Investor Inattention, Managerial Learning, and the Decline of Friday Merger Announcements

Transcription

1 Investor Inattention, Managerial Learning, and the Decline of Friday Merger Announcements Antonios Siganos Senior Lecturer in Finance University of Glasgow Betty (H.T.) Wu * Lecturer in Finance University of Glasgow Betty.Wu@glasgow.ac.uk This version: 20/12/2016 * Corresponding Author: Accounting and Finance Subject, Adam Smith Business School, University of Glasgow, West Quadrangle, Main Building, Glasgow, G12 8QQ, UK. 1

2 Investor Inattention, Managerial Learning and the Decline of Friday Merger Announcements Abstract We use US data between 1986 and 2013 and study whether managers select strategically the timing of Friday merger announcements. We find that there has been a significant reduction of Friday merger announcements overtime, and as an example 19% of mergers were on average announced on Fridays in 1986 while only 5% in We use two cut-off points to explain the pattern; the introduction of EDGAR in 1994 and the financial crisis in We find evidence indicating that investor inattention has increased since 1994, and that managers seem to consider prior investor inattention and strategically make a merger announcement on Fridays. In addition, the number of low quality merger deals has reduced since the financial crisis due to the limited access to external funding sources and managers seem to react accordingly by avoiding Friday merger announcements. Overall, our results support the investor inattention and managerial learning hypotheses, while showing evidence inconsistent with the stock market efficiency. Keywords: Friday Merger Announcements; Investor Inattention; Managerial Learning; Stock Market Efficiency. 2

3 1. Introduction It has been previously identified in politics arena that the timing to make public announcements is carefully selected. As an example an aide of the previous UK government gave the following advice at 9/11 terrorist attacks: It is now a very good day to get out anything we want to bury. Councillors expenses? (Sparrow, A. Sept.11, A Good Day to Bury Bad News, The Telegraph). Within this spirit, a number of studies in accounting and finance explore whether managers select strategically the timing to make public announcements based on investor inattention on Fridays (e.g., Patell and Wolfson 1982; Penman, 1987; Damodaran, 1989). According to Stone et al. (2012), most people feel like Thank God it s Friday, and there is therefore less attention at work activities and more interest in the weekend plans. Strategic managers would therefore select Fridays to make bad news announcements in order investors not to respond in full to available information. In this study, we find that there is a pronounced reduction on Friday merger announcements during our sample period ( ), and we argue that this pattern indicates that managers are strategic on the day of the week selected to make their merger announcements. Apart from very few studies (Louis and Sun, 2010; Louis and Sun, 2016) that explored investor inattention within mergers before, the majority of studies in the field test investor inattention on earnings announcements (e.g., Patell and Wolfson, 1982; Penman, 1987; Damodaran, 1989; Dellavigna and Pollet, 2009; Doyle and Magilke, 2009; DeHaan et al., 2015). We focus our paper in the context of mergers, since we believe that mergers offer an ideal setting to test investor inattention. Mergers are relatively unexpected events and managers may therefore have more pronounced opportunities to take advantage of investor inattention on Fridays in relation to other firm announcements. Within mergers, stock returns for both bidding and target firms are available where especially targets stock returns are pronounced on the announcement (typically over 10%), and potential investor inattention on 3

4 Fridays can therefore be more easily identified. The recent financial crisis in 2008 also offers a natural cut off points to explore how changes in the quality of merger announcements is related with the frequency of Friday merger announcements. We find evidence that the average quality of mergers got improved since the crisis, which we find to be related with a significant decrease of Friday merger announcements. We introduce to the literature a new pattern on the frequency of Friday merger announcements overtime. Using US merger deals between 1986 and 2013, we find that there is a pronounced decrease of Friday merger announcements during our sample period, as clearly shown at Figure 1. As an example, 19% of mergers were on average announced on Fridays in 1986, while only 5% in Figure 1 also shows the number of public-to-public deals undertaken during the period, indicating that our results are driven by relatively large numbers across the sample period. We construct a trend variable for Friday announcements in relation to other days of the week, which is found significantly negative at the 1% level. We find that the decrease pattern on Friday deals is more pronounced for bidders that experience negative abnormal stock returns in the announcement (-1,1 days), highlighting that the pattern is driven by managers opportunistic behavior. If managers would select the day of the week as a result of chance, it would be expected that around 20% of the merger announcements would take place on Fridays, equally across the days of the week. If managers would reduce their strategic selection of the day of the week, the percentage of merger announcements on Fridays would have increased overtime getting closer to the normal 20% level. Instead, the pronounced downward pattern indicates that managers have become more selective of Friday merger announcements during the sample period, reflecting managers efforts to take advantage of investor inattention on Fridays. [please insert Figure 1 here] 4

5 We argue that there are two main forces that made managers to reduce Fridays merger announcements overtime. These cut off points are the introduction of the Electronic Data Gathering, Analysis and Retrieval (EDGAR) system in 1994, and the financial crisis in We therefore distinguish three sub-periods, and expect that each sub-period should indicate a significant reduction on Friday merger announcements due to the different context that managers faced. The underlying reasoning of both cut off points offer some evidence that managers have reduced Friday merger announcements for opportunistic purposes. To elaborate, firms statements became electronically available with EDGAR system in On the one side, new technology may offer quick access to available information to investors. However, the share size of information was increased significantly since 1994, and therefore it became much harder for investors to follow available information. According to Internet World Stats, 89% of the North American population were internet users in June 2016, with an increase of 196% from 2000 to 2016 driven by mature citizens growth in online usage. 1 Also an increasing number of devices have been introduced such as ipads, smartphones, smart TVs that are all connected and generate noise. Within this digital setting, we argue that there should be more available opportunities for opportunistic managers to take advantage of investor inattention since 1994 who would select Fridays to announce mergers less often. Citizens have also been working for longer hours during a week overtime especially within managerial positions (Jacobs and Gerson, 2001), and therefore investors may show more evidence of inattention on Fridays within recent years due to the more pronounced anticipation of weekends. We support our expectations showing evidence that investor inattention tends to get poorer since 1994 as shown by exploring stock price reactions and trading volume on Fridays in relation to other days of the week. We find evidence that managers as if attempt taking 1 (last accessed November 2016). 5

6 advantage of that increasing level of investor inattention overtime by increasingly respond to prior inattention level. As an example, we find that more announcements are likely on Fridays of poor quality merger deals (those with negative abnormal stock returns around each merger announcement, -1 to 1 days) following high level of investor inattention. This relation is more pronounced within managers that are more likely to act opportunistically. We find that the relation is stronger within small size firms, in line with Chen and Mohan s (1994) results that it is more likely for small size firms managers to act opportunistically. We also find that opportunistic behavior is more prominent within merger deals with a short distance between the announcement and the completion of a merger, in line with Louis and Sun s (2016) result that mergers closer to completion offer the most potential benefit for managers to select the day of the week. In addition, we find that investors do not cover the initial underreaction on Fridays the days following and find that there has not been any improvement on investors coverage following the initial underreaction on Fridays overtime. Based on both investor inattention level on Fridays and investor slow responses the days following Fridays, our results indicate that the market has become more inefficient overtime. Our results further support the managerial learning hypothesis, according to which managers learn from available information as shown in this occasion by taking advantage of irrational investors. Our second cut off point is related with the recent financial crisis that would indicate that opportunistic managers would announce even less Friday merger announcements since Plenty of evidence suggests a significant decrease in bank lending in the post-financial crisis period. Campello et al. (2010) surveyed CFOs who reported having had significant difficulties accessing external funding during the crisis. More than half of the firms in their sample had difficulties accessing funding during the crisis. There was also a sharp decline of available funding from the start of the financial crisis, with Ivashina and Scharfstein (2010) 6

