Implementing and Integrating Policies on Performance-Based Pay: Coordinating the “One-Employer Approach” in a Swedish Municipality

ABSTRACT In Sweden, performance-based pay has gone furthest in the public sector. This development has placed demands on employers to have a transparent and salient salary review process with which to evaluate employees uniformly and fair – and to act in a unitary manner, as one employer. Pay policies aim to coordinate salary reviews in organizations. However, the transfer of policy to practice is not easy and research has shown a gap between intended and implemented policies. Drawing on a case study in a municipality, this article expands knowledge on policy transfer by exploring and describing how HR, unions, managers and employees perceive implementation and integration of the pay policies, and to what extent the municipality succeeds in a one-employer approach. The findings suggest that the municipality fails to act as one employer and that the units that are most successful in implementing the policy are those closer to top management.


Introduction
This article deals with one of the many challenges in organizations, namely coordination. Coordination is about "bringing activities and things together to create order" (Aspers, 2011, p. 19), which enables an organization to act unitary: as one employer. The oneemployer approach aims to have managers at all levels and in all departments carry out HRM strategies uniformly, in order to bring about social order within the organization (cf. Ahrne & Brunsson, 2009). Through a mixed methods approach, this article studies a Swedish municipality's attempts to implement a oneemployer approach, by focusing on the transfer of pay policies and thus the coordination of HRM in a large, departmentalized organization.
HRM aims to create order and consistency in how employees are handled and treated-regardless of department or manager (Armstrong, 2012;Bowen & Ostroff, 2004). HR policies communicate intentions to align and integrate, that is, coordinate people management. They provide guidelines for the organization to strive in the same direction; for managers to steer in line with the employer's overall goals, and to treat employees' uniformly and justly. Policies on pay are especially important as they aim to motivate employee performance (Perry, Engbers, & Jun, 2009;Schay & Fisher, 2013;Spano & Monfardini, 2018), and their motivational effects build upon their perception by employees as transparent and fair (Andersson-Str åberg, Sverke, & Hellgren, 2007;Colquitt, 2001;Kim, 2016;Mulvaney, McKinney, & Grodsky, 2012;Schay & Fisher, 2013).
Nationally bound public sector organizations should have all the prerequisites to act in a unitary manner, as they serve under the same institutional frameworks, where the coordination of HRM is not disrupted or blocked by variations of industrial relations and labor law (cf. Bamber, Lansbury, Wailes, & Wright, 2015). In Sweden, the employer organization, Swedish Association of Local Authorities and Regions (SALAR), also recommends that municipalities act as one employer. They emphasize justice, and their joint collective agreement with the unions articulates the one-employer approach in that pay policies, salary reviews, and performance appraisal criteria shall be distinct, transparent, and known to all employees (SKL, 2017(SKL, , 2009. However, as shown in previous research (Chow, 2012;Khilji & Wang, 2006;Op de Beeck, Wynen, & Hondeghem, 2018), there is often a gap between intended and implemented HR policies, between general policies formulated by top management and strategic HR and the practices of first-line managers and employees. Drawing on a case study, the present article explores a mid-sized Swedish municipality's endeavors to act as one employer in the salary review process, by analyzing how various actors perceive that the organization's pay policies and activities are implemented and integrated. The purpose is to explore and describe whether the municipality succeeds in the one-employer approach, that is, how and to what extent information and practices of salary settings are articulated and applied in a communal and consistent way throughout the organization.
The study was conducted using a mixed methods approach (Denscombe, 2008), including documents, interviews with HR, unions, managers, and employees, and an organization survey, to analyze perceptions of: (1) the implementation of the pay policy within the organization; (2) the integration of performance appraisal criteria; and (3) the implementation of a new model for the salary review talk. The remainder of the article is structured as follows. Below is a brief review of performance-based pay (PBP) in Swedish municipalities, followed by a discussion of the theoretical conceptualizations, and then a section on methods. The empirical findings are thereafter presented in three themes. The article ends with some concluding remarks.

