Abstract
We examine how research posted by “social media analysts” (SMAs)—individuals posting equity research online via social media investment platforms—is related to research subsequently produced by professional sell-side equity analysts. Using data from Seeking Alpha, we find that the market reaction to sell-side analyst research is substantially reduced when the analyst research is preceded by the report of an SMA, and that this is particularly true of sell-side analysts’ earnings forecasts. We further find that this effect is more pronounced when SMA reports contain more decision-useful language, are produced by SMAs with greater expertise, and relate to firms with greater retail investor ownership. We also provide evidence that the attenuated response to sell-side research is most likely explained by SMA research preempting information in sell-side research and that analysts respond to SMA preemption with bolder and more disaggregated forecasts. Collectively, our results suggest that equity research posted online by SMAs provides investors with information that is similar to but arrives earlier than sell-side equity research, and speak to the connected and evolving roles of information intermediaries in capital markets.
Similar content being viewed by others
Notes
Given our research objective, we focus on investment-related social media platforms (Seeking Alpha in particular). We do not use the term “crowdsourced” because, unlike venues such as Estimize or Glassdoor, the research, opinions, and analyses we examine are not aggregated or crowdsourced in any way.
As of May 2021, Seeking Alpha reports over 40 million monthly site visits per month spread across 15.2 million unique users (Seeking Alpha 2021).
Because of the flurry of news released by various financial market participants following firms’ disclosures of earnings news, we conduct our primary tests using a restricted sample of forecasts issued outside of periods when firms disclose earnings or earnings guidance, but in untabulated analyses we find generally consistent results using a full, unrestricted sample of forecasts.
“How Hedge Funds Rate Wall Street Analysts”, Alpha Magazine, November 21, 2005. This anecdotal evidence may initially appear inconsistent with the conclusion in Amiram et al. (2016) that sell-side analyst forecasts represent new information only to less sophisticated, retail investors. However, it is likely that the timing of the “high-touch” services provided to institutional clients does not correspond with the timing of analysts’ public forecasts. In this case, one could still observe the result in Amiram et al. (2016) even with a shift in focus towards institutional clients.
According to Chen et al. (2014), contributors on Seeking Alpha earn $10 per one thousand page views. Seeking Alpha also helps authors promote their work on major media outlets and hosts networking events, both of which help contributors build their reputations in the investment community and potentially monetize their skills through other means (Seeking Alpha 2019). In addition, Seeking Alpha hosts a “marketplace” where authors can sponsor their own “paid-for” research platform, which further incentivizes SMAs to produce high-quality analysis.
Seeking Alpha now puts most research behind a relatively inexpensive paywall. Users may still freely access current and recent analysis for stocks in the portfolios they maintain in their user accounts. During our sample period, the Seeking Alpha content we analyze was free to all users.
We do not collect content with “news” URLs, as those typically represent news flashes or dissemination of news published elsewhere.
One concern is that SMA reports that precede sell-side analyst reports systematically differ from those that do not. We evaluate whether this is the case in untabulated analyses. The mean (median) absolute returns for reports that precede analyst reports is 0.022 (0.013) percent, compared to 0.021 (0.013) percent for those that do not. We also find that SMA reports preceding sell-side analyst reports are slightly shorter than SMA reports not preceding them (average word count of 826 vs. 846, respectively). Thus, it is unlikely that significant differences in content contribute to our results.
We also consider a measure of forecast news derived from the consensus forecast and find similar results.
For SMA reports and sell-side forecasts issued after 4 p.m., we adjust the announcement date to the next trading day so that our return windows (described later) correctly identify the event day. We also delete a small number of sell-side forecasts dated after the firm’s earnings announcement, which likely reflect data errors.
Despite our best efforts to control for factors that likely contribute to the SMA’s decision to publish research and to the pricing of analyst research, the potential for an omitted variable remains. We evaluate parameters under which an omitted variable would alter our inferences using the method developed in Oster (2019). We implement this procedure as in Call et al. (2018). Untabulated analyses suggest that, in order to alter our inferences in Table 3, omitted correlated variables would need to be 1.1 to 5.2 times more important than the combined effect of the vector of controls we include. While omitted variables remain a limitation of our study, we believe that these diagnostics suggest that the likelihood of such a variable playing a significant role in our analyses is low.
