Public organisations’ benchmarking behaviour: an agency theory perspective

(1)

Public organisations’ benchmarking behaviour:

an agency theory perspective

University of Groningen

Faculty of Economics and Business

MSc Business Administration: Organizational & Management Control

Remi Rutgers

Student number: s2207753

Supervisor: dr. S. (Sandra) Tillema Co-assessor: prof. dr. H.J. (Henk) ter Bogt

Performance of public organisations has been of particular interest since the 1980s. The significantly lower efficiency rate in comparison to the private sector induced public organisations to apply benchmarking as a tool to search for best practices within and across industries to improve performance. As benchmarking is not without disadvantages and problems, this paper will apply agency theory to improve the understanding of the use of benchmarking information within the public sector. Data is collected by a questionnaire that has been distributed among Dutch municipalities and healthcare organisations. Regression analyses showed that a monitoring application of benchmarking increases the agent’s attention to benchmarking information. Moreover, the link between benchmarking scores and incentives neither does neither increase the agent’s attention to benchmarking information nor influence the agent’s attempts to influence reported performance. However, not using benchmarking for external comparisons induces the agent to focus especially on real organisational performance, which is most valuable for the organisation.

Key words: Agency theory, benchmarking, public sector, performance

(2)

2

1. Introduction

The increasing amount of attention that is being paid to improve performance in the public sector originates from the 1980s (Van Thiel and Leeuw, 2002). Several studies indicated that the public sector operates at a significantly lower efficiency rate than the private sector (e.g.: Rainey et al., 1976; Vining and Boardman, 1992; Fox, 1999). The fact that most public organisations are largely financed by the government implies that they would not go bankrupt. Therefore, compared to private sector organisations, less emphasis is placed on cost reductions and improvements of efficiency, as possible losses will be compensated by higher levels of government. Consequently, a continuous process of performance improvements for public organisations does not exist as explicitly as in the private sector. Furthermore, people could only select a specific municipality by moving towards the concerning area and not in another way. This is just an example of peoples’ limited freedom of choice between public organisations, which causes a lack of competition between public organisations. As competition drives towards higher levels of performance (Mia and Clarke, 1999), the lack of a competitive environment only complicates the performance issue for public organisations.

Global economic decline induced an urge to improve efficiency (Van Thiel and Leeuw, 2002), which has led to several methods and techniques to be applied in the public sector. A technique that was originally developed for the private sector, benchmarking, is one of the techniques used within the public sector for performance management. Adcroft and Willis (2005) state that two-thirds of the managers in the healthcare and education sectors are in some way involved in benchmarking, which indicates the high application of benchmarking. Elnathan et al. (1996) describes benchmarking as “the search for the best practices within and across industries to improve performance”. Organisations can use this technique to compare performance with threshold levels and also with other organisations’ performance. As a result, relative performance will be determined and a subsequent set of actions upon the results should be composed to improve or maintain that position.

Several problems have been identified regarding benchmarking practices in the public sector. Tillema (2010), for example, presented several reasons why the public sector might not benefit from benchmarking activities. Organisations’ dysfunctional behaviour may lead to improvements in their reported performance, while their actual performance does not necessarily increase as well. Another example is that organisations might emphasise good results and pay less attention to their lower scores. Moreover, Ball et al. (2000) argue that organisations could use benchmarking to defend performance rather than to improve performance. As a result, benchmarking does not necessarily lead to performance improvements, rather it could reduce the attention being paid to performance improvements.

(3)

3

sector organisations, where the distinction between agents and principals also exists, agency theory can be applied as well. Agency theory is used to describe agency problems, which arise in situations where an agent can make decisions that affect the principal in a forced and unsolicited manner. The most common cause for agency problems are diverging interests between agents and principals (Jensen and Meckling, 1976). As most problems do, agency problems are likely to deteriorate organisational performance. Therefore, agency problems are of interest in order to enhance organisational performance and efficiency.

The existing literature has examined multiple ‘solutions’ that should mitigate agency problems. Eisenhardt (1989) mentioned monitoring, bonding, and the use of incentives as strategies that could reduce agency problems. Moreover, Chua et al. (2009) argued that principals could solve agency problems by the use of signalling of the agent.

Because public organisations’ activities have in general a more profound impact on society, it can be stated that performance of the public sector is of even greater importance than the private sector’s performance (Wright, 2001). However, an examination of agency problems in the context of benchmarking is lacking in the existing literature. Therefore, this issue is of particular interest of this study, which aims to provide new insights and a more thorough understanding that could improve public organisational performance.

The empirical examination, by the use of questionnaire data from 69 public organisations, partly confirmed the hypotheses. Monitoring, which is a solution to agency problems (Eisenhardt, 1989), by the use of benchmarking information proved to be a useful approach to improve benchmarking efficiency. Furthermore, not using benchmarking for comparison with external organisations causes the organisation to focus on improving real performance rather than reported performance, which has several important implications. However, a stronger link between incentives, which is also a solution for agency problems (Eisenhardt, 1989), and benchmarking scores will neither induce the agent to increase the attention being paid to benchmarking information nor increase the focus on reported organisational performance.

As mentioned above, the existing literature has insufficiently examined the use of agency theory in the context of benchmarking of public organisations. This study tries to address this gap along the following research question:

How can agency theory be used to improve our understanding of the use of benchmarking information in the public sector and, thereby, of possibilities to enhance public organisations’ performance?

In order to provide a deliberate answer to the above stated research question, the following six sub-questions have been established.

1. What is agency theory? And in which ways could organisations respond to agency problems, according to this theory?

(4)

4

3. How can agency theory be applied to public sector benchmarking activities?

4. Which expectations does agency theory provide regarding the use of benchmarking in order to improve performance?

5. To what extent do observations in reality support these expectations?

6. Assuming that also differences will be found, what might explain these differences?

The remainder of this paper will be structured as follows. As a starting point, existing literature will be examined to provide a conceptual model and hypotheses. This conceptual model will, along with the research questions, provide the basis on which the empirical examination is founded. Subsequently, the methodology section will present the research design. Thereafter, the results will be presented. Finally, the discussion and conclusion section will provide a more in-depth analysis of this study’s findings, and limitations and further research recommendations will be discussed as well.

