The Young and the Restless: An International Study of CEO Age and Acquisition Propensity

The Young and the Restless: An International Study of CEO Age and Acquisition Propensity Teng Zhang Scheller College of Business Georgia Institute of Technology 800 West Peachtree Street NW Atlanta, GA 30332 Tel: (404) 385-1391 E-mail: Teng.Zhang@scheller.gatech.edu Stephen P. Ferris Trulaske College of Business University of Missouri 404F Cornell Hall Columbia, MO 65211 Tel: (573) 882-9905 E-mail: Ferriss@missouri.edu Narayanan Jayaraman Scheller College of Business Georgia Institute of Technology 800 West Peachtree Street NW Atlanta, GA 30332 Tel: (404) 894-4389 E-mail: Narayanan.Jayaraman@scheller.gatech.edu Sanjiv Sabherwal College of Business University of Texas at Arlington 701 S. West Street Arlington, TX 76019 Tel: (817) 272-5520 E-mail: Sabherwal@uta.edu 12 January 2015

The Young and the Restless: An International Study of CEO Age and Acquisition Propensity Abstract Using a large sample of CEOs of U.K. firms, we show that CEO age is a key determinant of acquisition activity. Younger CEOs are more likely to acquire another firm, acquire a firm outside their main line of business, and use cash for payment. We find that younger CEOs are not more likely to turn over, regardless of merger performance. Younger CEOs also spend more on capital expenditures than their older peers. Further, we determine that our evidence is inconsistent with both compensation incentive and overconfidence explanation. Rather, our results suggest executive signaling by young CEOs eager to distinguish themselves. Keywords: CEO; age; merger; acquisition JEL Code: G3; G34

The Young and the Restless: An International Study of CEO Age and Acquisition Propensity 1. Introduction In this study, we empirically examine how career and compensation concerns affect the acquisition activities of a set of Chief Executive Officers (CEOs) in a sample of U.K. firms. More specifically, we examine how these concerns vary with CEO age and whether they influence merger and acquisition decision-making. The effect of career concerns on CEO acquisition practices, however, has not yet been examined internationally. This study is the first to do so. There are several factors which make it unclear if the effect of career concerns on the merger decision-making of U.S. CEOs applies to U.K. executives. The national culture in which the CEO operates exerts a distinct effect, influencing how age, career concerns, and decision-making are reconciled. The efficiency and transparency of the labor markets varies significantly across countries, thus affecting the effectiveness of managerial incentives. Finally, patterns of insider ownership, executive hiring practices, limits on executive compensation, and pay-performance sensitivity differ globally. These differences reinforce our belief that the effect of career concerns on CEO decisions about acquisitions is not globally uniform. Consequently, our study of U.K. firms allows us to determine the extent to which previous findings regarding the effect of career and compensation concerns on CEO acquisition behavior hold internationally or are simply idiosyncratic to the U.S. Prendergast and Stole (1996) and Li et al. (2011) contend that a younger CEO is more likely to make bold decisions while older CEOs tend to resist dramatic decision making, fearful that previous decisions would be viewed as incorrect. This behavior is also consistent with Bertrand and Mullainathan s (2003) quiet life view of CEOs. Given their incentives to grow 1

the firm and enhance their reputation, Yim (2013) finds that younger CEOs of U.S. firms act more aggressively in their acquisition activity. This behavior is also consistent with the claim by Avery, Chevalier, and Schaefer (1998) that younger managers have a stronger incentive to expand the firm beyond its optimal size and build an empire to achieve greater prestige and power. Based on Li et al. (2011), we contend that career concerns influence the relationship between CEO age and acquisition activity in two opposite directions. Managerial career concerns might induce younger CEOs to be more restrained and less willing to undertake the risks associated with an acquisition. Such a result is called the CEO conservatism effect and implies a positive relationship between age and acquisition activity. That is, only older CEOs will be more likely to undertake acquisitions. Alternatively, young CEOs might wish to demonstrate their expertise and abilities by undertaking a large acquisition. Young CEOs might also respond to the compensation incentives associated with the leadership of a larger firm. We call this the CEO signaling-incentive effect and it posits an inverse relationship between age and acquisition activity. There are several aspects of the U.K. business environment that make it especially attractive for analyzing the effect of compensation on a CEO s decisions about mergers. Doukas and Petmezas (2007) point out that U.K. has the most merger activity after the U.S. and represents more than 65% of merger transactions in Europe. However, CEO compensation practices in the U.K. have some significant differences from those in the U.S. Conyon and Murphy (2000), Conyon et al. (2013), and Gerakos, Piotroski, and Srinivasan (2013) report that U.S. CEOs enjoy higher compensation than their U.K. peers. Becht et al. (2006) discuss that certain features of the U.K. governance environment may act as a barrier to CEO payment. For 2