Performance-based pay in Swedish municipalities
Swedish wage formation has undergone major transformations during the last four decades. The post-war Swedish model, with centralized national wage bargaining, was abandoned in the 1980s, and a loosely coordinated system of sectoral bargaining combined with local negotiations was established in the late 1990s (Lundh, 2004;Thörnqvist, 1999). The solidaristic principle of "equal pay for equal work", based on nationally coordinated tariffs, was gradually supplanted by more individualized desert-based principles of compensation (Baccaro & Howell, 2017). In Sweden, such PBP is called individual salary setting. This development, towards a greater scope for PBP, has gone furthest in the public sector and practically all public employees are covered by agreements giving space for PBP (Lapidus, 2015).
The employer organization SALAR negotiates collective agreements at the national level. In order to ensure prerequisites to act as one employer, there is a joint agreement on the salary review process (e.g. HÖK 2016). The agreement stress transparency, consistency, and justice in annual salary reviews; that pay policies, performance appraisal criteria and salary reviews should be known to all employees; that all employees are familiar with the municipality's goals, and that performance goals are aligned with these (SKL, 2017, 2009).
Swedish municipalities are self-governed and responsible for public services such as childcare, primary and secondary education, but also water supply, rescue services, and waste disposal. Like any other large organization with various businesses, municipalities are often organized into operational areas and districts. Accordingly, they face challenges of coordination, where HR policies from top management level are to be implemented by local management (Khilji & Wang, 2006;Op de Beeck, Wynen, & Hondeghem, 2017;Stanton, Young, Bartram, & Leggat, 2010).
The one-employer approach advocated by SALAR is similar to attempts by multinational corporations to establish and transfer company-wide HRM across sub-units and nations-a one-company approach (Ailon & Kunda, 2009). The aim is not only to align HRM to the overall business strategies in order to strengthen organizational performance, but to develop a joint corporate culture and enhance equity and organizational justice (Björkman & Lervik, 2007;Budhwar & Aryee, 2008), or to align HRM policies across company boundaries in multiemployer networks (Marchington, Rubery, & Grimshaw, 2011). Compared to MNCs, in which managers and employees are in different institutional contexts, the SALAR idea of a one-employer approach seems quite modest since the aim is not that all municipalities in Sweden should act as one employer, but only that each individual municipality should.

Theoretical conceptualizations
The theoretical point of departure in this study is that a one-employer approach implies a successful transfer and implementation of HR policies. To act as one employer-coherent, aligned, cohesive, and integrated -practices must be coordinated throughout the organization. Policies facilitate the coordination as they express which activities, practices, and rules of behavior are warranted (Armstrong, 2012;cf. Ahrne & Brunsson, 2009). Still, carrying out policies involves challenges. Even though a municipality is an organization in a delimited geographical area, embedded in a single institutional framework, it probably faces similar challenges as other organizations when top-level management are to transfer its policies to the divisions. The present study approaches the coordination of HR practices with inspiration from HR policy transfer from headquarters to subsidiaries in the hard case of MNCs (Björkman & Lervik, 2007).

Dimensions of transferring of HR policies
Björkman and Lervik (2007) specified three dimensions of policy transfer from headquarters to subsidiaries: implementation, integration, and internalization. Implementation concerns whether actors in subunits act in accordance with the policy. The first step in implementing a one-employer approach on PBP is to establish a joint and transparent policy. The purpose of a pay policy is usually three-fold: Firstly, it is a strategic document communicating what behaviors and performances the organization values and how pay will encourage and motivate high performance and continuous development in order to achieve organizational goals (Armstrong, 2012). Secondly, it is a guide to the salary review and how it should be carried out, and what responsibilities managers and employees have in the process (Mulvaney et al., 2012). Thirdly, it is a means to coordinate activities so that managers across the organization work similarly towards the same organizational goals (Aspers, 2011).
However, implementation is difficult. Research has shown a gap between intended HR policies and the implementation (Khilji & Wang, 2006). Top-down initiatives may fail due to lack of information and support throughout the implementation (Balogun, 2006). The top managers producing the policies tend to be passive, while middle managers have the difficult task of interpreting and adapting the policies locally with a lack of continuous information. Op de Beeck et al. (2018;cf., 2017) found that first-line managers perceived less support from HR than managers closer to the top, leading to less effectively implemented HRM. In addition, Stanton et al. (2010) showed that policy information spread via email or Intranet risked being blocked or sorted out, which led to gaps in implementation.