2. Literature review

As mentioned in the introduction, benchmarking is a technique that organisations use for performance management. More specifically, benchmarking is used to improve organisational performance. However, the usage of benchmarking is not restricted to this application. Benchmarking can be used as an instrument to provide accountability towards higher organisational levels and stakeholders as well (Eisenhardt, 1989). Here, benchmarking is part of the contractual relationship between different organisational levels. As agency theory explains relationships between different groups within an organisation, adopting an agency theory perspective could be fruitful to analyse public organisations' benchmarking activities.

2.1 Agency theory

Several theories are thought to be of relevance for explaining organisational behaviour. One of these theories is agency theory, which dates from the 1970s. Jensen and Meckling (1976), at the time entrepreneurs in this field, developed a theory that explains behaviour of agents and principals within their relations.

Jensen and Meckling (1976) defined the agency relationship as a contractual relation in which one party, the principal, outsources business activities to another party, the agent. In its most basic form, the principal is the owner of an organisation, and the agent is a hired external party that performs the job for the principal. Consequently, the principal delegates part of his or her decision making authority to the agent. The idea of inducing an agent to behave in such a way that the principal’s welfare will be maximised is quite general. Therefore, this theory can be applied to almost all organisations.

(5)

5

reasons for joining the relationship. Moreover, agents can be seen as utility maximisers (Lazear and Shaw, 2007), which further induces the agents to act in their own interests. Therefore, diverging interests are harmful in view of the principal.

The cases where diverging interests of agents and principals result into conflicts, are referred to as agency problems. However, diverging interests are not the only cause of agency problems. Agency theory indicated two other causes for this phenomenon (e.g. Jensen and Meckling, 1976; Eisenhardt, 1989). Agents and principals generally have different attitudes towards risks, which imply that the agent would not necessarily take actions in the way that is preferred by the principal. For example, a risk-averse agent will not take large risks, which automatically decreases the variability of profits. When the concerning principal is more risk seeking, a more risky alternative could have been preferred by the principal and, consequently, an agency problem is originated. The third cause for agency problems is information asymmetry. The problem of information asymmetry is especially relevant when a principal is evaluating the agent's performance afterwards. The principal cannot determine whether the good or bad performance is due to behaviour of the agent. Perhaps external influences could have been of particular relevance for the observed outcomes. However, this cannot be determined by the principal. Therefore, as the principal faces difficulties in measuring the agent’s performance, the agent could commit to self-interested behaviour, which could harm the principal’s interests.

However, agency problems do not necessarily harm organisations, as several solutions have been provided within the existing literature. The solutions can be categorised in three main categories: (1) monitoring, (2) bonding, and (3) incentives (Jensen and Meckling, 1976; Eisenhardt, 1989). The first solution, monitoring, is an instrument that a principal can apply to supervise the agent. Supervision should provide more insights in the way agents behave and the decisions they make. Consequently, the principal collects more information, which should decrease the information asymmetry. Bonding, which refers to the costs that the principal incurs to provide incentives that the agent will act in the principal’s interest, is the second solution. Contractual limitations on the agent’s power are an example of a bonding cost, which could be either monetary or non-monetary. Whereas monitoring is more focussed on creating insights in the way the agent acts, bonding is used to preclude specific types of agent behaviour. The third, and last, category of solutions is incentives. The use of incentives is to some extent similar to bonding, as it is used to adapt the agent’s behaviour. Incentives are, however, used as an inducement to motivate the agent to behave in a, for the principal, desired way. Incentives can consist either of financial or non-financial components. Financial incentives are for example bonuses or salary, and non-financial incentives could, for example, be a higher degree of empowerment or an increased chance on promotion, which intrinsically motivates the agent.

(6)

6

spent to ensure that the agent will behave towards the principal’s optimal destination is referred to as agency costs. In addition, as the relation between the principal and agent is based on a contract, which governs the relationship between the agent and principal, the focus should be to determine the most effective contract from the viewpoint of the principal (Eisenhardt, 1989).

2.2 Benchmarking

Traditionally, public organisations have focussed on rules and procedures as mechanisms for controlling organisations. However, various changes in management control issues in the public sector have occurred and, therefore, public organisations rely nowadays more on output controls (Verbeeten, 2008). Output controls are part of performance management, which is concerned with all activities that try to maintain or enhance organisational performance (Otley, 1999).

The public sector can be distinguished from the private sector on several aspects. Probably the most important difference is the absence of market incentives (Delfgaauw and Dur, 2008). The highly political environment of the public sector deemphasises efficiency, as those authors argued. Consequently, public organisations’ strategies, compared to strategies of private organisations, seem to be less focussed on competitive objectives. In addition, as Delfgaauw and Dur argued, the existence of multiple and conflicting goals of public organisations, a political context with a broader range of constituent groups, higher levels of accountability, and more rules and regulations are essential differences that provide incentives to apply a different or adapted structure of performance management.

To align the organisational focus with the performance objectives, several performance measures are available that could be applied. One of the techniques within the range of output controls is benchmarking. Benchmarking is becoming more popular since the 1980s in the public sector all over the world (Triantafillou, 2007). As mentioned afore, benchmarking can be defined as “the search for the best practices within and across industries to improve performance” (Elnathan et al., 1996). Askim et al. (2008) stated that benchmarking can be applied in two separate ways: (1) a vertical application in which the principal applies benchmarking as a performance monitor to oversee activities of the agent, and (2) a horizontal application, which is a voluntarily engagement of organisations where they compare their business characteristics with other organisations. Both applications are meant to improve organisational performance, as these authors argued. Elaborating on the contributions of Elnathan et al. and Askim et al., this study will consider benchmarking as the sum of activities that is performed in order to improve organisational performance, which typically implies a systematic measurement of organisations’ activities to improve their quality and efficiency (Triantafillou, 2007).

(7)

7

benchmarking. This study will focus on these problems and try to solve them in order to improve organisational performance.

Cowper and Samuels (1997) argued that benchmarking is not very appropriate for public organisations, as the lack of market incentives within this sector deemphasises the importance of out-performing competitors. For public organisations is it more important to achieve objectives effectively. Therefore, as they argued, there would have been little emphasis on performance improvements, which causes the comparison of results with similar organisations to be ineffective. Moreover, Kouzmin et al. (1999) stated that, although benchmarking with direct competitors could be effective, public organisations do not have direct competitors. Competitors are thought to face approximately the same circumstances, whereas selected similar organisations could face a different external environment, which could distort findings. Therefore, public organisations that compare benchmarking scores with other organisations do not compare results with competitors. These thoughts are shared by Kyrö (2003), who stated that public sector organisations focus on providing the best services as efficiently as possible, rather than emphasising competition between organisations. Therefore, the comparison of performance may be of minor interest. In addition, Askim et al. (2008) examined municipalities and their benefits from benchmarking activities. Their study concluded that municipalities with high levels of political competition received more benefits from benchmarking than the non-competitive municipalities did. This further emphasises the importance of competition, which is generally absent for public organisations. However, benchmarking could be a beneficial practice to adopt a certain competitive drive.

