Amsterdam Business School
Managers’ reliance on performance measures: the influence of
information asymmetry and the quality of indicators
Name: Elleke Dijkink Student number: 6051243
Thesis supervisor: prof. Dr. Victor S. Maas Date: 17 June 2016
Word count: 14295
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
Statement of Originality
This document is written by student Elleke Dijkink 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.
Abstract
Performance indicators is seen by many firms as an effective form of controlling and incentivizing their employees. A performance indicator should measure the actual performance of the employee. How effective this measure is, is determined by several quality components. In this study I examine if there is a positive association between the quality of performance measures and the managers’ emphasis on these measures. Whereas other studies focused on the effects of specific quality elements (e.g. employee participation in the development process), this study takes the effect of the overall quality into account. Moreover, new insight is provided regarding the influence of information asymmetry on the before mentioned relationship. A survey was completed by 103 pairs of employees and their direct supervisors. The three hypotheses formulated are: 1) Information asymmetry will lead to increasingly use of performance indicators by the manager in the evaluation of the employee. And 2) Managers will make more use of performance measures when they consider the indicators to be of higher quality. Moreover, quality of the indicators and information asymmetry together will strengthen this relation, since 3) Managers will rely even more on the outcomes of performance indicators in the presence of information asymmetry when they consider the indicators to be of higher quality. In this study I did not find support for the first hypothesis. However, I did find support for the second and third hypotheses. Therefore I conclude that the quality of performance measures is essential for the manager in order to rely on these outcomes. Information asymmetry is only positively associated with the emphasis on performance indicators in the presence of high quality measures. I discuss the implications of this study and propose suggestions for future research.
Contents
1 Introduction ... 7
2 Literature review ... 10
2.1.1 Why do companies have performance indicators? ... 10
2.1.2 Quality of performance indicators ... 11
2.1.3 The relation between the quality of performance measures and the evaluation .... 13
2.1.4 Information asymmetry ... 14 2.2 Hypothesis development ... 16 3 Research methodology ... 23 3.1 Empirical approach ... 23 3.2 Sample ... 23 3.3 Research question ... 24 3.4 Variable measurement ... 24 3.5 Equations ... 30 4 Results... 32 4.1 Preliminary analyses ... 32 4.2 Hypotheses testing ... 33 4.2.1 Information asymmetry ... 34
4.2.2 Quality performance measure ... 35
4.2.3 Interaction effect ... 35
4.2.4 Overview supported hypotheses ... 36
4.3 Additional tests ... 36
4.3.1 High and low information asymmetry ... 36
4.3.2 High and low quality of the performance measure ... 38
4.3.3 Conclusion additional tests ... 39
5.1 Summary findings ... 40 5.2 Theoretical implications ... 41 5.3 Contributions ... 43 5.4 Limitations... 44 5.5 Further research... 44 References ... 46
Table of figures
Figure 1 Hypothesis 1 ... 18
Figure 2 Hypothesis 2 ... 19
Figure 3 Hypothesis 3 ... 21
Figure 4 Hypothetical model ... 22
Figure 5 Interaction effects after additional tests ... 39
Figure 6 Results hypothetical relations ... 41
Table of tables Table 1 Characteristics respondents ... 24
Table 2 Factor analysis ... 26
Table 3 Descriptive variables ... 32
Table 4 Correlation table ... 33
Table 5 Results of the regression analyses ... 34
Table 6 Results additional test low IA – high IA ... 37
1 Introduction
In recent years there has been a change in the management structure of firms. Whereas the owners used to be actively involved in management, the management is now often separated from the owners. To ensure that management will make decisions that are in the interest of the owners, the incentives of the management should be aligned with that of the owners (Ding, Richard & Stolowy, 2008). In order to align the interests of the employees and the owners many firms now use performance measures. Performance measures are tools to ensure that a firm’s strategy, goals and objectives will be achieved (Ferreira & Otley, 2009). Complex organizations increasingly demand a stronger focus on performance indicators (Denis, Langley & Rouleau, 2007). The implementation of these indicators in the firm’s strategy is a way to control the behavior of the employees. Furthermore, it enables management to monitor the employees in their productivity and performance (Elrod, Murray & Bande, 2013). Evaluation of the performance of the employees is important, since it can increase their understanding of what is taking place and can improve their future performance (Perrin, 1998). Firms attach more value to performance measures that are more accurate (Holmström, 1979; in Indjejikian & Matĕjka, 2011). However, the process of setting performance indicators is rather complex. It is difficult to determine whether performance indicators provide a correct and accurate indication of one’s performance. Therefore, firms have embedded the possibility of errors and inaccuracies in their performance measures system (Murphy & Cleveland, 1995, and Murphy, Jako & Anhalt, 1993; in Roch, McNall & Caputo, 2011).
Groen (2012) describes the relation between the quality of the relevant performance indicators and the degree to which managers evaluate their employees based on these indicators. The quality of the performance measure depends on how well the indicator measures the actual performance of the employee (Abernethy, Bouwens & van Lent, 2004; Groen, 2012). Groen (2012) found a positive correlation between a higher quality of performance indicators and more intensive use of these indicators in the evaluation of the employees. This research suggests managers attach more importance to the performance indicators when these indicators give an accurate and correct representation of the performance of the employee. When a performance measure is of high quality due to the before mentioned conditions, it will enable the manager to gain trust in the decisions and findings of the employee (Moers, 2006).
Some researchers argue that the quality of performance measures is not the management’s primary concern (Hall, 2010). In an organization where decision making is centralized, management can make use of other forms of control such as observations (Abernethy, Bouwens & van Lent, 2010). In this thesis, I will research the motives for managers to put more emphasis
on performance indicators. I am specifically interested in the impact of information asymmetry and the quality of the indicators. In the presence of information asymmetry, the manager must be able to rely on the findings of the employee, since the employee possesses better information. The manager should therefore be convinced that the employee has the ability and knowledge to provide reliable information (Gallouj, 1997). This information can be obtained by evaluating the employee on the outcomes of his/her performance indicators. A performance indicator is of high quality when it is able to measure the employee’s actual performance (Abernethy et al., 2004; Groen, 2012). According to Groen (2012), managers rely more on the outcomes of performance indicators when they consider them of higher quality. Therefore I expect that managers rely even more on the outcomes of performance indicators when both assumptions are met; namely the high quality of the performance indicator and the presence of information asymmetry between the manager and the employee. Consequently, the research question will be the following:
To what extent do information asymmetry and the quality of performance indicators cause that managers evaluate their employees based on these indicators?
