Managerial ability and acquirer returns

Accepted Manuscript Title: Managerial Ability and Acquirer Returns Authors: Sheng-Syan Chen, Chih-Yen Lin PII: DOI: Reference:

S1062-9769(17)30048-0 https://doi.org/10.1016/j.qref.2017.09.004 QUAECO 1076

To appear in:

The

Received date: Revised date: Accepted date:

28-1-2017 14-7-2017 22-9-2017

Quarterly

Review

of

Economics

and

Finance

Please cite this article as: Chen, Sheng-Syan., & Lin, Chih-Yen., Managerial Ability and Acquirer Returns.Quarterly Review of Economics and Finance https://doi.org/10.1016/j.qref.2017.09.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Managerial Ability and Acquirer Returns by Sheng-Syan Chen1, and Chih-Yen Lin2 1. Sheng-Syan Chen is a professor in the Department of Finance, College of Management, National Taiwan University, No. 85, Sec. 4, Roosevelt Rd., Taipei, Taiwan. Tel: +886-2-33661083; fax: +886-2-33661519; e-mail: [email protected]. 2. Address correspondence to Chih-Yen Lin, Department of Economics, College of Social Sciences at Fu Jen Catholic University, No. 510, Zhongzheng Road, Xinzhuang Dist., New Taipei City, Taiwan. Tel: +886-2-29052695; Fax:+886-2-29052188;e-mail: [email protected]. We wish to thank Narjess Boubakri (the editor), two anonymous reviewers, Jonathan Batten, Yan-Shing Chen, and Chia-Wei Huang for helpful comments and suggestions. We thank Peter R. Demerjian, Baruch Lev, and Sarah E. McVay for providing the data on managerial ability. We are also indebted to Cláudia Custódio, Miguel A. Ferreira, and Pedro Matos for sharing their general ability data. Chih-Yen Lin gratefully acknowledges financial support from the Ministry of Science and Technology of Taiwan (MOST 104-2410-H-030-010).

Highlights  We examine the impact of managerial ability on the profitability of mergers and 







acquisitions. We use a new managerial ability measure that suffers less than conventional managerial ability measures from measurement error. We find a positive relationship between managerial ability, announcement abnormal returns and long-term buy-and-hold abnormal returns. We find a negative relationship between managerial ability and takeover premiums. High environmental uncertainty is an important scenario that should be included in studies investigating the influence of managerial ability.

ABSTRACT This paper examines the impact of managerial ability on the profitability of mergers and acquisitions. We find that acquisitions by firms with high managerial ability generate better announcement abnormal returns as well as better post-announcement abnormal returns than deals by firms with low managerial ability. We also find deals with high managerial ability pay significantly lower premiums than deals without. Further, we find that managers with high managerial ability perform better in scenarios with high environmental uncertainty, which suggests that high environmental uncertainty is an important scenario that should be incorporated into 1

studies of the influence of managerial ability. JEL classification: G14; G34 Keywords: Managerial ability; Mergers and acquisitions; Environmental uncertainty; Shareholder wealth

1. Introduction Economics and finance studies have demonstrated that heterogeneity of chief executive officers (CEOs) affects firm policies and associated firm performance (e.g., Adams, Almeida, & Ferreira, 2005; Bertrand & Schoar, 2003; Cain & McKeon, 2016; Kaplan, Klebanov, & Sorensen, 2012). Given the importance of CEOs in relation to the influence of shareholder wealth on mergers and acquisitions (M&As), a growing literature has examined whether managerial characteristics affect acquirer returns (e.g., Custódio & Metzger, 2013; Jacobsen, 2014; Malmendier & Tate, 2008). A number of studies have examined and demonstrated the importance of managerial ability based on the relative efficiency among industry peers on various financial aspects, such as bank liquidity creation and risk-taking analysis (Andreou, Philip, & Robejsek, 2016), corporate tax avoidance (Koester, Shevlin, & Wangerin, 2016), credit risk assessment (Bonsall, Holzman, & Miller, 2017), earnings quality (Demerjian et al., 2013), and relative peer quality (Francis et al., 2016). Few studies to date, however, have investigated the influence of such managerial ability on M&A returns. The question therefore remains whether managerial abiltiy is an important factor in the determinants of acquirer returns. Thus far, studies have given little attention to whether managerial ability matters

2

in M&A events.1 We respond to this void by focusing on managerial ability based on the relative efficiency among industry peers in generating revenues from given corporate resources. Our specific focus is on how managerial ability affects acquirer returns. In a recent paper, using director outside board seats to examine whether the market rewards ability or experience of a CEO in M&As, Harford and Schonlau (2013) show that experience, rather than ability, is positively significant related to the outside board seats received by acquirer CEOs. This result seems to suggest that managerial ability is not an important factor considered by the market. On the other hand, Custódio and Metzger (2013) find that when acquirers’ CEOs have experience in the target industry, deals with industry-expert CEOs enjoy higher announcement abnormal returns than those without such experience. They also find that this phenomenon is caused by such industry-expert CEOs’ ability to bargain for better deal terms. Thus, prior studies, from different perspectives, find mixed evidence about whether managerial ability affects shareholder wealth in M&As. It is unclear whether managerial ability based on relative efficiency affect acquirers’ returns or not. Also, prior studies do not explore whether there is any scenario that can enhance the value creation of managerial ability to acquirer returrns. We expect firms with managers of greater managerial ability to conduct more profitable acquisitions than firms with lower managerial ability managers for at least two reasons. First, more able managers have better knowledge of the trends in and the ongoing environment of the industry, can more precisely predict product demand, and invest more in positive net present value (NPV) projects (e.g., Bonsall, Holzman, &

1

A few recent exceptions include Custódio and Metzger (2013) and Harford and Schonlau (2013). These studies, however, do not take into account the influence of managerial ability based on the relative efficieny among industry peers. 3

Miller, 2017; Demerjian, Lev, & McVay, 2012; Demerjian et al., 2013; Koester, Shevlin, & Wangerin, 2016). Better understanding of industry needs can enable managers to use corporate resources more efficiently to identify and structure profitable acquisitions. Second, these more able managers are presumably considered to be more capable of achieving their goals (e.g., Koester, Shevlin, & Wangerin, 2016). Acquisitions conducted by these more able managers are therefore more likely to be profitable. Although theoretically intuitive, the empirical identification of managerial ability is quite challenging. Prior studies mainly use manager fixed effects to measure managerial ability (e.g., Bertrand & Schoar, 2003), and then move on to media coverage (e.g., Francis et al., 2008; Milbourn, 2003) or firm performance (e.g., Masulis, Wang, & Xie, 2007). These proxies, however, may have introduced a measurement error into the findings, because they also contain factors beyond managers’ control, and thus do not exclusively describe managerial ability (e.g., Demerjian, Lev, & McVay, 2012). To overcome this limitation, we borrow and use a new managerial ability proxy developed by Demerjian, Lev, and McVay (2012) to study the effect of managerial ability on merger and acquisition performance. The managerial ability proxy developed by Demerjian, Lev, and McVay (2012) is calculated using the data envelopment analysis (DEA) as to how well a firm’s revenues, compared to those of its industry peers, can be generated using given corporate resources. Demerjian, Lev, and McVay (2012) document that a firm’s ability to generate revenues is attributable to two factors: firm efficiency and managerial ability. By separating out the effect of firm efficiency, Demerjian, Lev, and McVay’s (2012) managerial ability is a better measure of how well managers use corporate resources to generate revenue. There are two main advantages to using this proxy. 4