7 empirically reported reductions in bank lending by 47 percent within a short period between the third and fourth quarters in Due to the difficulty for firms to borrow money to fund mergers, we argue that a significant reduction of merging activity would occur mostly within low quality merger deals. In line with our expectations, we empirically support that constrained bidders undertook relatively less mergers after the financial crisis, and that firms with financial constraints are linked with poor announcement returns. We also find that bidders announcement returns are on average insignificant after the financial crisis, while prior to the crisis the wealth effect is negative in line with conventional merger literature (e.g., Firth, 1980; Jarell and Poulsen, 1989; Agrawal et al., 1992; Kaplan and Weisback, 1992; Moeller et al., 2005). A decrease of Friday merger announcements after the financial crisis therefore indicates that managers did not seem to select Fridays to make merger announcements since relatively more merger deals are of good quality. We contribute to the literature in several ways. We first contribute regarding the day of the week that firms select to reveal new information. Louis and Sun (2010; 2016) are the only two studies available that have studied the inattention hypothesis on merger announcements. Louis and Sun (2010) used data between 1994 and 2006, and found that bidder stock returns are on average underreacted on Fridays merger announcements in relation to counterpart non-fridays announcements, supporting the investor inattention hypothesis. They reported that the underreaction on stock returns on Fridays is not covered on coming Mondays and managers could therefore take advantage of investor inattention. Louis and Sun (2016) updated their data from 1994 to 2010 and reported that the Friday effect is driven by mergers with a short distance in days between the announcement and the completion of the deals showing evidence that managers select Friday merger announcements opportunistically. However not all studies, mostly arriving from earnings announcement 7

8 literature, have supported that managers are opportunistic. For example, Michaely et al. (2016) reported that the lower stock return reaction on Fridays is the outcome of a selection bias of the firms that make announcements on Fridays. After controlling for firm characteristics, they reported that there are no differences in stock returns for Fridays announcements in relation to counterpart non-fridays announcements. Also, DeHaan et al. (2015) found that investors may underreact on Fridays in relation to other days of the week, but investors cover their initial underreaction over the following days. Our study offers new evidence by focusing on Friday merger announcements overtime. We highlight a significant reduction of Friday merger announcements between 1986 and 2013, which we argue that indicates managers strategic selection of the day of the week to announce merger deals. Our approach captures managerial decisions during the sample period, rather than focusing on the average decision, highlighting that there is a consistent pattern as we move along to more recent years that we argue is attributed to managers learning on prior investor inattention. Second, we contribute to the scant literature of the managerial learning hypothesis (e.g., Boot and Thakor, 1997; Subrahmanyam and Titman, 1999; Foucault and Gehrig, 2008), according to which managers learn from the response of their firms share price information to news. Most studies in the field are theoretical. Kau et al. (2008) is one of the few empirical studies found that US managers are likely to cancel investments if investors react negatively on the announcement. Within this spirit, Huang et al. (2016) empirically reported that Asian managers respond to share price reaction to H- rather to A-shares, with H-shares are those traded by sophisticated international investors while A-shares only traded by naïve domestic investors. We instead find that US managers learn from prior inattention level on Fridays. When inattention level is high (low) in past mergers, there are more (less) Friday merger announcements on poor quality merger deals. Not only the context of our study differs, but we also report that managers may learn from the share price reactions to information in other 8

9 firms. Overall, these results indicate that rational managers attempt to take advantage of irrational investors as previously supported within alternative contexts (e.g., Cooper et al., 2001; Shleifer and Vishny, 2003; Baker and Wurgler, 2004). Third, we offer some contribution into the debate whether markets have become more or less efficient overtime. Results are rather mixed in the literature within alternative markets and testing criteria. Bai et al. (2016) found that current stock returns may reflect in recent years more accurately US firms future cashflows over the 3 to 5 years later, but stock returns prediction remains similar during the sample period when predicting one-year future cashflows. These results show some support that the US market has become more efficient overtime. However, Hull and McGroarty (2014) used data of 22 international countries reported that there is no clear evidence that markets have become more efficient when exploring the Hurst exponent overtime. Also while Cajueiro and Tabak (2004) may find some evidence of improvement in market efficiency for international markets between 1992 and 2002, but they reported that there have been no significant changes at the level of US market efficiency. We compliment these studies by focusing on investor inattention on Fridays, showing some evidence that the US market has become less efficient overtime. Note that our results do not indicate that prices are less informative over the long term, but we instead show evidence of short-term underreaction to available information on Fridays. We argue that this increased investor inattention on Fridays may be related with the increased size of available information in more recent years. Investors also work for longer hours during the week who may therefore anticipate weekends more profoundly. Finally, our study contributes to the wealth announcement effect of bidders literature (e.g., Firth, 1980; Jarell and Poulsen, 1989; Agrawal et al., 1992; Kaplan and Weisback, 1992; Moeller et al., 2005). We find that there has been a relatively increase in bidder announcement stock returns in the post-financial crisis period. The difficulty to raise funding 9

10 for acquiring targets seem to have reduced the frequency of low quality mergers since the crisis and as a result, we report that bidders on average experience significantly stronger announcement returns after In particular, we find that the wealth announcement effect for bidders is insignificant after the crisis in relation to the typical negative bidder announcement stock returns for public-to-public merger deals in the conventional pre-crisis period. The remainder of the paper is structured as follows: Section 2 describes our data, while Sections 3 to 5 discuss the empirical results. Section 6 concludes. 2. Data We use US merger announcement data available from Thomson OneBanker between January 1986 and December 2013 to test our predictions. We start out data in 1986, since only limited number of mergers experienced earlier for an analysis that would compare Friday versus non- Friday announcements. For our main tests, we only select public-to-public merger deals, since we are interested on investor inattention at both bidder and target firms. 2 This is the first study that explores investor inattention on targets stock returns and trading volume, where targets experience significant announcement returns, and therefore potential investor inattention may be more easily identified. We set some commonly adopted data restrictions in the merger field (e.g., Agrawal and Nasser, 2012). In particular, we restrict our sample in deal values of at least $1 million, within acquisition of at least a 50% stake, and to transactions where the target company size is at least 1% of the market value of the bidder. 3 We get access to bidder and target stock returns and corresponding trading volume data from the Centre for Research in Security Prices (CRSP), where trading volume is calculated as a 2 As a robustness, we test results for public-to-private merger deals and for stock swap deals later in the study. 3 Note that we exclude deals classified as minority stake purchases, acquisitions of remaining interest, spinoffs, recapitalizations, and repurchases. 10

