Calculating Pay in Swedish Schools: Accounting, Performativity, and Misfires

This is a study of individual differentiated pay (IDP) for teachers in Sweden. We regard IDP as part of the observed accountingization of professional work. Our aim is to contribute to the understanding of accounting as a performative calculative device through the study of how performance is made an element of individual pay. As we sought to answer &#8220;How are merits and performance of professionals turned into a set IDP?&#8221; we observed the ways in which IDP had diverse performative consequences, which made the connection between performance and merit less influential in the process of setting wages. We contribute to the theorizing of accounting as a performative practice by introducing two propositions that indicate the importance of considering the extent to which accounting is turned into a dominant narrative and the extent to which the operations can be visualized in a relevant, stable manner. The propositions are the result of the approach to include visualizations as a focal dimension of performativity. The concept stresses the importance of recognizing the prerequisites of relevance and correspondence.


Abstract
This is a study of individual differentiated pay (IDP) for teachers in Sweden. We regard IDP as part of the observed accountingization of professional work. Our aim is to contribute to the understanding of accounting as a performative calculative device through the study of how performance is made an element of individual pay. As we sought to answer "How are merits and performance of professionals turned into a set IDP?" we observed the ways in which IDP had diverse performative consequences, which made the connection between performance and merit less influential in the process of setting wages. We contribute to the theorizing of accounting as a performative practice by introducing two propositions that indicate the importance of considering the extent to which accounting is turned into a dominant narrative and the extent to which the operations can be visualized in a relevant, stable manner. The propositions are the result of the approach to include visualizations as a focal dimension of performativity. The concept stresses the importance of recognizing the prerequisites of relevance and correspondence. on which level applies, and for whom, have become a matter of performance under the labels of performancerelated pay, pay for performance, and performance incentives (Ball, 2003;Chamberlin, Wragg, Haynes, & Wragg, 2002;Cutler & Waine, 2004;Forrester, 2001;Jones & Hartney, 2017). This presupposes measuring and evaluating merits, performance, and efforts. Although performance remuneration at unit or organizational level is well covered in the literature (Kastberg & Siverbo, 2007), less attention has been paid to the performance dimension of setting the level of the monthly individual salary. Our case is individual differentiated pay (IDP) for teachers in Sweden. IDP has been described as an attempt to "ensure that salaries reflect employees' individual contributions to the organizational outcomes and enable teachers for salary development" (SKL, 2015, p. 4).
We regard IDP as part of the observed accountingization and economization of professional work and organizations (Lapsley, 2007;Miller & Power, 2013;Power & Laughlin, 1992), where accounting has been described as providing a "dominant narrative of everyday doings in organizations" (Miller & Power, 2013). The study aims to contribute to the literature on performativity of accounting (Callon, 1998a;Themsen & Skaerbaek, 2018). The concept of performativity captures the active role of how accounting intervenes and mobilizes actors (Mouritsen, Hansen, & Hansen, 2009;Vosselman, 2013). In terms of accounting, performativity is about understanding the "productive" nature of accounting (Miller & Power, 2013). For example, Power (2015) illustrates the object formulation in the case of impact management for higher education institutions. Although a substantial number of studies illustrate the performative functioning of accounting, there is still a need for studies that explore the dynamics and "counterprocesses" at play (Vosselman, 2013). Accounting sometimes seems to be "co-opted" by the professional domain point in that direction (Kastberg & Siverbo, 2016). Recent studies (Boedker, Chong, & Mouritsen, 2019) indicate that there is a need for an advanced understanding of performativity because of observations of lack of performative effects.
Although object elaboration is central to the performativity thesis (Miller & Power, 2013;Power, 2015), drawing on the literature on professions and professionalism, we might expect to find measuring and valuing performance to be challenging (Freidson, 2001). The work is carried out in close relationships between client and professional worker and the specificity of each case makes it difficult to ex ante rule by standards and guidelines and ex post control by results (Hofstede, 1981;Mintzberg, 1983). This is certainly true also of teachers who have been described as conducting work in a closed classroom (Hargreaves, 2000;Helsby, 1995) with no or minimal interference by colleagues or principals. One might add that the maintaining of vagueness and obfuscating of actual performance have been described as a strategy by professional workers to safeguard autonomy and maintain control over work (Abbott, 1988;Alvesson, 2001). Therefore, the challenge can be described as one where measuring, qualifying, and valuing must be applied to something that is vague, intrinsic, and performed by actors who might even be opposed to the very idea. When scrutinizing IDP, we might therefore expect counter-processes that provide a foundation for an elaborated discussion on the reach of accountingization.
Our aim is to contribute to the understanding of accounting as a performative practice by studying how performance is made an element of individual pay. Therefore, we formulate our research question as "How are merits and performance of professionals turned into a set IDP?" We will first introduce a theoretical framework, where we expand the understanding of performativity in action with the help of the concept of visualizations. We observe in the analysis of our empirical case the way in which IDP had diverse performative consequences. This made the connection between performance and merit less influential in the wage-setting process. These observations allow us to contribute to the theorizing of accounting as a performative practice by introducing two propositions that indicate the importance of considering the extent to which accounting is turned into a dominant narrative and the extent to which the operations can be visualized in a relevant, stable manner. The propositions are the result of the approach to include visualizations as a focal dimension of performativity. The concept stresses the importance of recognizing the prerequisites of relevance and correspondence. In the next section, we develop a discussion on the relationship between the concepts of performativity and visualizations.