First, it has a strong correlation with manager fixed effects and thus can be better related to managerial characteristics. Second, it is a residual of firm efficiency, and is therefore less likely to correlate with managerial ability and corporate performance. As previously mentioned, this measure is better at identifying the relationship between managerial ability and corporate acquisition performance, and is less prone to measurement error. We follow the literature and use announcement abnormal returns to denote the outcome or profitability of an acquisition deal (e.g., Masulis, Wang, & Xie, 2007; Moeller, Schlingemann, & Stulz, 2004). This study also uses buy-and-hold abnormal returns (e.g., Krolikowski et al., 2017; Loughran & Vijh, 1997; Rau & Vermaelen, 1998) and takeover premiums (e.g., Hartzell, Ofek, & Yermack, 2004; Lin, Officer, & Zou, 2011) to denote the profitability of the M&A deals. The empirical results indicate that managerial ability can significantly affect an acquisition’s outcome. High managerial ability leads to deals with significantly positive announcement abnormal returns and long-term buy-and-hold abnormal returns. Further, we also find that deals with high managerial ability pay significantly fewer premiums than deals with low managerial ability. Besides testing the direct impact of managerial ability on the profitability of M&As, we also investigate what scenario can increase the effectiveness of managerial ability, an important topic that has yet to be investigated. We focus on environmental uncertainty. Our focus on environmental uncertainty is motivated by Bertrand and Schoar (2003), who find that managers perform differently on a given environment. We therefore expect that the extent to which more able managers can make more profitable acquisitions will be more likely to be observed in scenarios of high environmental uncertainty. The interaction terms between high environmental 5

uncertainty and managerial ability measures are found to be significantly positive in return analyses, and significantly negative in takeover premium analyses. These findings suggest that high environmental uncertainty is an important scenario that affects managerial ability and acquisition outcomes. We conduct a series of tests to check our results. We use a stochastic frontier analysis (SFA) to recalculate the managerial ability measure. We apply another managerial ability measure, the general ability index from Custódio, Ferreira, and Matos (2013). We also apply a propensity score matching technique to address the possible self-selection bias in our sample. We find similar results. This study makes several complementary contributions to the existing literature. First, this study contributes to a growing literature on whether and how managerial characteristics affect firm performance (e.g., Adams, Almeida, & Ferreira, 2005; Bertrand & Schoar, 2003; Custódio & Metzger, 2013; Jacobsen, 2014; Kaplan, Klebanov, & Sorensen, 2012; Malmendier & Tate, 2008). The current study suggests managerial ability can affect firm value through conducting M&As. In addition, few empirical studies have investigated whether managerial ability matters in corporate M&A events (e.g., Custódio & Metzger, 2013; Harford & Schonlau, 2013), and which circumstances increase or decrease the impact of managerial ability on M&As. This is the first study to evaluate the effect of managerial ability based on the relative efficiency on M&A performance, using a large sample and covering a longer period. Further, this study also takes into account the possible influence from the self-selection bias. In addition, this paper also suggests evidence on Custódio, Ferreira, and Matos (2013) as to why managers with higher general ability can receive more compensation, especially in high environmental uncertainty; those more able managers can make more profitable acquisitions. 6

The rest of the paper is organized as follows. Section 2 documents our data and methodology. Section 3 provides the empirical results as well as robustness checks. We provide our conclusions in Section 4. 2. Data and methodology 2.1. M&A sample We focus on domestic M&As and only retain acquirers that are U.S. firms. The initial sample is from the U.S. Mergers and Acquisitions Database of Thomson Financial’s Securities Data Corporation (SDC). The M&A announcements and other M&A related information come from SDC. As in prior M&A studies (e.g., Deng, Kang, & Low, 2013; Uysal, 2011), financial and utility firms are excluded. We also require the transaction value of the acquisition to be at least US$1 million. The final sample period covers the years 1991 to 2013. To complement the data analysis, we also extract relevant information from the Center for Research in Security Prices (CRSP), Compustat, and ExecuComp. 2.2. Managerial ability We use the managerial ability measure developed by Demerjian, Lev, and McVay (2012).2 Demerjian, Lev, and McVay (2012) adopt the following two steps to develop a managerial ability proxy (MA-Score). First, they use DEA to estimate firm efficiency. Firm efficiency measures how well a firm generates revenue from existing resources compared to firms within the same industry. The definition of industry is based on Fama and French (1997). The solution to the following optimization is obtained: 2

We thank Peter R. Demerjian for sharing the data on managerial ability on his website, http://faculty.washington.edu/pdemerj/data.html. 7

Max y λ 

Sales y1COGS  y2 SGA  y3 PPE  y4OPL  y5 R & D  y6GD  y7OTI

(1)

The optimization in the Eq. (1) identifies the optimal weights for the seven inputs as the firm-specific vector, y. This DEA produces the firm efficiency measure, λ, which ranges from 1 (the most efficient) to 0 (the most inefficient). Within the equation, there are seven input variables: COGS denotes cost of goods sold, SGA denotes selling and administrative expenses, PPE denotes net fixed assets, OPL denotes net operating lease, R&D denotes research and development, GD denotes purchased goodwill, OTI denotes other intangible assets. Second, because this firm efficiency measure contains both firm and manager efforts, to measure the managerial ability, a Tobit regression analysis is used to exclude firm characteristics. Specifically, the following regression is estimated by industry: FirmEffici ency   0  1 LN (TA)   2 MS   3 PFCF   4 LN ( AGE )   5 BSC   6 FCD  Years  

(2)

The residual from the Eq. (2) is the corresponding MA-Score. Within the equation, TA denotes total assets, MS denotes a firm’s market share within the industry, PFCF is a dummy variable that is equal to 1 if a firm has positive free cash flow, AGE denotes a firm’s age, BSC denotes the concentration of the business segments, FCD is a dummy variable that is equal to 1 if a firm has foreign operations, and Years denotes year dummies. 2.3. Short-term abnormal stock returns Following the literature (e.g., Masulis, Wang, & Xie, 2007; Wang & Xie, 2009), the short-run market reaction to an M&A event is measured as 5-day cumulative abnormal returns (CAR). CAR is measured from day -2 to day +2 and day 0 is the 8

corresponding M&A announcement date. Each day’s abnormal return is measured as its daily return minus its expected return. The expected return of each day is calculated using the parameters estimated from the market model, with the estimation period from day -210 to day -11. We use CRSP value-weighted index return as the market index proxy. CAR is the sum of these abnormal returns around the event window. 2.4. Long-term abnormal stock returns We use completed deals to calculate the associated long-term buy-and-hold abnormal returns to estimate the long-term abnormal stock performance after M&A events as suggested in the literature (e.g., Krolikowski et al., 2017; Loughran & Vijh, 1997; Rau & Vermaelen, 1998). We first calculate the separate 1-year buy-and-hold returns for sample firms and matching firms. The calculation period starts from the first trading day after the announcement date of the M&A event to the first anniversary, where each year is defined to be 250 trading days.3 Matching firms must meet the following criteria: listed on the same stock exchange as a sample firm and making no acquisition in the previous three years. We also require matching firms to be within the same industry as defined in Fama and French (1997), and in the same size decile and book-to-market quintile as the sample firm.4 Because using only one matching firm to calculate long-term returns may give rise to the noisy point estimate problem (e.g., Lyon, Barber, & Tsai,1999; Savor & Lu, 2009), we select the 10 matching firms with the closest book-to-market ratio as the 3

We have similar results if we start at the one trading day before or at the announcement date, or at the effective date to calculate the associated long-term abnormal stock performance. 4 We switch to require matching firms to be within the same size decile and book-to-market quintile as a sample firm if no firm can meet the industry match. Here, the size is measured as the market value of common stocks at two trading days before the announcement, and the book-to-market ratio is calculated as the ratio of book equity divided by its market capitalization at the end of the previous month (e.g., Savor & Lu, 2009). 9

sample firm (e.g., Savor & Lu, 2009). We take the average of the associated buy-and-hold returns of the 10 matching firms as the benchmark portfolio.5 We use the difference between the sample firm and its benchmark’s buy-and-hold returns to calculate the associated long-term abnormal returns (BHAR). 2.5. Takeover premium We follow the literature to require targets to be public firms in order to calculate the takeover premium. We calculate the takeover premium (Premium) as the offer price minus the target’s stock price measured at 4 weeks before the M&A announcement divided by the target’s stock price measured at 4 weeks before the M&A announcement (e.g., Hartzell, Ofek, & Yermack, 2004; Lin, Officer, & Zou, 2011).6 2.6. Environmental uncertainty For environmental uncertainty, we follow Custódio, Ferreira, and Matos (2013), who state that higher industry shocks lead to higher environmental uncertainty for business operations. Industry shocks are measured as the difference between each industry’s sale growth minus that of the benchmark, which is the average sales growth calculated using all industries. For each year, we rank the industry shocks into quintiles. Industries belonging to the highest industry shock quintiles are considered to be the highest environmentaly uncertain industries. High uncertainty is a dummy variable that is equal to 1 if a firm’s industry belongs to the highest quintile of the