example, U.K. shareholders can influence the board of directors more than the U.S. shareholders. Ozkan (2011) find that the pay-performance elasticity for U.K. CEOs is lower than their U.S. counterparts. Fernandes et al. (2012) report that there are substantial differences in how CEO compensation is structured around the world. They note, for instance, that fixed salary as a percentage of total CEO compensation is greater in the U.K. than in the U.S. Similarly, U.K. executives enjoy much smaller equity based pay and bonuses than their U.S. counterparts. Further, U.K. shareholders have greater authority over CEO pay compared to U.S. investors which can dampen executive salary levels (Ferri and Maber, 2013). These differences allow us to test with even more rigor how CEO career and compensation concerns effect their decisions about mergers. Our study finds that the behavior of U.K. senior executives is most consistent with the CEO Signaling-Incentive Effect. That is, we determine that younger CEOs pursue large scale acquisitions as well as commit to higher levels of capital expenditures than their older peers. We further discover that younger CEOs are more likely to undertake diversifying mergers, suggesting that they pursue riskier targets. They also use cash more often than their own firm s stock to pay for these acquisitions. We also separate the signaling from the incentive explanation for the inverse relationship between CEO age and acquisition levels. Unlike the results for U.S. CEOs reported by Yim (2013), we find only a weak and unstable relationship between acquisition activity and changes in CEO compensation. We do not find that the incentive of higher future compensation drives the acquisition decisions of young CEOs in our sample. This result holds for the salary, bonus, and equity components of CEO compensation. We also fail to find that younger CEOs are deterred from acquisition activity due to turnover concerns. We contend that younger CEOs 3

consider acquisitions without undue concern regarding turnover, regardless of the perceived quality of the merger. Finally, we find that overconfidence does not explain the acquisition practices of young U.K. CEOs. Overconfidence effects should be more pronounced in those firms having greater cash resources. Our results fail to find consistent evidence of this relationship in decisions by CEOs about large acquisitions, method of payment, or capital expenditures. Given the failure of compensation incentives or overconfidence to explain the inverse relationship between CEO age and acquisition activity, we conclude that executive signaling is the most likely explanation. That is, young CEOs want to demonstrate their expertise and abilities by undertaking large acquisitions. These U.K. CEOs use acquisition activity as a way to distinguish themselves from their peers and to enhance their reputational capital in the managerial labor markets. The remainder of this study is organized as follows. Section 2 discusses the conflicting CEO Conservatism and Signaling-Incentive Effects that are present in the career concerns of corporate executives. Section 3 contains a discussion of our data and describes the sample construction process. Section 4 explores the relationship between age and corporate acquisition activity. Section 5 explicitly tests the incentive effect by examining changes in compensation following a merger. Section 6 tests whether overconfidence might serve as an alternative explanation for greater acquisition activity by young CEOs. We conclude with a brief summary and discussion in section 7. 2. The Relationship between CEO Age and Acquisition Behavior 2.1 The CEO Conservatism Effect 4

The CEO Conservatism Effect notes that a younger CEO lacks a track record and faces greater risks when making acquisition decisions. Such an individual is likely to be perceived as incompetent if the acquisitions turn out poorly. Consequently, younger CEOs are reluctant to jeopardize their future earnings and therefore avoid projects with high levels of performance uncertainty. Such a result is consistent with the argument of Scharfstein and Stein (1990) that managers career concerns, particularly for younger CEOs, can lead them to avoid innovation or bold action. The conservative influence that career concerns play in managerial decision making has been reported in the literature examining mutual fund managers and security analysts. For example, Chevalier and Ellison (1999) and Hong, Kubik, and Solomon (2000) find that younger, less experienced mutual fund managers and security analysts are more likely to be fired due to bad performance. In response to these punishments, younger managers avoid taking risks and often exhibit herding behavior. These studies imply that if CEOs can be dismissed due to poorly performing acquisitions, then they might be reluctant to pursue them. 2.2 The CEO Signaling-Incentive Effect The CEO Signaling- Incentive Effect contends that younger CEOs purposely adopt a more aggressive acquisition strategy. Hence it predicts an inverse relationship between age and acquisition activity. However, this inverse relationship might be due to one of two explanations: signaling or incentives. Because younger CEOs wish to establish their reputations, they tend to exaggerate their reactions to new information and respond aggressively to signal their superior executive abilities. By undertaking highly uncertain, but promising activities, younger CEOs can distinguish 5

themselves among their peers and create reputational value for themselves. This is the skills signaling explanation for the inverse relationship between age and acquisition activity. The incentive explanation focuses on the reward structure faced by young CEOs and how it influences decisions about acquisition activity. A central component of that reward structure is CEO compensation. If acquisitions through their increase in firm size positively influence CEO compensation, then CEOs will have greater incentives to pursue acquisitions earlier in their career. Bebchuk and Grinstein (2005) and Yim (2013) document a positive and economically meaningful correlation between CEO compensation and the CEO s past acquisition activity. They argue that one of the most important incentives for CEOs to expand their firm size is to increase their compensation permanently. 3. Sample and Data 3.1 Sample Construction and Data Sources We draw our sample of CEO-firm years from a sample of firms located in the U.K. Our selection of this country motivated by two reasons. First, next to the U.S., the U.K. has the largest capital market in the world. Second, similar to the U.S., there is an active market in the U.K. for corporate control. Indeed Doukas and Petmezas (2007) note that U.K. mergers account for more than 65% of all European merger activity. This high level of activity in the U.K. merger market allows us to examine the role that CEO age and incentives have on corporate acquisition patterns. Further, our single country research design allows us to hold constant the various country-level cultural, legal and institutional factors that can influence corporate acquisition practices. As noted by Gerakos et al. (2013), our focus on one country allows us to obtain more 6

