Conceptualizing dysfunctional consequences of performance measurement in the public sector

ABSTRACT

Performance measurement (PM) has become increasingly popular in the management of public sector organizations (PSOs). This is somewhat paradoxical considering that PM has been criticized for having dysfunctional consequences. Although there are reasons to believe that PM may have dysfunctional consequences, when they occur has not been clarified. The aim of this research is to conceptualize the dysfunctional consequences of PM in PSOs. Based on complementarity theory and contingency theory we conclude that dysfunctional consequences of PM are a matter of interactions between PM design and PM use, between control practices in the control system and between PM and context.

KEYWORDS:

• Performance measurement
• dysfunctional consequences
• control system

Introduction

Performance measurement (PM) has become increasingly popular in the management of public sector organizations (PSOs) and is claimed to be an indispensable part of implementing strategies in these organizations (Bouckaert and Peters 2002; Johnsen 2005; Lapsley 2008; Broadbent and Guthrie 2008; van Helden, Johnsen, and Vakkuri 2008; Liguori, Sicilia, and Steccolini 2012; Arnaboldi, Lapsley, and Steccolini 2015; Pollanen et al. 2017). PM refers to measuring of output, outcome, efficiency, effectiveness and equity at various levels of organizations (cf. Johnsen 2005) and has a role in most organizations’ learning and motivation processes. It provides information on objectives and accomplishments and enables employees, managers and politicians to visualize success. To some, PM is considered vital for making the public sector more controllable by increasing accountability for performance among managers and professionals (Hood 1996; De Bruijn 2002; Diefenbach 2009; Van Hengel, Budding, and Groot 2014) and for enabling politicians and senior management to influence the production of public services with strategic prioritizations (Hood 1996; Johnsen 2005). PM has proliferated to the point where it is considered a natural part of the control systems in most PSOs (Johnsen 2005; Lapsley 2008).

This is somewhat paradoxical considering that in recent decades PM has been criticized for having dysfunctional consequences in the public sector. A core reason is that PM in PSOs is typically incomplete, that is, it misses important performance dimensions and fails to inflict behaviour in a manner consistent with organizational strategies and goals (Gibbons 1998). PM is commonly criticized for blocking innovations, causing gaming, leading to means-end inversion and producing inaccurate representations of performance (Bouckaert and Balk 1991; Smith 1995; De Bruijn 2002; Greener 2005; Chang 2006, 2015). Influential commentators have argued that PM may be a measure for pseudo-control (Hofstede 1981), a fatal remedy (Power 2004), and a source of dysfunctional and contentious effects (Lapsley 2008; Diefenbach 2009; Chang 2015).

The dysfunctional consequences arise as unwanted side effects of top management ambitions to influence behaviour through PM. The need for PM as well as other control practices rests on the assumption that coordinated, goal-congruent action in organizations does not take place automatically. To manage the performance of any organization, tools for information and motivation are needed. However, the public sector is an area of inherent complexity (Arnaboldi, Lapsley, and Steccolini 2015; Lapsley and Skaerbaek 2012), which means specific challenges when designing management tools (Boyne 2002; Jakobsen et al. 2018).

Although there are strong reasons to believe that PM may have dysfunctional consequences in PSOs, when they occur has not been clarified (cf. Broadbent and Guthrie 2008; Arnaboldi, Lapsley, and Steccolini 2015). On a conceptual level PM sceptics seem to suggest rather simple theories, where dysfunctional consequences from PM are more or less unavoidable. Others argue that dysfunctional consequences are dependent on the level of PM incompleteness. Still other findings indicate that dysfunctional consequences from PM are contingent upon how measurements are subsequently used by higher-level managers in appraisal and sanctioning processes (De Bruijn 2002; De Bruijn and Van Helden 2006; Greener 2005; Chang 2006, 2015).

We argue that the (implicit) theory of dysfunctional consequences of PM in PSOs is underdeveloped, with the result that PM risks are exaggerated and efforts to mitigate the problem are misled. The theory is incomplete in disregarding how other control practices and contextual conditions interact with PM. This is a recurrent problem in empirical research on PM and control which is continuously criticized for being reductionist (Otley 1980; Chenhall 2003; Malmi and Brown 2008; Cuganesan, Guthrie, and Vranic 2014). Some researchers have opened up for the possibility that PM dysfunctions may be reduced by inclusion of non-PM controls (Bevan and Hood 2006; Kelman and Friedman 2009), and a few empirical studies have noted interactions between control practices (Diefenbach 2009; Rautinen and Järvenpää 2012; Van Hengel, Budding, and Groot 2014; Arnaboldi, Lapsley, and Steccolini 2015). However, interactions between PM and other control practices and between PM and context have so far not been conceptualized in research.

The aim of this research is to conceptualize the dysfunctional consequences of PM in PSOs. Theoretically, our conceptualization builds on complementary theory and contingency theory which entail that dysfunctional consequences of PM are a matter of internal fit between PM and other control practices in the control system and external fit between PM and context (Drazin and Van de Ven 1985; Grabner and Moers 2013). We illustrate and refine our tentative conceptualization by conducting a case study of PM, other control practices, and context in a large Swedish local government organization.

In the next section, we provide our conceptualization of dysfunctional consequences of PM. This is followed by a method section where we describe how we use a case to illustrate and refine the conceptualization. Thereafter, we present the case. In the discussion section, we analyse the case results. The paper ends with conclusions and suggestions for future research.

Theorizing risks of dysfunctional consequences of performance measurement

In this section, we conceptualize the dysfunctional consequences of PM. We first account for complementary theory and contingency theory, on which our conceptualization rests. Then, we describe three different types of dysfunctional consequences. Thereafter, we theorize how dysfunctional consequences of PM are dependent on 1) interactions between PM and other practices in the control system and 2) interactions between PM and context. Last, we present a tentative conceptual model.

