Too Much of a Good Thing?
How Firm Performance and Equity-Based Compensation Influence Exploration
Vanessa Bienert S2330075 21 January 2019
MSc BA Change Management Supervisor: prof. dr. Jana D. R. Oehmichen
Second evaluator: drs. Heleen P. van Peet
Word count: 9,669
Abstract
Despite indications in previous research that equity-based compensation fosters organisational exploration, there are still pieces missing to creating a cohesive bigger picture. The
organisational level has so far mainly been neglected, in favour of an individual, CEO level, and research on contingency factors, like different levels of firm performance, is still scarce.
This study draws on assumptions of agency theory to hypothesise that equity-based compensation increases exploration and borrows complementing assumptions from the behavioural agency model to hypothesise that low levels of firm performance strengthen this relationship. For the data collection, a method was used that follows suggestions by
McKenny, Aguinis, Short, and Anglin (2018) to utilise computer-aided text analysis (CATA) in creating a strong measurement of exploration. To further incorporate the organisational level, top five manager compensation data was accessed via ExecuComp for the years 2000 to 2015. Ordinary least-squares regressions revealed support for both presented hypotheses.
While boards and shareholders will be pleased to know that equity-based compensation seems to have the intended effects on exploration, these findings should also raise awareness that in difficult times recalibration might be necessary to keep managers from going overboard.
However, future research should provide further practical implications before action is taken.
Keywords: exploration, equity-based compensation, firm performance, agency theory,
behavioural agency model
Too Much of a Good Thing?
How Equity-Based Compensation and Firm Performance Influence Exploration
“Innovation is the only way to win.” – Steve Jobs
Previous research has indicated that equity-based compensation (EBC) is associated with organisational exploration (Cho, 1998; Manso, 2011; Rajgopal & Shevlin, 2002). These suggestions can be substantiated based on agency theory, which proposes a) a misalignment between shareholders’ and managers’ risk preferences for organisations’ paths – due to their differential ties to and dependence on the given organisation (cf. Shapiro, 2005) – and b) the potential to remedy this misalignment to some extent by incentivising managers through equity-based compensation to take higher risks with the organisation (Eisenhardt, 1989) – like exploring for new opportunities (cf. He & Wong, 2004). However, four points can be made, that show that there is still important information missing from the picture and that, at a closer look, things might not always be that simple.
First of all, it has been claimed that, “despite the early attention given to the role of
incentives in March’s (1991) paper, there has been little follow-up by subsequent research on
the role of incentives in exploration” (Lee & Meyer-Doyle, 2017, p.20). This suggests that
evidence of potential effects and mechanisms between EBC and exploration is generally still
scarce. Second, the majority of compensation literature overall (Jensen & Murphy, 1990; Van
Essen, Otten, & Carberry, 2015), and EBC-exploration literature specifically (Cheng, 2004),
is heavily focused on an individual, CEO level, thereby neglecting the roles of further
members of the top management team (TMT) and the wider organisational level. Third,
throughout the literature, exploration has also been defined and operationalised differently
(Coles, Daniel, & Naveen, 2006; Rajgopal & Shevlin, 2002), which might cause difficulties in
drawing a general conclusion about strategic exploration. And fourth, it has been criticised for
a while now that agency theory treats managerial risk preferences as stable (Wiseman &
Gomez-Mejia, 1998), which leaves no room for contingency factors like varying levels of firm performance.
This study aims to make a start at filling in these gaps. Specifically, I will add to existing research by investigating the relationship between EBC and exploration on an organisational level, taking into account top five manager compensation data available through ExecuComp. Furthermore, I will draw on exploration data compiled through computer-aided text analysis (CATA) of Management Discussion & Analysis (MD&A) sections of 10-K filings, through which I hope to create a more straight-forward root measure of strategic exploration than previous operationalisations offered, like for example relative radicalness of innovation (Bierly & Chakrabarti, 1996) or self-report measures of
organisational exploration (Jansen, Van den Bosch, & Volberda, 2006; Kammerlander, Burger, Fust, & Fueglistaller, 2015). Finally, by leveraging arguments of the behavioural agency model (Wiseman & Gomez-Mejia, 1998), which draws on prospect theory
(Kahneman & Tversky, 1979) to extend agency theory’s restricted view on risk preferences, I will try to deepen the insights into the investigated relationship further by adding firm
performance to the equation, as a potentially important moderator.
In a world that is changing more and more rapidly and where innovation is of
importance to stay competitive (cf. Piao, 2010), it is relevant for companies’ survival that they understand how to trigger and foster exploration. This has been illustrated by Smith (2006) who reported that 75% of the revenues of successful companies are produced from new products or services that only emerged in the last five years. However, since exploration is connected to taking increased risk (Jensen et al., 2006), due to its less short-termed and more uncertain outcomes, it is usually not managers’ favourite approach to pursue (March, 1991).
As described at the beginning, based on agency theory, a lot of confidence has been
put into EBC already to increase managerial risk-taking (Eisenhardt, 1989; Feng & Rao,
2018; Sanders & Hambrick, 2007). Following from the same reasoning, EBC could possibly lead to more exploration as well. Indeed, Manso (2011) argued for an “optimal innovation- motivating incentive scheme” (p.1823), consisting of stock options with long vesting periods, managerial entrenchment, golden parachutes, and option repricing. However, his article only offers propositions based on his arguments and no empirical investigations or evidence, which he himself writes would be necessary to find out about the actual effects of the named
regulations. In a study he later conducts with a colleague (Ederer & Manso, 2013), they further investigate the suggested optimal incentive characteristics. But still, this investigation is done in a controlled laboratory setting using a student participant pool and measuring level of exploration by having participants play a game with different option combinations leading to different levels of success, which leaves outcomes of questionable generalisability to the outside business world.