To the extent that SMA reports increase the likelihood that news about the upcoming analyst forecast is disseminated either on social media (SMA[0,1]) or by the business press (BizPress[0,1]), these two controls may not be appropriate, as they are not predetermined with respect to our variable of interest (Whited et al. 2021). However, if we exclude them, then our results could plausibly be driven by dissemination (since business press coverage and social media coverage surrounding the forecast correlate with coverage prior to the forecast). If we exclude these two variables, our results are qualitatively similar (untabulated). In addition, we do not include a control for other analyst forecasts contemporaneous to the forecast of interest (i.e., ProfAnalyst[0,1]), because we collapse our dataset to the firm-day level. Finally, we note that results are similar if we use the logged count of business press articles and prior analyst forecasts rather than indicator variables (untabulated).
Results are robust to clustering standard errors one-dimensionally by industry-month-year instead (untabulated).
Leone et al. (2019) note that winsorization is not an effective method for addressing significant outliers. As such, we tabulate results after applying Cook’s distance. We obtain similar results if we exclude observations with studentized residuals exceeding 2 or estimate the models using robust regressions (with industry-year-month fixed effects). Consistent with winsorization not effectively addressing outliers, we find inconsistent results for the first four columns of Table 3 using winsorization alone.
As an additional means of comparing the effects of these various intermediaries, in untabulated analysis we include all possible interaction terms of SMA[−7,−1], ProfAnalyst[−7,−1], and BizPress[−7,−1]. We find that the attenuating effect of SMA reports is observed in the presence of an analyst report, a business press article, or both. In contrast, the main effects of ProfAnalyst and BizPress are not consistently significant.
Note that we estimate these models at the analyst level, so there can be more than one observation per trading day. In addition, we only estimate these models using the analyst forecast sample (i.e., excluding recommendations and price targets), because IBES scrambled analyst identifiers prior to our accessing data related to price targets and recommendations. This scrambling makes analyst fixed effects impossible in tests that combine research outputs, such as those in our Table 3 analyses.
To illustrate the column 1 calculation, 0.164 divided by 0.387 equals 0.424.
We use Python’s “textblob” package to measure sentence-level sentiment. This measure of sentiment has two parts: polarity, which is analogous to tone and varies between −1 (negative) and +1 (positive), and subjectivity, which varies between 0 and 1. We consider a sentence to generate higher uncertainty if it conveys relatively strong sentiment, defined as an absolute value of polarity that is greater than 0.75, and high subjectivity, defined as greater than 0.50. Note that we also considered dictionary-based measures of uncertainty from both Loughran and McDonald (2014) and General Inquirer dictionaries and observed no significant differences.
The correlation between institutional holdings and retail trading intensity is − 0.44, suggesting that these two proxies capture similar but not identical aspects of ownership structure.
We use sample partitions (fully interacted models) for these tests since our partitioning variable is defined for all observations. For the analyses discussed in Sect. 5.1.1, the partitioning variable (e.g., expertise) is only defined when SMA = 1. Therefore, we use SMAhigh and SMAlow in those analyses.
This method identifies retail trades using a regulatory restriction that retail orders can receive price improvements (measured in small fractions of a cent per share) but institutional orders cannot. Using TAQ data, we divide the transaction price by 1 cent (0.01). If the remainder is in the interval (0.0, 0.4], then we identify the trade as a retail sell transaction; if the remainder is in the interval [0.6, 1.0), then we identify the trade as a retail buy transaction. Trades that occur at a round penny (remainder = 0) or those with remainders that fall around the half-penny (remainders in the interval (0.4, 0.6)) are not categorized as retail for conservatism. We then aggregate retail trading volume over each trading day.
To evaluate whether SMAs generally agree with sell-side analysts, we compare the tone of SMA reports, computed using the Loughran and McDonald (2014) sentiment dictionary, to AF. As expected, we observe a significantly positive association (untabulated), suggesting that the tenor of the news contained in the reports of SMAs and those of sell-side analysts generally track one another.
In untabulated analysis, we find similar (non)results using intraperiod timeliness over the same three windows (though starting at day 0) to measure the efficiency of the price formation process (Twedt 2016).
We find similar inferences using the natural log of the number of SMA reports over the 90-day period instead of SMA[− 90,− 1].
Note that Bloomberg data begins in 2010, resulting in a smaller sample for this test.
References
Seeking Alpha. (2019). About Seeking Alpha. https://seekingalpha.com/page/about_us. Accessed January 2019.
