Antecedents of organizational ambidexterity:
“The influence of the distribution of CEO and top
management team compensation on organizational
learning strategies”
by NATHALIE STOKREEF Master's Thesis MSc Business Administration Change Management University of Groningen Faculty of Economics and BusinessFebruary 4th 2019
Student number 2775689
Supervisor: prof. dr. J.D.R. Oehmichen Co-assessor: A.A. Oleksiak
2
Abstract
As organizational ambidexterity, the simultaneous pursuit of exploitation and exploration, is perceived to be essential for an organization’s sustainable competitive advantage, the goal of this study was to explain the influence of CEO and top management team compensation on the different organizational learning strategies. Furthermore, the influence of a compensation gap was challenged. This study contributed to current organizational management theory as it (1) takes into account both the behavioral and economic perspective, (2) investigates two organizational levels and (3) tries to compare the findings between the different learning strategies. OLS-regressions were performed to statistically test our hypotheses. As the analyses show rather negative findings, this confirms the tension and complementarity between the behavioral and economic perspective and the required attention they need in both the field of compensation and organizational learning and innovation. Moreover, this research resulted in a better understanding of the complex concept of organizational ambidexterity by looking at the influence of a compensation gap. As managers strive for overall organizational success, they need to find a proper distribution of compensation and strategy to foster the desired employee motivation and effort.
3
Introduction
Organizations that succeed in renewing themselves through the simultaneous use of exploitation and exploration, which is known as organizational ambidexterity (Oehmichen, Heyden, Georgakakis & Volberda, 2017; Tushman & O’Reilly, 1996), are argued to be superior performers (Mudambi & Swift, 2014; Piao, 2010). In order to enhance performance and ambidexterity, an organization’s strategy and compensation system need to be aligned (Carpenter & Sanders, 2002; Hambrick & Mason, 1984). Certo, Lester, Dalton and Dalton (2006) argue that the top management team is an essential party in influencing an organization’s ability to combine exploitation and exploration. Moreover, CEOs have the most power in deciding on the strategy and structure of an organization (Hambrick, 1989). Therefore, this study is focused around the design of CEO and top management team compensation. Furthermore, the influence of a compensation gap between CEOs and the top management team on organizational learning strategies is challenged, taking into account both the behavioral and economic perspective on compensation incentives. The goal of this study is to explain the influence of CEO and top management team compensation on the different organizational learning strategies.
As competition in today’s businesses increases, organizations feel the pressure to focus on organizational learning strategies to sustain a competitive advantage (Floyd & Lane, 2000; March, 1991; Swift, 2016). This has led to extensive research on how to effectively combine and balance both exploitation and exploration (Tushman & O’Reilly, 1996), which have evolved into two directions. Whereas multiple researchers started to explore the antecedents of organizational ambidexterity, others were seeking to find a relationship between organizational ambidexterity and performance. Evidence was found for antecedents of organizational ambidexterity on various organizational levels, such as the connectedness of business units within the organization (Jansen, Van den Bosch & Volberda, 2006), functional background heterogeneity of the board of directors (Heyden, Oehmichen, Nichting & Volberda, 2015) or leadership characteristics (Beckman, 2006). Several other researchers followed Tushman and O’Reilly (1996), among whom Gibson and Birkinshaw (2004), who found that ambidexterity enhances sustainable performance and is required for long-term survival.
4
body of literature exists on organizational structures, such as mechanic or organic, that suit best with an organization’s goals. Hence, it can be argued that it is a major challenge for organizations to adapt to an alignment of two structures that appear on the other end of the same continuum (March, 1991; O’Reilly & Tushman, 2013; Swift, 2016) and compete for organizational resources (Simsek, Heavey, Veiga & Souder, 2009).
It is not surprising that this topic received a lot of attention in organizational literature, as the potential benefits of ambidextrous organizations are enormous with regard to performance (Raisch & Birkinshaw, 2008). O’Reilly, Main and Crystal (1988) already started to investigate what influences both CEOs and the top management team to pursue organizational goals. It was found that compensation of both CEOs and the top management team have a positive influence on organizational ambidexterity (Carpenter & Sanders, 2002; Hambrick & Mason, 1984). The alignment of CEO and top management team compensation enhances behavioral integration (Hambrick, 1995). This enables organizations to quickly adapt in competitive environments (O’Reilly & Tushman, 2013). However, compensation gaps can function as an incentive for individual performance as well (Bloom, 1999). As a result, compensation systems became a widely studied subject in strategic management literature (Finkelstein & Hambrick, 1989; Gomez-Mejia, Tosi & Hinkin, 1987). These studies focus in particular on CEOs and top management teams, as these higher organizational levels have the most influence on organizational strategies and performance (Certo et al., 2006, Finkelstein & Hambrick, 1989). Research found that there are two management perspectives with different implications on the influence of compensation gaps on potential organizational performance (Carpenter & Sanders, 2002). The behavioral and economic perspective are seen as complementary in explaining the relationship between compensation and organizational performance in this study. As both perspectives are based on different theories, e.g. agency and equity theory, this has implications regarding their view on the influence of compensation gaps (Gopalan, Milbourn, Song & Thakor, 2014; Henderson & Frederickson, 2001). According to agency theory, large compensation gaps create incentives for individual performance that solve monitoring problems (Henderson & Frederickson, 2001). However, based on equity theory large compensation gaps also cause feelings of deprivation among lower level employees which negatively influences commitment (Cowherd & Levine, 1992).
5
common. As compensation gaps are mostly present at the upper organizational levels, they play a major role in organizational ambidexterity research (Pitcher & Smith, 2001). However, research focused mostly on one organizational level and neglected the role of the top management team (Carpenter & Sanders, 2002; Hambrick, 1989). Second, a better understand of a compensation gap is found by taking into account both the behavioral and economic perspective. As both perspectives are based on different theories, e.g. agency and equity theory, this has implications regarding their view on the relationship between compensation gaps and organizational performance (Gopalan et al., 2014; Henderson & Frederickson, 2001). The importance of these implications and tensions is well known, but there is little agreement or studies available on how and why (Gomez-Mejia et al., 1987; Main, O’Reilly & Wade, 1993). Both theories are needed to explain and comprehend organizational outcomes (Henderson & Frederickson, 2001). Third, there will be strived for a better understanding of the influence of compensation gaps in organizational learning and innovation literature by considering exploitation, exploration and organizational ambidexterity separately. This way, the influence of compensation and gaps on the different learning strategies can be compared. According to Lazear and Rosen (1981) and Bloom (1999), the most favorable distribution has not yet been found. As compensation systems have an influence on strategic decision-making, different results are expected, which enhances knowledge on the complexity of the concept of organizational ambidexterity (He & Wong, 2004; March, 1991).
As mentioned, the goal of this study is to explain the influence of CEO and top management team (TMT) compensation on the different organizational learning strategies. This study has important implications for managers, as a proper distribution of compensation packages is the base for future organizational profits (Bloom, 1999). As managers want to strive for overall organizational success, a proper distribution is needed that fosters the desired employee motivation and effort (O’Reilly et al. 1988). Therefore, managers need to take this into account, also in combination with the organizational learning strategies the specific organization focuses on itself.