Contrary, Rajgopal and Shevlin (2002) investigated whether stock option risk incentives might influence the actions taken by CEOs of oil and gas firms to manage the uncertain success risk inherent in exploring for new reserves, which they indeed found to be supportive. This study already offers some more support for the proposition that EBC
increases actual organisational exploration. However, in combination with the one previously presented and other studies that operationalised exploration as, for example, relative
radicalness of innovation (Bierly & Chakrabarti, 1996) or merchandisers’ exploratory deals per day (Lee & Meyer-Doyle, 2017), this study is also another good example to underline the claim that research has not been very cohesive so far, regarding its definition and
operationalisation of organisational exploration. Furthermore, it aligns with the majority of studies investigating compensation that have focused only on CEOs (Bebchuk & Fried, 2003;
Gopalan, Milbourn, Song, & Thakor, 2014; Jensen & Murphy, 1990), while EBC is being
used as an incentive for a wider range of management positions (Kolb, 2012). The latter in
itself implies the expectation that these managers will have an influence on the organisational
strategy as well. And while CEOs arguably have a lot of influence in their organisations by themselves (Boivie, Lange, McDonald, & Westphal, 2010; Van Essen et al., 2015), they cannot make strategy all alone: “Setting up a formal innovation management system requires proactive, personal engagement by the top team” (Deschamps & Nelson, 2014, p.50).
Therefore, I further argue that the organisational level should not be overlooked in this matter.
Finally, considering the influence of the factor firm performance, it is necessary to keep in mind that risk cannot always mean the same thing for people – it depends on the situation they are currently in. While agency theory offers a mechanism to reduce managers’
risk aversion to some extent (Eisenhardt, 1989; Jensen & Meckling, 1976), it does not consider different reference frames that managers might use to evaluate their situation. Here, the behavioural agency model, proposed by Wiseman and Gomez-Mejia (1998), poses a useful extension to agency theory alone, as they draw on prospect theory (Kahneman &
Tversky, 1979) to add the concept of loss aversion to the picture and argue that, with
unsatisfactory anticipated firm performance ahead, managers will expect losses of wealth and likely react with even greater risk taking. This idea is supported by research (Audia & Greve, 2006; Bromiley, 1991; Hambrick & D’Avenii, 1988), as well as business examples (see the case of Lytro, Inc.; Robertson, 2018; Rosenthal, 2016) and might offer ground for the concern whether emphasising EBC could also backfire at some point. However, it has not been clearly investigated, yet, what exactly low firm performance’s influence is on the relationship
between EBC and exploration, so this will be of interest in this study as well.
Concluding, these studies offer an additional glimpse at the impression that
meaningful literature on the relationship between EBC and organisational exploration is still
rather scarce and scattered, as the earlier testimony already suggested (Lee & Meyer-Doyle,
2017). By elevating my investigations to the organisational, TMT level and using a refined
measure for exploration and perspective on risk for my arguments about the potentially
moderating factor of firm performance, I hope to be able to offer a valuable contribution to
the understanding of the relationship between equity incentives and exploration at different levels of firm performance. On the one hand, it has been proven to be important that
exploration is fostered in an organisation, in order to sustain innovation and stay competitive.
However, it is also only possible to spend an organisation’s money once, so in times of low firm performance and resource scarcity, too much of a good thing might also become a bad thing. Therefore, I will try to answer the following research question:
How do managerial, equity-based incentives impact organisational exploration?
In the following, I will first elaborate on the existing literature underlying my assumptions and build the arguments for my hypotheses. Afterwards, my methods and conduct will be explained regarding the data and my analyses. Finally, I will present my results and discuss them, as well as limitations to the study, and future research.
Theory and Hypotheses Agency Theory, Equity-Based Compensation, and Exploration
Agency theory and alignment through equity-based compensation. Agency theory dates back to the early 1970s (Pepper & Gore, 2015) and has become the most prominent theoretical framework regarding executive compensation in academic research (Bratton, 2005). Its assumptions are based on two problems, namely the agency problem and the problem of risk sharing, which both arise when principals (usually the shareholders of a firm) hand work to agents (usually the managers in a firm) who should carry it out (Delbufalo, 2018). The underlying factor here, that eventually causes these two problems, is that the principals and agents have different positions in this relationship and that different
expectations and preferences follow from it. While the principals are capable of diversifying
their portfolios, the agents are less flexible, as their employment and income depend on that
one company in question (cf. Shapiro, 2005). Regarding the agency problem, this causes the
issue that both parties differ in their goals and desired paths (for the firm), while it is difficult
for the principal to know exactly what the agent is doing for the company (Bosse & Phillips,
2016). Regarding the problem of risk sharing, their different positions result in different risk preferences because principals would like to see bold and risky approaches to make the company more successful, while agents are assumed to be risk averse (Tosi, Werner, Katz, &
Gomez-Mejia, 2000), due to their closer relationship to and dependence on the company.
To find a solution to the above-named problems, especially that of risk sharing, the framework has relied on more closely aligning agent interests with those of the principal (Seo, 2017). Researchers have suggested that, in order to achieve this closer alignment, an ‘optimal contract’ between shareholders (through the board of directors) and managers should be set up, which offers incentives that make managers more attracted to taking risks regarding their strategy in leading the company (Jensen & Meckling, 1976). More specifically, it has been hypothesised that basing a component of managers’ compensation on their company’s performance on the stock market (i.e., introducing EBC to the compensation package) will make them aim higher, in order to reap maximum benefits themselves through increasing firm performance (Feng & Rao, 2018; Harris, Johnson, & Souder, 2013). The idea is that,
logically, by making managers (future) shareholders themselves, their interest in taking on projects that do not only keep the company going, but possibly make it excel, will be raised.
The business world has taken up on the idea of EBC by covering large parts of the compensation package with forms of it, the two most prominent components being stock options and restricted stock (Carlson & Vogel, 2006). In fact, based on ExecuComp data, EBC comprised more than 50% of the compensation package for S&P 500 CEOs in the years from 1996 to 2010, with an upward trend (Kolb, 2012). And indeed, literature investigating whether this practise of using EBC components seems to work has yielded supportive evidence (Coles et al., 2006; Sanders & Hambrick, 2007), with Feng and Rao (2018)
concluding in their study that “managers appropriately respond to risk incentives by taking on
riskier projects” (p.162). Therefore, it can be assumed that EBC does support managerial risk
taking, in alignment with shareholder preferences.
Equity-based compensation and exploration. He and Wong (2004) state that
“[e]xploration implies firm behaviors characterized by search, discovery, experimentation, risk taking and innovation” (p.481). While experimentation further implies risk through the chance of failure, other authors have likewise described exploration as entailing the
characteristic of risk taking (Levinthal & March, 1993; Swift, 2016). Due to other characteristics, such as lack of short-term outcomes and high uncertainty of success, exploration is traditionally not managers’ favourite approach to pursue (March, 1991).
