781
Views
1
CrossRef citations to date
0
Altmetric
Articles

Common Ownership and Analyst Forecasts

, &
Pages 223-249 | Received 09 Dec 2021, Accepted 13 May 2022, Published online: 14 Jun 2022
 

Abstract

We examine the effect of the common ownership relation between brokerage houses and the firms covered by their analysts (referred to as co-owned brokerage houses, co-owned firms, and connected analysts, respectively) on analyst forecast performance. Common ownership can help the connected analysts have better access to co-owned firms, leading to higher-quality analyst research. However, common owners have incentives for higher valuation of the co-owned firms and thus can exert pressure on the connected analysts to issue optimistically biased research reports for these firms. We find that common ownership improves analyst forecast accuracy. This result is robust to a difference-in-differences design that exploits exogenous shocks to common ownership. The effects vary systematically with the quality of alternative sources of information that analysts can access for the co-owned firms. Overall, our paper contributes to the literature by documenting that common ownership can facilitate information communication.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 These studies have led to a hot debate on the antitrust effect of common ownership. For example, the antitrust regulatory bodies in the United States and Europe are contemplating the adverse impact of common ownership among industry peers on the extent of competition and customer welfare (Federal Trade Commission, Citation2018).

2 Figure A1 in the online appendixes provides an illustrative example for our research question. Vanguard Group Inc. holds beneficial stakes in several brokerage houses, including 6.65% of J.P. Morgan and 6.02% of General Electric Company (GE), among many others in 2019. An analyst, J. Inch, employed by J. P. Morgan issued earnings forecasts for GE in 2019. In this example, Vanguard is the common owner of the co-owned brokerage house, J. P. Morgan, and the co-owned firm, GE, and J. Inch is the connected analyst. Other analysts covering GE in 2019 who were employed by the brokerage houses that do not have common owners with GE are the non-connected analysts. In this paper, we explore how the common ownership relationship between J. P. Morgan and GE affects the quality of the forecasts issued for GE by the connected analyst, J. Inch, compared with non-connected analysts covering GE.

3 See SEC’s Final Rule: Selective Disclosure and Insider Trading at https://www.sec.gov/rules/final/33-7881.htm.

4 Details of these additional tests are presented in Table A3 of the online appendixes.

5 In untabulated tests, we investigate whether the Fair Disclosure (FD) regulation passed by the SEC in August 2000, which intends to prevent selective disclosure by publicly traded firms to market professionals and certain shareholders, has any effect on the association between common ownership and forecast performance. We do not find any mitigation effect of the FD regulation on the association. One interpretation for the result is that connected analysts obtain immaterial information from co-owned firms and thus this connection is not affected by the FD regulation.

6 While Kedia et al. (Citation2017) also examine the common ownership between a financial institution, Moody’s, and its rated firms, our paper differs from theirs in several important dimensions. First, unlike Kedia et al. (Citation2017), who document an adverse effect of common ownership on the credit ratings issued by Moody’s, we investigate the effect of common ownership on equity analysts’ forecast performance. Owing to the differences in regulatory and institutional environments for credit and equity analysts, the results documented in Kedia et al. (Citation2017) might not generalize to our setting. Second, focusing on analyst forecast performance, including both forecast bias and accuracy, allows us to examine both the positive and the negative effects of common ownership. Doing so would be difficult, if possible at all, in the credit rating setting. Lastly, we indeed document that common ownership improves analyst forecast performance.

7 Under the Global Analyst Research Settlement, the 10 largest U.S. investment banks involved in the settlement have been required to implement a series of reforms to improve analyst independence, such as separating research from investment banking business, linking analyst compensation to stock-picking ability, and disclosing any conflicts of interest faced by analysts in analyst reports. The SEC has also imposed various other disclosure and regulatory requirements to improve analyst research independence. See the SEC’s investor publication, “Analysing Analyst Recommendations,” available at https://www.sec.gov/tm/reportspubs/investor-publications/investorpubsanalystshtm.html.

8 Prior studies also provide evidence on the information communication role of common ownership in other settings, such as the credit market (Massa & Žaldokas, Citation2017).

9 The argument for the information hypothesis is not based on common owners’ incentives to increase the value of their investee firms or brokerage houses; instead, it is built on connected analysts’ incentives to improve their forecast performance.

10 We acknowledge that we lack direct empirical evidence to support such a conjecture. Tests would require a set of data on common ownership relationships between analysts’ brokerage houses and the firms that analysts follow across countries with different institutional environments in terms of the strengths of legal enforcements and investor protection. However, such tests are infeasible due to the limited availability of the data. We leave this question for future research when the data may become available.

11 For example, see Barroso et al. (Citation2018), Ge et al. (Citation2021), McCahery et al. (Citation2016), Edmans et al. (Citation2018), and Appel et al. (Citation2019).

12 This mentality is summarized succinctly by the former CEO of Vanguard Funds, F. William McNabb, in one of his speeches, “We’re going to hold your stock when you hit your quarterly earnings target. And we’ll hold it when you don’t. We’re going to hold your stock if we like you. And if we don’t. We’re going to hold your stock when everyone else is piling in. And when everyone else is running for the exits. That is precisely why we care so much about good governance.” https://corpgov.law.harvard.edu/2015/06/24/getting-to-know-you-the-case-for-significant-shareholder-engagement/.

13 See Appel et al. (Citation2019), McCahery et al. (Citation2016), and Fichtner et al. (Citation2017). Also see “Meet the new corporate power brokers: Passive investors,” Wall Street Journal. October 24, 2016. https://www.wsj.com/articles/the-new-corporate-power-brokers-passive-investors-1477320101.

14 The latest forecasts capture the information collected by analysts throughout the year and can thus better reflect analysts’ ability to collect and interpret information. In an untabulated test, we employ the first forecast issued by each analyst for a firm-year. The main references remain the same.