5

The benchmark portfolio may contain less than 10 matching firms if there are less than 10 matching firms available. We have similar results if we identify matching firms based on the criteria of size and book-to-market ratio only, or if we only use one matching firm. 6 The results are similar if we follow Wang and Xie (2009) to measure the takeover premium as the offer price minus the target’s stock price measured at 1 week before the M&A announcement divided by the target’s stock price measured at 1 week before the M&A announcement. 10

industry shock ranking.7 2.7. Control variables Following Cai and Sevilir (2012), Harford (1999), Malmendier and Tate (2008), Masulis, Wang, and Xie (2007), and Uysal (2011), we also include a set of control variables. The definitions of these control variables can be found in the Appendix. [Insert Table 1 about here] Table 1 presents the summary characteristics of the whole sample. We winsorize all continuous variables at the first and 99th percentiles to address the possible influence of outliers. Panel A contains the statistics for the dependent variables. The average CAR is about 0.3%, and the median BHAR is about -17% on a 1-year period after M&A announcements. These performance results are consistent with the literature; acquirers generally earn not statistically significant announcement abnormal returns and generally enjoy negative long-term abnormal stock performance (e.g., Cai & Sevilir, 2012; Rau & Vermaelen, 1998; Savor & Lu, 2009). Panel A also shows that the average takeover premium of the sample is about 34.6%, which is quite close to what the literature shows about takeover premiums; the average premium, for example, is about 34.8% on Hartzell, Ofek, and Yermack (2004), and about 38.4% on Lin, Officer, and Zou (2011). Panel B shows the characteristics for the whole sample. About 58% of the sample is from the same industry. About 13% of the whole sample is from high uncertainty industries. 3. Empirical analyses

7

The results are qualitatively similar if we instead use quartile or median to classify the sample. 11

3.1. The empirical results for short-term abnormal stock returns 3.1.1. Univariate analysis For each year, we use all Compustat firms to rank managerial ability (MA-Score). We classify these according to whether or not an acquiring firm is above the whole sample median.8 High managerial ability is a dummy variable which is equal to 1 whose firms have managerial ability above the median of the whole for Compustat firms. We classify the upper (lower) half as the high (low) ability group. For each group, we perform standard t-tests for the mean and Wilcoxon sign-rank tests for the median. For the difference between groups, we perform t-tests for the mean difference and Wilcoxon rank-sum tests for the median difference. Table 2 presents the corresponding analyses of 5-day CAR based on different managerial ability measures. The findings are similar if we use other time periods to calculate short-term abnormal returns, such as 2-day or 1-day. Panel A shows the results for the managerial ability measure. We find that high managerial ability has a significant positive impact on deal announcement abnormal returns. Deals with high managerial ability on average enjoy higher CAR (by 0.7%) than deals with low managerial ability, which is statistically significant at the 1% level. [Insert Table 2 about here] Panels B and C also present the test of the partitions based on the managerial ability measure and environmental uncertainty scenarios. We find that deals with high managerial ability continue to perform better than those without in both scenarios, and that the impact is more pronounced in high environmental uncertainty scenarios. The average difference between high managerial ability and low managerial ability on 8

The results are similar if we use the top quintile or quartile to denote high ability managers. 12

CAR is about 1.4% in high environmental uncertainty scenarios, whereas the average difference between high managerial ability and low managerial ability on CAR is about 0.7% in low environmental uncertainty scenarios.9 Further, we also find the difference between high managerial ability and low managerial ability in Panel B is higher than that in Panel A, with a difference of about 0.7% in the mean and 0.9% in the median. The results indicate that the market attributes greater value to managers in high environmental uncertainty scenarios. 3.1.2. Regression analysis Since univariate analysis may suffer from other confounding factors, we use a regression analysis to investigate the role of managerial ability associated with short-term abnormal returns on M&A events. We use standard errors adjusted for White heteroskedasticity (White, 1980) and clustered at the acquirer level, as suggested by Petersen (2009) to address possible serial correlation in standard errors. We also control for industry and year fixed effects. Model 1 includes only the managerial ability measure. Model 2 includes the managerial ability measure, the environmental uncertainty dummy, and their corresponding interaction term. Model 3 includes the proposed managerial ability measure, the environmental uncertainty dummy, the corresponding interaction term, and the other control variables. [Insert Table 3 about here] Table 3 shows the corresponding regression results analyzing the influence of High managerial ability on CAR, controlling for other confounding factors. In Model 1, we find that High managerial ability has a significant positive relationship to CAR

9

We use the bottom quintile of the industry shocks as the low environmental uncertainty scenario. We obtain qualitatively similar results if we use the bottom median or quartile of industry shocks to define the low environmental uncertainty. 13

at the 1% level (the coefficient estimate is 0.003). This result indicates that deals with high managerial ability create more value for shareholders. In Model 2, we find that the interaction term between High managerial ability and High uncertainty is significantly positive at the 1% level (the coefficient estimate is 0.011). This result indicates managerial ability is more valuable in high environmental uncertainty, which is consistent with Custódio, Ferreira, and Matos’ (2013) conjecture that firms generally give higher compensation to executives in high uncertainty environments in return for risks taken. Model 3 further includes other control variables. We still find that deals with high managerial ability create more value for shareholders, and the impact is more pronounced in high environmental uncertainty scenarios. Model 3 also shows some significant control variables. The coefficients of Firm size and Public are significantly negative, and the coefficients of Leverage and Subsidiary are significantly positive. These results are consistent with the findings of previous literature that deals with bigger acquirer size generally enjoy negative CAR (e.g., Moeller, Schlingemann, & Stulz, 2004), deals with public (subsidiary) targets generally receive negative (positive) CAR (e.g., Fuller, Netter, & Stegemoller, 2002), and deals with higher acquirer leverage enjoy higher CAR because the higher leverage can have a disciplinary effect on acquirers (e.g., Field & Mkrtchyan, 2017). 3.2. The empirical results for long-term abnormal stock returns 3.2.1. Univariate analysis Our results that deals with high managerial ability outperform deals with low managerial ability suggest managerial ability can be beneficial to the acquisition process. To check whether such phenomenon is only temporary, we also examine

14

whether the deals with high managerial ability still enjoy higher long-term abnormal stock performance.10 [Insert Table 4 about here] Table 4 presents the corresponding univariate results of BHAR. Panel A shows the results based on High managerial ability classification. We find deals with high managerial ability outperform deals with low managerial ability in an economically significant way. Over the 1-year horizon starting at the first trading day after the deal announcement date, acquirers with high managerial ability outperform those without by about 12.2%, on average. The results provide further support to our examination of short-term abnormal returns, and indicate that deals with high managerial ability continue to enjoy higher post-announcement returns than deals without. Panels B and C show the results based on the High managerial ability classification in relation to high environmental uncertainty and low environmental uncertainty scenarios, respectively. We find that deals with high managerial ability still enjoy higher BHAR than those without in both scenarios. Further, the impact of managerial ability on BHAR is more pronounced in high environmental uncertainty scenarios. On the subsample of high environmental uncertainty scenarios, we find that deals with high managerial ability perform significantly better than those without, with the mean (median) difference about 17.1% (20.6%). Conversely, the mean and median differences are about 13.1% and 10.9% in low environmental uncertainty scenarios. Further, we also find that the difference between high managerial ability and low managerial ability in Panel B is higher than that in Panel A, with differences of about 4.9% to the mean and 9.5% to the median. The results in Panels B and C suggest that managerial ability is more valuable in high environmental uncertainty 10

