Justina Klare - 1943731
Anticipated CEO Turnover and M&A Performance
Do retiring managers perform worse that average M&A deals?
First Supervisor:
Dr. K. J. McCarthy -k.j.mccarthy@rug.nl- Second Supervisor:
F. Noseleit -f.noseleit@rug.nl- Master Thesis
MSc Business Administration Strategic Innovation Management Rijksuniversiteit Groningen
Faculty of Economics & Business 12th July, 2013
1
ABSTRACT
This thesis studies the effect of retiring managers on merger and acquisition performance, using a sample of 37,530 U.S acquirers. This research is based upon the concept of Agency Theory, yet goes beyond its traditional view. Whereas in Agency Theory managers are seen as a constant variable, this study proposes that different types of managers have different effects on firm performance and shareholder wealth creation. The focus of the study lays on the special role of managers anticipating retirement, and proposes that these are behaving and performing differently than other types of managers. The concept of Legacy Theory is introduced as a possible explanation for the motives of retiring managers to perform last- minute acquisitions. The study uses an event study methodology following Brown and Warner (1985). The results indicate that managers anticipating retirement conduct last- minute deals in their final period in office, and do particularly poorly in these acquisitions.
This is an important finding as it shows that not only M&A performance influences management turnover, but that the effect is also true vice versa.
Key words: Mergers and Acquisitions, anticipated CEO turnover, agency cost, acquirer performance, retirement, legacy theory
2
TABLE OF CONTENTS
1.INTRODUCTION 3
2.LITERATURE REVIEW 5
2.1AGENCY THEORY AND CEOTURNOVER 5
2.2MANAGERIAL SELF-INTEREST IN THE RETIREMENT PROCESS (LEGACY THEORY) 7
2.3ANTICIPATED CEOTURNOVER &M&APERFORMANCE 9
2.4MODERATORS 10
2.4.1PAYMENT TYPE 10
2.4.2FIRM SIZE 11
4.DATA &METHODOLOGY 12
4.1SAMPLE SELECTION 12
4.2INDEPENDENT VARIABLE 12
4.3DEPENDENT VARIABLE 14
4.4MODEL SPECIFICATION 15
4.5MODERATORS 15
4.6CONTROL VARIABLES 16
5.RESULTS 17
5.1DESCRIPTIVE STATISTICS 17
5.2HYPOTHESES ONE:ANTICIPATED CEOTURNOVER AND M&APERFORMANCE 21
5.3HYPOTHESES TWO &THREE:MODERATING EFFECT OF PAYMENT TYPE AND FIRM SIZE 23
6.DISCUSSION 25
6.1PRACTICAL IMPLICATIONS 27
6.2LIMITATIONS 28
7.CONCLUSION 29
8.REFERENCES 30
9.APPENDIX 34
3
1.INTRODUCTION
In the past decades mergers and acquisitions have increasingly gained attention in international business and are seen as a promising source of value creation. Within the sixth merger and acquisition wave from 2003 to 2008, the worldwide volume of M&A amounted to approximately 15.5 trillion U.S. dollars (Statista, 2012). In 2004 alone, about 30,000 acquisitions were completed globally, equivalent to one transaction every 18 minutes (Cartwright & Schoenberg, 2006). Even though, M&A are seen as value enhancing, between sixty and eighty percent of all mergers are eventually considered as failures (Puranam & Singh, 1999). Despite this inevitable evidence, that a large majority of mergers fail to deliver profits, companies in every industry continue to see mergers and acquisitions as the answer to their problems (Haleblian, Devers, McNamara, Carpenter, & Davison, 2009). This paradox finds great attention in academic literature, yet still no panacea for M&A`s high failure rate could be found. In the light of these negative performance consequences for acquirers, more recent financial and strategic literature focused on antecedents of M&A, as scholars aim to undercover why firms merge (Haleblian et al., 2009). One school of thought which attempts to explain M&A antecedents is the theory of agency costs. Claiming that rational but self-serving managers act in order to maximise a private utility function, which fails to positively affect firm value, agency theory explains why an overwhelming percentage of M&A deals are value destructing (McCarthy, 2010).
Presently, literature solely focuses on current managers and their impact on firm performance. This paper will extend the management literature by considering the role of management turnover and its effect on performance. Reasons for management turnover can be diverse, but this study focuses on anticipated turnover and its effects. Anticipated turnover involves all cases when a CEO retires, is of old age, as well as when he/she stays with the company (Carapeto, Moeller & Faelten, 2010). This type of management turnover has a special psychological effect on retiring managers which is proposed to motivate them to perform last-minute acquisitions. CEOs anticipating retirement come into a psychological state of their life, where they have to accept the truth that time has come to step down and hand over power. For many, this truth is difficult to accept, and the fear of obsolescence and
4 losing power influences their rational decision making (Kets de Vries, 2003). The desire for power, prestige and empire, and the fear of obsolescence are proven to be examples of psychological motives for mergers which are most often considered to be value destructing (Ashton-James, McCarthy, & Dranca-Iacoban, 2010). In their last period in office, retiring managers are especially exposed to these psychological fears which therefore, most likely increase their motivation to make last-minute deals. This study will consider the different role and motivations of retiring managers and addresses the following research question:
Do retiring managers of the bidder firm make worse than average M&A deals?