To summarize, the following question will be addressed in this research: “How does CEO and TMT compensation affect organizational learning strategies and what influence does a compensation gap between the CEO and TMT have on this relationship?”.
6
described, followed by a report of the results on the influence of CEO and TMT compensation on organizational learning strategies. Finally, conclusions drawn from this research accompanied by possible limitations and options for future research can be found in the last section.
Theoretical Background
In this literature review, first of all the tensions of both the behavioral and economic perspective and the exploitation and exploration strategies regarding organizational ambidexterity are reflected upon. This is followed by a theoretical argumentation of the expected relationship between CEO compensation, TMT compensation and lastly the compensation gap on organizational learning strategies. Figure 1 represents the conceptual model of this study.
Figure 1. Conceptual model
Organizational Ambidexterity: The Tension between Exploitation and Exploration
Although exploitation and exploration strategies compete for organizational resources (Simsek et al., 2009), they are both essential to ensure short and long term success of organizations (March, 1991). An optimal balance is perceived to be crucial for survival of an organization (Gibson & Birkinshaw, 2004) and is called organizational ambidexterity (O’Reilly & Tushman, 2004). Organizational ambidexterity was found to be particularly important in management research (Gibson & Birkinshaw, 2004; Tushman & O’Reilly, 2004). However, after receiving a lot of attention, results on the effects of organizational ambidexterity on performance are still mixed (Cao, Gedajlovic & Zhan, 2009).
7
The payoff structure is less certain and more long term than with exploitation (Gibson & Birkinshaw, 2004). Moreover, exploitation contributes to profitability whereas exploration enhances organizational growth (Junni, Sarala, Taras & Tarba, 2013). Therefore, it is not surprising that the influence of both strategies on organizational performance differs (He & Wong, 2004) and Benner and Tushman (2003) were one of many that questioned the possibility of the simultaneous pursuit of both strategies. According to March (1991), exploitation and exploration need to be seen as contradictory. However, the duality of organizational structures and strategies is emphasized (Benner & Tushman, 2003), as research found that organizations that focus on productivity are less flexible to innovate and thus compete over time. Because of the focus on stable routines, anything other than incremental change is not common for those organizations anymore (Hannan & Freeman, 1984). So, it is important to develop dynamic capabilities to be able to both exploit and explore.
Lavie, Stettner and Tushman (2010) argued that looking at both learning strategies separately ignores the possible trade-offs and argued for multiple forms of the joint pursuit of both strategies. However, the actual positive performance influence of the evident challenge of organizational ambidexterity still lacks empirical evidence (He & Wong, 2004; Junni et al., 2013; Swift, 2016). This implies that there is still a lot of ambiguity regarding the interaction of both learning strategies and its influence on organizational performance (Swift, 2016). As organizational ambidexterity is seen as a multilevel phenomenon, meaning it has influence on multiple organizational levels (Raisch & Birkinshaw, 2008), interest in looking at both the CEO and TMT level is increased (Junni et al., 2013). Consequently, this study takes into account all three learning strategies when looking at the influence of CEO and TMT compensation to be able to compare the results.
CEO Compensation and Organizational Learning Strategies
Until now it has been argued that this study looks from the behavioral and economic perspective towards the influence of compensation on the different organizational learning strategies. This paragraph will further explain what other scholars have found regarding the influence of CEO compensation on organizational performance and with regard to the different learning strategies of this study. Moreover, the choice for looking at the concept of total CEO compensation, instead of focusing on short- or long-term types of compensation, is being discussed.
8
principals that want their agents, the CEOs, to perform well. When organizations grow larger, it is hard for shareholders to ensure oversight. This may result in agents putting less effort into organizational outcomes and start free riding (Henderson & Frederickson, 2001). Hence, to ensure mutual interests of CEOs and shareholders without the need for extensive monitoring, which is seen as unworkable (Lazear & Rosen, 1981), CEOs are compensated a major amount. This economic perspective has dominated research on CEO compensation (O’Reilly et al., 1988) and several critical notes on the large amounts of CEO compensation have been made (Jensen & Murphy, 1990). Taking the behavioral perspective, CEO compensation must reflect the responsibilities and complexity CEOs have to manage (Haleblian & Finkelstein, 1993). When organizations grow, coordination needs become greater to be able to process the increasing amount of information (Hambrick & Siegel, 2005). This also results in larger top management teams that need to be managed, which approves a higher CEO compensation reward as a means to motivate effort for organizational performance (Carpenter & Sanders, 2002). Moreover, social comparison theory is another explanation for the relationship between CEO compensation and organizational outcomes. As CEOs tend to compare their compensation with that of other CEOs, a perception of perceived inequity can result in lower job satisfaction and withdrawing effort (Wade, O’Reilly & Pollock, 2006). Individuals often compare themselves with others, which they view as slightly more qualified (Goodman, 1974), which increases desired compensation. As a result, justice and fairness of the compensation setting, positively influence effort for organizational outcomes (Carpenter & Sanders, 2002).
A lot of research found positive evidence for the relationship between CEO compensation and organizational outcomes (Carpenter & Sanders, 2002; Hambrick, 1995; Henderson & Frederickson, 1996; Tosi, Werner, Katz & Gomez-Mejia., 2000). As CEOs have a lot of authority in decision-making, CEO compensation directly enables and encourages CEOs to increase organizational performance (Jensen & Meckling, 1976). As this study focuses on different organizational learning strategies, more about this relationship will be explained with regard to exploitation and exploration specifically. To be able to compare the possible different relationship, the decision to take total CEO compensation will also be argued.
9
shareholders decide whether they focus on exploitation or innovation and apply the possible incentives accordingly (O’Reilly et al., 1988). As this study focuses on both learning strategies, as well as organizational ambidexterity, total CEO compensation enables comparability of the results. Second, incentives sometimes do not result in the desired behavioral outcomes (Hou, Priem & Goranova, 2015). According to Carpenter and Sanders (2002), this has to do with the organizational complexity a CEO has to manage. Some organizations have greater information-processing requirements than others, which complicates managerial work as CEOs need to make sure that the capabilities for communication and collaboration are well established (Haleblian & Finkelstein, 1993). This means that CEOs are not always able to control organizational outcomes, which weakens the relationship between a specific incentive and organizational learning strategies. Third, Hou et al. (2015) also found that there are more factors influencing a CEOs response to financial incentives. A CEO’s motivation and behavior are dependent on one’s beliefs and values, which are shaped by experiences (Hou et al., 2015). Moreover, Wowak and Hambrick (2010) also argue that individual characteristics of CEOs affect their choices and organizational outcomes. For example, risk-aversive CEOs would prefer salary and bonuses over stock options. They also already have a natural tendency towards exploitation (Kammerlander, Burger, Fust & Fueglistaller, 2015; Wowak & Hambrick, 2010). This contributes again to the possibility that incentives may not foster the desired organizational strategy and outcomes (Hou et al., 2015).