To develop the hypothesis I will use the agency theory, because of the relation between the principal and the agent. This theory assumes that when the agent is paired with the principle, he/she is either in possession of better information, has a higher degree of education or has more specific knowledge. Problems arise when the interests of the agent and the principle are not aligned and the agent chooses actions to maximize his own profits. In this research I look at the relation between the operational employee and his direct supervisor. I examine the effect of high qualitative performance measures and information asymmetry on the emphasis managers put on performance indicators.
This research is relevant since it sheds new light on the importance of the quality of performance measures. Due to more complex and fast growing businesses, firms make increasingly more use of performance indicators to control and incentivize the employees (Denis et al., 2007). This can result in information asymmetry and can be the reason for companies to put more emphasis on performance measures. However, do companies rely more on performance measures in the presence of information asymmetry? In this study I measure the impact of information asymmetry and the quality of performance indicators. Whereas previous research has mainly focused on aspects that determine the quality of performance measures, this study reveals the impact of high quality performance measures and information asymmetry on the reliance on these indicators.
Paragraph two provides the theoretical framework for information asymmetry and performance indicators. This section describes the hypotheses that will be tested. The research method is described in paragraph three. In this section the variables and models used for the study are explained. The results of the regression are provided in paragraph four. The last paragraph provides an analytical discussion and concludes.
2 Literature review
This paragraph will introduce the theory for the hypotheses. First, information about the performance indicators will be provided. The relevance of performance measures and the requirements for a high quality indicator are discussed. Next, the relation between the quality of the performance indicator and the degree to which this is used in the evaluation of the employee is illustrated. Then, the concept information asymmetry will be explained using the principal-agent theory. Since information asymmetry can have an impact on the information flow from employee to manager, it might influence the extent to which performance measures are used. Finally, the theory for the main hypothesis is discussed. It provides insights in the variables that influence the managers’ emphasis on performance measures.
2.1.1 Why do companies have performance indicators?
Performance measures have been defined as tools to ensure that an organization’s strategy, goals and objectives will be achieved (Ferreira & Otley, 2009). One of the reasons why companies make use of performance indicators is to provide incentives for employees. These incentives contribute to the aligning of interests between the employees and the owners of the firm. Performance indicators can influence employees’ behavior, since they will be rewarded and compensated on the basis of their decisions (Abernethy et al., 2010). Increasing performance incentives for employees raises their productivity (Kunz & Pfaff, 2002), generates commitment, increases motivation and enhances learning (Chenhall, Hall & Smith, 2014). Firms can use performance indicators as a tool to provide guidance to employees. Employees will be encouraged to deliver the required work if they are rewarded for the achieved targets (Groen, 2012).
In recent years there has been a shift from stakeholder model to shareholder model. The result of this general trend is that ownership is now dispersed. Previously, the owners were often the founder families and therefore actively involved in management. In the shareholder model, on other hand, the owners of the company are separated from the management. This means that owners are less or not at all involved in decision-making (Ding et al., 2008). Management is responsible for owner’s resources and it is therefore important that the interests of employees and the owners are aligned. Monitoring the performance of employees and sharing this information is a way to achieve this. It ensures that the employees are accountable towards the owners (Healy & Palepu, 2001). O’Dwyer and Boomsma (2015) stress the importance of transparency with regard to the firm’s performance. According to them, transparency contributes to building stakeholder trust. Moreover, by sharing information and knowledge, a firm’s flexibility and profitability can be improved (La Forem, Genoulaz & Campagne, 2007). To gather the information and to align the
interests, firms could use performance indicators (Abernethy et al., 2010). Performance indicators, however, provide not only incentives and guidance for employees’ actions, but also enable the manager to control the behavior of the employee and to focus on actions which are important (Jordan & Messner, 2012). Performance measures enhance the visibility and comparability of employees’ performance (Cruz, Scapens & Major, 2011). Abernethy et al., (2004) argue that performance measures minimize the loss of control problems when managers have delegated decision rights. Management will be able to measure the performance of and the actions taken by the lower-level employees when they make use of performance indicators. In this way they are able to exercise control. A comprehensive system for planning, monitoring and evaluating performance is required (O’Dwyer & Boomsma, 2015).
Ryan and Deci (2000) argue that a coerced use of performance measures might negatively impact the employees’ intrinsic motivation. Management and firms can be worse off when providing performance measures as incentives. However, the empirical analysis by Kunz and Pfaff (2002) shows that a large number of jobs are not valued highly enough to encourage high intrinsic motivation. According to their results, reward does have positive effects on performance. Not rewarding good performance would be perceived as unfair and could lead to frustration. This research underlines the importance of performance measures. A well designed incentive contract aligns the interests between participants. It furthermore minimizes the risks of conflict of interest and contributes to goal congruencies.
2.1.2 Quality of performance indicators
The agency theory explains that the main problem arises when the interests of the principal and the agent are not aligned (Zhang & Zenios, 2008). Performance measures should be developed to align these interests. These measures will incentivize employees to take actions that lead to good results for the firm on a short- as well as on a long-term basis (Lambert, 2001). Abernethy, Bouwens and van Lent (2013) argue that the extent to which performance indicators capture short- term and long-term activities will determine the quality of the indicator. The authors moreover argue that the quality of performance measures depends on congruity, sensitivity and precision. Congruity measures to what extent employees’ decisions improve the performance measure and the firm’s value. Low congruity means that the employees’ decisions improve their performance outcomes, but decreases the firm’s value. Consequently, a high quality performance indicator has high congruity. This means that employees’ decisions improve the performance outcomes and at the same time increase the firm’s value. Sensitivity measures the degree to which the actions of the employee influence the outcomes of the performance measure. If a performance indicator is very
sensitive, it gives an accurate and detailed reflection of the work effort and quality of the employee. Precision reflects the noise effect within the performance indicator (Abernethy et al., 2013). Performance measures are too noisy when actions taken elsewhere in the organization influence the outcome of the performance measure (Keating, 1997). The higher the quality of the indicator, the lower the noise effect should be (Abernethy et al., 2013). According to Groen (2012), the quality of the indicator might have a positive impact on the employee’s performance. When employees are aware that the indicator measures their own actions it will increase their performance. The findings of Moers (2006) support the quality components sensitivity and precision. The more sensitive the performance measures are to the actions of the employees, the higher the quality of the performance measure will be. The outcome of the performance measure will reflect the actual effort of the employee. Furthermore, Moers found that high quality performance measures have relatively high ‘verifiability’. In this case, the extent to which a performance measure reflects the employees’ actions is very specifically determined. This measurement process contributes to the verifiability of the indicator and indicates how the employee’s performance is measured.