granular data for our sample firms without concerns regarding availability and comparability across countries. Our study is also comparable to other researchers such as Kaplan (1994a, b), Conyon et al. (2013) and Gerakos et al. with its single country focus. We obtain data on CEO characteristics from BoardEx database. The BoardEx database contains biographical information on most board members and senior executives around the world. It includes data such as total compensation, salary, bonus, equity, age, and historical employment information. Using the historical average number of years as CEO at the current firm as a proxy for tenure, we construct a panel of CEO service by firm for the period of 1999-2012. To ensure that each firm has at least one observation, we include the chairman if there is no CEO observation for that firm during the sample period. We exclude observations when there is a co-ceo or an interim CEO. We obtain annual firm characteristics from Global Compustat. Share prices and the number of outstanding shares are collected from the Security Daily dataset of Global Compustat. We collect data on merger and acquisition transactions from Thomson s SDC Spectrum database. We include acquisitions of both US and non-us targets, and match each deal to a CEO-firm year observation using the announcement date of the acquisition. For each CEO-firm year observation, we assign a value of one to a binary variable named 5% Acquisition if the CEO announces an acquisition of 5% or more of the firm s market capitalization and zero otherwise. This threshold reflects an effort to restrict our acquisition sample to those observations likely to require the CEO s attention. A description of the independent variables used in our analysis is contained in the Appendix. 3.2 Sample Summary Statistics 7

Table 1 reports the summary statistics for our sample. Panel A contains the time-series of numbers of firms and acquisitions in the U.K. We observe that the greatest number of acquisitions occurs in 2001 with 993. The fewest number occur in 2010 with only 325. We provide an industry distribution of our sample acquisitions in Panel B based on the Fama-French 48 business classifications. We note that the greatest number of mergers occurs in the business services industry, distantly followed by the retail industry. Few mergers occur in the banking, coal, or candy and soda industries. Panel C contains descriptive statistics for select acquisition characteristics. The average deal size across the entire sample of firm-ceo years is $44 million. For those firm-ceo years in which a deal occurs, the mean deal size is $129 million. The relative deal size (i.e., relative to the acquirer s market capitalization) across all sample firm-ceo years is 0.09, but is 0.29 for those years in which a merger occurs. Finally, we note that the rate of large acquisition, (i.e., 5% acquisitions), is 16% across our sample of firms. Panel D provides summary statistics of firm and CEO characteristics for the full sample. Acquiring firms tend to be bigger in size (in terms of assets, market capitalization, and total capitalization). Further, acquiring firms tend to be older, have lower market-to- book ratios, and enjoy higher prior year returns. Proxied firm age is a measure of firm maturity (DeAngelo, DeAngelo, and Stulz 2006), since we lack firm age based on the IPO date for many sample firms. It is measured as retained earnings scaled by total assets. For those firms whose IPO date is identified, we calculate firm age using the difference between it and the year 2013. Acquiring firms also have younger CEOs who earn more. The median CEO age for the full sample is 51, with a median tenure of 4.8 years, and a median compensation of $304,000 annually. 8

4. CEO Age and Acquisition Propensity In this section, we test whether young CEOs undertake greater levels of merger activity. The CEO conservatism effect contends that young CEOs will avoid behaviors with highly uncertain outcomes such as mergers to avoid jeopardizing their future earnings or damaging their professional reputations. The CEO conservatism effect asserts that the relationship between age and acquisition activity is positive, with older CEOs engaging in higher levels of merger activity. The CEO signaling-incentive effect contends, however, that younger CEOs adopt an aggressive strategy towards mergers. This implies that the relationship between age and acquisition activity is negative, with younger CEOs making more acquisitions. 4.1 Age-Acquisition Relationship We begin our analysis by estimating a set of logit regressions with the dependent variable assuming a value of one if the CEO announces an acquisition deal whose value exceeds 5% of the firm s market capitalization and zero otherwise. Table 2 presents our empirical findings. Model 1 of the table includes only the age variable as the regressor. The coefficient for the CEO age regressor is less than one, indicating that CEO age and large scale acquisition activity are inversely related. That is, younger CEOs make more acquisitions. Model 2 controls for a set of firm characteristics that might be correlated with acquisition decisions. Although the market-to-book ratio, prior year return, and cash flow are statistically significant, CEO age remains significantly and inversely related to the likelihood of an acquisition. We add CEO tenure, which we define as the number of years the individual has held the CEO title at the current firm, in model 3. In addition to the inverse relationship between CEO age and acquisition activity, we find that CEOs with shorter tenure are also more likely to undertake large acquisitions. Mitchell and Mulherin (1996), Andrade, Mitchell, and Staford, (2001) and 9