We thank an anonymous reviewer for making this point and suggesting this examination. 15

scenarios. 3.2.2. Regression analysis Because the univariate analysis may suffer from the influence of other confounding factors, we also use a regression analysis to check our results. We use the same model setup as in Table 3. Table 5 presents the corresponding regression results of the analysis of managerial ability in relation to BHAR, controlling for other confounding factors. All results are robust to switch to a 3-year or 5-year period to calculate BHAR. [Insert Table 5 about here] Across different models, we continue to find that deals with high managerial ability enjoy statistically significant higher returns than deals without, ranging from 11.8% to 12.5%. Given that the average BHAR of the whole sample is about -0.1%, the results indicate that managerial ability can have an economically significant impact on acquisition deals. We also find that such phenomena are more pronounced in the high environmental uncertainty scenario. The interaction term between High managerial ability and High uncertainty is significantly positive at the 1% level, which provides further support for the proposition that managerial ability is more valuable in the high environmental uncertainty scenario. Table 5 also shows some significant control variables. The coefficients of Same industry dummy, All cash deal, and Subsidiary are significantly positive, while the coefficients of Firm size and Public are significantly negative. The results are consistent with the literature that has found that larger firms have inferior long-term abnormal stock performance (e.g., Alexandridis, Antoniou, & Petmezas, 2007), public targets have inferior long-term abnormal stock returns (e.g., Rau & Vermaelen, 1998), 16

and cash payment deals have better long-term abnormal stock returns (e.g., Dutta & Jog, 2009). 3.3. The empirical results for takeover premiums 3.3.1. Univariate analysis If deals with high managerial ability can enjoy higher returns, they are supposed to be less likely to overpay. We therefore conjecture that deals with high managerial ability, on average, pay lower premiums. [Insert Table 6 about here] Table 6 presents the corresponding univariate results of Premium. Panel A shows the results based on High managerial ability classification. The results indicate that deals with high managerial ability pay fewer premiums than deals without. The mean difference (about -11.7%) and median difference (about -10.7%) are both negatively statistically significant at the 1% level. One possible source of value creation is therefore that higher managerial ability can avoid overpayment to targets. Panels B and C show the results based on the High managerial ability classification in relation to high environmental uncertainty and low environmental uncertainty scenarios, respectively. We find that deals with high managerial ability still pay lower premiums in both scenarios. The average difference between high managerial ability and low managerial ability in relation to premiums is about -23.9% in high environmental uncertainty scenarios, whereas, in low environmental uncertainty scenarios it is about -11.5%. Further, we also find that the difference between high managerial ability and low managerial ability in Panel B is lower than that in Panel A, with a difference of about -12.2% in the mean and -16.3% in the median. 17

3.3.2. Regression analysis We also use a regression to analyze the influence of managerial ability on Premium, controlling for other confounding factors. Table 7 presents the corresponding regression results. All regressions control for industry and year effects. [Insert Table 7 about here] Our results for Model 1 suggest that higher managerial ability can significantly affect takeover premiums, and the coefficient estimate of High managerial ability is significantly negative at the 1% level (the coefficient estimate is -0.115). Further, our results on short-term and long-term stock performance suggest that higher managerial ability is especially valuable in the high environmental uncertainty scenario. We therefore also add an interaction term between High managerial ability and High uncertainty. The results for Model 2 indicate that for deals under high environmental uncertainty, the presence of high managerial ability can significantly lower the acquisition premiums. Model 3 further controls for other confounding factors, including acquirer and target characteristics. We find similar results. Table 7 also shows All cash deal is significantly positive. The results are consistent with the literature’s finding that cash payment deals pay higher takeover premiums (e.g., Lin, Officer, & Zou, 2011). The overall results in Tables 6-7 provide further support for our previous analyses of acquirer returns and provide one possible source of value creation for why high managerial ability can enhance acquirer returns—they can avoid the overpayment of deals. 3.4. Robustness checks 3.4.1. Parametric methods to estimate managerial ability 18

There is a well-established literature on efficiency measurements using either parametric models, such as SFA, or non-parametric models, such as DEA, in operational research and other studies (e.g., Chen, Delmas, & Lieberman, 2015; Kuosmanen & Johnson, 2010). Both methods are distinct, because non-parametric models such as DEA need not impose any distribution assumption, whereas parametric models such as SFA can be more likely to separate out random noise and to estimate efficiency (e.g., Baik et al., 2013; Chen, Delmas, & Lieberman, 2015). We therefore also use a stochastic frontier approach to check our previous results.11 SFA is a commonly used parametric method for checking the relative productive efficiency of firms (e.g., Baik et al., 2013; Casu, Girardone, & Molyneux, 2004; Chen et al., 2013). SFA has been used by scholars in finance on various aspects (e.g., Chen et al., 2013; Green, Hollifield, & Schürhoff, 2007; Habib & Ljungqvist, 2005). SFA generates an efficient frontier and measure efficiency of each firm relative to the efficient frontier (e.g., Baik et al., 2013). Any deviation from the efficient frontier of each firm can be attributed to a pure random shock (noise) and a technical inefficiency. Following the literature (e.g., Aigner, Lovell, & Schmidt, 1977; Baik et al., 2013; Habib & Ljungqvist, 2005), in a given year, a general SFA can be defined as follows: LN (yi)=Xi+vi-ui

(3)

Within the equation, yi is firm i’s production, X is a set of inputs used to produce output (y) for firm i, vi denotes a symmetric random error and have N (0, 𝜎𝑣2 ) distribution, vi is assumed to be independent and identically distributed (i.i.d.), ui denotes a non-negative i.i.d. half-normal random variable, and ui is independent of

11

We thank an anonymous reviewer for suggesting this investigation. 19

vi.12 ui accounts for the technical inefficiency within the production equation, and vi accounts for the noise. In a given year, we first select all firms within the same industry. The industry definition is based on Fama and French (1997). Then, we use the same output and input variables as in the Eq. (1). We estimate the Eq. (3) for each industry. After the parameters are estimated, we calculate E(exp(-ui)│ei) in order to obtain the efficiency measure of each firm in an industry, whereas ei=vi-ui. We replicate the above procedures across years to obtain the efficiency measures for each firm-year observation within the Compustat. Because our focus is on managerial ability, we also use the same procedure as in the Eq. (2) to obtain the residual as our managerial ability measure derived from SFA. Each year, we also use all Compustat firms to rank the managerial ability derived from SFA. We classify these according to whether or not an acquiring firm is above the whole sample median or not.13 High managerial ability (P) is the associated new measure. [Insert Table 8 about here] Table 8 shows the corresponding results for CAR. Panel A shows the univariate analysis of CAR based on the High managerial ability (P) classification. We find High managerial ability (P) still has a significantly positive impact on deal announcement abnormal returns. Deals with high managerial ability enjoy higher CAR than deals with low managerial ability, which is statistically significant at the 1% level on both mean and median difference. Panels B and C show the analysis of the CAR based on the High managerial ability (P) classification in high environmental uncertainty and