The involvement of the organizational members in designing the performance indicators contributes to a higher quality of the indicator (Groen, 2012). This is because organizational members have more knowledge about operational businesses than their managers. Their involvement increases the sensitivity and precision of the performance measure. The measure better reflects the employees’ actions and is less sensitive to noise. Chenhall et al. (2014) underline the importance of providing a context in which the employees can express their values and beliefs. The authors suggested creating a developing process in such a way that the employees feel comfortable enough to express their own ideas and suggestions.
Chenhall et al. (2014) argue that the quality of a performance measure depends on how accurately it measures the performance of the employee. The researchers experienced the frustration of organizational members when performance measures are not representing the impact of their work. Moreover, when the measure does not only capture the activities of the employee but also the actions from others, this performance indicator would provide less information to the manager (Abernethy et al., 2004). Incomplete performance indicators might also lose their information purposes. However, according to Jordan and Messner (2012), if managers have the flexibility to interpret the outcomes of the indicators this incompleteness is of little concern. Therefore, Jordan and Messner argue that incompleteness does not have to pose a problem provided that the manager is allowed some flexibility when handling the performance measures.
Setting up performance measures is a difficult process and may not always be achieved. O’Dwyer and Unerman (2008) argue that there is not always a black or white issue. According to the authors, there are situations that the target is reached without knowing whether it is because of the performed effort or because of external factors. This makes it difficult to measure and reward the employee. There is no certainty that the outcomes of the performance measures are the results of the desired actions. Employees would be unfairly rewarded for results beyond their own efforts. Therefore, Kunz and Pfaff (2002) suggest letting the employee bear some of the risk. Incorporating risk in an incentive contract better aligns the reward with the desired actions that are in fact taken (Holmström, 1979).
The findings of Franco-Santos, Lucianetti and Bourne (2012) show that effective performance measures should have targets that are aligned with the strategy. Management has to ensure that the obtained resources are used effectively and efficiently in line with the firm’s objectives (Ferreira & Otley, 2009). To fulfil the objectives, the indicators should have controllability, timeliness and they need to be validated (Franco-Santos et al., 2012). Franco-Santos et al. (2012) also underline the importance of creating a development process that is fair, transparent and that creates a setting where employees feel empowered and involved. They admit that the development process might be time-consuming, might increase costs and can include judgement biases. However, performance measures are important since they facilitate the implementation of business strategies. It provides employees with guidance on where to focus and on the actions that have to be taken to be in line with strategic goals.
2.1.3 The relation between the quality of performance measures and the evaluation
Performance evaluation has two functions. The first function is controlling the behavior of the employee. By providing incentives, the interests of the manager and employee can be aligned. Secondly, performance evaluation enables the manager to analyze the performance of the employee and to determine the effectiveness of the incentive contract (Kunz & Pfaff, 2002). A performance measure should capture all the necessary information to evaluate the employee. De Brandt (1995; in Gallouj, 1996) argued that the manager needs to know whether the employee has the required level of competences and experiences. He/she also needs to know whether the employee has the right skills and what his effort will be.
According to Abernethy et al. (2004), performance measures that not fully capture the lower-level employees’ performances provide less information to the manager. They also lose the purpose of incentivizing the employee. When a performance indicator is ‘noisy’, the employee will take actions that maximize his/her own performance which can have negative consequences for
the firm’s value. In this situation, the authors argue that managers will make less use of performance measures (Abernethy et al. 2004). Instead, they need other forms of controls such as education and selection procedures (Brickley, Smith & Zimmerman, 1997).
Ittner, Larcker and Meyer (2003) criticize the subjective role of managers when they evaluate their employees. The evidence of their research suggests that evaluating employees mainly depends on psychology-based explanations instead of economics-based explanations. This subjective approach might lead to bonuses that are based on preferences and favoritism instead of actual performance. It will create an unfair representation of reality and could therefore discourage employees to do their work properly. Nevertheless, introducing subjectivity in the evaluation can deduct the noise effect that arises from external factors. In this way, including subjectivity in the evaluation of performance measures can improve the quality of the indicators. However, critics argued that subjectivity will make it harder for employees to distinguish between good and bad performance. As a result, employees are less likely to attach value to performance indicators since the relation between efforts and rewards is not clear-cut (Ittner et al., 2003).
Groen, Celeste, Wilderom and Wouters (2016) found a positive relation between the evaluation of performance indicators and job performance. When managers evaluate their employees based on indicators, employees are more incentivized to deliver better performance. This is in line with the findings of Bloom and Van Reenen (2010). They underline the importance of performance evaluation. The findings of their study imply that many employees are not well enough informed about their own performance. This research therefore shows that employees who are evaluated will significantly improve their performance.
2.1.4 Information asymmetry
Several economists investigated the principal-agent model. In this model, ’the agent’ is the party that possesses private information which cannot be observed by the other party, ‘the principle’ (Zhang & Zenios, 2008). Within a typical agency model all individuals are expected to be driven by intrinsic motivation. However, agency problems arise when each individual interest differs from the firm’s objective (Baiman, 1990). Within this setting, the agent holds private information, for example about his knowledge or his effort level, which is not freely available to the principal. The model assumes that the agent will choose actions to maximize his/her earnings which might come at the expense of the owners (Bradford, 1987; Kunz & Pfaff, 2002). Identifying and resolving this conflict of interest can contribute to long-term sustainability of a firm’s performance (Bolos, Tudor & Christian, 2010). A fundamental assumption in agency theory is the concept of information asymmetry (Chia, 1995).
Some procedures require more delegation of decision rights than others (Scandura, Graen & Novak, 1986). Work floor employees are assumed to have a greater understanding of operational activities and processes than their supervisors (Kim, Sting & Loch, 2014). Because of that, the supervisors can give the work floor employees more decision rights (Moers, 2006). This might decrease the manager’s access to information. According to the researchers Zhang and Zenios (2008), this can have a negative impact on the principal’s choices since the principal is not able to assess the information in the same manner as the agent. The greater the difference in information between the manager and the employee, the higher the level of information asymmetry (Chia, 1995). Abernethy et al. (2010) argue that managers will delegate decision rights when there is information asymmetry between them and their lower-level employees. This is necessary since lower-level employees have more specific knowledge and private information. Delegating decision rights will result in better decision making processes. Locke and Schweiger (1979; in, Scandura et al., 1986) concluded that subordinate expertise is the most important reason for the delegation of decision rights. Abernethy et al. (2004) argue that delegating decision rights will result in more decentralization within the organization. Decentralization enables managers to have more responsibility and control over their activities by gaining access to specific information (Waterhouse & Tiessen, 1978).
In order to rely on the decisions and findings of the lower level employee, the manager should be able to trust the employee’s performance (Moers, 2006). According to Abernethy et al. (2004), managers who have delegated decision rights will rely more on the performance indicators. These indicators enable them to measure the employees’ performance. Therefore, performance measures could be useful for the manager in order to gain more trust in the decisions and findings of the employee. These findings indicate the existence of a positive relation between the delegation of decision rights and the reliance on performance measures by managers.
