Press Release Management around Accelerated Share Repurchases

ABSTRACT

Accelerated share repurchases (ASRs) have gained popularity in recent years. At the inception of an ASR, a firm receives most of the shares that would be repurchased at the current stock price. At the completion of the ASR, the repurchase price is adjusted to the average stock price during the ASR contract period. I find that managers increase the media coverage of negative press releases before the inception date but not during the contract period. While prior research has documented that managers deflate stock prices before open market repurchase announcements to reduce subsequent repurchase prices, this paper suggests that, in ASRs, managers rank accelerating share delivery higher than reducing the repurchase price.

Keywords:

• Accelerated share repurchases
• Press release management
• Media coverage
• Negative tone

Acknowledgements

I thank Reuven Lehavy and Florin Vasvari (guest editors), two anonymous referees, Christine Wiedman, Neil Brisley, Changling Chen, Darren Henderson, Gordon Richardson, Mikal Skuterud, and Tony Wirjanto for helpful comments and suggestions. A research grant from the Office of Research Services at Wilfrid Laurier University is gratefully acknowledged.

Notes

1 I read through available ASR contracts and do not identify any with a floor only.

2 There is a legislative reason for the slow buyback in OMR-only repurchase programs. A firm usually conducts OMRs in a manner that avails itself of the Rule 10b-18’s safe harbor protection (Atkins 2013). This rule governs the manner, timing, price, and volume of repurchases. Specifically, the aggregate repurchase on any given day cannot exceed 25% of the average daily trading volume during the preceding four weeks or only 0.12% of outstanding shares based on the World Bank’s data.

3 The OMR-only repurchase programs in Brockman et al.’s (2008) sample last 510 days on average from announcement to completion. Nain and Vijh (2016) raise the question of whether bad news management forecasts during 30 days before announcements can suppress stock prices for that long. In contrast, Gong et al. (2008) suffer less from the concern by using a sample of carry-through repurchase programs, that is, a program followed by actual buyback during the announcement quarter and/or the subsequent quarter.

4 The keywords used in the full-text search include (accelerated share repurchase) or (accelerated stock repurchase) or (accelerated share buyback) or (accelerated stock buyback) or (overnight share repurchase) or (overnight stock repurchase) or (overnight share buyback) or (overnight stock buyback) or (forward repurchase transaction) or (discounted share repurchase) or (discounted share buyback) or (discounted stock repurchase) or (discounted stock buyback).

5 Bargeron et al. (2011) attribute the decline in ASRs during 2008 and 2009 to the increased stock market volatility during the financial crisis, because the costs of lost flexibility in ASRs increase with the volatility of repurchasing firms’ stock prices.

6 Factiva provides a list of covered firms for each news article. I randomly select 100 articles to check whether they are firm-generated. I find that, if an article covers more than five firms, it is usually a note or alert issued by an analyst or a credit rating agency. I also find that an article is usually issued by either of the first two covered firms.

7 If I measure the net tone, the coefficients on PRE1 in Table 5 Panel B will not be significantly positive on a two-tail test (marginally significant on a one-tail test though), suggesting that my main results are sensitive to the tone measurement.

8 Although Henry’s (2008) word list addresses financial text too, it is developed from earnings press releases, a type of press release that accounts for only 10.2% of firm-generated press releases in my sample. Loughran and McDonald’s (2011) word list is developed from 10-Ks, a type of text close to firm-generated press releases in that both cover both financial and nonfinancial information. Thus Loughran & McDonald’s list is a better fit to this study. In addition, their list (354 positive and 2,355 negative words) contains more words than Henry’s list (105 positive and 85 negative words). In fact, 48 out of 85 negative words in Henry’s list are also included in Loughran & McDonald’s list. In a robustness check, I use a combination of the two lists to fit all types of press releases. All results in Section 4 and Section 5 remain qualitatively unchanged.

9 The average negative tone of my press release sample is lower than that of Davis and Tama-Sweet’s (2012) earnings announcement sample (1.19%). But those authors also note that the negative tone varies with disclosure outlets, due to the difference in the timing, form, function, and visibility of alternative outlets.

10 I use analyst consensus forecast, rather than reported earnings, because there is insufficient variation in reported earnings for firm-specific regressions using weekly data.

11 Only 86 out of my sample of 467 ASRs announce a motivation. The language they use is boilerplate in most cases (e.g., ‘create shareholder value’) and may not be accurate. One may argue that investors would be able to use the same method in this study to infer motivations of ASRs. However, investors may use different data and methods or make divergent judgements. For example, investors may not reach a consensus view on whether a firm is undervalued since this is a subjective judgement.

12 I search for takeover rumors using Capital IQ’s ‘key development–potential transactions and M&A rumors and discussions’ function.

13 The area under the ROC curve measures the predictive ability of a probit regression model. It ranges from 0.5, corresponding to random guessing, to one, corresponding to perfect prediction. Hosmer et al. (2013) suggest that a model with an area under the ROC curve of above 0.8 is excellent.

14 Because an ASR contract constitutes a material definitive agreement, an ASR firm must file an 8–K, issue a press release, or both to disclose the ASR within four business days after it enters into the contract. In fact, the mean (median) days between the ASR announcement date and the ASR inception date are 0.14 (0) days in my sample.

15 The results remain qualitatively unchanged if I use CAR estimated from the standard market model over the period from trading day [−252, −44] before the ASR inception date.

16 When calculating BHAR, I exclude trading day [0, +2] to remove the stock price run-up around ASR announcements.

17 In comparison, Bargeron et al. (2011) report an average three-day announcement return of 1.42% for ASRs from 2004 through 2008. Bonaimé (2012) reports a five-day announcement return of 1.93% for OMR-only repurchase programs from 1988 through 2007, with a lower announcement return of 1.19% in the more recent years of 2004 to 2007.

18 Tetlock’s (2007) finding suggests that the newspaper column contains no relevant information about fundamentals. Similarly, I also find that pre-ASR press release management is not associated with the improvement in operating performance over one year or two years following the ASR inception date.

19 Specifically, the coefficient on PRE1 is significantly positive (0.194 with a p-value of 0.039), and the coefficients on DUR and POST are not significantly different from zero.

Additional information

Funding

This work was supported by a research grant from the Office of Research Services at Wilfrid Laurier University.