Using accelerated share repurchases to meet or beat earnings expectations

Using accelerated share repurchases to meet or beat earnings expectations Nurul A Rafi and Clifford Stephens * January 15, 2018 Abstract We investigate the use of stock buybacks, specifically accelerated share repurchases, to meet or beat the analysts forecasts of EPS. We find that when the companies want to repurchase their stocks in an effort to meet or beat the EPS expectations, they are more likely to choose accelerated share repurchases (ASRs) rather than open market repurchases (OMRs). Interestingly, upon announcements, these EPS motivated ASRs experience strongest market reaction among all the ASRs. We also find evidence of a significant discount on the stock price of the ASR announcing firms that miss the analysts consensus EPS forecasts in the repurchasing quarters. we hypothesize that the habitual beaters among the repurchasing firms and firms whose earnings are consistently assessed by the investors to be highly informative, as measured by earnings response coefficients, are more likely to conduct EPS motivated ASRs. We also find results in support of signaling undervaluation hypothesis for ASRs. * Nurul Rafi is a Ph.D. candidate at Louisiana State University and Clifford Stephens is an instructor at Louisiana State University, Department of Finance. Csteph5@lsu.edu. We would like to thank Murali Jagganathan for work and thoughts on a much earlier version. 0

Introduction Accelerated share repurchases (ASRs) are a relatively new type of derivative share repurchase that is becoming increasingly more common. We analyze the recent proliferation and use of ASRs; specifically, we examine whether firms manage earnings using ASRs to meet or beat analysts earnings (EPS) forecasts. As the name implies, in an accelerated repurchase the firm repurchases the shares quickly, usually immediately. The firm enters an agreement with an investment bank where the investment bank sells the shares to the firm immediately, the investment bank takes a short position in the firm s stock and subsequently purchases shares in the open-market to unwind their position. Although ASRs are generally preceded by open-market repurchase program announcements and are frequently employed as part of the open-market repurchase program, ASRs differ from traditional open-market repurchases in several important ways. Firstly, and most importantly, the firm benefits from an immediate reduction in shares outstanding. Whereas in a traditional openmarket repurchase program the share are repurchased gradually over a period of months or perhaps years (Stephens and Weisbach, 1997), in an ASR the shares are repurchased, and frequently retired, immediately. Clearly, this immediate reduction in shares outstanding has more dramatic effect on earnings per share (EPS). The other differences between the two methods are more subtle. There is a potential off-balance-sheet liability associated with settlement of the agreement with the investment bank. To the extent that the firm s share price increased over the life of the agreement the firm will owe the investment bank the difference between the agreement price and the volumeweighted average share price. Finally, the repurchases made by the investment bank are not subject to the safe harbor provisions of SEC rule 10b-18 and the tax implications of these types of deals are indeterminate. A more detailed description and a numerical example are provided later to further aid the discussion. 1. ASR description, accounting for ASRs and their effect on EPS In this section we describe ASRs, how they work and their differences from open-market repurchases. We also provide a simple example to demonstrate the effect of ASRs on earnings per share (EPS). 1.1. Description of ASRs 1

An ASR is a type of open-market repurchase where the firm pays for the shares upfront and receives immediate benefit from the decrease in shares outstanding. Essentially, the firm enters an agreement with an investment bank where the investment bank sells the shares to the firm immediately (usually at the prevailing market price). The investment bank takes a short position in the firm s stock and purchases shares in the open-market to cover their position over the next 30 to 360 days. At the expiration of the agreement the firm settles with the investment bank based on the volume-weighted average price the investment bank paid for the shares. Generally, settlement can take place either in firm shares or in cash; if the stock price has increased the firm owes the investment bank the difference between the agreed price and the average price paid by the investment bank multiplied by the number of shares purchased. (Insert quote here from the WSJ about potential liability). Although non-trivial, we do not believe the potential liability posed by these transactions to be a material (significant) issue. However, the ASR could have a significant effect on the firm s earnings per share (EPS). A typical ASR announcement is provided in the appendix. 1.2. Accounting for ASRs The financial reporting requirements for ASRs are slightly different from other types of share repurchases. Pagach and Branson (2007) provide a detailed summary of accounting for accelerated stock repurchases. Essentially, the ASR is treated as two separate transactions: a repurchase of Treasury stock and a forward contract.1 Although the firms repurchases and pays for the shares immediately, a forward position arises as a result of agreement to pay investment bank the average price of their actual stock purchases made to cover the short position. Any changes in the market value of the forward contract are recorded as adjustments to owner s equity in the balance sheet rather than on the income statement.2 Although the ASR does not impact net income, it has a potentially dramatic effect on EPS.3 1 The Financial Accounting Standards Board s (FASB) Emerging Issues Task Force (EITF) Issue 99-7 addresses the prescribed accounting treatment for accelerated share repurchases. 2 EITF Issue 00-19 provides the prescribed accounting treatment for derivative contracts indexed or settled in a company s own stock. 3 It is not clear if there are any tax implications of structuring the repurchase as an ASR. Generally, gains and losses on forward, and other derivative contracts, are taxable events; however, since the forward contract is indexed to the company s own stock, ASRs may be non-taxable equity transactions. We believe any contract settled in stock would clearly be treated this way, but if the contract is settled in cash or must be settled in cash there may be a taxable event. We am still looking into this. 2