According to Kunz and Pfaff (2002), designing an incentive contract involves judgments. This is because the overall result depends on several factors that are not always influenced by the employee’s effort. Therefore the manager cannot measure the performance of the employee based on the overall result. Thus, an incentive contract must include at least some risk that the employee has to bear. An optimal incentive contract involves a variable component that reflects the employee’s effort level. This incentivizes employees to maximize their effort (Kunz & Pfaff, 2002). Fischer, Maines, Peffer and Sprinkle (2002) examined whether budgets as an incentive contract increase employees’ performance. Their findings show that using budgets as performance evaluation provides incentives for employees. Employees are more likely to share private information and to increase their work effort and performance.
When information asymmetry is less likely to occur in an organization, it is also less common to use performance measures. Abernethy et al. (2010) concluded that if decision making is more centralized managers are less likely to rely on performance indicators. They have fewer control loss problems and could use other techniques to monitor the employee. By directly observing the employee’s work, they can have a good impression of his/her work effort and quality.
However, directly observing is less likely when there is information asymmetry. Information asymmetry might increase distance between the manager and the employee. Therefore it is important to provide feedback and response to the effectiveness of the received information (Connelly, Certo, Trevis, Ireland & Reutzel, 2011). Connelly et al. (2011) describe the importance of providing feedback on the information the manager received from the employee. The manager can indicate whether the information is in line with his/her expectations. This can improve the quality of the information. It enables the manager to anticipate in an early stage information that does not meet his/her expectations. When the received information is not in line with the manager’s expectation, the manager can increase the frequency of providing feedback and he can increase the quality of the performance indicator.
2.2 Hypothesis development
Based on the theory described in the previous paragraph, I will now present my hypotheses. In early years the owners of firms were mainly the families. They were actively involved in management and had a long-term focus. During the last years there has been a shift from this stakeholder model to shareholder model. This new model is more short-term focused. This is because the owners are now the providers of capital and separated from the management. They are not actively involved in the decision-making process and should be able to rely on the decisions made by management. There is less of an information asymmetry problem for the providers of capital if they are better able to assess the firm’s future prospects (Ding et al., 2008).
For the first hypothesis I evaluated the concept information asymmetry. Information asymmetry can be explained by the principal-agent model. Within this model, the ‘agent’ possesses private information, which cannot be observed by the ‘principal’. The agent might want to hide information about his education or level of effort from the principal (Zhang & Zenios, 2008). It is assumed that the agent will choose actions that will maximize his/her earnings (Kunz & Pfaff, 2002).
Information asymmetry arises when employees possess better information than their supervisors (Kim et al., 2014), or when they have superior knowledge (Dehlen, Zellweger, Kammerlander & Halter, 2014). Transferring the information to the manager is unlikely, since he/she does not have this subordinate expertise. In order to have an optimal decision making process, the manager might delegate decision rights to the employee (Abernethy et al., 2010). By delegating decision rights the manager will lose part of his control. Whereas the manager is accountable for specific results and decisions he should be able to rely on the information and findings of the employee. He would have more confidence in the work effort of the employee when he is better able to observe his or her performance (Moers, 2006).
In order to observe the employee’s performance, the manager might evaluate the employee based on the outcomes of the performance indicators. It can increase the manager’s trust in the employee since he/she has a better understanding of the skills and effort of the employee (Moers, 2006). This can then result in the manager relying more on the information and findings produced by the employee. Abernethy et al. (2010) argue that delegating decision rights is associated with the use of performance indicators. They find that in organizations where decision making is centralized, managers put less emphasis on the use of performance measures. Performance measures would be less useful in this setting because the manager can exercise other forms of controls such as observation. The study of Abernethy et al. (2004) finds support for an increasing reliance on performance indicators associated with the delegation of decision rights. Delegation of decision rights might result in loss of control. To minimize the risks that accompany the loss of control, management is more or less forced to use performance measures (Abernethy et al., 2004). This is because performance measures enable the manager to exercise control (Jordan & Messner, 2012). Moreover, it is in the organization’s interest when managers delegate decision rights to the lower-level employees who possess more information as it increases the timeliness of the information and improves the decisions made (Chia, 1995).
Therefore, I assume that in the presence of information asymmetry management increases the use of performance measures. This leads to my first hypothesis:
H1: Information asymmetry will lead to increasingly use of performance indicators by the manager in the evaluation of the employee.
Figure 1 Hypothesis 1
The purpose of a performance indicator is that it measures the performance of the employee. The quality of a performance indicator is determined by how sensitive it is to the actions of the employee and how precisely it measures all relevant aspects of the performance of the employee (Abernethy et al., 2013; Groen et al., 2016). Abernethy et al (2013) argue that a performance indicator should measure the extent to which the employee’s decisions have improved his/her own performance as well as the value of the firm. The more sensitive the measure is, the better it reflects the work effort and the skills of the employee. When an indicator is very noisy, the overall result might not only have been affected by the employee himself, but has also been influenced by actions taken elsewhere in the organization. Thus, minimizing noise and incorporating sensitiveness in an indicator contributes to higher quality. This not only incentivizes the employee since he/she will be rewarded for his/her own performance (Groen, 2012), but it also provides the manager with tools to exercise control (Jordan & Messner, 2012). An optimal performance measure creates incentives and guidance for employee’s actions and enables the manager to control those actions (Jordan & Messner, 2012). When a manager considers the performance measure to be of high quality, he/she is better able to rely on the work of the employee (Abernethy et al, 2004). Groen (2012) investigated the impact of the involvement of organizational members in designing performance indicators. She found that it contributes to a higher quality of the indicator. Moreover, she found that managers were more likely to focus on the performance measures when the lower-level employees have participated in this developing process. Wouters and Wilderom (2008) studied the participation of managers in the developing process of performance measures. They found that problems of incomplete performance indicators can be addressed when managers actively participate in this developing process. The authors mentioned that this is because managers are more likely to accept this transparent system of performance setting. This might imply that when managers consider the performance measure to be of high quality, they are certain enough to make assumptions based on the outcomes of the performance measures. With regard to that, quality of performance measures could contribute to more reliance on performance measures by management.
For my second hypothesis, I expect that managers who consider the performance measure to be of high quality will make more use of these indicators. This is because managers are still able to incentivize and control the lower-level employee. This leads to the following hypothesis:
H2: Managers will make more use of performance measures when they consider the indicators of higher quality.
