Performance feedback and organizational risk-taking behavior: the moderating effects of
unabsorbed, absorbed and potential slack.
Master's Thesis (Final version)
Date 22th of June, 2018
Student Léonardus Martinus Vincent Jansen / Student no. 10878521
MSc. in Business Administration: Strategy
University of Amsterdam, Amsterdam Business School
Supervisor MSc. B. Silveira Barbosa Correia Lima
2 Statement of originality
This document is written by Student Léon Jansen who declares to take full responsibility for
the contents of this document.
I declare that the text and the work presented in this document is original and that no sources
other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of
3
Table of Contents
Table of Contents ... 3
Abstract ... 4
1. Introduction ... 5
2. Theory and Hypotheses ... 9
2.1. Performance feedback and organizational risk-taking behavior ... 9
2.2. Literature gap ... 13
2.3. Performance feedback and organizational slack ... 17
2.4. Research question ... 20
2.5. Performance below the aspiration level and organizational risk-taking behavior ... 21
2.6. Unabsorbed slack as a moderator ... 22
2.7. Absorbed slack as a moderator ... 24
2.8. Potential slack as a moderator ... 25
3. Methodology ... 27 3.1. Sample ... 27 3.2. Dependent variable ... 28 3.3. Independent variables ... 28 3.4. Moderating variables ... 29 3.5. Control variables ... 30 3.6. Statistical Model ... 31 4. Results ... 32
4.1. Sample overview and descriptive statistics ... 32
4.2. Regression analyses ... 36
5. Discussion ... 43
5.1. Findings of this study ... 43
5.2. Implications for research ... 46
5.3. Implications for practice ... 47
5.4. Limitations and suggestions for future research ... 48
6. Conclusion ... 50
4 Abstract
Behavioral theory states that when firm performance declines, decision-makers respond with
an attempt at solving the cause of the decline. While this process called 'problemistic search'
is well-grounded in the literature, its effect on managers inclination to take risk is not.
Contradictory evidence regarding the extent managers engage in risk-taking behavior when
faced with performance shortfalls, instigated a stream of research that attempts to identify
sources of heterogeneity among firms. By identifying such sources that moderate the
relationship between performance shortfalls and risk-taking behavior, I develop a more
thorough understanding of the mechanisms and attempt to reconcile the antagonistic views in
the debate. This study examines the roles of unabsorbed, absorbed and potential slack as
sources of heterogeneity among firms. This paper builds on earlier research that has
investigated the effect of one particular type of slack (financial) and continues that work by
hypothesizing that various types of slack moderate the aforementioned relationship
differently. First, I hypothesize that, generally speaking, performance below the aspiration
level triggers risk-taking behavior. Second, I propose that unabsorbed and potential slack
amplify this relationship and absorbed slack weakens it. I use financial data from firms in
manufacturing industries, gathered from the Wharton Compustat database, to test these
hypotheses. The results confirm a negative relationship between performance decline and
risk-taking behavior. Expectedly, unabsorbed slack strengthens this relationship. Insignificant
results were found regarding the moderating effects of absorbed and potential slack. Overall,
this paper contributes to a more comprehensive understanding of risk-taking behavior under
5 1. Introduction
The behavioral perspective of the firm has been extraordinary influential in guiding research
on organizations and strategy over the last decades (Gavetti, Greve, Levinthal, & Ocasio,
2012; Gavetti, Levinthal, & Ocasio, 2007; Shinkle, 2012). Performance feedback is an
important concept within this view (Gavetti et al., 2012). The premise of this concept is that,
because of bounded rationality, decision-makers use aspirations to determine whether firm
performance is satisfactory. This allows decision-makers to learn and constantly adapt their
organization, by evaluating their performance and using that evaluation as a basis for their
decisions. Consequently, performance feedback influences organizational behaviors, such as
organizational change and risk-taking behavior (Audia & Greve, 2006; Cyert & March, 1963;
Greve, 2003; March & Shapira, 1987; Shinkle, 2012).
Although much research has been done on the relationship between performance
feedback and organizational behaviors since Cyert & March (1963) first introduced it, there
still is an ongoing debate on the relationship between them. Specifically, when it comes to
performance below the aspiration level and the effect it has on organizational behaviors. Cyert
& March (1963) stress that when performance is below the aspiration level, decision-makers
are more likely to engage in problemistic search to instigate change with the aim of improving
organizational performance. Over the past decades scholars have identified a range of
behaviors that were influenced by these so-called performance shortfalls (Audia & Greve,
2006; Desai, 2008; Kuusela, Keil, & Maula, 2017; Shinkle, 2012).
Despite the absence of the concept of risk-taking behavior in A Behavioral Theory of
The Firm by Cyert & March (1963), the stream of literature on performance feedback has
demonstrated that firms sometimes respond with risk-taking behavior as a response to
performance below the aspiration level and sometimes with risk aversion (Audia & Greve,
6 Chen, 2004; Palmer & Wiseman, 1999; Shinkle, 2012; Sitkin & Pablo, 1992; R. M. Wiseman
& Bromiley, 1996). Evidently, there is a mixed empirical picture when it comes to the effects
of performance below the aspiration level on risk-taking behavior. Researchers indicate that
this mixed picture is a result of different conditions that firms are susceptible to (Audia &
Greve, 2006; Desai, 2008; Kim et al., 2015; Kuusela et al., 2017; Lim & McCann, 2014).
Firms are inherently heterogeneous (Desai, 2008), and therefore respond differently to
performance feedback. Consequently, firms sometimes respond with risk aversion and rigidity
and sometimes with risk-taking behavior and organizational change as a response to
performance shortfalls (Audia & Greve, 2006; Desai, 2008).
Clearly, there is a need for the identification of additional boundary conditions on the
theory of performance feedback to help reconcile the opposing views in this debate. A
particularly interesting debate, because it appears that some level of risk-taking behavior is a
requirement for firms to remain competitive, adaptive and to enhance organizational
performance (Desai, 2008).
Some scholars have already identified organizational conditions that influence
decision-makers responses to performance feedback, that partly explain the inconsistent
findings (Audia & Brion, 2007; Audia & Greve, 2006; Greve, 2008; Kacperczyk, Beckman,
& Moliterno, 2015; Kim et al., 2015; Kuusela et al., 2017; Lim & McCann, 2014).
For instance, firm size has been identified as a moderator of the relationship between
performance below the aspiration level and risk-taking behavior (Audia & Greve, 2006).