(8)

8

the extent to which benchmarking actually increases organisational performance. Therefore, principals should aim to maximise benchmarking efficiency.

Furthermore, Bowerman et al. (2002) stated that in the private sector benchmarking is a free choice whereas it is in the public sector much more an imposed exercise. This is considered to be a pitfall as compulsory benchmarking activities are thought to be ineffective. Moreover, Ball et al. (2000) argued that organisations could use benchmarking to defend performance rather than to improve it. This could be due to the imposed exercise of benchmarking, which may result in demotivated agents who perform their benchmarking activities reluctantly. Defending areas of high performance can be regarded as less difficult than improving areas of low performance. Consequently, higher results of the own organisation are emphasised and attention to performance improvements is reduced. This finding is supported by Tillema’s (2010) study in which she found that public organisations might emphasise good results and pay less attention to their lower scores. Furthermore, as Tillema argues, organisations’ dysfunctional behaviour may lead to improvements in their reported performance, while their actual performance does not necessarily increase as well. This will further distort the intention of benchmarking, and thereby its results.

Kouzmin et al. (1999) mentioned the fact that benchmarking generally focuses on organisational results rather than organisational practices. As results are more easily comparable than practices, a focus on results is more obvious. Moreover, benchmarking’s focus on results is aligned with the general broader focus of public organisations (Verbeeten, 2008), which is a focus on results as well. However, it is likely that practices differ more than results between organisations in the public sector. Therefore, a focus on practices should be more relevant in identifying the ‘best of the class’ experience.

The above mentioned problems for public organisations regarding benchmarking activities can be divided into four categories: a lack of competitive drive, a lack of desire to change, dysfunctional behaviour of agents, and a focus on results rather than practices. The presence of these problems suggests that the efficiency of public organisational benchmarking could be hampered. These problems should be of particular interests to public organisations and these problems should be solved to improve organisational performance. Therefore, as prior literature hinted at similarities based on agency theory and public organisation’s benchmarking, examining benchmarking in the public sector with an agency perspective is suggested to be fruitful in order to solve the benchmarking-related problems, which will eventually improve organisational performance.

2.3 Linkage benchmarking with agency theory

(9)

9

public organisations are to a much larger extent characterised by well-defined agent-principal relationships at various levels compared to the private sector organisations. Therefore, applying agency theory to public organisations will be of particular value.

Bowerman et al. (2002) indicated that the trade-off between voluntary and compulsory benchmarking within the public sector is an indication of an agency problem. In the public sector, the compulsory benchmarking variant is used more often. This benchmarking method implies that the agent is obliged by his or her principal to fulfil benchmarking activities. Benchmarking scores are of utmost importance for the principal, as those scores are one of the few means through which a principal can verify the agent’s performance. As an agent will not always benefit from performance evaluations, the agent would not perform the benchmarking activities when he or she is not obliged to do so. Therefore, compulsory benchmarking can, despite the fact that interests differ between the agent and principal, be viewed as a solution to an agency problem, which is in this case information asymmetry. This solution is of particular interest, as, according to Christen et al. (2006), employee performance is likely to decrease in case of an agency problem. Moreover, decreasing information asymmetry will at the same time improve employee performance, as the agent knows that he or she is being watched.

Coinciding with benchmarking’s possibility to verify the agent’s performance, Eisenhardt (1989) and Jensen and Meckling (1976) indicated ‘monitoring’ as one of the solutions to agency problems. Monitoring enables the principal to gather information about the agent’s behaviour and performance, which, consequently, diminishes information asymmetry. A principal can track and assess behaviour, activities and performance of the agent by the use of benchmarking activities, which Askim et al. (2008) call a performance monitoring application of benchmarking. As a result, agency problems will occur less when principals apply more monitoring. Thus it is suggested that the verification of the agent’s performance by a principal’s use of benchmarking information would increase the attention being paid to benchmarking information by the agent. Therefore, it is hypothesised:

H1: When a principal pays more attention to benchmarking information, the attention that the agent is paying to benchmarking information is higher.

Prior literature indicated that public organisations’ benchmarking activities may be confronted with dysfunctional behaviour of agents due to the imposed exercise of benchmarking (Bowerman et al., 2002). This dysfunctional behaviour may induce a less motivated attitude of agents towards benchmarking activities. One of the solutions that agency theory has suggested for solving agency problems could be used here as well. Incentives, which agency theory mentions to align interests of agents and principals (Jensen and Meckling, 1976; Eisenhardt, 1989), could motivate agents to increase their usage of benchmarking information, which will improve benchmarking efficiency.

(10)

10

activities and agency problems will be diminished. More specifically, when the incentives are directly linked to benchmarking scores, the agent is encouraged to improve the benchmarking scores to maximise his or her own remuneration. Therefore, this study suggests that incentives will be most effective when they are directly linked to benchmarking scores. Consequently, the ultimate goal of the incentives in view of the principal will be achieved, that is improving the benchmarking scores, which coincides with improvements in organisational performance. Therefore, it is hypothesised:

H2: A stronger link between benchmarking scores and incentives will increase the attention that the agent is paying to benchmarking information.

As indicated by Tillema (2010), dysfunctional behaviour of agents could lead to a distorted view of performance. Dysfunctional behaviour is the ability of agents to improve organisations’ reported benchmarking scores while at the same time the underlying processes remains unchanged. Consequently, where the second hypothesis suggests a closer link between an agent’s rewards and the organisational benchmarking scores, dysfunctional behaviour could confuse the increase in benchmarking efficiency. When the agent focuses its activities on improving the reported benchmarking scores, it could even be that the organisation’s real performance would not increase at all (Tillema, 2010). In this case, the agent aligns activities with its own interests. Consequently, the incentives have an adverse effect. Therefore, the usage of incentives to improve benchmarking efficiency could be inappropriate. Consequently, it is hypothesised:

H3: A stronger link between benchmarking scores and incentives will stimulate the agent to attempt to influence the reported organisational performance in particular.