Complementary theory and contingency theory

Both complementary theory and contingency theory are occupied with the notion of ‘fit’, but while contingency theory relates to how control practices fit with context, complementary theory relates to how control practices fit with each other (Grabner and Moers 2013).

Control systems consist of two or more interdependent control practices. A control practice is the actual use of management control tools, such as budgets, performance measurements and rules (cf. Grabner and Moers 2013). Control practices are choice variables for individual organizations, which means that they can be changed if considered necessary. Interdependence means that the effect of one control practice (e.g. dysfunctional consequences of PM) depends on the existence of other control practices (Drazin and Van de Ven 1985; Milgrom and Roberts 1995; Grabner and Moers 2013). Interdependence can play out in two ways. Control practices are complements if the effect on an outcome variable – which may be learning, motivation, efficiency or something else – increases when both control practices are present (are used to a high extent). They are substitutes if the outcome of one control practice decreases when the other control practice is present (is used to a high extent), and vice versa (Grabner and Moers 2013).

Besides the overall guidance provided by complementary theory, there is no detailed theory about how dysfunctional consequences of PM are influenced by other control practices. Hypotheses about interactions remain to be posed and tested. There are, however, empirical studies that to some extent are informative about interactions between control practices and therefore useful for tentative theorizing. In the conceptualization below, we chose to focus on how control practices exacerbate or buffer dysfunctional consequences of PM. We refrain from using the concepts ‘complements’ and ‘substitutes’ since they become somewhat confusing when the outcome variable of interest is not beneficial for the organization but an unwanted side effect.

Contingency theory rests on the notion that the degree of fit between contextual variables and control practices has performance consequences (Chenhall 2003; Gerdin and Greve 2004). This means that dysfunctional consequences of PM are not only a matter of interactions within the control system, but on how control practices fit with context. Control systems may cause dysfunctional consequences in one context, but not in another. Inversely, in one context a control system may cause only minor dysfunctional consequences, but in another context, these consequences may be substantially exacerbated. Contextual conditions are exogenous to the individual organization and therefore not choice variables. This means that civil servants and politicians in PSOs must adapt control practices to context.

Dysfunctional consequences

In the PM literature, dysfunctional consequences of PM are described with many different concepts, which makes a choice necessary. To capture all dimensions of dysfunctional consequences we chose the all-embracing typology presented by Merchant and Van der Stede (2017). They divide dysfunctional consequences into dysfunctional behaviour (gamesmanship and behavioural displacement) and negative attitudes.

Gamesmanship is defined as behaviour where accountable persons knowingly try to manage or manipulate the control system to look more favourable and reap unearned positive consequences (Jaworski and Young 1992; Merchant and Van der Stede 2017). Gamesmanship follows as a risk from decentralization, as it involves making resources available to subordinates. PM may be involved in gamesmanship by being actively managed to ensure unearned tangible or intangible rewards.

Behavioural displacement occurs when managers and employees are misled by incomplete PM. In many organizations, PM has an important role in providing information, which means that the potential risk of behavioural displacement as a consequence of incomplete PM is substantial (Otley 2003; Horngren 2004). Typical examples of behavioural displacement caused by PM are myopia, suboptimization and means-end inversion (Smith 1995; Merchant and Van de Stede 2017).

Negative attitudes are related to job tension, conflict, frustration and resistance (Merchant and Van der Stede 2017). Such attitudes may be the outcome of PM systems containing unrealistic or unfair targets, an incomplete set of performance measures that fails to register and appreciate effort, and unattractive rewards. This may cause a lack of motivation and unwillingness to cooperate, e.g. it may decrease subordinates’ interest in using their knowledge to further develop the organization. If negative attitudes arise from PM it is a serious paradox effect since the intended purpose of PM is to increase motivation (Otley 2003; Horngren 2004).

Interaction between PM and other control practices

A common opinion among public sector scholars is that PM in PSOs is typically incomplete, i.e. cannot capture all relevant dimensions of organizational performance (Gibbons 1998). Despite ambitious attempts in many PSOs to develop better PM through comprehensive performance measurement systems, such as the balanced scorecard and similar techniques, the complexity (Boyne 2002; Jakobsen et al. 2018) and low measurability (Johansson and Siverbo 2018) of many activities means that performance cannot be fully captured. Although PM can be made less incomplete, the remaining incompleteness means that there will always be a risk that managers hit the target but miss the point. The fundamental problem of incompleteness is what makes many scholars fear that PM inevitably causes gamesmanship, behavioural displacement and negative attitudes.

However, from a complementary theory and systems perspective, dysfunctional consequences are not only a matter of PM design, but also on how PMs are used and how the PM practice interacts with other control practices in the control system. Starting with the use aspect of PM, as mentioned in the introduction, previous research on PSOs has suggested interactions between PM design and how tightly PM is used by superiors in the organisation. It is generally understood that tight control¹ exacerbates dysfunctional consequences of incomplete PM (Gibbons 1998; De Bruijn 2002; De Bruijn and Van Helden 2006). Control tightness increases as superior managers or politicians use PM as a basis for evaluating subordinate managers and when PM results are used for reimbursement purposes, e.g. performance-based budgeting (cf. Jakobsen et al. 2018). Greener (2005) and Chang (2006, 2015) show how tight control related to UK NHS [National Health Service] trusts’ waiting time targets, in the form of strong political pressure, ‘naming and shaming’ procedures, financial incentives and threat of replacing managers, resulted in gaming, reduced focus on quality and negative attitudes. Kastberg and Siverbo (2007) and Conrad and Uslu (2011) illustrate how the introduction of activity-based payment increased attention paid to PM, sometimes with dysfunctional consequences. The explanation appears to be based on economics and agency theory, since dysfunctional consequences are expected to occur when incentivized (Gibbons 1998; Kerpershoek, Groenleer, and de Bruijn 2016). Conceptually this means that tight use of PM creates or exacerbates dysfunctional consequences of PM and that loose control buffers from these consequences.