15 Prior research suggests that the 13F institutional ownership data are subject to some quality problems (e.g. missing information in 13F reports and incomplete coverage of securities) for the period after June 2013 (Lewellen & Lowry, Citation2021; He et al., Citation2020a). To ensure that our inferences are not affected by these data problems, we replicate the analyses using the data from the 1990–2013 period. The inferences remain the same. Please refer to the following documents for detailed information about the data quality problems and the potential fix for these problems:https://wrds-www.wharton.upenn.edu/documents/533/Research_Note_-Thomson_S34_Data_Issues.pdf; https://wrds-www.wharton.upenn.edu/documents/952/S12_and_S34_Regenerated_Data_2010-2016.pdf.

16 The common ownership variables have relatively tight distributions because we require common ownership to be at least 5%. When a firm shares multiple common owners with an analyst’s affiliated brokerage house in a year, we use the average ownership these common owners have in the co-owned firm and brokerage house in the analyses.

17 While these common owners are likely important clients of co-owned brokerage houses, they are also likely important clients of non-co-owned brokerage houses. Our empirical design of comparing forecast performance between connected and non-connected analysts following the same firm-year controls for the potential client catering effect these common owners have on analyst forecast performance.

18 Following prior studies (e.g. Cheong & Thomas, Citation2011; Kini et al., Citation2009), we also use total assets per share or the range of analyst forecast error (the difference between the maximum and minimum forecast error) as the deflator and obtain the same inferences. The same applies to the forecast bias measure.

19 Gormley and Matsa (Citation2014) show that controlling for firm-year fixed effects yields more consistent estimates than adjusting both the dependent and independent variables by their corresponding firm-year means as in other papers in the analyst literature (e.g. Call et al., Citation2009; Clement, Citation1999). When we use the mean-adjusted specification, our inferences remain the same.

20 We also note that different from the definition of common ownership relationships in our study, the percentage firms having common ownership relationships in He and Huang (Citation2017) and Park et al. (Citation2019) are for firms in the same industry. Relatedly, Matvos and Ostrovsky (Citation2008) report 15.5% of institutional shareholders simultaneously own both a target firm and an acquirer firm.

21 The highest variance inflation factor score for the variables in the regression analyses is much smaller than the conventional cut-off value of 10, suggesting that multicollinearity is unlikely to be a concern that would overturn our results.

22 The negative association between analyst firm experience and forecast accuracy is surprising, and inconsistent with Clement (Citation1999). However, this association is also shown in Kini et al. (Citation2009), suggesting that the direction of the relationship may vary depending on the research setting and the sample used in testing the relationship.

23 Note that these are not mergers of brokerage houses, which do not lead to a formation of common ownership between a brokerage house and a firm.

24 We acknowledge that the propensity score matching method is matched on observables and thus is limited in mitigating selection bias due to unobservables. Therefore, readers need to be cautious in generalizing the inferences of our results.

25 In the model, we remove analysts with equal to or less than one standard deviation of the number of forecasts (seven forecasts) to alleviate the estimation biases due to insufficient time series in high-dimensional fixed effects model (deHaan, Citation2021; Phillips & Sul, Citation2007). We note that we do not control for analyst fixed effects in our baseline regression model to avoid the same issues. In an untabulated test, we also conduct a within-analyst test and obtain the same inferences.

26 In untabulated tests, we explore the economic incentives that can strengthen or weaken the common ownership-induced conflicts of interest. First, as optimistically biased analyst forecasts can help uphold high stock prices, it is possible that common owners have stronger incentives to induce connected analysts to issue optimistic forecasts before selling their shares of the co-owned firms. We use the ex-post reduction in common owners’ holdings in the firm to capture their trading incentives. Second, prior research suggests that institutional investors with a short investment horizon care more about short-term stock price movements and focus more on the trading gains than those with a long investment horizon (e.g. Bushee & Goodman, Citation2007). Therefore, it is possible that the effect of common ownership on analyst forecast bias is more pronounced when common owners have a shorter investment horizon. We follow Gaspar et al. (Citation2005) and use the frequency that common owners balance their positions on all of the stocks in their portfolios in a quarter, referred to as the churn rate, as the proxy for their investment horizon. However, we do not find any results consistent with the predictions.

27 When a firm-year-analyst has more than one common owner, we use the average of ownership held by all common owners in the firm (brokerage house) in the regressions.

28 The finance literature commonly uses 10% or 5% ownership to define the control interests of institutions, funds, or family owners (e.g. Becht et al., Citation2009). Note that we only keep common owners whose holdings in both brokerage houses and firms are at least 5% of the outstanding shares to increase the power of the test in the sample selection process.

29 Note that HIGH_DD and HIGH_COPX discussed below are measured at the firm-year level and are thus not included in the regression owing to the inclusion of firm-year fixed effects. In contrast, MGT_FC discussed below is measured at the firm-year-analyst level and is thus included in the regression.

30 Mayew (Citation2008) argues that managers can use their discretion to give some analysts more opportunities to ask questions during firms’ conference calls, while discriminating against others.

31 We repeat the analyses on the differential market reaction upon stock recommendations issued by connected analysts as opposed to non-connected analysts. Untabulated results suggest an incremental positive (negative) and significant market reactions upon strong buy (strong sell) recommendations issued by connected analysts, consistent with stock recommendations issued by connected analysts being more credible than those by non-connected analysts.

Reprints and Corporate Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

To request a reprint or corporate permissions for this article, please click on the relevant link below:

Academic Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

Obtain permissions instantly via Rightslink by clicking on the button below:

If you are unable to obtain permissions via Rightslink, please complete and submit this Permissions form. For more information, please visit our Permissions help page.