1.3. ASRs and EPS By reducing the number of shares outstanding all share repurchases increase EPS, ceteris paribus. However, the immediate reduction is shares outstanding resulting from the ASR has a much more significant effect on EPS than the gradual reduction in shares outstanding resulting from open market repurchases. Pagach and Branson (2007) provide a numerical example of the affect ASRs on diluted EPS; we provide a simplified adaptation of that example below. Assume that as of December 31 (fiscal year end) R&S Enterprises has 5 million common shares outstanding. R&S then enters an ASR agreement with their investment bank to repurchase 1 million shares at the current market price of $50 per share; R&S pays the investment bank $50 million, retires the 1 million shares and the investment bank takes a short position of 1 million shares to be unwound over the next 6 months. The following additional assumptions are made in the subsequent EPS calculations: The investment bank purchases shares evenly over the six month life of the agreement. If R&S utilized an open-market repurchase instead of the ASR, these repurchases would also be made evenly over the 6-month period. R&S has net income of $5 million for the current and each of the next four quarters. R&S s stock price is $60 and $75 per share, respectively, at the end of the two quarters immediately following the ASR (March 31 and June 30). The forward contract between the firm and investment bank is settled in cash at expiration. If no share repurchases are made the EPS is $1 ($5,000,000 / 5,000,000 shares). If the firm chooses to employ an ASR, the average number of shares outstanding is 4.5 million shares in the current quarter and 4.0 million shares in every subsequent quarter (ignoring the dilutive effect of the ASR agreement). Consequently, EPS is $1.11 in the current quarter and $1.25 in the subsequent quarters. Alternatively, the firm could choose to repurchase the shares in the open-market rather than through an ASR. In this case, the average number of shares outstanding is 5.0 million in the current quarter, 4.75 million in the first quarter subsequent to beginning the repurchase program, 4.25 million in the second quarter and 4.0 million shares in the final two quarters after completing the repurchase program. EPS would then be $1.00 in the current quarter, $1.05 in the following quarter, $1.18 in the next quarter and $1.25 in the final two quarters. 3

Ceteris paribus, both types of repurchases have a positive impact on EPS; however, the ASR clearly has a more dramatic effect on EPS, particularly in the periods immediately following the ASR. The immediate decrease in the number of shares outstanding resulting from the ASR has a concurrent, but muted (due to the averaging of the number of shares outstanding) on EPS. The full impact of the ASR on EPS is not observed until the period immediately following the ASR; however, the full impact of analogous open-market repurchase would not be observed for two more quarters. If the agreement between the firm and the investment bank allows the forward contract to be settled in the firm s stock, then the contract is likely to be dilutive. If the stock price has decreased since the inception of the ASR, then the investment bank will owe the firm cash or shares at expiration of the agreement and the forward contract is not dilutive. However, if the stock price has increased since the inception of the ASR, as is likely to be the case, then the forward contract is dilutive. The number of shares outstanding will have to be adjusted to reflect the dilutive effects of the ASR in calculating fully-diluted EPS. In our example, further assume that R&S s stock price is $60 and $75 per share, respectively, at the end of the two quarters immediately following the ASR (March 31 and June 30). We can then calculate the number of contingent shares associated with the forward contract and the firm s diluted earnings per share. There are 83,333 contingent shares associated with the forward contract at the end of the first quarter following the ASR. 1.0 million shares x ($55 average market price - $50 ASR price) = $5.0 divided by the current market price of $60 per share. During the second quarter the average purchase price is $67.50 per share and the volumeweighted average purchase price over the first two quarters is $61.25 per share. There are 150,000 contingent shares associated with the forward contract at the end of the second quarter. 1.0 million shares x ($61.25 average market price - $50 ASR price) = $11.25 million dividend by the current market price of $75 per share. Consequently, for purposes of calculating EPS, the firm has 5.0 shares outstanding at the end of the current quarter, 4.083 million shares at the end of the fourth quarter and 4.15 million shares at the end of the second and subsequent quarters (since the firm will issue 150,000 shares to the investment bank at expiration of the agreement). The resulting EPS figures are $1, $1.22 and 4

