International Earnings Announcements: Tone, Forward-looking Statements, and Informativeness

Abstract

This paper examines two attributes of earnings press releases issued by firms cross-listed on U.S. stock exchanges: the tone and frequency of forward-looking statements. A more conservative tone and a greater proportion of forward-looking statements are often viewed as contributing to more credible disclosures. Our analysis indicates that culturally and institutionally more distant firms are generally less positive in their disclosures and include more forward-looking statements than U.S. firms. Further, we find that the tone and frequency of forward-looking statements of cross-listed firms’ earnings announcements are more informative than those of U.S. firms in predicting future firm performance, and this informativeness generally increases with the cultural and institutional distance of the home country from the U.S. In explaining market reaction to earnings announcements, tone informativeness in particular increases with the cultural distance of the home country from the U.S. Overall, in the context of home bias theory, we interpret our findings as suggesting that a cautious disclosure tone and more forward-looking information serve to mitigate potential home bias-related credibility and asymmetric information concerns arising from cultural and institutional distance.

Keywords:

• Disclosure
• Tone
• Forward-looking statements
• Credibility
• Home bias
• Cross-listed firms

JEL codes:

• M41
• G30
• G40

Acknowledgements

We thank Hervé Stolowy (Associate Editor) and two anonymous reviewers for their helpful feedback and suggestions. We also thank Özgür Arslan-Ayaydin, Catherine D’Hondt, Frederiek Schoubben, Kristof Struyfs, Cynthia Van Hulle and Beibei Yan for their valuable comments. We greatly acknowledge the comments of the participants at the 2019 JIAR Conference (Québec, Canada).

Supplemental Data and Research Materials

Supplemental data for this article can be accessed on the Taylor & Francis website, https://doi.org/10.1080/09638180.2021.1963795.

Table A1: Correlation Matrix

Table A2: Post-U.S. GAAP reconciliation

Table A3: Alternative measures of tone and unexpected tone

Table A4: Separate Hofstede cultural dimensions

Table A5: Analysis with distance fixed by the single-dimension level

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

1 We use the term ‘cross-listed firms’ to refer to non-U.S. firms that are listed on a U.S. stock exchange. SEC regulations also refer to cross-listed firms as foreign private issuers (FPIs). FPIs can list their securities on a U.S. stock exchange either directly or via a depositary receipt DR (i.e., an American Depositary Receipt, ADR). This article has been republished with minor changes. These changes do not impact the academic content of the article.

2 Our paper differs from that of Lundholm et al. (2014) in several ways. In contrast with their focus on disclosure readability and numerical intensity, our focus is on different disclosure attributes, namely, a cautious tone and more FLS. In addition, we focus on cultural distances from the U.S. as well as institutional distances. Moreover, in contrast with their focus on the outcome of whether U.S. institutional investors buy more of certain cross-listed firms’ stock, our outcome measures are whether more cautious and forward-looking disclosures made by more distant cross-listed firms are more informative in terms of predicting future firm performance and incrementally explaining market reaction.

3 While we do not presuppose that U.S. investors are the only audience for cross-listed firms’ English language earnings announcements, we argue that U.S. investors are a particularly important audience because of the relative size of the U.S. investor base compared to most cross-listed firms’ home markets.

4 Annual reports typically include ‘boilerplate’ resulting in increased semantic and linguistic duplicates within and across firms (Dyer et al., 2017). To the extent that annual reports are increasingly used as compliance documents rather than a means of conveying firm-specific information, they are less valuable for a comparative analysis of disclosure.

5 For example, cross-listed firms are generally exempt from the SEC regulation most pertinent to earnings announcements – Regulation G, which restricts the manner in which U.S. firms disclose non-GAAP metrics. Regulation G requires SEC registrants using non-GAAP measures to provide ‘a presentation, with equal or greater prominence, of the most directly comparable financial measure calculated and presented in accordance with GAAP’ (SEC, 2003). Cross-listed firms are exempt from Regulation G if they are listed on an exchange outside the U.S., have also made disclosures outside the U.S., and do not derive the non-GAAP measure from U.S. GAAP-based amounts (SEC, 2017).