Seeking Alpha. (2021). Who reads seeking alpha? https://seekingalpha.com/page/who_reads_sa. Accessed May 2021
Amiram, Dan, Edward Owens, and Oded Rozenbaum. (2016). Do information releases increase or decrease information asymmetry? New evidence from analyst forecast announcements. Journal of Accounting and Economics 62 (1): 121–138. https://doi.org/10.1016/j.jacceco.2016.06.001.
Armstrong, Robert. (2018). The unmourned death of the sellside analyst. Financial times. https://www.ft.com/content/8699b996-f141-11e7-b220-857e26d1aca4. Accessed April 2019
Bartov, Eli, Lucile Faurel, and Partha Mohanram. (2018). Can twitter help predict firm-level earnings and stock returns? The Accounting Review 93 (3): 25–57. https://doi.org/10.2308/accr-51865.
Ben-Rephael, Azi, Zhi Da, and Ryan Israelsen. (2017). It depends on where you search: institutional investor attention and underreaction to news. The Review of Financial Studies 30 (9): 3009–3047. https://doi.org/10.1093/rfs/hhx031.
Beyer, Anne, Daniel Cohen, Thomas Lys, and Beverly Walther. (2010). The financial reporting environment: review of the recent literature. Journal of Accounting and Economics 50 (2/3): 296–343. https://doi.org/10.1016/j.jacceco.2010.10.003.
Blankespoor, Elizabeth, Ed. de Haan, and Ivan Marinovic. (2020). Disclosure processing costs and investors information choice: a literature review. Journal of Accounting and Economics 70 (2): 1–45.
Boehmer, Ekkehart, Charles Jones, Xiaoyan Zhang, and Xinran Zhang. (2020). Tracking retail investor activity. Journal of Finance Forthcoming. https://doi.org/10.1111/jofi.13033.
Bradshaw, Mark, Yonca Ertimur, and Patricia O’Brien. (2016). Financial analysts and their contribution to well-functioning capital markets. Foundations and Trends in Accounting 11 (3): i–iv. https://doi.org/10.1561/1400000042.
Bradshaw, Mark. (2011). Analysts forecasts: what do we know after decades of work? SSRN Working paper. https://doi.org/10.2139/ssrn.1880339
Brown, Lawrence, Andrew Call, Michael Clement, and Nathan Sharp. (2015). Inside the ‘black box’ of sell-side financial analysts. Journal of Accounting Research 53 (1): 1–47. https://doi.org/10.1111/1475-679X.12067.
Call, Andrew, Gerald Martin, Nathan Sharp, and Jeron Wilde. (2018). Whistleblowers and outcomes of financial misrepresentation enforcement actions. Journal of Accounting Research 56 (1): 123–171. https://doi.org/10.1111/1475-679X.12177.
Campbell, John L., Matthew DeAngelis, and James Moon. (2019). Skin in the game: personal stock holdings and investors’ response to stock analysis on social media. Review of Accounting Studies 24: 731–779. https://doi.org/10.1007/s11142-019-09498-9.
Chen, Hailiang, YuHu. De Prabuddha, and Byoung-Hyoun. Hwang. (2014). Wisdom of crowds: the value of stock opinions transmitted through social media. Review of Financial Studies 27 (5): 1367–1403. https://doi.org/10.1093/rfs/hhu001.
Clement, Michael. (1999). Analyst forecast accuracy: do ability, resources, and portfolio complexity matter? Journal of Accounting and Economics 27: 285–303. https://doi.org/10.1016/S0165-4101(99)00013-0.
Clement, Michael, and Senyo Tse. (2005). Financial analyst characteristics and herding behavior in forecasting. The Journal of Finance 60 (1): 307–341. https://doi.org/10.1111/j.1540-6261.2005.00731.x.
Cowen, Amanda, Boris Groysberg, and Paul Healy. (2006). Which types of analyst firms are more optimistic? Journal of Accounting and Economics 41 (1/2): 119–146. https://doi.org/10.1016/j.jacceco.2005.09.001.
Daniel, Kent, Mark Grinblatt, Sheridan Titman, and Russ Wermers. (1997). Measuring mutual fund performance with characteristic-based benchmarks. The Journal of Finance 52 (3): 1035–1058. https://doi.org/10.1111/j.1540-6261.1997.tb02724.x.