The model of this hypothesis is structured as follows:
Figure 2 Hypothesis 2
In the presence of information asymmetry, the manager might have more interest in performance indicators (hypothesis 1). The purpose of hypothesis 2 is to find out if there is a relation between the use of performance measures and the quality of the measure.
The agency theory is based on two assumptions: conflicts of interest and information asymmetry between the agent and the principal. Without information asymmetry the principal would be able to directly propose and control the desired action. Without conflict of interest motivation would be needless since the required actions already fit in with the interests of both the agent and the principal. Therefore, in the presence of information asymmetry an optimal incentives contract needs to motivate an agent and enables the principal to control behavior. Increased performance incentives are associated with a raise of the agent’s productivity (Kunz & Pfaff, 2002). Thus, the better aligned the interest between the agent and the principal in the presence of information asymmetry are, the more importance is attached to incentive contracts.
Performance measures are used to incentivize employees (Jordan & Messner, 2012). As mentioned before, the quality of a performance measure depends on sensitiveness, congruity and precision. This would suggest that high quality performance indicators should accurately measure the employee’s actions, contribute to an increase in the firm’s value and should not be influenced by actions taken elsewhere in the organization (Abernethy et al., 2013).
Incompleteness of performance indicators does not have to be a problem as long as the manager is allowed flexibility when handling the indicator (Ahrens & Chapman, 2004). According to the researchers Jordan and Messner (2012), it enables the manager to interpret the outcomes based on his knowledge and experience instead of treating them as ends. However, the results from this study show that allowing flexibility poses problems when managers are feeling coerced to pay more attention to such indicators. These findings indicate that an increased emphasis on
performance indicators enforce higher quality performance measures. In the case of loss of control, which can be caused by information asymmetry, top management may want to underline the importance of indicators. In such situations the problem of incompleteness becomes more apparent (Jordan & Messner, 2012).
Based on the before mentioned arguments, the third and final hypothesis is developed. I expected that when managers put more emphasis on performance measures, they will attach more value to the quality of performance measures. In this main hypothesis, I expect that this relationship will be strengthened in the presence of information asymmetry. Information asymmetry might increase the distance between the manager and the employee. Relying on performance measures enables management to measure the employees’ performance and to exercise control (Jordan & Messner, 2012). I expect that the manager is better able to control the actions and behavior of the employee when the quality of the performance indicator is high. The higher the quality of the performance indicator, the better it reflects the effort of the employee (Abernethy et al., 2013). According to Jordan and Messner (2012), the better the indicator reflects the actual performance and enables the manager to keep control, the more likely the manager delegates decision rights. In this setting, I expect that the quality of the performance measure is positively associated with the delegation of decision rights and thus with a degree of information asymmetry. Together these assumptions contribute to an increasing use of performance measures by management.
Hence, I expect that managers’ emphasis on performance measures is strengthened when both conditions are met: (1) in the presence of information asymmetry, and (2) when they consider the indicators to be of higher quality. Therefore, my third hypothesis will be the following:
H3: Managers will rely even more on the outcomes of performance indicators in the presence of information asymmetry when they consider the indicators of higher quality.
Figure 3 Hypothesis 3
To summarize, I expect that in the presence of information asymmetry, the manager is more interested in performance measures. I also expect that the quality of the indicator is positively associated with a greater emphasis on performance measures. Overall, I assume that because of information asymmetry and high quality performance measures, managers will attach more value to the performance indicators. I expect this because the independent variables have a moderating effect on the dependent variable. This means that the relation between the quality of the performance measure and the emphasis on performance measures will be strengthened by information asymmetry. Therefore, the main question will be: To what extent do information asymmetry
and the quality of performance indicators cause that managers evaluate their employees based on these indicators?
Figure 4 Hypothetical model
H1: Information asymmetry will lead to increasingly use of performance indicators by the manager in the evaluation of the employee.
H2: Managers will make more use of performance measures when they consider the indicators of higher quality.
H3: Managers will rely even more on the outcomes of performance indicators in the presence of information asymmetry when they consider the indicators of higher quality.
3 Research methodology
3.1 Empirical approach
This study is motivated by a desire to increase our understanding of the quality of performance indicators, in the presence of information asymmetry and the use of performance measures. Since there is no public data available to test the hypotheses, a survey has been used in order to gather the specific data. This paragraph indicates which sample is used, what the research question is and how it will be answered. The variables used for testing the hypotheses and the used methods are described in detail.
3.2 Sample
For this study, the hypotheses were tested with a survey. The survey was filled out by pairs of operational employees and their direct supervisors. Both the operational employee as well as his/her direct supervisor needed to meet the following criteria: (1) the employee fulfils operational activities; (2) both have been working in their current position for at least one year; (3) the supervisor uses performance indicators to measure the performance of the employee. The survey that was conducted was part of a project of the Faculty of Economics and Business (FEB) 2015/2016. Together with other students I had the opportunity to use this existing survey provided that we would gather at least six pairs of respondents. Each respondent was aware that his or her answers would be treated as strictly confidential. The final database consists of 103 pairs of employees and their managers. The outcomes are collected by other students and myself. There were more employees than their supervisors who have completed the survey: we received the answers from 113 employees, while 103 managers filled out the survey. Hence, in total we received data from 103 complete pairs (91%).
Both the managers and employees had approximately two months to complete the survey. To check for non-response bias I compared ‘early’ and ‘late’ respondents. In this comparison I noticed that early respondents scored a significantly lower ‘Emphasis on PM’ (M=5.27, SD=1.02) than late respondents (M=5.69, SD=.81), t(91)=-2.26 p=.026. This result could indicate that late respondents are busier than early respondents and could therefore better rely on the outcomes of the performance measures. This might explain the higher score for ‘Emphasis on PM’.
I created a table including the characteristics of the respondents for a complete overview of the data in this research. Table 1 gives an overview of these characteristics.
Table 1 Characteristics respondents
Characteristics Employee Manager
Sex 51% male
49% female 69% male 31% female Education 16% low level high school
37% high level high school/BSc 47% MSc or higher
16% low level high school 27% high level high school/BSc 57% MSc or higher
Age Mean=33 (SD=9.6) Mean=39 (SD=9.6)
Span of control N/A Mean=14 (SD=14.7)
Employees in
organization Mean=2,366 (SD=8,606) Min 3; max 70,000 Idem
Job type 14% production
29% mass services 57% professional services
N/A
3.3 Research question
Given the elaborate motivation in the prior section, the following research question will be answered in this thesis:
To what extent do information asymmetry and the quality of performance indicators cause that managers evaluate their employees based on these indicators?