2
THEORETICAL FRAMEWORK: PERFORMATIVITY AND VISUALIZATIONS Callon's (1998b) performativity thesis focalizes on the sociotechnical construction of economics. The core argument is that economics is not based on empirical evidence but must instead be understood as a device that structures and orders social interaction. Recognizing economics as a performative urges us to scrutinize the construction of economic actions (Callon, 1998b;MacKenzie, Muniesa, & Siu, 2007a). Performativity, therefore, addresses not only descriptions of how an economic process is performed but also how such economic processes are subject to alterations/manipulations, or "reconstructions" by the use of such descriptions (MacKenzie, 2006). Callon (1998bCallon ( , 2007Callon ( , 2010 illustrates the argument by pointing to the (re)construction of "homo oeconomicus," claiming that it does not have an empirical correspondence. Theory (re)constructs how the "rationality of economic man" and how s/he would behave in the making of economy.
The concept has been used within the accounting literature as well, helping us follow the structuring effects of accounting or a "statement" (Themsen & Skaerbaek, 2018). The dual consequences of calculative technologies have been described in studies of risk management (Themsen & Skaerbaek, 2018), waste management (Corvellec, Ek, Zapata, & Campos, 2018), innovations (Busco & Quattrone, 2018b), and impact measurement (Power, 2015). Following such reasoning, accounting as a performative calculative technology may therefore not only represent the processes that unfold in everyday situations, but it also shapes and reshapes humans and nonhumans upon their consent in object formation (Revellino & Mouritsen, 2015). Therefore, the IDP may be expected to bring about performative implications.
When illustrating the relation between "economics" and option pricing, MacKenzie (2006, pp. 17-20) points to Barnesian and counter-performativity. In the former, economics, a theory, model, or idea "alters (economic) processes and/or their outcomes to make them more like their depiction by economics." On the other hand, in counterperformativity, "the use of economics undermines its claims to empirical accuracy" (MacKenzie, Muniesa, & Siu, 2007b, p. 9). Likewise, Callon (2010) calls such counter-performativity "misfires" and claims that they are "the rule of the game" (p. 165) because of the fragility and rareness of existing proper sociotechnical agencements. Such "misfires" (or overflowings) always exist, or will be visible later, and only when the overflowings are invisible and inactive may the performativity become successful (Callon, 2010). We accordingly regard accounting (in the form of IDP) as a (scientific or epistemic; Busco & Quattrone, 2018a) potential performative in the way Callon (1998b) regards "economics." Based on this, we expect performative aspects in our search of the connection of merits, performance, and salary levels. The performativity of accounting may be successful or fail (counter-performativity) in terms of how the actors are located in a concrete-contingent sociotechnical network (Vosselman, 2013). Such reasoning may then have implications for the existence of different actors and their actions not only inscribing the expectancies of a "good teacher" but also shaping and reshaping the individual teacher when determining the pay.
In the search of performative consequences of IDP for the teachers, it is necessary to understand which aspect of the professional work becomes solid ground for assessment. This means that assessment of teacher's performance, merits, and skills within the IDP process would be expected to be (re)constructed by the assessor in order to find the matching salary for the teacher. In this respect, in order to provide a more specific conceptual point of departure for the analysis, we draw on Latour's discussion of visualizations (Latour, 1986;Robson, 1992). The trick that the employer has to perform when valuing and grading the professional worker is to turn the vague and intrinsic (Alvesson, 2001) everyday performance into numbers and rankings, that is, visualizations. The challenge is to turn a three-dimensional world into two-dimensional charts and figures and, inversely, to turn two-dimensional templates into operational actions.
The linked activities are perceived as steps in a translation process. Latour (1998) visualizes the process as a chain.
One important notion is that when something is accounted for, nothing is transmitted directly, instead accounts are representations. Each step means distancing it from the original practice through reductions, while simultaneously amplifying it. Latour's own example is the producer of a map (Latour, 1986). In order to make any sense, nature has to be Electronic copy available at: https://ssrn.com/abstract=3712359 turned into a two-dimensional blueprint. A picture would have been more in line with what the cartographer observes, but less useful for orientation. Furthermore, the map must be produced in such a way that it is understandable (mapping rules), transportable (physical requirements), and solid (not deteriorate if it gets wet). Each step means a reduction of the local context, but also that dimensions are added.
The merit of adding the discussion of visualizations to our understanding of performative processes is that the concept addresses the requirements that facilitate and structure the process of visualization. These might be regarded as requirements of "correspondence" and of "relevance" (Latour, 1986). Both are important because they must be traceable throughout the chain of translations. If the translations lose the correspondence with what is claimed to be represented, there is a risk of being accused of fraud (Latour, 1998) or even of "bullshitting" (Macintosh, 2009). This is important because it ties accounting, object elaboration, and legitimacy together. Likewise, if the relevance is lost, we end up with representations not considered by anyone. In order to pass as adequate representations, they must be also regarded as relevant and useful by the network of interested actors (Latour, 1986). They must fit into truth regimes and be regarded as legitimate knowledge (Kastberg & Ek Österberg, 2017).
The main implication is that the requirements form the translations being made, that is, they have performative effects (Qu & Cooper, 2011;Robson, 1992;Robson & Bottausci, 2017). From an organizational point of view, the rerepresentations might have to correspond to the strategy and objectives of the management, and in our case politicians at the school board. Looking even wider, norms and guidelines from more distant stakeholders like The Swedish National Agency for Education and education departments at universities must be considered. There are established ideas about what counts as facts and what procedures are perceived as valid (Robson, Humphrey, Khalifa, & Jones, 2007).
The discussion of visualizations assists in refining our understanding of performativity. That is, the discussion helps stress that performativity presupposes relevance and correspondence, and that our analysis must be expanded to the relevant situated network. There needs to be a network strong enough to support accounting ideas as being relevant and a legitimate truth. Turning to correspondence, in order for performative effects to become effective, there must be visualizations adjudicating right and wrong behavior in a perceivable manner. The very basic idea of the performative thesis is that there is an established correspondence between local practices and templates.