Second, dysfunctional consequences of PM may depend on how PM interacts with budget control, which is a fundamental control practice in PSOs (Johansson and Siverbo 2014; Arnaboldi, Lapsley, and Steccolini 2015). The budget control practice is about allocating financial resources, measuring spending and acting on overspending. It is similar to PM practice with the important exception that budget control is about restrictions and not about operational performance. PM is often subordinated to budget control in the sense that managers’ first priority is managing budget targets (Lapsley 2008; Rautiainen and Järvenpää 2012; Van Hengel, Budding, and Groot 2014; Arnaboldi, Lapsley, and Steccolini 2015). A suggested explanation is the centrality of budgets in the life of public sector organisations, due to tradition, the size of PSOs and the level of sophistication in financial control (Arnaboldi, Lapsley, and Steccolini 2015). What impact the budget control practice has on dysfunctional consequences of PM, however, is still an open question. On the one hand, if budget control is tight, that is, compliance with budget targets is a primary goal, it may exacerbate the dysfunctional behaviour of PM. The reason would be that controlees without necessary funding must cut corners to at least seemingly manage operational targets. On the other hand, tight budget control may mean that all attention is put on finances. In this case, there may be less interest in (fewer incentives to) acting dysfunctional as a consequence of PM. Rather, budget control would buffer dysfunctional consequences.

Third, the extent to which PM causes dysfunctions may well depend on the co-existence of cultural controls implemented by the PSO. These controls are often emphasized when complex and qualitative services are provided. Through the establishment of common views, based on organizational core values, cultural controls inform in an indirect way when interpretation of the situation at hand is required and create a kind of mutual monitoring within work groups (Merchant and Van der Stede 2017). In situations where PM incentivizes dysfunctional consequences, this will not occur because controlees are primed with norms and values that contradict dysfunctional consequences and possible violations will be detected by peers. Examples of measures PSOs may take to establish and maintain a positive culture are careful recruitment, internal socialization efforts and communication of vision, missions and value statements. In other words, dysfunctional consequences of PM are less likely if the PSO practices cultural controls that ‘create’ stewards (see Davis, Schoorman, and Donaldson 1997; Kerpershoek, Groenleer, and de Bruijn 2016). Since cultural controls may alleviate the negative effects of PM, they buffer possible dysfunctional consequences induced by PM.

Fourth, most if not all PSOs develop internal behavioural controls which are more or less prescriptive instructions about what is allowed and how to act. They represent a direct form of control, reducing the leeway and areas for interpretation for employees, which may be appropriate when quality, fairness and equal treatment are priorities. Examples of internal behavioural controls are standard operating procedures (SOPs) and local policies, rules and regulations (Kennedy and Widener 2008; Macintosh and Daft 1987). The primary role of behavioural controls is to provide information that directly guides behaviour in order to increase efficiency or reduce risk. In a similar vein as cultural controls, internal behavioural controls may have a buffering effect on the relationship between PM and dysfunctional consequences. For example, if PM incentivizes under-treatment of patients in healthcare organizations, e.g. in order to maintain productivity, this dysfunction may only appear in the absence, but not in the presence, of internal behavioural controls.

As mentioned, more theorizing is necessary for making formal predictions about how combinations of control practices affect dysfunctional consequences. We provide tentative ideas on how PM design interacts with PM use and with other control practices of which some may exacerbate and some may buffer dysfunctional consequences.

Interaction between PM and context

The literature on context–control fit is huge. Typical context variables that have been examined in the general management control literature are environmental uncertainty, technology, structure, size, strategy and national culture (Chenhall 2003). However, as is the case with complementary theory, there is no specified theory about context–PM implications for dysfunctional consequences in PSOs. There are strong reasons to expect that such implications exist, but at this point, detailed predictions of the importance of individual context variables cannot be made. Still, tentative reasoning is possible to illustrate how the effects of PM on dysfunctional consequences may be moderated by context.²

When considering context, PSOs’ long tradition of being bureaucracies must be noted. It resembles a core value in public administration of rectitude, i.e. the rule of law (Hood 1991; Hood and Jackson 1991). Governmental legislation and guidelines are highly attention directing for employees and managers within PSOs, which means that bureaucracy is an important contextual condition that impinges on the controlling effect of PM and other control practices. Since members of bureaucracies are reluctant to violate rules, dysfunctional behaviour induced by PM which means non-compliance with legislation, guidelines and other rules become less likely. However, it is important to underline that the extent and intensity of bureaucracy vary within and between PSOs depending on the type of service provided. Complex services such as education, social services, elderly care and health care are often more regulated than hard, technical services. Less regulated services can rely less on bureaucracy to buffer from dysfunctional consequences of PM. At the same time, it is important to acknowledge that complexity requires ‘case-specific judgment, knowledge, and discretion’ (Jakobsen et al. 2018, 132) which means that the set of bureaucratic rules neither can nor should be extensive enough to completely inhibit dysfunctional behaviour.

Consequently, for many public sector activities, it is mandatory to employ professionals, that is, workers with authority and credibility, a code of ethics and a joint culture of values and norms (Greenwood 1957). Especially in complex services, it is mandatory to recruit professionals. They are expected to be intrinsically motivated to act in the best interest of society and to practice self-control (Freidson 2001; Merchant and Van der Stede 2017). For some activities, almost only professionals or ‘stewards’ (Davis, Schoorman, and Donaldson 1997) are recruited and they have shown inclination to obstruct managerialism reforms if these are perceived to cause adverse effects (Lapsley 2008; Kerpershoek, Groenleer, and de Bruijn 2016). In accordance with the reasoning about stewards above, the degree of professionalization has implications for how severe it is reasonable to expect the dysfunctional consequences of PM to be. In control contexts where professionals are active on the operational level, it is reasonable to assume that the space for dysfunctional consequences of PM becomes narrow (cf. Freidson 2001). This justified assumption, however, has not been subject to much empirical research and the limits of professionalization in buffering dysfunctional consequences remain to be studied. For instance, while Kastberg and Siverbo (2016) observed that professionals at a large university hospital refused to give priority to waiting time targets when these targets contradicted their ethos to treat the sickest patient first, the professionals in the NHS seemed to fail to do the same (Greener 2005; Chang 2006, 2015).