Lack of information and support is connected to the second dimension of transfer discussed by Björkman and Lervik (2007). Integration concerns how policies are communicated and discussed, which facilitates the transfer and helps involved parties to a mutual understanding of what performances, behaviors, and attitudes are warranted (Mulvaney et al., 2012: cf. Op de Beeck et al., 2017. Thus, information dissemination is an important aspect of integration and coordination of activities. It refers to how policies, strategies, and core activities are interconnected. On an organizational level, integration is about internal consistency in HRM and its alignment with the overall strategies. Accordingly, integration connects to the third purpose of pay policies discussed above: to ensure that managers (and employees) strive towards the same overall goals.
Also relevant for the current study is the fact that integration may include and accent a local adaption of policies. In a divisional organization, goals may vary due to differences in operations, as is the case for a municipality that covers a lot of different services. Therefore, general pay policy and the performance evaluation instruments need specifying in relation to different operational areas and occupational groups to be useful (Bowen & Ostroff, 2004;Mulvaney et al., 2012). In the municipality under study, this point is illustrated by a decision that the general performance criteria should be concretized and adjusted vis-à-vis local conditions. Adaption relates to the notion of sensemaking, which explains why a policy is ignored or partly decoupled; that is, edited to fit with local interpretations and practices (Sahlin & Wedlin, 2008;Weick, 2001).
The third dimension of transfer, internalization, refers to when a practice is "taken for granted and accepted by employees, and when they see the value of using the practice" (Björkman & Lervik, 2007, p. 322). A policy is fully internalized only when actors are truly committed and attribute value to it. If a practice is performed only as a ritual, it "is likely to yield disappointing results and is more likely to be discontinued" (ibid.). This dimension concerns the underlying difficulty of motivating managers and employees to approve to a certain practice. It connects to the acceptance and legitimacy of a policy and, in the case of pay policy and PBP systems, part of the motivation effect has to do with them being perceived as just and fair (Andersson-Stråberg et al., 2007). However, as discussed by organizational justice theory (Colquitt, 2001), fair pay is not only about distributive justice. More closely connected to the one-employer approach is procedural justice, whereby policies and performance appraisal instruments are used in a consistent and transparent way for all employees so that they can influence the outcome through their own work efforts. Besides, informational justice includes ensuring that employees are equally informed about what performances, behaviors, and attitudes that are valued and how evaluations are done.

Materials and methods
This study draws on a case study approach (Merriam, 1988). For research-ethical reasons, the Swedish municipality under study is not named. It has 75,000−150,000 inhabitants and employs 7,500−10,000 people. The municipality has 10 business-oriented boards and three district boards, each which with local management and administration. The district boards are responsible for the neighborhoods' childcare, education, family and care, while business operating areas such as Living, Environment and Traffic and Business and Work are community common and thus placed closer to the municipal executive.
The overall salary review process was studied broadly in the entire municipality during 2016 via a survey questionnaire, documents (policies, instructions, performance appraisal criteria, audits, etc.), and more deeply within one of the districts where qualitative interviews were conducted and the locally adjusted performance appraisal criteria accessed and analyzed. The analyses in this article are of survey data and the qualitative interview data, while the documents are used primarily for information and illustrations.
For the qualitative interview study, 12 interviews were conducted with 17 individuals: one central and one local HR professional, five representatives from three unions, and three line managers and five employees within the operating area Childcare. The area of childcare was selected in communication with the district management, which also helped in recruiting line managers and employees. The interview guide was semi-structured and thematic. The interviewees were asked questions about the salary review process and more specific questions about the pay policy, salary reviews, and performance appraisal criteria. The interviews lasted 35−60 minutes and were held at the interviewees' office or in meeting rooms. The interviews were transcribed verbatim and coded into seven major themes using NVivo. The present article focuses especially on the themes of support and information, performance appraisal criteria, and salary review talk. Respondents were promised confidentiality by not including disclosing quotes or information that reveals where the study was conducted.