The paper will consist of seven chapters structured as follows; Chapter 2 is a review of relevant academic literature on, agency theory, CEO turnover, and M&A performance.
Chapter 3 considers potential moderators. Chapter 4 explains data, samples and methodology used for the study. In Chapter 5 the empirical results of the study are presented. Chapter 6 then discusses these results and limitations, and gives practical implications. In the end, Chapter 7 serves as a reflection and summary of the study’s findings and will give insights into future research.
5
2.LITERATURE REVIEW
2.1AGENCY THEORY AND CEOTURNOVER
Estimates suggest that managers are responsible for nearly one third to one half of all merger failures (Ashton-James et al., 2010). One explanation for this is agency theory.
Agency theory is assuming rational but self-serving managers which deliberately account for net losses in order to reinforce their job positions, or increase firm size to maximise a private utility function (Marris, 1963; Amihud & Lev 1981; Rhoades 1983; Black, 1989, Shleifer & Vishny, 1989). Generally, agency theory can be divided into two different groups:
the first, managerial entrenchment theory, assumes that M&A fail as managers primarily make investments that minimise the risk of replacement (Shleifer & Vishny, 1989). The second group of agency theory, empire building, stresses that managers are only motivated to invest in the growth of their firms revenues if this investment is subject to a minimum self-profit (Marris, 1963).
Thus far, agency theory explains the different objectives of shareholders and managers;
however it misses a comparison between different types of managers. In agency theory managers are seen as a constant, where all managers are the same type with the same effect on firm performance. This study extends the concept of agency theory by considering that current and outgoing CEOs behave and perform differently, and have different motivations to acquire.
Reasons for CEO turnover can have different causes and send out different signals for change (Huson, Malatesta, & Parrino, 2004). As shown in Figure 1, CEO turnover is generally classified as either “forced” or “voluntary”. “Forced” turnover is related to disciplinary actions taken by the board of directors due to disappointing performance, hostile takeovers, or bankruptcy (Carapeto, Moeller, & Faelten, 2010). “Voluntary” turnover, on the other hand, is associated with all “normal” turnovers as a result of for instance retirement (Carapeto, et al., 2010). As a CEO has no influence on “forced turnover” and in
6 most cases cannot foresee to be leaving a company soon, this type of turnover is not considered.
“Voluntary” turnover, on the other hand can be anticipated. According to the definition of Carapeto et al. (2010), anticipated turnover involves all cases when a CEO retired/is of old age as well as cases when he/she stayed with the company as a director/other senior position. Anticipated turnover is generally a planned event and a major task of many CEOs approaching retirement is finding a successor (Puffer & Weintrop, 1995). This transition process provides a great avenue for self-interested behaviour of the outgoing manager (Kets de Vries, 2003).
CEO TURNOVER
FORCED
INTERNAL EXTERNAL
ANTICIPATED UNANTICIPATED
VOLUNTARY
Disciplinary actions taken by the board due to or example bad performance of CEO
When the company becomes the target in a takeover process (M&A or hostile takeover), or through bankruptcy
All cases when a CEO retired, is of old age as well as when he/she stayed with the company
Death, illness, or poaching of the CEO by another firm
Figure 1. Types of CEO Turnover
7 2.2MANAGERIAL SELF-INTEREST IN THE RETIREMENT PROCESS (LEGACY THEORY)
Even though, retirement is an inevitable process, many CEOs struggle to accept that their time has come to step down (Kets de Vries, 2003). In order to accept this truth, many CEOs have to overcome some hidden inner fears, which are on the one hand the denial of death and the fear of losing power, and on the other hand, a profound wish to leave a legacy (Kets de Vries, 2003). The fear of obsolescence, the desire for power, prestige and empire are proven to be examples of psychological motives for mergers (Ashton-James et al., 2010).
Managers facing retirement are facing these psychological states, and are thus likely to engage in mergers close to their retirement date. It is proposed that these managers conduct a last-minute deal to leave behind a legacy in order to overcome their fear of being obsolete. For a CEO the chance to leave behind such a reminder of one’s accomplishments can amount to defeating death, a deep psychological desire (Kets de Vries, 2003). This remainder in form of a merger or acquisition may be valuable for the outgoing manager. For the firm however, these mergers are assumed to be value destructing, due to two reasons.