The role of professionalization relates to the institutional logics framework (Thornton 2004; Reay and Hinings 2009) where it is emphasized that the outcome of managerial ideas is contingent on the prevailing institutional logics, that is, the existing taken-for-granted rules guiding behaviour. Since the advent of New Public Management (NPM), the professional logic in PSOs has been contested by other logics, for instance, market and corporation logics (Saz-Carranza and Longo 2012: Van Den Broek, Boselie, and Paauwe 2014). Several researchers have pointed out that the outcome of this challenge may be multiple logics within PSOs (Thornton and Ocasio 2008; Reay and Hinings 2009; Pache and Santos 2010; Van Den Broek, Boselie, and Paauwe 2014). This means that the response to PM may be different in PSOs depending on how the professional institutional logic has managed to ward off competing logics. This is exemplified in a case study by Rautiainen and Järvenpää (2012) which indicated that those multiple logics make different responses to PM possible and that PM is more used in organizational units characterized by business logics than professional logics.

Summing up

Our tentative conceptualization is summarized in model form in Figure 1. It suggests that the dysfunctional consequences of PM are contingent on how the PM design element interacts with the PM use element. More specifically, a combination of incomplete PM and tight use by superiors is a high-risk PM practice when it comes to dysfunctional consequences. However, the dysfunctional consequences of PM also depend on interactions with other control practices and context. We suggest the relationship is affected by the level of bureaucracy and professionalization.

Figure 1. The conceptual model.

In the next stage, the tentative conceptual model should be subject to empirical studies, both for illustrating that dysfunctional consequences are contingent on system interactions and context and for the cause of refining the model.

Method

To illustrate the conceptual model on a comprehensive level and to refine it we conducted a qualitative case study of the municipality CoG, one of the largest municipalities in Sweden. Case research is not appropriate for formally testing theoretical models, but is suitable for theory refinement (Vaivio 2007). We do not use the case for standard (statistical) generalizations, but for analytical generalization based on details from the case.

Swedish local government organizations (municipalities and counties) have far-reaching autonomy, to some extent protected by the constitution. They are governed by directly elected politicians and mandated to make decisions on taxes, fees and management systems, for example, PM systems. Within the frame of national laws and regulations, municipalities are mandated to decide on the magnitude, direction and quality of their activities. We selected CoG because of its long experience of working with an ambitious, extensive and formalized PM system. The work had not been without difficulties, and the system had been subject to numerous internal evaluations. Representatives of CoG granted us access to study their PM system as well as other control practices and provided us with recent reports about the control system and with the budget document, which contained information on budget allocations and political objectives.

CoG is organized in a number of city districts, which are responsible for preschool, elementary school, elderly care, social care and leisure activities. City districts are organized in sectors related to the services provided. Each city district has a political committee, which appoints a district manager. Managerial positions beneath the district manager are sector manager, area manager and unit manager. Our main empirical data come from two city districts in CoG: District East and District West. This means we have predominantly collected data about PM in activities characterized by high complexity, that is, where PM typically is incomplete. In District East, there are about 53,000 inhabitants and the department’s yearly spending is about 300 million EUR. District West has 63,000 inhabitants and a turnover of almost 200 million EUR.

At District East, we interviewed five unit managers, five area managers, one accountant, one controller and the city district manager. At District West, we interviewed two unit managers, two area managers, two sector managers, one controller, the HR director and the district manager. In addition, at the central level, we interviewed seven administrators, the finance director, two internal auditors and one union representative (see Appendix for details). In total, we interviewed 33 persons in CoG.

The primary ambition when selecting respondents was to maximize the likelihood of capturing different perspectives on and opinions about the PM practice. Therefore, we interviewed managers responsible for dissimilar activities, managers at different hierarchical levels, staff personnel, users of PM information and system designers at the central level, etc. In the selection of respondents, to some extent, we were advised by previously interviewed persons who were knowledgeable about other persons with interesting perspectives on PM. Except for one HR manager and one HR specialist who referred to time constraints, no one denied participation.

Through the interviews, we wanted to learn about experiences with PM design and use and how it related to other control practices. Of interest was the relative importance of control practices and statements of interactions between them, e.g. accounts like ‘we put all efforts into being productive and have no time to bother about internal policy documents’ (fabricated quote), and reasoning about contextual impacts. Naturally, we were especially interested in experiences of dysfunctional behaviour and negative attitudes related to PM. Our strategy was to not explicitly ask about dysfunctional consequences but await spontaneous statements. The reason was that direct questions about dysfunctional behaviour, especially gamesmanship, could have been insulting. Instead we relied on the conviction that long personal conversations about PM (average 68 min) where there was time to establish a relationship with the respondent, would give good conditions for open and honest descriptions about dysfunction. To increase the likelihood for this even more, we promised all respondents anonymity. Since our respondents gave us several examples of dysfunctional behaviour (but, as will be described later, not because of PM), our strategy seemed to work.

The interviews were organized and analysed assisted by the software program NVivo. Nodes were created for all variables in the conceptual model. Transcribed interviews were analysed, and quotes were referred to the nodes. In this way, all respondents’ statements about each control practice or contextual condition (existence, importance, interactions) were gathered in separate data files. This simplified the treatment of the large data material and reduced the risk of missing important statements in the interviews.

The case of CoG

Our case description starts with a presentation of the formal PM system implemented in CoG, how managers at various levels experienced PM use and to what extent they connected PM with dysfunctional consequences. It continues with a description of other control practices and how the system interacted with context.