The questionnaire, a total organization survey, was distributed as an anonymous web survey by e-mail to all employees and managers in the municipality. For confidentiality reasons, the actual size of the population is not indicated, as that would identify the municipality. Since it was a total population survey, the response rate is considered acceptable and the number of completed surveys (1980) sufficient for statistical analyses (response rate in the range of 15−30%). A rough response analysis shows that the representativeness for general categories of employees and operating areas was adequate, albeit with a slightly higher non-response rate for unprivileged groups; seventy-three percent of respondents were female, 9% were born in foreign countries, and 4% had temporary contracts (compared to 78% females, 19% foreign-born, and 4% on temporary contracts in the total population). Seventy percent of the respondents had higher education, 25% had vocational or secondary school education, and 5% had primary school education only. Seventy-seven percent of the respondents were regular staff, while 23% had some kind of managerial duties, but only 9% had responsibility for salary reviews.
Three survey questions were selected for analyzing different aspects of policy transfer. The first question, "How did you get information about the salary review process?", aimed to explore the organizations ability to disseminate information, which is as an aspect of policy integration (Björkman & Lervik, 2007) and relates to informational justice (Colquitt, 2001). The question offered six response options, and the respondents were able to give multiple answers ( Table 1). For the statistical analysis, an index called Degree of Received Information was created, ranging from 0−2, where 0 indicated no information received, 1 indicated that information was received through one of the two employer channels, and 2 indicated that information was received through both channels. By using the index as the dependent variable, and organizational and individual level variables as independent variables, an OLS-regression analysis was conducted.
The second question, "What kind of wage criteria are used at your unit?", explored the extent to which performance appraisal criteria were adapted to the local context. The questions also relate to aspects of procedural justice. The respondents had four options (Table 3). Since the particular interest was in those respondents who had specified performance criteria, a logistic regression studied the odds of having adapted criteria. The independent variables were the same as above.
The third question, "Have you been offered and participated in any of the following salary talks?", is an example of how well the organization has implemented policies and also an aspect of procedural justice. The respondents could choose one of four options ( Table 5). The dependent variable in the logistic regression was Odds of having had salary review dialogue, and the explanatory variables were the same as in the other analysis.

Findings
The municipal assembly and executives decided on a new compensation policy in 2014, to be implemented in 2015. The general approach was that of one employer, revealed in writings such as "[The Town] is an attractive employer with the will to…", "…a large employer", and "The workplace is characterized by…". The document states that the municipality has a "clear wage policy" and that the salary review process aims to spur employees to good performances, commitment and job satisfaction, which in turn will improve efficiency, productivity, and quality. However, the policy does not say anything about the salary review process or how employees are to be informed about it. Nor does it contain any specific information about the performance appraisal criteria or the salary review talk process. It does stress the core values that make up the basis of the municipality's general performance criteria.
This study explores how the municipality realizes its intentions to act as one employer by looking at the outcome of three important elements in the salary review process. They are examples of the transfer of policies for implementation and integration (Björkman & Lervik, 2007) and illuminate aspects of informational and procedural justice (Colquitt, 2001). Thus, the focus in the present study is on perceptions of how and to what extent (1) information about the salary review process has been spread throughout the organization; (2) general performance appraisal criteria are integrated and adapted locally; and (3) the new "manageremployee dialogue" model is implemented for the salary review talk. Each of these is initially described through extracts from policy documents and findings from the interviews and explored more closely through the survey data and the statistical analysis.

Dissemination of information on the salary review process
The joint collective agreement states that thorough and equal information on the salary review process is important and a way for the municipality to act as one employer. Moreover, that "a precondition for a well-functioning salary setting is that the employer has a long-term pay policy/ -/known and discussed with the employees" (HÖK 2016, p. 25. Our translation). The agreement highlights that the employees' familiarity with the policy, process, and criteria, is important to create a "positive relation between wage, motivation and output" (ibid.). Thus, the accounts in the collective agreement connect strongly to the ideas of informational and procedural justice (Colquitt, 2001).
