Taken together, these nuanced results suggest that editorial content increases trading activity by creating new information, and that when this information corroborates (or contradicts) the tone of the firm’s news it speeds up (or slows down) the price reaction. Intuitively, this could be because corroboratory information increases consensus among investors and contradictory information increases disagreement that takes time to resolve. Similarly, agreement (or disagreement) between firm and media could increase (or decrease) investors’ assessments of earnings credibility. Perhaps surprisingly, the positive treatment effect on both prices and volume is absent among low word difference articles. This evidence contradicts the idea that WSJ articles fill a dissemination role and could be due to the relatively high level of dissemination of S&P 500 firms’ news by other outlets. Because word and tone differences are likely confounded by firm-level factors, I use the amount of competing earnings and industry news as instrumental variables and find similar results.
To provide additional insight into how articles create information, I next compare several linguistic characteristics of the high and low word difference articles. The high difference articles are slightly longer, are more readable and specific, include more references to the industry and economy, repeat less “stale” news published in previous WSJ articles, and quote more investor and expert sources. All these features are consistent with high difference articles creating more value-relevant information that could enhance investors’ ability to understand the implications of earnings.
My study’s most direct contribution is to research on the role of media coverage surrounding firms’ earnings announcements. My findings suggest the editorial content reporters provide helps investors better understand firms’ earnings, instead of simply entertaining or increasing awareness. In addition, I provide novel evidence on the specific aspects of earnings-related editorial content (e.g., readability, quotes, etc.) that help investors. But an important caveat is that the WSJ events only allow me to estimate the treatment effect of WSJ coverage of the S&P 500, so my results do not necessarily generalize to other firms or media outlets.
This study also complements the many studies that find the media creates information in other settings, such as by monitoring firms for fraud or excessive CEO pay. These investigative efforts to uncover new information first, which journalists often refer to colloquially as “exclusives” or “scoops,” can take weeks or months and make or break a reporter’s career. The evidence in my study suggests reporters also create value-relevant information while covering “breaking” news such as routine earnings coverage, a setting in which the stakes are relatively low but that provides only a few hours to produce an article. These findings may partially explain the continued existence of in-depth articles about earnings announcements vis-a-vis the proliferation of low-cost alternatives.
In addition, I contribute to the broader literature on the capital market roles of information intermediaries. My findings suggest journalists, like sell-side analysts, create information about firm value. Although both create information, journalists can likely enrich a firm’s information environment in ways analysts cannot because journalists reach a broader audience and have incentives that often make them more credible. Moreover, my results relate to the many papers finding larger risk premiums and/or mispricing among smaller stocks. Specifically, if the media’s earnings editorial content can improve market efficiency in the rich S&P 500 information environment, as my results suggest, then the potential returns to processing earnings are likely even greater among smaller firms.
Nicholas Guest is an Assistant Professor of Accounting at the Johnson Graduate School of Management, Cornell University.
This post is adapted from his paper, “The Information Role of the Media in Earnings News,” available on SSRN.