This study contributes to existing literature since it can provide new insights in the relation of the quality of the performance indicators and the degree to which managers use these indicators in the evaluation of their employees’ performance. Groen et al. (2016) investigated the relation between the quality of performance measures and the emphasis on these measures in the evaluation. While she already found a positive relation, this research will include another aspect. This new item is information asymmetry. Information asymmetry might increase the use of performance measures. Moreover, I expect that when information asymmetry and quality of performance measures are taken together, it strengthens the use of performance measures in the evaluation. Firms and managers with a high degree of information asymmetry and firms that often make use of performance measures can use the findings of this study. They can do so by attaching more importance to the quality of performance indicators.
3.4 Variable measurement
The survey used for this study was a pre-developed questionnaire. The overall aim of the survey was to investigate why performance measures are used: namely, to help employees instead of to
control them. The questionnaire measures several variables, such as self-reported job performance, information asymmetry and measurement properties of performance indicators. As a participant of this survey project, I have used three of these constructs for further analysis. These three constructs are the variables for my thesis. The three variables are information asymmetry (IA), quality of the performance measure (QPM) and the degree to which performance indicators are considered in the evaluation of the employee (Emphasis PM). These three main variables are incorporated in the survey by asking several sub-questions that measure IA, QPM and Emphasis PM. I calculated the Cronbach’s Alpha and performed a factor analysis.
The Cronbach’s Alphas of the constructs IA, QPM and Emphasis PM are respectively 0.94, 0.91 and 0.94. These high Cronbach’s Alphas reflect the internal consistency of the indicators measuring the three constructs.
After that I performed a factor analysis. A factor analysis verifies the scale construction. It tests if the item scores within a construct reflect one underlying factor. It is different from the Cronbach’s Alpha in that it seeks for underlying unobservable variables that are reflected in one or each of the constructs. In fact, a factor analysis is a correlation matrix that looks for variables that are highly correlated. Items that load on one underlying factor will correlate very highly with each other (Field, 2000). In this setting, I expect the factor analysis to reveal three constructs explaining that each construct has only one underlying factor that affected the item scores. The results of the factor analysis are, however, inconsistent with the three items as a scale. Table 2 provides the results from the factor analysis. The first two columns show the classification of the variables based on the Cronbach’s Alpha. The variables are the questions which are included in appendix A. The last six columns are the six constructs that are unrolled from the factor analysis.
Table 2 Factor analysis Component 1 2 3 4 5 6 Emphasis PM A.1.1 0.103 0.773 0.008 0.261 0.413 0.312 A.1.2 0.359 0.434 0.668 -0.208 0.044 0.335 A.1.3 0.925 0.084 0.254 0.082 0.221 -0.004 A.1.4 0.896 0.233 0.223 -0.075 0.015 -0.203 A.1.5 0.381 0.553 0.246 0.543 0.203 -0.176 A.1.6 0.925 0.022 0.223 0.044 -0.035 0.087 A.1.7 0.921 0.155 0.273 0.046 0.016 -0.054 A.1.8 0.933 0.057 0.222 0.006 0.108 0.140 QPM A.2.9 0.569 0.114 0.017 0.728 0.090 0.265 A.2.10 -0.335 0.219 0.251 0.801 -0.065 0.325 A.2.11 0.331 0.211 0.323 0.543 0.578 0.068 A.2.12 0.040 0.273 0.027 0.935 0.055 -0.143 A.2.13 0.112 0.356 -0.341 0.135 0.772 -0.060 A.2.14 0.237 0.793 -0.046 0.014 0.319 -0.208 A.2.15 0.133 0.671 0.471 0.496 -0.008 -0.061 A.2.16 -0.082 0.771 0.543 -0.052 -0.085 -0.242 A.2.17 -0.081 0.822 0.011 0.302 0.148 0.257 A.2.18 -0.090 0.888 0.020 0.301 0.227 0.000 A.2.19 0.446 0.760 0.073 0.065 -0.404 0.042 A.2.20 0.309 0.617 0.091 0.081 0.548 0.160 A.2.21 0.075 0.369 -0.247 -0.012 0.775 0.363 A.2.22 0.164 0.895 0.129 0.134 0.191 0.186 A.2.23 0.004 0.261 -0.005 0.573 0.266 0.653 IA A.3.24 0.563 0.072 0.718 0.230 0.004 0.027 A.3.25 0.612 0.063 0.677 0.151 0.018 0.119 A.3.26 0.581 -0.025 0.689 0.127 -0.073 -0.214 A.3.27 0.355 0.144 0.833 0.099 -0.075 0.084 A.3.28 0.230 0.085 0.842 0.179 -0.272 -0.269 A.3.29 0.280 -0.024 0.674 -0.017 -0.598 0.257
The results show a six items construct. The highest correlated value for each question is marked. It can be revealed from the table that the first construct has five items that are affected by the same underlying factor ‘emphasis on performance measures’. The underlying factor for the second construct is ‘quality of the performance measure’ and for the third construct the underlying factor is ‘information asymmetry’. Except for IA, there are differences between the results from Cronbach’s Alpha and the factor analysis. Some questions which were grouped together based on the Cronbach’s Alpha are dispersed after applying the factor analysis.
For ‘emphasis PM’, the questions A.1.1, A.1.2 and A.1.5 are not clustered in this construct. The questions that belong to A.1.1 and A.1.5 are: I attach very high importance to the performance
indicators in the evaluation of my employee’s performance and in determining potential salary increases. These questions are strongly correlated with QPM. That these questions are strongly correlated with QPM means that managers are consistent in scoring the questions that measure QPM and the questions A.1.1 and A.1.5. This can be interpreted as that managers who have scored high on QPM also attach high importance on PM specific in the evaluation of employee’s performance and in determining potential salary increases. This is in line with the findings of Abernethy et al. (2004). According to this research, managers attach more importance to PMs when they consider them of high quality. This might explain why the questions A.1.1 and A.1.5 are strongly correlated with QPM.
Question A.1.2 measures the importance managers attach to performance indicators in periodic performance reports. With regard to table 2, this is strongly correlated with IA, while based on Cronbach’s Alpha, I expected a correlation with emphasis PM. This indicates that managers rely more on performance indicators in periodic performance reports when there is a high level of IA. This can be explained by the study of Abernethy et al. (2004). They found that managers attach more importance to PMs when there is a high level of information asymmetry.
The variables that differ from the construct ‘QPM’ are A.2.9, A.2.10, A.2.11, A.2.12, A.2.13, A.2.21 and A.2.22. The questions A.2.9, A.2.10, A2.12 load on the same construct and question A.2.11 is also not much dispersed from this construct. These questions are the reversed questions and measure how strongly performance expressed in the indicators is affected by: changes in economic conditions, decisions made in other parts of the organization, changes in behavior of parties outside the organization, and factors behind the responsibility of the employee. The purpose of these questions is to measure how sensitive a performance indicator is. However, with regard to existing literature, sensitiveness is an aspect of the quality of the performance measure (Abernethy et al., 2013). The same applies for questions A.2.13, A.2.21 and A.2.23. These questions identify to what extent the performance indicators only measure what the employee can actually influence and if all the activities and tasks the employee performs are included in the performance measure. These questions reveal how accurately performance indicators measure the performance of the employee. According to Chenhall et al. (2014) this is one of the condition for high quality performance indicators.