Whether this is realized is questionable according to the interviewees. When asked about how the municipality informs employees about the pay policy and the salary review process, the central HR stated that the information is published on the Intranet and accessible to everyone: "We have a lot [of information] on the Intranet so that you easily can get access to our pay policy" (i5). Accordingly, the responsibility for being informed lies upon the employees, as acknowledged by a manager who said: "The municipality really stresses this: it is the employee who must absorb information and has great responsibility to be informed" (i7). This was also acknowledged by the unions, although one union representative explained that "Our members do not look for that kind of information" (i3) and that some occupational groups did not have easy access to the Intranet on the job. The same respondent stated that the employer was probably even not interested in raising awareness about salary as "the employer is grateful as long as you don't discuss pay, as it is a neglected area" (i3) (cf. Spano & Monfardini, 2018). The unions try to raise awareness through e-mails and information meetings, albeit without much success: "We invited, when the agreement arrived, to explain what it meant and everything. Eight persons came and we have 1800 members" (i4).
The employees said that they were quite wellinformed about the review process and the criteria, and confirmed that they had been informed by line managers at workplace meetings: "It has been mentioned that there a new ones [criteria] and that one should look them through" (i10); "They [criteria] are e-mailed to us/ -/I wish that one could talk about them on a workplace meeting" (i12). Some had also received information from the unions. Even so, it seemed as if the pay policy and the salary review process were not as thoroughly discussed as stipulated in the collective agreement. One employee was actually unaware of the policy: "What do you mean by pay policy? […] I can't answer that. I think we get some mail from the Union" (i11).
The survey data partly confirmed this lack of knowledge. As shown in Table 1, employees were informed in different and sometimes multiple ways; 56% said that they had been informed by their manager, while 44% had informed themselves via the Intranet or e-mail. In total, 77% recalled having been informed through any of these employer channels, whereas some of the others reported having received the information from colleagues (16%) and/or their trade union (14%). However, 15% could not recall receiving any information at all. To explore variations of how well-informed the employees were, an OLS-regression analysis departing from the index of received information was conducted ( Table 2). The hypothesis was that in a municipality that had been successful in the one-employer approach, employees in various subunits received information similarly-there would thus be less variation depending on organizational variables compared to individuallevel variables. However, even though the explained variation is quite low, the analysis shows that the organizational-level variables explained more (R 2 = 0.057) than the individual-level variables (R 2 = 0.017). The overall tendency was that people closer to the core were better informed. Firstly, employees in the districts, with their own local administration and HR, were less informed than employees in the joint municipality (Model 1). Yet, the effects are quite low and the significance for two districts are also weakened when individual variables are controlled for (Model 3). Secondly, managers, and particularly those with salarysetting tasks, are better informed than employees with open-ended contracts, who in turn are better informed than the temporary employed. These confirmed differences remain when individual variables are controlled for (Model 3). Thirdly, there are significant differences for the operating area Business and Work, which is part of the Joint Municipality. There is a tendency, albeit one that is not statistically confirmed, for employees in operational areas linked closely to the municipality executives, Municipality and Politics, are better informed than those in Housing, Environment and Traffic, and the operations organized by the districts, namely, Children and Education and Family and Care. Taking all the organizational variables together, employees close to municipality top management, where policies are designed and decided upon and informational material are produced, seem to have a better chance of being well-informed about the salary review process than those in the more peripheral organizational positions (cf. Op de Beeck et al., 2017Stanton et al., 2010).
Turning to the individual-level variables, even though the explanatory value is low (Model 2), there are some statistically confirmed effects. First, employees born outside of Europe are less informed than employees from Swedish origin. Second, the educational level has an effect in that the higher educated are better informed. Still, the effect of country of origin and educational level may have interlinked causes: access to computers and thus the possibility to partake information on the Intranet is dependent on both educational level and language skills. Employees with lower educational levels are found in occupations that seldom require computer work (cleaners, janitors, kitchen aids, etc.), which are also the kind of occupations in which many migrants are employed (Statistics Sweden, 2014).