Garfinkel and Hankins (2011) report that mergers usually occur in waves within a given industry. Therefore, following Yim (2013), we include industry and year fixed effects via an interaction term in all of the models estimated in Table 2 and subsequent tables. Overall, the findings in Panel A of Table 2 are consistent with the claim that younger CEOs pursue large scale acquisitions more frequently than their older peers. That is, we find that the CEO age-acquisition relationship is inverse. Thus, these findings are most supportive of a signaling-incentive effect present in the decision-making of younger CEOs. In Panel B of Table 2, we observe a robust effect of CEO age on merger propensity. In this panel, we investigate the effect of ADR status on CEO age by adding an ADR binary variable and an interaction term between ADR status and CEO age. The ADR variable has a value of one if the firm is crosslisted in the U.S. Since the ADR binary variable has a strong positive effect on acquisition activity, these results imply that firms with ADRs are more likely to acquire bigger targets. Once the interaction term is added, the positive effect of the ADR binary variable on acquisition activity vanishes and the effect of the interaction term is also not significant. This implies that the ADR status does not affect the decision-making of young CEOs. The act of listing on a U.S. exchange does not change the merger propensity of younger U.K. CEOs. 4.2 Nonlinear Effect of CEO tenure Yim (2013) reports a nonlinear effect of tenure on acquisition propensity in her analysis of U.S. CEOs. Consequently, in Table 3, we test for a nonlinear effect of tenure on the acquisition behavior of our sample CEOs and examine if the inverse relationship between age and acquisition propensity is robust to the inclusion of nonlinear effect of tenure. Our results indicate that the nonlinear effect of tenure is present in the U.K. as well. Model 1 examines tenure based on terciles, with the lowest tercile group omitted by default. 10

Based on the distribution of tenure observations, we use 3.5 and 7.3 years as our tercile boundaries. We observe that the effect of tenure on acquisition propensity is only significant for the middle range. The coefficient for the highest tenure tercile is statistically insignificant, which is consistent with older CEOs having less desire to acquire. Models 2 and 3 compare the effect of an additional year of tenure for CEOs who have less than 5.2 years relative to those with 5.2 or more years of tenure. We use 5.2 years to construct these subsamples since that is the median tenure of our sample CEOs. Model 2 shows that for CEOs with a short tenure, the effect of an additional year of tenure on acquisition propensity is significantly positive. Indeed, the odds of such a CEO undertaking a merger increase by 7.3% annually. This result is consistent with the argument that CEOs with a shorter tenure want to increase their influence and build their reputation before pursuing risky acquisitions. So at the margin, an additional year of tenure is more valuable to them than to CEOs with longer tenure. Model 3 shows that the effect of a year of tenure on acquisition likelihood dissipates for long-tenured CEOs. This result is consistent with the findings of Yim (2013) and might reflect an end-of-tenure effect whereby CEOs who are nearing retirement or departure fail to initiate acquisitions. The effect of age on acquisition propensity continues to be negative and statistically significant in all the three models. This shows that the relationship between age and acquisition activity remains inverse even in the presence of possible nonlinearities in CEO tenure. 4.3 CEO Age and Acquisition Behavior In this section we investigate whether young and old CEOs pursue acquisitions differently. Due to different psychological factors and characteristics, we expect younger CEOs 11

to undertake mergers more aggressively, including more frequent cash payment and acquiring targets outside the acquirer s industry. Table 4 contains our empirical findings. We examine the relationship between age and the industry identity of the target in the first set of models. In models 1 and 2, the dependent variable is one if the target firm is from an industry different from that of the acquirer. In both the models, we include CEO age, ADR listing, CEO tenure, firm size, market-to-book ratio, ROA, prior year return, cash flow, and proxied firm age as independent variables. In model 2, we also include an interaction between ADR listing and CEO age. In both models, we find a negative effect of age on the likelihood of a CEO pursuing a diversifying merger. The effect is statistically significant in model 2. Firm size, the return on assets, the prior year s return, and the cash flow are also statistically significant. These results show that younger CEOs are more likely to pursue the challenges resulting from acquiring a target outside its normal line of business. In the next set of models we examine how the age of the CEO influences the medium of payment. In models 3 and 4, the dependent variable is one if CEOs use all cash as payment to complete the deal and zero otherwise. The independent variables in models 3 and 4 are the same as those in models 1 and 2, respectively. Models 3 and 4 show that younger CEOs are more likely to use cash to pay for an acquisition than their older peers. We also find that larger firms, firms with better prior stock return performance, and older firms tend to use cash more often when paying for acquisitions. Overall, the results in Table 4 provide strong evidence that younger and older CEOs pursue acquisitions differently. Younger CEOs are more likely to undertake acquisitions which are outside their line of business, suggesting that they pursue more risky targets. Further, they use cash more often than their own firm s stock to pay for these acquisitions. This latter finding 12