METHODOLOGY
The aim of the study is to observe the practice of connecting the skills, merits, and performance with the pay. Based on our theoretical starting points, we have adopted a processual ANT-inspired approach for data gathering and analysis.
This means that we are interested in how relationships are established and how practices emerge, rather than in steady states (Mouritsen, Mahama, & Chua, 2010). In our case, this means scrutinizing the network of actors and practices involved in IDP. The ambition is to explicate practice and to relate actions to the concept of IDP (Llewelyn, 2003). Thus, this study is based on a case study (Flyvbjerg, 2006) of IDP for teachers.
The data were collected between April and June 2018, through 16 semistructured interviews with the principals and teachers at different primary schools in the Gothenburg municipal area. Additional interviews were held with municipal HR managers in municipal districts, and teacher union representatives. The choice of respondents in terms of their roles complies with ANT framework, insofar as the intention of "following the actor" remains (Latour, 2005). The principals and teachers are inherently included, because the measurement and pay determination work unfolds primarily between the principal and teacher. However, as we proceeded with the study, it turned out that the other actors, HR administrators and union representatives, are also involved in IDP. This led us to extend the range of respondents in terms of their roles. Table A1 in the appendix summarizes the numbers and locations of the interviewees.
Electronic copy available at: https://ssrn.com/abstract=3712359 Although the data have been gathered from different schools, the study does not primarily concern the differences in locations, because the IDP process is the same for all schools. Our aim was merely to gain a deep and nuanced understanding. IDP is common practice in Swedish public sector organizations, and the overall process is the same, regardless of the jurisdictional location of the organization. The process is the same for organizations under municipal (i.e., schools), regional (i.e., hospitals), and national (i.e., universities) jurisdiction (IDP guideline;SKL 2015). This means that the IDP process is regulated nationally by the law (Collective Agreement, 2018) and does not change depending on the location, size, or type of school. Likewise, the local context in Gothenburg shows similarities with other municipalities in the rest of Sweden. On the other hand, there are some differences between municipalities, such as the salary criteria and the municipal (performance) priorities (SKL, 2013).
The semistructured interviews (Kvale, 2006) were conducted at the respondents' workplaces. The length of interviews ranges between 40 and 90 min. The interviews were guided by the broader themes to allow the respondents to reflect on their daily practices. By keeping the interviews open ended, we aimed to grasp thick descriptions of the IDP process. The interviews were recorded and transcribed verbatim for analysis.
Documents were gathered in two different ways and are listed in Table A2 in the appendix. First, some documents were gathered from the available resources on the municipality's websites, unions, The Swedish Association of Local Authorities and Regions (SKL), and the trade unions. These documents helped us to understand what the IDP process looks like and provided us with background information as we prepared for the interviews. The other documents include the salary criteria, guidelines for performance appraisal and salary meetings, and notes. These notes were gathered from the respondents during and/or after the interviews. Background information from the documents regarding HR processes, salary negotiations, and collective agreements was helpful in contrasting with the way in which the IDP process unfolds as described by the respondents in the interviews.
The empirical material from interviews and the documents were analyzed using categorizations produced in an iterative process moving between collecting, readings of the empirical material, sorting, and coding, to the conceptualizations of themes, to theoretical readings, and back to revisions of the empirical material (Schurz, 2008). We proceeded with the analysis until we were able to reconstruct a coherent story of IDP process based on the empirical descriptions.
This resulted in basic empirical themes such as the different sources of differentiation of pay and how visualizations were present in relation to them.

EMPIRICAL ANALYSIS
IDP in the Swedish education sector could be traced back to the early 1990s (Larsson, 2004;Nilsson & Ryman, 2005).
Following the decentralization of collective bargaining processes from the central to the local actors on municipal and organizational levels, pay structures have gradually changed from fixed ladders to individually flexible ones (Andersson & Schager, 1999). The reasons were that the IDP would better reflect individual accomplishments and encourage efforts. In this sense, IDP might be regarded as part of the New Public Management (NPM) movement. In this section, we describe the result of our empirical investigation and illuminate how the process of determining pay unfolds as a process. We have observed three potential sources for differentiated pay: (a) setting the level in the entry, (b) revising the level annually, and (c) (re)setting the levels by promotions and job changes. In the setting of pay, teachers, principals, administrators (HR managers), and union representatives play key roles. We also observe an interplay among different levels. Accounts of merits, skills, and performance follow from the school to the municipal levels in attempts to visualize performance, whereas the salary criteria, areas of priorities, and the salary levels follow from the municipality to the school. This means that the pay determination unfolds in negotiation between principal and teacher and organization and field levels. Figure 2 summarizes the process. We present our empirical findings in terms of the three identified sources with potential to cause differentiation in wage levels.
Electronic copy available at: https://ssrn.com/abstract=3712359 The negotiations at different levels