Integration and adaption of performance appraisal criteria
The second example of a one-employer approach concerns the performance appraisal criteria. PBP put great demands on evaluation criteria in order to be motivating (Mulvaney et al., 2012), but also for the evaluation procedure to be perceived as fair, transparent, and consistent (Andersson-Stråberg et al., 2007). The importance of well-designed performance appraisal criteria and consistency is stressed in the collective agreement: Salaries should be determined by objective factors such as the complexity of tasks, competence, professionalism and other requirements associated with the tasks, as well as the employee's skills and performance against business objectives. (HÖK 2016, p. 25. Our translation.) Index range = 0-2 (0 = no info; 1 = info by one channel; 2 = info by two channels).
The municipality has five general criteria that are presented in three different documents. The idea is that they are to be further developed and specified locally, in relation to operational areas, occupational groups, and tasks. For example, the core value professionalism, which is a common performance criteria in the public sector, needs to be defined in order to capture what it implies in different occupations (Larsson, Ulfsdotter Eriksson, & Adolfsson, 2017). This was also stressed by a union representative who participated in the adaption of criteria in one of the districts: We experienced problems with the criteria being so general … Regardless of whether I was in the kitchen, the janitorial, the school, the preschool or the administration, it was the same [criteria], and that was very weird … We looked at one of the professional groups and made performance appraisal criteria based on that [group]. And there we sat, union representatives and managers and HR. … Then, you get criteria that are targeted at the job, which makes it easier to discuss them in [relation to] the operational area. Otherwise, they become too plain. (i3) The local adaption of performance appraisal criteria was also commented on by HR, who suspected that it may be blocked somewhere "along the way", yet hoped that it would work "reasonably well" (i5). In the district where the interviewees worked, the performance appraisal criteria had been specified in relation to at least some occupational groups. The criterion professionalism was used as an example to describe how a criteria may be interpreted differently depending on the occupation at hand. One of the HR stressed that adapted criteria more fairly evaluate performances within different occupations, as the prerequisites and expected outcomes differs: "The child-minders' top-performance is only compared with other child-minders, and the pre-school teachers with themselves" (i2). The employees also seemed satisfied, but still found the criteria a bit abstract: In the survey questionnaire, respondents were asked whether they had experience of general or specified (adapted) performance criteria in the salary review.
The aim was to find out to what extent the criteria, in line with ambitions within the organization, had been locally adapted. About 40% of the respondents reported that they had used specified criteria, and an equal amount had used the general ones (Table 3). Moreover, 10% stated that there were no clear performance criteria, and an additional 10% did not know whether there were any criteria or not. This shows that not even the general criteria are known to all employees, perhaps as a result of the lack of information dissemination discussed above. The fact only 40% of the respondents reported that their unit used specified criteria indicates an inability of the municipality to act as one employer. It also raises questions about whether some units have been more  responsive in integrating the criteria. In order to analyze the probability (chance) of having specific performance criteria, a logistic regression using similar models as in the previous OLS-regression was conducted. The result indicates a reversed logic from the previous analysis since District Pink followed the call to adapt better than did the centrally placed unit Joint Municipality (Table 4). Moreover, it is less likely that employees in Living, Environment and Traffic, and Municipality and Politics, units placed close to top management, have specific criteria in comparison with Children and Education, an operating area managed by the districts. Once again, the organizational level variables explain more than the individual level variables. The explained variance (R 2 ) is quite low, but the differences remain significant when controlling for individual variables. Regarding the individual variables, only sex is significantly confirmed and shows higher chances that adapted performance criteria had been used for female employees. This, and also drawing on the fact that the districts seem more inclined to adaption, may be explained with references to that it is the large occupational groups, teacher, child-minders, and social workers employed within the districts that have pursued and adapted the general criteria.

Implementation of "manager−employee dialogue"
The third aspect of the municipality's ability to act as one employer concerns the implementation of a new model for the annual salary review talk, developed in alignment with PBP. The manager−employee dialogue gives line managers the responsibility to set the salary raise for each employee, in contrast to the traditional model in which the proposed pay increase is negotiated with the local union. The dialogue model is approved in the collective agreement that stresses a well-functioning performance appraisal dialogue between managers and employees (HÖK 2016). Even though the new model was introduced a few years ago, managers have had the opportunity to choose between the new and the traditional model. However, in the instructions for the salary review process in 2016, the municipality stressed that all managers should use the new model, with the exception of newly appointed managers who should stick with the traditional salary talk.