However, it has also been shown that, if successful, exploration can yield a large number of benefits for organisations, such as competitive advantage, increased firm performance, and diversification (Furman, Gawer, Silverman, & Stern, 2017) and is nowadays an essential part of organisations’ survival due to that. Therefore, it is important to investigate factors that could potentially influence the level of exploration an organisation engages in and find out what fosters managers’ willingness to do so.
Based on the previously presented support for the relationship between EBC and decreased managerial risk aversion, and the cited definition for exploration, I claim that the same effect might be evident for EBC and exploration. Since a number of its characteristics suggest risk taking as a necessity for exploration, I expect that the positive effect of EBC on managerial risk taking will overarchingly also have an enhancing effect on exploration. As discussed in the beginning, previous research has already offered supportive indications for a positive relationship between EBC and exploration specifically (Ederer & Manso, 2013;
Manso, 2011; Rajgopal & Shevlin, 2002). And while, as further discussed, the listed studies might not all be convincing in themselves, yet, I still assume that the general tendency can be taken seriously, based on both these academic indications and similar findings with regards to risk taking alone. Therefore, my first hypothesis is:
H1: EBC increases the amount of exploration in organisations.
Prospect Theory, the Behavioural Agency Model, and Firm Performance
Agency theory has evidently been helpful in explaining principal-agent problems in organisations. However, researchers have argued for a while that its view on risk is too simplistic (Wiseman & Gomez-Mejia, 1998), which makes it hard to argue for outcomes regarding the involvement of different contingency factors that might have an influence on managerial risk perceptions, like firm performance. In order to account for this shortcoming, prospect theory has been suggested as a complementary approach to issues of agency (Pepper
& Gore, 2015), and married together with agency theory in the behavioural agency model by Wiseman and Gomez-Mejia (1998). Prospect theory, developed by Kahneman and Tversky (1979), theorises that people code outcomes as either gains or losses relative to a reference point and that this reference point (and the resulting coding) “can be affected by the
formulation of the offered prospects, and by the expectations of the decision maker” (p.274).
Further research by the authors has shown that people are loss averse, which means that their fear of loss is bigger than their interest in gains, resulting in the acceptance of higher risk, if necessary, to avoid losses (Tversky & Kahneman, 1986). For the formulation of the
behavioural agency model, Wiseman and Gomez-Mejia (1998) have translated these assumptions into an agent context and concluded that “when forecasted performance is unsatisfactory, executives may anticipate losses to wealth (e.g., raises may be withheld, the value of stock options may fall, and so forth) and therefore entertain greater strategic risks on behalf of the firm” (p.137).
Previous research has supported this notion. Hambrick and D’Avenii (1988) found that, while overall levels of risk-taking were similar for successful and declining
organisations, declining organisations had far more varying levels of effort at expansion into new domains, suggesting a bolder approach to risk taking. Furthermore, by testing a time series model, Bromiley (1991) found that firm risk got increased through low firm
profitability and that of its industry. Also, Audia and Greve (2006) found that unsatisfying
firm performance tended to decrease risk taking in small firms, while it tended to increase risk
taking in bigger firms. They hypothesised that this difference might be due to smaller organisations being more sensitive to levels of decline that would not worry bigger organisations and that they also have more limited resources to begin with.
In the context of this study, the presented reasoning would mean that, faced with the potential loss of parts of, or even all their income and their job in the event that an
organisation’s performance is declining and maybe not recovering, managers might take even bigger risks in an attempt to turn things around. While exploration can actually lead to
positive outcomes, helping with a recovery, for example, through the development of new products or acquiring new resources (Morrow, Sirmon, Hitt, & Holcomb, 2007), too much exploration in a situation that does not allow excessive spending on uncertain outcomes could also be the final nail in an organisation’s coffin. Therefore, and remembering the numbers about how much of managers’ compensation is actually already based on EBC, it is important to find out more about this relationship, in order to be able to understand and try to control it.
Based on previous arguments, I assume:
H2: Low firm performance strengthens the relationship between EBC and exploration.
Figure 1 summarises my hypotheses and the expected relationships.
Method Sample and Datasets
The data used for the study at hand was merged into one set from a) a set containing data on organisational exploration, exploitation, and ambidexterity and b) a set containing data on top management compensation. After merging the two datasets, a total of 2,540 observations were available. However, due to missing data for some of the companies or single years, the final numbers of observations considered in the two conducted regressions were 2,310 and 2,299.
The first set was provided by my supervisor and contained data from the WordStat,
LIWC, Datastream, and EDGAR databases on 284 U.S. companies that provide 10-K filings,
ranging from 2000 to 2015 (see Appendix A for the identifier list). The data covered common company-related variables like company size, return on assets, and year dummies, as well as more strategy-related variables regarding level of exploration, exploitation, and
ambidexterity. Specifically, the strategy part of the dataset was compiled in a way following suggestions by McKenny, Aguinis, Short, and Anglin (2018). The authors argue for CATA to be a useful tool for “content analysis that enables the measurement of constructs by
processing text into quantitative data based on the frequency of words” (McKenny et al., 2018, p. 2910). Its advantages include greater confidence in internal validity, as data is not collected via sources like self-report measures or archival data, and in external validity, as generalisability is enhanced through the possibility to collect large amounts of data across units. However, like other data collection methods, CATA is also subject to measurement error variances, which result from, for example, transient error (related to persistent temporal factors influencing authors’ word choices at the time of writing, like emotional state or
business climate) or specific factor error (arising from the potentially idiosyncratically chosen words on researchers’ word lists). McKenny et al. (2018) suggest remedies for the discussed errors, which are reflected in the provided variables. These are, for example, limiting word searches to Management Discussion & Analysis (MD&A) sections of 10-K filings, which are usually written quite formally and with great consistency, or refining word lists through different means. More specific information on all variables that were used will be provided in the variable descriptions below.