These three hypotheses suggest that when a principal is paying more attention to benchmarking information, the attention that the agent pays to benchmarking information is higher. Secondly, a stronger link between benchmarking scores and incentives will increase the attention that the agent is paying to benchmarking information. Furthermore, incentives will stimulate the agent to attempt to influence the reported organisational performance. As improvements in the reported organisational performance do not necessarily reflect improvements in real performance, it remains unclear which specific methods could be applied to improve the real organisational performance, which is most interesting after all.

(11)

11

external organisation. Consequently, when agents perform benchmarking activities, knowing that their organisation’s scores will be compared to external organisations’ scores, their main goal is to provide an image that will be in favour of their own organisation. Therefore, improvements in real organisational performance, which is one of the main goals of benchmarking in view of the principal, can be achieved by concentrating the agent’s benchmarking activities on the internal usage of benchmarking. Agents would then not have to improve benchmarking scores to provide benchmarking scores that favour them over external organisations. Consequently, external comparison should be eliminated, as this would induce agents to focus on reported organisational performance. Therefore, it is hypothesised:

H4: Not using benchmarking scores for comparison with external organisations will dissuade the agent to attempt to influence the reported organisational performance in particular.

H5: Not using benchmarking scores for comparison with external organisations will induce the agent to attempt to influence the real organisational performance.

The five hypotheses suggest a fruitful examination of agency theory, as described in section 2.1, in the context of benchmarking of public organisations, as described in section 2.2. A schematic overview of the proposed relationships and the directions of the hypotheses are presented in Figure 1, which will be the starting point for the data gathering and empirical research in this study. Characteristics about the methodology and data gathering will be explained in the subsequent section.

Principal’s attention to

benchmarking H1: +

H2: +

Agent’s attention to benchmarking Link between benchmarking

scores and incentives H3: +

H4: - Agent’s attempts to influence reported performance No external application of

benchmarking H5: +

Agent’s attempts to influence real performance Figure 1. Overview of the relationships between the variables

3. Methodology

(12)

12

elaborate on the method of data collection, the variables, and the data analysis, which provides several characteristics of the data as well.

3.1 Research method

The cycle of research consists of two parts (Eisenhardt and Graebner, 2007), theory development and theory testing. The second part of this research cycle, theory testing, consists of studies that test previously developed theories. As this study elaborates on previously developed theories, a theory testing approach is applied.

Furthermore, as benchmarking efficiency is more a personal interpretation than a fact, the subjective research approach is thought to be an appropriate approach, which automatically implies that a questionnaire is a suitable method for data collection (Holden and Lynch, 2004). Furthermore, as those authors argue, subjectivism is associated with a tendency towards regulation, rather than change. This coincides with the global trend of public organisations, as they have a tendency to avoid change (Magd and Curry, 2003).

3.2 Data collection

In collaboration with another researcher, questionnaires were distributed among 904 public organisations to collect primary data. Combining the data gathering with another researcher was intended to collect a higher number of responses. The use of questionnaires has the advantage that the interviewer cannot bias responses. Moreover, geographic distances will not be problematic and the costs to spread questionnaires is low (Evans and Mathur, 2005). However, Evans and Mathur mentioned a low response rate as the most relevant weakness of email questionnaires. The importance of this disadvantage is thought to be of minor relevance, as the distribution of a sufficient high number of questionnaires should result in enough responses to conduct reliable statistical analyses.

In advance, a list of approximately 300 municipalities and 200 healthcare organisations was obtainable, which was a first incentive to focus on those two subsectors. However, the dominant reason that municipalities are of special interest is that they are of major importance in daily society. Municipalities have interactions with virtually all people and their performance is, therefore, thought to be of great importance. In contrast, healthcare organisations are in contact only with those people who need their service. Both the performance of healthcare organisations and municipalities is often questioned in the media, which suggests that performance could, and should, be improved. Furthermore, based on the assumption that healthcare organisations’ performance is more crucial than municipalities’ performance, it is expected that healthcare organisations apply less external comparison. Consequently, healthcare organisations should be more concerned about real organisational performance.

(13)

13

hospitals, special care clinics and nursing homes have received the questionnaire on a general email address. The accompanied letter appealed the receiver to send the mail forward to the most appropriate person or department in view of the researches. The respondent should be directly involved with benchmarking. Consequently, respondents are thought to have sufficient knowledge and experience to answer the questions deliberately. When a respondent has finished multiple benchmarking projects in the past three years, he or she could base their answers on different projects. Therefore, respondents are explicitly asked to focus on their main benchmarking project of the past three years.

The questionnaire, of which an abbreviated version is presented in the Appendix, starts with an introduction that explains the research setting to the respondent. The questions are clustered by variable and positioned by relevance. Moreover, the more direct and personal questions are positioned at the end of the questionnaire. Essential definitions are provided, which assured the respondents to interpret terms equally. Consequently, consistency among respondents is ensured and bias due to different circumstances will be diminished. Put differently, the internal consistency of the questionnaire is improved (Van Aken et al., 2012).

The questions are measured on a Likert scale, which is a widely applied response scale and thought to be most popular among respondents (Schiele and McCue, 2006). Usage of a Likert scale should further increase the response rate. Furthermore, a Likert-scale causes the data to be ratio-scaled, which allows the data to be quantitative and this enables the possibility to apply statistical analyses. A specific Likert scale that is most useful in all circumstances does not exist (Cox, 1980; Garland, 1991). However, a seven-point scale is thought to be not too narrow and not too broad for respondents in that they have too many options (Cox, 1980). Therefore, respondents are required to provide their answers based on a seven-point Likert scale.

Eventually, 119 respondents responded, which corresponds with a response rate of 13.1%, which is below the mean response rate for online questionnaires of 20% (Baruch and Holtom, 2008). Of the respondents, 69 completed the questionnaire, which resulted in a useful response rate of 7.6%. This number of respondents should be sufficient, as it enables at least some generalisability (Van Aken et al., 2012). The low response rate could be caused by the high number of questionnaires that those public organisations receive, as was reported by several respondents for not filling in the questionnaire. Furthermore, the period of data collection could have been a little too short. Finally, the use of general email addresses could have refrained people from responding, as the initial receiver was generally not able to fill in the questionnaire.

3.3 Variables

(14)

14

variables are located in existing questionnaires. Moreover, most of the consulted studies did not present their questionnaire in their publication.