When the time comes to step down and hand over power, the outgoing CEO may worry that a successor will disrespect the legacy and destroy what he or she has carefully built up over the years (Kets de Vries, 2003). As studied by Puffer & Weintrop (1995), CEOs have a psychological need to leave a legacy of accomplishments, which they wish to be perpetuated by their successors. In order to ensure one’s legacy’s survival, the CEO may look for a successor who will carry on in exactly the same way. This search for a “clone” is however doomed to fail, given that a company needs changes over time, and therefore what is right for the present may be a pitfall for the future (Kets de Vreis, 2003). Moreover, it is proven that position imprints, the legacies left behind by the incumbents, constrain subsequent position holders (Burton & Beckman, 2007). This is a first potential explanation for value destruction. Secondly, some CEOs may secretly raise the hope that their successors will fail, as failure would be proof of their own indispensability, and they will make sure, unconsciously or not, to set their successor up for failure (Kets de Vries, 2003).
8 These literature insights give reason to believe that outgoing managers engage in mergers and acquisition as a result of self-interested behaviour in order to build up and protect their legacy. Facing retirement, the CEOs are either likely to choose a successor which will continue in the same manner as them in order to carry on their legacy, which will in turn negatively affect a firm’s future prospect to change, or the incumbents might set their successors up for a failure to resemble their own indispensability. Their desire to leave and/or protect a legacy might be greater than their desire to increase shareholder value.
Therefore, the following concept of Legacy Theory as an extension to agency theory in explaining reasons for M&A failure is proposed:
Proposition 1: Legacy Theory - Retiring managers pursue last-minute M&A deals not in an effort to maximise firm value, but in an effort to leave a legacy, and/or potentially setting up their successor for failure as a proof of their own indispensability.
9 2.3ANTICIPATED CEOTURNOVER AND M&APERFORMANCE
As this literature review has shown, there is empirical ground to believe that CEO turnover and M&A performance are in a way interrelated. Past studies however, solely focused on the relation of CEO turnover post M&A performance. Scholars could find proof that target firms experience a higher than normal turnover of management after an M&A deal is conducted (Haleblian et al., 2009). Additionally Lehn & Zaho (2006) proved that also in bidder firms CEO turnover is related to the amount of value CEOs create or destroy through M&A. Following an acquisition 57% of CEOs were replaced, of which 83% were replaced within five years (Lehn & Zaho 2006). Research is however lacking an understanding of the diverse relationship, namely how CEO turnover in the bidder firms affects M&A performance.
Literature provides evidence to suggest that a change in the bidder firm’s management might have a negative effect on M&A performance. Warner & Watts (1987) for instance investigated that, there is an inverse relation between the probability of a management change and a firm's share performance. In a later study Beatty & Zajac (2006) examined that announcements of CEO changes are typically associated with a reduction in the value of the firm, as reflected in the perceptions of the stock market. It could moreover be proven, that external CEO succession is associated with M&A deterioration (Carapeto et al., 2010).
Based on these assumptions and the aforementioned legacy theory, the following is hypothesised:
Hypothesis 1: Retiring managers make worse than average M&A deals in search of legacy and therefore negatively influence M&A performance.
10 2.4MODERATORS
Given that bidder firms generally fail to create value from mergers and acquisitions, research has increasingly focused on developing a better understanding of the specific conditions that moderate the management and M&A performance relationship (Haleblian et al., 2009). Also for this research moderating affects will be taking into account, focusing on deal and firm specific characteristics.
2.4.1 PAYMENT TYPE
As proposed in the concept of Legacy Theory, a retiring manager’s desire to leave behind a legacy leads him to think comparatively irrational. This may also impact his/her decision making on what type of payment method to choose in order to finance a deal. When paying for their acquisitions, managers face a trade-off between paying by cash or stock shares (equity), which has shown to provide an avenue for agency conflict probable to be resulting in value destruction (Harford et al, 2010). The ultimate payment method may be based on the value the acquirer places on his firm. Managers who finance their deals with cash, signal believe in their firm stock being undervalued, whereas managers making equity payments signal an overvaluation of their firm’s stock value (King, Dalton, Daily, & Covin, 2004).
Scholars proved that cash-financed deals are more profitable for shareholders, than equity- financed deals (Travlos, 1987; Loughran & Vijh, 1997; Carow, Heron, & Saxton, 2004).
However, still many overvalue their firm stock and chose to pay with equity, which is according to Hirschenschliefer (2001) a result of biased emotional decision making.
Retiring managers are considered to be especially emotional biased in their last period in office. They are more likely to overvalue their firm stock, as a greater stock value may represent their own power and status. Based to these facts, the following moderating effect is hypothesized:
Hypothesis 2: Financing through equity payments will increase the negative effect of retiring managers on M&A performance.