11 firm s daily volume divided by the number of outstanding shares. Abnormal stock returns are estimated as a firm s market adjusted stock returns using the interval period -250 days to -100 days prior to each merger announcement. Abnormal trading volume is estimated using the standardized trading volume per firm above its normal level of transaction volume in the interval period from -250 to -100 days prior to each merger announcement. Table 1 shows the number and the percentage of merger announcements per day of the week. In line with Louis and Sun (2010; 2016), we find that there is no equal distribution of merger announcements per day in a week that would have supported a random selection. There may be around 20% of the announcements on the days between Tuesday and Thursday (in line with random selection), but there are significantly less announcements on Fridays and more on Mondays. We find that only 11.53% of the mergers are on average announced on Fridays, and 33.04% on Mondays. [please insert Table 1 here] Table 2 reports the descriptive statistics of our control variables. The selected control variables are normally used within merger literature (e.g., Agrawal and Nasser, 2012; Louis and Sun, 2010; 2016). For example, we control for deals that are competitive that takes one for deals with multiple bidders according to SDC otherwise zero, for diversified deals with diversifying being a dummy variable equal to one where targets and bidders do not share the same first two-digit SIC code, for the relative size of the target to the bidder, for the size of the bidder as captured by the natural logarithm of the market value of equity, for the percentage of stock payment using the natural logarithm, for the profitability (return on assets) of the acquirer, and for bidders who are listed in Nasdaq. We also control for macroeconomic conditions as measured by the GDP growth in the US market. [please insert Table 2 here] 11

12 3. Establishing the reduction of Friday merger announcements overtime 3.1 Results in the full sample of public-to-public mergers This section explores empirically the decreasing pattern of Friday merger announcements overtime. We first estimate univariate statistics by exploring the distribution of merger announcements per day of the week during , , and As stated earlier, 1994 is linked with the introduction of EDGAR system, while 2008 with the recent financial crisis. We split the period into two sub-periods to ensure that the Dot-Com bubble does not influence significantly our results. Table 3 shows that there are insignificant increases in the frequency of merger announcements on most days between Mondays and Thursdays and that there is a significant decrease on Fridays merger announcements overtime. 4 We find that 14.90% of merger announcements were on Fridays between 1986 and 1993, 12.51% between 1994 and 2000, 10.83% between 2001 and 2007, and only 6.68% between 2008 and The drops in the frequency of Friday merger announcements are economically and most are statistically significant. The Dot-Com bubble in early 2000s does not seem to influence significantly the findings. [please insert Table 3 here] We undertake multivariate analysis to explore the frequency of Friday merger announcements during our sample period. We first estimate logistic regressions where the dependent variable is a dummy that takes one for Friday merger announcements, otherwise zero. Our interest is on the trend variable that captures the frequency on Fridays merger 4 In untabulated results, we explore a trend for Monday, Tuesday, Wednesday and Thursday in relation to Friday s merger announcements. We find no evidence to support that managers decreased merger announcements on Fridays in relation to any particular day of the week. The increases in merger announcements overtime between Monday and Thursday are economically small, and statistically insignificant. 12

13 announcements in relation to non-fridays announcements. Table 4 shows that the parameter coefficient of trend is significantly negative at the 1% level as shown at columns (1) and (2) without and after controls, respectively. In untabulated results, we estimated marginal effects showing that in relation to other days of the week, there is a reduction by 0.39% of Friday merger announcements per year during our sample period. The decrease in the frequency of merger announcements on Fridays is thus economically and statistically significant. [please insert Table 4 here] At column (3) of Table 4, we instead use a dummy for merger announcements that took place between 1994 and 2000, 2001 and 2007, and those between 2008 and 2013, while using merger deals took place between 1986 and 1993 as the base period. After controls, we find that our parameter coefficients are insignificant in , , but significant at the 5% level in The frequency of Friday merger announcements is therefore dropped as we move along to more recent sub-periods, being at the lowest level after the financial crisis. 3.2 Results in alternate merger samples We then explore results within alternate merger samples. We first explore the reduced frequency of merger announcements on Fridays in stock swap and public-to-private merger deals. Note that stock swap merger deals are a sub-sample of public-to-public merger deals used earlier by selecting only those mergers with at least 50% funded with stock. These samples are used to test the robustness of our results when following data restrictions in line with Louis and Sun (2010, 2016). [please insert Table 5 here] 13

14 Table 5 shows that our results hold in the new samples as for public-to-public merger deals earlier. As an example, the parameter coefficient for trend is (significant at the 5% level) and the parameter coefficients in is significant at the 10% level for stock swap merger deals. The corresponding parameter coefficients for public-to-private merger deals are and , both significant at the 1% level. Interestingly, the magnitude of the parameter coefficient in trend for stock swaps merger deals is more negative, and for public-to-private merger deals less negative, in relation to public-to-public merger deals (although these differences are insignificant in conventional statistical level). Bidding firms tend on average to experience negative abnormal stock returns in the announcement on stock swap merger deals and positive abnormal stock returns in the announcement of public-to-private merger deals (e.g., Travlos, 1987; Chang, 1998). We thus test whether bidders abnormal announcement returns are related with the magnitude of the reduction of Friday merger deals within the full sample of public-to-public merger deals. We estimate bidders abnormal stock returns in the announcement, -1 to 1 days, and split merger deals into positive and negative. If managers select Fridays to announce poor merger deals, and that trend is more prominent overtime, we should find that Friday merger announcements are reduced more prominently on poor merger deals. Figure 2 indeed shows that the slope is more negative for poor quality merger deals in relation to good quality deals, indicating that managers select Friday merger announcements strategically. In untabulated results, we also estimate univariate logistic regressions with the dependent variable takes one for Friday merger announcements otherwise zero, focusing on the magnitude of the parameter coefficient trend. We find that the relevant parameter coefficient is (significant at the 1% level) for negative merger deals, and (significant at the 10% level) for positive merger deals. [please insert Figure 2 here] 14

15 Managers may not manage to identify that a merger deal is bad/good news in deals with abnormal stock returns in the announcement close to zero. We therefore explore the magnitude of trend variable in more extreme negatively/positively performing bidders. As expected, we find that the reduction on Friday merger deals is more pronounced in very poorly performing bidders, and as an example the parameter coefficient on trend is even more negative at (significant at the 1% level) for the bottom 30% performing bidders. Figure 2 as an example illustrates the trends for the bottom versus the top 20% performing bidding firms indicating that the slope is negative for very poorly performing firms, while it remains stable for well performing firms. Overall, our results indicate that managers select Fridays intentionally to announce poor deals and such selection drive the overall reduced pattern on Friday merger announcements overtime. 4. The reasoning for the reduction of Friday merger announcements 4.1 Overtime investor inattention In the following sections, we explore the reasoning behind the reduction of merger announcements on Fridays using the full sample of public-to-public merger deals. The new era of internet may make information easily available across investors, but the magnitude of information has been increased significantly, making harder for investors to follow all available information. Citizens have also been working for longer hours during a week overtime who may have stronger anticipation of weekends overtime. We thus expect that investor inattention on Fridays merger announcements should have gradually been increased. If that is true, and to the extent that managers attempt to take advantage of investor inattention, managers would be more selective on Friday merger announcements experiencing less frequent Friday merger announcements after EDGAR. 15