3.3.1 Independent variables

The extent to which a principal pays attention to benchmarking information is verified by four items. As no specific questions have been located in other studies’ questionnaires, these items are loosely connected with prior literature. Furthermore, as municipalities’ and healthcare organisations’ principals cannot be designated by a single group, a nuance has to be applied. In the case of a municipality, a municipal council is likely to represent the principal, and for healthcare organisations the principal is likely to be represented by a supervisory body.

Two separate items in the questionnaire examined between the link between financial and non-financial incentives with benchmarking scores. In addition, the extent to which benchmarking scores are included in an agent’s job evaluation is determined by a third item.

For the third independent variable, not using benchmarking for comparison with external organisations, characteristics of Geerlings et al.’s (2006) distinction between three levels of benchmarking have been used. The lowest level of benchmarking uses benchmarking only for internal purposes, which coincides with this study’s variable. Therefore, the characteristics of this lowest level of benchmarking are used to verify whether organisations are benchmarking with external organisations.

3.3.2 Dependent variables

The three dependent variables are: the attention that the agent pays to benchmarking information, the agent’s attempts to influence reported organisational performance in particular, and the agent’s attempts to influence real organisational performance. For both municipalities and healthcare organisations, agents will be the employees of the concerning organisation. The items for the dependent variables have been established based on a loosely connection with the prior literature as referred to in previous sections.

3.3.3 Control variables

The relationships between the independent and dependent variables could be influenced by control variables. Therefore, multiple control variables are examined to provide a more thorough examination of the hypothesised relationships.

(15)

15

benchmarking scores, could influence the hypothesised relationships. Furthermore, the agent’s age and gender, and the specific subsector could influence an agent’s behaviour.

3.4 Data analysis

Of the 69 respondents, 53 represent a municipality and 16 represent an organisation within the healthcare subsector. The median respondent is between 41 and 60 years old, the organisation that he or she works for has a size between 100 and 250 FTEs, and he or she has been involved with two benchmarking projects in the past three years. These numbers suggest that respondents are well informed and that they have sufficient knowledge about benchmarking to answer the questions deliberately. Furthermore, 46 males and 23 filled in the questionnaire and they mentioned both an average and median score of 5, based on a 1-7 scale, on their benchmarking projects.

After collecting, the data is analysed with SPSS version 22. Firstly, a factor analysis has been conducted to verify whether the questions empirically reflect the variables that are intended to be measured. Subsequently, a reliability analysis has been performed, which verified along the Cronbach’s Alpha score whether the questions provide a reliable image. Although Bland and Altman (1997) stated that there is no general score that is sufficient for all nature of studies, it is generally believed that a Cronbach’s Alpha value equal to or above 0.60 is thought to be sufficient (Hair et al., 1998). Subsequently, the data is analysed by the use of linear regression analyses, which indicates whether the proposed hypothesis are confirmed or rejected.

4. Results

This section elaborates on the performed analyses. Firstly, the results of the factor and reliability analyses will be presented. Secondly, the linear regression analyses’ results indicate whether the hypotheses are confirmed or rejected. Moreover, the examination of control variables will indicate whether the relationships in the hypotheses are influenced by those control variables.

4.1 Factor and reliability analysis

(16)

16

items into factors should not be problematic. Despite the KMO and Bartlett’s Test of Sphericity measures were sufficient for the extent to which organisations compare their benchmarking scores externally, one item has been removed because it appeared afterwards that this item did not coincide with the aimed variable. Consequently, as only one item remained, neither a factor analysis nor a reliability analysis could be performed. Therefore, this variable is neither represented in Table 1, which presents an overview of each item’s high loading on the concerning factor, nor in Table 2.

Table 1

Results factor and reliability analysis

Variable and Cronbach’s Alpha Items Loading

Principal’s attention to benchmarking (ATT_PRINCIPAL)

Cronbach’s Alpha = 0.907

Principal studies benchmarking scores carefully 0.902 Principal finds benchmarking scores important 0.898 Principal continuously emphasised importance 0.874 Principal insisted to improve the scores 0.868

Agent’s attention to benchmarking (ATT_AGENT)

The organisation takes benchmarking seriously 0.960 My colleagues and I take benchmarking seriously 0.960

Link between benchmarking scores and incentives (LINK_INC_SCORES)

Agent’s financial incentives depends on scores 0.866 Agent’s non-financial incentives depends on scores 0.812 Scores are included in job evaluation 0.782

Influence reported performance (REPORTED_PERF)

Benchmarking to improve reported performance Reported performance is the main purpose

0.842 0.842

Influence real performance (REAL_PERF)

Benchmarking to improve real performance 0.932 Real performance is the main purpose 0.932

Reliability of the data is verified along each variable’s Cronbach’s Alpha, which verifies the internal consistency by the degree to which the multiple underlying items relate to each other and whether they could be combined into a factor. General rule of thumb is that a minimum Cronbach’s Alpha of 0.6 is acceptable (Hair et al., 1998). Even a Cronbach’s Alpha above 0.5 should be sufficient, according to Berendsen et al. (2009). Therefore, the Cronbach’s Alphas for the variables, which vary between 0.590 and 0.913, are sufficient, see Table 1. As a result, both the factor and reliability analysis guarantee a high construct validity of this research and combining the items into factors is unproblematic.

Table 2

Descriptive statistics of the dependent and independent variables

Variable Items Theoretical range Actual range Mean SD

(17)

17

After transforming the individual items into sum variables, of which descriptive characteristics are presented in Table 2, correlations and linear regression analyses are used to test the hypotheses.

4.2 Linear regression analyses

As presented in Table 3, the correlations between the variables range from -0.325 till 0.612 and the correlations between the independent variables range from -0.154 till 0.251. Therefore, there is no need to assume that multicollinearity will have adverse effects on the coefficients in the regression analyses (Mansfield and Helms, 1982).

As reflected by the correlations of the variables that compose the hypotheses, which are bold in Table 3, an increased attention towards benchmarking information is associated with an increase in the attention that the agent pays to benchmarking. Also, a stronger link between incentives and benchmarking scores is both associated with a decrease in the agent’s attention towards benchmarking and a reduced focus on reported performance. Both of these correlations are, however, insignificant. Finally, the external usage of benchmarking is associated with an increased focus on reported performance and an even stronger increased focus on real organisational performance.