11 2.4.2FIRM SIZE
A second moderating effect that will be taken into account is the influence of firm size on the relationship between anticipated CEO turnover and M&A performance. As investigated by Moeller, Schlingeman, & Stulz (2004), M&A performance decreases with size, where managerial hubris play a greater role in the acquisition decision of large firms than those of small. Moreover, managers in small companies typically have more ownership than managers in large companies, which increases the potential for agency costs (Demsetz &
Lehn, 1985). Also the effect of retiring managers on performance is hypothesized to be more negative in large firms. During their time in office, CEOs in big firms enjoyed a greater amount of power and had more responsibility for a larger group of employees than those of small firms. In line with the reasoning of Legacy Theory, CEOs retiring from large firms are thus likely to have a greater fear of obsolesce and losing power. Their wish to leave behind a legacy should therefore also be greater, giving them an even stronger motivation to perform last-minute acquisitions. Another factor, supporting this notion is the shorter tenure of CEOs in large firms. Ronghui & Hailin (2007) showed that large firms tend to remove their top-level executives more frequently than do small firms. This leads to suggest that managers in large organizations have a greater felling of being replaceable than CEOs of small firms. As a consequence their desire to leave behind a legacy should be greater.
It is therefore hypothesized that firm size has an effect on the relation between CEO retirement and M&A performance. As agency costs are on average higher in large firms, and CEOs retiring from large firms have a stronger feeling to leave behind a legacy, the following moderating effect is hypothesized:
Hypothesis 3: Large firms will increase the negative effect of retiring managers on M&A performance
12
4.DATA&METHODOLOGY 4.1SAMPLE SELECTION
To select a sample in order to test the hypotheses of outgoing managers on M&A deal performance, deal specific data from the Thomson Financial SDC Merger and Acquisition database was employed. The sample includes deals satisfying the following conditions: (1) the deal is announced between January 1, 1990 and 31 December, 2012; (2) the acquisition is completed; (3) the acquirer is located in the United States; (4) the target is located both within and outside of the United States; (5) the acquirer and the target are not owned by the same ultimate parent; (6) variables necessary to run the analysis are not missing.
Additionally, firm specific data was attained from DataStream, including firm data on Return Index and Tobin’s Q. The selection process produced a sample of 37,530 M&A deal observations.
4.2INDEPENDENT VARIABLE
CEO turnover resulting from retirement is used as the independent variable in this study.
The main sources of CEO turnover information are proxy statements, company reports, and press releases attained from LexisNexis over a period of 23 years (1990-2012). The information on retirement was carefully analyzed. Only cases where the manager was
“voluntarily” retiring were selected. A dummy variable for retired managers was created where all retired managers whose companies matched the base data were given the value of one. A total of 2,111 retired managers were found. These were further coded by the year, corresponding to their retirement date. Additionally, the retirement dates were coded per timeframe that they lay within the merger date effective, as visualized in Figure 2 for the first two timeframes. The retirement dates were coded with the value of one if the retirement was within half a year before or after the merger, one year before or after the merger, two years before or after the merger, three years before or after the merger, or four years before or after the merger.
13 Figure 2. Timeframe Date Retirement and Date M&A Effective
14 4.3DEPENDENT VARIABLE
The dependent variable used for this study is firm performance. Given the studies focus on agency theory and therefore, shareholder wealth creation, the accumulated abnormal return (CAR) to acquiring firms around the announcements of a mergers and acquisitions will be used as a performance metric. The accumulated abnormal return (CAR) reflects investors’ response to the announcement of an acquisition, based on present expectations about the future cash flows of a combined firm (bidder+ target). Following the event study methodology of Brown & Warner’s (1985), The CARs will be measured over several event windows. These include the abnormal return on the date announced [0], the CAR measured one trading day before through one trading day after the announcement [-1, 1], the CAR measured five trading days before through one trading day after the announcement [-5, 1], and the CAR measured 20 days before through one trading day after the announcement [-20, 1]. Following Fuller, Netter & Stegmoller (2006) the following market adjusted model is estimated:
= -
In the equation, represents the acquirer i’s abnormal return, is the stock return on acquirer ἱ, and is the return of the market index which is based on the S&P500 composite index. The total share price reaction to an event can be obtained by calculating the abnormal return per firm:
=
The time period used in the regression is the event window, returning the abnormal return (AR) for acquirer ἱ, where 1 is the start (-1, -5, and -20 days) and 2 the end (+1 days) of the event window. The resulting cumulative abnormal return provides a measurement of the acquirer’s performance.