is consistent with the conjecture that younger CEOs are more positive about the value of their firm s equity and hence less willing to use it as payment in a merger. We further examine the issue of CEO and the undertaking of major corporate investments with an analysis of capital expenditures in Table 5. That is, we test whether CEO age has a broader effect on corporate projects that involve growth. If CEO age is a proxy for the firm s desire to grow, then it should be also inversely related to corporate capital expenditures. We examine this issue for both large and small capital investment projects by using five percent of the firm s assets as a project threshold measure. In Table 5 we estimate two models of the relationship between age and large capital expenditures. Model 1 examines CEOs undertaking capital expenditures of at least 5% of the firm s assets in the current year. The dependent variable in this model is a binary variable with a value of one if the capital expenditure is 5% or more of the firm s assets. In model 2, the dependent variable is a dummy variable which is one if capital expenditure is smaller than 5% of the firm s capitalization. We find that the relationship between age and expenditure activity is significantly negative in model 1 for large capital expenditures but not in model 2 for smaller capital expenditures. This finding shows the young CEOs are more interested in large capital expenditures, and use them as signals to enhance their professional reputation. In model 3 and model 4, we test the relationship between CEO age and the size of acquisition. The dependent variable is 5% Acquisition in model 3 and <5% Acquisition in model 4. We find a negative relationship between age and acquisition activity in both the models. These results show that young CEOs are interested in both small and large acquisition deals. We conclude from this analysis that younger CEOs commit higher levels of corporate spending to capital expenditures, but only for larger projects. This might occur since only large 13

projects or acquisitions are adequate for purposes of skills signaling or responding to compensation incentives. In the following section we attempt to determine whether skills signaling or the desire for greater compensation is the more correct explanation for this relationship between CEO age and acquisition activity. 5. Acquisition Incentives The findings reported in the previous sections establish that CEO practices regarding mergers differ based on age with signaling-incentive effects present in these decisions. But our empirical analysis has been unable to determine whether signaling or the response to compensation incentives is the more compelling explanation. In this section, we explicitly test whether compensation incentives can explain age varying behavior regarding acquisition activity. Yim (2013) contends that CEOs experience permanent and substantial gains in compensation following acquisitions, usually in the forms of salary, bonus, or equity. These increases are likely to provide a larger lifetime increase in total compensation for younger CEOs because of their longer projected remaining career. Yet such an effect might not hold in the U.K. due to the additional scrutiny and constraints placed on CEO compensation (Ferri and Maber, 2013). Perhaps most importantly, the use of equity based incentive pay for CEOs is much less common outside the U.S. (Cheffins, 2003; Conyon et al. 2013). Consequently, we explore whether changes in CEO compensation can explain why younger CEOs more aggressively pursue acquisitions. 5.1 Acquisition Activity and Changes in Total CEO Compensation 14

In Panel A of Table 6, we examine the compensation awarded to CEOs after acquisition activity. We estimate the following OLS model to examine the change in compensation following acquisition activity: where: %ΔCompijt = β0acqijt + β1acqijt-1 + β2acqijt-2 +... + βracqijt-r + Xitδ + ϵijt; (2) % ΔCompijt is the percentage change in the total compensation of CEO j of firm i from year t to t+1. Acq are indicator variables for contemporaneous and lagged acquisition activity. Acqijt-r equals one if CEO j of firm i announces an acquisition whose deal value exceeds 5% of the firm market capitalization in year t-r and 0 otherwise. Xit is a matrix of controls for firm characteristics influencing CEO compensation In this model we test whether cumulative acquisition activity has an effect on CEO compensation. Specifically, β0 represents the contemporaneous effect on compensation of a 5% acquisition. We interpret β1 as the effect on compensation today of a 5% acquisition one year ago. The coefficients β2 to βr can be interpreted similarly. Therefore, we interpret the sum β0 + β1 + β2 +... + βr as the cumulative compensation effect of a 5% acquisition. This sum will be positive if acquisitions have a permanent compensation effect as argued by Yim (2013). Alternatively, if acquisitions are associated with only a one-time compensation increase, we might see a positive β0 followed by a negative β1, as compensation reverts to previous levels. In this case the sum of coefficients might be no different from zero. We use the industry-adjusted percent change in compensation as the dependent variable in our empirical analysis. Model 1 in Panel A shows that there is an increase in compensation following a 5% acquisition. The results, however, are not statistically significant. If we introduce a control for acquisition activity occurring in year t-1, the year t coefficient remains positive, but remains statistically insignificant. In model 3, we include measures of acquisition activity for years t-1 and t-2. In model 4, we extend past acquisition activity to the past previous three years. 15

We find that the year t effect and the prior year effects continue to be statistically insignificant, suggesting that past acquisition activity does not influence CEO compensation. The results from models 1 through 4 are inconsistent with a permanent increase in compensation following an acquisition. They do not support the claim that the incentive of higher future compensation drives the acquisition decisions of our sample of young U.K. CEOs. We further test the relationship between compensation and acquisition activity by separately examining the issue of completed and withdrawn mergers. If previous acquisition activity accounts for CEO compensation increases, then we should observe a significantly positive coefficient for completed mergers and no discernible effect for withdrawn mergers. Our findings for model 5 show no significant coefficients for either type of merger. We interpret this result as further evidence against the presence of a compensation effect in the acquisition decisions of young global CEOs. 5.2 Acquisition Activity and Changes in CEO Compensation Components Although total CEO compensation does not appear responsive to acquisition activity, it might be that specific components do adjust. Consequently, in Panel B we separately examine changes in the salary, bonus, and equity-based components of CEO compensation following acquisitions. In models 1 and 2, the dependent variable is the annual percent change in fixed salary. In models 3 and 4, the dependent variable is the annual percent change in bonus. In the last set of specifications, models 5 and 6, the dependent variable is the equity component of compensation which includes option grants and restricted stock. All of the percentage changes are net of corresponding industry median adjustments. We obtain several interesting findings in Panel B. First, we observe that acquisition activity does not positively influence CEO fixed salary levels. Indeed, we find some evidence in 16