Performance measurement design and use

The PM process in CoG started with the budget document, which was decided on by the municipal council. The budget was a comprehensive and superior control document for all activities in the municipality. It consisted of spending limits and objectives, directions (detailed strategies) and tasks for political committees and boards to break down and translate into action. For the year 2017, the council decided on 18 objectives, 160 directions and 85 tasks for committees and boards. The process of breaking down and communicating objectives, directions and tasks was supported by a software called Rappet. With objectives, directions and tasks in mind, managers at all hierarchical levels were required to plan and decide on performance measures for their operations. Plans were developed after communication between higher-level and subordinate managers and with support from controllers. Reporting was a bottom-up process where subordinate managers’ reports constituted the bases for higher-level managers’ reports and so on. PM figures were not aggregated automatically but manual work was required. At the final stage, formal reports were handed into the municipal council by all committees and boards.

The general attitude among the interviewed managers was that the intended purpose of the PM system was legitimate. It was reasonable that the political level communicated priorities and received information on performance. However, a typical problem for managers in the selection of PMs was that it was hard to find measures that truly reflected the objectives of the city districts. Many managers were pessimistic about the possibilities to find measures for their activities that really captured the complexities involved.

When it comes down to our level and our staff, we are operative, and it is hard to get into the general objectives. We want targets, relevant for us. But they are hard to find. [Area manager, Property and security]

An area manager of meal services in schools felt she could not find PMs that showed how her activity contributed to fulfilment of overall school objectives of good grades. This made her vulnerable to being pushed towards other priorities, like environmental sustainability.

My manager’s manager started to focus on vegetarian food. She said to my manager that we should make sure that we served more of this. It is cheaper, and it fits well with increasing environmental sustainability, which is something that is a performance area for our municipality. But just doing that, we know that more food is thrown away, not served or not eaten from the plates. We should also focus on pupils eating well to have energy for school. If we serve organic food, they go buy candy instead. [Area manager, Internal service]

In addition, the PM process itself was a concern. One manager claimed that more use of PM for their own purposes required participation from employees in selecting targets, but participation was hindered by a tight time schedule and system difficulties.

I would like to have the unit plan as a living document that is always with us at our workplace meetings. But there is always a lack of time. [Unit manager, Elderly care]

Practical concerns thereby worked together with measurement difficulties, creating a view of PM as a bad reflection of operations for many areas and units.

However, from the interviews, it was evident that the incomplete PM system was rather loosely used. Higher-level managers were relaxed about missed performance targets if responsible managers could explain the reasons why. Poor goal attainment registered in the PM system did not mean managers were replaced. Replacements were primarily caused by budget overruns or if the staff was so discontented that they ‘revolted’. It seemed clear that PMs were not used to evaluate subordinate managers’ performance in an absolute and rigid manner.

When it comes to the municipality objectives…. It is more about that you continue working with them [if they are not reached]. [Unit manager, School]

Performance measurement and dysfunctional consequences

Another impression from the interviews was that dysfunctional behaviour from PM (gamesmanship and behaviour displacement) was not a major problem or concern. According to a controller, PM simply did not influence or change behaviour very much.

It does not matter what performance measurements we use. We work and then we see what ends up in the reporting system. I have never heard anyone say that ‘Oh, now we have a new performance measure, we have to redirect our operations’. Instead it is more like ‘Oh, now we have to measure this instead’. [Controller, Culture]

However, although it seemed clear from the interviews that PM was not tightly used, there were a few stories told of when single PMs abruptly were emphasized. In one case the political level wanted a specific measurement of children’s work environment.

Suddenly, the politicians demand that now you should be a toxin-free preschool, or labelled as ‘green flag’ [‘Eco-Schools’]. They do not say be a sustainable preschool, it is ‘green flag’. (Unit manager, Preschool)

The unit manager saw such initiatives as disturbing. Even if they were perceived as important, they were not considered vital for the local unit and they troubled the local development processes by demanding a specific focus in the ongoing work. This was an example of negative attitudes and risk of behaviour displacement, from a tighter use of PM.

Overall, the PM system appeared to cause more problems with negative attitudes than dysfunctional behaviour. A source of frustration related to PM was the fact that PM caused work without adding any value. Several unit and area managers were dissatisfied with the lack of feedback from local and central politicians on reported performance. The managers questioned if someone at the political level paid attention to their reports. The reason was that, in their opinion, nothing happened at the political level because of reported problems.

I think it [the PM system] is difficult. Is there really anyone that cares about what we report? I must put a lot of effort into making it sound understandable and controllable. But I don’t think anyone reads it. [Area manager, Property and security]

Also, school representatives were displeased that the internal PM system overlapped the national legislation. The priorities of safe school environment and good grades that were included in the municipality’s internal PM system were also included in the Education Act. Besides receiving the same control signal twice, they had to report the same information twice.

The performance measures we have are fairly adapted to what is already in the Education Act and are nothing that contradicts what we work with in the school. Grades, security, a good place to study and health issues. We know that it is important. [Unit manager, School].

Another reason for dissatisfaction was that managers could not focus solely on core purposes related to their activities, such as ‘a worthy life’ and ‘active consumption of culture’, but also had to deal with central priorities such as environmental sustainability, equality, diversity and digitalization.

Summing up on PM, the primary aim of the PM system was to communicate political objectives and to measure achievements. The PM system was accepted by managers but not appreciated. Its incompleteness meant it did not add value, only costs, and caused risks of behavioural displacements. However, the loose use of PM meant low impact and, consequently, minor problems with dysfunctional behaviour. Still, though, the unmet expectations on feedback on reported PM were a source of negative attitudes.

Budget control

If PMs were not used much for manager evaluation, it was totally different with budget control. CoG’s municipal council had formally decided that committees and boards must realize objectives and directions within the economic frames and this decision was very clearly communicated to all managers in the organization. In CoG the tight budget control practice was known to give consequences for individual managers who did not stay within spending limits. Stories of managers being fired for not managing their budgets were well-known to the respondents and contributed to the budget being a major concern.