Drake, Michael, Jacob Thornock, and Brady Twedt. (2017). The internet as an information intermediary. Review of Accounting Studies 22 (2): 543–576. https://doi.org/10.1007/s11142-017-9395-1.
Drake, Michael, Peter Joos, Joseph Pacelli, and Brady Twedt. (2019). Analyst forecast bundling. Management Science Forthcoming. https://doi.org/10.1287/mnsc.2019.3339.
Fama, Eugene, and James MacBeth. (1973). Risk, return, and equilibrium: empirical tests. The Journal of Political Economy 81 (3): 607–636. https://doi.org/10.1086/260061.
Farrell, Michael, Clifton Green, Russell Jame, and Stanimir Markov. (2020). The democratization of investment research and the informativeness of retail investor trading. SSRN. Working paper. https://doi.org/10.2139/ssrn.3222841
Frankel, Richard, S.P. Kothari, and Joseph Weber. (2006). Determinants of the informativeness of analyst research. Journal of Accounting and Economics 41 (1/2): 29–54. https://doi.org/10.1016/j.jacceco.2005.10.004.
Gleason, Cristi, and Charles Lee. (2003). Analyst forecast revisions and market price discovery. The Accounting Review 78 (1): 193–225. https://doi.org/10.2308/accr.2003.78.1.193.
Gomez, Enrique, Frank Heflin, James Moon, and James Warren. (2020). Can financial analysis on social media help level the playing field among investors? Evidence from Seeking Alpha. SSRN. Working paper. https://doi.org/10.2139/ssrn.3274644
Green, Clifton, Russell Jame, Stanimir Markov, and Musa Subasi. (2014). Broker-hosted investor conferences. Journal of Accounting and Economics 58 (1): 142–166. https://doi.org/10.1016/j.jacceco.2014.06.005.
Groysberg, Boris, and Paul Healy. (2013). Wall Street research: past, present, and future. Stanford University Press.
Guan, Yuyan, Congcong Li, Lu. Hai, and Franco Wong. (2019). Regulations and brain drain: evidence from wall street star analysts career choices. Management Science 65 (12): 5766–5784. https://doi.org/10.1287/mnsc.2018.3182.
Hales, Jeffrey, James Moon, and Laura Swenson. (2018). A New Era of voluntary disclosure? Empirical evidence on how employee postings on social media relate to future corporate disclosures. Accounting Organizations and Society 68–69: 88–108. https://doi.org/10.1016/j.aos.2018.04.004.
Jackson, Andrew. (2005). Trade generation, reputation, and sell-side analysts. The Journal of Finance 60 (2): 673–717. https://doi.org/10.1111/j.1540-6261.2005.00743.x.
Jame, Russell, Rick Johnston, Stanimir Markov, and Michael Wolfe. (2016). The value of crowdsourced earnings forecasts. Journal of Accounting Research 54 (4): 1077–1110. https://doi.org/10.1111/1475-679X.12121.
Jame, Russell, Stanimir Markov, and Michael C. Wolfe. (2017). Does crowdsourced research discipline sell-side analysts? SSRN. Working Paper. https://doi.org/10.2139/ssrn.2915817
Kadan, Ohad, Leonardo Madureira, Rong Wang, and Tzachi Zach. (2009). Conflicts of interest and stock recommendations: the effects of the global settlement and related regulations. Review of Financial Studies 22 (10): 4189–4217. https://doi.org/10.1093/rfs/hhn109.
Lang, Mark, and Russell Lundholm. (1996). Corporate disclosure policy and analyst behavior. The Accounting Review 71 (4): 467–492.
Lee, Justina. (2019). Research analysts’ existential crisis enters MiFID II Era. Bloomberg. 2019. https://www.bloomberg.com/news/articles/2019-01-03/the-research-analyst-s-existential-crisis-enters-mifid-ii-era. Accessed April 2019
Leone, Andrew, Miguel Minutti-Meza, and Charles Wasley. (2019). Influential observations and inference in accounting research. The Accounting Review 94 (6): 337–364.
Li, Xi. (2010). The impacts of product market competition on the quantity and quality of voluntary disclosures. Review of Accounting Studies 15 (3): 663–711. https://doi.org/10.1007/s11142-010-9129-0.