model 1 that it actually reduces the level of fixed salary. Second, we determine that CEO bonuses do not increase with merger activity. Finally, contrary to the results for U.S. CEOs, we find that acquisition activity has a consistently negative effect on the equity incentive component of compensation. Overall, the results in Panel B fail to support a claim that compensation incentives explain the acquisition activities of CEOs operating in the U.K. We conclude from the results in Panel B that CEO equity wealth is not a determinant of subsequent acquisition activity. That is, those CEOs whose compensation is more concentrated in equity are not more likely to engage in acquisitions. As a further test of the incentive effects of compensation on CEO acquisition activity, we examine the role that CEO equity wealth plays in acquisition activity. If compensation is a major incentive for CEO acquisition decisions, then the strength of the CEO age-acquisition relationship should vary across levels of CEO equity wealth. In the following analysis the dependent variable is equal to one if the firm announces an acquisition whose deal value exceeds 5% of the firm's market capitalization and zero otherwise. Table 7 contains our empirical findings. Since Yim (2013) reports that the post-acquisition compensation response is primarily equity-driven, it follows that young CEOs who get paid mainly in the form of equity are especially likely to anticipate a decline in their post-acquisition compensation. Therefore, in models 1 and 2 of Table 7, we construct terciles of CEO-firm year observations based on the fraction of total compensation three years prior received by CEOs in the form of stock and option grants. In models 1 and 2, we interact CEO age with these terciles to determine the effect that 17

CEO wealth exerts on younger CEOs. In both the models, the interactive effect is statistically insignificant. In models 3 and 4, we use an alternative measure of anticipated CEO equity-based compensation. Our equity wealth variable is calculated as the sum of the values of restricted and unrestricted stock holdings and option holdings. We reason that CEOs who have accumulated high levels of equity-based wealth have a history of receiving large equity-based grants, and likely have greater concerns of losing equity compensation. The corresponding interaction term, Age*High Tercile, in models 3 and 4 again is statistically insignificant. Models 5 and 6 explore whether the age effect is stronger in firms where CEOs have greater ability to influence post-acquisition compensation. Bebchuk and Fried (2003) as well as Jensen and Murphy (1990) observe that compensation is a negotiated process between the CEO and the board. Although a board might hire compensation consultants, the ultimate CEO payment is influenced by the CEO s bargaining power relative to the board. Models 5 and 6 use CEO tenure as a proxy for the CEO s power to influence the board. As noted by Hermalin and Weisbach (1998), CEOs with longer tenure have a greater ability to hire senior management, appoint directors, and generally influence the board. Therefore, in models 5 and 6, we examine whether the age-acquisition relationship is stronger among more tenured CEOs. The interaction terms of CEO age with terciles based on CEO tenure are again statistically insignificant. The results in Table 7 in combination with those presented earlier in Table 6 are inconsistent with the finding based on U.S. data that CEOs are rewarded with large increases in equity compensation for merger activity. For instance, Grinstein and Hribar (2004) report that CEOs are paid large one-time bonuses for completing acquisitions. Yim (2013) finds that acquisitions are followed by large increases in the CEO s total compensation, especially in the 18

equity component. Rather these results are consistent with previous findings (e.g., Towers Perrin, 2001; Cheffins, 2003; Conyon et al. 2013) that report less use of incentivized compensation for CEOs around the world. 5.3 Post-Acquisition Turnover In Table 8, we examine whether the threat of executive discipline through CEO turnover influences the age-acquisition relationship. In this table, we run a set of logit regressions that estimate the likelihood of CEO turnover in the year following an acquisition. For a firm i in year t, the dependent variable Turnover equals one if the CEO leaves office in year t+1 and zero otherwise. The independent variables include CEO age and other determinants of CEO turnover, such as CEO tenure, firm size, and prior stock and operating performance as well as industry and year fixed effects. To control for the effect of acquisition activity on turnover, we also include the binary variable 5% Acquisition. The coefficients are reported as odds ratios. In Panel A, we examine turnover using CEO-firm year data. In model 1 we test the entire sample of CEOs and find that age is positively correlated with turnover, which suggests that older CEOs are more likely to turn over. In models 2 and 3, we consider separate young and old CEO samples. These results fail to show that younger CEOs are more likely to change employment. The models in Panel A also show that the coefficient for the 5% Acquisition binary variable is statistically insignificant. This suggests that even after large acquisitions, the probability of turnover does not increase for CEOs regardless of their age. The overall importance of these findings is not that older CEOs turn over more, but that younger CEOs do not. Hence, the threat of dismissal is less of a factor in explaining the merger decisions of younger CEOs. 19