Table 3 Correlation matrix 1 2 3 4 5 6 7 8 9 10 11 12 1: ATT_PRINCIPAL 1 2: ATT_AGENT .439*** 1 3: LINK_INC_SCORES .077 -.024 1 4: EXTERN_USAGE .251** .612*** -.154 1 5: REPORTED_PERF .131 .368*** .015 .277** 1 6: REAL_PERF .137 .517*** .071 .437*** .425*** 1 7: GENDER .045 .106 .162* _.280*** _.264** _.227** ₁ 8: AGE -.325*** _-.178* _-.030 _-.212** _-.165* _-.232** _-.035 ₁ 9: SIZE .092 .279** .036 .134 .159* -.051 0 -.018 1 10: SUBSECTOR .094 .335*** .033 .273* .141 .204** .267** -.075 .149 1 11: #_PROJECTS .110 .090 -.241** .179* .230** -.136 -.018 -.019 .240** -.151 1 12: SCORES .031 .336*** -.169* .062 .364*** .181* -.114 .082 .211** .013 .147 1 *: p < 0.10; **: p < 0.05; ***: p < 0.01 (one-tailed). N = 69.

(18)

18

and the benchmarking scores (β = 0.737 and respectively β = 0.447). Agents of healthcare organisations are, compared to agents of municipalities, paying more attention to benchmarking information. Furthermore, higher benchmarking scores are associated with an increased attention being paid to benchmarking information.

Table 4

Regression results hypothesis 1 and 2

Model 1 Model 2 β SE β SE Intercept 2.809*** .626 -.377 1.287 Independent variables ATT_PRINCIPAL .493*** .123 .414*** .119 LINK_INC_SCORES -.060 .114 -.027 .109 Control variables GENDER .146 .271 AGE -.119 .216 SIZE .207 .157 SUBSECTOR .737** .308 #_PROJECTS .005 .114 SCORES .447*** .157 R square .196 .406 Δ R square .210 **:_{p < 0.05;}***:_{p < 0.01} Table 5

Regression results hypothesis 3 and 4

Model 1 Model 2 β SE β SE Intercept 2.945*** .618 .178 1.220 Independent variables LINK_INC_SCORES .057 .116 .094 .112 EXTERN_USAGE .235** .098 .099 .100 Control variables GENDER .590** .285 AGE -.293 .212 SIZE .004 .159 SUBSECTOR .139 .320 #_PROJECTS .181 .116 SCORES .547*** .159 R square .080 .315 Δ R square .235 **:_{p < 0.05;}***:_{p < 0.01}

(19)

19

significant relationship disappeared, and only the agent’s gender and the benchmarking scores were significantly influencing (β = 0.590 and respectively β = 0.547) the agent’s attempts to influence reported performance. As a result, H4 is rejected with the note that without control variables the relation would have been significant.

Finally, it has been indicated that the external application of benchmarking significantly influences (β = 0.354) the agent’s attempts to influence reported performance, which confirms H5. In addition, higher benchmarking scores show a significant increase (β = 0.333) in the attempts of the agent to influence reported performance.

Table 6

Regression results hypothesis 5

Model 1 Model 2 β SE β SE Intercept 3.262*** .466 2.951*** 1.102 Independent variable EXTERN_USAGE .354*** .089 .320*** .096 Control variables GENDER .302 .271 AGE -.316 .206 SIZE -.151 .153 SUBSECTOR .084 .310 #_PROJECTS -.204 .110 SCORES .333** .153 R square .191 .324 Δ R square .133 **:_{p < 0.05;}***:_{p < 0.01}

5. Discussion and Conclusion

The goal of this study is to examine whether agency theory could be used to improve the understanding of public organisations’ benchmarking information, which should lead to improvements in organisational performance as an ultimate goal. This section will elaborate on the results by reconciling back to the literature review section. Moreover, theoretical and managerial implications will be provided to place the results into perspective. Finally, limitations and recommendations for further research will finish this paper.

5.1 Discussion

(20)

20

information is improving benchmarking efficiency, which will eventually lead to organisational performance improvements.

Whereas the principal’s attention towards benchmarking information has a significant influence on the attention being paid by the agent towards benchmarking information, the link between benchmarking scores and incentives is insignificantly related to the agent’s attention towards benchmarking information. Moreover, the indicated relationship was slightly negative, which contrasts with the positive relationship that has been proposed. Therefore, the reasoning from which the hypothesis originates is not confirmed. Consequently, it is suggested that the use of incentives to increase benchmarking efficiency is improper.

Furthermore, the attention being paid to benchmarking information by the agent is dependent on the specific subsector within the public sector in which the agent’s organisation operates. This study indicated that agents of healthcare organisations are paying more attention to benchmarking information compared to agents of municipalities. The reason why this distinction has been observed is unclear. Furthermore, the organisational benchmarking scores are also influencing the agent’s attention. Higher scores cause an agent to pay more attention to benchmarking information. An explanation could be that agents want to emphasise their good performance by paying more attention to the information from which their performance is extracted.

Based on relevant prior studies, also a second influence of the link between benchmarking scores and incentives was expected, which is the influence on the agent’s attempts to influence reported organisational performance in particular. Tillema’s (2010) findings were used as base for this relationship. Dysfunctional behaviour of agents was expected to cause self-interested behaviour, which would rather result in improvements of reported performance than in improvements of real performance. As the hypothesised relationship has been rejected, the issue of dysfunctional behaviour in this study’s setting could be questioned. Further research should address the issue of dysfunctional behaviour to explore whether its existence is dependent on specific contextual factors.

(21)

21

for public organisations, as indicated by Magd and Curry (2003), could be triggered by the external comparison of benchmarking scores.

The influence of not using benchmarking for external comparison is, however, expelled by the influence of two control variables. Therefore, it should be mentioned that the reasoning as provided above should be carefully interpreted. The attempts of agents to influence reported performance in particular is also dependent on the agent’s gender and the organisational benchmarking scores. Although the influence of the agent’s gender was only significant at a 5%-significance level, it is indicated that men are, compared to women, less likely to influence reported organisational performance. As the consulted prior literature did not report any differences based on gender diversity within a benchmarking context, no explanation could be provided for why men put more effort in attempts to influence reported performance. Furthermore, the level of benchmarking scores influences the agent’s attempts to influence reported performance as well. As those attempts are likely to improve reported scores, a self-enhancing process is activated. However, as the reported performance also includes real organisational performance, the underlying processes could be improved as well. Nevertheless, according to the reasoning as provided above, it is suggested that reported performance improvements reflect rather increased scores than improvements in the underlying processes.