$1.20 for the current quarter, first quarter following the ASR and all subsequent quarters, respectively. The ASR has a dramatic effect on EPS even after accounting for the dilution associated with the forward contract. Although all share repurchases are likely to increase EPS, an ASR is a private, and not particularly well disclosed, transaction that results in an immediate and significant decrease in shares outstanding. In many instances the ASR does not even need to be formally approved by the board; many of these firms have ongoing open-market repurchase programs whose authorization generally also allow for privately negotiated repurchases, such as ASRs or block repurchases. Consequently, an ASR may be a particularly attractive tool for manipulating EPS. The purpose of our study is document and describe the recent proliferation of the ASR use, examine the market s reaction to ASRs and associated events and provide some evidence of whether ASRs are used to manipulate EPS. Meeting or beating analysts expectation of earnings has been a well-documented motive among the corporate managers. Kasznik and McNichols (2002) find evidence of a value premium for the firms which consistently meet or beat earnings expectation (i.e., habitual beaters). Bartov, Givoly and Hayn (2002) find evidence of a value premium for the firms that meet or beat analysts estimates, irrespective of the firms prior forecast beating performance, and even if the meeting or beating of earnings expectation is managed. In a survey of more than 400 financial executives, Graham, Harvey and Rajgopal (2005) find that 73.5% of the CFOs surveyed agree or strongly agree that analyst consensus estimate of earnings per share (EPS) is an important benchmark for them in a quarter. 83.6% of the surveyed participants believe that meeting benchmarks increase credibility with the capital market. More than 80% agree that meeting benchmarks helps maintain or increase the firm s stock price. Contrary to the economic theory, Graham, Harvey and Rajgopal (2005) find that mangers are willing to make sacrifices in value in order to meet or beat the earnings expectations of analysts. For example, 78% of the surveyed executives indicated that they have strong preferences for smooth earnings over volatile earnings, and as a result would give up economic value in exchange for smooth earnings. In other words, managers seem to decide on a trade-off between delivering earnings for the capital market incentive in the short-term and making profitable investmentdecision for value-maximizing in the long-term. 5

Extant academic literature that studies this type of myopic behavior (Stein 1989) examines the use of earnings and expectation management techniques by the corporate managers before the earnings announcement. Few studies have looked at the use of payout policies by the corporate managers to meet or beat analysts expectation of EPS. Share repurchases have been integral parts of corporate payout policies. In addition to fulfill other common motives like signaling undervaluation, reducing agency problem of free cash flow, restructuring the capital structure, repurchases of stocks can be used by managers to boost up the EPS. Stock repurchases decrease the number of shares outstanding by the number of shares repurchased and consequently, increase EPS. Almeida, Fos and Kronlund (2016) find that the probability of share repurchases that increase EPS is sharply higher for firms that would have just missed the EPS forecast in the absence of repurchase, when compared with the firms that just beat the EPS forecast. The immediate and committed nature of ASR gives an even stronger incentive to the repurchasing firms, which want to use repurchases to meet or beat EPS. Using a matched sample of OMRs for the ASRs announced between January 2004 and June 2015, we find that repurchasing firms that seem to repurchase their shares with a motive to meet or beat the analysts consensus forecast of EPS in a quarter are more likely to conduct ASRs than OMRs. We also find that firms which consistently meet or beat analysts EPS forecasts and firms whose earnings are consistently assessed by the investors to be highly informative, as measured by earnings response coefficients, are more likely to use ASRs as an EPS management mechanism (i.e. a tool to meet or beat analysts consensus forecast of EPS). In addition to the two main results, we investigate the factors influencing the decision of the repurchasing firms to undertake ASRs. We find that election of ASRs over OMRs is significantly negatively associated with illiquidity, leverage and significantly positively associated with preannouncement stock performance. ASRs have important implications not only to managers, but also to market participants. We examine the abnormal returns at the announcements of ASRs and the matched OMRs and find that ASRs have significantly higher abnormal return both in short-term and long term. More interestingly, we find that the ASRs which seem to be conducted to meet or beat the analysts consensus forecasts of EPS experience the highest abnormal return of all at the announcement of the repurchase programs. We find that 3-day announcement return of an ASR is significantly positively associated with the repurchasing firm s Q ratio and is significantly negatively associated 6

with the repurchasing firm s past return and Cash holdings. These results along with the finding of higher abnormal returns for ASRs in the long run imply that signaling undervaluation might be an important motive for announcing ASRs. We also find that, investors discount the stock price of the ASR announcing firms which miss the analysts consensus forecasts of EPS in the quarters of repurchase and the discount on the 3-day announcement return is 2.1%. This paper contributes both to the repurchase literature and earnings management literature in a number of ways. First, very few studies have examined ASRs, which is surprising because ASRs have been used by companies in increasing frequency in recent years. As Bargeron, Kulchania and Thomas (2011) documented, over the period 2004-2008, the frequency of ASRs exceeded that of privately negotiated repurchases, fixed price self-tender offers, Dutch-auction self-tender offers and large special dividends. The number of ASRs in more recent years have only increased. By discussing the common determinants of and the market reaction around repurchases, this paper examines how companies use ASRs differently than OMRs. Second, the most important contribution of this paper comes from the fact that this paper recognizes and investigates the use of ASRs as a tool for meeting or beating expected EPS. And, finally, this paper investigates why some companies might use ASRs as an EPS management tool even though ASRs are a relatively costly form of repurchases. Rest of the paper proceeds as follows. In sections 2, we review the related literature and develop our testable hypotheses. We describe my sample construction in section 3 and report the results of my empirical analysis in section 4. We conclude in section 5. 2. Literature review and hypotheses development Open-market share repurchases are the most common method repurchase method representing about 99% Of all repurchase programs and about 95% of the dollar amount of repurchases. 4 However, accelerated share repurchases (ASRs) have become increasingly popular (Cook and Kim, 2006). There has been a fair amount of empirical research on share repurchases and payout policy in general; however, little is known about the use of accelerated stock repurchases. Early research by Dann (1981) and Vermaelen (1981) observed that share repurchases are generally received positively by the market. This research focused on tender offer repurchases, relatively common at the time, 4 See Grullon and Michaely (2004) 7