12

We follow the literature and use a half-normal distribution assumption for ui (e.g., Baik et al., 2013; Chen et al., 2013). However, our results are robust to switch to an exponential or a truncated normal distribution. 13 The results are also similar if we use the top quintile or quartile to denote high ability managers. 20

low environmental uncertainty scenarios, respectively. We find deals on average with high managerial ability still enjoy higher CAR on both scenarios, and the impact of high managerial ability on CAR is more pronounced in the high environmental uncertainty scenario. Further, we also find the difference between high managerial ability and low managerial ability in Panel B is higher than that in Panel A, with differences of about 0.1% in the mean and 0.3% in the median. Panel D shows the corresponding regression results on the analysis of the influence of managerial ability on CAR, controlling for other confounding factors. We have omitted the coefficient estimates of the intercept and control variables to save space. The regression results are similar to Table 3; those deals with high managerial ability perform better than those without, and the impact is more pronounced in the high environmental uncertainty scenario. In unreported results analyzing the impact of High managerial ability (P) on BHAR and Premium, we also find results consistent with previous findings that deals with high managerial ability enjoy higher long-term abnormal stock returns and pay fewer takeover premiums than those without. The overall results suggest that our results are robust to use a parametric method such as SFA. 3.4.2. Another managerial ability measure We also borrow the general ability index from Custódio, Ferreira, and Matos (2013) to measure managerial ability.14 High general ability represents the group of firms whose CEOs have general ability above the median of the whole for Compustat

14

We are indebted to Miguel A. Ferreira for sharing the general ability data on his website, http://docentes.fe.unl.pt/~mferreira/. The general ability index measures a manager’s general management skills and is a first principal component of the following five variables: the number of different positions a CEO has held, the number of different firms a CEO has worked for, the number of different industries a CEO has worked in, whether a CEO has worked as a CEO at other firms, and whether a CEO has worked for multi-division firms before. 21

firms in a year.15 Table 9 presents the corresponding results. Panel A shows the univariate analysis of CAR based on the High general ability classification. We find that deals with high general ability enjoy higher CAR than those without, and the mean and median difference are about 0.7% and 0.5% respectively, and both are statistically significant at the 1% level. Panels B and C further provide the results on the analysis of the CAR based on the High general ability classification in high environmental uncertainty and low environmental uncertainty scenarios, respectively. We find that deals with high general ability still enjoy higher CAR in both scenarios, and the impact of high general ability on CAR is more pronounced in the high environmental uncertainty scenario. Further, we also find that the difference between high general ability and low general ability in Panel B is higher than that in Panel A, with differences of about 2.5% in the mean and 2.6% in the median. [Insert Table 9 about here] Panel D reports the regression results. We have omitted the coefficient estimates of the intercept and control variables to save space. The results are similar to those of the other tests; managers with high ability enjoy higher short-run market reactions, and the impact is more pronounced in high environmental uncertainty scenarios. In unreported results analyzing the impact of High general ability on BHAR and Premium, we also find results consistent with previous findings that deals with high general ability enjoy higher long-term abnormal stock returns and pay fewer takeover premiums than those without. 3.4.3. Reverse causality and self-selection bias Our results, however, may suffer from the possibility of confounding effects

15

The results are similar if we use the top quintile or quartile to denote high general ability managers. 22

from reverse causality in the relation and self-selection bias. As for the reverse causality concern on the relation between managerial ability and firm performance, for example, firms with good performance can be more likely to hire managers with higher managerial ability, and such firms will therefore show better performance. Because M&As are largely unanticipated events, the use of announcement abnormal returns on unexpected events such as M&As to measure performance can potentially reduce the reverse causality concern (e.g., Deng, Kang, & Low, 2013). As for the self-selection bias, for example, firms involved in M&A activities could possibly be those with high managerial ability. To address the concern of the self-selection bias explicitly, we follow the literature and use a propensity score matching procedure (e.g., Dehejia & Wahba, 2002; Faccio, Marchica, & Mura, 2011). We use three different methods to conduct the propensity score matching estimation: nearest neighborhood, Gaussian kernel, and local linear regression (e.g., Deng, Kang, & Low, 2013). We first use all Compustat firms to conduct a probit regression with the dependent variable being a dummy variable which is equal to 1 if a firm in investigation has conducted an M&A in a year. The matching variables include Firm size, Tobin’s Q, Cash flow, Leverage, Same industry dummy, All cash deal, Public, Subsidiary, industry dummies, and year dummies. The industry dummies are set according to the industry definition in Fama and French (1997). We match each sample firm of high managerial ability with a matching firm which is of low managerial ability and has the closest propensity score as the acquirer with high managerial ability, with replacement. We then calculate the corresponding difference in CAR, BHAR, and Premium between our treatment firms (i.e., acquirers with high managerial ability) and matching firms (i.e., acquirers with low managerial ability).

23

As in Bae, Kang, and Wang (2011), Deng, Kang, and Low (2013), and Lee and Wahal (2004), the associated t-statistics are calculated on the basis of bootstrapped standard errors, with the bootstrapping based on one hundred replications. [Insert Table 10 about here] Table 10 shows the corresponding results. We find similar results.16 Panel A shows the results on CAR. Deals with higher managerial ability still continue to enjoy higher announcement abnormal returns and are statistically positively significant at the 5% level or better across the three matching techniques. Panels B and C show the results on BHAR and Premium, respectively. We continue to find evidence that deals with higher managerial ability continue to enjoy higher BHAR and pay lower Premium. The results on Table 10 suggest our previous results seem to be less affected by the self-selection bias. 4. Conclusion In this paper, we examine the impact of managerial ability on mergers and acquisitions. We find that deals made by firms with high managerial ability enjoy higher short-term stock market reactions and better post-announcement abnormal returns than deals with low managerial ability. Further, deals made by firms with high managerial ability also pay fewer takeover premiums than deals without. We also find that managers perform better in high environmental uncertainty, which suggests that markets attribute greater value to managers in high uncertainty scenarios. The results of this paper also suggest that high environmental uncertainty is an important scenario that should be included in studies on the impact of managerial ability.

16

We conduct a series of tests to check our findings and results are unchanged. First, following Bae, Kang, and Wang (2011), Deng, Kang, and Low (2013), and Smith and Todd (2005), we drop 2% of observations that has the lowest propensity score in order to make sure the quality of matching. In further tests, we also conduct tests based on a requirement that matching firms must be from the same industry and year as the acquirer of higher managerial ability. 24

References Adams, R. B., Almeida, H., & Ferreira, D. (2005). Powerful CEOs and their impact on corporate performance. Review of Financial Studies, 18, 1403-1432. Aigner, D., Lovell, C. A. K., & Schmidt, P. (1977). Formulation and estimation of stochastic frontier production function models. Journal of Econometrics, 6, 21-37. Alexandridis, G., Antoniou, A., & Petmezas, D. (2007). Divergence of opinion and post-acquisition performance. Journal of Business Finance & Accounting, 34, 439-460. Andreou, P. C., Philip, D., & Robejsek, P. (2016). Bank liquidity creation and risk-taking: Does managerial ability matter? Journal of Business Finance & Accounting, 43, 226-259. Bae, K. H., Kang, J. K., & Wang, J. (2011). Employee treatment and firm leverage: A test of the stakeholder theory of capital structure. Journal of Financial Economics, 100, 130-153. Baik, B., Chae, J., Choi, S., & Farber, D. B. (2013). Changes in operational efficiency and firm performance: A frontier analysis approach. Contemporary Accounting Research, 30, 996-1026. Bertrand, M., & Schoar, A. (2003). Managing with style: The effect of managers on firm policies. Quarterly Journal of Economics, 118, 1169-1208. Bonsall, S. B., Holzman, E. R., & Miller, B. P. (2017). Managerial ability and credit risk assessment. Management Science, 63, 1425-1449. Cai, Y., & Sevilir, M. (2012). Board connections and M&A transactions. Journal of Financial Economics, 103, 327-349. Cain, M. D., & McKeon, S. B. (2016). CEO personal risk-taking and corporate policies. Journal of Financial and Quantitative Analysis, 51, 139-164. Casu, B., Girardone, C., & Molyneux, P. (2004). Productivity change in European banking: A comparison of parametric and non-parametric approaches. Journal of Banking and Finance, 28, 2521-2540. Chen, S. S., Chen, Y. S., Liang, W. L., & Wang, Y. (2013). R&D spillover effects and firm performance following R&D increases. Journal of Financial and Quantitative Analysis, 48, 1607-1634. Chen, C. M., Delmas, M. A., & Lieberman, M. B. (2015). Production frontier methodologies and efficiency as a performance measure in strategic management research. Strategic Management Journal, 36, 19-36. Custódio, C., Ferreira, M. A., & Matos, P. (2013). Generalists versus specialists: Lifetime work experience and chief executive officer pay. Journal of Financial Economics, 108, 471-492. Custódio, C., & Metzger, D. (2013). How do CEOs matter? The effect of industry expertise on acquisition returns. Review of Financial Studies, 26, 2008-2047. Dehejia, R. H., & Wahba, S. (2002). Propensity score-matching methods for nonexperimental causal studies. Review of Economics and Statistics, 84, 151-161. Demerjian, P. R., Lev, B., & McVay, S. E. (2012). Quantifying managerial ability: A 25