Another study researches the moderating effect of organizational experience, legitimacy and
age on the mentioned relationship (Desai, 2008). More recently, Lim & McCann (2014)
identified CEO and outside director stock options as a moderator. Many scholars are still
indicating the importance of identifying additional boundary conditions of the original theory
7 2015; Kuusela et al., 2017; Lim & McCann, 2014).
One study in particular (Kuusela et al., 2017), indicated that financial slack can affect
the direction of change in reaction to performance decline. There is not much prior research
on the boundary conditions of performance feedback that incorporates slack. In fact, Kuusela
et al (2017) are one of the first scholars to specifically research the moderating effect of slack
on the relationship between performance below the aspiration level and organizational
change. They indicate that the moderating effects of slack, despite the appearance of slack in
A Behavioral Theory of the Firm (Cyert & March, 1963), have not been fully researched yet.
Kuusela et al. (2017) only focus on the moderating effect of financial slack, indicating that the
level of financial slack attenuates the negative relationship between performance below the
aspiration level and the amount of acquisitions and attenuates the positive relationship
between performance below the aspiration level and the amount of divestments (Kuusela et
al., 2017). However, despite indicating the importance of organizational slack in this stream
of research, no distinction between different types of slack was made. Yet, evidence of prior
research suggests that various types of slack exert different influences on organizational
behaviors, directly but also as moderators (Kraatz & Zajac, 2001; Singh, 1986; Tan & Peng,
2003; Voss, Sirdeshmukh, & Voss, 2008). Prior research on performance feedback and
organizational slack indicates that this field may be more complex than initially thought and
urges scholars to engage in future research in this field (Kuusela et al., 2017). In this paper I
continue the work of Kuusela et al. (2017) by going beyond their study. In doing so, I propose
to make a distinction between different types of slack.
Following prior research, I distinguish between three types of organizational slack;
unabsorbed, absorbed and potential slack (Bourgeois, 1981; Bromiley, 1991; Greve, 2003;
Sharfman, Wolf, Chase, & Tansik, 1988). These types of slack are differentiated based on the
8 The rate to which slack is absorbed in an organization determines the extent to which the
resources are ready to be used for a different alternative or are tied to current operations (Voss
et al., 2008). To reconcile the antagonistic views in this debate, I use the shifting-focus model
of risk-taking (e.g. Audia & Greve, 2006; Desai, 2008; March & Shapira, 1987, 1992). This
study examines the moderating effects of unabsorbed, absorbed and potential slack on the
relationship between performance below the aspiration level and risk-taking behavior. Based
on the shifting-focus model, I argue that unabsorbed and potential slack lower the perception
of threat to the organization when performance declines, because these types of resources are
easier to use for alternative courses of action. Therefore, when a firm is buffered with high
unabsorbed or potential slack an increase in risk-taking behavior among decision-makers is
expected. I argue that firms with higher levels of absorbed slack, engage in less risk-taking.
Absorbed slack is stuck in the firm (Voss et al., 2008), and is therefore more likely to raise the
perception of threat.
This paper makes two main contributions to the existing literature. Firstly, it
contributes to the stream of literature on behavioral theory by extending prior research in this
area and identifying additional moderators of the relationship between performance below the
aspiration level and risk-taking behavior. Consequently, this study enhances the
understanding of the mechanisms of performance feedback and risk-taking behavior. It also
helps to reconcile the two sides in the debate by showing conditions under which these
mechanisms change. Secondly, this paper contributes by showing a more in-depth research
into the effects of different kinds of organizational slack. Specifically, this paper contributes
to a richer understanding of unabsorbed, absorbed and potential slack in the context of
performance feedback.
The following structure is adopted in this paper. Firstly, in the second chapter the
9 feedback in general and its connection to risk-taking behavior. Then the debate, and the need
to identify organizational conditions that influence the aforementioned relationship are further
explained. Subsequently, the concept of organizational slack, its different forms and its
connection to performance feedback is described, which leads to the different hypotheses. The
methodology is outlined in chapter three, which consists of the sample, the operationalization
of the concepts and a description of the statistical model used in this paper. Afterwards, the
results are shown in chapter four and discussed and interpreted in chapter five. Chapter five
also indicates the implications of this study for both theory and practice and provides
limitations of this study along with suggestions for future research. Lastly, the paper ends
with a conclusion.
2. Theory and Hypotheses
In this chapter of the paper, I first describe the roots of the theory on performance feedback
and its connection with risk-taking behavior. Secondly, I describe the ongoing debate in the
literature on performance below the aspiration level in-depth and elaborate on work that has
been done in this stream of literature up until now. Thirdly, I introduce organizational slack,
unabsorbed, absorbed and potential, its connection with the debate and the possible
moderating influence it can exercise on the relationship between performance decline and
risk-taking behavior. Then, I integrate the theoretical concepts described above. By bringing
the theories together and articulating my own interpretation, I derive four hypotheses from the
literature. Lastly, I provide an overview of the proposed hypotheses in a conceptual model.
2.1. Performance feedback and organizational risk-taking behavior
Originally, the stream of research on performance feedback initiated from A Behavioral
10 three views that attempt to explain the relationship between these aspirations and
organizational outcomes, each with their own mechanisms. One of which is behavioral theory,
founded on some of the principles of the Carnegie School, such as adaptive learning,
satisficing, sequential attention, search and bounded rationality (Shinkle, 2012).
Behavioral theory is founded on the assumption that decision-making drives
administration and humans are the ones making the decisions. Importantly however, humans
are limited in their cognitive abilities (Shinkle, 2012). In other words, there is bounded
rationality. Precisely the limited cognitive ability of human beings is the reason
decision-makers use aspirations to evaluate their performance and base their decisions on. This helps to
understand the relevancy of performance feedback and also captures the behavioral character
of firms (Gavetti et al., 2012).
The main argument of performance feedback is that setting targets makes performance
comparable to an organization's past performance or to that of other firms in the same
industry (Cyert & March, 1963; Desai, 2008). Since decision-makers of firms are boundedly
rational, they require these aspirations to evaluate if the current performance is satisfactory
(Desai, 2008). In other words, they receive feedback to make sense of past performance and
determine the strategic behavior of the firm (Shinkle, 2012).
With regard to aspirations, decision-makers can compare their current performance to
their past performance, in which they make use of an historical aspiration level. They can also
compare their current performance to that of other firms in the same industry, which is a
social aspiration level (Audia & Greve, 2006; Cyert & March, 1963; Desai, 2008; Gavetti et
al., 2007; Greve, 2003; Shinkle, 2012). According to the behavioral perspective of the firm,
organizations learn from these performance feedback mechanisms. It allows decision-makers
to determine whether or not the firm performs adequately and provides information to give
11 words, performance feedback aids decision-makers in figuring out what action to undertake
next.