The second set was downloaded and processed by me and two other students, working
on similar projects, from the database ExecuComp. This database, dating back to 1992, is
being maintained and offered by Standard & Poor’s and provides comprehensive information
on the compensation of CEOs and the four following top managers of companies (Kolb,
2012). Using the company identifier list (see Appendix A) from the previously described,
provided dataset, we downloaded all available compensation data for these five managers for
the same timeframe, 2000 to 2015. For the purpose of the study at hand, we then aggregated all individual compensation variables into one organisational variable (per company, per fiscal year) by averaging the total of all managers listed for the company by the number of the listed managers. Additionally, we manually included the variable of firm age, which will also be described in more detail below.
Measures
Independent variables. To measure EBC, I summed up the aggregated,
organisational-level versions of the individual-level variables restricted stock holdings and unexercised stock options provided by ExecuComp. Regarding stock options, both
unexercisable and exercisable unexercised stock options were summed up to yield one
measure of all unexercised stock options. Since the study considers level of firm performance as a moderator, I decided to use the average number of restricted stock and stock options held by the top management teams, instead of the average value. That is because the value varies in line with and therefore, to some extent, reflects firm performance itself. However,
regarding managerial behaviour, owning less or more restricted stocks and stock options should have a differential and interesting interaction effect with varying levels of firm
performance, as the impact on the amount of received compensation will be smaller or bigger and therefore of smaller or bigger concern.
Moderator. Firm performance was measured as the companies’ annual return on assets (ROA), as it is a) a widely used variable to measure firm performance (Heyden, Oehmichen, Nichting, & Volberda, 2015; Oehmichen, Heyden, Georgakakis, & Volberda, 2017), and b) accounting-based measures (like ROA) and market-based measures (like, for example, Tobin’s q) were found to be of equal suitability in measuring performance (Martin, 1993). The variable was provided in the organisational strategy dataset.
Dependent variables. Organisational exploration was measured using a variable
from the described dataset that was replicated using CATA, following suggestions by
McKenny et al. (2018). The specific variable was set up as word counts resulting from word lists used on MD&A sections, providing the sum of exploration words per identifier and year.
Furthermore, this count was cleaned through keyword-in-context-analysis to see whether all detected keywords actually referred to the concept under investigation (i.e., strategic
exploration). Later on, a similar variable was used for a robustness check, which followed the suggestion to clean the data through removing word stems from the word list, that were often found to lead to false positives regarding the use of the words, and instead adding conjugated words or short word combinations. However, for the main regression analysis, I chose the variable described before, as it seemed to be the one that has the most advantages regarding measurement errors.
Control Variables. To consider potential effects of confounding variables, different control variables were included in the analysis as well. On the one hand, the two variables related to the top management team were age (in years, averaged throughout the team) and a dummy variable for gender (0 = female, 1 = male, averaged throughout the team). These two variables are commonly used control variables in studying human behaviour and could have an influence in that age has been found to modestly decrease people’s relative risk aversion (Bellante & Green, 2004) and women have been found to be more risk averse than men (Borghans, Golsteyn, Heckman, & Meijers, 2009). On the other hand, the three organisational control variables were firm size (number of employees), firm age (in years, reported relative to the respective fiscal year), and a dummy variable for industry (0 = manufacturing, 1 = services). Here, previous research has shown that older and larger companies usually follow different orientations regarding strategies than younger and smaller companies (Tushman &
Romanelli, 1985), and that firm size also explains CEO compensation, as an increase in firm
size was connected to an increase in compensation (Nourayi & Mintz, 2008). Both variables
could account for variances in strategy that are not being explained by TMT compensation
(alone). Furthermore, due to the differential set-up and mechanisms of manufacturing and
service companies, these two dummies were included to account for industry effects. Finally, another dummy variable year was included to account for year effects in numbers.
Age and gender were derived from ExecuComp, while firm size and industry
dummies were included in the provided dataset and marked as derived from Datastream and EDGAR, respectively. Firm age was created manually by subtracting the companies’
founding years from the respective fiscal year. Since different definitions exist regarding founding years, considering mergers, name changes, and other aspects, and firm age was not supposed to be a main variable in this study, we decided to use the Google indication
1of the founding years of our companies for reasons of replicability and consistency regarding the source. The advantages and disadvantages of this way of conduct will be discussed later in the limitations section.
Model
To investigate the influence of firm performance on the relationship between long- term compensation and exploration, an ordinary least-squares (OLS) regression was used.
This procedure is commonly utilised to test moderation models and draw conclusions about the interactions between various variables (cf. Cohen, Cohen, West, & Aiken, 2003). Prior to the analysis, the respective statistical assumptions were checked and after the analysis, a robustness check was conducted with an alternative variable for exploration, in order to test for replicability beyond the chosen variable.
Results Preliminary Analyses
Prior to any analyses, the assumptions for regression analyses were checked and a
number of variables were adjusted accordingly. Both the EBC and the exploration variable
were logged, in order to remedy issues with normality. The same was done with the control
variable of firm age. Furthermore, the variable of firm performance was winsorised at the 5%
level, due to extreme outliers in the variable but a sufficiently normal distribution (cf.
McGuire, Oehmichen, Wolff, & Hilgers, 2017).
All means, standard deviations, and pairwise correlations can be found in Table 1. The majority of correlations is statistically significant, suggesting potential issues due to
multicollinearity. However, an additional check for all variables’ variance inflation factors (VIF) yielded no higher results than 2.20, which is far below the concerning threshold of 10, so multicollinearity did not jeopardise further results or conclusions (Neter, Kutner,
Nachtsheim, & Wasserman, 1996). Furthermore, a number of the correlations could be expected to be at least approaching significance, due to previously named and other earlier findings. These are, for example, the significant positive correlations between, on the one hand, company size and, on the other hand, exploration (r = 0.33), EBC (r = 0.30), and ROA (r = 0.23), connected to the reported claim that bigger companies have more resources at their disposal, which gives them a wider action range, and they have to be less concerned about minor bumps than smaller firms (Audia & Greve, 2006). Also, the significant positive
correlation between EBC and exploration (r = 0.16) gives a first supportive indication towards the hypothesised relationship between the two variables (H1). More surprising were the significant negative correlations between gender and exploration on the one hand (r = -0.10), and gender and EBC on the other hand (r = -0.10), as they contradict earlier suggestions about, for example, women being more risk averse than men (Borghans et al., 2009) and the claim about a pay gap between the two sexes (Poulsen, 2011). I will therefore get back to these two relationships in the Discussion.