new measure and validity tests. Management Science, 58, 1229-1248. Demerjian, P. R., Lev, B., Lewis, M. F., & McVay, S. E. (2013). Managerial ability and earnings quality. Accounting Review, 88, 463-498. Deng, X., Kang, J. K., & Low, B. S. (2013). Corporate social responsibility and stakeholder value maximization: Evidence from mergers. Journal of Financial Economics, 110, 87-109. Dutta, S., & Jog, V. (2009). The long-term performance of acquiring firms: A re-examination of an anomaly. Journal of Banking and Finance, 33, 1400-1412. Faccio, M., Marchica, M. T., & Mura, R. (2011). Large shareholder diversification and corporate risk-taking. Review of Financial Studies, 24, 3601-3641. Fama, E. F., & French, K. R. (1997). Industry costs of equity. Journal of Financial Economics, 43, 153-193. Field, L. C., & Mkrtchyan, A. (2017). The effect of director experience on acquisition performance. Journal of Financial Economics, 123, 488-511. Francis, J., Huang, A. H., Rajgopal, S., & Zang, A. Y. (2008). CEO reputation and earnings quality. Contemporary Accounting Research, 25, 109-147. Francis, B., Hasan, I., Mani, S., & Ye, P. (2016). Relative peer quality and firm performance. Journal of Financial Economics,122, 196-219. Fuller, K., Netter, J., & Stegemoller, M. (2002). What do returns to acquiring firms tell us? Evidence from firms that make many acquisitions. Journal of Finance, 57, 1763-1793. Green, R. C., Hollifield, B., & Schürhoff, N. (2007). Financial intermediation and the costs of trading in an opaque market. Review of Financial Studies, 20, 275-314. Habib, M. A., & Ljungqvist, A. (2005). Firm value and managerial incentives: A stochastic frontier approach. Journal of Business, 78, 2053-2094. Harford, J. (1999). Corporate cash reserves and acquisitions. Journal of Finance, 54, 1969-1997. Harford, J., & Schonlau, R. J. (2013). Does the director labor market offer ex post settling-up for CEOs? The case of acquisitions. Journal of Financial Economics, 110, 18-36. Hartzell, J. C., Ofek, E., & Yermack, D. (2004). What’s in it for me? CEOs whose firms are acquired. Review of Financial Studies, 17, 37-61. Kaplan, S. N., Klebanov, M. M., & Sorensen, M. (2012). Which CEO characteristics and abilities matter? Journal of Finance, 67, 973-1007. Koester, A., Shevlin, T., & Wangerin, D. (2016). The role of managerial ability in corporate tax avoidance. Management Science, forthcoming. Krolikowski, M. W., Adhikari, H. P., Malm, J., & Sah, N. B. (2017). Inter-firm linkages and M&A returns. Quarterly Review of Economics and Finance, 63, 135-146. Kuosmanen, T., & Johnson, A. L. (2010). Data envelopment analysis as nonparametric least-squares regression. Operations Research, 58, 149-160. Jacobsen, S. (2014). The death of the deal: Are withdrawn acquisition deals informative of CEO quality? Journal of Financial Economics, 114, 54-83. Lee, P. M., & Wahal, S. (2004). Grandstanding, certification and the underpricing of venture capital backed IPOs. Journal of Financial Economics, 73, 375-407. Lin, C., Officer, M. S., & Zou, H. (2011). Directors’ and officers’ liability insurance and acquisition outcomes. Journal of Financial Economics, 102, 507-525. 26

Loughran, T., & Vijh, A. M. (1997). Do long-term shareholders benefit from corporate acquisitions? Journal of Finance, 52, 1765-1790. Lyon, J. D., Barber, B. M., & Tsai, C. L. (1999). Improved methods for tests of long-run abnormal stock returns. Journal of Finance, 54, 165-201. Malmendier, U., & Tate, G. (2008). Who makes acquisitions? CEO overconfidence and the market's reaction. Journal of Financial Economics, 89, 20-43. Masulis, R. W., Wang, C., & Xie, F. (2007). Corporate governance and acquirer returns. Journal of Finance, 62, 1851-1889. Milbourn, T. T. (2003). CEO reputation and stock-based compensation. Journal of Financial Economics, 68, 233-262. Moeller, S. B., Schlingemann, F. P., & Stulz, R. M. (2004). Firm size and the gains from acquisitions. Journal of Financial Economics, 73, 201-228. Petersen, M. A. (2009). Estimating standard errors in finance panel data sets: Comparing approaches. Review of Financial Studies, 22, 435-480. Rau, P. R., & Vermaelen, T. (1998). Glamour, value and the post-acquisition performance of acquiring firms. Journal of Financial Economics, 49, 223-253. Savor, P. G., & Lu, Q. (2009). Do stock mergers create value for acquirers? Journal of Finance, 64, 1061-1097. Smith, J. A., & Todd, P. E. (2005). Does matching overcome LaLonde’s critique of nonexperimental estimators? Journal of Econometrics, 125, 305-353. Uysal, V. B. (2011). Deviation from the target capital structure and acquisition choices. Journal of Financial Economics, 102, 602-620. Wang, C., & Xie, F. (2009). Corporate governance transfer and synergistic gains from mergers and acquisitions. Review of Financial Studies, 22, 829-858. White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48, 817-838.

27

Table 1 Summary statistics. Variable

N

Mean

Median

Q1

Q3

STD.

Panel A: Dependent variables CAR

36555

0.003

-0.005

-0.036

0.032

0.094

BHAR

35013

-0.001

-0.171

-0.026

0.115

0.311

Premium

2696

0.346

0.285

0.129

0.443

0.384

36555

0.554

1.000

0.000

1.000

0.497

High managerial ability (P) 36555

0.411

0.000

0.000

1.000

0.492

High general ability

14950

0.465

0.000

0.000

1.000

0.499

High uncertainty

36555

0.128

0.000

0.000

0.000

0.334

Firm size

34794

6.035

5.946

4.498

7.443

2.167

Tobin's Q

34793

2.543

1.762

1.319

2.607

3.969

Cash flow

34640

0.003

0.040

-0.007

0.077

0.220

Leverage

34794

0.228

0.200

0.042

0.349

0.213

Same industry dummy

36555

0.580

1.000

0.000

1.000

0.494

All cash deal

36555

0.153

0.000

0.000

0.000

0.360

Public

36555

0.080

0.000

0.000

0.000

0.272

Panel B: Firm and deal characteristics High managerial ability

Subsidiary 36555 0.307 0.000 0.000 1.000 0.461 This table presents the summary statistics of the whole sample. The sample period covers the years 1991 to 2013. Panel A shows the results for the dependent variables, and Panel B shows the statistics for the firm and deal characteristics. Variable definitions are detailed in the Appendix.