Behavioral theory illustrates a mechanism which is called problemistic search. This
process is best described as ''search that is stimulated by a problem ... and is directed toward
finding a solution for that problem'' (Cyert & March, 1963, p. 121). The general idea is that
whenever performance is above the aspiration level, there is no need for problemistic search.
Whenever performance is below the aspiration level, there is (Gavetti et al., 2007). In that
sense, the original theory by Cyert & March (1963) states that when firm performance
exceeds the aspiration level, decision-makers do not make an attempt to change the existing
routines (Desai, 2008; Miller & Chen, 2004; R. M. Wiseman & Bromiley, 1996). In contrast,
whenever a firm performs below an aspiration level, the decision-makers of the firm do
engage in problemistic search. This means that decision-makers attempt to find alternative
courses of action that can improve organizational performance (Greve, 2008).
The original behavioral theory by Cyert & March (1963) only indicates that firms
engage in problemistic search to find alternative courses of action to improve performance.
The theory does not specify what type of action decision-makers undertake. Other theories
that can accompany behavioral theory, do predict this. Therefore, many researches in the field
of performance feedback have incorporated for example prospect theory, self-enhancement
theory or threat rigidity theory (Audia & Greve, 2006; Chen & Miller, 2007; Iyer & Miller,
2008; Jordan & Audia, 2012; Kahneman & Tversky, 1979; Miller & Chen, 2004).
The findings of these studies, incorporating several of the aforementioned theories,
found that decision-makers often engage in alternative courses of action that involve new
routines or new strategic approaches. Examples of this are, engaging in innovation or in
acquisitions (Audia & Greve, 2006; Iyer & Miller, 2008; Jordan & Audia, 2012; Miller &
12 that decision-makers become rigid and risk averse when performance declines and are more
likely to stick to their current operations (Sitkin & Pablo, 1992; Staw, Sandelands, & Dutton,
1981).
As mentioned before, even though there was no mention of risk-taking in A
Behavioral Theory of the Firm (Cyert & March, 1963), a substantial amount of research on
performance feedback has found that performance below the aspiration level influences the
amount of risk-taking behavior (Argote & Greve, 2007; Audia & Brion, 2007; Audia &
Greve, 2006; Bromiley, 1991; Chen & Miller, 2007; Desai, 2008; Gavetti et al., 2012, 2007;
Greve, 1998; Iyer & Miller, 2008; Kacperczyk et al., 2015; Kim et al., 2015; Kuusela et al.,
2017; Lim & McCann, 2014; March & Shapira, 1987, 1992; McKinley, Latham, & Braun,
2014; Miller & Leiblein, 1996; Mone, Mckinley, & Barker, 1998; Palmer & Wiseman, 1999;
Shinkle, 2012; Singh, 1986; Sitkin & Pablo, 1992; R. M. Wiseman & Bromiley, 1996; Robert
M. Wiseman & Catanach, 1997).
In contrast, many of these studies found that when performance rises above the
aspiration level, decision-makers are less inclined to take risk (March & Shapira, 1987).
Decision-makers have no need to change existing routines, which involves risk, because that
might jeopardize current performance (Bromiley, 1991; Greve, 2003).
However, when it comes to performance below the aspiration level, the empirical
findings in the literature appear to be inconsistent. Some studies have found that
decision-makers are more inclined to engage in risky actions to overcome the decline in performance
(Audia & Greve, 2006; Boyle & Shapira, 2012; Bromiley, 1991; Desai, 2008; Lim &
McCann, 2014; Miller & Chen, 2004; Palmer & Wiseman, 1999). These findings are in line
with behavioral theory and prospect theory (Audia & Greve, 2006; Desai, 2008; Kahneman &
Tversky, 1979; Lim & McCann, 2014). Other studies have found evidence to the contrary,
13 performance below the aspiration level (Audia & Greve, 2006; Greve, 2003; McKinley et al.,
2014; Sitkin & Pablo, 1992; R. M. Wiseman & Bromiley, 1996). These findings are not in
line with the behavioral theory of the firm and the prospect theory, but they do coincide with
threat rigidity theory (Audia & Greve, 2006; Staw et al., 1981).
Altogether, behavioral theory and prospect theory suggest that performance below the
aspiration level triggers a search for solutions causes risk-taking behavior and change for
these firms. Prior research shows evidence of these theories. However, prior research also
indicates that sometimes performance below the aspiration might lead to risk-aversion and a
rigid response. This is not in line with behavioral theory and prospect theory, but is in line
with the threat rigidity theory.
2.2. Literature gap
As described before, a significant number of scholars have investigated the effects of
performance below the aspiration level on risk-taking behavior. Yet, the results appear to be
inconsistent. Some scholars have found evidence for risk-taking behavior when performance
falls below the aspiration level (Audia & Greve, 2006; Boyle & Shapira, 2012; Bromiley,
1991; Miller & Chen, 2004; Palmer & Wiseman, 1999; R. M. Wiseman & Bromiley, 1996).
In other words, these scholars found evidence that decision-makers engage in a riskier action
whenever performance decreases below the aspiration level. These findings complement
behavioral theory and prospect theory (Cyert & March, 1963; Kahneman & Tversky, 1979).
Other scholars have found evidence for risk aversion whenever performance falls below the
aspiration level. (Audia & Greve, 2006; Greve, 2003; Sitkin & Pablo, 1992; R. M. Wiseman
& Bromiley, 1996). This means that decision-makers are likely to respond with a non-risky
action, without changing established routines. This evidence is in line with the argumentation
14 The logical explanation of these conflicting perspectives is as follows.
Decision-makers of firms that underperform, believe that by allocating resources to activities, that
involves a certain amount of risk, it can enhance future performance (Desai, 2008; Greve,
1998; Singh, 1986; R. M. Wiseman & Bromiley, 1996). While, scholars on the other side of
the debate argue that ''low performance is perceived not as a repairable gap but rather as a
threat to organizational survival or other valued outcomes'' (Desai, 2008, p. 595).
Consequently, when faced with performance below the aspiration level, decision-makers tend
to focus on their usual operations and responses that they are familiar with. Managers do not
opt for risky decisions, since they believe it would bring further threat to the continuity of the
firm (Audia & Greve, 2006; Desai, 2008). Evidently, the search for alternative strategic
actions that can be undertaken is constraint by this process. Managers respond risk aversely
and also in a more rigid way (McKinley et al., 2014).
