Earnings Management in Domestic and Foreign IPOs in the United States: Do Home Country Institutions Matter?

Abstract

Using a large sample of domestic and foreign IPOs in the US, we investigate how threats of enforcement by the Securities and Exchange Commission (SEC) and private litigation influence earnings management in IPO prospectuses. We propose that perceptions of foreign institutions may influence SEC enforcement action and private litigation. We provide evidence that enforcement and litigation threats are negatively related to the strength of legal institutions in the foreign IPO’s country of origin. We find earnings management is more pronounced in foreign IPOs from countries with strong legal institutions. We further explore whether earnings management is priced in the IPO market and find no relation between IPO proceeds and earnings management. Our results are consistent with upward earnings management as in Stein (1989), the magnitude of which is reduced when the anticipated cost of enforcement and litigation is higher. Collectively, our results cast doubt on the validity of the bonding hypothesis.

Keywords:

• Foreign IPOs
• bonding
• perceptions of foreign institutions
• SEC enforcement
• abnormal accruals

JEL Classifications:

• G18
• M41
• M43
• M44

Acknowledgements

We thank two anonymous reviewers and Thomas Jeanjean (the Editor), Dan Amiram, Mary Barth, Michael Dambra, Katerina Hellstrom, Amir Licht, Angela Pettinicchio, Matthew Pinnuck, Giulia Redigolo, Jay Ritter, Naomi Soderstrom, and Anne Wyatt. We also thank seminar participants at the BEROC Conference Minsk 2012, EAA Annual Meeting 2013, AAA Annual Meetings 2013 and 2015, Bocconi University, Cass Business School, ESSEC Paris, Griffith University, New Economic School, Oulu University, Stockholm School of Economics, Tsinghua University, University of Edinburgh Business School, University of Melbourne, University of New South Wales, University of Padua, University of Queensland, University of Southampton, University Technology Sydney, and Warwick Business School for their helpful comments on earlier versions of this paper.

ORCID

Igor Filatotchev http://orcid.org/0000-0003-4291-7768

Jonathan Jona http://orcid.org/0000-0003-4417-868X

Gilad Livne http://orcid.org/0000-0003-2270-6453

Notes

1 See Karolyi (2012) for a review of the evidence on the bonding hypothesis.

2 Our paper, however, does not provide evidence pertaining to the importance of contracting in an IPO’s home country.

3 We elaborate on the role of perceptions in Section 2.3.

4 Perhaps the most relevant section of the 1933 Act is Section 11. It stipulates that damages can arise when an investor relies on a prospectus and the award to the successful plaintiff is larger when the difference between the offer price and the sell price is greater, or the price at the time of the lawsuit. Lowry and Shu (2002) found the incidence of litigation against IPOs stands at about 6% and that most of the suits were brought under this section.

5 Ndubizu (2007) found foreign IPOs manage earnings more than mature US firms. Lee and Masulis (2011) found IPOs select underwriters to reduce information asymmetry, which, in turn, reduces earnings management. However, both papers use post-IPO quarterly accruals. We control for underwriters’ reputation in our robustness tests (see Section 5.2).

6 This regulatory ‘lighter touch’ is likely motivated by the desire to attract foreign listings to the US.

7 Consistent with this, Cheng et al. (2014) reported that foreign firms are sued at a rate of 2.16% whereas in a sample of matched US firms the rate is 4.10%.

8 Research in accounting has largely overlooked the role of perceptions in the IPO market. An exception to this is a recent study by Blankespoor, Hendricks, & Miller (2017) that used an experimental research design to investigate the effect of perceptions on IPO pricing.

9 We provide evidence on the threat of litigation and enforcement in Table 4.

10 Although the SEC vets IPO documents prior to listing, the SEC audits financial statements only periodically following an IPO.

11 We apply a similar procedure throughout the paper.

12 Due to the limited sample size, and in line with Peek, Meuwissen, Moers, & Vanstraelen (2013), we combine several years for each industry. Specifically, we use three consecutive years as a single period, as follows: 1990–1994, 1995–1998, 1999–2001, 2002–2005, and 2006–2009. Model 1 is estimated to require at least ten observations.

13 Silvers (2016) did not report a measure of enforcement of US firms. Hence, we run Model 3a on the foreign IPOs subsample. In using Silvers’ (2016) Table 2, we implicitly assume that SEC enforcement of mature foreign firms is similar to its enforcement of foreign IPOs. If this assumption is violated, then SEC_ENF is measured with noise and we are less likely to establish significant relations.

14 We also assume that private litigations rates reported in Cheng et al. (2014) of mature firms are similar for IPOs.

15 The industries that are more prone to US litigation are identified in Francis, Philbrick, & Schipper (1994).

16 We use the actual proceeds because we do not have a measure for target proceeds. Additionally, the results remain the same if instead of the offer price we use is first-trading-day price.

17 If Model 1 is correctly specified, CFO is orthogonal to abnormal accruals (the error term in (1)). However, if Model 1 suffers from an omitted variable problem, ABNACC and CFO may be correlated. It is plausible that without CFO, Eq. (3) may suffer from the same problem. Hence, including CFO in (3) helps to mitigate the omitted correlated variable problem.

18 We confirmed that the SEC defines foreign IPOs in our sample as foreign registrants (see https://www.sec.gov/divisions/corpfin/internatl/companies.shtml).

19 These periods were also used in estimating Model 1.

20 In our sample there are only nine IPOs that report under IFRS. Nevertheless, in Section 5.5 we investigate whether IFRS adoption in the IPO’s home country affects the results.

21 Retrieved from http://www.imf.org/external/data.htm#data on April 2011.

22 In the remainder of the paper, ABNACC is measured with respect to the IPO sample.

23 To address the differences between the samples, we run robustness tests (unreported) that restrict the domestic US-firm sample to observations with LTA and LPROC of at least the minimum values of these variables in the foreign sample. These did not yield any material differences in our findings.

24 For PLIT we have fewer observations as we do not have data for all countries (mainly Canada).

25 We also verified that these results are robust to self-selection. We discuss self-selection in Section 5.1.

26 A similar approach has been used in the cross-listing literature (e.g., Doidge, Karolyi, & Stulz, 2004, 2010; Loureiro, 2010).

27 We use the country enforcement index as measured and presented in Brown, Preiato, & Tarca (2014) by the country of origin of the foreign IPO and the IPO year. This index includes the US as well as our IPO foreign countries. This measure is excluded from the other models, as we do not expect enforcement in the country of origin to affect earnings management in US filings.

28 We report pairwise correlations of HOME_ENF with other variables used in this study in Table 3. Notably, the correlation of HOME_ENF with FOREIGN is highly negative (at 0.60), since enforcement is strongest in the US. The correlation with INST is 0.40, as would be expected. In addition, the correlation with SOX is 0.60 reflecting increasing enforcement scores over time as documented in Brown et al. (2014).

29 Insider ownership is similar in foreign IPOs from countries with strong home institutions and in foreign IPOs from countries with weak home institutions.

30 http://bear.warrington.ufl.edu/ritter/ipodata.htm

31 In Column 6 of Panel B of Table 4 (the subsample of foreign IPOs from weak institutions countries) PLIT turns insignificant once we control for VC backing. However, PLIT remains negative in the entire foreign sample.

32 This multi-listing involves 19 foreign countries, with the UK leading the other markets (15 observations).

33 These findings hold also when we drop FOREIGN and INST from the model specification and when we control for self-selection in Column 1.

34 A further analysis of the pricing implications of FOREIGN is beyond the scope of this paper.