15 4.4MODEL SPECIFICATION
The effect of management turnover on M&A performance is tested with OLS regression. The acquirers’ abnormal return is modeled using the following specifications:
+
where = acquirer i’s cumulated abnormal return = a dummy variable indicating a retired manager
= a vector with firm-specific control variables (described below) = moderating variables (described below)
= normally distributed error term
In all three estimation windows it is controlled for unobserved effects, as year dummies are included. Furthermore, the model is adjusted to account for heteroscedacity using the robust estimation of variance according to White (1980).
4.5MODERATORS
The variables payment type and firm size are the moderators in this study. In order to observe for the effect of payment type on the relation between CEO turnover and M&A performance the acquirer’s EBITAD (Earnings before interest, taxes, depreciation and amortization) is used. To control for the acquirer’s firm size the deal value is used as a proxy. The data for these variables was employed from the Thomson Financial SDC Merger and Acquisition database.
16 4.6CONTROL VARIABLES
Prior literature shows that a number of firm-and deal-specific characteristics affect M&A behavior and M&A performance (King et al., 2004). Therefore, control variables are included in the model to account for unobserved effects.
Public versus Private: Carpon & Shen (2007), provide evidence that acquirers of private targets perform better than acquirers of public targets on merger announcements.
Therefore, the study controls for the target by status using a dummy variable.
Historical Performance: Scholars have stressed the importance of the role of historical performance in acquisition events (Heron & Lie, 2002; Haleblian et al, 2009). One way to measure this performance is by calculating Tobin’s Q, which represents the ratio between the market value and replacement value of the same physical asset. Tobin’s Q is calculated as follows: (Equity Market Value + Liabilities) / (Equity Book Value + Liabilities Book Value). Serevaes (1991) reported that bidders’ abnormal returns are higher when their Tobin’s Q ratios are higher. Moreover, Rau & Vermeulen (1989) show that firms with agency problems, so low Tobin’s Q ratios, generally invest in negative net present value projects, including acquisitions
Similarity: Haleblian et al. (2009) found that positive transfer is dependent on bidder-to- target similarity. Similarity is controlled for in this study by comparing bidder and targets on two-digit SIC codes (Standard Industrial Classification).
Cross-border: Datta & Puia (1995) findings on cross-border acquisitions by U.S. firms suggest that cross-border acquisitions, on average, do not create value for acquiring firm shareholders.
.
17 0 0 7 5 5
0 8 13 12 4
18 26 6
24 16 14 74 71
132
93 77
93 112
69
0 20 40 60 80 100 120 140
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Year
5.RESULTS
5.1DESCRIPTIVE STATISTICS
After calculating the cumulative abnormal return, the sample is reduced to a total of 11,039 observations, where in 10,262 observations there was no case of retirement in the acquiring firm, and in 777 cases a manager retired from the acquiring firm during the M&A process. Figure 3 displays the number of retiring managers per year, and shows an increase in CEO retirements in the last decade. In the years 2008 and 2012 the greatest amount of CEOs retired.
Figure 3. Number of Retired Managers per Year
18 The retiring managers were furthermore coded per timeframe in which their retirement lies before or after the merger was effective. As shown in Table 1, in case of retirement, 50 of the 777 managers (6,43%) left the bidder firm within half a year before or after of the M&A was effective, 152 (19,56%) retired within one year before or after the M&A, 83 (10,68%) retired within two years before or after the M&A, 83 (10,68%) retired within three years before or after the M&A, and 63 (8,11%) retired within four years before or after the M&A.
Table 1. Number of (Retired) Managers per Timeframe
At the industry level, the results show that the majority of CEOs retired in Manufacturing, where 392 out of 777 CEOs retired (50,5). Also for the non-retiring managers the largest amount of CEOs is represented in Manufacturing, with 3,047 CEOs (29,7%), followed by Services, with 2,310 CEOs (22,5%).
Timeframe No.
Retired Managers
No. Same
Managers Retired managers in
%
Half Year 50 10,989 0,45%
One Year 152 10,885 1,38%
Two Years 83 10,955 0,75%
Three Years 83 10,955 0,75%
Four Years 63 10,970 0.57%
19 Table 2. Number of (Retired) Managers per Industry
In Table 3 the correlation matrix of the event windows is shown based on Pearson Correlation, where CAR5 and CAR20 show the strongest correlation between the event windows.
Table 3. Relationship between the event windows, Pearson Correlation
Table 4 displays the descriptive statistics of the variables employed in this study with the corresponding mean and standard deviation. What stands out is the high variance in (deal) value.
Industry No.