28

Table 2 Announcement abnormal returns based on managerial ability classification. Panel A: Managerial ability High managerial ability Low managerial ability Difference Mean 0.622%*** -0.085%* 0.707% Median 0.007%*** -0.817%*** 0.825% N 20248 16307 Panel B: Managerial ability in high uncertainty scenario High managerial ability Low managerial ability

*** ***

Difference

Mean

1.068%***

-0.333%***

1.402%***

Median

0.418%***

-1.333%***

1.752%***

N

1981 2683 Panel C: Managerial ability in low uncertainty scenario High managerial ability Low managerial ability Mean Median

0.662%*** -0.009%

Difference

-0.063%

0.724%***

-0.926%***

0.917%***

N 1980 2682 This table presents the 5-day cumulative abnormal returns based on managerial ability classification. High (Low) managerial ability represents the group of firms that have managerial ability above (equal to or less than) the median of the whole for Compustat firms. High (Low) uncertainty represents an industry belonging to the top (bottom) quintile ranking for industry shocks. Panel A reports the results based on managerial ability. Panel B reports the results based on the partitions of the managerial ability measure and high environmental uncertainty. Panel C reports the results based on the partitions of the managerial ability measure and low environmental uncertainty. ***, **, and * indicate 1%, 5%, and 10% statistical significance levels, respectively.

29

Table 3 Regression analysis of announcement abnormal returns. (1) High managerial ability 0.003*** (2.71) High uncertainty

(3) 0.003 (0.94) -0.004 (-1.56) High managerial ability × 0.013*** High uncertainty (2.99) Firm size -0.005*** (-10.47) Tobin's Q -0.001 (-1.34) Cash flow 0.021 (1.38) Leverage 0.009** (2.12) Same industry dummy -0.002 (-1.09) All cash deal 0.002 (1.39) Public -0.024*** (-11.71) Subsidiary 0.005*** (2.58) Intercept -0.007 -0.007 0.018 (-0.99) (-0.97) (1.29) Industry fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 36555 36555 34640 Adjusted R2 0.008 0.008 0.033 This table presents the findings of regression analysis of 5-day cumulative abnormal returns based on managerial ability classification. High managerial ability represents the group of companies that have managerial ability above the median of the whole for Compustat firms. High uncertainty represents an industry belonging to the top quintile ranking for industry shocks. The industry dummies are set according to the industry definition in Fama and French (1997). The definitions of other independent variables can be found in the Appendix. The t-statistics are presented in parentheses, adjusted for White heteroskedasticity and firm clustering. ***, **, and * indicate 1%, 5%, and 10% statistical significance levels, respectively.

30

(2) 0.001 (0.72) -0.003 (-1.60) 0.011*** (3.77)

Table 4 Long-term abnormal returns based on managerial ability classification. Panel A: Managerial ability High managerial ability Low managerial ability Difference Mean 5.358%*** -6.848%*** 12.206% *** Median 0.163%*** -10.900%*** 11.063% *** N 19437 15576 Panel B: Managerial ability in high uncertainty scenario High managerial ability Low managerial ability Difference Mean 11.200%*** -5.902%*** 17.102%*** Median 9.400%*** -11.200%*** 20.600%*** N 2599 1907 Panel C: Managerial ability in low uncertainty scenario High managerial ability Low managerial ability Difference Mean 5.653%*** -7.482%*** 13.135%*** Median 0.041%*** -10.850%*** 10.891%*** N 2576 1887 This table presents the 1-year buy-and-hold abnormal returns based on managerial ability classification. High (Low) managerial ability represents the group of firms that have managerial ability above (equal to or less than) the median of the whole for Compustat firms. High (Low) uncertainty represents an industry belonging to the top (bottom) quintile ranking for industry shocks. Panel A reports the results based on managerial ability. Panel B reports the results based on the partitions of the managerial ability measure and high environmental uncertainty. Panel C reports the results based on the partitions of the managerial ability measure and low environmental uncertainty. ***, **, and * indicate 1%, 5%, and 10% statistical significance levels, respectively.

31

Table 5 Regression analysis of long-term abnormal returns. (1) High managerial ability 0.125*** (14.91) High uncertainty

(3) 0.124*** (18.56) -0.006 (-0.40) High managerial ability × 0.076*** High uncertainty (3.98) Firm size -0.004** (-2.27) Tobin's Q -0.006 (-1.18) Cash flow 0.050 (1.57) Leverage 0.023 (1.60) Same industry dummy 0.009* (1.69) All cash deal 0.016*** (2.70) Public -0.112* (-1.86) Subsidiary 0.020*** (3.35) Intercept 0.146** 0.140** 0.262** (2.09) (1.99) (2.10) Industry fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 35013 35013 33178 2 Adjusted R 0.058 0.059 0.082 This table presents the findings of regression analysis of 1-year buy-and-hold abnormal returns based on managerial ability classification. High managerial ability represents the group of companies that have managerial ability above the median of the whole for Compustat firms. High uncertainty represents an industry belonging to the top quintile ranking for industry shocks. The industry dummies are set according to the industry definition in Fama and French (1997). The definitions of other independent variables can be found in the Appendix. The t-statistics are presented in parentheses, adjusted for White heteroskedasticity and firm clustering. ***, **, and * indicate 1%, 5%, and 10% statistical significance levels, respectively.

32

(2) 0.118*** (22.85) -0.014 (-1.10) 0.054*** (3.34)

Table 6 Takeover premiums based on managerial ability classification. Panel A: Managerial ability High managerial ability Low managerial ability Difference Mean 29.499%*** 41.157%*** -11.658% *** Median 27.780%*** 38.490%*** -10.710% *** N 1524 1172 Panel B: Managerial ability in high uncertainty scenario High managerial ability Low managerial ability Difference Mean 12.027%*** 35.930%*** -23.903%*** Median 11.390%*** 38.400%*** -27.010%*** N 175 122 Panel C: Managerial ability in low uncertainty scenario High managerial ability Low managerial ability Difference Mean 31.281%*** 42.789%*** -11.508%*** Median 28.490%*** 39.500%*** -11.010%*** N 216 131 This table presents the analysis of takeover premiums based on managerial ability classification. High (Low) managerial ability represents the group of firms that have managerial ability above (equal to or less than) the median of the whole for Compustat firms. High (Low) uncertainty represents an industry belonging to the top (bottom) quintile ranking for industry shocks. Panel A reports the results based on managerial ability. Panel B reports the results based on the partitions of the managerial ability measure and high environmental uncertainty. Panel C reports the results based on the partitions of the managerial ability measure and low environmental uncertainty. ***, **, and * indicate 1%, 5%, and 10% statistical significance levels, respectively.

33

Table 7 Regression analysis of takeover premiums. (1) High managerial ability -0.115*** (-7.24) High uncertainty

(3) -0.097*** (-5.47) -0.052 (-1.31) High managerial ability × -0.134*** High uncertainty (-2.85) Firm size-acquirer 0.010 (1.05) Tobin's Q-acquirer 0.002 (0.55) Cash flow-acquirer -0.017 (-0.29) Leverage-acquirer -0.033 (-0.36) Firm size-target -0.005 (-1.57) Tobin's Q-target -0.001 (-0.69) Cash flow-target 0.007 (0.88) Leverage-target 0.009 (0.66) Same industry dummy -0.006 (-0.34) All cash deal 0.071*** (3.75) Intercept 0.479*** 0.488*** 0.536*** (6.52) (6.17) (5.78) Industry fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 2696 2696 2696 2 Adjusted R 0.049 0.053 0.062 This table presents the findings of regression analysis of takeover premiums based on managerial ability classification. High managerial ability represents the group of companies that have managerial ability above the median of the whole for Compustat firms. High uncertainty represents an industry belonging to the top quintile ranking for industry shocks. The industry dummies are set according to the industry definition in Fama and French (1997). The definitions of other independent variables can be found in the Appendix. The t-statistics are presented in parentheses, adjusted for White heteroskedasticity and firm clustering. ***, **, and * indicate 1%, 5%, and 10% statistical significance levels, respectively.