but also reported that abnormal returns in the neighborhood of 3% are observed in the days surrounding the announcement of open-market share repurchases. Comment and Jarrell (1991) observed that the announcement of open-market repurchase programs tend to be preceded by a recent decline in stock price. They report that the abnormal returns observed around the announcement of an open-market repurchase program are correlated to the announced size of the repurchase program and the amount of the preceding stock price decline. The observed abnormal returns around the announcement of tender-offer repurchases (both Dutch auction and fixed price tender offers) are also correlated with the size of the repurchase, but there is not an analogous price decline preceding the tender offer. Both the abnormal returns and program size are substantially larger (about 4 times) than that of open-market repurchases. Unlike tender offer repurchases, open-market repurchase announcements are merely a statement of intent to repurchase 5 and do not obligate the firm to repurchase any shares. However, the vast majority of firms actually repurchase a significant number of shares, frequently more than the initial announcement. Stephens and Weisbach (1997) not only report that most firms follow through with their announced intentions to repurchase stock, but generally do so quickly. Most of the repurchasing activity takes place within several quarters of the announcement and actual share repurchases are correlated with cash flows and market conditions (less repurchases following increases in the firms share price). Although it is well documented that the market generally greets share repurchases positively, the source of the observed abnormal returns and the long-run performance of repurchasing firms are more ambiguous. Ikenberry, Lakonishok and Vermaelen (1995) report that repurchasing firms outperform their peers over several years following the repurchase announcement and that value (low M/B) firms that announce repurchase programs outperform their peers by 44% over the subsequent four years. However, their does not appear to be any corresponding improvements in operating performance. Grullon and Michaely (2003) and Jagannathan and Stephens (2003) report observing no increases in ROA or operating cash flows following the announcement of an openmarket repurchase program. 5 All of the major U.S. exchanges require that the firm publicly announce their intent to repurchase shares on the openmarket. SEC rule 10b-18 does not specifically require an announcement but does provide safe harbor provisions which impose the following restrictions:. 8

In terms of payout policy, share repurchases, specifically open-market share repurchases, appear to be a partial substitute for dividends or dividend increases. Fama and French (2001) show that the propensity to pay dividends has decreased over time and that share repurchases account for some of this decline. Grullon and Michaely (2002) suggest that repurchases are a substitute, but imperfect substitute, for dividends. Similarly, DeAngelo, DeAngelo and Skinner (2000) conclude that share repurchases are only partially responsible for the decline in special dividends. As suggested as far back as Litner (1956) dividends appear to be paid out of permanent earnings. Jagannathan, Stephens and Weisbach (2000) report that dividends end to be utilized by firms with high, relatively stable operating income, whereas repurchases are utilized by firms with more volatile operating income and high non-operating income. Rather than commit the firm to a higher dividend level, the firm preserves flexibility by disgorging excess temporary cash flows in the form of share repurchases. Similar to a privately negotiated or block repurchase, an ASR involves the immediate repurchase of a large number of shares. Unlike other types of repurchases, the returns observed around the announcement of block repurchases are not statistically significant (Peyer and Vermaelen, 2005). Jagannathan and Stephens (2007) find positive abnormal returns, on average, around the announcement of a block repurchase, but the direction and magnitude of these returns varies by the type of block holders. Repurchases from insider or affiliated block holders are generally received more favorably than repurchases from outside block holders. However, unlike a block repurchase, an ASR does not remove or reduce the influence of a large block holder with potential monitoring abilities. ASRs and derivative repurchases in general are a relatively new phenomena. As reported by Cook and Kim (2006), in February 1991 the SEC effectively began allowing firms to write put options on their own stock providing firms a new avenue for repurchasing shares. Although not exactly common there were 264 instances of firms writing put options on their own shares between 1991 and 2001. Atanasov, Gyoshev, Szewczyk and Tsetsekos (2004) find that the vast majority of these put options go unexercised and Cook and Kim note that this practice largely disappeared in the early 2000s and was replaced by the use of ASRs. ASRs can be viewed as an expedited and more credible form of OMR. This immediacy and higher credibility in ASR come with a loss of flexibility inherent in OMR. Bargeron, Kulchania, and Thomas (2011) describe firms undertaking of ASRs using flexibility hypothesis and immediacy and 9