Retired Managers
No. Same
Managers Total Retired managers in %
Agriculture 0 28 28 0,0%
Mining 37 421 458 8,08%
Construction 1 81 82 1,22%
Manufacturing 392 3,047 3,439 11,40%
Transp., Utility 60 421 481 12,47%
Communication 4 413 417 0,96%
Wholesale, Retail 22 589 611 3,60%
Insur., Real Est. 80 1,572 1,652 4,84%
Finance other 77 1,373 1,450 5,31%
Services 104 2,310 2,414 4,31%
Public Admin. 0 4 4 0,0%
Total 777 10,262 11,039
Event Windows CAR1 (-1,1) CAR5 (-5,1) CAR20 (-20,1) CAR1 (-1,1) 1.0000
CAR5 (-5,1) 0.5798*
0.0000 1.0000
CAR20 (-20,1) 0.4102*
0.0000 0.7190*
0.0000 1.0000
*significant at α=0.05
20 Table 4. Descriptive Statistics
Variable Observations Mean Std. Deviation Min. Max.
retired 11039 .0703868 .2558094 0 1
halfYr 11039 .0045294 .0671512 0 1
oneYr 11037 .0137719 .116548 0 1
twoYrs 11038 .0075195 .0863922 0 1
threeYrs 11038 .0075195 .0863922 0 1
fourYrs 11033 .0057101 .0753528 0 1
AqSIC 11036 5281.29 1923.9 139 9511
T_SIC 11023 5419.272 1972.614 112 9661
AqSIC2 11036 52.30845 19.1587 1 95
AqSIC3 11036 527.7572 192.3671 13 951
T_SIC2 11023 53.74535 19.6965 1 96
T_SIC3 11023 541.5943 197.2737 11 966
Value 11039 312.2239 2081.636 .5 67285.7
crossborder 11039 .8662016 .340451 0 1
private 11039 .5150829 .4997951 0 1
tobinsQ 8783 1.98946 5.788384 -51.41082 456.6024
T_EBITDAD 2356 86.83309 473.9361 -6320 7953.808
AqEBITDAD 9924 633.701 2620.799 -4440 61276
CAR1 11039 16.72074 591.305 -5226.482 10094.05 CAR5 11039 43.70282 1025.821 -11025.37 29457.44 CAR20 11039 145.1309 2644.931 -35994.3 50664.99
21 5.2HYPOTHESES ONE:ANTICIPATED CEOTURNOVER AND M&APERFORMANCE
Hypotheses 1 divides managers of the acquiring firm in two categories, based on whether they retired from the firm or not. The hypothesis suggests that outgoing managers make worse than average M&A deals in search of legacy and therefore negatively influence M&A performance. Managers that stay within the firm throughout the whole M&A process are on the other hand suggested to not engage in these value destructing last-minute deals. The hypothesis is tested by looking at the coefficient of the two categories in an OLS regression.
The independent was first considered in a univariate context, using a selection of event windows as shown in Table 5. Model 3 provides evidence in support of Hypothesis 1, and provides the best fit. Retired managers show a negative significant relationship (at α=0,05) with M&A performance. The explained variance in the performance of the model is significant, yet small. The coefficients show a very strong negative direction, indicating that for one retired manager, M&A performance decreases by -78.42 points in the univariate model.
Table 5. Test of Hypothesizes 1, Univariate Model
Variable Model 1
Event Window (-1,1)
Model 2 Event Window (-20,1)
Model 3 Event Window (-5,1)
retired -14.82
(19.54) -129.4
(97.59) -78.42**
(32.48)
N(Obs) 11,039 11,039 11,039
R² 0.000 0.000 0.000
F 0.58 0.1850 5.83
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Table 6 replicates Model 3, and adds a multivariate specification. The increase in R-squared shows that the multivariate model increases the amount of variance performance explained. In Model 4 Retired managers show a negative significant relationship (at α=0,01) with M&A performance. Again, the coefficient shows a very strong negative direction, where for one retired manager, M&A performance decreases by -110.2 points.
22 Table 6. Test of Hypothesizes 1, Multivariate Model
These results give proof to the suggestion that outgoing managers negatively affect M&A performance, and as the coefficients show, this negative effect on performance is severe.
Based on the results of Model 3 and Model 4 Hypotheses1, on the negative effect of retiring managers on M&A performance is supported.
Variable Model 3
Event Window (-5,1)
Model 4 Event Window (-5,1)
retired -78.42**
(32.48) -110.2***
(36.08)
AqSIC 12.77***
(4.434)
crossborder 20.76
(29.28)
T_SIC3 -0.0844
(0.0706)
AqSIC3 -127.7***
(44.33)
private -45.54**
(22.29)
TobinsQ -0.430
(0.440)
N(Obs) 11,039 8,719
R² 0.000 0.002
F 5.83 2.66
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
23 5.3HYPOTHESES TWO AND THREE:MODERATING EFFECT OF PAYMENT TYPE AND FIRM SIZE
In order to test the hypothesis on the role of payment type and firm size two interaction regressions were performed, where the acquirer’s EBITAD, or free cash flow, was used as a substitute for payment type, and deal value as an equivalent for firm size. Table 7 reports the results of these two models. Model5 shows the interaction effect of payment type and Model6 the interaction effect of firm size, both for retiring and non-retiring managers.