The before mentioned arguments show explanations for difference in constructs. However, the arguments provide also support for elaborating on the three existing constructs. The dispersed questions for the construct Emphasis PM that correlate with other constructs can be included in the construct extracted from the Cronbach’s Alpha. That they correlate with other
variables is more an outcome or a result; managers attach more value to PMs when they consider them of high quality (Abernethy et al., 2004). The dispersed questions for the construct QPM that correlate with other constructs can also be included under this construct QPM. The variables measure the sensitiveness and how accurately they are. Since sensitiveness and accuracy of PMs are conditions for quality, it can also be included in the construct QPM. Furthermore, all dispersed questions correlate (positively) with the other variables in the original construct. This indicates that the dispersed questions can be included in the original construct extracted from Cronbach’s Alpha. I performed an additional test. Based on the latest six constructs I performed a Cronbach’s Alpha analysis. The values of Cronbach’s Alphas decreased or remained the same after applying these constructs. The alpha for IA decreased from 0.94 to 0.87, the alpha for QPM decreased from 0.91 to 0.88 and the alpha for Emphasis PM remained the same (0.94). Taken these arguments together and because I used a pre-developed questionnaire, I decided not to exclude any sub-questions from the data. Therefore the three constructs remain the same for this study. All items used in this study are shown in appendix A.
The dependent variable in this research is ‘the degree to which performance indicators are considered
in the evaluation of the employee’ (Emphasis PM). The survey contains one subject with several
sub-questions that measures how much importance a manager attaches to the performance indicators in the evaluation. For example, an actual statement was:
‘I attach very high importance to the performance indicators in increasing my employee’s chance of promotion’.
The manager was required to indicate on a 7-point Likert1 scale how much importance he/she
attaches to performance indicators in periodic reports, in the rating of the performance, in discussions, in salary increases and bonuses, in promotion and in increasing employee’s authority within the organization.
The independent variable in the first hypothesis is ‘information asymmetry’ (IA). One part of the survey compares whether the employee or the manager possesses better information. Six questions have to be answered on a 7-point Likert scale. One of these six statements was:
‘In comparison with me he/she is better able to assess the potential impact on his/her activities of factors external to his/her area of responsibility.’
1 The 7-point Likert scale consists of the following possible answers: (1) Totally disagree, (2) Disagree, (3) Moderately
These sub-questions compare the employee with the manager in their familiarity with the client and technical facets and their understanding of what can be achieved.
The independent variable in the second hypothesis is ‘the quality of the performance indicator’
(QPM). The manager is asked to answer several sub-questions about how sensitive and noisy the
performance indicators are. An actual statement was:
‘Devotion and effort in the job leads to better performance on the performance indicators.’ It measures to what extent the performance indicators represent all facets that are important for the employee’s work. Moreover, it analyses whether the performance indicators summarize the employee’s performance. The questions are measured on a 7-point Likert scale.
Both independent variables have a moderating effect on the other relation. I expect that QPM moderates the relation between IA and the Emphasis PM. Furthermore, I expect that IA moderates the relation between the QPM and the Emphasis PM. The main hypothesis examines whether both moderators strengthen the original relation. This entails that I assume that the manager relies more on the outcomes of the performance measures when he/she considers them to be of higher quality, and in the presence of information asymmetry.
Control variables are included because they may give rise to alternative explanations for the use of performance measures. As control variables I take into account the facts that could influence managers’ decisions to use performance indicators in the evaluation. Since all three hypotheses measure the emphasis managers attach to performance indicators, the control variables for the three hypotheses are the same. Whereas information asymmetry is the independent variable in hypothesis 1, for hypothesis 2 this variable is used as control variable. Furthermore, quality of performance measure is the independent variable in hypothesis 2, while this variable is included as control variable in hypothesis 1. This means that the regression model for the first and second hypotheses are the same. Another control variable I included is the variable is the ‘leader-member exchange’. The leader-member exchange reflects the amount of effort both the supervisor and the employee put in their relationship (Maslyn & Uhl-Bien, 2001). The leadership style of the manager could influence his or her decision to use performance measures in the evaluation. According to Abernethy et al. (2010), this impact could be bilateral. On the one hand, the manager might prefer to communicate more informally with the employee instead of using formal settings of performance measures. This would indicate that having performance measures contributes to a more formal relation setting. However, on the other hand, performance measures might increase the informal communication between manager and employee. According to Groen (2016) this is because an interactive setting of developing and evaluating performance measures contributes to
a more informal relation between the manager and the employee. Scandura et al. (1986) found that managers who gave their employees a high performance rating and perceived high qualitative leader-member exchange delegated more decision rights. These arguments taken together can imply that the leadership style of the manager can have a bilateral impact on the emphasis on performance measures. This means that a high level of leader-member exchange might increase or decrease the use of performance measures. The questionnaire requires the manager to indicate on a 7-point Likert scale to what extent he/she has an informal relationship with the employee. Questions such as ‘I understand his/her problems and needs in his/her work’ and ‘I have a good working relationship with him/her’ have to be answered.
Furthermore, as control variable I also take into account how many subordinates the manager has. I assume that managers with more subordinates are more likely to evaluate their employees based on performance measures. Finally, I identified from prior literature the importance of the manager’s ability to trust the employee (Abernethy et al., 2004). The ability to trust is captured using education and the age of the employee as proxy. I assume that managers will make more use of performance measures when the employees are older or have a higher degree of education. This is because I expect managers to have more trust in older and higher educated employees due to their work experience and knowledge.