credibility hypothesis. According to the flexibility hypothesis, the choice to undertake an ASR in a repurchase program represents the early exercise of the flexibility option inherent in OMR (e.g., Ikenberry and Vermaelen (1996) and Oded (2005)). Therefore, factors affecting the costs of early exercise of the flexibility option inherent in an OMR should also affect the firm s choice to conduct an ASR. This hypothesis predicts that all else equal, the costs of an ASR are greater for firms with greater stock price uncertainty, less stock market liquidity of its shares, less predictable cash flows, greater imbalance between cash flow and investment needs, and smaller program authorizations. The credibility and immediacy hypothesis predicts that the enhanced credibility and immediacy of ASRs incentivizes the repurchasing firms in signaling their willingness to reduce the agency problem of free cash flow by distributing to the shareholders in an expedited manner, signaling undervaluation (e.g., Chemannur, Cheng and Zhang (2010)), defending against unwanted takeover attempts (e.g., Akyol, Kim, and Shekhar, 2009), increasing leverage towards a target ratio ( e.g., Hovakimian, Opler, and Titman (2001)), or even managing reported earnings per share with a motive for higher executive compensation (e.g., Cheng, Harford, and Zhang(2010) and Marquardt, Tan and Young (2011)). The agency problem of free cash flow (Jensen (1986)) arises when firms have a large amount of cash in hand but don t have enough investment opportunities to act on. When firms have excess cash in hand compared to the investment need, they can distribute this excess cash to the shareholders in the form of repurchases (Jagannathan, Stephens, and Weisbach (2000); Grullon and Michaely (2004)). If ASRs are conducted with a motive to reduce the agency problem of free cash flow, the market reaction to repurchases should be negatively associated with Q (a proxy for firm s investment opportunities) and positively associated with cash holdings (measured by Cash/Assets ratio). On the other hand, the signaling undervaluation hypothesis (Comment and Jarrell (1991), Asquith and Mullins (1986), Ofer and Thakor (1987), Constantinides and Grundy (1989)) predicts that firms may purchase its own stocks from the investors, when they think their stocks are relatively underpriced. Thus, according to this hypothesis, the market reaction to repurchases should be negatively associated with preannouncement stock returns of the repurchasing firms. There has been a considerable amount of research done to study earnings management around corporate events like initial public offerings (IPOs), seasoned equity offerings (SEOs) and repurchases, but little has been done to discuss the use of these financial policies by firms to meet or 10

beat the analysts forecast of earnings. Marquardt, Tan, Young (2011) find that firms are more likely to choose ASRs over OMRs when the repurchase is accretive to EPS, when annual bonus compensation is explicitly tied to EPS performance, when CEO horizons are short, and when CEOs are more entrenched. In a study on bonus-driven repurchases, Cheng, Harford, and Zhang (2010) find that a firm is more likely to conduct a buyback and the magnitude of the buyback tends to be larger when the firm s CEO s bonus is directly tied to EPS. In most of the earnings management articles, earnings management around important corporate events is hypothesized to be driven by either market response or managerial opportunism. Market response hypothesis, often referred to as managerial incentive hypothesis, predicts that firms manage their earnings before corporate events (announcements of SEO or repurchase) as a market response to the unfavorable future price movement at the events occurrence. Shivakumar (2000) finds that that mangers of SEO announcing firms, in anticipation of a decline in price at the SEO announcements, tend to overinflate the firms earnings before the announcements to increase the price of the firm s stocks artificially. On the other hand, managerial opportunism hypothesis says that managers seek to mislead the investors by overstating (understating) the earnings before an SEO (repurchase) program (Rangan (1998), Teoh et al. (1998), Chiu Liang (2015)). Unlike market response hypothesis, this hypothesis predict that managers do not anticipate an unfavorable price movement at the announcement of SEOs or repurchases driven by the investors assessment of earnings management. Market response hypothesis does not make a case for ASRs, since ASRs, like other repurchases, are usually met with positive announcement returns. Managerial opportunism hypothesis, on the other hand, seems to make a strong case for ASRs, like OMR repurchases. In fact, ASR, with its immediate and committed nature provides even stronger incentives to the repurchasing firms for understating the earnings before repurchase announcements. Gong, Louis, Sun (2008) find evidence of pre-announcement earnings understatement by the OMR announcing firms. Additionally, they find pre-announcement understatement of earnings is a determinant of post-announcement superior return as well as post announcement improvement in operating performance of the repurchasing firm. Following their work, Chiu, Liang (2015) look for the existence of pre-announcement earnings understatement by the ASR announcing firms. Contrary to reasonable belief, they find no 11