34

(2) -0.098*** (-5.78) -0.051 (-1.31) -0.148*** (-3.16)

Table 8 Different estimation procedures for managerial ability. Panel A: Managerial ability (P) High managerial ability (P) Low managerial ability (P) Difference Mean 0.658%*** 0.061% 0.598% *** Median -0.198% -0.537%*** 0.338% *** N 15038 21517 Panel B: Managerial ability (P) in high uncertainty scenario High managerial ability (P) Low managerial ability (P) Difference Mean 0.880%*** 0.211% 0.669%** Median 0.221%* -0.407% 0.628%** N 1828 2836 Panel C: Managerial ability (P) in low uncertainty scenario High managerial ability (P) Low managerial ability (P) Difference Mean 0.636%*** 0.047% 0.589%** Median -0.020% -0.787%*** 0.767%*** N 2054 2608 Panel D: Regression analysis (1) (2) (3) High managerial ability (P) 0.005*** 0.002** 0.001 (3.38) (2.16) (0.58) High uncertainty -0.004 -0.004 (-1.53) (-1.49) High managerial ability (P) × 0.007** 0.008** High uncertainty (2.08) (2.01) Intercept Yes Yes Yes Controls No No Yes Industry fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 36555 36555 34640 Adjusted R2 0.006 0.007 0.022 This table presents the findings of analysis of 5-day cumulative abnormal returns based on another managerial ability classification derived from a stochastic frontier analysis. High (Low) managerial ability (P) represents the group of firms that have managerial ability above (equal to or less than) the median of the whole for Compustat firms. High (Low) uncertainty represents an industry belonging to the top (bottom) quintile ranking for industry shocks. Panel A reports the results based on managerial ability. Panel B reports the results based on the partitions of the managerial ability measure and high environmental uncertainty. Panel C reports the results based on the partitions of the managerial ability measure and low environmental uncertainty. Panel D reports the regression analysis. We have omitted the coefficient estimates of the intercept and control variables to save space. The control variables include Firm size, Tobin’s Q, Cash flow, Leverage, Same industry dummy, All cash deal, Public, and Subsidiary. The industry dummies are set according to the industry definition in Fama and French (1997). The definitions of control variables can be found in the Appendix. The t-statistics are presented in parentheses, adjusted for White heteroskedasticity and firm clustering. ***, **, and * indicate 1%, 5%, and 10% statistical significance levels, respectively.

35

Table 9 Analysis of announcement abnormal returns based on general ability. Panel A: General ability High general ability Low general ability Difference Mean 0.819%*** 0.091% 0.728% *** Median 0.568%*** 0.029% 0.539% *** N 6946 8004 Panel B: General ability in high uncertainty scenario High general ability Low general ability Difference Mean 3.462%*** 0.234% 3.228%*** Median 3.264%*** 0.098% 3.166%*** N 1054 1025 Panel C: General ability in low uncertainty scenario High general ability Low general ability Difference Mean 0.542%** -0.100% 0.642%** Median 0.594%*** 0.090% 0.504%** N 827 1221 Panel D: Regression analysis (1)

High general ability

(2)

0.007*** (5.37)

(3)

0.004 0.002 (1.37) (0.53) High uncertainty -0.001 -0.003 (-0.60) (-0.80) High general ability × 0.026*** 0.025*** High uncertainty (6.46) (4.30) Intercept Yes Yes Yes Controls No No Yes Industry fixed effect Yes Yes Yes Year fixed effect Yes Yes Yes N 14950 14950 13211 Adjusted R2 0.013 0.013 0.053 This table presents the findings of analysis of 5-day cumulative abnormal returns based on general ability classification. High (Low) general ability represents the group of firms that have general ability above (equal to or less than) the median of the whole for Compustat firms. High (Low) uncertainty represents an industry belonging to the top (bottom) quintile ranking for industry shocks. Panel A reports the results based on general ability. Panel B reports the results based on the partitions of the general ability measure and high environmental uncertainty. Panel C reports the results based on the partitions of the general ability measure and low environmental uncertainty. Panel D reports the regression analysis. We have omitted the coefficient estimates of the intercept and control variables to save space. The control variables include Firm size, Tobin’s Q, Cash flow, Leverage, Same industry dummy, All cash deal, Public, and Subsidiary. The industry dummies are set according to the industry definition in Fama and French (1997). The definitions of control variables can be found in the Appendix. The t-statistics are presented in parentheses, adjusted for White heteroskedasticity and firm clustering. ***, **, and * indicate 1%, 5%, and 10% statistical significance levels, respectively.

36

Table 10 Test of differences between high managerial ability firms and matching firms. Variable

Nearest neighborhood

Gaussian kernel

Local linear regression

Panel A: Announcement abnormal returns CAR

0.645%*** (3.42)

0.573%** (2.11)

0.623%** (2.29)

10.188%*** (4.11)

11.201%*** (5.12)

Panel B: Long-term abnormal returns BHAR

9.142%*** (4.39)

Panel C: Takeover premiums Premium

-8.994%*** -9.147%*** -10.319%*** (3.63) (3.81) (4.22) This table presents the tests of differences between high managerial ability firms and low managerial ability firms. We match each acquirer with high managerial ability with an acquirer with low managerial ability using nearest neighborhood, Gaussian kernel and local linear regression matching techniques. The matching variables include Firm size, Tobin’s Q, Cash flow, Leverage, Same industry dummy, All cash deal, Public, Subsidiary, industry dummies, and year dummies. Variable definitions are in the Appendix. The industry dummies are set according to the industry definition in Fama and French (1997). Panel A shows the results of 5-day cumulative abnormal returns. Panel B shows the results of 1-year buy-and-hold abnormal returns. Panel C shows the results of takeover premiums. The t-statistics are calculated on the basis of bootstrapped standard errors, with the bootstrapping based on one hundred replications. ***, **, and * indicate 1%, 5%, and 10% statistical significance levels, respectively.

37

Appendix A. Table A1 Variable defintions. Variable Definition Panel A: Dependent variables CAR The 5-day cumulative abnormal returns, calculating from 2 trading days before to 2 trading days after the acquisition announcement. BHAR The 1-year buy-and-hold abnormal returns relative to the acquisition announcement. Premium The takeover premium, measured as the offer price minus the target’s stock price measured at 4 weeks before the M&A announcement divided by the target’s stock price measured at 4 weeks before the M&A announcement. Panel B: Firm and deal characteristics High managerial A dummy variable which is equal to 1 whose firms have managerial ability ability above the median of the whole for Compustat firms, where managerial ability is that defined by Demerjian, Lev, and McVay (2012). High managerial A dummy variable which is equal to 1 whose firms have managerial ability (P) ability above the median of the whole for Compustat firms, and the managerial ability is derived from a stochastic frontier analysis. High general ability A dummy variable which is equal to 1 whose firms have general ability above the median of the whole for Compustat firms, and the general ability data is from Custódio, Ferreira, and Matos (2013). High uncertainty A dummy variable which is equal to 1 for an industry belonging to the top quintile ranking for industry shocks. Industry shocks are measured as the difference between each industry’s sale growth minus that of the benchmark, which is the average sales growth calculated using all industries. Firm size The logarithm of annual sales, lagged by 1 year. Tobin’s Q The ratio of the market value of total assets to the book value of total assets, lagged by 1 year. Cash flow A ratio measured as sales minus COGS, SGA, working capital change, then scaled by total assets, lagged by 1 year. Leverage The long-term debt divided by total assets, lagged by 1 year. Same industry A dummy variable that is equal to 1 if an acquirer and its target dummy belong to the same Fama and French (1997) industry. All cash deal A dummy variable that is equal to 1 if the payment of the M&A is all cash, and 0 otherwise. Public A dummy variable which is equal to 1 if the target is a public firm, and 0 otherwise. Subsidiary A dummy variable which is equal to 1 if the target is a subsidiary, and 0 otherwise.

38