AMOUNTS
In order to understand the wage-setting process, we have to start at the national level. The collective bargaining agreement between teacher unions and employer representatives is the basic framework for annual salary revision. In Sweden, there is a tradition that the agreement between industrial employers and the industrial workers union forms the starting point and the benchmark for the rest of the negotiating parties. The annual salary revision follows the agreed mark by the employer and worker representatives on the national level, in our case with the national teacher unions and the employer organizations. The central agreement at the national and municipal levels (field level) travels to the local agreements in municipal districts and then to the schools. Figure 1 summarizes this journey from central (national) to local (municipal, district, and school) contexts. There is little acceptance for deviations in relation to the overall national marks. The collective national agreement therefore plays an important role in setting the guidelines for further local and individual agreements. It specifies the amount (as a percentage) to be redistributed in the municipal, district, and school contexts. The total fixed amount for salaries is redistributed first among municipalities, then in municipal districts, and then to the specific school. On the other hand, although the process is top-down from central to local contexts, within the local context, the previous district and school's budgets guide the budget formation for the next period, which in practice follows a bottom-up feedback flow.
Because it is the average that is considered, this leaves room for differentiating between individuals where some get more and others less. Apart from the regular IDP, the teachers also get salary increase from promotions. These promotions are made possible by government reforms concerning teachers' profession (first teacher reform) and salaries (skilled teacher reform). Therefore, promotions constitute a parallel redistribution process in addition to the collective agreement. Neither promotions nor entry-level salary setting has to consider the national mark. Within the local context, on the other hand, the pay determination for individual teacher includes different steps that take place between Unions monitors and intervenes -Teacher demands from principal the salary level -Principal negotiates the amount with teacher -Principal negotiates on the amount with HR administrator -HR administrator approximates the amount -Principal informs the amount to teacher -Teacher accepts/refuses -Entry level salary becomes the basis for revision -Task allocation, appraisal talk -Annual salary becomes the basis of new entrylevel pay when teacher changes the job -Based on annual assessments, the principal approximates the annual salary increase -HR administrator informs the principal about the benchmark -HR administrator monitors the amount and coordinates with unions -Principal informs the teacher about annual increase during revision meeting -Principal differentiates the amounts among teachers -Additional increase forms the basis of new entrylevel pay and amount subject to the annual revision -HR administrator distributes separate funding to the principals F I G U R E 2 The salary setting process in between organization and field levels actors in municipal administration and school. Figure 2 illustrates the IDP process among teacher, principal, and HR administrator for determining the entry level salary, annual pay increase, and promotions for an individual teacher.

Setting entry level salary
In the wage-setting process, the first marker is the entry level. The term "entry level" applies to both newly graduated teachers who are new to the labor market, and experienced ones who change schools. The entry-level salary is determined according to two sets of negotiations: (a) between the teacher and the principal and (b) between the principal and the HR administrator. The entry-level pay determination starts with negotiations between the job applicant and the principal. The applicant teacher demands an amount as entry-level salary. After discussions with the responsible municipal administrator, the principal then makes a new offer that the applicant either accepts or refuses.

Negotiations between teacher and principal
The principal first assesses the applicant according to accreditation. Accredited and nonaccredited teachers constitute two different groups for entry-pay determination, because different wage levels apply to these categories. In the case of noncertified teachers, the entry level pay determination is more like an offer that the applicant either accepts or rejects. The pay determination therefore differs between nonaccredited and accredited teachers, where the accreditation becomes the visualized evidence of performance, merits, and skills. Another aspect of entry-level pay is the teacher's experience. The new graduates and experienced teachers constitute two different categories. Previous work experience shapes starting salary in at least two ways. First, teachers have described that when they change jobs, it is often the case that principals demand reference letters from their previous workplaces. This means that merits, skills, and performance assessments start, to a certain extent, before the recruitment. Second, the teacher's salary at the previous workplace represents the minimum in entry-level salary determination, because the bargaining for a new Electronic copy available at: https://ssrn.com/abstract=3712359 entry salary does not normally fall below the previous salary. Therefore, the previous salary impacts (re-)setting the entry-level pay. We will describe this aspect in greater detail in the following sections.
A related observation is about the salary gap between new and experienced teachers. Teachers, principals, and union representatives explained to us that the entry salaries for new teachers are higher than for experienced ones.
In our case, this gap was caused by the scarcity of teachers on the labor market. Such a "market factor" was evident when one teacher described to us his/her negotiation for an entry wage. Accredited teachers have more capacity to bargain for better entry salaries because of such scarcity. This also demonstrates the difference in the salary determination between noncertified and accredited teachers. The pay determination finds itself within two dynamics: Finding the best-matching entry pay relates more to situated technicalities (accreditation) and market factor (scarcity), but less to individual merits, skills, and performance. Among these individual assets, the accreditation alone becomes visualized. Such situated factors disable the principal (or ac/counter) to handle, or control, the individual teacher as a specific case, prompting reductions.