16 We initially explore investor inattention level on Fridays versus non-fridays along our sample period. We estimate abnormal stock returns and abnormal trading volume on merger announcements on Fridays minus the corresponding stock returns and trading volume in the announcement of mergers in non-fridays. In line with investor inattention hypothesis, we expect that stock return reactions are of lower magnitude on Fridays in relation to non- Fridays, and that there is lower trading volume on Fridays (Louis and Sun, 2010). More importantly, in this study, we focus on investor inattention during the sample period that we expect to be more prominent in recent years. We split results for positive and negative bidders announcement effect on Fridays and on non-fridays, since the difference in stock returns would indicate the opposite relation. As long as there is an increase in inattention overtime, we expect that there is an increase in abnormal stock returns on Fridays versus non-fridays for positive announcement bidders, and a decrease for negative announcement bidders. We only report the difference in stock returns on Fridays versus non-fridays in the announcement of target firms with positive stock returns, since only very few target firms experienced losses. Since trading volume in a firm is non-negative, we expect that the difference in trading volume on Fridays versus non-fridays should decrease overtime to indicate investor inattention on Fridays. We use 120 prior merger deals to identify those with positive and negative announcement effect. 5 Figure 3 plots the differences in stock returns and in trading volume for Fridays and non-fridays bidders with positive and negative announcement effects, and for targets with positive announcement effect. In line with our expectation, we find evidence that investor inattention has increased overtime. There is the tendency the difference in stock returns to increase in negative merger deals for bidders and to increase in positive merger deals for 5 In untabulated results, we find that results are similar when using 100 or 150 prior merger deals to identify those with positive and negative announcement returns. 16

17 bidders and targets. The difference in trading volume on Fridays and non-fridays tends to decrease overtime for both bidders and targets. [please insert Figure 3 here] 4.2 Do managers attempt taking advantage of investor inattention? We now explore whether managers attempt learning from past investor inattention level. Studies have earlier shown that rational managers attempt taking advantage of irrational investors (e.g., Cooper et al., 2001; Shleifer and Vishny, 2003; Baker and Wurgler, 2004). Also a number of mostly theoretical studies have shown that managers do respond to stock return information (Boot and Thakor, 1997; Subrahmanyam and Titman, 1999; Foucault and Gehrig, 2008; Kau et al., 2008; Huang et al., 2016). If investor inattention has on average increased overtime as shown in the above section, and as long as managers attempt exploiting investor inattention, managers should be more selective on merger deals announced on Fridays. The managerial learning hypothesis could therefore explain the less frequent merger announcements on Fridays overtime. Since only very few firms undertake a number of mergers within a short period, we explore whether managers learning is based on investor inattention in all prior mergers. We expect that managers learning is more pronounced after the introduction of EDGAR in 1994, where increased level of investor inattention was present, offering more notable opportunities for managers to exploit. As long as managers learn from prior investor inattention level, they are more likely to announce poor quality merger deals on Fridays following high investor inattention. We use a dummy to separate merger announcements of good and poor quality merger deals based on the cumulative abnormal stock returns and abnormal trading volume around each merger announcement day. We identify good quality merger deals as those with 17

18 non-negative cumulative abnormal stock returns, while counterpart poor quality merger deals as those with negative cumulative abnormal stock returns. We also define high prior investor inattention when there is an underreaction on Friday stock returns and trading volume in relation to the same period stock price reaction on other days of the week. We use dummies to separate the groups. We initially estimate a logit model where the dependent variable is a dummy variable that takes one when a Friday merger announcement is experienced, while the key independent variables are a dummy for negative announcement bidders, and a dummy for bidders that the difference in stock returns / trading volume on Fridays minus non-fridays was negative in prior 120 merger deals. We interact these two variables and add all our control variables. We estimate this regression over the following interval periods: , , , , and Figure 4 shows the parameter coefficient of the interaction term (prior inattention on abnormal stock returns and trading volume * merger deals with poor announcement of bidders) in each of these sub-periods, while all other parameter coefficients are available upon request. We find that the relevant parameter coefficient tends to increase overtime, indicating that managers are more likely to announce poor quality merger deals on Fridays following high investor inattention in more recent years. [please insert Figure 4 here] We further add a third term in the interaction within the full dataset. Since we use the introduction of EDGAR in 1994 as the initial stage of internet usage, we interact prior inattention * merger deals with poor announcement of bidders * merger deals after Once again the dependent variable is a dummy variable that takes one for Friday merger announcements, otherwise zero. If managers have become more opportunistic after 1994, the relevant parameter coefficient of the three interactive terms should be significantly positive. We report results with and without control variables as shown at Table 6. Our results indeed 18

19 support our hypotheses. We find that the relevant parameter coefficient is and significantly positive at the 1% level without controls, while and significant at the 10% level after controls when considering investor inattention on prior abnormal stock returns. Results are similar when exploring investor inattention on prior abnormal trading volume, although the parameter coefficient on the three interactive terms is slightly insignificant after controls (p-value =0.110). [please insert Table 6 here] Settings with more or less pronounced relation We further explore in which settings our three interactive terms may be more or less pronounced in line with the likely managers to act opportunistically. According to Chen and Mohan (1994) who undertook a questionnaire across 846 financial officers, it is more likely for small size firm managers to change the day of poor earnings announcements. In addition, according to Louis and Sun (2016), it is also more likely managers to opportunistically announce poor merger deals on Fridays with a short distance between the announcement and the completion. The benefit of a merger that arrives from the day of the week selected to make the announcement is more important for merger deals that will be completed. Managers would thus select more opportunistically the timing of the announcement for deals that are very close to completion. We re-estimate Table 6 s results separately for small versus large size bidders and for merger deals with a short versus long distance in days between the announcement and the completion. In each occasion, we use the median value to split the full sample into different groups. Note that there were not many large firms with positive abnormal stock returns and so we are unable to generate the three-way parameter coefficient in particular modes. Table 7 19

20 shows the results. As hypothesized, we find that the relevant parameter coefficient that interacts bidders announcement effect, prior investor inattention and merger deals after 1994 is more pronounced for small (rather than large) size bidders and for merger deals with a short (rather than long) distance in days between the announcement and the completion. The significance of the interaction term for small size bidders is at the 5% level, while it is insignificant for large size bidders. Also the parameter coefficient for merger deals with a short distance between the announcement and the implementation is significant at the 5% level, while once again insignificant for long distance mergers. The magnitude of the parameter coefficients also indicates the same story in line with the corresponding statistical significant levels such as that the parameter coefficient of the interactive terms for small size bidders is 3.296, while for large bidders only These results support the managerial learning hypothesis, showing that the pattern is more pronounced amongst managers who were expected to act more opportunistically. [please insert Table 7 here] We also test managerial learning on prior investor inattention for bidding firms with relatively good versus poor corporate governance. We hypothesize that managers that work in poor corporate governance firms are more likely to explore prior investor inattention to select the day of the week to make the announcement since stock abnormal returns in the merger announcement reflect amongst others managers quality. To test this, we access data from Boardex, which are only available after 2000, and proxy bidding firms corporate governance using (1) CEO s duality and (2) board independence. These measures are commonly used in the literature as proxies of a firm s corporate governance quality (e.g., Dechow et al., 1996; Carcello et al., 2002). CEO s duality is a dummy that indicates poor corporate governance when a firm s CEO also holds the position of the Chairman of the board. We also split the sample of bidding firms into high and low board independence based 20