To conclude, the relationship between CEO compensation and organizational outcomes is not as straightforward as one might think. The intended strategic outcomes or desired CEO focus of financial incentives may be influenced by other factors (Wowak & Hambrick, 2010). For that reason, this study focuses on total CEO compensation with regard to the different organizational learning strategies. This results in the following hypothesis:
H1: Total CEO compensation has a positive influence on (a) exploitation, (b) exploration and (c) organizational ambidexterity.
TMT Compensation and Organizational Learning Strategies
10
The top management team of an organization is responsible for the actual execution of the set strategy and by that, contribute to organizational outcomes as well (Hambrick & Mason, 1984). It is emphasized that the top management team has a significant influence on information processing capabilities required for organizational performance (Finkelstein & Hambrick, 1989). Moreover, this team has a lot of resources that are important for organizations. To be able to utilize these resources, the top management team needs to be motivated to improve organizational outcomes (Certo et al., 2006). Also, looking from social comparison theory, TMT compensation can have an influence on organizational performance (Wade et al., 2006). When individuals are making these social comparisons and perceive inequity, this has implications on their perceptions and behavior (Wade et al., 2006). Lack of fairness may result in lower job satisfaction, sometimes accompanied by less effort and lower product quality (Wade et al., 2006). These consequences lead to less organizational performance at the bottom-line (Carpenter & Sanders, 2002). Looking at these arguments, they are very similar to those of CEO compensation. This is not surprising, as Henderson and Frederickson (1996) already suggested that organizations are not only managed by the effort of one single executive, but rather the top management team. Moreover, in complex organizations, decision-making responsibilities are also given to lower level executives. This contributes to social integration (Smith, Smith, Olian, Sims, Bonnan & Scully, 1994) and decision-making speed (Hambrick & Mason, 1984), which explains the particular relationship found between top management teams and organizational performance (Carpenter & Sanders, 2002).
11
may differ, the total compensation of the top management team is used to compare with that of CEOs. Therefore, both learning strategies are taken into account separately, as well as combined, to compare the effect of total TMT compensation with CEO compensation. The following hypothesis will be tested:
H2: Total TMT compensation has a positive influence on (a) exploitation, (b) exploration and (c) organizational ambidexterity.
The Moderating Role of the CEO – TMT Compensation Gap
Until now it has been argued that CEO and TMT compensation have a positive influence on organizational learning strategies. However, the question arises why organizations do not engage in paying the CEO and top management team a same amount. This paragraph will explain what influence a compensation gap between the CEO and the top management team may have on the relationship between compensation of both organizational levels and organizational learning strategies. This study views a compensation gap as the difference between CEO compensation and average TMT compensation (Bloom, 1999) and thus considers one hierarchical level. It is argued that a large compensation gap has a positive influence on the relationship between CEO compensation and organizational learning strategies, but is perceived negatively for the top management team.
12
From a behavioral perspective, a large compensation gap does not contribute to the desired wholeness and unity of a top management team (Carpenter & Sanders, 2002). This emphasizes the importance of alignment between CEOs and the top management team. A large compensation gap namely has a negative influence on an individual’s feeling of appreciation (Wade et al., 2006). This means that members of the top management team may feel inequity with regard to their compensation and that of their CEO when there is a large gap (Cowherd & Levine, 1992; Kim and Mauborgne, 1996). The feeling of procedural justice has influence on the top management team’s level of collaboration and behavioral integration (Carpenter & Sanders, 2002; Hambrick, 1995), as they become less committed to organizational goals (Henderson & Frederickson, 2001). This behavioral integration is needed for organizations in competitive environments because it enables them to rapidly change (Carpenter & Sanders, 2002). This is something that agency theory ignores, as it focuses only on cooperative action instead of complete team effectiveness (Hambrick, 1995). All in all, bottom-line performance is not ensured with large compensation gaps, as this results in a divided and not integrated top management team, which limits collective effectiveness and use of valuable resources (Carpenter & Sanders, 2002; Jensen & Meckling, 1976). Compensation alignment between CEOs and top management teams was found to be a determinant for unity of effort and dynamic capabilities (Benner & Tushman, 2003). Integrated teams are a valuable and rare resource for information-processing capabilities of organizations (Hambrick, 1995; Henderson & Frederickson, 1996). Carpenter and Sanders (2002) explain the importance of integrated top management teams as well and argue that this is essential to be able to adapt to changing competitive environments and to engage in both exploitation and exploration (Benner & Tushman, 2003).
13
To conclude, a large compensation gap has a positive influence on the relationship between CEO compensation and organizational learning strategies, as it enhances individual effort by CEOs. Moreover, due to social comparison theory, CEOs need to be compensated according to a general amount other CEOs receive as well, to be properly motivated. However, a large compensation gap has a negative influence on the feeling of appreciation of the top management team when they compare themselves with the CEO of their organization. This leads to a less productive workforce and a decrease in willingness to collaborate. This results in the following hypotheses:
H3: The compensation gap between the CEO and TMT has a positive influence on the relationship between CEO compensation and organizational learning strategies ((a) exploitation, (b) exploration, (c) organizational ambidexterity).
H4: The compensation gap between the CEO and TMT has a negative influence on the relationship between TMT compensation and organizational learning strategies ((a) exploitation, (b) exploration, (c) organizational ambidexterity).
Methods
Research Plan
This study focuses on a theory testing approach, as there is already a substantial amount of literature available but there is still a gap that needs to be filled. This study will aim to fill this gap. A quantitative approach will be used to test the hypotheses, through a statistical analysis of a sample of secondary data using the statistical program STATA 15. The sample was realized by merging two datasets on financial and organizational ambidexterity data.
Data Collection
14
data, which means that multiple variables are measured for multiple firms over multiple years. This enhances the generalizability (McKenny et al., 2018). Moreover, recent research encouraged studies on organizational ambidexterity to use longitudinal data (Certo et al., 2006; Henderson & Frederickson, 1996; Swift, 2016).
Measurements
To measure the concepts used in this research, they need to be operationalized. This was done according to existing theory.
15
to calculate the third variable. This means that for organizational ambidexterity, both exploitation and exploration need to be at a high level (Simsek et al., 2009). Therefore, to measure organizational ambidexterity, first both the exploitation and exploration measures were adjusted by the length of the text analyzed by dividing the total word count of the text by the word count of either exploitation or exploration. Organizational ambidexterity was then measured by the multiplication of adjusted exploitation times adjusted exploration.
Independent variables. CEO compensation and TMT compensation are used as independent variables in this study. This study consciously chose for total compensation, which includes both short- and long-term compensation. Many studies emphasize the importance of either short- or long- term incentives for CEOs to either foster exploitation or exploration. However, the research of Balkin, Markman and Gomez-Meija (2000) explains that it is essential for CEOs to be rewarded both short- and term to be motivated to completely manage new product innovations. As the payoff of long-term innovations is often uncertain, CEOs need to feel responsible and able to influence their compensation. When CEOs are rewarded on both the short- and long-term of innovation, this is expected to have a positive effect on the financial performance of organizations (Balkin et al., 2000). Moreover, because an extended stock option vesting period positively influences the stay of employees at organizations, the value of option grants was preferred over the value of options exercised as a measure for the long-term compensation component of total compensation (Yanadori & Marler, 2006). Based on this, the following measurement of total CEO compensation, acquired from the database of Compustat ExecuComp was used: (Salary + Bonus + Other Annual + Restricted Stock Grants + LTIP (long term incentive plan) Payouts + All Other + Value of Option Grants).