Hypothesis Testing
Following a common procedure (Aiken & West, 1991), all independent variables were centred for the conducted OLS regression analysis, prior to generating the interaction term.
These centred independent variables and the resulting interaction term were used for the
analysis investigating the second hypothesis of this study. An overview of the results can be found in Table 2.
The first hypothesis that was being tested, predicted an increase in the amount of exploration through a higher amount of EBC. Indeed, a significant supporting outcome was found (b = 0.05, SE
b= 0.02, F(2288) = 20.85, p < 0.05; see model 1 in Table 2), statistically confirming the predicted positive relationship.
The second hypothesis in this study predicted that low firm performance would further strengthen the relationship between EBC and exploration. Again, statistical evidence was found that supported the predicted effect of firm performance (b = -0.49, SE
b= 0.02, F(2275)
= 20.44, p < 0.001; see model 2 in Table 2). The reported negative relation of the interaction term with EBC and exploration means that, with growing firm performance, the slope
describing the positive relationship between EBC and exploration will get flatter. In turn, this also means that, with decreasing firm performance, the slope describing the positive
relationship between EBC and exploration will get steeper, confirming the prediction of hypothesis 2. A visual representation of this relationship can be found in Figure 2.
Robustness check
In order to investigate whether the reported findings are robust across different variables suggested by McKenny et al. (2018), the OLS regression was conducted again, using the second presented variable, based on the clean-up concerning word stems. After conducting preliminary investigations once more that should rule out concerns about the assumptions of regression analyses, the analyses were conducted with the logged alternative exploration variable, which was again centred prior to the interaction analysis.
The results of both analyses show the same outcomes for both hypotheses and are
presented in Table 3. Again, a significant positive relationship was found between EBC and
exploration (b = 0.05, SE
b= 0.02, F(2288) = 20.85, p < 0.05; see model 1 in Table 3) and the
relationship was strengthened in the same way under the influence of the moderator firm
performance (b = 0.07, SE
b= 0.02, F(2275) = 20.44, p < 0.01; see model 2 in Table 3). These outcomes support the reliability of the results and of the concept of exploration across the established variables.
Discussion Purpose and Results
Research based on agency theory and the concern around the misalignment of risk preferences between shareholders and managers of a firm has brought forward numerous studies already, investigating the decreasing effects of EBC on risk aversion (Coles et al., 2006; Feng & Rao, 2018; Sanders & Hambrick, 2007). Much less effort has yet been invested into the question, whether this effect might also go one step further, resulting in more
readiness to take the risk of emphasising organisational exploration as well (Ederer & Manso, 2013; Rajgopal & Shevlin, 2002). I pointed out four aspects about research into this matter that still need further attention, in order to paint a more complete and cohesive picture of the relationships and mechanisms in place – scarcity of research in general, a too narrow focus on the individual, CEO level, too scattered ideas about the concept of exploration to offer
sufficient generalisability, and the lack of room for contingency factors in agency theory’s view on risk. By elevating my research to the organisational, TMT level, leveraging a CATA- based approach to defining and operationalising exploration (McKenny et al., 2018), and using ideas underlying the behavioural agency model (Wiseman & Gomez-Mejia, 1998) as a theoretical basis complementary to agency theory, I hoped to be able to offer some more insights aiding to fill these gaps the literature still has.
The hypotheses I proposed were that, on the one hand, EBC was expected to increase the amount of exploration in an organisation (H1), and that, on the other hand, low firm performance was expected to further strengthen this relationship between EBC and
exploration (H2). My analyses revealed that these assumptions indeed seem to be supported
by data and that these effects also hold across two different ways of constructing the variable
measuring exploration. Following theory-based lines of reasoning, the results suggest two things. First of all, derived from assumptions underlying agency theory (Eisenhardt, 1989), it seems like the mechanism that is theorised to decrease managers’ risk aversion by introducing EBC components into their compensation packages, also still reaches one step further,
resulting in just as reduced concerns about taking the risk of emphasising organisational exploration. Second of all, under consideration of assumptions underlying the behavioural agency model (Wiseman & Gomez-Mejia, 1998; based on prospect theory; Kahneman &
Tversky, 1979), the reasoning that loss aversion concerning compensation and job safety might further excel risk taking, seems to equally translate into the relationship between EBC and exploration as well. Therefore, my results offer overall support for the extendibility of these theories to further concepts related to risk taking. Furthermore, with the gained insights I can answer my research question – How do managerial, equity-based incentives impact organisational exploration? – by stating that these types of incentives seem to increase not only managerial risk taking but, consequently, also organisational exploration, and even more so under circumstances of low firm performance.
Two more results of which I would like to remind the reader, are the negative
correlations between gender and EBC, and gender and exploration. Despite previous literature suggesting the opposite directions (Blau & Kahn, 2017; Borghans et al., 2009; Poulsen, 2011), these two yielded significant results indicating that female members on the TMT were connected to both higher EBC and exploration. In the following, I will discuss a possible explanation for each of these effects.
Even though it is commonly criticised that a pay gap exists between male and female employees, leaving the latter at a disadvantage (Blau & Kahn, 2017; Poulsen, 2011), it actually also seems to be the case that this is not true for the highest levels in organisations.
Even though this field of research (into the advantages that female managers might actually
have) is still young, due to the number of female managers and CEOs only gradually
increasing over the last years (Donovan, 2015), Hill, Upadhyay, and Beekun (2015) indicated that on the highest level of employment, the pay gap tables might actually be turned, leaving the male CEOs at a disadvantage. In a follow-up study by Gupta, Mortal, and Guo (2018) this indication could not be confirmed, as they could not find a pay gap discriminating against either of the two sexes. However, as a potential pay gap disadvantaging male managers seems to also be suggested in the data investigated in this study, future research should consider taking another, closer look at the effects that might be evident on this organisational level.
