Performance feedback on risk-taking behavior : does performance feedback lead to more, or less risk-taking behavior?

(1)

Performance feedback on risk-taking behavior:

Does performance feedback lead to more, or less risk-taking

behavior?

Student: Katrine Folkersen/ Student nr: 11390611 MSc. Business Administration, Strategy track

University of Amsterdam, Faculty of Economics and Business Supervisor: MSc. B. Silveira Barbosa Correia Lima

Date: 23th June 2017

Final version submitted

(2)

Page 1 of 45

Statement of originality

This document is written by Student Katrine Folkersen 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 completion of the work, not for the contents.

(3)

Page 2 of 45

Abstract

This paper introduces the effect of different performance feedback models on organizational risk-taking behavior. Moreover, this paper presents two activities associated with different levels of risk, to specify under which conditions, that decision makers will take more (or less) risk. Building on The Behavioral Theory of The Firm (Cyert & March, 1963) and using data on public listed U.S manufacturing firms, this paper argues that firms will increase risk-taking when performance is below an aspiration level. Furthermore, it contributes with the insight that firms invest more in low risk activities compared to high risk activities. Thus, showing that firms take less risk when performance is falling short relative to an aspiration level. Additionally, this paper argues that inconsistency in performance feedback signals affect the level of risk-taking, depending on the managerial interpretation of the exact configuration of the underlying performance feedback discrepancies. Specifically, the result shows that when firms are doing well compared to the previous year, but is being outperformed by their competitors – this will lead to an increase in high risk activities – because decision makers see an opportunity of displacing their competitors and are confident that the firm has the capabilities to do so. For the opposite case of inconsistent performance feedback, the result showed no effect on risk-taking in terms of low risk activities. These findings are mainly consistent with this papers predictions and suggest that risk-taking behavior should be considered at different risk levels, when analyzed in relation to performance feedback theory.

Key words: Performance feedback; inconsistency; risk-taking behavior; the behavioral theory of the firm; R&D; capital expenditures

(5)

Page 4 of 45

Introduction

Building on the Behavioral Theory of The Firm (Cyert & March, 1963), various scholars have demonstrated that organizations learn from experiences to change firm behavior, and enhance organizational performance (Cyert & March, 1963; Levitt & March, 1988; Levinthal & March, 1999; Argote & Miron-Sepctor 2011). Especially, the effect of learning from performance feedback is a central argument that outsprings from the behavioral theory of the firm (Cyert & March, 1963). Performance feedback entails that decision makers use an aspiration level to evaluate their current performance, and furthermore, that the performance relative to aspiration levels will influence organizational changes and risk-taking (Audia & Greve, 2008; Cyert & March 1963; Bromiley, 1991). An aspiration level is defined as the smallest outcome that a decision maker would consider satisfactory (Schneider 1992), and is formed based on past performance and on peers in the industry (Greve, 1998; Shrinkle, 2012; Lant, 1992). Most research on performance feedback and how it affects organizational behavior has particularly investigated organizational risk-taking behavior, which are actions of central interest to strategic management research (Audia & Greve, 2006; Bromiley, 1991). In relation to this, considerable studies have used arguments from Prospect Theory (Kahneman & Tveraky, 1979), which assumes that individuals’ risk preferences vary around a single reference point, which is similar to an aspiration level. Prospect Theory proposes that firms performing, above their aspiration level are risk averse, whereas firms that perform below their aspiration level are risk seeking (Kahneman & Tversky, 1997).

Most studies agree that firms are not motivated to take risk when performance exceeds an aspiration level, however, the effect of changes and risk-taking when performance is below an aspiration level is open to debate (Iyer & Miller, 2008; Bromiley & Wiseman, 1996; Greve, 2003; Greve, 1998). A specific focus is held upon two opposing arguments; One argument

(6)

Page 5 of 45 builds on the prospect theory and the behavioral theory of the firm, suggesting that performance below aspiration levels stimulate the need for improvements, which will encourage decision makers to act risk seeking (Cyert & March 1963; Kahneman & Tversky, 1979), whereas the counter argument builds threat rigidity theory, proposing that short falling performance will increase the awareness of danger and thus leads to risk aversion for decision makers (Staw et al. 1981; Jordan & Audia, 2012).

Existing studies have however tried to reconcile these predictions on risk-taking by arguing that firms may respond conservatively in some situations based in their organizational condition (e.g. Chen & Miller, 2007; Iyer & Miller, 2008). Such studies have focused on the moderating effect of contingencies which has proven that organizations interpret performance feedback in distinct ways, leading to difference in firm’s risk-taking behavior (Lim & MaCann, 2014). For example, Iyer and Miller (2008) investigated distance to bankruptcy as a source of rigidity in relation to the likelihood of making acquisitions. Their result showed that when performance is below an aspiration level, and the closer firms are to bankruptcy, the less likely they are to acquire and take risk. So, even though the behavioral theory of the firm says that decision makers are more risk seeking when performance is falling short, some decision makers might focus more on the survival rather than the growth of the firm (March & Shapira, 1978; Greve, 2008; Iyer & Miller, 2008)

The debate regarding the effect of short falling performance on risk-taking behavior has received substantial attention (March & Shapira, 1987, Bromiley, 1991). However, considerable research has often assumed change behavior to be risky, and thus failed to distinguish between change and risk (Kacperczyk et al. 2015, Greve, 2011). Change can consist of different levels of risk, which may be affected differently by performance feedback (Greve, 2011). Recent research therefore suggests that a richer understanding of how performance

(7)

Page 6 of 45 feedback affects risk-taking behavior, lies in investigating activities associated with different risk levels. This paper will therefore focus on two distinct risky activities to enhance the understanding of whether firms will take more, or less risk when performance is below an aspiration level.

Additionally, this paper enhances the understanding of performance feedback and levels of risk-taking by changing the premise that decision makers evaluate performance relative from one aspiration level, to a premise stating that decision makers can have multiple goals: such as responding to a historical – and a social aspiration level simultaneously. The consequence of responding to multiple goals is however that these might contradict each other which bring inconsistent feedback signals to decision makers (Joseph et al., 2015). While some studies have covered the issue of inconsistent signals by using attention rules to show which aspiration level that a decision makers must pay attention to first (Greve, 1998,2008; Lucas et al. 2012). Thus, following the idea that decision makers are motivated by performance feedback to change organizational behavior (Cyert & March, 1963). It might instead be useful to evaluate both signals simultaneously and shift the focus to the managerial interpretation of these signals (Lucas et al. 2015). Since strategic choice is not only driven my motivation but also by a firm’s presence of opportunities and the capabilities to improve future performance (Lucas et al. 2015; Baum et al., 2005). This paper argues that decision maker’s interpretation on inconsistent performance feedback can affect risk-taking differently based on the firm’s capabilities and opportunities to develop their business.