16
Moderator. The compensation gap between the CEO and the top management team is derived from both independent variables. This variable was calculated as follows: (1 – (Average TMT total compensation / CEO total compensation)). This results in a variable with values ranging from 0 to 1, where 0 equals no compensation gap and 1 means absolute inequality. Research of Bloom (1999) indicated this as a possibility to measure compensation differences between individuals. By calculating the moderator as a ratio, it is comparable across different organizational sizes. This is valuable, as executives of larger organizations often receive a higher compensation due to the complexity of managing the organization (Henderson & Frederickson, 2001), so the monetary gap would be hard to interpret.
Control variables
This study controls for several other variables that may have an influence on the relationship between CEO and TMT compensation and organizational learning strategies.
Population ecologists believe that organizational performance is influenced by the external environment (Hannan & Freeman, 1977). Therefore, this study takes into account industry effects for several reasons. First of all, commitment and engagement from individuals is especially essential for organizational performance in dynamic environments (Junni et al., 2013). The information-processing capabilities in these dynamic environments are crucial to manage the complex work of CEOs (Certo et al., 2006). In more stable environments, dominant CEOs may also be able to minimize communication and conflict which they view is not necessary for the process. Hence, industry effects can have influence on the compensation gap between CEOs and top management teams, as their compensation packages may differ due to the higher responsibilities of CEOs or top management teams (Henderson & Frederickson, 2001). Second, Wang and Li (2008) have argued that the actual rewards of exploitation or exploration may also differ from the expected rewards, due to competition in the industry. Lastly, exploitation and exploration influence organizational performance mainly in manufacturing and service industries (Junni et al., 2013; Kammerlander et al., 2015), whereas exploitation is dominant in manufacturing and exploration in service industries. Therefore, an industry dummy variable is included which separates all organizational data in either manufacturing or service industry.
17
performance and thus the compensation gap between CEOs and the top management team (Henderson & Frederickson, 2001). Moreover, organizational size influences organizational growth and this could be related to the availability of more resources (Jansen et al., 2006). This means that organizational size might be correlated with exploration and thus organizational ambidexterity as well.
Besides industry effects and organizational size, Research & Development (R&D) investments is also considered as an important control variable. Exploitation and exploration are only possible when an organization invests in R&D. Swift (2016) argues that both learning strategies are R&D-based and that more R&D expenditure is correlated with a shift from exploitation to exploration. R&D departments always need to consider which resources to assign to either invest in exploitation or exploration. In general, it is expected that more R&D expenditure results in a more exploratory focus of the organization. However, in high-technology industries, a lot of variance in R&D investments still exists, although one might expect this to be high as these organizations focus on innovation (Yanadori & Marler, 2006). Lastly, more R&D investments is an indicator of more and large innovation projects, which increases importance of team effort in coordinating this. Therefore, this study controls for R&D investments.
As learning strategies and organizational ambidexterity stand in relation with organizational performance, two common control variables in organizational performance literature are included as well. These are return on assets (ROA) and Tobin’s Q (Finkelstein & Hambrick, 1989). These are indicators of financial performance and have also been used in other research on organizational ambidexterity as a means to check for external validity (Gibson & Birkinshaw, 2004). A high ROA means that an organization was able to earn a lot from its assets and resources. An organization’s strategic orientation, on either exploitation or exploration, may change based on its financial performance (Heyden et al., 2015). ROA was provided in the organizational ambidexterity data as a financial performance variable. Tobin’s Q is used as a control variable as well, as it captures both short- and long-term performance (Uotila et al., 2009), which is also the case in this study because of the use of total compensation and different learning strategies. Tobin’s Q was measured by McKenny et al. (2018) by the available financial variables in the organizational ambidexterity data as follows: (market capitalization + total assets – total shareholder’s equity) / total assets. Lastly, this study controls for unobserved year effects through the use of year dummies (Yanadori & Marler, 2006).
Analysis Plan
18
Analysis and Results
This section first looks at the descriptive statistics and correlations of the variables used. Moreover, the hypotheses will be addressed by discussing the results of the regressions.
Descriptive Statistics and Correlation Matrix
Table 1 shows the descriptive statistics of the variables that were analyzed for this study. To get a comprehensive understanding of the consequences of winsorizing and taking the natural logarithm of a variable, all descriptives are included in the table. Winsorizing a variable was used to take care of outliers and taking the natural logarithm compensates for skewness and reduces heteroscedasticity (He & Wong, 2004; Henderson & Frederickson, 2001). This resulted in taking the natural log of all three dependent variables. Moreover, the exploration variable was winsorized beforehand as well, to take care of outliers. Furthermore, both compensation variables were winsorized and afterwards logged for the same reasons. This is common for compensation variables (Carpenter & Sanders, 2002). To continue, the moderator has also been winsorized and logged. Lastly, besides the year and industry dummies, all control variables were either logged or winsorized and logged as well.