To reconcile these opposing perspectives in this debate, the shifting-focus model is
used. Other scholars in this particular field have used the same model for reconciliation
purposes (e.g. Audia & Greve, 2006; Desai, 2008; Kim et al., 2015; Lim & McCann, 2014).
The shifting-focus model of risk -taking, developed by March & Shapira (1992), is relevant
and important because the focal point of decision-makers determines their response to
performance feedback (Audia & Greve, 2006; March & Shapira, 1992). In other words,
March & Shapira (1992) argue that whether decision-makers respond with risk-taking
behavior or with risk aversion depends on the focus of their attention. According to the
shifting-focus model, managers do not focus their attention on one specific target, but instead
focus on either the aspiration level or the survival point (Audia & Greve, 2006). This goes
against the behavioral theory of the firm and the prospect theory, since they indicate that
decision-makers only focus on one target, namely the aspiration level (Audia & Greve, 2006;
15 performance below the aspiration level but above the survival level. In which, reaching the
survival point means the organization goes bankrupt. In the first scenario decision-makers
focus on the survival point (Audia & Greve, 2006; March & Shapira, 1992). By focusing on
the survival point, decision-makers interpret a decrease in performance as a threat because it
means getting closer to failure. According to March & Shapira (1992), this scenario leads to
risk aversion because of two reasons. Either, managers make low-risk decisions on purpose to
avoid failure of the firm, or managers are unable to make a risky decision because of the
perception of threat (Audia & Greve, 2006; March & Shapira, 1992).
In the second scenario, the focus is on the aspiration level rather than on the survival
point, since there is little or no perception of threat to the survival of the organization. This
leads to risk-taking behavior because managers believe that the firm can overcome
performance decreases if they opt for riskier decisions (Audia & Greve, 2006; March &
Shapira, 1992). The attention of managers shifts between either the survival point or the
aspiration level according to different rules (March & Shapira, 1992). These rules determine
the perception of threat to the organization and therefore determine on what points they focus
their attention.
Over the past years scholars in this field of research have identified some of these
rules, otherwise known as determinants, factors or organizational conditions (Audia & Greve,
2006; Desai, 2008). Importantly, firms are heterogeneous, which means they are susceptible
to different conditions (Audia & Greve, 2006; Desai, 2008). Therefore, different rules or
factors apply to different organizations. This is an explanation for the conflicting findings of
past research on performance feedback. As I mentioned before, scholars have initiated a
search for conditions under which performance below the aspiration level might lead to higher
or lower levels of risk-taking behavior (Audia & Greve, 2006; Desai, 2008; Kim et al., 2015;
16 influences below, that already have been investigated, to gain a deeper understanding of
performance feedback and its outcomes.
Audia & Greve (2006) attempt to address this gap by adding a moderator that
influences the relationship between performance below the aspiration level and risk-taking
behavior. They have found evidence to support risk-taking behavior when performance falls
below the aspiration level with firm size as a moderator (Audia & Greve, 2006). In which
small firms are more likely to be risk-averse, when performance falls below the aspiration
level, than large firms. Other, more recent studies, have also made an attempt to identify
boundary conditions (Kim et al., 2015). Kim et al. (2015) indicate that different aspirations,
historical and social, have differing effects on risk-taking behavior. Another recent study
found that CEO and outside director stock options moderate the relationship between
performance shortfalls and risk-taking behavior (Lim & McCann, 2014). When performance
is below the aspiration level, high values of option grants increases risk-taking behavior of
outside directors and increases risk aversion of CEOs (Lim & McCann, 2014). Desai (2008)
found that lower levels of experience and legitimacy lead to higher levels of risk-taking when
performance is poor. Additionally, although the effect is weaker, older firms are less likely to
take risk when performance declines.
Similarly, Kuusela et al. (2017) identified boundary conditions of the theory on
performance feedback. They have found evidence for the moderating effect of financial slack
on the relationship between low performance and different types of organizational change.
Here, the focus was either on divestments (resource-freeing change) or on acquisitions
(resource-consuming change) and how these are affected by performance below the aspiration
level. Aside from finding evidence to support slack's moderating effect, the study also
17 Consequently, some boundary conditions have already been identified. Yet, many
scholars point out that the field of research on performance feedback may be more complex
than initially thought. This field has a lot of potential for future research, particularly with
regard to finding and empirically testing more conditions under which performance below the
aspiration level lead to risk-taking behavior or risk aversion (Audia & Greve, 2006; Desai,
2008; Kuusela et al., 2017; Lim & McCann, 2014; Shinkle, 2012).
2.3. Performance feedback and organizational slack
As Kuusela et al. (2017) point out, despite the first appearance of organizational slack in A
Behavioral Theory of the Firm (Cyert & March, 1963) to date the moderating effects of this
concept are not yet thoroughly researched (Kuusela et al., 2017). Specifically, in the context
of performance decline and risk-taking behavior. Prior research shows evidence that the
nature of organizational slack influences the relationship between performance feedback and
risk-taking behavior and change (Audia & Greve, 2006; Kuusela et al., 2017). However, prior
research missed an important consideration when it comes to slack. Kuusela et al. (2017) have
only researched the moderating effects of financial slack, but have not made a distinction
between different types of slack. Consequently, to what extent different kinds of
organizational slack have a moderating influence on the relationship between performance
shortfalls and risk-taking behavior, has yet to be researched. The logic that these different
types of slack might have different moderating effects, lies in the fact that slack can differ in
the way the resources can be redeployed or recovered (Bourgeois, 1981). Firms with higher
levels of some type of slack might find it easier to engage in risk-taking behavior, because a
certain buffer of slack resources is easier to redeploy than with other types of slack (Voss et
18 In this paper, organizational slack in general can be seen as the availability of marginal
resources (Wiseman & Bromiley, 1996; Wiseman & Catanach, 1997). Organizational slack is
a widely used concept (Bourgeois, 1981). According to Bourgeois (1981, p. 30) the following
definition best fits organizational slack:
Organizational slack is that cushion of actual or potential resources which allows an
organization to adapt successfully to internal pressures for adjustment or to external
pressures for change in policy, as well as to initiate changes in strategy with respect to
the external environment.
Organizational slack includes resources, such as over-qualified personnel, undiscovered
improvements in current technologies and over-designed equipment (Levinthal & March,
1981). Slack is basically a stock of excess resources (Voss et al., 2008). According to Voss et
al. (2008) slack accumulates overtime because of different reasons. It can be because of a
planned buffer, inadequate planning or because of organizational performance in preceding
periods.
