EU Regulation and open market share repurchases: new evidence

ABSTRACT

This paper re-examines the impact of the EU Market Abuse Directive (MAD) on the market reaction around share repurchase announcements. We use a unique hand-collected dataset of firms listed on the Athens Stock Exchange, and we find evidence that contrasts with previous conclusions for large European economies. The implementation of the MAD is followed by a significant increase in announcement abnormal returns, which is more pronounced in initial repurchase programs. Our results remain robust to a series of robustness tests. We attribute our findings to cross-country differences in institutional framework and pre-MAD existing national laws. Collectively, our results support the notion that EU directives do not have a uniform effect across Member States. Thus, the impact of such reforms should also be examined in individual capital market studies.

KEYWORDS:

• Share repurchases
• Market Abuse Directive
• safe harbor
• signaling undervaluation

JEL classifications:

• G14
• G15
• G35

Acknowledgements

The authors thank the participants at the Hellenic Finance and Accounting Association Conference (HFAA, 2015), the International Conference on Business and Economics - Hellenic Open University (ICBE-HOU, 2015), and the Financial Engineering and Banking Society National Conference (FEBS, 2014) for their helpful comments and suggestions. Funding by the Research Center of Athens University of Economic and Business (Projects EP-2254-01 and EP-2599-01) is gratefully acknowledged. Any remaining errors are our own.

Disclosure statement

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

Notes

1 A partial literature includes: Andriosopoulos and Lasfer (2015) for France, Germany and UK; González and González (2004) for Spain; Ginglinger and L’her (2006) for France; Ikenberry, Lakonishok, and Vermaelen (2000) for Canada; Kang et al. (2011) for Japan; Otchere and Ross (2002) for Australia; Rau and Vermaelen (2002) for the UK; Von Eije and Megginson (2008) for Europe.

2 In the literature there are many (and not necessarily mutually exclusive) motives for initiating SRPs, such as signaling undervaluation (Vermaelen 1981), distribution of free cash flows (Jensen 1986), dividend substitution (Grullon and Michaely 2002), capital structure adjustment (Dittmar 2000; Lie 2002; Bonaimé, Öztekin, and Warr 2014), option exercise in stock option plans (Fenn and Liang 2001; Kahle 2002; Bens et al. 2003), and takeover defense (Denis 1990).

3 Shareholder meeting approval is required in most developed European economies (Manconi, Peyer, and Vermaelen 2019).

4 See the Appendix of Siems and De Cesari (2012). The authors had sent a questionnaire to the HCMC regarding market manipulation regulations and safe harbor provisions for stock repurchases in Greece.

5 Regulation (EC) No 2273/2003, Article 3.

6 Repurchase programs must be authorized by the shareholders in a general meeting. The board of directors proposes the agenda topics, and publicizes them at least 20 days ahead of the meeting. Our data start in June 2000 when the Daily Official List of the Athens Stock Exchange became available on the Exchange’s website.

7 Stocks with fewer than 30 observations during the estimation period are excluded.

8 For repurchase announcements in 2000 and 2001, we have examined whether the repurchasing firms had announced any SRPs in the previous 24 months.

9 Our results remain similar even when we use the same classification of initial programs as before August 2007.

10 Similar results were obtained when we used the market model, the mean-adjusted return model, and the market model with the Scholes-Williams beta estimation method. The estimation period ranges from 200 to 21 days before the announcement date [−200, −21].

11 For robustness, we also use the standardized cross-sectional test and the generalized sign test. The results are qualitatively the same.

12 Results are similar if we use White (1980) heteroscedasticity robust standard errors.

13 In absolute values, the highest correlation coefficient is between Cash and Leverage (−0.39).

14 Low Tobin’s q × Cash is not included in the analysis, because it worsens the quality of matching between the two groups. The exclusion of this variable ensures that all mean differences in covariates between post-MAD and pre-MAD matches are insignificant.

15 Matching is done with replacement (yielded 143 matches in model 1 and 109 matches in model 2). Results are qualitatively similar if we match with no replacement, or if we use more than one matches.

16 We report ATTs on the announcement CARs over the [−1, +4] event window. Results are similar if we use the other three windows.

17 To identify the non-SRP firms, we exclude all firms from the initial sample of 615 SRPs.

18 We have also compared abnormal returns between initial and subsequent programs at the univariate level. We have done the comparisons for all the hypothetical sub-periods as well as the actual pre-MAD period. In all cases, mean differences across the two subsamples were insignificant.

Additional information

Funding

This work was supported by the Research Center of Athens University of Economic and Business [grant numbers EP-2254-01, EP-2599-01].