Furthermore, despite the previous finding that women tend to be more risk averse (Borghans et al., 2009), the results of this study also suggest a significant contradicting relationship between gender and exploration, indicating that TMTs with a higher number of female members might actually engage in more exploration. An explanation for this
surprising finding could lie in the different leadership approaches that men and women tend to exhibit. Avolio and Bass (1991) developed the Full Range Leadership Model, which captures eight leadership approaches, ranging from passive-ineffective (i.e., categorised as more transactional) to active-effective (i.e., categorised as more transformational). Research by Eagly and Carli (2003) into gender differences along this continuum showed that women are more likely to engage in approaches that are associated with transformational leadership, while the contrary was found for men. As transformational leadership is characterised by communicating a vision to followers and encouraging them to engage in new ways of thinking (Antonakis, Avolio, & Sivasubramaniam, 2003; Kark, Waismel-Manor, & Shamir, 2012), intuitively this also seems to fit better with the exploration of new opportunities than with exploitation of existing products and services, possibly explaining the found correlation.
However, in both lines of arguments it should be taken into account that the overall
number of female managers in this study was very low compared to male managers and that
gender was only investigated as a control variable. Therefore, both suggestions for an
explanation should be investigated more carefully in future research before drawing any premature conclusions.
Theoretical and Practical Implications
As it has been found that EBC increases exploration and that this effect is even stronger in the context of low firm performance, a number of implications can be derived from this.
Theoretically, as already described above, the results of this study offer extended grounds for assumptions based on agency theory and the behavioural agency model, since both hypotheses based on these were supported. Despite the increased attention and popularity agency theory has enjoyed over the years, it probably received just as much critique for being too simplistic (Wiseman & Gomez-Mejia, 1998) and for predicting outcomes of EBC that are not actually evident in reality (Ross, 2004). This study, however, strengthens the confidence in this theory and its assumptions and suggests that it might even be useful for a wider range of predictions than just those directly related to risk taking. And while not being as harshly criticised as agency theory, nor being as widely used, yet, in the context of EBC and risk taking, this study also offers support for the behavioural agency model (Wiseman & Gomez-Mejia, 1998), with its underlying assumptions of prospect theory (Kahneman & Tversky, 1979), through the confirmed effects of firm performance on the relationship between EBC and exploration.
Furthermore, this study contributes to a wider insight into managerial behaviour, due to its increased radius of investigation. Since previous literature leaned heavily towards the individual, CEO level (Cheng, 2004; Jensen & Murphy, 1990; Rajgopal & Shevlin, 2002;
Van Essen et al., 2015), the influence of further members of the TMT organisational strategy
was neglected and potentially important nuances might have been overlooked. By taking into
account the top five managers of organisations in this study, I hoped to contribute to the
literature in offering somewhat richer findings about the relationship between EBC and
exploration. A similar reasoning precedes my method of measuring exploration. While
previous research mainly focused on aspects of exploration, like exploring for new oil and gas reserves (Rajgopal & Shevlin, 2002), relative radicalness of innovation (Bierly & Chakrabarti, 1996), or merchandisers’ exploratory deals per day (Lee & Meyer-Doyle, 2017), I tried to utilise a more straight-forward root measurement of organisational exploration. This was done through CATA used on MD&A sections of 10-K filings, which should yield an indication of the organisations’ overall intentions regarding exploration, which can then subsequently manifest themselves in different organisational actions, like the ones named above. The robustness check that yielded the same result as the main analyses further confirms that the variables McKenny et al. (2018) suggested in their investigations of the CATA procedure are qualitatively similar and could be used interchangeably.
Regarding managerial implications, this study suggests that the idea underlying agency theory’s proposition that EBC can remedy some of the misalignment between
shareholders’ and managers’ risk preferences (Feng & Rao, 2018; Harris et al., 2013), might be a bit more complicated in reality. Even though the results supported a direct relationship between EBC and exploration, they also showed that contingency factors can further
influence this relationship and make it more complicated than just an increase in exploration in response to higher EBC. Shareholders might desire a higher rate of exploration in order to maximise returns and market value of the organisation, which will lead to higher value of shares benefitting them. However, under the precarious circumstances that the organisation’s performance is already at a low level and remembering that money can only be spent once, it should be reconsidered, whether an adjustment in the TMT’s pay package might be necessary in these harder times to avoid excessive spending on exploration that might not contribute to a recovery anymore but rather make the situation even worse.
Limitations and Future Research
Despite the qualities of this study, it also still has some important limitations that should be addressed. First of all, regarding the discussed managerial implication that caution might be important, and calibrating the TMT’s pay package under different circumstances, it also needs to be considered that this study does not offer enough information about the practical significance of the findings to already be truly alarming for boards and shareholders.
Statistically significant indications might have been found in the data but the underlying dimensions that these mechanisms seem to influence cannot reliably be derived from the outcomes, yet. Therefore, further investigations will be needed into the actual consequences these implications might have before premature actions should be taken.
Another limitation, that somewhat relates to the previous one, is that in the study at hand it was not considered what the respective firm performances were like in previous years.
Despite the data being longitudinal data, only a general trend was investigated in the analyses and previous performance was not an investigated variable in this study. However, based on the assumptions underlying the theoretical background, it could be proposed that loss aversion might only cause the predicted effects in the event that the reference point changes to a
situation that suggests future losses, and not if the situation stands constant and is just at a lower level than would be claimed optimal in general. Therefore, future research should extend this investigation further by considering trends in firm performance, in order to yield a more nuanced picture of the mechanisms and triggers in place.
A limitation that was already announced earlier is the use of founding year data
accessed through Google information. As different definitions can be brought forward as to
when a company was founded (e.g., first listing, post-merger, name changes, or even first
garage meeting and prototype-development) and it is not obvious which information Google
uses to establish the displayed founding years on its website, a standardised approach to
finding out about the founding years cannot be offered through this approach. If firm age were
to be used as a main variable in an investigation, its establishment should be conducted more
carefully and considerate. However, since firm age only served as a control variable in this study and the number of organisations investigated was too high to justify a more thorough approach, a high level of accuracy and standardisation was sacrificed to offer a more repeatable approach instead, by offering all necessary information about the procedure
1.
Finally, using longitudinal compensation data offered through ExecuComp provides a great opportunity to access a vast database of organisational information and draw
conclusions that are more reliable than if only one year or a small number of years would have been considered in the analyses. However, as it got obvious from preliminary
investigations of the raw data, databases like these are also not fully complete and might lack certain amounts of information regarding some provided variables and organisations. It was taken care in this study that only rich enough information was used but future research should still consider other methods to access data that are more reliable and complete.