Meeting the budget is the first priority, it is something we must do. If you do not meet your budget, then you can be removed from your position – that goes for both unit managers and area managers. [Area manager, Education]

The budget is more important than the political performance targets. I have asked around. [Area manager, Health care]

The tight budget practice seemed to cause dysfunctional consequences, as one manager reported.

I know now that we have a budget deficit coming up, but in our forecast, I still feel the pressure to declare that we are in balance. I am really nervous about this; how should I handle this when this budget year ends? When is the time to announce that we are losing money? Two months ago, I gave a realistic forecast to my manager, and then I got the response that this is nothing that you can move forward with. I must report that I see a surplus. And then I need to have an action plan to achieve it. And of course, we work with an action plan, but we won’t make it the whole way, that is my feeling and that is what I would like to report. [Area manager, Internal service]

In contrast, a few managers reported that budget deficits could be accepted if the reason was accomplishment of new and important objectives and tasks. Examples given were a new target for the increased number of full-time positions and a target about organically produced food. However, overall it appeared as if budget control was tightly used, contributing to PM being downplayed.

Internal behavioural controls

According to the budget document, all internal behavioural controls decided by the municipal council were compulsory but subordinate to the budget, that is, budget frames and objectives, directions and tasks. Thus, formally, if there were conflicts between PM that reflected budget goals and internal standard operating procedures, policies and regulations, PMs were supposed to be prioritized.

The interpretation of the importance of internal rules varied between managers and between different rules. Critical commentators argued that mandatory purchasing procedures and joint administrative rules added bureaucracy and did not add much value to the operations. It was generally reported that internal behavioural controls were not always complied with. Sometimes procedures were not understood, sometimes they were not found, and sometimes they were simply not obeyed.

We have lots of policy documents. There are central policy documents and then these are broken down locally. Then they, in the best of worlds, are implemented in a decent way in the units. No, in reality, I would not say we are much controlled by them. [Sector manager, Elderly care]

We have policies about all possible things, smoking…. They are innumerable. I can tell you, I don’t think about them every day. [Area manager, Property and security]

However, attitudes were more positive about quality assurance systems with SOPs and procedures for dealing with deviations.

We have a quality management system that consists of different parts and an important part is deviation management. In our organization, we try to keep it prevalent and work to create a culture where it is okay to make errors. Then we write a deviation report and learn from it. [Unit manager, Elderly care]

Overall, internal behavioural controls did not seem to be more attention catching than PM and there were no stories told of situations in which dysfunctional consequences of PMs were mitigated or exacerbated by internal behavioural controls.

Cultural controls

The central level of CoG had initiated various forms of cultural control. Ambitions were high to establish common ground for how managers and employees in the organization should meet the residents and be open to cooperation and development. This was described in the document ‘Our approaches’ which was communicated recurrently at training programs and recruitments. For the interviewed managers, proper treatment of residents was the core message in the approaches. It was the responsibility of all managers to work with the values of the employees in their activity. What constituted good treatment was discussed at general meetings with staff and at specific meetings with individual employees when problems occurred. In elderly care and health care ‘basic values’ were added, adapted to their specific conditions. So-called dignity guarantees were given to customers where it was promised that staff would call before showing up, wear badges and carry out patient-centred care. In education, managers tried to ‘walk the talk’ by being good examples when it came to treatment.

The way of controlling by influencing basic values was considered important by some respondents, but not by all. Positive managers pointed out the crucial role of cultural controls in complementing targets and rules. According to them, it reminded employees to be customer centred and what the meaning of customer centredness was. Negative or indifferent managers argued that it was difficult for managers to know if the work had been successful and that recent research had shown poor results as a consequence of working with basic values. It was also claimed that the work with implementing ‘Our approaches’ and basic values was less intensive at the time of the interviews than it had been in previous years.

I think it has calmed down. I thought it was very much [five years ago].… There was a lot that came from the central level, you felt ‘this is something you just have to do’. [Area manager, Health care]

One sector manager admitted that he had forgotten the four core value concepts in ‘Our approaches’.

Although the ambitions behind cultural controls had only partly been successful in the city districts the core implication for the purpose of our study was that they sometimes seemed to fill the gaps in formal control systems. In theory, this means they could buffer the dysfunctional consequences of PM, but we received no explicit statements or examples of this in the interviews. The dominant impression was that cultural control did not have any tangible impact on the consequences of PM.

Context of bureaucracy and professionalization

In the studied city districts, many activities were conducted in bureaucratic contexts where national law and regulations were highly influential on daily operations. Core services under the responsibility of city districts, such as schools, social services, health care and elderly care were bureaucratically controlled by, for instance, the Education Act, the Health Care Act and the Social Care Act and by other regulations from governmental agencies. Governmental inspection bodies conducted regular site visits where they evaluated compliance with laws and regulations. Facilities that did not live up to standards were criticized and if they did not improve they could face a fine or even be closed.

Naturally, the districts also worked in accordance with national legislation on recruitment of professionals for operations where this was mandated, even though sometimes it was difficult to recruit eligible employees. Formal education and/or certification was required for working with health care, social care or in schools. In the experience of managers of these services, their professionals generally thought and acted correctly, had a caring attitude and were motivated from being able to help others. The apprehension among managers was that applicants for jobs characterized themselves as professionals whose identity was related to the specific line of work. In general, managers relied on the fact that subordinates wanted to contribute to their workplace. Superior managers’ focus was to enable employees’ work and to try to remove issues that could undermine their motivation.

External regulations were often mentioned as being important to professional identity. In interviews with managers of professionals, a connection between being professional and complying with bureaucratic rules was indicated. Managers stated that it was assumed that professionals and managers of professional service units pay close attention to legislation and guidelines. They were trained to do so when educated. This did not mean, however, that professionals were rule-following bureaucrats. The legislation also contained basic values which complemented rules.

In this environment, the unanimous opinion among the managers was that laws and regulations were the dominant source of information about what to do. For them, working in a public sector organisation was working in a highly externally regulated environment, and for most of them, this was not a concern. On the contrary, conducting operations in accordance with bureaucratic rules was to work evidence-based and quality assured.