19
Table 1. Descriptive statistics
Measure obs mean s.d. min max
1 Organizational ambidexterity (OA) 2.433 1.05 1.18 0 18.06
2 OA (w) 2.433 1.04 1.08 0 11.41 3 OA (log) 2.360 -.35 .98 -3.91 2.89 4 OA (w + log) 2.364 -.36 .98 -3.91 2.43 5 Exploitation 2.433 10.89 9.73 0 82 6 Exploitation (w) 2.433 10.85 9.52 0 71 7 Exploitation (log) 2.386 2.07 .86 0 4.41 8 Exploitation (w + log) 2.390 2.07 .86 0 4.26 9 Exploration 2.433 12.10 11.13 0 101 10 Exploration (w) 2.433 11.97 10.39 0 77 11 Exploration (log) 2.407 2.19 .80 0 4.6 12 Exploration (w + log) 2.410 2.19 .80 0 4.34 13 CEO compensation 2.521 6407.19 12668.54 0 293097.3 14 CEO compensation (w) 2.521 5910.49 7891.71 19.09 47462.53
15 CEO compensation (log) 2.517 7.91 1.72 -6.91 12.59
16 CEO compensation (w + log) 2.521 7.96 1.33 2.95 10.77
17 TMT compensation 2.583 2476.04 4164.43 26.96 98716.21 18 TMT compensation (w) 2.583 2341.38 2859.92 221.64 18843.91 19 TMT compensation (log) 2.583 7.29 .97 3.29 11.5 20 TMT compensation (w + log) 2.583 7.28 .94 5.40 9.85 21 Compensation gap 2.506 -74034.75 1714025 -.00 .98 22 Compensation gap (w) 2.506 -.47 6.56 -57.30 .93
23 Compensation gap (log) 2.506 -74034.75 1714025 -.00 .98
24 Compensation gap (w + log) 2.208 -.70 .64 -7.94 -.07
25 R&D investments 2.433 9.47 4.28 0 16.24
26 R&D investments (w) 2.433 9.47 4.28 0 15.59
27 R&D investments (log) 2.079 2.39 .17 1.44 2.79
28 R&D investments (w + log) 2.079 2.39 .17 1.44 2.79
29 Organizational size 2.433 7.77 1.69 .69 12.98
30 Organizational size (w) 2.433 7.78 1.64 4.25 11.83
31 Organizational size (log) 2.433 2.03 .23 -.37 2.56
32 Organizational size (w + log) 2.433 2.03 .21 1.45 2.47
33 ROA 2.408 4.15 19.27 -173.62 157.88 34 ROA (w) 2.408 4.24 15.83 -68.93 45 35 ROA (log) 1.868 2.02 .90 -2.81 5.06 36 ROA (w + log) 1.868 2.02 .89 -2.81 3.81 37 TobinsQ 2.433 2.92 2.05 .22 21.83 38 TobinsQ (w) 2.433 2.90 1.94 .75 10.73 39 TobinsQ (log) 2.433 .89 .59 -1.51 3.08 40 TobinsQ (w + log) 2.433 .89 .58 -.29 2.37 41 Year 2.646 2007.541 4.54 2000 2016 42 Industry effects 2.433 .71 .45 0 1
20
Table 2. Correlations between measures
Measure 3 7 12 16 20 24 27 32 36 40 41 42 3 OA (log) 1.00 7 Exploitation (log) .64*** 1.00 12 Exploration (w + log) .57*** .31*** 1.00
16 CEO compensation (w + log) .04* .11*** .25*** 1.00
20 TMT compensation (w + log) .05* .09*** .32*** .67*** 1.00
24 Compensation gap (w + log)
-.03 -.00 .07** .45*** .15*** 1.00 .
27 R&D investments (log) .11*** .13*** .42*** .53*** .74*** .20*** 1.00
32 Organizational size (w + log) .19*** .26*** .30*** .46*** .62*** .20*** .78*** 1.00
36 ROA (w + log) .06** .08*** .03 .01 .05* .02 .02 .01 1.00
40 TobinsQ (w + log) .08*** -.02 .03 .09*** .18*** .00 .07** -.03 .43*** 1.00
41 Year -.20*** -.03 -.01 .07*** .09*** -.01 .03 .04* -.01 .01 1.00
42 Industry effects -.03 .07** -.11*** -.17*** -.09*** -.08*** -.13*** .01 -.13*** -.03 -.00 1.00
21 Hypothesis Tests
Table 3 shows the ordinary least squares regression results with exploitation as the dependent variable including all control variables, both independent variables and the interaction terms. The same holds for table 4 and 5, whereas the former uses exploration and the latter organizational ambidexterity as the dependent variable. Moreover, to look whether multicollinearity was present, the variance inflation factors (VIF) of each regression were determined. The maximum value for the VIFs was 4.13. This is lower than 10, which is universally regarded as the threshold after which multicollinearity must be dealt with (O’Brien, 2007).
CEO compensation. Hypotheses 1a, 1b and 1c address the direct relationship between CEO compensation and (a) exploitation, (b) exploration and (c) organizational ambidexterity.
There was not found a significant relationship between CEO compensation and exploitation (β = .020, p > .05) or exploration (β = .013, p > .05). This means that hypotheses 1a and 1b are rejected. CEO compensation does have a significant influence on organizational ambidexterity, but rather negative (β = -.052, p < .05) according to our model (F(7, 1517) = 38.11, p < .001, adjusted R2 = .146). This implicates that more CEO compensation, results in less organizational ambidexterity. This research expected this relationship to be positive as, according to an economic perspective, more compensation elicits more individual effort towards organizational goals (Carpenter & Sanders, 2002). This results in the rejection of hypothesis 1c.
TMT compensation. Hypotheses 2a, 2b and 2c address the direct relationship between TMT compensation and (a) exploitation, (b) exploration and (c) organizational ambidexterity.
22 Table 3. Results of the OLS-regression for exploitation (log)
Exploitation (log)
(1) (2) (3) (4) (5)
b (s.e.) b (s.e.) b (s.e.) b (s.e.) b (s.e.)
Intercept -10.271 (9.754) -10.147 (9.814) -14.11 (9.790) -16.951 (10.294) -19.00 (10.31)†
Independent variables
CEO compensation (w + log) .020 (.019) -.099 (.034)**
TMT compensation (w + log) -.148 (.035)*** -.142 (.039)***
Interaction effects
CEO compensation (w + log) * Compensation gap (w + log)
-.003 (.005)
TMT compensation (w + log) * Compensation gap (w + log)
-.012 (.005)*
Control variables
R&D investments (log) -.890 (.214)*** -.961 (.222)*** -.417 (.241)† -.321 (.256) -.207 (.259)
Organizational size (w + log) 1.642 (.181)*** 1.640 (.183)*** 1.782 (.184)*** 1.821 (.191)*** 1.847 (.191)***
ROA (w + log) -.084 (.026)** -.088 (.027)** -.088 (.026)** -.077 (.027)** -.078 (.028)**
TobinsQ (w + log) -.048 (.046) -.044 (.047) -.010 (.047) .007 (.050) .017 (.050)
Year dummies Yes Yes Yes Yes Yes
Industry dummy Yes Yes Yes Yes Yes
F-statistic 23.27 *** 20.25 *** 22.50 *** 21.31 *** 21.97 ***
R2 .082 .085 .093 .111 .114
Adjusted R2 .079 .081 .089 .106 .109
N observations 1.552 1.541 1.546 1.378 1.378
Note. This table reports the result for 5 ordinary least squares (OLS) regressions using ‘exploitation’ as dependent variable. Regression (1) is the dependent variable with the
23 Table 4. Results of the OLS-regression for exploration (w + log)
Exploration (w + log)
(1) (2) (3) (4) (5)
b (s.e.) b (s.e.) b (s.e.) b (s.e.) b (s.e.)