As described above, since organizational slack differs in the way the resources can be
redeployed or recovered, scholars have identified three different kinds of slack (Bourgeois,
1981). These types are unabsorbed, absorbed and potential slack (Bourgeois, 1981; Singh,
1986). The definitions of these types of slack depicted in Greve (2003) are adopted in this
paper. Unabsorbed slack consists of financial reserves ''which an organization can maintain by
holding cash of financial instruments (unabsorbed slack)'' (Greve, 2003, p. 5). Unabsorbed
slack is financial in nature, such as cash that is not ''absorbed'' in a part of the organization.
Absorbed slack ''exists as a use of administrative resources beyond what is necessary for the
19 absorbed slack are; R&D facilities, extra time for staff members to engage in development
(Greve, 2003). In past research unabsorbed slack has also been labeled as high-discretion
slack and absorbed as low-discretion slack (Nohria & Gulati, 1996; Sharfman et al., 1988). In
which discretion refers to the extent to which the resources are easy to recover, where
high-discretion means the slack is easy to recover (Mishina, Pollock, & Porac, 2004; Sharfman et
al., 1988). Lastly, potential slack refers to the amount of resources a firm can lend in the
future (Greve, 2003). This is for example new debt. In more recent research, potential slack
has also been described as high-discretion slack (Mishina et al., 2004; Sharfman et al., 1988).
In light of the debate, and the use of the shifting-focus model to reconcile opposing
views in the debate, it is important to stress why unabsorbed, absorbed and potential slack
each have different moderating effects on the relationship between performance below the
aspiration level and risk-taking behavior. Even though the specific mechanisms of these
effects and how they differ are described below in the hypotheses section, it is necessary to
stress the underlying logic why different slack types moderate differently.
Evidently, as for example Audia & Greve (2006) have pointed out, stocks of resources
impact the perception of managers with regard to the position of the survival point. They
indicate that ''a large stock of resources lowers the performance level at which the
organization's survival is in danger (i.e., the survival point)'' (Audia & Greve, 2006, p. 86). In
contrast, whenever there are limited resources, it raises the survival point of firms.
Consequently, this means that decision-makers of firms with large amounts of resources
engage in riskier actions and vice versa (Audia & Greve, 2006).
However, intuitively it is understandable that it makes a difference whether the
resource stock is immediately available or is available in the shape of an idle machine that
cannot be redeployed for anything other than what it is made to do. For example, usually
20 these resources are trapped in the organization, it might make decision-makers risk averse
since these kinds of slack resources cannot help them (Audia & Greve, 2006; Nohria &
Gulati, 1996; Sharfman et al., 1988; Tan & Peng, 2003; Voss et al., 2008). In other words, the
logic behind this reasoning is that the specific characteristics of slack resources influence the
way decision-makers respond to performance shortfalls.
2.4. Research question
Overall, the main aim of this study is to reconcile contradictory views in the debate on
performance below the aspiration level and risk-taking behavior. In order to do so, this paper
contributes by identifying boundary conditions of the relationship between performance
below the aspiration level and risk-taking behavior. This will further resolve the ongoing
debate in the literature and increase the understanding of the conditions firms are susceptible
to. Evidence of prior research suggests that different types of slack, namely unabsorbed,
absorbed an potential slack may moderate the relationship between performance shortfalls and
risk-taking behavior differently (Nohria & Gulati, 1996; Sharfman et al., 1988; Tan & Peng,
2003; Voss et al., 2008). To understand the process of performance feedback, particularly the
relationship between performance decline and organizational risk-taking behavior under
conditions of unabsorbed, absorbed and potential slack, means that the following research
question needs to be answered:
What are the moderating effects of unabsorbed, absorbed and potential slack on the relationship between performance decreases below the aspiration level and risk-taking behavior?
21 2.5. Performance below the aspiration level and organizational risk-taking behavior As discussed in the sections above, whenever a firm performs below the aspiration level either
one of two things can occur regarding risk-taking behavior and organizational change.
According to behavioral theory, it is likely that decision-makers initiate a search for
alternative strategic actions that can improve organizational performance (Cyert & March,
1963). This concept of ''problemistic search'' is the driving mechanism of organizational
change and prior research has already indicated that this search involves risk-taking behavior
(Audia & Brion, 2007; Audia & Greve, 2006; Bromiley, 1991; Greve, 2003, 2008;
Kacperczyk et al., 2015; Kim et al., 2015; Kuusela et al., 2017; Lim & McCann, 2014; Mone
et al., 1998; R. M. Wiseman & Bromiley, 1996; Robert M. Wiseman & Catanach, 1997).
Similarly, prospect theory predicts an increase in risk-taking behavior when
performance declines (Kahneman & Tversky, 1979). In that sense, it is logical to expect a
negative relationship between performance below the aspiration level and risk-taking
behavior in this study as well. In which a negative relationship indicates that when
performance below the aspiration decreases, risk-taking behavior increases (Audia & Greve,
2006; Lim & McCann, 2014).
However, as pointed out before, according to the threat rigidity theory firms
sometimes react with risk aversion as a response to performance shortfalls (Audia & Greve,
2006; Kahneman & Tversky, 1979). Furthermore, Audia & Greve (2006) have attempted to
address this conflict as well. Their study indicated that the risk-aversion hypothesis was
confirmed. Yet, they also stress that conflicting past results may stem from differences in
samples. A sample focusing only on firms near bankruptcy might indicate risk-taking as a
response, since doing nothing (responding risk aversely) will lead to inevitable failure of the
organization (Audia & Greve, 2006; Ketchen & Palmer, 1999). Altogether, as prior research
22 to test the particular relationship between performance shortfalls and risk-taking behavior
with this sample. I argue that decision-makers engage in more risk-taking behavior when
performance declines, which is consistent with behavioral theory and prospect theory.
Therefore, I propose the following Hypothesis.
H1a: Performance below the aspiration level and risk-taking behavior are negatively related.
However, conflicting evidence has been found in the past and threat-rigidity theory dictates
that decision-makers respond in a risk averse manner. So, even though I propose that the
findings will be in line with behavioral theory, I set up an alternative Hypothesis that supports
the threat-rigidity theory.
H1b: Performance below the aspiration level and risk-taking behavior are positively related.