The Education Act is the most important thing I have. I look at it all the time. And then I have the curriculum for schools, and the occupational safety regulations. I need to know that we really work by them. Yesterday, I had a problem with a pupil, he was in a fight with a teacher. There is a section on what I can do to have a safe work environment, and I am writing a decision with support from that section. [Unit manager, School]

We are very much under the control of rules. The Health Care Act and regulations from the National Board of Health and Welfare. It is a mix of what we must do and what we should do, but we are very influenced by these instructions. Our profession [nursing] is very happy with rules and regulations. [Unit manager, Health care]

Our respondents were aware of the specific character of their work. Equality between residents and fair treatment were basic features of their work, and bureaucratic rules were seen as support in handling this.

The law provides confidence. I know what to do, and the parents know that their child is treated right. [Unit manager, School].

The presence of bureaucratic rules was in general seen as increasing the legitimacy and status of the service and the employees. It provided support and reduced uncertainty for managers in their decision-making. Only a few times in our interviews, bureaucratic rules were considered a source of problems. These examples related to the sheer volume of regulations. For example, the Education Act, which was the main source of guidance for principals, had expanded in recent years, and now contained above 600 rather detailed sections. It was considered a challenge to keep up to date.

Bureaucratic rules were sometimes found to conflict with the PM, but when this was the case it was obvious that managers gave priority to laws and regulations. For instance, one manager had observed a conflict between the local PM target to treat citizens with social or drug problems locally, and the legislation which stated that such patients must be sent to national institutions if necessary. Although the local PM target indicated a local political desire, the unit would not choose to break the law, i.e. violate bureaucratic rules. Another example was the PM to keep the use of temporary staff below five per cent in home care. This target was noticed but rarely met since as soon as there was a regulated task to carry out without ordinary staff available, temporary staff had to be used. To increase the ordinary staff to avoid this would instead conflict with the budget, which was not considered an option.

When managers noticed that local PMs and legislation overlapped, no one claimed the law was superfluous, only the PMs. This indicated that the legislation was taken for granted and the presence of overlapping PMs was seen as unnecessary. There were similar opinions about overlaps between external bureaucratic rules and internal behavioural controls. Things were somewhat less clear cut when bureaucratic rules conflicted with the tight budget control practice, but in most cases, the strong position of bureaucracy overrode budget control.

Overall, bureaucratic rules dominated managers’ daily work and were much more attention directing and disciplining than PM. There were not many stories about contradictions between professional ethos and the PM system, but sometimes there were clashes between professionalization and budget control.

Concluding discussion

In this section, we first analyse how the case illustrates and refines the conceptual model. Then, we draw conclusions and present ideas on future research.

Discussion

A first important note is that the case primarily is an example of PM not causing severe dysfunctional consequences. To some extent, dysfunctional consequences in the form of negative attitudes could be noted but very few examples of dysfunctional behaviour. We found perceived risks of behavioural displacement as a result of exaggerated interest in the measurement of grades, toxin-free preschool and organic food. However, no manager mentioned episodes where managers or employees, in fact, had been so overly inclined to hit distorted PM targets that ends had become less important than means.

The first question to discuss is to what extent the conceptual model can explain the general lack of problems with dysfunctional behaviour. It is fairly obvious that the explanation cannot be found in the design of the PM system itself. The set of PMs was reported to be incomplete and thereby risky in the sense that it gave decision makers an erroneous or at least incomplete understanding of achievements and quality of services. Rather, the explanation for absence of dysfunctional behaviour appears to be the loose control style practised by superior managers related to operational targets included in the PM system. In agreement with the conceptual model, dysfunctional consequences from PM were ‘buffered’ by loose control. If the incomplete PM system had been tightly used, behavioural displacement, gamesmanship and negative attitudes would have been probable, as reported from the PM practice in the NHS (Greener 2005; Lapsley 2008; Chang 2006, 2015) and the Dutch DRG system (Kerpershoek, Groenleer, and de Bruijn 2016). This is also indicated by our study since the alleged risk for behaviour displacement seemed to presuppose tightness in use in the form of extra funding linked to PM (grades and organic food).

The reported negative attitudes to the PM work because of overlaps between PM and external regulations and lack of feedback, appeared to be connected to both PM design and use. Overlaps may be possible to reduce through design changes where some PMs are excluded, while negative attitudes due to lack of feedback from politicians can only be dealt with through changed usage of PM information at the political level or by altering managers’ expectations of feedback.

In addition, the notably disciplining tight budgetary control practice was an influential factor since managers had to pay close attention to their spending and could not care as much about PM targets. Besides the fact that there was less room for managers to bother about PM results, it was clear that PM targets had lower priority when they conflicted with financial budget targets. Although operational targets occasionally could be in the political spotlight and therefore prioritized, the superiority of budget targets over operational targets was default. Lack of uncertainty about the priority order between budget targets and PM targets meant managers were not pressured to manage inconsistencies through dysfunctional behaviour, but could refer to lack of funding when PM targets were not met. This illustrates how tight budget control can buffer the dysfunctional behaviour of PM. Budget control simply takes attention away from PM.

It is not obvious, however, that tight budgetary control functions as a buffer if tight control related to PM targets is practised simultaneously. On the contrary, this may create an overly demanding and inconsistent control system for managers to handle, where dysfunctional consequences become an almost inevitable outcome, as in the case of PM in the NHS, where an impossible situation for managers was seen as a licence for dysfunctional behaviour (Greener 2005). This presumed three-way interaction between PM, tight budgetary control and tight PM control should be subject of further research.

Notably, the interaction between the PM practice and budget control practice did not buffer negative attitudes stemming from PM.³ Even if the budget was substantially more eye-catching for managers, they still had to do the PM work, which, as reported, was not perceived as very meaningful. Possibly the tight budget control practice even exacerbated the negative attitudes related to PM. The reason would be that managers who were pressured to manage tough spending limits became more frustrated when mandated to carry out work they perceived as fairly meaningless.