Intercept 19.929 (8.23)* 20.169 (8.287)* 20.648 (8.307)* 14.055 (8.803) 14.908 (8.822)†
Independent variables
CEO compensation (w + log) .013 (.016) .051 (.029)†
TMT compensation (w + log) .027 (.030) .071 (.034)*
Interaction effects
CEO compensation (w + log) * Compensation gap (w + log)
-.006 (.005)
TMT compensation (w + log) * Compensation gap (w + log)
-.001 (.005)
Control variables
R&D investments (log) 1.969 (.183)*** 1.92 (.190)*** 1.882 (.207)*** 1.879 (.224)*** 1.830 (.226)***
Organizational size (w + log) .037 (.152) .033 (.154) .013 (.155) -.023 (.163) -.035 (.163)
ROA (w + log) .015 (.022) .009 (.022) .016 (.023) .008 (.024) .009 (.024)
TobinsQ (w + log) -.129 (.039)** -.124 (.040)** -.138 (.040)** -.124 (.043)** -.129 (.043)**
Year dummies Yes Yes Yes Yes Yes
Industry dummy Yes Yes Yes Yes Yes
F-statistic 71.09 *** 60.36 *** 60.72 *** 50.05 *** 50.29 ***
R2 .215 .214 .215 .224 .225
Adjusted R2 .212 .211 .211 .220 .221
N observations 1.569 1.558 1.563 1.394 1.394
Note. This table reports the result for 5 ordinary least squares (OLS) regressions using ‘exploration’ as dependent variable. Regression (1) is the dependent variable with the
24 Table 5. Results of the OLS-regression for organizational ambidexterity (log)
Organizational ambidexterity (log)
(1) (2) (3) (4) (5)
b (s.e.) b (s.e.) b (s.e.) b (s.e.) b (s.e.)
Intercept 96.085 (10.609)*** 93.964 (10.643)*** 88.980 (10.571)*** 82.479 (11.118)*** 79.853 (11.140)***
Independent variables
CEO compensation (w + log) -.052 (.021)* -.180 (.037)***
TMT compensation (w + log) -.231 (.038)*** -.218 (.042)***
Interaction effects
CEO compensation (w + log) * Compensation gap (w + log)
-.003 (.006)
TMT compensation (w + log) * Compensation gap (w + log)
-.018 (.006)**
Control variables
R&D investments (log) -1.605 (.236)*** -1.459 (.244)*** -.866 (.262)** -.881 (.281)** -.818 (.284)**
Organizational size (w + log) 2.241 (.197)*** 2.323 (.199)*** 2.465 (.199)*** 2.569 (.207)*** 2.578 (.207)***
ROA (w + log) .005 (.029) -.011 (.029) -.002 (.028) -.007 (.030) -.009 (.030)
TobinsQ (w + log) .156 (.050)** .171 (.050)** .219 (.051)*** .259 (.054)*** .267 (.054)***
Year dummies Yes Yes Yes Yes Yes
Industry dummy Yes Yes Yes Yes Yes
F-statistic 43.09 *** 38.11 *** 42.95 *** 35.89 *** 36.31
R2 .145 .150 .165 .175 .176
Adjusted R2 .141 .146 .161 .170 .172
N observations 1.536 1.525 1.530 1.366 1.366
Note. This table reports the result for 5 ordinary least squares (OLS) regressions using ‘organizational ambidexterity as dependent variable. Regression (1) is the dependent
25
Compensation gap between CEO and TMT. Hypotheses 3a, 3b and 3c expect a positive influence of a compensation gap on the relationship between CEO compensation and (a) exploitation, (b) exploration and (c) organizational ambidexterity. Hypotheses 4a, 4b and 4c expect that this influence is negative when looking at TMT compensation and (a) exploitation, (b) exploration and (c) organizational ambidexterity.
The compensation gap did not yield any significant results regarding its moderating effect on the relationship between CEO compensation and exploitation (β = .003, p > .05), exploration (β = -.006, p > .05) or organizational ambidexterity (β = -.003, p > .05). This means that hypotheses 3a, 3b and 3c are rejected. However, it is important to mention that the moderator does change the direct relationship between CEO compensation and organizational learning strategies. For CEO compensation and exploitation for example (F (8, 1369) = 21.31, p < .001, adjusted R2 = .106), the relationship becomes significant (β = -.099, p < .01). The same holds for CEO compensation and organizational ambidexterity (β = -.180, p < .001), when the compensation gap was included (F (8, 1357) = 35.89, p < .001, adjusted R2 = .170). For CEO compensation and exploration, the presence of the compensation gap variable changes the direct relationship from non-significant to marginally significant, which means that it approaches significance (β = .051, p < .1) (Pritschet, Powell & Horne, 2016). This means that CEO compensation only has a significant influence on exploitation and organizational ambidexterity when a compensation gap is present. Note here that this relationship is rather negative, which could mean that the economic perspective does not explain this relationship.
26
rejection of hypothesis 4c as well. To conclude, the compensation gap had a significantly negative influence on the relationship between TMT compensation and exploitation and organizational ambidexterity. This is in line with our expectation that a compensation gap may negatively influence individual effort due to perceived inequality, coming from the behavioral perspective.
Lastly, the regression model with exploration as the dependent variable has an adjusted R-square between 21.1% and 22.5%. This means that this model explains at least one fifth of our data. Moreover, the R-square becomes slightly better in every model including the interaction of the compensation gap. In the regression model with exploitation as a dependent variable, this changes from 8.1% and 8.9% to 10.6% and 10.9%. This indicates that a compensation gap does have an explanatory function with regard to CEO and TMT compensation and organizational learning strategies.
Robustness Tests
27 Table 6. Results of the robust standard errors OLS-regression for exploitation (log)
Exploitation (log)
(1) (2) (3) (4) (5)
b (r.s.e.) b (r.s.e.) b (r.s.e.) b (r.s.e.) b (r.s.e.)
Intercept -10.272 (10.136) -10.147 (10.215) -14.11 (10.17) -16.951 (10.852) -19.00 (10.86)
Independent variables
CEO compensation (w + log) .020 (.0192) -.099 (.034)**
TMT compensation (w + log) -.148 (.036)*** -.142 (.040)***
Interaction effects
CEO compensation (w + log) * Compensation gap (w + log)
-.003 (.006)
TMT compensation (w + log) * Compensation gap (w + log)
-.012 (.005)*
Control variables
R&D investments (log) -.890 (.226)*** -.961 (.240)*** -.417 (.251)† -.321 (.273) -.207 (.275)
Organizational size (w + log) 1.642 (.194)*** 1.640 (.194)*** 1.782 (.195)*** 1.821 (.209)*** 1.847 (.209)***
ROA (w + log) -.085 (.030)** -.088 (.031)** -.088 (.030)** -.077 (.032)* -.079 (.032)*
TobinsQ (w + log) -.048 (.047) -.044 (.047) -.010 (.048) .007 (.050) .017 (.050)
Year dummies Yes Yes Yes Yes Yes
Industry dummy Yes Yes Yes Yes Yes
F-statistic 20.36 *** 17.60 *** 20.14 *** 18.47 *** 18.77 ***
R2 .083 .085 .093 .111 .114
N observations 1.552 1.541 1.546 1.378 1.378
Note. This table reports the result for 5 robust standard errors ordinary least squares (OLS) regressions using ‘exploitation’ as dependent variable. Regression (1) is the
28
Table 7. Results of the robust standard errors OLS-regression for exploration (w + log)
Exploration (w + log)
(1) (2) (3) (4) (5)
b (s.e.) b (s.e.) b (s.e.) b (s.e.) b (s.e.)