We continue our analysis of CEO turnover and acquisition activity in Panel B where the unit of analysis is deal-ceo firm-year. Model 1 is for the full sample of CEOs whereas models 2 and 3 are for subsamples of young and old CEOs, respectively. Models 4 through 6 follow models 1 through 3 but also include controls for the quality of the 5% acquisitions. We do so by assigning the acquisition announcement day returns to low, middle or high return terciles. Binary indicator variables are created to account for assignment to the bottom, middle and top return tercile. Models 1 and 4 for the full sample of CEOs show a consistent significant relationship between age and turnover, indicating that it is older and not younger CEOs who have the highest likelihood of turnover. Although the results from these models suggest some limited increase in turnover likelihood, there is no evidence to suggest that this increase is concentrated among younger CEOs. That is, even in the presence of low quality acquisitions as measured by their announcement period returns, young CEOs do not face an increased likelihood of turnover. Overall, the analysis in Table 8 does not support the claim that younger CEOs are deterred from acquisition activity due to turnover concerns. In that sense, these results are inconsistent with the CEO conservatism effect. They suggest that younger CEOs can consider acquisitions without the fear of discipline through termination, regardless of the perceived quality of the merger. 6. A Possible Alternative Explanation An alternative to career concerns as an explanation for the difference in acquisition behavior between young and older CEOs relates to changing personal characteristics. Psychological and physiological changes that occur with aging can directly affect a CEO s 20

inclination to pursue acquisitions. One such characteristic that is relevant for our study of acquisitions is overconfidence. Roll (1986), Heaton (2002) and Malmendier and Tate (2005, 2008) propose that CEOs suffer from overconfidence. These studies use the tendency of younger CEOs to overestimate their ability to increase firm value through acquisitions to explain the prevalence of valuedestroying mergers. This overconfidence propensity plausibly changes with time and might serve as the basis for an alternative explanation to signaling. Overconfident CEOs systematically overestimate their ability to create value from their investments and believe their firm is undervalued. Therefore, they tend to overinvest when the firm s internal resources are plentiful. When firms are capital-constrained, however, they are reluctant to issue equity to raise cash (Roll, 1986; Heaton, 2002; Malmendier and Tate, 2005, 2008). Malmendier and Tate (2005, 2008) find that overconfident CEOs are more likely to undertake poor acquisitions, especially when their firms are cash-rich. Overconfident CEOs should be more likely to make investments if their firms are cash-rich. Yim (2013) tests for the presence of overconfidence in the acquisition activities of U.S. CEOs by arguing that (1) overconfidence declines with age and (2) the age-acquisition relationship is stronger in cash-rich firms. We follow this approach in our subsequent analysis. Although we find evidence in Table 5 that younger CEOs are more likely to make large acquisitions than older CEOs, we fail to find that this relationship is more pronounced in cash rich firms. In Table 9, we divide the sample firms into two groups based on their cash flows and examine the two groups separately. The first two models examine the relationship between age and large acquisitions for the two groups of firms. The first model shows that the age effect is 21

statistically strong even among firms whose cash flow is below the median, which is inconsistent with overconfidence. Although the second model shows that among the cash-rich firms, age still plays a role; our results for cash-constrained firms suggest that over confidence does not explain the merger decision. In the next two models we examine a further implication of CEO overconfidence. Overconfident CEOs will use cash more often to pay for mergers since they tend to view their equity as undervalued (e.g., Malmendier and Tate, 2005). Further, this effect should be stronger in cash rich firms. Our results for the models (3) and (4) show that there is no difference between the two cash flow groups. That is, the logit coefficients for CEO age are statistically insignificant both the groups, with a value of one. This means that CEO age has no effect on the likelihood of a cash bid. These results are inconsistent with an overconfidence explanation for the acquisition practices of young CEOs. Finally, in the last two models of Table 9, we test whether overconfidence might explain the age and capital expenditure behavior of young CEOs. We find a significant relationship between age and capital expenditures for cash constrained firms. Again, these results are strong evidence against overconfidence being the explanation for our results. We conclude from Table 9 that overconfidence does not explain the acquisition practices of young CEOs. Overconfidence effects should be more pronounced in firms with greater cash resources. Our results fail to find consistent evidence of this in decisions about large acquisitions, the method of acquisition payment, or the level of capital expenditures. 22