3.5 Equations
A regression analysis can predict an outcome variable from one or more predictor variables (Field, 2009). For the first hypothesis, I will use a linear regression. The model I will use for this hypothesis is the following:
𝑈𝑆𝐸 𝑜𝑓 𝑃𝑀𝑖 = 𝛽0+ 𝛽1𝐼𝐴𝑖+ 𝛽2𝑄𝑃𝑀𝑖 + 𝛽3𝑆𝑈𝐵𝑂𝑅𝐷𝑖 + 𝛽4𝐴𝐺𝐸𝑖 + 𝛽5𝐸𝐷𝑈𝐶𝑖 + 𝛽6𝐿𝑀𝑋𝑖 + 𝜀𝑖𝑈𝑆𝐸 𝑜𝑓 𝑃𝑀
The outcome variable is the dependent variable ‘the degree to which performance indicators are
considered in the evaluation of the employee’. The predicted variable is the independent variable ‘information asymmetry’ (β1). The control variables are quality of the performance measure, the number of
subordinates of the manager, the age and education of the employee and the leadership style of the manager. Apart from the latter, I expect the control variables to have a positive influence on the use of performance measures. For the leader-member exchange I assume that it can have a positive or a negative impact on the use of performance measures. As mentioned before, more leader-member exchange can result in less use of performance measures since other informal communication tools might be preferred (Abernethy et al., 2010). However, an increase in
leader-member exchange might also lead to more use of performance measures. This is due to the fact that more leader-member exchange contributes to an interactive setting of developing and evaluating performance outcomes. The residual (ε) is the difference between the predicted and the observed value which cannot be explained by the model. Both the dependent variable and the independent variable are measured on a quantitative – ordinal – scale (the 7-point Likert scale).
For the second hypothesis, I measure the impact of the quality of the PM on the use of the PM by the manager. The estimated regression model for this hypothesis is:
𝑈𝑆𝐸 𝑜𝑓 𝑃𝑀𝑖 = 𝛽0+ 𝛽1𝑄𝑃𝑀𝑖 + 𝛽2𝐼𝐴𝑖 + 𝛽3𝑆𝑈𝐵𝑂𝑅𝐷𝑖 + 𝛽4𝐴𝐺𝐸𝑖 + 𝛽5𝐸𝐷𝑈𝐶𝑖 + 𝛽6𝐿𝑀𝑋𝑖 + 𝜀𝑖𝑈𝑆𝐸 𝑜𝑓 𝑃𝑀
The outcome variable is the dependent variable ‘The degree to which performance indicators are
considered in the evaluation of the employee’. ‘The quality of the performance measure’ is in this case the
predicted variable (β1). The control variables are the same as for the first hypothesis. Whereas information asymmetry is the independent variable in hypothesis 1, for hypothesis 2 this variable is used as a control variable.
The third regression is the main regression. The outcomes of this regression will provide an answer to the research question. The regression for the third and also the main question is the following:
𝑈𝑆𝐸 𝑜𝑓 𝑃𝑀𝑖 = 𝛽0 + 𝛽1𝐼𝐴𝑖+ 𝛽2𝑄𝑃𝑀𝑖 + 𝛽3𝐼𝐴 ∗ 𝑄𝑃𝑀𝑖 + 𝛽4𝑆𝑈𝐵𝑂𝑅𝐷𝑖 + 𝛽5𝐴𝐺𝐸𝑖 + 𝛽6𝐸𝐷𝑈𝐶𝑖+ 𝛽7𝐿𝑀𝑋𝑖 + 𝜀𝑖𝑈𝑆𝐸 𝑜𝑓 𝑃𝑀
The moderating effect of information asymmetry and the quality of performance measures is included. The β3 measures whether both assumptions together - quality of performance measure and information asymmetry - strengthen the independent relation of the variables with the outcome variable. Thus, if in the presence of information asymmetry the quality of the performance measure is high; will the manager then make more use of the performance measure than when there is no information asymmetry or when the quality is low? This answer will be provided by including this last equation in the regression model.
4 Results
In this paragraph the results of the hypotheses will be described. Table 3 shows an overview of the main and control variables used in the study. The correlation between the constructs, including all control variables, is included in table 4.
4.1 Preliminary analyses Table 3 Descriptive variables
Variables N Min Max Mean Median SD
Quality performance measure 102 3.20 7.00 5.27 5.39 0.77
Information asymmetry 100 2.00 7.00 5.10 5.33 1.42
Leader-member exchange 102 1.86 7.00 5.87 5.86 0.71
Emphasis on performance measures 101 2.38 7.00 5.48 5.75 0.94 Number of subordinates managers 103 0 80 13.99 8.00 14.69
Education employee 106 1 7.00 5.96 6.00 1.32
Age employee 105 18 62 33.02 30.00 9.58
The preliminary analyses showed that the two independent and the dependent variables are skewed to the left. The mean is a bit lower than the midpoint of the scale. This indicates that more than 50% of the respondents have scored higher than the average score. This non-normality does not have to be a problem. Therefore, I performed additional analyses to test for outliers. The data must not contain significant outliers or other unusual points that could reflect the different impact they have on the relation. This can change the output since it is not included in the regression model and can have a negative impact on the regression (Field, 2009). The independent variable ‘quality performance measure’ and the dependent variable ‘emphasis on performance measures’ had some outliers. After deleting these outliers the skewness decreased and the adjusted R² of the overall model increased. Based on these improvements, I excluded the outliers from the data. Table 3 presents the details of the variables without the outliers from ‘quality performance measure’ and ‘emphasis on performance measures’.
Table 4 presents the correlation table. This table provides a first indication for the existence of multicollinearity. The data must not show multicollinearity, this means that the independent and control variables should not correlate too much with each other (Field, 2009). According to table 4, there are no values above 0.800 between the independent variables. This might implicate that there is no multicollinearity. Multicollinearity is tested in more detail with VIF and tolerance. When performing the collinearity statistics all variables had a tolerance just below 1 and a VIF just
above 1. Multicollinearity is expected at a tolerance < 0.200 or at a VIF > 10.00 (Field, 2009). Therefore, in this study multicollinearity does not exist.
Table 4 Correlation table
Item 1 2 3 4 5 6 7
1. Quality performance measure 1 2. Information asymmetry 0.256**
0.201* 1 3. Leader-member exchange 0.286**
0.265** 0.323** 0.310** 1 4. Emphasis on performance measure 0.560**
0.616** 0.321** 0.298** 0.375** 0.286** 1 5. Number of subordinates managers -0.094
-0.118 -0.034 -0.109 -0.069 -0.041 0.033 0.005 1 6. Education employee 0.195 0.217* 0.170 0.227* -0.064 -0.059 0.178 0.191 -0.259* -0.238* 1 7. Age employee -0.103 -0.114 0.186 0.227* 0.016 -0.029 -0.146 -0.120 0.175 0.178 -0.246* -0.092 1 Pearson correlations are the first line numbers, Spearman’s correlations are on the second line. **. Correlation is significant at the 0.01 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed). 4.2 Hypotheses testing
Table 5 provides an overview of the results of the regression analysis for the hypotheses 1, 2 and 3. How the emphasis on performance measures is influenced by the variables is tested with two models. The first model describes the effect of information asymmetry and the effect of the quality of performance measures without control variables. Model 2 provides the results of these effects including control variables. The control variables as mentioned in the previous paragraph are: number of subordinates, age and education of the employee and the leader-member exchange.