Compared to the effect of the interaction between PM design and PM use and between the PM practice as a whole and budgetary control, interactions with internal behavioural controls and cultural controls appeared less important. Internal behavioural controls were formally subordinated to political objectives, and no manager described any contradictions between internal behavioural controls and PM. Cultural controls were established for CoG and variously noticed by managers, but even though some managers saw benefits with cultural controls we could not capture any story about the interaction between such controls and PM. Cultural controls in CoG overlap with parts of a strongly established professionalization, which appears to override cultural controls. This means that, in this study, we cannot connect the general absence of dysfunctional consequences of PM to either cultural controls or internal behavioural controls.

As has been clarified, the control system under study made managers in the two city districts predominately budget focussed, and not PM focussed. However, the case also unanimously illustrates how a high degree of bureaucracy and professionalization in the units made PM even more subordinated. The case showed that bureaucracy and professionalization contributed in buffering against dysfunctional behaviour. Laws and regulations were substantially more attention catching and disciplining than PMs and, most often, budget targets. Any incompleteness in PM was compensated by bureaucratic rules which decreased the risk for PM to cause behavioural displacement.⁴ The high degree of professionalization further contributed to this by complementing bureaucratic rules in informing managers. In addition, the case did not display many trials of strength between PM and professional values and was therefore different from the experiences of PM in the NHS. It is also reasonable to assume that professionalization mitigated negative attitudes from frustration over unmet expectations of feedback on PM; professionals are intrinsically motivated and less dependent on political feedback.

Conclusion

In previous public sector research, dysfunctional consequences stemming from interactions between PM and other control practices and between PM and context have not been conceptualized. Therefore, the aim of this research was to conceptualize the dysfunctional consequences of PM in PSOs. Based on complementarity theory and contingency theory we provided a tentative conceptualization, which was illustrated and refined through an empirical study. The illustration supports the founding idea that dysfunctional consequences of PM – dysfunctional behaviour (gamesmanship and behavioural displacement) and negative attitudes – are a matter of interactions between PM design and PM use, between control practices in the control system and between PM and context.

The empirical study refined the model by showing how tight budgetary control assisted in buffering dysfunctional behaviour through reducing managerial attention paid to PM. Our study also contributed by displaying how negative attitudes from PM, to some extent caused by design problems, were exacerbated by low-intensity use (feedback) at the political level, but buffered, or at least mitigated, by bureaucracy and professionalization. Dysfunctional consequences of interactions between PM and other control practices are still an open question, but appear to be a matter of how attention catching these practices are.

Naturally, a limitation of this research is that a one-case study cannot illustrate all possibilities in a comprehensive conceptual model. In the future research, additional refinement of the conceptual model is warranted. However, already at this stage, a formal test of the possible three-way interaction between PM incompleteness, PM control tightness and tight budgetary control in different bureaucratic and professionalized contexts appears justified.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by Forskningsrådet om Hälsa, Arbetsliv och Välfärd [2015–00758].

Notes on contributors

Sven Siverbo

Sven Siverbo is Professor of Accounting and head of the business administration research group at University West, Sweden. He researches management accounting and management control, mainly in public sector organizations, and teaches management accounting and management control at undergraduate and graduate courses.

Mikael Cäker

Mikael Cäker is associate Professor of Accounting at School of Business, Economics and Law, University of Gothenburg, Sweden and also affiliated with Trondheim Business School at NTNU, Norway. Cäker’s research interest is management accounting and management control in both private and public sector organizations. He teaches at both undergraduate and graduate level.

Johan Åkesson

Johan Åkesson is lecturer in Accounting at School of Business, Economics and Law, University of Gothenburg, Sweden. Åkesson is doing research on design issues regarding management control systems and teaches management control and management accounting, behavioural aspects of management accounting and control and basic management accounting: concepts, definitions and relationships.

Notes

1. Tight control may cause dysfunctions independent of whether PM is incomplete or not. For instance, profitability performance measures may be manipulated although they sometimes are fairly complete and aligned measures of performance in private sector organizations.

2. It is important to note that this paper deals with the dysfunctional effects of PM and not with antecedents to different PM practices in PSOs. If we were interested in explaining variation in PM practices it would have been necessary to model all context variables as explanatory variables for PM practices.

3. Although not the main focus in this research, it should be noted that our study corroborates previous findings about a main effect of tight budgetary control on dysfunctional consequences (see e.g. Hartmann 2000).

4. Even if this could be seen as bureaucratic rules imposing on strategic alignment communicated through PM, the fact remains that CoG must comply with laws and any misalignment between the two is a misspecification of PM.

Appendix.

Respondents

Table

	Position	Activity	Length
District East
1	Unit Manager	Preschool	70 min
2	Unit Manager	School	65 min
3	Unit Manager	Meal services	58 min
4	Unit Manager	Elderly care	48 min
5	Unit Manager	Health care	55 min
6	Area Manager	Health care	82 min
7	Area Manager	Administration	67 min
8	Area Manager	Property and security	50 min
9	Area Manager	Education	70 min
10	Area Manager	Internal service	92 min
11	Accountant	Development	50 min
12	Controller	Culture	60 min
13	City District Manager		71 min
District West
14	Unit manager	School	53 min
15	Unit manager	Elderly care	60 min
16	Area manager	Preschool	83 min
17	Area manager	Elderly care	75 min
18	Sector manager	Elderly care	63 min
19	Sector manager	School	47 min
20	District controller		85 min
21	HR manager		68 min
22	District manager		37 min
Other respondents
23	School Coordinator		72 min
24	Internal Auditor		68 min
25	Internal Auditor II		63 min
26	Finance Director		110 min
27	Union Representative		65 min
28	Central administrator		72 min
29	Central administrator		69 min
30	Central administrator		72 min
31	Central administrator		80 min
32	Central administrator		60 min
33	Central administrator		88 min
	Mean		68 min