7. Conclusions Based upon a sample of U.K. firms, we examine whether career concerns can explain the relationship between a CEO s age and corporate acquisition decisions. We find that career concerns do effect mergers, but that CEO conservatism does not. Rather, we determine that younger CEOs pursue large scale acquisitions more frequently than their older peers. That is, we find that the age-acquisition relationship is inverse. This relationship also holds for capital expenditures. We further determine that younger and older CEOs pursue acquisitions differently. Younger CEOs are more likely to undertake acquisitions which are outside their line of business, suggesting that they pursue more risky targets. They also use cash more often than their own firm s stock to pay for these acquisitions. In aggregate, our initial findings support the presence of a CEO signaling-incentive effect in the relationship between age and acquisition activity. But the inverse relationship between age and acquisition due to the CEO signalingincentive effect can reflect either signaling or compensation incentive explanations. Consequently, we undertake additional empirical analysis to separate these two possible explanations. We accomplish this by examining the relationship between compensation and acquisition levels. Unlike Yim s (2013) results for U.S. CEOs, we find no relationship between acquisition activity and changes in CEO compensation. We determine that the incentive of higher future compensation does not drive the acquisition decisions of young CEOs. Our findings hold across the salary, bonus, and equity components of CEO compensation. This result emphasizes how sensitive the effect of compensation on merger activity is to international differences in design and payout level. We also fail to find that younger CEOs are deterred from acquisition activity due to turnover concerns. It appears that younger CEOs can pursue 23

acquisitions without excessive concern about turnover, regardless of the perceived quality of the merger. An alternative explanation for the relationship between age and acquisition activity outside of the career concerns is overconfidence. But we determine that overconfidence does not explain the acquisition practices of young CEOs. As noted by Yim (2013), overconfidence effects should be more pronounced in those firms that have greater cash resources. Our results fail to find consistent evidence of this across large acquisitions, the method of payment for acquisitions, and capital expenditure levels. Consequently, we cannot support overconfidence as an explanation for the decision of younger CEOs to pursue acquisitions more intensely. Based on these empirical findings, we conclude that executive signaling best explains the inverse relationship between CEO age and acquisition activity. Young CEOs might wish to demonstrate their expertise and abilities by pursuing a large acquisition. By undertaking highly uncertain, but promising activities, younger CEOs can distinguish themselves among their peers and create reputational value for themselves. Overall, this paper contributes to the literature examining distortions in corporate decision-making that originate from the career concerns of CEOs. By focusing on a set of CEOs of U.K. firms, we are able to provide new evidence that younger CEOs send active signals to the managerial labor markets by investing in the growth of the firm. Further, our research highlights the difference between the U.S. and the U.K. regarding the role that CEO compensation plays in mergers. We conclude that the impact of CEO characteristics such as age on firm policy is a global phenomenon, but that the nature of its specific effect is likely to vary across countries. 24

Appendix: Description of Independent Variables 5% Acquisition is a binary variable that equals one if the CEO announces an acquisition that exceeds 5% of the market cap. 5% Acquisition [t-r] is a binary variable that equals one if the CEO announces an acquisition whose value exceeding 5% of the market cap in year t-r. 5% Acquisition [Completed] is a binary variable that equals one if the CEO announces an acquisition that exceeds 5% of the market cap and is completed after announcement. 5% Acquisition [Withdrawn] is a binary variable that equals one if the CEO announces an acquisition that exceeds 5% of the market cap and is withdrawn after announcement. 5% CapEx [t-r] covariates are binary variables that equal one if the current CEO undertakes capital expenditures that exceed 5% of the firm's assets in year t-r. <5% CapEx [t-r] covariates are binary variables that equal one if the current CEO undertakes capital expenditures whose values are smaller than 5% of the firm's assets in year t-r. ADR is a dummy variable with a value of one if the firm has ADR. ADR_Age is an interaction term of ADR and CEO Age. Assets is book value of assets. Cash Flow is EBITDA less interest, taxes, and dividends. CEO Age is CEO's age reported in BoardEx dataset in each observation year. CEO Compensation is CEO's total direct compensation. CEO Tenure is the number of years the CEO has held the CEO title at the current firm. Deal Value is the total deal value of acquisitions announced by the CEO in the firm-year over the full sample. Deal Value >0 is the total value of acquisitions over the subsample of firm-ceo years when there is a merger. Equity Fraction is the fraction of prior 3-year compensation that CEOs have received in the form of stock and option grants. Equity Wealth is the sum of the value of the CEO's stock and option holdings. Firm Age is the current year less the firm's first year of IPO in Compustat. 25

Firm Age Proxy is retained earnings scaled by total assets. Low_3DayReturn is a binary variable that equals one if the 3-day (t-1, t, t+1) cumulative abnormal return of the day t acquisition announcement falls in the lowest tercile group. Mid_3DayReturn is a binary variable that equals one if the 3-day (t-1, t, t+1) cumulative abnormal return of the day t acquisition announcement falls in the lowest tercile group. Market Capitalization is the market value of the outstanding equity. Market to Book is the market value of assets (total capitalization) divided by book value of assets. Market Leverage is the ratio of total debt to total equity. In Table 7, "Middle Tercile" and "High Tercile" are binary variables that correspond to terciles of the variables "Equity Fraction," "Equity Wealth," and "CEO Tenure," in columns 1, 2, and 3, respectively. Prior Year Returns are cumulative monthly returns from the prior year. Relative Deal Value is the deal value normalized by the beginning of year market capitalization. Relative Deal Value >0 is the total deal value of acquisitions normalized by the beginning of year market capitalization over the subsample of firm-ceo years when there is a merger. ROA is net income divided by assets. Total Capitalization is book assets plus market value of equity less book value of equity. 26