21 on median. Firms with high board independence are considered high corporate governance firms having a significant percentage of outside directors not to be affiliated with the top executives. Table 8 shows the results. In line with our expectations, we find that managerial learning is more pronounced within low corporate governance firms. We find that the parameter coefficient that interacts prior investor inattention and negative deals are both significant at the 5% level for firms with CEO s duality, and on low board independence firms. The parameter coefficient of the interactive term is instead insignificant in counterpart scenarios. [please insert Table 8 here] 4.3 Do managers manage to take advantage of investor inattention? We have previously shown that investors inattention has increased overtime and that managers have increasingly considered prior inattention to decide whether to announce poor quality merger deals on Fridays. In this section, we explore whether managers manage to take advantage of investor inattention by exploring investor reaction on the days following the initial underreaction. If investors underreaction to information remains over the following days, managers strategic decision on announcing poor merger deals on Fridays would indicate that the share price of bidders did not reach its fair price for the benefit of bidding firms. Panel A of Table 9 compares the average returns on Fridays versus non-fridays as well on the days following the merger announcement for bidders and targets over the full sample period. As an example, CAR(0,1) indicates the cumulative abnormal stock returns for bidders and targets between Friday and coming Monday for Friday merger announcements. 21

22 We find that there is on average no cover of the initial reduced abnormal stock returns on Fridays. The difference between Fridays and non-fridays on the merger announcement day (Day 0) is 0.66% for bidders and -3.16% for targets, and there are only little signs that such underreaction got reduced the following days. At Panel B of Table 9, we further explore the changes of these differences across the sample period. We mainly explore whether investors reactions have improved/worsen as we move to more recent years. We find that there is no particular pattern within the first few subperiods, however the covering up got significantly worse after the financial crisis. Overall, these results indicate that managers seem able to take advantage of investor inattention on Fridays. [please insert Table 9 here] 5. The significance of the financial crisis on the decline of Friday merger announcements Although managerial learning hypothesis explains the reduction on Friday merger announcements overtime, we put forward a further argument that highlights a reduction on Friday merger announcements after the financial crisis in There is evidence that a decrease in bank lending took place since the crisis (Campello et al., 2010; Ivashina and Scharfstein, 2010), and we hypothesize that proportionately higher quality mergers took place after the crisis. Managers would reduce the percentage of Friday merger announcements after the crisis as long as they consider the quality of a merger to take such decision. We initially explore the channel according to which firms financial constraints are related with the quality of mergers experienced after the financial crisis due to the difficulty to raise money from external sources. We use Kaplan and Zingales (1997) financial 22

23 constraint index as a measure of bidders financial constraints. Although KZ index has received criticism in the literature, it is probably the most conventional measure of a firm s financial constraints (e.g., Chen and Wang, 2012). High KZ values indicate high probability of a firm facing constraints. We identify the highest 30% financially constrained bidders during the sample period, and explore their distribution before versus after the financial crisis period. In untabulated results, we find that 51% of bidders were financially constrained versus 49% without constraints before the crisis, while 44% of bidders faced financial constraints versus 56% without constraints after the crisis. This result indicates that bidders financial condition had no relation with the probability to undertake a merger before the crisis, while relatively less bidders with financial constraints undertook mergers after the crisis. In line with our argument, bidders that faced financial problems would thus found difficult to raise funds from external sources after the crisis. Our evidence also shows that it is more likely that financially constrained bidders undertake low quality merger announcements. We find that (1) financial constrained bidders are more likely to make Friday merger announcements, with 14% of financially constrained bidders made their announcement on Fridays versus only 9% of Friday merger announcements for financially unconstrained firms, and (2) financially constrained bidders are positively correlated with negative abnormal returns in their merger announcement. We thus expect that bidders on average experience relatively better abnormal stock returns in their merger announcements after the crisis, since relatively less number of mergers with financially constrained bidders took place. To test this prediction, we compare the announcement effect of bidders before versus after the crisis and in line with our expectation, we find that on average the cumulative abnormal stock returns (-1,1 days) are -1.50% and significant at the 1% level before the crisis 23

24 while -0.15% and insignificant after the crisis. We report at Table 10 abnormal stock returns of bidders after the crisis when adding commonly used control variables. We find that bidders experience 1.79% higher abnormal announcement returns after the crisis and that difference in stock performance is significant at the 1% level. In line with the literature (e.g., Firth, 1980; Jarell and Poulsen, 1989; Agrawal et al., 1992; Kaplan and Weisback, 1992; Moeller et al., 2005), we find that public-to-public mergers destroy on average acquirers shareholder wealth before the crisis. We contribute to the literature by highlighting that bidders announcement returns got relatively improved after the crisis, while we identified the channel of this change by showing that financially constrained bidders had a difficulty to raise funds from external sources after the financial crisis thus undertook less merging activity. Relatively low quality merger deals thus reduced since the financial crisis, and managers reduced accordingly the percentage of Friday merger announcements. [please insert Table 10 here] 6. Conclusions We explore whether firm managers strategically select merger announcements on Fridays based on the investor inattention hypothesis (e.g., Patell and Wolfson 1982; Penman, 1987; Damodaran, 1989), and find that there is a significant decline on Fridays US merger announcements from 1986 to 2013 driven by low quality merger deals. With the introduction of EDGAR system, a significant increase of information got available to investors, and therefore it got harder for investors to follow all available information. Managers seem to attempt learning from prior investor inattention level when selecting Fridays to announce mergers. Also since the financial crisis, the quality of mergers got on average improved due to the difficulty that firms faced to raise funds (Campello et al., 2010; Ivashina and 24

25 Scharfstein, 2010). Managers did not seem proportionally to select Fridays to make merger announcements since 2008 reflecting the change of the quality of merger deals. The results of this study are of interest to academics, practitioners and regulators. We introduce to the literature a new pattern regarding the frequency of Friday merger announcements overtime, and we made an effort to explain its sources. We report that rational managers attempt taking advantage of irrational investors by selecting the day of the week to make public announcements. While we support the investor inattention and the managerial learning hypotheses, we offer evidence against the stock market efficiency by showing that investors typically underreact to new information on Fridays and that this pattern becomes more prominent overtime. Interestingly, we also report that the wealth announcement effect of bidders got relatively improved since the recent financial crisis in