2.6. Unabsorbed slack as a moderator
Unabsorbed slack, as opposed to absorbed slack, does not constrain an organization (Voss et
al., 2008). It is the easiest of the three types of slack to redeploy into a new project (Sharfman
et al., 1988; Tan & Peng, 2003; Voss et al., 2008). Greve (2003) indicates that higher levels of
unabsorbed slack also make it more relaxed for decision-makers to engage in R&D projects,
because uncertainty plays a less important role. As pointed out by past research, higher levels
of slack allow decision-makers to respond more risky to the environment and it also allows
firms to experiment (Bourgeois, 1981; Hambrick & Snow, 1977). I argue that this is
particularly the case with the unabsorbed dimension of organizational slack. Given that this is
the most flexible of the three, it is likely that unabsorbed slack amplifies the relationship
between performance below the aspiration level and risk-taking behavior.
23 by March & Shapira (1992). As soon as performance drops below the aspiration level,
decision-makers can either engage in risk-taking behavior or risk aversion. As explained
above, decision-makers respond in a risk averse manner when faced with the perception of
threat or because choosing a risky alternative might bring further harm to the firm (Audia &
Greve, 2006). They respond in a risk-taking way when they see the decline as a repairable
gap. These mechanism are influenced by the amount of unabsorbed slack.
Since unabsorbed slack can be used for all kinds of purposes and is readily available, it
lowers the perception of threat and therefore lowers the survival points of the organization.
This means that decision-makers interpret the failure of the organization to be further away
(Audia & Greve, 2006; March & Shapira, 1987, 1992). Consequently, this leads to
decision-makers focusing on the aspiration level rather than on the survival point. In other words,
unabsorbed slack acts as a buffer and creates the perception that decision-makers can
overcome the performance shortfall by engaging in riskier decisions. Moreover, higher levels
of unabsorbed slack can also relieve the actual threat to the firm, since slack can be seen as
''internal shock absorbers'' (Bourgeois, 1981, p. 30)
Furthermore, Audia & Greve (2006) indicate that firms with a larger buffer are more
likely to respond with risk-taking behavior. This indicates that resources support firms to
respond in a riskier manner. I argue that this is specifically the case for effortlessly available
resources. Thus, I believe that the nature of unabsorbed slack allows decision-makers of firms
to engage in risk-taking behavior when firms are faced with performance decline. The
resources can be easily relocated to another place in the organization (Sharfman et al., 1988;
Tan & Peng, 2003; Voss et al., 2008), which means that these resources simply allow
decision-makers to take risky actions instead of restraining them. Decision-makers can
explore new options, without perceiving it as threat to the continuity of the organization. So,
24 with the aim of improving performance and reaching the aspiration level again. This leads to
the following hypothesis:
H2: Unabsorbed slack strengthens the negative relationship between performance below the aspiration level and risk-taking behavior.
2.7. Absorbed slack as a moderator
In contrast to unabsorbed slack, when slack is more absorbed in nature, it does constrain
organizations. In other words, organizations are more prone to stick to their current course of
action under conditions of absorbed slack (Voss et al., 2008). Voss et al. (2008) indicate that
when the rate of absorption is high, firms are also less likely to engage in exploratory
activities. When it comes to absorbed slack, Voss et al. (2008) indicate that these resources
are tied to the current operation. There are structural constraints of these resources that makes
them less- or even unavailable to redeploy for innovation purposes (Greve, 2003; Sharfman et
al., 1988; Tan & Peng, 2003; Voss et al., 2008). Absorbed resources are not expected to allow
organizations to search for new alternatives in the face of performance shortfalls. In fact,
sometimes slack can even be seen as (sunk) costs (Miller & Leiblein, 1996; Nohria & Gulati,
1996; Zinn & Flood, 2009). I argue that this applies to the absorbed dimension of slack.
The moderating effect of absorbed slack on the relationship between performance
below the aspiration level and risk-taking behavior make more sense when applied to the
shifting-focus model. Generally speaking, resources should allow managers to see
performance shortfalls as a gap that can be fixed through risky action or investment (Audia &
Greve, 2006). At least, that is the buffering effect that Audia & Greve (2006) described when
they measured firm size as a moderator of the aforementioned relationship. However, I argue
that this is not the case with absorbed slack. Even when there is slack available,
25 reason for this is that when performance declines, the perception of threat will increase for
firms with high levels of absorbed slack. Mainly, because absorbed slack is difficult to
withdraw from the firm and is not easy to use for different purposes (Bourgeois, 1981; Tan &
Peng, 2003; Voss et al., 2008). Consequently, decision-makers feel like they should be careful
with their activities and investments and are likely to respond in a risk averse manner to
prevent damaging the survival of the firm.
The second reason is that decision-makers cannot afford to experiment, because there
is no cash available. In fact, it may even jeopardize current operations if absorbed slack is
used for different purposes (Voss et al., 2008). Therefore, under high levels of absorbed slack,
decision-makers are more inclined to focus on the survival point rather than on the aspiration
level. Thus, because of the restricting and constraining nature of absorbed slack, it is expected
that decision-makers do not engage in risk-taking behavior when faced with performance
shortfalls. Therefore, I propose the following hypothesis.
H3: Absorbed slack weakens the negative relationship between performance below the aspiration level and risk-taking behavior.
2.8. Potential slack as a moderator
Potential slack is different from the other two forms of slack, in that it is not yet a resource of
the firm. In fact, it is the ability of a firm to gain external resources (Bromiley, 1991). This
can be done by raising new equity or debt. Potential slack can also be described as leverage
(Bourgeois, 1981; Greve, 2003; Iyer & Miller, 2008; Kuusela et al., 2017). One of the
advantages of potential slack is its flexible nature. Similar to unabsorbed slack, potential slack
does not constrain organizations (Kuusela et al., 2017). Since the resources are not
immediately available, they are also not committed to any ongoing exploitative operations.
26 resources must be secured first. It appears that potential slack does not directly help the
process of innovation, which can be seen as a risky way to increase performance, but it does
play a role in the decisions whether to continue in R&D projects or not (Greve, 2003). Prior
research does indicate a moderating effect of potential slack on the relationship between
performance decline and the rate of acquisitions (Iyer & Miller, 2008). Logically, when
performance declines, less resources are available and decision-makers are inclined to search
for resources outside the organization. Therefore, having potential slack might assist in this
process.
Like the arguments for unabsorbed and absorbed slack, the underlying logic when it
comes to the moderating effects of potential slack is best understood through the application
of the shifting-focus model. Since potential slack plays a comparable role to unabsorbed slack
in the organization, it makes sense that potential slack also lowers the survival point for
decision-makers. In other words, when there is a lot of potential slack, decision-makers
perceive performance shortfalls as less of a threat than under conditions of low potential
slack. Therefore, instead of focusing on the survival point and perceiving the performance
shortfall as a threat to the continuity of the firm, decision-makers focus on the aspiration level.