Negotiations between principal and administrator
The formal procedure is that the principal takes the offer for entry-level salary to the municipal HR administrator. The HR department, together with unions, considers the entry salaries in terms of the teacher's age and professional work experience. The HR administrator uses statistical information on entry wage levels in the specific municipal district and in the Gothenburg municipal area. Because the entry-level salary constitutes not only the basis of the amount to be revised annually but also the benchmark for the further entry-level salaries when the teacher changes schools, the lower/higher deviations are followed by the municipal district administration. However, the shortage of teachers has had certain implications for how the wage-setting unfolds in practice. Some principals have described to us that they agree upon salary amounts without consulting HR administrators.
Determining the salary in dialogue with the HR department is not always easy in recruitment. It must go fast, otherwise teachers accept another job. If I do not get an answer from HR the same day, then I decide alone. I

can't wait because I must have a teacher for the pupils' classroom. (Principal 1)
Although this can be due to coordination problems between administrator and principal, the existence of such a market factor paves the way for the principal's improvisation. Although teachers make use of the scarcity in the market by selecting the school and pushing the negotiations to get the entry salary that they demand, principals act alternatively to the formal structure to make faster recruitments. The case from the HR administration's side prompted a similar observation. One administrator described that their involvement is not only about compliance with salary averages but also about antidiscrimination protection of women, foreigners, and older workers: Entry-level pay setting between the principal and HR administration is not only about the coordination of the salary levels. It also has to do with the (un)fairness of the salaries among different groups. We have observed how certain criteria such as age and gender are emphasized by the administrator, showing that different requirements are understood to be legitimate. On the other hand, for principals, the administrator's work has much to do with statistics.

… It does not matter for HR if I say "here we have a teacher who is well-performing, and very skilled with good references." It is pretty much statistical information of working years and age that concerns HR. (Principal 1)
Thus, teachers' merits, skills, and performance assessments remain of little importance in the interaction between the HR administrator and principal. The criteria and attributes considered do not concern actual performance or merits but, for example, gender and accreditation. We have observed that there is a network situated in between the school and municipality, certain technicalities that relate the teacher to certain wage levels, and market dynamics that alter the formal structures. This shows that for entry-level pay determination, some factors delimit the handling of an individual teacher as a specific case. Furthermore, contrary to the idea of differentiation based on merits, skills, and performance, we have observed that fairness, appropriateness, and correctness (finding the best-matching salary) are elements used in practice.

Salary revision/review
Once the entry pay is set, it is subject to annual salary revision. The collective agreement between union representatives and employer decides the percentage of increase locally, but not for the individual. Similar to entry-level salary, decisions on revision move back and forth iteratively between the teacher, principal, HR administrator, and union representatives ( Figure 2). In revising the salary, performance of the teacher becomes visualized during the academic year in terms of benchmarks, work objectives, and salary criteria.
Before revising the salaries, principals get annual benchmarks in percentages from the municipal administration in order to evaluate below-, normal, and good performance. One respondent describes the revision as a process among the principal, administration, and unions. Thus, we see how the HR administration intervenes in the principal's assessments by providing benchmarks, as how the unions intervene by questioning them. The benchmarks become a legitimate way for the valuation of "good" or "bad" performance. The performance of the teacher thus becomes visualized in terms of these benchmarks.

Before the salary revision, I get the information from my
The starting point of annual salary revision is the work-task allocation. These include school-specific objectives, such as education of newly arrived foreigners in Sweden for schools located in socially deprived districts, as well as annual municipal objectives in school development such as digitalization, special education, and quality excellence in scoring techniques. Teachers and principals explain that digitalization is concerned with the use of technological devices in education.

Salary criteria for all employees in Gothenburg municipality
A-Competence Indicators 1) Required/qualifying education 2) Relevant further education with results, diploma, certificate, or similar 3) Sectoral experience 4) Other relevant experience B-Performance Indicators 1) Knowledge about the tasks, user's needs, regulations and rules, and processes 2) Care/concern about work tasks, users, and others

3) Team work and collaboration 4) Thinking new 5) Structured and effective work
Salary criteria for all employees in city district

1) Creativity 2) Social competence 3) Sense of initiative and independent work 4) Quality 5) Adaptivity
Salary criteria for education sector 1) Systematically developing teaching 2) Developing quality work 3) Improving student's attendance 4) Creating protection and study peace 5) Collaborating with parents 6) Improving student's participation to teaching 7) Making complementary efforts for all students in need of support or special support 8) Taking responsibility in administrative development to enable pedagogical leadership

School-specific criteria Organization level
Field level

F I G U R E 3 The salary criteria for teachers in organization and field levels
We have two focus areas this year: study peace and digitalization. These are the frames and a kind of indicator that the teacher is skilled in these aspects to move development forward and work for the whole school. Hence, this should be prioritized in making a good performance. (Principal 1) When we consider how work-task allocation forms part of the setting of pay, it is obvious that it is tasks performed outside the classroom that are rewarded, while visualizations of classroom practice were lacking.