Conclusion
In this study I investigated whether EBC leads to higher levels of exploration and whether low levels of firm performance further strengthen this relationship. My claims were based on assumptions underlying agency theory and the behavioural agency model (drawing on prospect theory), and substantiated by previous literature suggesting these hypothesised relationships. Both hypotheses were statistically supported by the data. Based on my findings, I suggest that boards and shareholders of organisations exhibiting low firm performance should take a moment to consider recalibrating TMT pay packages regarding EBC, in case extensive exploration seems to be a critical issue endangering organisational survival.
However, in order to draw more reliable conclusions about this study and its implications, and
offer stronger advice to managers, future research should first confirm these findings and
provide concrete practical implications of the outcomes.
References
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage.
Antonakis, J., Avolio, B. J., & Sivasubramaniam, N. (2003). Context and leadership: An examination of the nine-factor full-range leadership theory using the Multifactor Leadership Questionnaire. The Leadership Quarterly, 14(3), 261-295.
Audia, P. G., & Greve, H. R. (2006). Less likely to fail: Low performance, firm size, and factory expansion in the shipbuilding industry. Management Science, 52(1), 83-94.
Avolio, B. J., & Bass, B. M. (1991). The full range of leadership development: Basic and advanced manuals. Binghamton, NY: Bass, Avolio & Associates.
Bebchuk, L. A., & Fried, J. M. (2003). Executive compensation as an agency problem.
Journal of Economic Perspectives, 17(3), 71–92.
Bellante, D., & Green, C. (2004). Relative risk aversion among the elderly. Review of Financial Economics, 13(3), 269-281.
Bierly, P., & Chakrabarti, A. (1996). Generic knowledge strategies in the US pharmaceutical industry. Strategic Management Journal, 17(S2), 123-135.
Blau, F. D., & Kahn, L. M. (2017). The gender wage gap: Extend, trends, and explanations.
Journal of Economic Literature, 55(3), 789–835.
Boivie, S., Lange, D., McDonald, M. L., & Westphal, J. D. (2011). Me or we: The effects of CEO organizational identification on agency costs. Academy of Management Journal, 54(3), 551-576.
Borghans, L., Heckman, J. J., Golsteyn, B. H., & Meijers, H. (2009). Gender differences in risk aversion and ambiguity aversion. Journal of the European Economic Association, 7(2-3), 649-658.
Bosse, D. A., & Phillips, R. A. (2016). Agency theory and bounded self-interest. Academy of
Management Review, 41(2), 276-297.
Bratton, W. (2005). The academic tournament over executive compensation. California Law Review, 93(5), 1557-1584.
Bromiley, P. (1991). Testing a causal model of corporate risk and performance. Academy of Management Journal, 34(1), 37-59.
Carlson, R. L., & Vogel, T. J. (2006). Restricted stock versus stock options: The case of Jones Apparel Group, Inc. Issues in Accounting Education Teaching Notes, 21(4), 8-24.
Cheng, S. (2004). R&D expenditures and CEO compensation. The Accounting Review, 79(2), 305-328.
Cho, M. H. (1998). Ownership structure, investment, and the corporate value: An empirical analysis. Journal of Financial Economics, 47(1), 103-122.
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/
correlation analysis for the behavioral sciences. UK: Taylor & Francis.
Coles, J. L., Daniel, N. D., & Naveen, L. (2006). Managerial incentives and risk-taking.
Journal of Financial Economics, 79(2), 431-468.
Delbufalo, E. (2018). Agency theory and sustainability in the global supply chain. Cham:
Springer.
Deschamps, J. P., & Nelson, B. (2014). Innovation governance: How top management organizes and mobilizes for innovation. John Wiley & Sons.
Donovan, J. (2015). Women fortune 500 CEOs: Held to a higher standard. New York, NY:
American Management Association.
Eagly, A. H., & Carli, L. L. (2003). The female leadership advantage: An evaluation of the evidence. The Leadership Quarterly, 14(6), 807-834.
Ederer, F., & Manso, G. (2013). Is pay for performance detrimental to innovation?.
Management Science, 59(7), 1496-1513.
Eisenhardt, K. M. (1989). Agency theory: An assessment and review. The Academy of
Management Review, 14(1), 57-74.
Feng, H., & Rao, R. P. (2018). Cash holdings and CEO risk incentive compensation: Effect of CEO risk aversion. International Review of Financial Analysis, 60, 162-176.
Furman, J., Gawer, A., Silverman, B. S., & Stern, S. (Eds.). (2017). Entrepreneurship, innovation, and platforms. Bingley, UK: Emerald Group Publishing.
Gopalan, R., Milbourn, T., Song, F., & Thakor, A. V. (2014). Duration of executive compensation. The Journal of Finance, 69(6), 2777-2817.
Gupta, V. K., Mortal, S. C., & Guo, X. (2018). Revisiting the gender gap in CEO
compensation: Replication and extension of Hill, Upadhyay, and Beekun’s (2015) work on CEO gender pay gap. Strategic Management Journal, 39(7), 2036–2050.
Hambrick, D. C., D'Aveni, R. A. (1988). Large corporate failures as downward spirals.
Administrative Science Quarterly, 33(1), 1-23.
Harris, J. D., Johnson, S. G., & Souder, D. (2013). Model-theoretic knowledge accumulation:
The case of agency theory and incentive alignment. Academy of Management Review, 38(3), 442-454.
He, Z. L., & Wong, P. K. (2004). Exploration vs. exploitation: An empirical test of the ambidexterity hypothesis. Organization Science, 15(4), 481-494.
Heyden, M. L., Oehmichen, J., Nichting, S., & Volberda, H. W. (2015). Board background heterogeneity and exploration‐exploitation: The role of the institutionally adopted board model. Global Strategy Journal, 5(2), 154-176.
Hill, A. D., Upadhyay, A. D., & Beekun, R. I. (2015). Do female and ethnically diverse executives endure inequity in the CEO position or do they benefit from their minority status? An empirical examination. Strategic Management Journal, 36(8), 1115–1134.
Jansen, J. J. P., Van den Bosch, F. A. J., & Volberda, H. W. (2006). Exploratory innovation, exploitative innovation, and performance: Effects of organizational antecedents and environmental moderators. Management Science, 52(11), 1661-1674.
Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency
costs and ownership structure. Journal of Financial Economics, 3(4), 305-360.