The major contribution of this paper is, first of all, to provide extensive insight on the relationship between performance feedback and risk-taking behavior introduced by Cyert and

(8)

Page 7 of 45 March (1963). By emphasizing that distinct activities can be associated with different levels of risk, this paper contributes to a better understanding of whether performance below an aspiration level will lead to more, or less organizational risk-taking. Moreover, this paper sheds light on how decision makers can gain additional information by evaluating performance relative to both historical and social aspiration levels simultaneously compared to focusing on either of their aspirations in isolation. The ambiguity related to inconsistent performance feedback signals is decreased as this paper contributes with findings on decision makers interpret inconsistent performance. Specifically, this paper show that decision makers depend their strategic choices, in terms of investing in high or low risk activities, on the presence of opportunities the firm’s capabilities to improve performance.

Literature and hypotheses

The following section discuss relevant literature on performance feedback and risk-taking behavior, which constitute as the basis for this papers hypotheses. First the nature of performance feedback and how it influences risk-taking behavior is presented. Next, the literature gap and this paper’s research question will be addressed. Then, there will be accounted for two risk activities which will be related to the fundamental principles of the behavioral theory of the firm, leading up to the first hypotheses. Subsequently, the relevant literature on inconsistent performance feedback will be outlined. Finally, two inconsistent combinations that affect risk-taking on different levels will be presented together with the remaining hypotheses.

(9)

Page 8 of 45

Performance feedback and risk-taking behavior

Existing literature has proven that learning from experiences has an important impact on organizational behavior (Cyert & March, 1963; Levinthal & March, 1981; Argote & Greve, 2007; Argote & MironSpektor, 2011). Especially, learning from performance feedback has gained substantial attention from various scholars (Greve 1998; Greve 2003; Cyert & March, 1993; Levienthal & March, 1981; Chen & Miller, 2007). Performance Feedback Theory has its out springs from the behavioral theory of the firm, introduced by Cyert & March (1963). The behavioral theory of the firm builds on the assumption that individuals are bounded rational, which means that decision makers have limited capacity to evaluate all available information. Though, in order to overcome these cognitive limitations, decision makers learn from performance feedback (Greve, 1998; Jordan & Audia, 2012; Miller & Chen, 2004). A premise in performance feedback theory is that that decision makers evaluate current performance relative to an aspiration level. An aspiration level is defined as “the smallest outcome that would be deemed satisfactory by the decision maker" (Schneider 1992). Aspirations are formed based on two reference points, which decision makers use to access the effectiveness of their actions (Cyert & March, 1963) A comparison of the firm’s own performance history is known as a historical aspiration level, and additionally, a comparison of the industry peers is known as a social aspiration level (Cyert & March, 1963). Thus, the available information combined with simple decision processing rules is used to create an aspiration level that will cope with future expectations (Greve, 1998). Furthermore, the commitment to achieve a certain aspiration level lies in the discrepancies of the firm’s current performance relative to their aspirations. Decision makers will therefore be motivated to change their organizational behavior depending on how decision makers perceive their performance feedback signals (Cyert & March, 1963; Argote & Greve, 2007; Greve, 1998).

(10)

Page 9 of 45 In relation to this, the behavioral theory poses that when firms will be stimulated to engage in problemistic search behavior when performance is below an aspiration level. That is, searching for solutions to solve the problem of short falling performance (Leventhal & March, 1981). Problemistic search behavior therefore makes the firm generate new activities to improve performance, leading to a reduction of the negative discrepancy. Performance feedback theory does however not predict what kind of commitment, or which specific actions that decision makers must adapt in the attempt to minimize discrepancies between performance and aspirations. Therefore, performance feedback theory must be complemented with different theories to identify alternative solutions that decision makers could incorporate when performance is falling short. Such theories concern among others; prospect theory, self-enhancing theory, threat of rigidity theory, and organizational learning theory (E.g., Greve 2003; Bromiley, 1991; Chen & Miller, 2007; Audia & Greve, 2006; Jordan & Audia, 2012). Following this reasoning, prior research suggests that alternative solutions to low performance are; adoption of new routines and strategies, such as refinements in existing technology, investments in new technologies or engaging in acquisitions (Levinthal & March, 1981; Greve 2003; Chen & Miller, 2007; Iyer & Miller, 2008 Greve, 2006).

When performance exceeds an aspiration level, decision makers are however not induced to engage in problemistic search, because they are satisfied with the existing practice (Greve, 2003). Fhe firm will instead generate slack resources, which means ‘using resources beyond what is necessary for the short-time operation’ (Greve 2003; Bromiley, 1991). This includes among others; overqualified employees, relaxed managerial control and undiscovered efficiency improvements on technology (Levienthal & March, 1981). Subsequently, managers

(11)

Page 10 of 45 are interesting in increasing the organizational performance to go above their aspiration levels, but they are not willing to take risk in order to do so, because they are afraid that these risky actions eventually will result in performance falling below their aspiration level (Kahneman & Tversky, 1979; March & Shapira, 1978; Iyer & Miller, 2008).

Existing literature has additionally argued that performance feedback affects organizational taking (Bromiley, 1991; Audia & Greve, 2006; Miller & Chen, 2004). Organizational risk-taking is related to actions of central interest in strategic management research, which makes it an important discussion point (Bromiley; 1991). Though, scholars have conceptualized risk differently (March & Shapira, 1987; Kahneman & Tversky, 1979; Bromiley, 1991). For example, classical decision theory define risk as “reflecting variation in the distribution of possible outcomes, their likelihoods, and their subjective values” (March and Shapria pp. 1401, 1978; Bromiley, 1991). Meaning that risk is considered when the variance in a firm’s business outcome increases or decreases, because the outcomes becomes unpredictable (Bromiley, 1991). Thus, investments are considered as highly risky when there is a high variance in positive or negative outcomes (Bromiley, 1991). March and Shapira (1978) on the other hand, argue that decision only associate risk with negative outcomes and potential losses, which is defined as downside risk (March & Shapira, 1987). This managerial definition for risk can therefore be described as the likelihood that the result outcome of a decision will fall below the target value. This research adopts the managerial risk definition (March & Shapira, 1978).