Furthermore, high levels of potential slack make it easier to obtain additional resources. In
that case, making risky decisions does not bring further damage to the firm. This also lowers
the survival point of the firm, prompting managers to focus on the aspiration level.
As described before, by focusing on the aspiration level, decision-maker engage in
risky actions to overcome the decline in performance (Audia & Greve, 2006; March &
Shapira, 1992). However, as indicated before, in the case of potential slack, the resources do
have to be secured first. Therefore, the effect of potential slack is presumably not as strong as
27 evidence of prior studies and the arguments presented above, I propose the following
hypothesis.
H4: Potential slack strengthens the negative relationship between performance below the aspiration level and risk-taking behavior
3. Methodology
In this chapter the research design of this study is explained. Firstly, the sample of this study
is explained. Then, the operationalization of the control, the independent, the dependent and
the three moderating variables is discussed. Lastly, I explain the models that are used to
analyze the data.
3.1. Sample
This study uses secondary data gathered from the Wharton Compustat database. The
financial data available in this database is used to measure the variables. The database consists
of observations of firms from the manufacturing industry with SIC codes from 2000 to 3999.
This ensures a relatively large sample, but the industries within the sample are still
28 2007; Iyer & Miller, 2008). Using the SIC codes from the manufacturing industry allows for
comparison to prior research in this field (Lim & McCann, 2014). Furthermore, the database
is comprised of observations from 1980 until 2014. The original sample consists of 81,635
observations, but after the cleaning data and excluding the missing values, there are 40,675
observations left. Table 1 in chapter 4 shows the industry frequencies across the sample.
3.2. Dependent variable
The dependent variable in this study is taking behavior. To measure organizational
risk-taking behavior, I use R&D intensity as a proxy. This is consistent with prior research (Chen
& Miller, 2007; Greve, 2003; Kacperczyk et al., 2015; Lim & McCann, 2014). R&D intensity
is calculated by dividing R&D expenditures by sales. Additionally, investing in R&D
expenditures is uncertain and risky, because the payoffs are not immediate or sometimes
non-existent altogether (Lim & McCann, 2014; Palmer & Wiseman, 1999). According to Chen &
Miller (2007) the behavioral theory of the firm indicates that the theoretical in this paper only
apply to firms with regular R&D and sales activities. Therefore, similar to other studies, I
excluded firm observations with an R&D intensity less than or equal to 1.0 and more than or
equal to 0 (Chen & Miller, 2007; Lim & McCann, 2014). This ensures that R&D specialists,
where R&D expenditures exceed sales expenditures, are excluded. Table 1 in the results
section indicates a full overview of industry frequencies in the sample.
3.3. Independent variables
The independent variable in this study is firm performance. To measure performance, a
popular accounting measure of performance is used in this study. This is pre-tax income
divided by total assets (ROA). ROA is the most frequently used measure of performance in
the literature on performance feedback, therefore the use of this measure is consistent with
29 2003; Kuusela et al., 2017; Lim & McCann, 2014). Following past studies, this paper includes
both the historical and the social aspiration level in the overall aspiration level (Greve, 2003).
The mean ROA of a firm's past three years is used to calculate the historical aspiration level.
The mean ROA of the other firms in the sample is used to calculate the social aspiration level.
Note, that these performance variables are measured at year t-1, which means they were
lagged by one year (Lim & McCann, 2014), since performance precedes R&D.
Then, to measure performance feedback, the difference between a firm's actual
performance and the historical and social aspiration level are calculated. This means for the
historical performance feedback ROA (performance) - ROA mean of the last three years of
the same firm (historical aspiration level). For the social performance feedback this means,
ROA (performance) - ROA mean of the other firms in the same industry (social aspiration).
Subsequently, I create spline variables, one for performance below the aspiration level and
one for performance above the aspiration level (Desai, 2008; Greve, 1998). In other words,
within the variable ROA, I use performance feedback above 0 as an indication that
performance is above the aspiration level and performance below 0 as an indication that
performance is below the aspiration level. Additionally, I include performance above
aspiration level for a full specification of the model (Desai, 2008)
3.4. Moderating variables
The moderating variables in this study are unabsorbed, absorbed and potential slack. Similar
to past studies I measure the three variables in the following way. Unabsorbed slack is
measured by calculating the ratio of quick assets to liabilities (Bromiley, 1991; Greve, 2003).
Basically, more cash on hand means more unabsorbed slack, since that is the most flexible
kind of slack. Absorbed slack is measured as the ratio of selling, general and administrative
expenses to sales (Bromiley, 1991; Greve, 2003). Essentially, this operationalization measures
30 are absorbed in the organization and cannot easily be redeployed to different parts of the firm.
Lastly, potential slack is calculated as the ratio of debt to equity (Bromiley, 1991; Greve,
2003), because this ratio indicates the amount of leeway a firm has to gain additional debt. All
three measures are accepted proxies of slack and have been used in past research (Bourgeois,
1981; Singh, 1986).
Additionally, according to Greve (2003, p. 691) these terms measure ''excess resources
without adjusting for the normal resource requirement for a given business''. Therefore, a
study should either control for firm effects or research one particular industry. In this study,
firm effects are taken into account.
3.5. Control variables
To make sure that the results are not influenced by other variables that are known to influence
risk-taking behavior, I control for the following variables.
Firm size. Firm size is the first control variable I consider, since this variable might
influence organizational change and risk-taking behavior (Audia & Greve, 2006; Desai, 2008;
Lim & McCann, 2014). Firm size is measured as the logged number of employees, which is
consistent with other studies (Audia & Greve, 2006; Lim & McCann, 2014). Logging the
number of employees makes this measure more suitable, because then it measures the effect
of a percentage increase in employees instead of an absolute increase (Audia & Greve, 2006).
Distance from bankruptcy. When firms are facing bankruptcy, they tend to be more
risk averse (March & Shapira, 1992). Therefore, following past studies, I also control for this
variable by measuring distance from bankruptcy with the Z-score (Altman, 1983; Chen &
Miller, 2007; Iyer & Miller, 2008; Lim & McCann, 2014; Miller & Chen, 2004).
Industry R&D intensity. I also control for industry R&D intensity, to account for
industry effects. To measure this, I calculated the average of firm R&D intensity in the
31 influence investments decisions that managers make. Therefore, it is included as a control
variable.