26 References Agrawal, A., J. Jaffe, and G. Mandelker (1992) The post-merger performance of acquiring firms: A re-examination of an anomaly, Journal of Finance, 47, Agrawal, A., and T. Nasser (2012) Insider trading in takeover targets, Journal of Corporate Finance, 18, Bai, J., T. Philippon, and A. Savov (2016) Have financial markets become more informative? Journal of Financial Economics, 122, Baker, M., and J. Wurgler (2004) A catering theory of dividends, Journal of Finance, 59, Boot, A. and A. Thakor (1997) Financial system architecture, Review of Financial Studies, 10, Cajueiro, D. and B. Tabak (2004) The Hurst exponent over time: Testing the assertion that emerging markets are becoming more efficient, Physica, 336, Campello, M., J. Graham, and C. Harvey (2010) The real effects of financial constraints: Evidence from a financial crisis, Journal of Financial Economics 97, Carcello, J., D. Hermanson, T. Neal, and R. Riley (2002) Board characteristics and audit fees, Contemporary Accounting Research 19, Chang, S. (1998) Takeovers of privately held targets, method of payment, and bidder returns, Journal of Finance 53, Chen, C. and N. Mohan (1994) Timing the disclosure of information: Management s view of earnings announcements, Financial Management, 23, Chen, S. and Y. Wang (2012). Financial constraints and share repurchases, Journal of Financial Economics, 105,

27 Cooper, M., O. Dimitrov, and O. Rau (2001) A Rose.com by any other name, Journal of Finance, 56, Damodaran, A. (1989) The weekend effect in information releases: A study of earnings and dividend announcements, Review of Financial Studies, 2, Dechow, P., R. Sloan, and A. Sweeney (1996) Causes and consequences of earnings manipulation: an analysis of firms subject to enforcement actions by SEC, Contemporary Accounting Research, 13, DeHaan, E., T. Shevlin, and J. Thornock (2015) Market (in)attention and the strategic scheduling and timing of earnings announcements, Journal of Accounting and Economics, 60, Dellavigna, S., and J. Pollet (2009) Investor inattention and Friday earnings announcements, Journal of Finance, 64, Doyle, J., and M., Magilke (2009) The timing of earnings announcements: An examination of the strategic disclosure hypothesis, Accounting Review, 84, Firth, M. (1980) Takeovers, shareholder returns, and the theory of the firm, Quarterly Journal of Economics, 94, Foucault, T, and T. Gehrig (2008) Stock price informativeness, cross-listings and investment decisions, Journal of Financial Economics, 88, Huang, S., H. Zou, G. Hu, and J. Xiao (2016) What prices to learn from? A new test of the managerial learning hypothesis, Working Paper. Hull, M., and F. McGroarty (2014) Do emerging markets become more efficient as they develop? Long memory persistence in equity indices, Emerging Markets Review, 18,

28 Ivashina, V., and D. Scharfstein (2010) Bank lending during the financial crisis of 2008, Journal of Financial Economics, 97, Jacobs, J., and K. Gerson (2001) Overworked individuals or overworked families? Work and Occupations, 28, Jarrell, G., and A. Poulsen (1989) The returns to acquiring firms in tender offers: Evidence from 3 decades, Financial Management, 18, Kaplan, S., and M. Weisbach (1992) The success of acquisitions: Evidence from divestitures, Journal of Finance, 47, Kaplan, S. and L. Zingales (1997) Do investment-cash flow sensitivities provide useful measures of financing constraints? Quarterly Journal of Economics, 112, Kau, J., J. Linck, and P. Rubin (2008) Do managers listen to the market? Journal of Corporate Finance, 14, Louis, H., and A. Sun (2010) Investor inattention and the market reaction to merger announcements, Management Science, 56, Louis, H., and A. Sun (2016) Abnormal accruals and managerial intent: evidence from timing of merger announcements and completions, Contemporary Accounting Research, 33, Michaely, R., A. Rubin, and A. Vedrashko (2016) Selection bias and investor inattention on Friday, Journal of Financial Economics, 122, Moeller, S., F. Schlingemann, and R. Stulz (2005) Wealth destruction on a massive scale? A study of acquiring-firm returns in the recent merger wave, Journal of Finance, 60,

29 Patell, J., and M. Wolfson (1982) Good news, bad news, and the intraday timing of corporate disclosures, Accounting Review, 57, Penman, S. (1987) The distribution of earnings news over time and seasonalities in aggregate stock returns, Journal of Financial Economics, 18, Roll, R. (1986) The hubris hypothesis of corporate takeovers, Journal of Business, 59, Shleifer, A., and R. Vishny (2003) Stock market driven acquisitions, Journal of Financial Economics, 70, Stone, A., S. Schneider, and J. Harter (2012) Day-of-the-week mood patterns in the United States: On the existence of Blue Monday, Thanks God it s Friday and weekend effects, Journal of Positive Psychology, 7, Subrahmanyam, A., and S. Titman (1999) The going public decision and the development of financial markets, Journal of Finance, 54, Travlos, N. (1987) Corporate takeover bids, methods of payment, and bidding firms stock returns, Journal of Finance 42,

30 Table 1 Merger announcements per day of the week Monday Tuesday Wednesday Thursday Friday Number of obs % This table shows the number and the percentage of merger announcements per day of the week. We explore US public-to-public merger deals. Our sample period is between 1986 and

31 Table 2 Descriptive statistics of control variables Mean Median Min Max Number of obs. Size Stock payment Relative size Competitive Nasdaq ROA Diversified GDP growth Average Size Stock payment Relative size Competitive Nasdaq ROA Diversified GDP growth This table shows the descriptive statistics of the control variables used. Competitive are deals that take one for deals with multiple bidders according to SDC otherwise zero, diversified is a dummy that takes one for targets and bidders which do not share the same first two-digit SIC code, relative size is the ratio of target to bidder market capitalization, size is bidders market capitalization as captured by the natural logarithm of the market value of equity, stock is the percentage of stock payment as captured by its natural logarithm, ROA is the return on assets profitability of the acquirer, Nasdaq is a dummy that takes one for bidders being listed in Nasdaq otherwise zero, and GDP growth captures US economy s growth. We explore US public-to-public merger deals. Our sample period is between 1986 and

32 Table 3 Merger Announcements in four sub-periods (%) (1) (2) (3) (4) (1) vs (2) (1) vs (3) (1) vs (4) (2) vs (3) (2) vs (4) (3) vs (4) Monday Tuesday Wednesday * 1.08 Thursday Friday * -8.22*** *** -4.15** This table shows the percentage of merger announcements per day of the week. We explore US public-to-public merger deals. Our sample period is between 1986 and *, ** and *** indicate significance at the 10, 5 and 1% level respectively. 32

33 Table 4 Friday merger announcements: Multivariate results Y=0/1 where 1 for Friday merger announcements (1) (2) (3) Trend *** *** (-4.18) (-2.80) (-0.54) (-1.32) ** (-2.3) Size (-0.16) (-0.43) Stock payment (-1.37) (-1.33) Relative size (0.75) (0.68) Competitive (-1.13) (-1.01) Nasdaq ** ** (-2.10) (-2.26) ROA (0.72) (0.76) Diversified * * (-1.93) (-1.89) GDP growth (0.18) (-0.19) Pseudo R Number of obs This table shows Friday merger announcements after controlling for number of characteristics. We estimate logistic regressions where the dependent variable is a dummy that takes one for Friday merger announcements, otherwise zero. Our interest is on the trend variable that captures the frequency on Fridays in relation to non- Fridays merger announcements. Competitive are deals that take one for deals with multiple bidders according to SDC otherwise zero, diversified is a dummy that takes one for targets and bidders which do not share the same first two-digit SIC code, relative size is the ratio of target to bidder market capitalization, size is bidders market capitalization as captured by the natural logarithm of the market value of equity, stock is the percentage of stock payment as captured by its natural logarithm, ROA is the return on assets profitability of the acquirer, Nasdaq is a dummy that takes one for bidders being listed in Nasdaq otherwise zero, and GDP growth captures US economy s growth. T-statistics are shown in parentheses. We explore US public-to-public merger deals. Our sample period is between 1986 and *, ** and *** indicate significance at the 10, 5 and 1% level respectively. 33