A large body of research have adopted the performance feedback theory and extended it with theories of organizational risk-taking (Audia & Greve, 2006; Iyer & Miller; 2008; Greve, 2003). Prospect Theory introduced by Kahneman and Tversky (1979) has for example guided

(12)

Page 11 of 45 much research on risk-taking behavior. The components of prospect theory are similar to the components of the behavioral theory of the firm in three ways: Both theories states that decision makers focus on an aspiration level to evaluate performance feedback. In the behavioral theory, an aspiration level is form by social and historical comparisons, whereas in prospect theory, this aspiration is status quo (called reference point). Secondly, the simplified information obtained by decision makers, is categorized outcomes in terms of success and failure (Kahneman & Tversky, 1979). Accordingly, when performance is above an aspiration level, decision makers will code it as success and when performance is below an aspiration decision makers will code it as failure. Last component assumes that the desire to overcome failure is larger than the desire to overcome success, leading to increased acceptance of risk-taking when performance is below an aspiration level compared to when performance is above (Kahneman & Tversky, 1997; Audia & Greve, 2006; Cyert & March, 1963). Scholars that has been guided by the prospect theory (Kahneman & Tversky, 1979), agree that decision makers are willing to take more risk in making new investments, launching products and adopting new routines, to improve the current short falling performance (e.g., Audia & Greve, 2006; Lim & McCann, 2014; Miller & Chen, 2004; Bromiley, 1991). When performance is above aspiration levels, decision makers are not inclined to take risk because they wish to reduce the chance for falling below their aspiration level (Greve, 2003)

Literature gap and research question

Considerable research has used performance feedback to explain risk-taking behavior (Iyer & Miller, 2008; Bromiley, 1991; March and Shapira,1978; Greve, 2003; Chen & Miller, 2007). Although risk aversion is widely accepted when performance is above aspiration levels, the claims when performance is falling short relative to an aspiration level, on risk behavior, has been controversial in existing literature. Particularly, some studies propose that organizations

(13)

Page 12 of 45 will take greater risk when performance is below as aspiration level (Greve, 2003; Chen & Miller, 2007; Cyert & March, 1963; Kahneman &Tversky, 1979; Gooding et al. 1996), whereas other studies argue for no risk-taking (Staw et al. 1981; Jordan & Audia, 2012). The present conflict in existing literature therefore concerns whether decision makers will act risk seeking or risk averse when performance is below aspirations. Respectively, these opposing perspectives involves two theoretical frameworks that both have supportive evidence through prior studies – the behavioral theory of the firm contra the threat rigidity theory. The behavioral theory poses that low performance will lead to adaptation of routines, problematic search behavior, and risk-taking behavior (Cyert and march, 1963). Threat rigidity theory argues on contrary that low performance leads to a decrease of the ability to process information, centralized decision making, and enhanced organizational rigidity (Staw et al. 1981). One explanation of these conflicting results is that decision makers are facing competing pressures, when performing below aspiration level (March & Shapira, 1992). Short falling performance thus activates preferences for risky alternatives, but the perception of potential crises triggers predictions by threat rigidity theory in terms of cognitive and organizational changes (Cyert & march, 1963; Kahneman & Tversky, 1979; Staw et al.1981; March & Shapira, 1992; Jordan & Audia, 2012).

Supportive evidence that follow the behavioral theory of the firm, suggest that individuals have a stronger desire to overcome performance failure compared to the desire of extending success (Lant, 1992; Fiegenbaum, 1990). Thus, building on the reasoning that decision makers will act risk seeking in terms of initiating activities that can increase the currently short falling performance. Various scholars have provided evidence on this prediction (Fiegenbaum, Hart, & Schendel, 1996; Bromiley, 1991; Gooding et al. 1996; Miller & Chen,

(14)

Page 13 of 45 2004), in which they demonstrate that activities such as adapting new routines, launching new products and making acquisitions will be initiated (Greve, 2003; Bromiley, 1991; Chen & Miller, 2007; Lim & McCann, 2014; Greve 1998). Hence, these predictions challenge the threat rigidity view by stating that decision makers will take risk to improve short falling performance.

However, there is also evidence showing that when performance is below aspiration level, organizations will act risk averse (Wiseman & Bromiley, 1996). In line with threat rigidity theory, Staw et al. (1981) claim that specific tendencies exist for individuals as well as organizations to behave rigid in threatening situations. Specifically, psychological stress and anxiety are major factors caused by threats, which decreases the ability to process information. This makes individuals rely on well-learned or dominant responses and therefore reduces the likelihood of change. Whereas on an organizational level, threats lead to simplification of communication and centralization of control (Staw et al. 1981).

Subsequently, other studies connect to threat rigidity theory by focusing on psychological processes (Jordan & Audia, 2012) and managerial interpretation of low performance as a fundamental threat (March & Shapira, 1992). Particularly, Jordan & Audia (2012) acknowledge the assumptions springing from the behavioral theory by underlying the fact that individuals can be problem-solvers, because they will start searching for solutions as a problem-solving response to the perceived performance. On the other hand, they contribute to the literature by arguing that individuals also have a need to see themselves in a positive light, which is called the ‘self-enhancement motive.’ The self-enhancing mode means that decision makers feel responsible for the low performance and therefore use retrospectively adjusted standards of evaluation - to assess performance as more favorable. The self-enhancing behavior will therefore make decision makers less likely to make organizational changes and take risk, because they are not acknowledging the negative performance feedback. Initially, it

(15)

Page 14 of 45 shows risk averseness and may result in organizational rigidity, because it slows down the process of acting (Jordan & Audia, 2012).

Shapira and March (1992) provides different insight that also build on threat rigidity theory. They argue that decision makers do not only evaluate performance relative to one fixed reference point as stated in prospect theory and the behavioral theory of the firm. Instead they suggest that decision makers evaluate performance relative to one aspiration level close to performance and an additional aspiration level, called a survival point. The survival point indicates a vital threat for the firm (i.e. indicating that the firm is close to bankruptcy), indicating that firms close or below their survival point will take less risk (Shapira & March, 1992). Hence, studies that follow the threat rigidity theory argue that when firms will take less risk when performance is low.

These contradictory findings in the existing literature on performance feedback and risk-taking behavior has constituted to further investigation on this relationship. Consequently, considerable research is trying to reconcile these opposing perspectives, by arguing that firms interpret performance feedback differently based on their organizational condition. Such contingencies have established boundary conditions under which performance feedback affects taking, and has a moderating effect on performance feedback and organizational risk-taking behavior (Audia & Greve, 2006; Iyer & Miller, 2008; Greve, 2011). For example, Audia and Greve (2006) use organizational size as a moderator for organizational risk preferences, and obtain their evidence by building on the insights from March and Shapira (1986,1992). By using the contingency of firm size, the result reveals that; when performance is below aspirations, decision makers of small firms are more likely to focus on a survival point rather than a point close to current performance, which enhance performance reduction and invoke

(16)

Page 15 of 45 risk-aversion. On the other hand, large firms focus more on aspirations close to current performance and thereby respond with risk-seeking behavior. Small firms have limited resources to take on risk compared to large firms, which indicate that they are more vulnerable and threatened by low performance compared to firms with larger endowments (Audia & Greve, 2006). Contingencies that moderates a risk-taking behavior is therefore a way to enhance the understanding of the contradicting arguments on risk-taking behavior when performance falls short.