evidence of pre-announcement earnings understatement by ASR announcing firms. Both these studies, like other on earnings management, examine earnings management by the repurchasing firms before repurchases via accrual management techniques. This study, on the hand, investigates the use of ASRs to meet or beat analysts EPS forecasts by the repurchasing firms. Though this is not an example of earnings management per se, this seems to be an example of EPS management if one considers the motive and the effect of these repurchases. We hypothesize that ASRs are widely used by repurchasing firms to meet or beat analysts consensus EPS forecasts. More specifically, we hypothesize that EPS-motivated repurchases are more likely to be ASRs than OMRs. This hypothesis is in line with the managerial opportunism hypothesis in a sense that EPS motivated repurchasing firms announce ASRs with an anticipation that the investors won t be able to predict the EPS management motivation of the repurchasing firms at the time of announcements. Managers of the repurchasing firms act opportunistically by conducting ASRs to meet or beat analysts EPS forecasts. From a broad perspective, this hypothesis seems to be consistent with the idea of myopic behavior of corporate managers induced by the capital market pressure presented by Stein (1989). Mangers of the repurchasing firms, induced by the capital market pressure, seem to act myopically to boost EPS. But, the increase in value through EPS motivated ASRs rarely come at a cost of future losses. ASRs are too costly to be conducted unless the benefits from immediacy and credibility of ASRs are higher than flexibility lost from not using OMRs. Also, the investors will easily find out any inherent EPS management motivation in the repurchases at the earnings announcement of that quarter and, discount the stock price accordingly. ASR conducting firms, even with genuine motives of signaling information to the investors could possibly face investors backlash if their meeting or beating the EPS forecasts in a quarter seem to be driven by the conducted ASRs in that quarter. This possibility of investors backlash is lower for the firms which have strong track record of meeting or beating analysts EPS forecasts and whose earnings are assessed by the market to be highly informative, evident from high earnings response coefficients (ERC). High ERCs also incentivizes the firms to meet or beat EPS forecasts since their returns at the earnings announcements are highly associated with their unexpected portion of earnings. Given these consequences, we hypothesize that firms which consistently meet or beat analysts EPS forecasts and firms whose earnings are consistently assessed by the investors to be highly informative are more likely to conduct ASRs to meet or beat EPS forecasts. 12

3. Sample construction My sample includes ASRs announced during the period starting from January 2004 and ending in June, 2015. ASRs are from the repurchase database of Thomson Reuters. This database does not contain all ASRs, so to include any missing ASR, we search the mergers and acquisitions database of Thomson Reuters using a word search technique in the synopsis of the reported mergers and acquisitions. In total, there are 300 ASR announcement during the sample period. Among the ASR announcements, some are announced in the same year by the same firm. In the cases of multiple announcements, we keep only the first ASR announcement by a firm in a year. My final sample has 269 unique firm-year ASR observations. Table 1 presents my ASR sample by year (panel A) and by 2-digit standard industry classification (SIC) (panel B). we create a matched sample of OMR announcements for the ASR announcements in my sample based on 2-digit SIC codes of the repurchasing firms, repurchase authorizing years and size of the total assets of the repurchasing firms. We use 2-digit SIC codes instead of 3-digit SIC codes to get the highest number of matches possible. Even after using 2-digit SIC codes to match ASR announcing firms to OMR announcing in a year, we still have 10 ASR announcing firms, for which we don t have any matched OMR announcing firm in a year. For those 10 ASR announcing firms, we find matches based on the SIC divisions. Data on the financial characteristics of the repurchasing firms are from Compustat and capital market data is from CRSP. We collect data on the analysts consensus forecasts of quarterly EPS and the actual values of EPS from the IBES summary file. 4. Empirical Analysis 4.1. Univariate analysis Table 2 presents means, medians and differences in means and medians of some important variables for my ASR sample and the matched OMR sample. All the financial characteristics in table 2 are calculated using values from the fiscal year prior to the repurchase announcements. Assets is the Compustat value of total assets at the end of the fiscal year prior to the repurchase announcement. Ln(Assets) is the natural log of Assets. Market Cap is the product of number of common shares outstanding and closing price of each share. Ln(Market Cap) is the natural log of market Cap. P/E is measured as market cap divided by net income. MV /BE is measured as market cap divided by book value of common equity. Q is measured as the sum of book value of liabilities 13

and market value of common equity divided by book value of assets. ROA is measured as net income divided by lagged assets. Debt/assets is measured as total debt divided by lagged assets. Total debt is the sum of total long-term debt and debt in current liabilities. Cash flow/assets is measured as the sum of net income and depreciation divided by lagged assets. Cash/assets is measured as total cash and cash equivalents divided by lagged assets. Dividend payout is measured as cash dividends divided by net income. Net PPE / assets is measured as net PPE divided by lagged assets. Authorized shares percentage is percentage of shares outstanding authorized to repurchase. Program size is measured as total value authorized divided by lagged equity. Ln(Illiquidity) is the log of Amihud (2002) measure of illiquidity estimated as the absolute percentage price change per dollar of daily trading volume measured over the period of 255 trading days prior to the repurchase announcement and ending 46 days prior to the repurchase announcement. Past return volatility is measured as standard deviation of daily returns measured over the period of 255 trading days prior to the repurchase announcement and ending 46 days prior to the repurchase announcement. Beta is measured using 36 monthly returns during 3 years prior to the year of repurchase announcement from CRSP. Past return (-40, -5 days) is the market model cumulative abnormal returns over days - 40 through -5 prior to the repurchase announcement. Past return (-6m,-2m) is the market model cumulative abnormal return over months -6 through -2 prior to the repurchase announcement. We test the differences in means (medians) using a t-test (Wilcoxon rank sum test). We find that companies that announce ASRs are significantly larger in size than the matched companies that announce OMRs in the same year. The difference in mean (median) asset size is $9474.9 ($1302.76) millions and is significant at a significant at 10% (5%) level. Both ASR and OMR announcing firms hold same percentage of their total assets in cash and cash equivalents. The value of the mean (median) Cash/Assets ratio for both ASRs and OMRs is 0.16 (0.11). Since a typical ASR announcing firm is larger in size than a similar OMR announcing firms and both have similar Cash/Assets ratio, the actual value of cash holdings by an ASR announcing firm should be higher than that of a similar OMR announcing firm. These results are expected since ASRs are large amount of immediate repurchases, usually relatively larger companies with higher amount of cash holdings will conduct ASRs. We don t find any significant difference in the level of investment opportunities for the ASR announcing firms and the matched OMR announcing firms before the repurchase announcements. The values of the mean (median) Q for ASRs and OMRs are 1.91 (1.61) and 1.80 (1.47) respectively, 14