With the new model, the municipality stresses that PBP is continuous and not just a once-a-year occasion. As pointed out by HR: "[Wage] is not something you negotiate in one review, but something you show during the year" (i2). For this to happen, there has to be an explicit usage of the performance criteria in the salary talk, as explained by HR: "It's decidedly so with the dialogue-model, that one needs to have clear criteria … for it [performance evaluations] to become clear in the [salary] talk" (i5). A union representative clarified the connection: "Both employee and manager fill in [performance appraisal criteria], and then have a dialogue around it" (i3). A manager described the situation in more detail: For Professionalism, then I have read through, looked, thought about it … and then I evaluate that person … Then at the salary talk I put up my evaluation form and we have looked at it … and talked about it. What I see, what you have done and worked with, and discussed that (i7).
Thus, the new dialogue model departs from the performance appraisal criteria and, after a discussion, the talk ends with the manager announcing the pay increase and the new wage to the employee. An employee summarized the model as follows: "I said what I thought, and she said what she thought, and then after a while, she said what I would get in pay" (i11). The interviewees were quite fond of the new model and felt that it made it easier to understand why you got a certain pay increase, and also how to influence the salary in the future: A lot has changed. Before we could get a note in the hallway with the new salary, and with no chance to influence at all. And even if the differences are not that major, you want to know why you get a certain amount and what you can do to increase it even more. (i9) However, the interviews revealed that not all of the employees have had the dialogue model and both HR and the unions were aware that the implementation was not as far-reaching as intended. According to the survey data, roughly half of the employees had been offered a traditional salary talk; 38% were offered the dialogue model, and 10% reported not having been offered a salary talk at all ( Table 5). The lack of implementation could possibly be explained by a very high turnover of managers, with many new managers not being allowed to decide salaries without local negotiation with the union. However, survey data shows that only 11 of the 172 managers in wage-setting positions had worked less than two years. Therefore, the vast majority should have had practiced the new model, if the implementation had been successful. Since the dialogue model is what the municipality advocates, a logistic regression was conducted to analyze the chance that an employee had this particular model (Table 6). According to Model 1, focusing on the organization level variables, it is less likely that employees in District Pink have had the dialogue model compared to employees at the Joint Municipality. Those employed in Municipality and Politics and Business and Work were more likely to have had the dialogue model than those in Children and Education. Temporary employees were also less likely to have had the new model than permanent employees. The significance remains, and even increases in the case of operational areas, when individual-level variables are controlled for (Model 3). With regard to the individuallevel variables, education has significant effects in that the higher the education, the greater the chance of having had the dialogue model. One may note that employees born in "another country in Europe" had a greater chance of having had the new model, but there were only a few such respondents.
The overall tendency in the analysis of the implementation of the manager−employee model is in line with the analysis of information dissemination (Table 2). Units closer to the top management act more in accordance with the one-employer approach than operating areas like Children and Education and Family and Care, which are governed by the districts.

Concluding remarks
This article has studied a mid-sized Swedish municipality's endeavors to act as one employer in the salary review process. The theoretical point of departure was that implementation and integration of HR policies are a necessity for coordinating a departmentalized organization when pursuing a one-employer approach (Björkman & Lervik, 2007;Marchington et al., 2011). This study analyzed the extent to which the organization informs about pay policies, integrates performance appraisal criteria, and implements the manageremployee dialogue across different administrative units and operating areas. Drawing on the survey data, the main conclusion is that the municipality has not succeeded in the one-employer approach; it is only half-way there. There are a lot of variations in the extent to which employees are informed about the pay policy and salary process, have been offered the new model of salary talk, and also had used specified performance appraisal criteria. Since the organizational level variables have greater effects in the regressions than the individual level variables, the variations exist between units and positions, rather than individuals. In other words, employees are managed differently, not because of individual characteristics but because they belong to different units, which stands in contrast to the central idea of the one-employer approach.