Jensen, M. C., & Murphy, K. J. (1990). Performance pay and top-management incentives.
Journal of Political Economy, 98(2), 225-264.
J Rosenthal. (2016, April 4). Why I lit up Lytro and scrapped the strategy as CEO. [web log comment]. Retrieved from https://petapixel.com/2016/04/04/lit-lytro-scrapped- strategy-ceo/
Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decisions under risk.
Econometrica, 47(2), 262-291.
Kammerlander, N., Burger, D., Fust, A., & Fueglistaller, U. (2015). Exploration and
exploitation in established small and medium-sized enterprises: The effect of CEOs' regulatory focus. Journal of Business Venturing, 30(4), 582-602.
Kark, R., Waismel-Manor, R., & Shamir, B. (2012). Does valuing androgyny and femininity lead to a female advantage? The relationship between gender-role, transformational leadership and identification. The Leadership Quarterly, 23(3), 620-640.
Kolb, R. W. (2012). Too much is not enough: Incentives in executive compensation. Oxford University Press.
Lee, S., & Meyer-Doyle, P. (2017). How performance incentives shape individual exploration and exploitation: Evidence from micro-data. Organization Science, 28(1), 19-38.
Levinthal, D. A., & March, J. G. (1993). The myopia of learning. Strategic Management Journal, 14(S2), 95-112.
Manso, G. (2011). Motivating innovation. The Journal of Finance, 66(5), 1823-1860.
March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2(1), 71–87.
Martin, S. (1993). Advanced industrial economics: Solutions manual. Blackwell.
McGuire, J., Oehmichen, J., Wolff, M., & Hilgers, R. (2017). Do contracts make them care?
The impact of CEO compensation design on corporate social performance. Journal of
Business Ethics, 1-16.
McKenny, A. F., Aguinis, H., Short, J. C., & Anglin, A. H. (2018). What doesn’t get
measured does exist: Improving the accuracy of computer-aided text analysis. Journal of Management, 44(7), 2909-2933.
Morrow Jr, J. L., Sirmon, D. G., Hitt, M. A., & Holcomb, T. R. (2007). Creating value in the face of declining performance: Firm strategies and organizational recovery. Strategic Management Journal, 28(3), 271-283.
Neter, J., Kutner, M. H., Nachtsheim, C. J., & Wasserman, W. (1996). Applied linear statistical models (Vol. 4). Chicago: Irwin.
Nourayi, M. M., & Mintz, S. M. (2008). Tenure, firm's performance, and CEO's compensation. Managerial Finance, 34(8), 524-536.
Oehmichen, J., Heyden, M. L., Georgakakis, D., & Volberda, H. W. (2017). Boards of directors and organizational ambidexterity in knowledge-intensive firms. The International Journal of Human Resource Management, 28(2), 283-306.
Pepper, A., & Gore, J. (2015). Behavioral agency theory: New foundations for theorizing about executive compensation. Journal of Management, 41(4), 1045-1068.
Piao, M. (2010). Thriving in the new: Implication of exploration on organizational longevity.
Journal of Management, 36(6), 1529-1554.
Poulsen, K. A. (2011). Exploring pay and management gaps for women. New York: Nova Science Publishers, Inc.
Rajgopal, S., & Shevlin, T. (2002). Empirical evidence on the relation between stock option compensation and risk taking. Journal of Accounting and Economics, 33(2), 145-171.
Robertson, A. (2018, March 27). VR camera maker Lytro is shutting down, and former employees are going to Google. The Verge. Retrieved from
https://www.theverge.com/2018/3/27/17166038/lytro-light-field-camera-company-
shuts-down-google-hiring
Ross, S. (2004). Compensation, incentives, and the duality of risk aversion and riskiness. The Journal of Finance, 59(1), 207-225.
Sanders, W. G., & Hambrick, D. C. (2007). Swinging for the fences: The effects of CEO stock options on company risk taking and performance. Academy of Management Journal, 50(5), 1055-1078.
Seo, J. (2017). Board effectiveness and CEO pay: Board information processing capacity, monitoring complexity, and CEO pay-for-performance sensitivity. Human Resource Management, 56(3), 373-388.
Shapiro, S. (2005). Agency theory. Annual Review of Sociology, 31, 263-284.
Smith, D. (2006). Designing an innovative Britain. ESRC: The Edge, 22(2).
Swift, T. (2016). The perilous leap between exploration and exploitation. Strategic Management Journal, 37(8), 1688-1698.
Tosi, H., Werner, S., Katz, J., & Gomez-Mejia, L. (2000). How much does performance matter? A meta-analysis of CEO pay studies. Journal of Management, 26(2), 301-339.
Tushman, M. L., & Romanelli, E. (1985). Organizational evolution: A metamorphosis model of convergence and reorientation. Research in Organizational Behavior, 7, 171-222.
Tversky, A., & Kahneman, D. (1986). Rational choice and the framing of decisions. Journal of Business, 59(4), S251-S278.
Van Essen, M., Otten, J., & Carberry, E. J. (2015). Assessing managerial power theory: A meta-analytic approach to understanding the determinants of CEO compensation.
Journal of Management, 41(1), 164-202.
Wiseman, R. M., & Gomez-Mejia, L. R. (1998). A behavioral agency model of managerial
risk taking. The Academy of Management Review, 23(1), 133-153.
Footnotes
1
The founding years were ascertained via Google.com, entering “founding year [company
name derived from the dataset]”. The respective founding dates were then displayed in the
information box at the top of the page. Accessed October 2018.
Table 1
Means, Standard Deviations, and Pairwise Correlations for all Model Variables
Variables M SD 1. 2. 3. 4. 5. 6. 7.
1. Exploration (log) 2.19 0.80 1.00
2. EBC (log) 5.94 1.18 0.16* 1.00
3. ROA (w) 4.72 11.46 0.02 -0.03 1.00
4. Gender 0.92 0.18 -0.10* -0.10* 0.01 1.00
5. Age 50.66 5.21 0.08* 0.10* 0.08* -0.03 1.00
6. Company size 7.77 1.69 0.33* 0.30* 0.23* -0.04 0.30* 1.00
7. Firm age (log) 3.11 0.76 0.14* -0.00 0.18* -0.03 0.35* 0.44* 1.00