The debate regarding the effect of short falling performance on risk-taking behavior has received substantial attention (March & Shapira, 1987, Bromiley, 1991). However, considerable research has often assumed change behavior to be risky, and thus failed to distinguish between change and risk (Kacperczyk et al. 2015, Greve, 2011). Changes that has been analyzed are often routines or incremental changes, which often have little/ or no impact on risk levels (Kacperczyk et al. 2015; Greve, 2005). Though, changes can be associated with different levels of risk, which makes it important to distinguish between the two concepts. Subsequently, one must distinguish between risk at different levels depending on the magnitude of downside risk associated with the change activity (Greve, 2011). Some changes are incremental with little risk associated, whereas other activities are associated with a significant probability of potential losses or failures, such changes are much riskier (Kacperczyk et al. 2015; Markovitch, 2005). Studies that follows these predictions ask for further investigation on how performance feedback affect activities with different risk levels (Greve, 2011; Kacperczyk et al. 2015, Iyer & Miller, 2008), and how these outcomes might differ from each other. Hence, this will be the aim for this study. First this paper will use a performance feedback model that test whether performance below an aspiration level will lead to more, or less risk-taking. Next,

(17)

Page 16 of 45 the aim of this study is to enhance the notion of performance feedback and different risk levels, by investigating other performance feedback models that provide inconsistent signals, respectively.

Additionally, prior studies have followed the assumption that decision makers evaluate performance relative to their historical and social aspirations independently. Meaning that, in the so far discussed papers on performance-aspiration discrepancies and risk, the response on performance feedback are treated and evaluated in isolation (Greve 2003; Iyer & Miller, 2008; Audia and Greve, 2006; Bromiley, 1991; Chen & Miller, 2007). Yet, firms might seek to meet multiple goals, that is, evaluating performance relative to both their historical and social aspiration level simultaneously. A consequence of this is however that performance feedback signals can be inconsistent; when performance is above one aspiration level but below another at the same time (Lucas et al. 2012; Baum, et al, 2005). How decision makers will respond to inconsistent performance feedback signals is thus related with much ambiguity, which ask for enhanced understanding on risk-taking behavior. Specifically, decision makers base their strategic choices on the future opportunities of the firm and whether the firm has the means and capabilities of acting on these opportunities. Such interpretation on inconsistent performance feedback might lead to different levels of risk-taking.

Hence, this paper seeks to extend the understanding of risk-taking behavior by considering different risk-taking levels on the relationship between performance below an aspiration level and risk-taking behavior. Since different activities are associated with different levels of risk (Larazza et al. 2007; Markovitch et al. 2005), it is reasonable to suggest that the outcome of these activities will differ from each other when firms are performing below their aspirations - or are facing inconsistent feedback signals. The aim of this study is to provide an

(18)

Page 17 of 45 in-depth understanding on when decision makers take more, or less risk when responding to negative performance. Moreover, by building on the research upon multiple goals (Greve, 2008, 2011, 1998; Baum et al., 2005; Lucas et al. 2015), this paper seeks to advance the understanding of how high and low risk activates are affected when decision-makers must respond to contradictory performance feedback signals. Therefore, the research question that this paper seeks to answer is as follows:

Will performance below an aspiration level and(or) inconsistent performance feedback lead to more, or less risk-taking behavior?

Performance feedback and levels of risk-taking

Decision makers invest in capital expenditures such as new equipment, production facilities and plants to improve product lines and extend their processes, which are made to enhance organizational performance and thus market position (Larraza et al. 2007). However, investments in capital expenditures often require substantial expenditures and are associated with significant downside risk as the failure of such investment will lead substantial losses for firms. Therefore, investments in capital expenditures is defined as a high-risk activity (Larazza et al. 2007).

Additionally, decision makers invest in R&D investments to explore new innovations to enhance their organizational performance and market position through strengthening capabilities and exploiting current knowledge (Markovitch, 2005). R&D investments are considered as a low risk activity compared to short run enhancements of product lines (e.g. capital expenditures and acquisitions), because R&D investments can produce basic learning from organizations and provide incremental improvements of firm capabilities compared to the

(19)

Page 18 of 45 success or failure related to a product line investment. Hence, R&D investments are associated with less downside risk, and can be considered as a less risky decision for firms to take.

Though, investments in capital expenditures and R&D have both been recognized as activities that improve organizational performance (Larazza et al, 2007; Markovitch, 2005; Lucas et al. 2012). Capital expenditures in the way that firms can produce more efficiently from new machinery/equipment and extending production facilities. R&D investments in a way that exploring new innovations can enhance a firm’s capabilities to exploit existing knowledge (Markovitch, 2005). Hence, in accordance to the behavioral theory of the firm and prospect theory, when organizations are performing below aspiration levels, decision makers resort to risk-taking activities that could accelerate growth or safeguards to new profits in attempt to ensure performance turnarounds (Greve, 2003). Following this reasoning, hypothesis 1 is presented:

H1: When performance is below an aspiration level, performance is negatively related to investments risk activities, such that further decrease in performance relative to aspirations leads to an increase in risk activities

However, the main characteristic that distinguish different risk levels is depending on the magnitude of downside risk related to the activity. It is therefore expected that performance feedback will affect dissimilar risk activities differently. In the investigation on acquisitions, Iyer and Miller (2008) found that firms will make less acquisitions when performance is below a firm’s aspiration level and close to the threat of bankruptcy. Unlike what the behavioral theory proposes, Iyer and Miller (2008) thus found that that organizations will take less risk when performance is falling short. This is a significant contribution to the performance feedback

(20)

Page 19 of 45 literature, because the study base its findings on acquisitions, which is a highly risky activity that is associated with significant downside risk (Markovitch, 2005). Following the predictions of Iyer & Miller (2008), this paper expects that organizations will invest more in low risk activities compared to high risk activities when performance is below an aspiration level because low risk activities are associated with less downside risk, which provides greater probabilities of surviving if the investment fails. Hypothesis 2 is thus presented:

H2: When performance is below an aspiration level, organizations will invest more in low risk activities compared to high risk activities.