and they are not significantly different. The values of the mean (median) MV/BE is for ASRs and OMRs are 3.55 (2.36) and 4.95 (2.17) respectively, and they are not significantly different either. ROA of the ASR announcing firms tend to be slightly higher than the matched OMR announcing firms. The difference in mean (median) ROA is 1% (1%) and is significant at a significant at 10% (10%) level. Firms conducting ASRs pay larger portions of their net income as dividends than similar firms announcing OMRs do. The median dividend payout ratio of ASR announcing firms is significantly (at 5% level) higher that of the matched OMR announcing firms by 0.09. ASR announcing firms have significantly lower leverage ratio than the OMR announcing firms. The difference in mean (median) Debt/Assets ratio is -0.06 (-0.04) and is significant at 1% (1%) level. Both ASR and OMR announcing firms have negative abnormal return on average prior to the announcements. The mean (median) abnormal returns during a period of 5 months ending 2 months prior to the repurchase announcements of ASRs and OMRs are -2.7% (-1.7%) and -5.1% (- 3.6%) respectively. The mean (median) abnormal returns during a period of 36 days ending 5 days prior to the repurchase announcements of ASRs and OMRs are -0.2% (0%) and -3.2% (-3%) respectively. Where ASRs and OMRs do not have any significant difference in the past 5-month mean cumulative abnormal return (CAR), they have a very significant (1% level) difference of 3% (1.3%) in the means (medians) of past 36-day CARs. Though both ASRs and OMRs experience negative stock performances before the repurchase announcements, improving stock performance close to the repurchase announcements for ASRs implies that the market is recognizing the undervaluation of the ASR announcing firms and has started to catch up. Firms conducting ASRs have significantly higher liquidity (implied by lower log of illiquidity) and past return volatility than firms conducting OMRs. The mean (median) Ln(Illiquidity) for ASR announcing firms is significantly (at 1% level) lower than that of OMR announcing firms by 0.48. The mean (median) past return volatility for ASR announcing firms is significantly (at 1% level) lower than that of OMR announcing firms by 0.0015. The difference in mean (median) past return is 0.03 (0.013) and is significant at 1%(5%) level. Also, ASR announcing firms have significantly lower systematic risk, measured as beta, than the matched OMR announcing firms in my sample. The difference in mean (median) beta is -0.10 (-0.07) and is significant at 10% level. Lower stock return volatility along with lower Beta of ASR announcing firms, compared to those of OMR announcing firms imply relatively lower uncertainty of the expected near-term operating performance of the ASR announcing firms. As previously mentioned, ASR announcing firm may have to compensate 15

the investment bank later based on the future movement of the repurchasing firm s stock price in the market. Relatively lower expected near term uncertainty and higher liquidity of stocks thus incentivize the repurchasing firm to elect ASRs over OMRs. We find that ASR programs are smaller in relative size on average than OMR programs announced by matched firms. This result doesn t imply that ASRs are smaller in actual size since these are relative sizes (offer value as a percentage of equity) and a typical ASR announcing firm is larger than a matched OMR announcing firms. The OMR programs often tend to be very large in size in order to benefit from the inherent flexibility of the program. Marginal benefit of flexibility decreases with the offer size as long as the repurchasing company is expected to repurchase the whole of the authorized shares on time. As Stephens and Weisbach (1998) show that 3 years after the OMR announcement only slightly above half of the OMR announcing firms repurchase the number of shares they authorized, it is evident that OMR announcing firms often tend to announce an offer size which they eventually do not repurchase in full. ASR sizes are big largely due to the fact that conveyed signal goes stronger with a bigger size. But, immediacy of an ASR program and the possibility of higher compensation to the investment bank for a positive price movement of repurchasing firm s stocks in future may cause the average relative size of an ASR program to be significantly smaller than that of an OMR program. 4.2. Choice of ASR versus OMR Table 3 presents the coefficients from the logistic regressions where the dependent variable is an indicator variable which takes a value of 1 if the announced repurchase is an ASR and 0 otherwise. In Regression (1), we use Ln(Assets), Debt/Assets, Cash/Assets, Net PPE/Assets, Ln(Illiquidity), Q, ROA, dividend payout ratio, past stock performance, past return volatility and program size as explanatory variables to see the effect of these variables on the choice between ASRs and OMRs. All the explanatory variables except past stock performance, past return volatility and program size are calculated using the Compustat values form the quarter immediately prior to the repurchase announcements. All the asset scaled variables are divided by lagged assets. Past stock performance, past return volatility and program size are estimated as in table 2. As we find in the univariate analysis, Firms with higher Debt/Asset ratio are significantly less likely to conduct ASRs than OMRs. This result also suggests that repurchasing companies below their target leverage ratios may adopt ASRs to increase their leverage ratios immediately. Like the 16