Intercept 19.929 (9.02)* 20.169 (9.102)* 20.65 (9.100)* 14.055 (9.441) 14.908 (9.486)
Independent variables
CEO compensation (w + log) .013 (.015) .051 (.028)†
TMT compensation (w + log) .027 (.029) .071 (.032)*
Interaction effects
CEO compensation (w + log) * Compensation gap (w + log)
-.006 (.005)
TMT compensation (w + log) * Compensation gap (w + log)
-.001 (.004)
Control variables
R&D investments (log) 1.969 (.176)*** 1.92 (.185)*** 1.882 (.196)*** 1.879 (.230)*** 1.830 (.221)***
Organizational size (w + log) .037 (.151) .033 (.151) .013 (.154) -.023 (.164) -.035 (.163)
ROA (w + log) .015 (.022) .009 (.022) .016 (.022) .008 (.023) .009 (.023)
TobinsQ (w + log) -.129 (.041)** -.124 (.041)** -.138 (.041)** -.124 (.043)** -.129 (.043)**
Year dummies Yes Yes Yes Yes Yes
Industry dummy Yes Yes Yes Yes Yes
F-statistic 71.18 *** 60.91 *** 61.31 *** 50.35 *** 50.52 ***
R2 .215 .214 .215 .224 .225
N observations 1.569 1.558 1.563 1.394 1.394
Note. This table reports the result for 5 robust standard errors ordinary least squares (OLS) regressions using ‘exploration’ as dependent variable. Regression (1) is the
29
Table 8. Results of the robust standard errors OLS-regression for organizational ambidexterity (log)
Organizational ambidexterity (log)
(1) (2) (3) (4) (5)
b (r.s.e.) b (r.s.e.) b (r.s.e.) b (r.s.e.) b (r.s.e.)
Intercept 96.085 (10.908)*** 93.964 (10.6961)*** 88.980 (10.944)*** 82.479 (11.526)*** 79.853 (11.550)***
Independent variables
CEO compensation (w + log) -.052 (.023)* -.180 (.035)***
TMT compensation (w + log) -.232 (.038)*** -.218 (.041)***
Interaction effects
CEO compensation (w + log) * Compensation gap (w + log)
-.003 (.007)
TMT compensation (w + log) * Compensation gap (w + log)
-.018 (.006)**
Control variables
R&D investments (log) -1.605 (.222)*** -1.459 (.244)*** -.866 (.244)** -.888 (.263)** -.818 (.264)**
Organizational size (w + log) 2.241 (.192)*** 2.323 (.192)*** 2.465 (.196)*** 2.569 (.205)*** 2.578 (.207)***
ROA (w + log) .005 (.031) -.011 (.031) -.002 (.030) -.007 (.032) -.009 (.032)
TobinsQ (w + log) .156 (.050)** .171 (.051)** .219 (.052)*** .259 (.052)*** .267 (.052)***
Year dummies Yes Yes Yes Yes Yes
Industry dummy Yes Yes Yes Yes Yes
F-statistic 42.32 *** 37.54 *** 40.66 *** 34.91 *** 34.74
R2 .145 .150 .165 .175 .176
N observations 1.536 1.525 1.530 1.366 1.366
Note. This table reports the result for 5 robust standard errors ordinary least squares (OLS) regressions using ‘organizational ambidexterity as dependent variable. Regression
30
31 Table 9. Results of the OLS-regression using another measure of exploitation (w + log)
Exploitation (w+ log)
(1) (2) (3) (4) (5)
b (s.e.) b (s.e.) b (s.e.) b (s.e.) b (s.e.)
Intercept -26.055 (9.870)** -24.810 (9.874)* -28.121 (9.950)** -32.008 (10.267)** -33.800 (10.291)**
Independent variables
CEO compensation (w + log) .082 (.019)*** -.092 (.034)**
TMT compensation (w + log) -.081 (.036)* -.125 (.039)**
Interaction effects
CEO compensation (w + log) * Compensation gap (w + log)
-.003 (.006)
TMT compensation (w + log) * Compensation gap (w + log)
-.011 (.005)*
Control variables
R&D investments (log) -.380 (.218)† -.642 (.225)** -.114 (.247) .081 (.257) .165 (.261)
Organizational size (w + log) 1.226 (.182)*** 1.164 (.183)*** 1.298 (.185)*** 1.501 (.190)*** 1.514 (.189)***
ROA (w + log) -.066 (.027)* -.063 (.027)* -.067 (.027)* -.047 (.028)† -.048 (.028)†
TobinsQ (w + log) -.066 (.047) -.068 (.047) -.045 (.048) -.001 (.050) .006 (.050)
Year dummies Yes Yes Yes Yes Yes
Industry dummy Yes Yes Yes Yes Yes
F-statistic 26.48 *** 25.46 *** 23.37 *** 25.55 *** 25.95 ***
R2 .094 .105 .097 .131 .132
Adjusted R2 .090 .101 .093 .125 .127
N observations 1.542 1.531 1.536 1.371 1.371
Note. This table reports the result for 5 ordinary least squares (OLS) regressions using another measure of ‘exploitation’ as dependent variable. Regression (1) is the
32
Table 10. Results of the OLS-regression using another measure of exploration (w + log)
Exploration (w + log)
(1) (2) (3) (4) (5)
b (s.e.) b (s.e.) b (s.e.) b (s.e.) b (s.e.)
Intercept 60.288 (9.662)*** 60.065 (9.737)*** 61.701 (9.748)*** 53.081 (10.380)*** 54.529 (10.398)***
Independent variables
CEO compensation (w + log) -.006 (.0191) .068 (.035)†
TMT compensation (w + log) .058 (.035)† .104 (.039)**
Interaction effects
CEO compensation (w + log) * Compensation gap (w + log)
-.010 (.006)†
TMT compensation (w + log) * Compensation gap (w + log)
-.004 (.005)
Control variables
R&D investments (log) 2.105 (.217)*** 2.118 (.225)*** 1.919 (.244)*** 1.928 (.265)*** 1.830 (.267)***
Organizational size (w + log) -.006 (.182) .002 (.184) -.060 (.185) -.121 (.195) -.146 (.195)
ROA (w + log) -.023 (.027) -.024 (.027) -.021 (.027) -.026 (.029) -.025 (.029)
TobinsQ (w + log) -.040 (.047) -.038 (.047) -.055 (.048) -.058 (.051) -.068 (.051)
Year dummies Yes Yes Yes Yes Yes
Industry dummy Yes Yes Yes Yes Yes
F-statistic 59.75 *** 50.62 *** 51.21 *** 39.61 *** 40.08 ***
R2 .193 .193 .194 .193 .195
Adjusted R2 .190 .189 .190 .188 .190
N observations 1.503 1.493 1.497 1.334 1.334
Note. This table reports the result for 5 ordinary least squares (OLS) regressions using another measure for ‘exploration’ as dependent variable. Regression (1) is the
33
Table 11. Results of the OLS-regression using another measure of organizational ambidexterity (w + log)
Organizational ambidexterity (w + log)
(1) (2) (3) (4) (5)
b (s.e.) b (s.e.) b (s.e.) b (s.e.) b (s.e.)