34 Table 5 Results when using alternative merger data restrictions Y=0/1 where 1 for Friday merger announcements Stock swaps Public-to-private Trend ** *** (-2.39) (-4.52) (-0.10) (-0.06) ** (-0.96) (-2.04) * *** (-1.87) (-3.40) Size *** *** (0.44) (0.2) (-3.33) (-3.65) Stock payment *** *** (-0.42) (-0.32) (-3.31) (-3.21) Relative size (1.31) (1.27) Competitive ** ** (-1.61) (-1.60) (-2.18) (-2.08) Nasdaq *** *** (-1.15) (-1.32) (-2.64) (-2.80) ROA (0.61) (0.70) (0.25) (0.36) Diversified (-0.84) (-0.84) (0.96) (1.02) GDP growth (-0.32) (-0.78) (1.12) (0.02) Pseudo R Number of obs This table shows whether results of Table 4 holds within alternative merger data restrictions. We report results for stock swaps which are deals with at least 50% funded with stock, and for public-to-private merger deals. We estimate logistic regressions where the dependent variable is a dummy that takes one for Friday merger announcements, otherwise zero. Our interest is on the trend variable that captures the frequency on Fridays in relation to non-fridays merger announcements. Competitive are deals that take one for deals with multiple bidders according to SDC otherwise zero, diversified is a dummy that takes one for targets and bidders which do not share the same first two-digit SIC code, relative size is the ratio of target to bidder market capitalization, size is bidders market capitalization as captured by the natural logarithm of the market value of equity, stock is the percentage of stock payment as captured by its natural logarithm, ROA is the return on assets profitability of the acquirer, Nasdaq is a dummy that takes one for bidders being listed in Nasdaq otherwise zero, and GDP growth captures US economy s growth. T-statistics are shown in parentheses. Our sample period is between 1986 and *, ** and *** indicate significance at the 10, 5 and 1% level respectively. 34

35 Table 6 Managerial learning on prior investor inattention Y=0/1 where 1 for Friday merger announcements (1) (2) (3) (4) DNegativeCAR(-1,1) * DNegative returns prior inattention * *** 1.642* (2.66) (1.83) DNegativeCAR(-1,1) * DNegative volume prior inattention * ** (2.42) (1.60) DNegativeCAR(-1,1) * (1.51) (0.79) (1.76) (1.25) DNegative returns prior inattention 0.952* (1.84) (1.19) DNegative volume prior inattention (1.39) (1.52) (0.75) (0.57) (0.80) (0.90) DNegativeCAR(-1,1) * DNegative returns prior inattention ** (-2.34) (-1.19) DNegativeCAR(-1,1) * DNegative volume prior inattention ** (-2.35) (-1.61) DNegativeCAR(-1,1) * * * (-1.79) (-1.17) (-1.81) (-1.08) DNegative returns prior inattention * ** * (-2.18) (-1.67) DNegative volume prior inattention * * * (-1.86) (-1.75) Size (-0.88) (-0.78) Stock payment (-1.00) (-1.14) Relative size (0.31) (0.35) Competitive (-0.85) (-0.83) Nasdaq ** ** (-2.50) (-2.50) ROA (0.68) (0.65) Diversified * * (-1.80) (-1.94) GDP growth (1.44) (1.64) Pseudo R Number of obs This table shows whether managers learning on prior investor inattention has improved after We use a dummy to separate merger announcements of good and poor quality deals based on the cumulative abnormal stock returns around the merger announcement day (-1 and 1 days, where day 0 is the merger announcement). We identify good quality merger deals as those with non-negative cumulative abnormal stock returns, while counterpart poor quality merger deals as those with negative cumulative abnormal stock returns. We also define high prior investor inattention when there is an underreaction on Friday stock returns and trading volume in relation to the same period stock price reaction on other days of the week. We use dummies to separate the groups. We estimate a logit model where the dependent variable is a dummy variable that takes one when a Friday merger announcement is experienced, while the key independent variables are a dummy for negative announcement bidders, a dummy for bidders that the difference in stock returns on Fridays minus non-fridays was negative in prior 120 merger deals, and a dummy after the introduction of EDGAR since We interact these three variables; we interact return/volume prior inattention * merger deals with poor announcement of bidders * merger deals after All control variables are added (for further details, please study earlier tables). T-statistics are shown in parentheses. We explore US public-to-public merger deals. Our sample period is between 1986 and *, ** and *** indicate significance at the 10, 5 and 1% level respectively. 35

36 Table 7 Managerial learning on prior investor inattention; Settings with more/less pronounced results Y=0/1 where 1 for Friday merger announcements Low date diff High date diff Low date diff High date diff Small size Large size DNegativeCAR(-1,1) * DNegative returns prior inattention * ** 1.657* (2.08) (1.82) DNegativeCAR(-1,1) * DNegative volume prior inattention * * ** (1.88) (1.58) (2.34) (0.02) DNegativeCAR(-1,1) * (1.11) (1.08) (1.20) (1.29) (1.66) (0.19) DNegative returns prior inattention (1.59) (1.12) DNegative volume prior inattention (1.15) (0.90) (1.57) (0.28) (0.54) (0.66) (0.56) (0.58) (1.13) (0.00) DNegativeCAR(-1,1) * DNegative returns prior inattention * (-1.54) (-1.79) DNegativeCAR(-1,1) * DNegative volume prior inattention * ** (-1.57) (-1.78) (-2.18) (-0.10) DNegativeCAR(-1,1) * * (-1.60) (-1.16) (-1.48) (-1.16) (-1.66) (-0.08) DNegative returns prior inattention * * (-1.85) (-1.36) DNegative volume prior inattention * * ** (-1.72) (-1.00) (-1.98) (-0.23) Pseudo R Number of obs This table shows managers learning on prior investor inattention after 1994 within small versus large size bidders and within for merger deals with a short versus a long distance in days between the announcement and the completion. In each occasion, we use the median value to split the full sample into different groups. We use a dummy to separate merger announcements of good and poor quality deals based on the cumulative abnormal stock returns around the merger announcement day (-1 and 1 days, where day 0 is the merger announcement). We identify good quality merger deals as those with non-negative cumulative abnormal stock returns, while counterpart poor quality merger deals as those with negative cumulative abnormal stock returns. We also define high prior investor inattention when there is an underreaction on Friday stock returns and trading volume in relation to the same period stock price reaction on other days of the week. We use dummies to separate the groups. We estimate a logit model where the dependent variable is a dummy variable that takes one when a Friday merger announcement is experienced, while the key independent variables are a dummy for negative announcement bidders, a dummy for bidders that the difference in stock returns on Fridays minus non-fridays was negative in prior 120 merger deals, and a dummy after the introduction of EDGAR since We interact these three variables; we interact prior return/volume inattention * merger deals with poor announcement of bidders * merger deals after T-statistics are shown in parentheses. We explore US public-to-public merger deals. Our sample period is between 1986 and *, and ** indicate significance at the 10 and 5% level respectively. 36