Finally, it has been indicated that not using benchmarking for external comparisons influences the degree to which agents attempt to influence real organisational performance. As the relationship is positive, it is suggested that performance improvements are of high importance, which extends the findings of Van Thiel and Leeuw (2002). In the case of excluding external comparison of benchmarking scores, both principals and agents seem to value real organisational performance over reported performance, which is reflected by a rejection of the fourth hypothesis and a confirmation of the fifth hypothesis. More specifically, an internal application of benchmarking induces agents to reinforce their attempts to improve real organisational performance. As principals will prefer real organisation improvements over reported performance improvements, the distinction as made by Geerlings et al. (2006) should get more attention. In their distinction, the lowest level of benchmarking that only reflects an internal application should be central, as this will maximise performance for public organisations. Moreover, one of the two different applications of benchmarking that Askim et al. (2008) mentioned should be central, that is the monitoring application of benchmarking. As a result, this study’s results support the findings of Kyrö (2003) and Cowper and Samuels (1997), who stated that comparisons with similar organisations are not very effective.

(22)

22

5.2 Conclusion

Returning to this study’s research question, ‘How can agency theory be used to improve our understanding of the use of benchmarking information in the public sector and, thereby, of possibilities to enhance public organisations’ performance?’, at least some applicability of agency theory is indicated. Firstly, the efficiency of benchmarking seems to be improved by a monitoring application of benchmarking, which enables the principal to track performance of the agent. Consequently, the agent is induced to improve benchmarking scores, as those scores reflect his or her own performance, which thus at the same time results in improvements in organisational performance.

Secondly, whereas a stronger link between benchmarking scores and incentives was expected to improve the agent’s attention towards benchmarking information and to stimulate the agent to influence reported performance, these relationships have not been confirmed. Therefore, the link between benchmarking scores and incentives would not be a helpful grip in increasing benchmarking efficiency.

Furthermore, it is stated that in a situation where benchmarking scores are not compared to external organisations the agent’s main focus is to influence real organisational performance. Obviously, the agent’s and principal’s interests are more or less aligned. As a result, no further specific actions are required to align interests regarding benchmarking, which supports the previous findings that the link between benchmarking scores and incentives has no influences.

To conclude, this study indicated that the examination of agency theory’s use in the context of benchmarking for public organisations provided several important implications for managers. The highest benchmarking efficiency seems to be achieved when the principal pays much attention to benchmarking information and when benchmarking is not used for external comparison. Both agents and principals of public organisations should incorporate, or at least be aware of, those implications to achieve an increased benchmarking efficiency, which, consequently, improves organisational performance.

5.3 Limitations

(23)

23

sector would have their own results as well. Deviating results could be caused by different characteristics of the subsectors. For example the level of competitiveness could be relevant, as indicated earlier. Moreover, the degree to which an organisation is resistant to change is suggested to have an important influence on benchmarking efficiency, which will further impact the results. In addition, as this study’s respondents consist of only Dutch agents that represent Dutch organisations, findings are reflecting only a minor part of the whole public sector. Therefore, applying the results to the entire public sector is hazardous. More specifically, generalising the results to other Western countries would not have to be problematic. However, applying the results to Eastern countries could be out of place, as those countries differ considerably in, for example, culture. Consequently, further research should address the influence of contextual factors, including cultural factors. Finally, the research instrument of this study has not been applied before. Reliance on a self-constructed questionnaire could be risky, as its quality has not been proved yet. However, as no existing instruments have been located in prior literature, there was no other option than to construct a new questionnaire loosely connected to prior literature.

5.4 Recommendations for further research

The limitations that have been mentioned are on the basis of directions for further research. As can be stated, the limited number of respondents together with the selective choice of subsectors within the public sector restrained generalisability. Therefore, further research should explore whether the application of agency theory would improve the understanding of public organisations’ use of benchmarking information in a more general context. Replicating this study’s research design in a different contextual setting would increase generalisability and, therefore, improve our understanding of agency theory’s synergy with the use of public organisational benchmarking.

Furthermore, whereas Tillema (2010) has indicated dysfunctional behaviour in her study, the issue of dysfunctional behaviour has not been proved in the current research setting. Consequently, it became obscure whether dysfunctional behaviour of agents is a relevant issue for public organisations. Therefore, further research should examine whether dysfunctional behaviour exists in a broader research setting, which could be achieved by composing a research design based on Tillema’s (2010) paper and the present paper and applying that on a different set of subsectors or an international research setting to extend the current results.

Although this paper contains several limitations and further research should extend this paper’s findings to a broader context, this paper is regarded as a step in the right direction. Eventually, a thorough understanding of how agency theory could be used to improve the understanding of public organisations’ benchmarking information will be achieved, which will lead to improved organisational performance ultimately.

(24)

24

References

Adcroft, A., Willis, R., 2005. The (un) intended outcome of public sector performance measurement. International Journal of Public Sector Management 18(5), 386-400.

Andersen, B., Henriksen, B., Spjelkavik, I., 2008. Benchmarking applications in public sector principal-agent relationships. Benchmarking: An International Journal 15(6), 723-741.

Askim, J., Johnsen, Å., Christophersen, K. A., 2008. Factors behind organizational learning from benchmarking: Experiences from Norwegian municipal benchmarking networks. Journal of Public Administration Research and Theory 18(2), 297-320.

Ball, A., Bowerman, M., Hawksworth, S., 2000. Benchmarking in local government under a central government agenda. Benchmarking: An International Journal 7(1), 20-34.

Baruch, Y., Holtom, B. C., 2008. Survey response rate levels and trends in organizational research. Human Relations 61(8), 1139-1160.

Berendsen, A. J., Groenier, K. H., De Jong, G. M., Meyboom-de Jong, B., Van der Veen, W. J., Dekker, J., Schuling, J., 2009. Assessment of patient's experiences across the interface between primary and secondary care: consumer quality index continuum of care. Patient Education and Counseling 77(1), 123-127.

Bland, J. M., Altman, D. G., 1997. Statistics notes: Cronbach's alpha. Bmj, 314(7080), 572.

Blanton, H., Buunk, B. P., Gibbons, F. X., Kuyper, H., 1999. When better-than-others compare upward: choice of comparison and comparative evaluation as independent predictors of academic performance. Journal of Personality and Social Psychology 76(3), 420-430.

Bowerman, M., Francis, G., Ball, A., Fry, J., 2002. The evolution of benchmarking in UK local authorities. Benchmarking: An International Journal 9(5), 429-449.