6 Our measure includes the five Hofstede cultural dimensions we consider to be most relevant to financial disclosure choices. The results (untabulated) remain qualitatively similar when including all dimensions. In addition, this paper, like the vast majority of related literature, relies on the cultural index formula utilized in Kogut and Singh (1988). The Kogut and Singh distance measure has been shown to be a ‘special’ case of the Mahalanobis distance that assumes that the dimensions used in computing cultural distance are not correlated (Kandogan, 2012). Several papers rely on the Mahalanobis measure (e.g., Beugelsdijk et al., 2018), whereas others take the square root of the index, relaxing the assumption of uncorrelated dimensions (e.g., Karolyi, 2016). We examine various alternatives in composing the cultural distance measure, including the Mahalanobis distance and the square root, and our results remain qualitatively similar.

7 Following Lundholm et al. (2014), we subtract the grand mean of the cultural and the institutional distance measures of the foreign sample from all the non-U.S. observations. As such, the mean of the distance variables is zero both in the U.S. sample and in the sample of cross-listed firms. This ensures that the dummy variable DF isolates the main effect of ‘being cross-listed,’ whereas the distance measures control for the incremental impact of cultural and institutional distance.

8 Event studies often use a three-day window around the earnings announcement (e.g., Huang et al., 2014; Henry, 2008); however, we extend the trading window in our context because the SEC's filing deadline for foreign private issuers’ current reports is more relaxed. Cross-listed firms are not subject to the four-day deadline applicable to U.S. issuers’ current reports on Form 8-K (SEC, 2004) and instead are required to file Form 6-K ‘promptly after the material contained in the report is made public. Interim reporting for foreign private issuers parallels the requirements of the issuer's home country regulatory and stock exchange practices’ (SEC, 2013).

9 We elect to hand collect earnings announcements because SEC Form 6-K is used for cross-listed firms’ interim reports as well as many other types of disclosures, and unlike the domestic equivalent (Form 8-K), Form 6-K currently has no topical indexing system. The structure of the actual filings thus hinders automated data collection (see, e.g., Lundholm et al., 2014).

10 For example, if there are 120 firms available for a certain country, we keep the largest 100 firms since 100 is greater than 30% of 120. However, if there are 400 available firms, we keep the largest 120 since 120 (i.e., 30% of 400) is greater than 100.

11 This negative value is due to two earnings press releases that report poor earnings numbers. To ensure that our results are not driven by outliers, we repeat our main analysis on cultural and institutional distance using (i) a robust linear regression and (ii) a subsample of firms located in countries for which we have at least 25 observations. The results (untabulated) are qualitatively similar.

12 For all regressions, we control for correlated errors, and we compute our standard errors within each country and year following Lundholm et al. (2014). This approach to compute standard errors is considerably more conservative than clustering by firm and year. In addition, if we cluster by industry and year, then our significance levels are at least as high as those reported in the tables. Finally, although we have directional hypotheses, we report p-values for two-tailed t-tests.

13 Due to the significant differences between the characteristics of U.S. and non-U.S. firms in our sample, we performed additional tests to ensure that our results are not driven by non-linearities in the relationship between the primary test variables and the control variables. Specifically, we assessed interactions of our test variables with each of the control variables displayed and repeated the key analysis. The results (untabulated) confirm our main findings.

14 We thank an anonymous reviewer for this suggestion.

15 An examination of these alternative explanations is beyond the scope of this paper, and we leave this question for future research.

16 Our main analysis uses Δ EARN as the measure for earnings news. Alternatively, we compute the forecast error as the difference between the achieved earnings per share (EPS) and the mean analysts’ EPS consensus prior to the issuance of the results, which substantially reduces the sample size, particularly for cross-listed firms. The results (untabulated) are similar to the main results.

17 People located in countries with weak future time references tend to perceive the future as something near that they would not be willing to jeopardize, whereas under strong language-time disassociation, the future would be perceived as something more distant. Firms located in countries with weaker future time references in language have been shown to engage less in myopic behaviour, such as accruals earnings management (Kim et al., 2017). By analogy, we conjecture that cross-listed firms from such countries communicate differently with regard to TONE and FLS in financial disclosure.

18 A potentially interesting avenue for future research could be to compare the disclosures of cross-listed firms targeted specifically to a home-country audience with disclosures targeted to the country of listing. This could allow tests of whether and how cross-listed firms adjust their communication.

19 Prior empirical research on reporting choices also shows that the Hofstede cultural dimension of Uncertainty Avoidance is negatively associated with earnings management, while Individualism is positively associated with earnings management (Han et al., 2010).