Salary criteria
The salary criteria form a formal foundation for performance assessments. There are three levels of criteria: municipaloverall and particular district as well as profession-specific criteria. Figure 3 summarizes the variety of indicators at different levels, and they form a considerable list. The assessments based on these criteria occur during the academic year. We have observed that they are not necessarily made individually, but relatively, in comparison to "who is better." This shows us how "good" and "bad" performance become a matter of comparison. However, the extensive list of criteria did not seem to be used in an elaborate manner. One principal admits that it becomes easier to assess the teachers whom s/he has known and worked with previously. This reminds us of the distance between the assessor and teacher or put differently the inability of principals to evaluate actual work. The distance grows when the principal does not know the teacher; reductions in comparisons come into play. Regarding performance information, one principal explains that s/he hears things second-hand.

… One part it is that you hear, from pupils, parents, all employees, cooperating partners. So, you hear things. And you get a somewhat general picture. (Principal 4)
The information about performance comes from inside and outside schools' boundaries including parents, and cooperating partners. The relationships among the principal, teacher, and parents are relevant for assessments because not only is communication with parents a performance criterion for teachers (Figure 3), but the parents are also an information source.
Well, I get so much information from parents. The parents are often frustrated, unsatisfied with the teachers …

But I cannot tell a teacher that "you got a worse salary offer because a student's parents are unsatisfied" unless the communication with the parents is lacking, which constitutes a salary criterion. (Principal 1)
What parents say about the teacher is one indication of the latter's performance. Similarly, visibility also comes into play in the assessments, and "what becomes visible becomes a matter for" the performance assessment. However, because of the distance, the principal cannot observe teacher's actions. One way to reduce the distance and to make the unseen visible is the classroom visit.

I have my own rule of thumb to visit at least three classrooms every week. I must know what is going on behind closed doors. Before a salary revision meeting, I must know if someone is not good in certain aspects. (Principal 1) I try to visit all my teachers three times a year from the beginning to the end of their lessons to get an understanding of who has the skills in the classroom, and who is better than the others. But from which of them the students learn the most, that I don't know! (Principal 4)
Another tool to visualize performance is the intranet portal, Hjärntorget (Brain square). Teachers update their daily and weekly activities on this portal. Although it is designed for the parents to get daily/weekly information about their children's activities at school, it is also a source of information for the principal to assess the teachers' performance.

I think that I have pretty good judgment about what teachers actually do in the classroom because you listen,
take meeting notes, talk to pupils, talk to parents, see the information on the intranet, and so forth. Teachers share their academic term plans in their subjects-for instance, you can go in and see them there on the net.

(Principal 3)
Hjärntorget is therefore one visualization of the performance that takes place behind the classroom door. The principal approximates the assessments with visible fragments of teachers' work, whereas rumors, word of mouth, and complaints become proof of "good" or "bad" performance, which is not necessarily teaching related. Instead, the focus turns to elements possible to visualize, often outside the classroom practice, which lend themselves to correspondence. The engagement in digitalization, for example, becomes an indication of efforts. As the distance among teacher, principal, HR administrators and unions grows, the visualization of individual teachers' performance becomes problematic for the principal. However, it may also be argued that such distance results in further consequences for visualizations. On one hand, the principals have said that it is easier for them to assess teachers who they have worked with in the past, which perhaps signals a bias toward them. On the other hand, because of the number of actors, the Electronic copy available at: https://ssrn.com/abstract=3712359 already existing distance between the other (perhaps newer) teachers and principal grows even larger, which makes it more difficult to make visualizations that would constitute the ground for the assessments.

Resetting the salary level
Job changes and promotions (re)set the salary levels. When the teacher changes job, a new entry-level salary is set.
Likewise, when the teacher is promoted as first teacher and/or to boost his/her salary by the skilled teacher reform, new wage levels apply. The re/setting is not documented as a "formal" aspect in the IDP process; however, we have observed that job changes and promotions are central components in the practice of IDP.

Salary tourism or wage carousel
One of the determining factors for salary levels is the municipal boundaries. Because the collective agreement is municipal, changing job between municipalities gives the teachers a possibility to increase their salaries. The teachers call this a "wage carousel": moving to a school in the adjacent municipality, before returning to the school where the teacher actually wants to work, with an even higher salary.

If someone comes in and suddenly gets a good salary, for which I could have worked in a place forever, I would get
my 700 or 1000 SEK. But someone new comes in and gets far more! You could say "but how unfair!" and it does not mean that person is competent. But it is also good for you if you take the carousel that salaries go around.