Christen, M., Iyer, G., Soberman, D., 2006. Job satisfaction, job performance, and effort: a reexamination using agency theory. Journal of Marketing 70(1), 137-150.

Chua, J. H., Chrisman, J. J., Bergiel, E. B., 2009. An agency theoretic analysis of the professionalized family firm. Entrepreneurship Theory and Practice 33(2), 355-372.

Cowper, J., Samuels, M., 1997. Performance benchmarking in the public sector: the united kingdom experience. Benchmarking, Evaluation and Strategic Management in the Public Sector, 11-32.

Cox III, E. P., 1980. The optimal number of response alternatives for a scale: a review. Journal of Marketing Research 17(4), 407-422.

(25)

25

Eisenhardt, K. M., 1989. Agency theory: an assessment and review. Academy of Management Review 14(1), 57-74.

Eisenhardt, K. M., Graebner, M. E., 2007. Theory building from cases: Opportunities and challenges. Academy of Management Journal 50(1), 25-32.

Elnathan, D., Lin, T. W., Young, S. M., 1996. Benchmarking and management accounting: a framework for research. Journal of Management Accounting Research 8, 37-54.

Evans, J. R., Mathur, A., 2005. The value of online surveys. Internet research 15(2), 195-219.

Fox, K. J., 1999. Efficiency at different levels of aggregation: public vs. private sector firms. Economics Letters 65(2), 173-176.

Garland, R., 1991. The mid-point on a rating scale: is it desirable. Marketing bulletin 2(1), 66-70.

Geerlings, H., Klementschitz, R., Mulley, C., 2006. Development of a methodology for benchmarking public transportation organisations: a practical tool based on an industry sound methodology. Journal of cleaner production 14(2), 113-123.

Hair, J.F., Anderson, R.E., Tatham, R.L., Black, W.G., 1998. Multivariate Data Analysis, fifth ed. Printice Hall, New Jersey.

Holden, M. T., Lynch, P., 2004. Choosing the appropriate methodology: understanding research philosophy. The Marketing Review 4(4), 397-409.

Jensen, M. C., Meckling, W. H., 1976. Theory of the firm: managerial behavior, agency costs, and ownership structure. Journal of Financial Economics 3(4), 78-79.

Kouzmin, A., Löffler, E., Klages, H., Korac-Kakabadse, N., 1999. Benchmarking and performance measurement in public sectors: towards learning for agency effectiveness. International Journal of Public Sector Management 12(2), 121-144.

Kyrö, P., 2003. Revising the concept and forms of benchmarking. Benchmarking: An International Journal 10(3), 210-225.

Lazear, E. P., Shaw, K. L., 2007. Personnel economics: the economist's view of human resources (No. w13653). National Bureau of Economic Research.

Magd, H., Curry, A., 2003. Benchmarking: achieving best value in public-sector organisations. Benchmarking: An International Journal 10(3), 261-286.

Mansfield, E. R., Helms, B. P., 1982. Detecting multicollinearity. The American Statistician 36(3a), 158-160.

(26)

26

Otley, D., 1999. Performance management: a framework for management control systems research. Management accounting research 10(4), 363-382.

Rainey, H. G., Backoff, R. W., Levine, C. H., 1976. Comparing public and private organizations. Public Administration Review 36(2), 233-244.

Schiele, J. J., McCue, C. P., 2006. Professional service acquisition in public sector procurement: a conceptual model of meaningful involvement. International Journal of Operations & Production Management 26(3), 300-325.

Tillema, S., 2010. Public sector benchmarking and performance improvement: what is the link and can it be improved? Public Money & Management 30(1), 69-75.

Triantafillou, P., 2007. Benchmarking in the public sector: a critical conceptual framework. Public Administration 85(3), 829-846.

Van Aken, J., Berends, H., Van der Bij, H., 2012. Problem solving in organizations: A methodological handbook for business and management students, second edition. Cambridge University Press.

Van Thiel, S., Leeuw, F. L., 2002. The performance paradox in the public sector. Public Performance & Management Review 25(3), 267-281.

Varol, M., 2011. Assessment of heavy metal contamination in sediments of the tigris river (turkey) using pollution indices and multivariate statistical techniques. Journal of Hazardous Materials 195, 355-364.

Verbeeten, F. H., 2008. Performance management practices in public sector organizations: impact on performance. Accounting, Auditing & Accountability Journal 21(3), 427-454.

Vining, A. R., Boardman, A. E., 1992. Ownership versus competition: efficiency in public enterprise. Public Choice 73(2), 205-239.

(27)

27

Appendix

I: Questionnaire (abbreviated)

Dear respondent,

First of all, we would like to thank you for taking time to respond to this questionnaire. This is of utmost importance to our researches. The subject of the questionnaire is benchmarking in the public sector. Completion of the questionnaire will take about 5-10 minutes. The provided information will be treated with care. Consequently, anonymity will be guaranteed. If you would like to receive a summary of the researches’ results, please send an email to; Remco Klapdoor: r.klapdoor@student.rug.nl; or Remi Rutgers: r.rutgers.1@student.rug.nl, and we will contact you in due course. Hereby, we are able to ensure that your email address cannot be linked to the entered information which assures your anonymity.

1. Principal’s attention to benchmarking (ATT_PRINCIPAL)

Please indicate the extent to which you agree with the following statements (1 = strongly disagree, 7 = strongly agree).

‒ The principal studied the benchmarking scores very carefully. ‒ The principal finds the benchmarking scores important.

‒ The principal was continuously emphasising the importance of good benchmarking scores. ‒ The principal stimulated to improve benchmarking scores.

2. Agent’s attention to benchmarking (ATT_AGENT)

‒ My organisation is taking benchmarking very seriously. ‒ My colleagues and I study the benchmarking scores explicitly.

3. Link between benchmarking scores and incentives (LINK_INC_SCORES)

‒ The agent’s financial compensation depends on benchmarking scores. ‒ The agent’s non-financial compensation depends on benchmarking scores. ‒ Benchmarking scores are included in an agent’s job evaluation.

4. No external application of benchmarking (EXTERN_USAGE)

‒ Our organisation compares benchmarking scores with external organisations. ‒ External organisations influence our benchmarking projects. *

5. Influence reported performance (REPORTED_PERF)

(28)

28

‒ The main improvements due to benchmarking are improvements in reported scores.

6. Influence real performance (REAL_PERF)

‒ We use benchmarking to improve internal business processes.