results from the univariate analysis, we find significant negative coefficients for Ln(Illiquidity) (-.546, significant at 1% level) and program size (-10.818, significant at 1% level) and a significant positive coefficient for prior stock performance (3.13, significant at 1%). Firms with lower liquidity (implied by higher Ln(illiquidity)) are significantly less likely to conduct ASRs than OMRs. Firms that want to conduct a larger repurchase program are significantly less likely to conduct ASRs than OMRs. Firms with better prior stock performance are significantly more likely to conduct ASRs than OMRs. These results agree with the flexibility hypothesis according to which cost of flexibility lost in an ASR declines with an increase in liquidity (implied by a decline in Ln(Illiquidity)) of the repurchasing firm s stocks and with a decline in uncertainty of its stock price. The significant positive association between the choice of ASR and the repurchasing firm s past stock performance does not necessarily mean that signaling undervaluation cannot be a motive of ASR announcements. However, as in the univariate analysis, it does imply that undervaluation might not be the only reason to choose ASRs over OMRs. It indicates that among the repurchasing firms of similar size and in the same industry, the ones with comparatively higher returns close to the announcements have higher probability to choose ASRs over OMRs. Later in the abnormal return analysis, it will be clear that signaling undervaluation can well be an important motive for conducting an ASR. In regression (2), (3) and (4), we add indicator variables Miss, Beat through repurchase and Beat without repurchases respectively as explanatory variables and use all the variables from regression (1) as control variables. Miss is an indicator variable that takes a value of 1 if the repurchasing firm misses the mean EPS forecast from analysts in the quarter of the repurchase announcement and 0 otherwise. Beat through repurchase is an indicator variable that takes a value of 1 if the repurchasing firm beats the mean EPS forecast from analysts in the quarter of the repurchase announcement by repurchasing its stocks (would have missed the EPS forecast without repurchases) and 0 otherwise. Beat without repurchases is an indicator variable that takes a value of 1 if the repurchasing firm beats the mean EPS forecast from analysts in the quarter of the repurchase announcement even without repurchasing any stock in that quarter and 0 otherwise. We calculate the EPS before repurchases as EPS from the quarter of repurchase multiplied by the common shares outstanding at the end of that quarter of repurchase and divided by the common shares outstanding at end of the quarter immediately prior to the repurchase. For the analysts consensus EPS forecasts, we take the last available mean EPS forecast before the earnings reporting date for that firm for that quarter from IBES summary statistics file. Then we estimate the 17

difference between the EPS before repurchases and the mean EPS forecast for that quarter from the IBES summary file to determine if the repurchasing company would miss or beat that forecast without repurchases. We find a significant positive coefficient (.936, significant at 1% level) for the indictor variable Beat through repurchase in regression (3) and no significant coefficient for either Miss in regression (2) or Beat without repurchases in regression (4). This result satisfies my hypothesis that repurchasing firms often use accelerated share repurchases to meet or beat the analysts consensus forecasts of EPS. More specifically, repurchasing firms that use repurchases to meet or beat the analysts consensus forecasts of EPS are more likely to conduct ASRs than OMRs. 4.3. Abnormal returns observed around the announcement of ASRs We look at the short-term market reaction around the repurchase announcements of ASRs and OMRs. Panel A of Table 4 presents the 3-day (-1,+1 day) mean market model cumulative abnormal returns 6 (CARs) around the repurchase announcement days of ASRs and matched OMRs. Panel B of table 4 presents 3-day mean market model CARs around the announcement of ASRs in different EPS categories. As described before, the EPS categories are Miss, Beat through repurchase and Beat without repurchases. The 3-day mean CAR around the repurchase announcements of ASRs is significantly higher than the 3-day mean CAR around the repurchase announcements of OMRs. The 3-day mean CAR around the ASR announcements is 1.84% and that around the OMR announcements is 1.49%. Both are significant at 0.1% level. Among all the ASRs in our sample, the ones announced by firms in Beat through repurchase category have the highest 3-day mean CAR around the announcement dates. The 3-day mean CAR around the announcements of ASRs announced by firms in the Beat through repurchase category is 2.30% and is significant at 0.1%. The same return measure for the ASRs announced by firms in the Miss category is 0.94% and is significant at 1%. The same return measure for the ASRs announced by firms in the Beat without 6 Cumulative abnormal returns are defined as sums of daily (monthly in monthly CARs) abnormal returns over the period of interest. the cumulative abnormal return (CAR) of a security i over (τ 1, τ 2 ) period is defined as τ CAR i,(τ1,τ 1 ) = 2 AR t=τ i,t 1 Where, AR i,t = R i,t - E [R i,t ] Here, E [R i,t ] is the expected return for security i and over a period t, and is estimated using market model: R i,t = α i,t + β i R m,t + ε i,t Parameters of the market model for a security i are estimated using the security returns and market returns over a period of 255 days ending 46 days prior to the event date with a minimum estimation length of 3 days. 18