(Teacher 3)
When changing between different municipalities, the new entry-level salary increases more than the annual revision. Because the salary increase resulting from the job change is not necessarily related to the annual increase based on revision, it interrupts the assessments and resets the salary. Therefore, the "salary carousel" activates a connection between performance and pay. Regardless of how well the teacher performs in the previous school, changing workplace results in a salary increase.
Newly employed teachers generally have a higher salary than those who have been secure in their work for a very long time. The only way to raise your salary by several thousand is to change schools. (Principal 3) In this case, the (re)setting of the entry-level salary is related to the correctness of wages and enabling fairness between new and experienced teachers. On the other hand, skills, merits, and performance are of little account. Moreover, such a carousel also has implications for the principals when assessing the applicant. The term carousel has a counterpart for the principals: salary tourism.
We managers call them salary tourists. If I were to proceed with the recruitment, then you understand. "Okay, here is someone who is trying to increase his or her salary" So, when I ask, "What is your salary demand?" and there maybe they put it at 47,000 SEK. "Okay, you're asking for a salary which is higher than mine." And there you understand that it is not plausible, but then they are not interested, and they go away. (Principal 1) Administrators have a role in protection against competition between municipalities. The purpose of such salary policies against unfair salary competition between different municipalities is to tackle the problem of ever-increasing salaries. Therefore, we observed that handling the pay determination faces a market factor. The job changes reset not only the wage levels, but also their connection to the previous assessments, consequently delimiting the performance component in the IDP. This was also evident in the salary review statistics conducted by the teacher unions, where "changing jobs between municipalities results in monthly salary increases between 3,000 and 5,000 SEK on average, whereas the levels of annual salary review for teachers working in the same school increase on average between 600 and 1,500 SEK" (Union representative). This shows that job changes reset the salary levels, and the salary increase from changing job exceeds the one from annual salary.

Promotions
The government has introduced two sets of reforms that impact salaries: skilled teacher and first teacher reforms.
First, the skilled teacher reform of 2016 was introduced with the aim of increasing the preschool, elementary school, and high school teacher salaries beyond the yearly revision. The eligible teachers promoted by the principal get a fixed increase of 2,500 Swedish SEK per month. Second, the first teacher governmental reform was introduced in 2013. The ambition of this reform was to strengthen teachers' professional work through creating alternative career paths for them. First teachers are promoted upon the application of a teacher to a vacant position, and the recruiting process includes employment interview. Those who are promoted get an additional 5,000 SEK as fixed monthly increase. The reforms have separate national funding and considerably higher impact on the salary increase than the salary revision.
They also have implications for the future salary of the teacher, because the reforms increase not only the monthly pay, which is subject to annual salary revision, but also the average salary levels with its reflection on the next entry salaries.
Concerning the teacher reforms, one principal explains that the salary increase from the reforms can be the source of but also the "correction of unfairness." In Thus, performance is only one aspect in the redistribution, and the promotion also becomes a means of correcting salaries perceived as too low. It is used to make the salaries "fair" and "even" between teachers with lagging wages who have worked for many years at the same school and the newly graduated teachers who start working with better salaries. Therefore, while the promotions generate an overall increase of salaries to a certain level, the increase is only to some extent related to skills, merits, and performance. We have observed a twofold connection between performance and merits: on the one hand, the first teacher reform aims to promote the teachers based on their differentiation of performance, merits, and skills. Because first teachers are promoted via a separate recruitment process, there are limited numbers of first teachers in schools. On the other hand, as illustrated in our case, the skilled teacher promotions are widespread in schools because the reform is to a large extent used for correction of salaries.
This has implications for how the salary differentiation unfolds in practice: the first teacher promotion distinguishes the "first teacher" from the "ordinary" one and "differentiates" not only the salaries but also the teacher profiles.
However, because of the wide distribution of skilled teacher reform, the differentiation in the salary from the first teacher promotion in some cases becomes invisible and void. Particularly, the skilled teacher promotion "evens out" the ordinary teacher's salary with the first teacher. Therefore, the reforms "make" and "efface" the differentiation at the same time, and the connection of performance, merits, and skills with pay ends up putting a counter-process into play.