The results fail to support Hypothesis2 and Hypothesis3 on the moderating effect of payment type and firm size for retiring managers. Contrary to expectations, no evidence could be found to support the conclusion that financing through equity payments and large bidder firms negatively moderates the relationship of outgoing managers on M&A performance.
Table 7. Test of Hypothesizes 2 and 3, Interaction Model
Variable Model 5
Event Window (-5,1)
Model 6 Event Window (-5,1)
retired -107.8**
(40.04) -116.7***
(36.26)
AqSIC 14.92***
(4.652) 12.76***
(4.434)
crossborder 9.426 (30.94) 20.88
(29.29)
T_SIC3 -0.0976
(0.0764) -0.0844 (0.0706)
AqSIC3 -149.1***
(46.51) -127.6***
(44.32)
private -49.55**
(23.13) -46.12**
(22.51)
TobinsQ -1.304
(1.217) -0.426
(0.438)
N(Obs) 8,034 8,719
R² 0.003 0.003
F 2.51 2.34
Predictor
retired x AqEBITDA 0.000615 (0.00419)
retired x value 0.00751
(0.00830)
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
24 However, the opposite is true for all non-retiring managers. Table 8 (in Appendix) shows that the interaction of free cash flow with non-retiring managers is negative and statistically significant with a β of -0.00406 at α=0, 1 level. This suggests that managers which do not depart from the acquiring firm throughout the M&A process destroy shareholder value if they finance mergers through cash payments. This is in line with Jensen’s (1986) free cash flow hypothesis. Accordingly, there is an interaction effect of free cash flow on the relationship of non-retiring managers on M&A performance. Furthermore, the results indicate that the interaction of deal value with non-retiring managers is negative and statistically significant with a β of -0.00274 at a significance level of α=0, 05. This shows that for non-retiring managers the bigger the deal value (and thus the bigger the acquirers firm size) the poorer the M&A performance.
25
6.DISCUSSION
The results of this study aim at identifying what effect retiring managers of the bidder firm have on merger performance. This relationship was analyzed with an event study methodology, in which the cumulated abnormal return is used as a measure of performance. For this a sample of 11039 mergers with 777 retiring managers over a period of 23 years was used. In addition, two moderating effects were considered: the payment type the outgoing manager chooses in order to finance a merger or acquisition, and the size of the acquiring firm. Based on the results of the previous section, the following key findings were found:
1. Retiring managers make value destructing M&A deals
The results support the hypothesis that retiring managers destroy shareholder value through last-minute M&A deals. Performance was measured in three different event windows ([-1,1], [-5,1], [-20,1]) where the hypothesized effect could be proven for the CAR measured five trading days before through one trading day after the merger announcement.
Graph1 shows the average relationship between the cumulated abnormal return for the event window [-5, 1] and retired managers, where 95% of the firms fall within the shaded area. The graph shows that retired managers significantly decrease performance.
Graph 1. Average Relationship between Performance and Retired Managers
26 No significant results could be found for the other two event windows. Choosing for various event windows is a common practice in the event study research (Menzar, Night, Kwok, 1994, event window [-30,10]; Mahoney & Mahoney, 1993, event window [-50,10]; Turk, 1992 event window [-50,5]; Seth, 1990, event window [-40,5], Jacobson, 1994 event window [-10,1]). According to McWilliams & Siegel (1997), two problems are likely to occur with this practice, which are suggested to also explain the insignificant results of the other two event windows from this study. Firstly, using a long event window such as [-20,1]
significantly reduces the power of the test statistic as it increases the difficulty of controlling for confounding effects (McWilliams & Siegel, 1997). The majority of firms used for this study are large, diversified, multinational firms. Therefore, it is likely that significant events occur quite frequently which increases the likelihood for confounding effects, which can overshadow the actual event under investigation (McWilliams & Siegel, 1997). These cofounding effects are likely to explain the insignificant results for the [-20,1] event window of this study. In very small event windows [-1,1], on the other hand, there is a possibility that significant effects on the event are not captured due to information leakage (McWilliams & Siegel, 1997). As leakage of information about the mergers is likely in this study, according to McWilliams & Siegel (1997) the event window should include some time prior to the announcement of the event so that abnormal returns related with the leakage will be captured. An Event window should be long enough to capture the significant effect of the event, yet short enough to eliminate confounding effects which is probable to explain why only the middle Event window [-5,1] of this study shows significant results.