37 Table 8 The significance of corporate governance in the relation between managerial learning and prior investor inattention Y=0/1 where 1 for Friday merger announcements CEO s duality=1 CEO s duality=0 CEO s duality=1 CEO s duality=0 Low board independence High board independence Low board independence High board independence DNegative returns prior inattention* ** ** DNegativeCAR(-1,1) (2.16) (0.19) (2.42) (0.57) DNegative volume prior inattention* DNegativeCAR(-1,1) (0.14) (0.54) (0.28) (-0.25) DNegativeCAR(-1,1) * ** (-1.67) (-0.55) (0.00) (-0.76) (-2.25) (-1.16) (-0.38) (-0.36) DNegative returns prior inattention ** (-2.30) (-0.03) (-1.26) (-1.02) DNegative volume prior inattention * (-1.77) (0.53) (-0.52) (0.21) Pseudo R Number of obs This table shows managers learning on prior investor inattention within firms with high and low corporate governance. We capture corporate governance using CEO s duality and board independence. Firms with high corporate governance are those with non-ceo s duality and with high board independence. We use a dummy to separate merger announcements of good and poor quality deals based on the cumulative abnormal stock returns around the merger announcement day (-1 and 1 days, where day 0 is the merger announcement). We identify good quality merger deals as those with non-negative cumulative abnormal stock returns, while counterpart poor quality merger deals as those with negative cumulative abnormal stock returns. We also define high prior investor inattention when there is an underreaction on Friday stock returns in relation to the same period stock price reaction on other days of the week. We use dummies to separate the groups. We estimate a logit model where the dependent variable is a dummy variable that takes one when a Friday merger announcement is experienced, while the key independent variables are a dummy for negative announcement bidders, a dummy for bidders that the difference in stock returns on Fridays minus non-fridays was negative in prior 120 merger deals. T-statistics are shown in parentheses. We explore US public-to-public merger deals. Our sample period in this test is between 2000 and 2013 in line with Boardex data availability of CEO s duality and board independence. *, and ** indicate significance at the 10, and 5% level respectively. 37

38 Table 9 Investor reaction in the days following initial underreaction Acquirers Targets Friday Non-Friday Diff Friday Non-Friday Diff Panel A: Full period results ** ** CAR(0,1) ** CAR(0,2) ** CAR(0,3) ** CAR(0,4) * Panel B: Sub-period results * CAR(0,1) CAR(0,2) CAR(0,3) CAR(0,4) * * This table explores whether investors cover up the initial underreaction to Friday merger announcements on the following days. Panel A compares the average returns on Fridays versus non-fridays as well on the days following the merger announcement for bidders and targets over the full sample period. As an example, CAR(0,1) indicates the cumulative abnormal stock returns for bidders and targets between Friday and coming Monday for Friday merger announcements. Panel B explores the changes of these differences across the sample period. We explore US public-to-public merger deals. T-statistics are shown in parentheses. Our sample period is between 1986 and *, and ** indicate significance at the 10 and 5% level respectively. 38

39 Table 10 Wealth announcement effect of bidders after the financial crisis CAR(-1,1) *** (3.50) Size -0.69*** (-7.07) Stock payment -0.65*** (-7.88) Relative size -0.99*** (-6.30) Competitive (-0.49) Nasdaq -1.10*** (-3.23) ROA 3.20*** (3.77) Diversified (-1.07) GDP growth 0.11 (1.01) Pseudo R Number of obs This tables shows the announcement wealth effect of bidders after the financial crisis. CAR(-1,1) indicates the interval period around each merger announcement (-1,1 days where 0 is the merger announcement day). We control for competitive which are deals that take one for deals with multiple bidders according to SDC otherwise zero, for diversified that is a dummy that takes one for targets and bidders which do not share the same first twodigit SIC code, for relative size that is the ratio of target to bidder market capitalization, size that is bidders market capitalization as captured by the natural logarithm of the market value of equity, for stock that is the percentage of stock payment as captured by its natural logarithm, for ROA that is the return on assets profitability of the acquirer, for Nasdaq that is a dummy that takes one for bidders being listed in Nasdaq otherwise zero, and for GDP growth that captures US economy s growth. T-statistics are shown in parentheses. We explore US public-to-public merger deals. Our sample period is between 1986 and *** indicates significance at the 1% level. 39

40 Figure 1 Annual % of Friday merger announcements during the sample period This figure shows the annual percentage of Friday merger announcements and the number of public-to-public announcements during our sample period,

41 Figure 2 Annual % of Friday merger announcements for good and poor quality merger deals This figure shows the annual percentage of Friday merger announcements separately for good merger deals as indicated with the positive abnormal stock returns on the merger announcement (CAR(-1,1)) versus for poor merger deals, also for merger at the bottom 20% announcement returns versus the top 20%. We explore US public-to-public merger deals. Our sample period is between 1986 and

42 Figure 3 Investor inattention during the sample period A. Bidders B. Targets This figure shows investor inattention over the sample period. We estimate abnormal stock returns and abnormal trading volume on merger announcements on Fridays minus the corresponding stock returns and trading volume in the announcement of mergers in non-fridays. We split results for positive and negative bidders announcement effect on Fridays and on non-fridays, since the difference in stock returns would indicate the opposite relation. We use 120 prior merger deals to identify those deals with positive and negative announcement effect. We explore US public-to-public merger deals. Our sample period is between 1986 and

43 Figure 4 Managerial learning and prior investor inattention per sub-period This figure shows sub-period results whether managers learn from prior inattention on stock returns and trading volume to announce a bad merger deal later. We estimate a logit model where the dependent variable is a dummy variable that takes one when a Friday merger announcement is experienced, while the key independent variables are a dummy for negative announcement bidders, and a dummy for bidders that the difference in stock returns on Fridays minus non-fridays was negative in prior 120 merger deals. We interact these two variables and add all our control variables; Competitive are deals that take one for deals with multiple bidders according to SDC otherwise zero, diversified is a dummy that takes one for targets and bidders which do not share the same first two-digit SIC code, relative size is the ratio of target to bidder market capitalization, size is bidders market capitalization as captured by the natural logarithm of the market value of equity, stock is the percentage of stock payment as captured by its natural logarithm, ROA is the return on assets profitability of the acquirer, Nasdaq is a dummy that takes one for bidders being listed in Nasdaq otherwise zero, and GDP growth captures US economy s growth. We estimate this regression over the following interval periods: , , , , and This figure only reports the parameter coefficient of the interaction term; prior inattention on stock returns / trading volume * merger deals with poor announcement of bidders per sub-periods. We explore US public-to-public merger deals. 43