DISCUSSION AND CONCLUDING REMARKS
This study aims to contribute to the understanding of accounting as a performative practice. This was because studies indicate accounting both as a practice with potential performative effects, and as a practice causing counterperformative effects. Although the idea of IDP is about "making the salary individual and differentiated" (Collective Agreement 2018, p. 7), our case shows how it unfolds differently in practice. Formulated as two propositions, we here Electronic copy available at: https://ssrn.com/abstract=3712359 discuss how performativity of accounting may be successful or fail (counter-performativity) depending on to what extent the actors located in a concrete-contingent sociotechnical network (Vosselman, 2013) perceive the accounting rationale to be legitimate and to what extent the actors succeed in producing relevant visualizations.
In the search of how merits and performance are set into the IDP for the teachers in Sweden, we observed that there are multi-layered actors involved in the entrance pay determination, salary revision, and promotions. In conceptualizing the process of setting wages, we have observed an iterative process. The direct negotiations between principals and teachers are during the process embedded in a web of relations. This is evident in the "lifting" of issues to municipal-level negotiations where the municipal as employer, represented by HR administrators, and unions are involved. Although the principals largely determine entry level salaries, there are limits regarding discretion. Wider municipal concerns (degree of spread, age, and gender) had to be considered when it comes to annual revisions. In our case, these concerns competed with ideas of differentiated pay related to merits and performance. We have therefore observed obfuscation of skills, merits, and performance in this process (Latour, 1986;Robson, 1992), providing a foundation for our discussions on the performativity thesis (Callon, 2010;MacKenzie, 2006).
Analyzing performativity based on our developed framework is about capturing how practices are shaped by the visualizations of work practices as part of the IDP process. In our case, this would be about how professional work performed by the teachers is made measurable and possible to value and rank, practices that in turn should affect behavior. As indicated in the previous section, there were expressed criteria and the principals used them as one part of the evaluation of the teachers. Still, when they described how they valued teachers, it became evident that key sources of information were not measurements and stabilized criteria (such as counting journal publications for scholars), but rather qualitative and general perceptions gathered throughout the year. Rumors, complaints, and "what one hears" formed a part of these qualitative appraisals. Neither merits nor actual results seemed to be the focus. Instead, performance in relation to well-defined explicit strategic ambitions attracted attention, such as digitalization. Some principals did try to generate a perception through visiting lessons-a method contested due to the superficiality of the situation (Alvunger, 2015). The lack of valuation of the classroom work might potentially be explained as a consequence of a lack of willingness by professional staff to open up their practice to evaluation (Abbott, 1988;Covaleski, Dirsmith, & Rittenberg, 2003). All in all, teaching performance represented a very small part of the wage setting. Because the base for evaluating teachers work was a mixture of rather intangible and tacit practices (hearsay) and a focus on specific nonteaching related aspects (involvement in Information Technology (IT) projects), it is difficult to arrive at the conclusion that IDP affects teaching practices. This because there was a lack of visualizations regarded as relevant and corresponding to the actual work of teachers. If anything, performative effects would arise around outside classroom activities signifying a performative effect.
In our case, the factor that produced significant wage differences was the market situation, putting teachers applying for new positions in a favorable situation. Recent graduates could receive much higher salary levels than well-performing teachers with a long record of accomplishment at the very same school. By many of the respondents, this was perceived as unfair and unreasonable. Besides that, the promotion reforms contributed to differences as well. However, this rather triggered a use of IDP aimed at evening out differences. In this sense, IDP brought about counter-performativity. We have demonstrated the principals' efforts to allocate sums as evenly as possible for the salary revision, promoting teachers (skilled teacher reform) to correct the "wrongly set" entry salaries. HR managers' involvement against discrimination is another example. These are actions that potentially obscure the relationship between wage and the merits, skills, and performance of the teacher in pay determination and more certainly narrows down differences. There seemed to be little support among teachers for too wide a spread when it comes to salary.
Rather than being a source of differentiation, the IDP was used as a counter-performative to narrow the wage span.
The differences resulting from the market-oriented negotiations were largely neither perceived as legitimate nor as related to performance and skills. Justice, fairness, and/or equality existed together with the idea of the idea of rewarding performance and merits, and IDP was used to enhance these values as well.
In order to sum up, and to make our observations more explicit, we present two propositions in respect of accounting as a performative. The propositions are related to our extended theorizing where we used the concept of visualization and the prerequisites of relevance and correspondence to advance the reasoning. We thereby answer the call to further refine our understanding of how and why performatives sometimes misfire (Boedker et al., 2019). Of course, formulating propositions based on one qualitative study is rather bold and we aim only for a small contribution, by pointing out a direction for further studies. The use of propositions helps explicate our findings.
The "first proposition" is that performativity will only be possible if it really gains the status of dominant narrative deemed relevant by a supportive network. The rationale related to the performative must be regarded as a legitimate truth. When it does, transformative change might come about (Kastberg & Ek Österberg, 2017). In our study, this was not the case, and there were competing ideas regarding what to consider as legitimate. We could observe competing narratives about fairness, which open up for counter-performative use of IDP. When scrutinizing the concretecontingent local network (Vosselman, 2013), the acceptance of wide wage spans was limited. "The second proposition" we formulate is that performativity presupposes the ability to produce stable visualizations of core dimensions of the operations. If the first proposition concerns the legitimacy of the narrative, the second regards the legitimacy of what is accounted for. When producing accounts, there is always a selection and a reduction, as we described in the theoretical section. Still, what is selected must be regarded as relevant and must correspond with what is claimed to be represented. In our study, merits and performance escaped being accounted for, which may be less of a surprise because we have described the tacit and hard-to-externalize character of professional work. What we instead observe is how IDP was used for pursuing fairness and equality, and rested upon vague visualizations such as years of faithful service.
When focal dimensions are missed, the risk of misfires increases. This would be the conclusion, and it is in line with observations in other studies (Boedker et al., 2019).
Finally, we return to the practical implication of our study. Similar to what Wise (1994) argues, we have observed how the IDP system is loosely coupled with the individual and organizational productivity measurements, and also that limited space is provided for managerial discretion. Our study also potentially explains observations (Eriksson, Sverke, Hellgren, & Wallenberg, 2002) that IDP to a limited extent affects individual motivation, performance, and organizational effectiveness. The IDP system in action met a "concrete-contingent" network (Vosselman, 2013) limiting the functioning of the performance and merit-oriented system. We argue that we, based on our study, cannot determine how much IDP actually affects behavior, only that there are counteracting forces. Of course it is possible that the communication of the criteria has a governing effect, which in combination with a minor salary raise is enough to trigger aligned behavior, although what to align with remained hard to grasp in our study. Another possibility would be that the multiplicity of actors moderates the effects, and by that might mitigate the potential negative side effects that have been reported as a consequence of performance orientation of professional services (Arnaboldi, Lapsley, & Steccolini, 2015;Brorström, Hallin, & Kastberg, 2004;Smith, 1995). Considering the limitations of this study, our ambition is to point to a direction for further research concerning the performativity of accounting by showing the coexistence of (counter)performativity and practice. To understand how the performativity arises in sociomateriality will require further research that shows time and space dimensions.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

CONFLICT OF INTEREST
The authors declare no conflict of interest.