One may ask whether the municipality actually has a "two-employer approach". Firstly, administrative units and operational areas that are closer to the top management have been more successful at implementing and integrating the pay policies than the districts and the operational areas managed within them. This center-periphery difference is in line the theoretical assumptions and previous research (Op de Beeck et al., 2017Stanton et al., 2010). From the conceptual perspective of MNCs (Björkman & Lervik, 2007), headquarters are more likely to implement and integrate new policies while sub-units are more likely to either reject or delay the implementation because of geographical distance and/or differences in institutional framework (cf. Bamber et al., 2015;Marchington et al., 2011). This is, of course, not the case for a municipality. However, cultural and symbolic differences may be present and influence organizational processes (Ailon & Kunda, 2009). It may appear obvious that policies are edited, rejected, or de-coupled when traveling out to the local capillary units, since they have to make sense of policies on behalf of their specific conditions (Sahlin & Wedlin, 2008;Weick, 2001). Cultural differences may be present even within a municipality as joint operational areas and districts work under different conditions and with such different tasks. Top management's strategic and unitary ambitions may clash with the districts' daily operational endeavors. The districts are responsible for the kind of public service that meets citizens' needs on a daily basis. PBP systems, and the manager −employee dialogue are time-consuming and line managers in the Swedish public sector, especially in areas like childcare, elderly care, and education are under a heavy workload and pressure just making the daily operations work (Björk, 2013). For example, finding a substitute fill-in for a sick child-minder may be more pressing than informing about pay policies or adapting the performance criteria thoroughly. Björk (2013) showed that line managers in femaledominated areas have a larger control span than those in male-dominated areas. The latter operations are communal, while the former are operated by the districts. Hence, the gap in culture is being that top management does not fully realize the efforts needed locally to implement and integrate the pay policies fully. Op de Beeck et al. (2017 stressed the importance of support to line managers in carrying out people management; support from HR, middle managers, and co-workers is required. It seems as though the municipality lacks support in many respects. HR seemed to be satisfied by having the pay policies on the Intranet, and some managers only e-mailed the performance appraisal criteria to the employees. There is also uncertainty about how some of the instructions actually are communicated within the organizationsuch as the call to integrate performance criteria locally and the use of the dialogue model-as these aspects of the salary review process are not mentioned in the overall pay policy but only in separate documents distributed to managers. The lack of communication from top management and central HR was also present in a study of implementation in public hospitals were line managers complained about receiving important information via e-mail or the Intranet (Stanton et al., 2010). Thus, it seems as though, in order for policies to have a coordinating function within an organization, one should not underestimate face-to-face information dissemination. A case study conducted in a US public sector organization also showed that employees that were engaged in developing the performance evaluation instrument were more satisfied with the appraisal process (Mulvaney et al., 2012). Such an approach is not only a means to reach consensus, but also a way to provide "training, support and resources to managers to enable them to develop their people management practices" (Stanton et al., 2010, p. 589), and also development for employees, HR, and unions.
In sum, the point of departure in this study was a municipality's pursuit, in line with the collective agreement and the call from the employer organization, to act as one employer. Coordination of practices-that is, orderly and unified people management-is facilitated by policies (Ahrne & Brunsson, 2009;Aspers, 2011). The present study shows, in line with previous research, a gap between intended and implemented HR policies (Chow, 2012;Khilji & Wang, 2006;Op de Beeck et al., 2017Spano & Monfardini, 2018). Yet, with support from this case study it is arguable, firstly, that gaps may be the result of differences in interpretations due to cultural differences within a departmentalized organization (Sahlin & Wedlin, 2008;Weick, 2001). Policies may be intentionally vague in order to spur local adaption. Secondly, as also shown by Op de Beeck et al. (2017: cf. Stanton et al., 2010, implementation of policies need continuous support from HR and top management. According to Balogun (2006), top management need to take part not only in the launch of new strategies or policies, but also in the implementation in order to participate in the organization's sensemaking of the policy. Hence, to facilitate a policy implementation, local participation and face-to-face activities should be included so that management can recurrently explain the intentions and facilitate if local interpretations appear to be taking unintended tracks (Balogun, 2006;cf. Mulvaney et al., 2012;Stanton et al., 2010). This may also promote coordination and social order as these activities enable communication, relationships, and trust.

Funding
This work was supported by Swedish Research Council (Vetenskapsrådet).