Inconsistent performance feedback

In the so far discussed literature on performance feedback and risk-taking, performance feedback from historical and social aspirations has mostly been treated as independent signals (Chen & Miller, 2007; Audia & Greve, 2006; Iyer & Miller, 2008). Some studies have treated the performance feedback as interchangeable (Chen & Miller, 2007), whereas other studies have operationalized performance feedback as a single signal treated in isolation (Lant & Montgomery, 1981; Chen and Miller, 2007). These studies respect the premise that decision makers are evaluating performance relative to one aspiration at a time, indicating that performance becomes unambiguous once it has been interpreted relative to an aspiration level (Joseph et al. 2015). To this point, this paper has therefore not distinguished between historical and social performance aspiration levels, though these two performance referents may influence decision maker’s risk-taking differently compared to what either would do alone (Baum et al. 2005). Hence, this paper seeks to propose an alternative treatment of performance feedback, when decision makers treat both aspiration levels simultaneously.

(21)

Page 20 of 45 There exist different combinations, when treating multiple aspiration levels simultaneously. Consistent performance feedback indicate that historical and social aspiration levels are following the same patterns: meaning that current performance is below or above both historical and social aspiration levels. Considered together, it indicates that if performance is below both historical and social aspiration level, the performance feedback is currently underperforming compared to the firm’s own track record and its industry peers.

Existing literature has investigated the effect of inconsistent performance feedback on multiple goals in performance feedback, meaning that historical and social performance feedback is not following the same patterns which have different propensities for risk-taking behavior. One position is that historical performance is below aspiration level and social performance is above the firm’s aspiration level. This combined feedback indicates the firm is currently underperforming relative to its own performance last year, but at the same time is outperforming its industry peers. A consequence of inconsistent performance feedback is that decisions makers have difficulties coming up with an unambiguous conclusion to the contradicting signal. The responsiveness of the firm thus becomes ambiguous which can have a vital impact of the survival of the firm (Joseph et al. 2015). Existing literature studies have addressed this issue in different ways.

Greve (1998) found that decision makers may shift their attention between historical and social aspiration when performance feedback is inconsistent, depending on how the firm is performing relative to each aspiration level when performance feedback is inconsistent. Greve (1998) proposed two attention rules that decision makers use when evaluating contradicting performance feedback signals. These decision rules are conceptualized as heuristic short-cuts when decision makers are processing complex information or facing challenging situations,

(22)

Page 21 of 45 which will determine the response of the decision makers. One is a “fire alarm rule” in which decision makers will focus on the one aspiration level above performance that provides negative feedback and ignore the other (Greve, 1998). By focusing on the negative performance feedback, it leads to an enhance in decision maker’s responsiveness and follow the behavioral theory of the firm (Cyert & march, 1963) as well as prospect theory (Kahneman & Tversky, 1797). This build on the assumption that decision’s makers respond by taking action that potentially will root out problems (Cyert & March; Greve, 1998). Thereby decision makers seek to minimize any negative discrepancies between the current and desired performance (Lucas et al. 2015). In contrast, decision makers might follow the self-enhancing rule, in which the decision maker’s attention shifts to the aspiration level below current performance that provides positive feedback. This build on the assumption that people no longer respond by seeking for solutions to solve problems, but instead focus on enhancing their self-image (Jordan & Audia, 2012). The motive behind the self-enhancement rule is the desire to see one-self as a winner, which makes decision makers assess performance feedback with a rationality of retrospectives by refining standards of evaluating performance (Jordan & Audia, 2012). As decision makers see themselves as a reflecting of the performance of the firm, negative discrepancies become a threat to the self-image and will prone them to self-enhancement. In the study by Greve (1998) no evidence was found on attention shifting. Additionally, Greve (2008) investigated multiple goals in terms of performance and size goals, in which he addressed the issue by introducing an additional attention rule, defined as a sequence attention rule. This rule indication that decision makers attend one goal at a time and move on to the next goal when performance on the first is above the aspiration level (Cyert and March, 1963: pp 117-119). The result showed that size goals can either be activated or deactivated when performance is below an aspiration level; when a firm experiences high losses, the size goal

(23)

Page 22 of 45 affects growth, whereas the size goal affects growth less, when the firm is experiencing low losses.

However, instead of using the attention rules introduced by Greve (1998;2008) decision makers may combine information from social and historical aspirations as an alternative when interpreting performance feedback (Baum et al. 2005). The effect of multiple goals is hence investigated simultaneously, in which the information sources from both historical and social aspirations differs in relevant respects (Lucas et al. 2015). Baum et al. (2005) follow this approach in a study on how market share and network status both affect network tie initiation. The result show that banks with network-status below historical but above social aspirations had greater propensity for nonlocal ties, which also accounted for banks with network-status performance below social but above historical aspirations.

While prior literature mostly discusses what sort of performance feedback signals that will lead to change in organizational behavior (Greve, 2003; 1998), it might be beneficial to turn attention towards how decision makers interpret performance feedback in terms of the presence of opportunities and capabilities to take risk in the attempt to increase future performance and social position (Lucas et al. 2012). Following the combined approach of performance feedback with both historical and social aspiration, there exists two inconsistent performance feedback combinations that affect risk-taking differently.

In the situation where the organization’s performance is below social aspiration level, yet above historical aspiration levels. This combination implies that although the organization is currently performing worse compared to its counterpart in the industry, the decision makers are receiving signals of realizing a more favorable social position, because the organization currently is improving its own performance compared to previous years. Thus, the decision

(24)

Page 23 of 45 makers are confident that the firm has the means and thus the capabilities to displace similar, yet higher-performing competitors in the industry (Baum et al. 2005). Increasing high risk activities is therefore a suitable strategy for firms that intend to approach such opportunities signaled by market competitors. Furthermore, because these firms also experience a positive trend in their own performance, this leads to the conclusion that the firm has the capabilities to invest in high risk activties (Lucas et al. 2012). It is therefore expected that decision makers will increase high risk investments, when the organization perform below social aspiration but above historical aspiration level. This prediction follows research suggesting that decision makers increase their confidence and striving goals when performance is above historical aspiration level (Lucas et al. 2015, Levinthal and March, 1981; Lant 1992). The hypothesis is thus formulated:

H3a: The further an organization’s performance is below social aspirations but above historical aspiration, the more the organization will invest in high risk activities