2. There is no moderating effect of Payment Type and Firm Size on the relationship between CEO turnover and performance
The results reject the hypothesized moderating effects of payment type and firm size on the relationship between outgoing managers and performance. In other words, the payment type a retiring manager chooses to finance his or her deals and the size of the firm he/she operates in do not significantly affect performance in any way. Also other moderating effects were tested which all show insignificant results for the retiring managers (see
27 Table5 in Appendix). This leads to suggest, that the effect of retiring managers on performance in itself is very robust.
3. Payment Type and Firm Size moderate the effect of non-retiring managers on performance
Findings support Jensen’s (1986) free cash flow hypotheses, which suggest that high levels of free cash flow increase managerial freedom, and therefore make it possible for managers to peruse self-serving acquisitions. As the results have shown also in this study a great amount of free cash flow negatively moderates a mangers impact on performance.
Furthermore the results support the notion “the bigger the deal, the poorer the merger performance” (Carline, Linn, & Yadav, 2002). In this study the deal value was used as a substitute for firm size. The results therefore also confirm with Moeller et al. (2004) who stress that M&A performance decreases with size, where managerial hubris play a greater role in the acquisition decision of large firms than of small. Overall, the results of this section confirm with existing management theory, suggesting that the data set is highly representative.
6.1.PRACTICAL IMPLICATIONS
In the context of CEO retirement the results reveal new insights for the acquirer firm, impacting managerial decision making in the turnover process. Based on these results the following practical implications are suggested:
1. Retiring managers should be prohibited to perform last-minute M&A deals
The results imply that managers anticipating retirement still conduct mergers or acquisitions in their final period in office and do particularly poorly in these acquisitions.
Measures should be taken in order to prohibit retiring managers from performing these last-minute deals.
28 2. Retiring managers should not have the majority say in naming a successor
If a company is experiencing “normal” turnover, the outgoing CEO is most likely to be involved in the succession and change process (Puffer & Weintrop, 1995). As has been stressed, CEOs anticipating retirement are on the one hand likely to choose a successor which will continue in the same manner as them in order to carry on their legacy, which will in turn negatively affect a firm’s future prospect to change. On the other hand, they might set their successors up for a failure to resemble their own indispensability. In either way, the outgoing CEO shouldn’t have a majority say in the appointment of a successor.
Other shareholders should lead this process in order to choose what is right for the firm’s future prospect.
6.2LIMITATIONS
The study is subject to some limitations, which should be taken into account. Firstly, for the US the results appear encouraging, however future research should investigate if the effect of retiring managers is also generalizable across other nations.
Secondly, the dependent variable of this study was performance, which was measured in the form of abnormal return, in a function of the share price reaction to an event. The measurement of stock price reactions to an event does however, not come without limitations. The abnormal return reflects the difference between the actual return, and the expected return of an individual stock, and therefore reflects expected and not actual performance. Nevertheless, previous studies relied on the use of stock price effects as a performance measure (Kothari & Warner, 2006). Common models for performance calculations are the Fama-French model, the market-adjusted model, or the capital asset pricing model. Therefore, also in this study stock price reactions are used as a measure for performance.
29
7.CONCLUSION
The purpose of this paper was to investigate the effect of CEO turnover on M&A performance. Based on the research question and a literature review on agency theory, merger performance, and CEO turnover, a number of hypothesizes were formulated and tested. Using an initial sample of 37,530 deals with 2,111 retiring managers over a period of 23 years, a number of conclusions can be drawn.
The study proves that retiring managers make last-minute value destructing M&A deals.
With this finding the study strongly contributes to the value destructing theories on mergers and acquisitions, especially agency theory. Thus far, agency theory has solely focused on managers which stayed with a firm. The findings of this study extend the theory by adding that also managers facing retirement perform value destructing last-minute deals. If this is due to self-interested motives by the retiring managers was not investigated.
It has however, been proposed that retiring managers pursue last-minute M&A deals not in an effort to maximise firm value, but in an effort to leave a legacy, and/or potentially setting up their successor for failure as a proof of their own indispensability. If there is ground for such a Legacy Theory or what other motives could drive retiring mangers to pursue last- minute M&A deals should be investigated in future research.
Next to agency theory this paper furthermore contributes to managerial turnover theory.
Whereas prior research merely focused on post M&A performance and its impact on CEO turnover (Lehn & Zao, 2006), this study complements by adding the reverse relationship, namely the impact of CEO turnover on M&A performance.
The research question of this paper, do retiring managers of the bidder firm make worse than average M&A deals?, can thus be answered with a yes. It could be proven that retiring managers make value destructing M&A deals. As this study did not find significance for moderating effects on the relationship between retiring managers and performance, future research should lay an emphasis on finding other possible moderating effects on this relationship.