Li, Edward, K. Ramesh, Min Shen, and Wu. Joanna. (2015). Do analyst stock recommendations piggyback on recent corporate news? An analysis of regular-hour and after-hours revisions. Journal of Accounting Research 53 (4): 821–861. https://doi.org/10.1111/1475-679X.12083.
Lin, Hsiou-wei, and Maureen McNichols. (1998). Underwriting relationships, analysts earnings forecasts and investment recommendations. Journal of Accounting and Economics 25 (1): 101–127. https://doi.org/10.1016/S0165-4101(98)00016-0.
Loughran, Tim, and Bill Mcdonald. (2014). Measuring Readability in Financial Disclosures. The Journal of Finance 69 (4): 1643–1671. https://doi.org/10.1111/jofi.12162.
Mayew, William. (2008). Evidence of management discrimination among analysts during earnings conference calls. Journal of Accounting Research 46 (3): 627–659. https://doi.org/10.1111/j.1475-679X.2008.00285.x.
Miller, Greg, and Doug Skinner. (2015). The evolving disclosure landscape: how changes in technology, the media, and capital markets are affecting disclosure. Journal of Accounting Research 53: 221–239. https://doi.org/10.1111/1475-679X.12075.
Morris, Stephen. (2017). Banks to cut $1.2 billion in research spending, mckinsey says. Bloomberg. https://www.bloomberg.com/news/articles/2017-06-21/banks-to-cut-1-2-billion-in-research-spending-mckinsey-says. Accessed April 2019
Oster, Emily. (2019). Unobservable selection and coefficient stability: theory and evidence. Journal of Business & Economic Statistics 37 (2): 187–204. https://doi.org/10.1080/07350015.2016.1227711.
Petersen, Mitchell. (2009). Estimating standard errors in finance panel data sets: comparing approaches. Review of Financial Studies 22 (1): 435–480. https://doi.org/10.3386/w11280.
Pumfrey, Mark. (2019). Technology will sharpen the trader’s competitive edge. Financial times. https://www.ft.com/content/1c766615-5e31-309d-b055-87327f37619d. Accessed April 2019.
Ramnath, Sundaresh, Steve Rock, and Philip Shane. (2008). A review of research related to financial analysts forecasts and stock recommendations. Foundations and Trends in Finance. https://doi.org/10.1561/0500000023.
Rees, Lynn, Nathan Sharp, and Brady Twedt. (2015). Who’s heard on the street? Determinants and consequences of financial analyst coverage in the business press. Review of Accounting Studies 20 (1): 173–209. https://doi.org/10.1007/s11142-014-9298-3.
Tang, Vicki. (2018). Wisdom of crowds: cross-sectional variation in the informativeness of third-party-generated product information on twitter. Journal of Accounting Research 56 (3): 989–1034. https://doi.org/10.1111/1475-679X.12183.
Twedt, Brady. (2016). Spreading the word: price discovery and newswire dissemination of management earnings guidance. The Accounting Review 91 (1): 317–346. https://doi.org/10.2308/accr-51129.
Weil, Roman, Patricia O’Brien, Michael Maher, and Clyde Stickney. (1994). Accounting: The language of business. T. Horton and Daughters
Whited, Robert, Quinn Swanquist, Jonathan Shipman, and James Moon. (2021). Out of control: the (over)use of controls in accounting research. The Accounting Review Forthcoming. https://doi.org/10.2308/TAR-2019-0637.
Acknowledgements
We thank two anonymous reviewers, John Campbell, Ted Christensen, Eric Condie, Braiden Coleman, Matthew DeAngelis, Peter Easton (editor), Mac Gaulin (discussant), Ilan Guttman, Chad Ham, Mirko Heinle, Wen Hi (discussant), Mike Iselin, Ajay Kohli, Josh Madsen, Saby Mitra, Joseph Pacelli, Marlene Plumlee, Christina Zhu (discussant), and workshop participants at Georgia Tech, University of Calgary, University of Minnesota, University of Oregon, the 2020 Financial Accounting and Reporting Section Midyear Meeting, the 2020 UTS Summer Accounting Conference, and the 2020 Utah Winter Accounting Conference for helpful comments and suggestions.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix A
Appendix A
Rights and permissions
About this article
Cite this article
Drake, M.S., Moon, J.R., Twedt, B.J. et al. Social media analysts and sell-side analyst research. Rev Account Stud 28, 385–420 (2023). https://doi.org/10.1007/s11142-021-09645-1
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11142-021-09645-1