On contrary, firms can be performing above their social aspirations but at the same time below their historical aspirations. This combination of feedback implies that even though the organization is currently outperforming its competitors, the organization is vulnerable to displacement by the currently lower-performing peers in the industry, due to its own declining performance (Baum et al. 2005). Based on arguments from prospect theory (Kahneman and Tversky, 1979) decision makers are triggered to take risk in such situation because the firm is experiencing performance short falls, and on the same time facing the threat of losing their social position. Hence, the potential loss will lead to risk seeking behavior (Kahneman & Tversky, 1979) Yet, due to the negative trend in firm performance and the lack of capabilities to explore new opportunities, this paper expect that firms will increase investments in low risk

(25)

Page 24 of 45 activities. Low risk activities such as R& D investments are associated with less downside risk compared to high risk activities, which provides the decision makers with the possibility to improve the short falling performance without threatening the survival of the firm. This prediction follows literature arguing that firms will take less risk, when a possible fail will threaten the survival of the firm (Lucas et al. 2015; Iyer & Miller, 2008; March & Shapira, 1978). Therefore it is hereby hypothesized:

H3b: The further an organization’s performance is below historical aspirations but above social aspirations, the more it will invest in low risk activities

Methodology

The following will account for the sampling strategy and research design used to conduct this investigation. Moreover, the dependent, independent, and controlling variables will be explained, followed by an introduction of the statistical model.

Sampling

The sample consist of secondary data collected from the combined CRSP-Campustat database. The dataset includes financial and industry data in the time frame 1980 to 2014. Moreover, the analysis was restricted to restricted to public North American manufacturing firms from SIC 2000-3999. The selection process allows for a homogeneous treatment of the dataset, even though it is acknowledged that not all industries are completely similar. The nature of this dataset is panel data, in the time 1997-2014. Thus, this dataset observes firms over different years. A sample has been drawn from the initial dataset by excluding firms that invest more on R&D compared to what they are selling. The dataset was thus reduced from 81,632

(26)

Page 25 of 45 observations to a sample of 53,375 observations. Firms that invest more than they sell are typically in a start-up phase of their business. These types of firms do not fall under the reasoning of the behavioral theory by Cyert and March (1963), because this theory concern firms with ongoing sales and production (Chen & Miller, 2007). The three biggest indystries are Chemicals, Industrial and Commerical machinery and computer equipment, and other electrical equipments, respectively. These industries account for 15% 18%, and 22% of the total two-digit SIC industries. Using secondary data allows for a great number of observations and is an efficient way of collecting the needed sample taking time and resources into consideration. The limitation however lies within the external validity and hence, the generalizability of the result. Since dataset only includes financials of public firms a part of the population is excluded in terms of private firms.

Dependent variables

This paper consistent of two dependent variables; one is a proxy for high risk activities/ investments and the other is a proxy for low risk activities. The two dependent variables reflect the type of activity a decision maker will engage in as response to the received performance feedback. By following the assumption presented by e.g. Greve (2011) and Kacperczyk et al., (2015); that change can be associated to different levels of risk. This study will use different activities that are associated with different levels of risk. Larraza et al. (2007) made a questionnaire regarding which activities a CEO will associate with high and low risk. In line with the study by Larraza et al. 2007, this research will use (1) annual change of capital expenditures as a proxy for high risk, and (2) annual change in R&D expenditures as a proxy for low risk. A comparison between performance feedback and the two dependent variables will be made in order to test for more or less risk-taking.

(27)

Page 26 of 45 By following previous studies (Iyer & Miller, 2008; Lim & McCann. 2014; Chen & Miller, 2007), performance below historical aspiration level and below social aspiration level will be interpreted such that positive coefficients for organizations performing below their aspirations will indicate that the negative discrepancy will become larger. Hence, the organization will investments less in low and high risk activities. Whereas negative coefficients imply that organizations will invest more in low and high risk activities.

Independent variables

Firm performance is measured by using ROA (net income divided by total assets), which is a frequently used measure for firm profitability within the manufacturing industries, and has been applied in a vast of prior studies (Bromiley, 1991; Greve 2003; Audia & Greve, 2006; Iyer & Miller, 2008). When evaluating performance feedback, performance is compared to both historical- and social aspiration levels. Therefore, a variable that measures the focal firm’s past performance (Historical aspiration level), as well as the past performance for the industry peers (social aspiration level) is included. The firm’s historical aspiration level is constructed by average ROA for the past three years. Whereas the social aspiration level is calculated by the mean ROA for each industry, which consistent of a four-digit SIC level. Actual performance is measured at year 1 and each aspiration level at year (t-1).

The performance feedback is computed by finding the discrepancies following earlier studies (Baum et al. 2005; Greve, 1998). The historical performance feedback is calculated by subtracting the historical aspiration level (firm’s average ROA for the last three years) from the actual performance (ROA) from. And the social performance feedback is computed by subtracting the social aspiration level (the mean ROA for each industry, which consistent of a four-digit SIC level) from the firm’s actual performance (ROA) minus. A spline function is

(28)

Page 27 of 45 used to compare the performance slopes above and below an aspiration level following previous studies e.g. (Desai, 2008; Greve, 1998; Lucas et al. 2015; Baum et al. 2005). The discrepancies are split into two variables being either above or below the aspiration level variable. The performance below the aspiration variable consist of observation smaller than zero, showing a negative discrepancy. whereas, performance above the aspiration variable carry values greater than zero - showing a positive discrepancy.

Inconsistent performance feedback variables follow the approach from previous studies that look at combined information from both historical and social performance feedback simultaneously. Thus, building on previous studies by (Baum et al., 2005 and Lucas et al. 2012) the inconsistent feedback is calculated by creating four interactions between the four performance variables. The consistent performance feedback interactions are calculated as follows: Social aspiration >0 x Historical aspiration >0, Social aspiration <0 x Historical aspiration <0. The inconsistent performance feedback is calculated as follows: (H3a) Social aspiration >0 x Historical aspiration <0, (H3b) Social aspiration <0 x Historical aspiration >0. This paper will use the last two interaction terms that concern inconsistent performance feedback. If the estimate coefficient for the interaction term Social aspiration >0 x Historical aspiration <0 is positive, this indicates that organizations who performance above their industry peers but below their own past performance have greater propensity to make low risk related investments. On the counterpart, a positive estimate coefficient for the interaction term Social aspiration <0 x Historical aspiration >0, indicates that organizations who are performing worse compared to their industry peers but above their own past performance will take make more investments in high risk related activities.

(29)

Page 28 of 45

Controlling variables

Industry growth: This paper will account for industry effect by following the study by Chen & Miller, 2007. An industry-level variable at the four-digit SIC level is thus computed computed as the percentage change in industry sales from t-1 to t.