3.6. Statistical Model
Since the database consists of panel data, the research needs to account for time series and
cross-section in the choice of a correct estimation model. Firstly, I ran a Breusch-Pagan
Lagrange to check if there are individual firm specific effects. The result of this was
significant (χ2(4) = 80828.45, P < 0.01), indicating that the null Hypothesis is rejected. This
shows that there are indeed individual firm specific effects that need to be taken into
consideration. Therefore, an ordinary least squares (OLS) is not the appropriate regression
model for this study. The generalized least squares (GLS) method is used, which is
appropriate for panel data.
Additionally, for normal regression, the Breusch-Pagan Lagrange test also checks for
time effect, but that does not apply to panel data regression. Therefore, I account for
heteroscedasticity in my regression model, by utilizing robust standard errors against
heteroscedasticity. Then, to determine whether to run this regression with fixed effects or with
random effects, I ran a Hausman specification test. The result of this test was also significant
(χ2
(1) = 345.78, P < 0.01). This indicates that, due to firm effects, running a GLS regression
with random effects significantly differs from running the regression using fixed effects.
Accordingly, I use a (within) fixed effects model to control for these effects. Additionally,
before creating the interaction terms I mean centered the independent and moderating
variables in this study. Doing so, reduces the possibility of multicollinearity (Aiken & West,
1991). When running this regression, I also account for time effects by adding an independent
variable that indicates the fixed effects over the years.
32 4. Results
In this chapter I discuss the results of this study. In the first part of the result section, I present
the distribution of the industries in the sample. Additionally, I provide an overview of the
descriptive statistics and highlight some of them. In the same table, the correlations between
the different variables are presented on the diagonal, of which I also outline the most
important findings. Furthermore, I show the findings of the regression analysis and provide an
overview of the analysis in two tables, one for the historical and one for the social aspiration
level. In the latter part, I also indicate whether the hypotheses are supported or not.
4.1. Sample overview and descriptive statistics
Firstly, the sample is more thoroughly described. Given that this data set consists of many
observations in 20 different industries, it is useful to indicate how the observations are
distributed across the industries. The sample that was used in this study consisted of 40,675
observations. The observations ranged from 1980 until 2010. Industry frequencies are
indicated in Table 1. The sample consisted only of firms in manufacturing industries (SIC
2000-39000). Notably, the largest industry in industry with 9,367 (23.03%) observations is
electronic and other electrical equipment and components (SIC 3600). The second largest
industry in this sample measuring, analyzing and controlling Instruments (SIC 3800), which
consists of 7,688 (18.90%) observations. The third largest industry, with 7,493 (18.42%)
observations, is industrial and commercial machinery and computer equipment (SIC 3500).
The two smallest industries in this sample are tobacco products (SIC 2100) with 60 (0.15%)
observations and leather and leather products (SIC 3100) with 161 (0.40%) observations.
Table 2 shows the descriptive statistics of the variables and the Pearson product-moment
correlation coefficient. The correlations between the dependent variable, R&D intensity, and
33 is negative (r = -0.296). Between R&D intensity and distance from bankruptcy there is a
weak positive correlation (r = 0.046). Furthermore, the correlation between R&D intensity
and Industry R&D intensity is stronger and positive (r = 0.219).
When considering the correlations between the independent variables, Table 1
indicates a strong and significant positive correlation between performance below aspiration
(historical) and performance below the aspiration (social) (r = 0.827). Therefore, following
past research, I separate the models for historical and social aspiration level in the regression
analyses to prevent altering parameter estimates (Chen & Miller, 2007; Iyer & Miller, 2008;
Lim & McCann, 2014).
The correlations between the dependent variable and the moderating variables are
interesting, given that they are not all significant. The correlation between unabsorbed slack
and R&D intensity is not very strong, but positive and significant (r = 0289). R&D intensity
and absorbed slack have a slightly stronger positive correlation, which is also significant (r =
34 R&D intensity (p > 0.10). Additionally, the correlations between the moderating variables and
the independent and control variables should also be considered. Unabsorbed slack is
significantly correlated with all the independent and control variables, except for performance
below the aspiration(historical). That correlation is insignificant. Between unabsorbed slack
and performance below aspiration (social) this correlation is very weak and positive (r =
0.029). Unabsorbed slack is negatively correlated with firm size (r = -0.285), has a stronger
positive correlation with distance from bankruptcy (r = 0.485) and has a weaker positive
correlation with industry R&D intensity (r = 0.079). Absorbed slack has a weak but
significant correlation with both independent variables, performance below aspiration
(historical) (r = -0.142) and performance below aspiration (social) (r = -0.235). There are
weak but significant correlations between absorbed slack and the control variables, firm size
(r = -0.246), distance from bankruptcy (r = -0.037) and industry R&D intensity (r = 0.111).
Lastly, it appeared that potential slack has no significant correlations with any of the control,
independent, dependent or moderating variables. Please refer to Table 2 for a full overview of
36 4.2. Regression analyses
Table 3 and 4 indicate the results of the regression analyses. The first table consists of Models
1 through 6 and the second table of Models 7 through 11. Table 3 and 4 are structured
similarly, the difference is that Table 3 only includes the performance variables measured
with the historical aspiration level and Table 4 only includes the performance variables
measured with the social aspiration level. Additionally, Model 1, which measures the effect of
the control variables of R&D intensity is only included in the Table 3. Model 2 and Model 7
are the baseline models without the interaction terms. They include the dependent variable,
control variables, the three types of slack, the independent variable and performance above
the aspiration level to complete the model. Model 3, 4 and 5 include the same variables as
Model 2 plus each a different interaction term between performance below the aspiration level
and one of the types of slack. As pointed out before, Model 8, 9 and 10 are similar to these
models, but incorporate the social aspiration level. Lastly, Model 6 and 11 include all the
variables and the three interaction terms.
Hypothesis 1a states that when performance below the aspiration level declines,
risk-taking increases. Table 3 shows the results of the regression analysis of the models described
above, using the historical aspiration level. Note that, according to Lim & McCann (2014),
when the coefficient involving performance below the aspiration level is positive, it means
that when performance decreases, R&D intensity decreases as well. Conversely, a negative
coefficient indicates that when performance declines, R&D intensity increases. Model 1
indicates that the control variables explain 13.5% (p < 0.01) of the variance in R&D intensity.
Model 2 explains an additional 10.2 % (p < 0.01) of the variance in R&D intensity compared
to Model 1. Furthermore, Model 2 indicates a significant negative association (b = -0.020, t =
-2.11, p < 0.05) between performance below the aspiration level and R&D intensity.