Intercept 119.187 (11.800)*** 117.445 (11.886)*** 114.213 (11.867)*** 102.861 (12.575)*** 100.661 (12.621)***
Independent variables
CEO compensation (w + log) -.004 (.023) -.166 (.042)***
TMT compensation (w + log) -.144 (.042)** -.185 (.048)***
Interaction effects
CEO compensation (w + log) * Compensation gap (w + log)
-.003 (.007)
TMT compensation (w + log) * Compensation gap (w + log)
-.017 (.007)*
Control variables
R&D investments (log) -.940 (.265)*** -.939 (.275)** -.474 (.298) -.375 (.322) -.362 (.326)
Organizational size (w + log) 1.758 (.221)*** 1.784 (.223)*** 1.892 (.224)*** 2.135 (.236)*** 2.129 (.236)***
ROA (w + log) -.015 (.033) -.019 (.033) -.020 (.033) -.010 (.034) -.010 (.034)
TobinsQ (w + log) .243 (.057)*** .248 (.057)** .286 (.058)*** .340 (.062)*** .343 (.062)***
Year dummies Yes Yes Yes Yes Yes
Industry dummy Yes Yes Yes Yes Yes
F-statistic 36.92 *** 31.29 *** 33.18 *** 27.85 *** 27.72
R2 .131 .131 .138 .147 .146
Adjusted R2 .128 .127 .133 .141 .141
N observations 1.471 1.461 1.465 1.307 1.307
Note. This table reports the result for 5 ordinary least squares (OLS) regressions using another measure of ‘organizational ambidexterity as dependent variable. Regression (1)
34
Discussion and Conclusion
The concept of organizational ambidexterity and its antecedents and outcomes has been widely studied by multiple scholars (He & Wong, 2004; March, 1991; Oehmichen et al., 2017). However, as the transition to a so-called ambidextrous organization and the manageability are so complex, this concept still receives a lot of attention (Certo et al., 2006; O’Reilly & Tushman, 2013). As CEOs and the top management team play an essential role in the decision-making of an organization’s strategy, this research considered their compensation and the distribution of the compensation gap between both organizational levels. This research tried to enhance organizational management theory by looking at the influence of compensation on exploitation, exploration and organizational ambidexterity. The tensions of the behavioral and economic perspective with regard to the influence of compensation gaps were taken into account. It was expected that a compensation gap has a strengthening influence on the positive relationship between CEO compensation and organizational learning strategies, but rather negative when looking at this relationship on top management level.
The results showed that CEO compensation has a negative influence on organizational ambidexterity, which challenges the economic perspective that more compensation elicits more effort for organizational outcomes. Moreover, a negative effect between CEO compensation and exploitation was found, when including the compensation gap in the regression model. The compensation gap itself did not yield a significant effect. Furthermore, TMT compensation has a negative relationship with both exploitation and organizational ambidexterity, which also challenges the economic perspective. To continue, regarding both learning strategies, the interaction effect of the compensation gap is also negative. When looking at TMT compensation and exploration, this direct relationship only becomes significant when the compensation gap, as a non-significant interactor, is present. This indicates the influence of the behavioral perspective when looking at a compensation gap.
35
management team do have the needed capabilities, there must be another reason that explains why compensation does not result in more effort. Traditional agency theory has already been challenged in the last couple of years. Due to the large amount of research done into this theory that provide mixed findings (Bosse & Phillips, 2016), agency theory was argued not to account for bounded self-interest. This might also explain the negative relationships found in this research. Bounded self-interest means that CEOs minimize their effort just to the level where it is not yet perceived unfair, as a means to maximize their own interests (Bosse & Phillips, 2016). This indicates that the amount of compensation does not elicit more individual effort and that organizational ambidexterity therefore, is not enhanced. Third, higher CEO compensation may also have a negative relationship with organizational ambidexterity because of the effect on the opinion of the top management team on this compensation. This implies that the behavioral perspective plays a role in the direct relationship between CEO compensation and organizational learning strategies. When CEO compensation increases, this could result in confusion and cynicism among lower level employees (Henderson & Frederickson, 2001). This discontent and perceived feeling of inequity by the top management team might result in less effort and thus also less organizational performance. As organizational performance is an outcome of overall organizational effort by all employees (Bloom, 1999), this could explain the negative relationship between CEO compensation and organizational ambidexterity. Moreover, this could argue why a compensation gap could have a negative interaction influence. This is something which can, with caution, be interpreted from the results of this research. The direct relationship between CEO compensation and (a) exploitation and (c) organizational ambidexterity were both significant in the case of the presence of a compensation gap. This compensation gap negatively influenced the direct negative relationship. At last, the negative influence of TMT compensation and organizational learning strategies may also be explained based on the fact that the economic perspective is not suitable. Colvin and Boswell (2007) argue that the alignment of interests of the top management team and the organization should not be sought in extrinsic rewards such as compensation, but rather in non-monetary intrinsic work motivators such as meaningfulness. This indicates that job design theory could be the possible explanation for the negative relationship and the negative interaction effect of a compensation gap. It is not surprising that when the top management team is not internally satisfied with their job, compensation does not have a positive influence on organizational learning strategies. Moreover, a compensation gap would strengthen this negative relationship with organizational outcomes, as this results in more dissatisfaction and feelings of inequity, following the behavioral perspective.
36
However, this sample only controls for service and manufacturing firms. Moreover, it may be possible that executive characteristics play a role in this as well. Wowak and Hambrick (2010) namely found that risk-averse individuals do not directly react on compensation and always prefer exploitation over exploration as this yields more results on the short term.
To conclude, the complementarity of the behavioral and economic perspective regarding the influence of compensation on organizational learning strategies is evident. The economic perspective alone does not explain the relationship between compensation and organizational learning strategies and the behavioral perspective does not explain the hypothesized interaction of compensation gaps between both organizational levels. The non-significant and negative results that yielded from the analyses show that other theories must be of more influence. Moreover, the results differ across the dependent variables of exploitation, exploration and organizational ambidexterity. This does indicate the tensions between both learning strategies and the complexity of the concept of organizational ambidexterity. Although this research yielded different significant results than hypothesized, this does contribute to the goal of this study to get a better understanding of the relationship between compensation, a compensation gap, and different learning strategies. As the behavioral and economic perspective do not explain the negative effects alone, further research into other possible explanations must be done.
Limitations and Future Research
This research has certain limitations that will now be discussed and the possibilities for future research will be addressed.
First of all, although the need for longitudinal research (Henderson and Frederickson, 2001) is met, this research did not account for firm-fixed effects. This has implications for possible endogeneity, also due to the correlation of multiple variables in this research. Future research may consider the use of the GMM estimation in STATA 15 (Ioannis et al., 2013) instead of the traditional technique of an OLS-regression to solve this.
Moreover, the robustness check performed with another measurement for the organizational learning strategies found a significant different positive relationship between CEO compensation and exploitation, which challenges the heteroscedasticity of this research and CATA-analysis. Future research must deal with the definition and measurements of the concepts of organizational learning to enhance validity (Lavie et al., 2010). Raisch and Birkinshaw (2008) argue it may be useful to look at industry specific contexts with regard to relevant measurements.