Size: Since researchers have found that organizational size might increase a firm’s risk-taking (Lim & McCann, 2014; Audia & Greve, 2006), this paper will control for firm size. The measurement will be based on number of employees, which has been argued for a reliable measure of overall firm size in a given industry (Audia & Greve, 2006). To obtain an approximately normal distribution, this paper use the logarithm of employees following the studies by Audia & Greve (2006); Lim & McCann (2014), and Gentry& Shen (2013).

Slack resources: Existing literature has proven, that decision maker’s responsiveness to changes is depending on the slack resources available (Singh, 1986). This study will follow Bromiley (1991), Chen & Miller (2007), Wiseman & Bromiley, (1996) and Singh (1986) by selecting current ratio (current assets divided by current liabilities) to control for unabsorbed slack, and working capital to sales ratio as proxy for absorbed slack. Both variables have been normalized and added together as a slack index.

Distance to bankruptcy: Based on the study of Iyer and Miller, 2008. This paper will measure distance to bankruptcy through the Altman's (1983) Z-score, which is calculated: (1.2 X working capital divided by total assets) + (1.4 X retained earnings divided by total assets) + (3.3 X income before interest expense and taxes divided by total assets) + (0.6 X market value of equity divided by total liability) + (1.0 X sales divided by total assets). A high Altman's Z-score imply that a firm is not at risk of bankruptcy. Yet a low Z-score means that the firm is on the verge of bankruptcy.

(30)

Page 29 of 45

The statistical model

Since the data set is based on panel data, it is first valuable to test whether the analysis must be conducted using a random effect model or a fixed effect model. Random effect model is a time series, cross sectional regression model that includes an intercept that is an error term. This error term is independent from the error for a specific observation, which means that the estimating parameters are described by the distribution of which each unit’s intercept is drawn (Yaffee, R. 2003). The fixed effect model assumed that the intercept is constant over time. The model however allows for heterogeneity between firms by including multiple dummy variables. Meaning that the fixed effect estimators only deviate from the group means. To test the relevance of using random effect on the data model, meaning that we want to assure that the all error terms are random, a Breusch-Pagen Lagrange multiplier test has been used for both dependent variables (Low/High Risk). The test results for the dependent variable Low Risk rejected the null hypothesis (!" 1 = 2.58, < 0.05). Though the null hypothesis was not rejected for the model using High Risk as a dependent variable !" 1 =0.04, p< 0.41). Indicating that a simple OLS estimation is not appropriate of the first model but could be appropriate for the second model. To clarify whether the analyses should be conducted within groups (fixed effect) or between groups (random effect), a Hausman test has furthermore been conducted. The test showed significant results for both dependent variables, Low Risk: (!"(8)= 89.90, < 0.00), and High Risk: (!"(8)= 91.96, < 0.00). Hence, the null-hypothesis must be rejected and it can be concluded that the random effect is not appropriate to account for firm- and time series effect in the data. The analyses will thus use the within (fixed effect) model, which allows to control for both firm and time series effect. Unlike the random effect, this approach will only allow us to make interpretations of the actual data, but not the whole population that this sample has been drawn upon. This paper use STATA to test all hypotheses.

(31)

Page 30 of 45

Results

The following section will represent this paper’s research analysis. To provide an overview of the data, the descriptive statistics are first presented. Next, a correlation matrix is conducted, where all the significant correlations are exposed. Finally, several multiple regression models have been conducted for testing this papers hypotheses.

Descriptive and correlation matrix

Table 1 presents descriptive statistics together with a bivariate correlation analysis. To summarize the strength of association between the variables a Pearson product moment correlation has been conducted. First, the correlation between the dependent variable of low risk will be explained. Followed by the correlation of positive and negative discrepancies between performance relative historical- and social aspiration levels. Lastly, the same structure follows for correlations between the dependent variable, high risk, and the same variables.

(32)

Page 31 of 45 When studying the correlation matrix, it can be inferred that most variables are significant at the .01 level. Furthermore, it must be kept in mind that the majority of the correlations are weak, which means that there is almost no association found bettwen the dependent variables and the study variables. It is assumed that it is possible to make test the given hypotheses and make certsin interpretations despite the weak associatens. When considering the correlation between the dependent variable Low Risk and firm size there is a negative association (r = -0.06). Though, the dependent variable is positively correlated with slack resources (r = 0.06). Moreover, Low Risk has almost no relation with the variable distance to bankruptcy (-0.01). In regards to Industry sales growth, industry, this show a correlation coefficient of (r=0.03), thus positively but weakly correlated to Low Risk. Performance relative to historical aspiration level show correlation coefficients of (r=0.02) and (r=0.05) in relation to performance below and above historical aspiration, respectively. Whereas there is almost no relation between performance and social aspiration levels since performance below social aspiration shows a coefficient of (r=0.03) and above social aspiration (r=0.00).

Additionally, the dependent variable High Risk, is negatively correlated with firm size (r =-0.04) as well as slack resources (r=-0.04). Next, High risk is negatively correlated to distance of bankruptcy (r=-0.02) yet, the variable is positively correlated with the control variable for industry (r=0.01). The relation between High Risk and historical aspiration below performance is negative (r=-0.02) but positive for historical aspiration above performance (r=0.03). Social aspiration below performance is negatively/almost no correlation (r=-0.01), and social aspiration above performance has no association with the High Risk variable (r= 0.00).

(33)

Page 32 of 45

Regression analysis

Table 2 and table 3 illustrate a summary of the parameter estimates for the dependent variables high risk and low risk, respectively. Model 1 and model 5 only consists of control variables and thus show as a base model for both dependent variables. Model 2, 3, 6 and 7 show the main relationship which means that the splined performance variables has been added to the models. Finally, model 4 and 8 shows the effect of inconsistent performance feedback, by adding performance interactions.

(34)

Page 33 of 45 Regarding the effect of performance below an aspiration level and risk-taking, Model 2 shows the coefficient for performance below their aspiration level is negative and insignificant ((= -0.01, p< 0.947). Hypothesis 1 is not supported. Model 3 shows that the coefficient for performance below social aspiration is positive but not significant (( =0.23, p< 0.313) thus, not supporting Hypothesis 1. The effect of performance below aspirations on high risk activities is therefore not supported. Model 6 shows positive and significant coefficient estimate (( =0.28, p<0.000), indicating that the further an organization falls below its historical aspiration level, the bigger the increase in low risk investments. Model 7 also support hypothesis H1b at a 5% significance level and a positive coefficient estimate ((=0.08, p< 0.045), indicating that the further the organization falls below its social aspiration level, the more the organizations will invest in low risk activities. Hence, the hypothesis 1b is significantly supported, stating that when performance is below an aspiration level, organizations will increase R&D expenditures. Based in the information provided from the results of hypothesis 1, these results will now be compared to test whether hypothesis 2 can be supported.

Hypothesis 2 claims that organizations will invest more in low risk activities compared to high risk activities when performance is below an aspiration level. By comparing performance below the historical aspiration (model 2 and 6) the result show that the coefficient for high risk investments (( =-0.01) is negative but more positive compared to low risk investments (( =-0.15). Thus, the result supports H2 by indicating that when performance is below historical aspiration level, organizations will experience as larger increase in low risk activities compared to high risk activities and thereby take less risk. When comparing the effect of social performance feedback (model 3 and 6) on high and low risk. It can be seen that both the coefficient for high risk (( =0.23) and low risk ((=0.15) are positive. Positive coefficients

(35)

Page 34 of 45 indicate that organizations decrease their investments in both activates. Though the decrease is less for low risk, which mean that this result somewhat supports hypothesis 2.

Hypothesis 3a suggests that the further an organization’s performance is below social aspirations but above historical aspirations, the more the organization will invest in high risk activities. Model 4 show a positive and significant coefficient estimate (( =2.15, p<0.002) for the variable Social aspiration <0 x Historical aspiration >0. Hypothesis 3(a) is thus supported. This combined performance feedback will lead to increase in high risk investments the further an organizations performance will fall below its social aspirations but above historical aspiration level. Plotting the interaction for firms performing below social aspirations and above historical aspirations took the value of one standard deviation below (i.e. low level) and above (i.e. high level) the mean. The plot of the interaction is shown in figure 1. To illustrate the effect, -1 and 0 was used to present the values of performance below social aspiration levels in figure 1. Hence, the figure shall be read from right to left: when performance below historical aspirations drops from 0 to -1 it indicates that firms are experiencing further performance shortfalls relative to their social aspirational level. Consistent with hypothesis H3a, figure 1

(36)

Page 35 of 45 shows a positive relationship between performance below social aspirations when firms perform highly perform highly above historical aspiration levels (( = 11.10, p < 0.00), meaning that organizations increase their high risk activities as a response to negative social performance feedback signal, when the firm is performing better compared to its own history record. Moreover, the figure shows a negative relationship between performance below social aspirations and high risk when firms perform slightly above the historical aspirations (( = -10.03, p <0.00).

Finally, Hypothesis 3b postulates that the further an organization’s performance is below historical aspirations but above social aspirations, the more it will invest in low risk activities. Model 8 show a positive yet insignificant coefficient estimate for the variable Social aspiration >0 x Historical aspiration <0, (( =0.88, p<0.376). This combined performance feedback will lead to an increase in low risk investments the further an organizations performance will fall below historical aspirations but above social aspiration level. This result is however not significantly supported, meaning the null hypothesis for H3b cannot be rejected. It can therefore not be inferred by the result that decision makers will invest in low risk when being threating in losing their social position.

Discussion

The aim for this research was to examine whether performance below an aspiration and inconsistent performance feedback leads to more, or less risk-taking. With a few exceptions, the overall results are consistent with the suggested predictions. For this discussion section, the major findings and contributions will first be discussed. Next, this study’s limitations will be presented, followed by some recommendations for future research.

(37)

Page 36 of 45

Major findings and contributions

There has been considerable debate on the effect of performance feedback on organizational behavior, specifically on risk-taking behavior. Some studies argue that performance falling below a firm’s aspiration level, motivates decision makers to take risk in the attempt to improve performance (Bromiley, 1991; Wiseman & Bromiley, 1996; Audia & Greve, 2006; Miller & Chen, 2004). On contrary, other researchers have claimed that decision makers act risk averse when the organization is performing below their aspiration levels (Iyer & Miller, 2008; Wiseman & Bromiley, 1996). This paper tried to reconcile these contradicting results by specifying under which conditions firms are likely to take more, or less risk. However, because considerable research on performance feedback and risk-taking behavior have failed to distinguish between change and risk-taking, there exist limited notion on how the outcomes of distinct activities associated with different levels. Therefore, this study used two specific activities to determine different levels of risk, (1) investments in capital expenditures, defined as high, and (2) investments in R&D, defined as low risk. Consistent with existing literature that base their predictions on the behavioral theory of the firm (Greve, 2003; Audia & Greve, 2006), this paper found an increase in risk-taking when a firm’s performance is below an aspiration level. This result was however only supported for investments for low risk activities.

Moreover, it was investigated whether firms take more, or less risk, when performance is falling short. The result showed that firms will invest more in low risk activities compared to high risk activities. The result was consistent with previous studies that argue that organizations take less risk when performance is falling short (Iyer & Miller, 2008). Particularly, the result enhances the understanding of risk-taking, because it demonstrates that underperforming firms increase low risk investments more than high risk activities. Indicating that decision makers are more sensible for taking high risk when performance is falling short. Since activities with

(38)

Page 37 of 45 significant downside risk can threaten the survival of the firm, if the investments fail, decision makers are more confident in investing in low risk in the attempt to improve the short falling performance. Hence, a difference is found in the outcomes of activities associated with different levels of risk, which support prior studies that argue for change being related with different levels of risk (Kacperczyk et al. 2015, Greve, 2011).

Furthermore, past research argue that it is likely that decision makers respond to multiple goals when evaluating current performance (Greve, 2008; Greve, 1998; Joseph et al. 2015; Lucas et al. 2015). When responding to multiple goals, it might be the case that multiple goals, such as historical and social aspiration levels, deliver inconsistent performance feedback. When performance is above one aspiration level but below the other aspiration level, the conclusion becomes ambiguous in terms of the response of organizational behavior, and more specifically, decision makers’ risk-taking (Joseph et al, 2015).

Some studies have proposed that decision makers use attention rules to shift the attention between the historical and social aspiration level. One rule is the fire alarm rule that shifts the decision maker’s attention to aspiration that level that is above the current performance. The self-enhancing rule is the other attention rule that shifts decision maker’ s attention towards the aspiration level that is below performance. This study however following studies that propose that performance feedback should be evaluated relative to multiple aspiration levels simultaneously (Baum et al. 2005; Lucas et al. 2015). The argument is that both historical and social aspirations provide useful information in the determination of decision making. Historical aspiration provides information on the trend of the performance, and thus indicating whether an organization’s performance is stable, improving or worsening over time. Social aspiration level provides information about how the firm is performing

Performance feedback on risk-taking behavior : does performance feedback lead to more, or less risk-taking behavior?