Men are Always Better? How Swedish Municipalities Justify Pay Differences in Gender Pay Audit Reports

ABSTRACT This study analyses how pay inequality among Swedish municipal employers is justified in their mandatory pay audit reports. Critical discourse analysis was used to examine the overarching message that there are no unjust salary differences between women and men, since the differences found have justifiable explanations. Normative critique for reproducing gender salary gaps can be addressed to the reports, since salary differences between individual men and women doing similar work were justified by reference to men’s experience or special tasks and the organizational policy of individual salaries. Salary differences between groups of men and women were explained by men’s experience and their special and demanding tasks. Market explanations for salary differences were frequently used as a justification of men’s higher salaries. The normative critique can be supplemented by an explanatory critique, as overarching societal discourses, such as employers being held responsible for gender pay gaps in the labour market and technical areas having more value than care areas were identified in the texts. We argue that through pay audit reports inequality regimes in regard to pay are discursively reproduced.


Introduction
Gender inequalities in pay have been researched from various and intersecting perspectives, including economic, institutional, sociological and organizational ones (Milner et al., 2019;Rubery & Grimshaw, 2015).The causes of gender pay gaps have been discussed in relation to women's investment in education, the caring role of women in families, gender segregation in the labour market, traditions of collective wage bargaining, and gender biased organizational practices, to mention a few (Eveline & Todd, 2009;Koskinen Sandberg, 2018;Rubery & Grimshaw, 2015).
At the Swedish national political level, gendered pay has been a recurring topic for decades.The current discourse at national level represents the problem as being partly about women's choices of low-paid areas of work.Partly the discourse on gendered pay is about the wage-setting policies of individual employers.A requirement of gender pay audits undertaken by employers has existed in the Swedish legislation since 1994.The current legal requirement (Discrimination Act SFS 2008:567) obliges all employers with ten or more employees to conduct yearly gender pay audits, by examining wage-determining policies and processes and by comparing men's and women's salaries at individual and group levels to come to terms with any undue differences in pay.
Gendered pay is the only part of the Discrimination Act where employers' obligations are stated in such a detailed way.
Gender pay audits are an interesting example of societal initiatives to come to terms with inequalities in the Swedish labour market.This study does not deal with salary setting or pay policies in the Swedish labour market as such (for such studies, see for example Erikson, 2021;Hensvik, 2014;Linehagen, 2022;Säve-Söderbergh, 2019).Instead, we ask: How are the existence and reproduction of inequality regimes in the pay setting of employers motivated in explanations of pay differences between men and women?
We do this by applying critical discourse analysis (CDA) to the gender pay audit reports that employers need to produce to document the results of their gender pay audit processes.To answer our overarching research question we empirically investigate how municipal employers: (1) Explain their salary policies in relation to gender pay gaps (2) Explain salary differences between individuals (3) Explain salary differences between groups (4) As public sector employers relate to gender pay differences in relation to the private sector

Pay Gaps as Aspects of Inequality Regimes in Organizations
This study illustrates the results and perpetuation of inequality regimes, which is Acker's (2006) concept to demonstrate how gender, race and class inequalities are reproduced in organizations, and why it is difficult to change them.She defines inequality regimes as "loosely interrelated practices, processes, actions, and meanings that result in and maintain class, gender, and racial inequalities within particular organizations (p. 443)."Here we concentrate on gender and gender pay gaps as part of the larger and likewise pervasive issue of gender inequalities in work organizations.
Inequality regimes are based on organizational segregation.Abrahamsson (2002) uses a similar concept of an organizational gender order, which first makes a social construction of two different genders and their characteristics and then sort people according to these notions of gender.The social construction of gender characteristics largely builds on gender essentialism (Ellingsaeter, 2013), and results in a notion that different genders are suitable for different tasks.This gendering of characteristics is part of a larger societal gender order, which makes changes in an organization difficult.Reskin (1993) stated that that female-typed occupations generally confer fewer benefits and lower pay than occupations that are regarded as typically male.Reskin, as well as Acker (2006), have also pointed out that even in mixed gender occupations work tasks are often gender segregated, and women more often get to do tasks with lower pay.Reskin's results are still valid and are a basis of the two steps in the gender pay audits that Swedish employers have to take: examining pay differences between individuals in the same occupational group and examining pay differences between male and female dominated occupational groups.Following Reskin, pay differences in both comparisons can be expected.
Gender pay differences are also to be expected according to Acker (2006), who states that "individualized wage-setting produces inequality (p.450)".When it comes to gender, she observes that not only are tasks male and female typed, but in the organizational construction of the two genders men are constructed as deserving higher wages.This means that even if work tasks are not segregated, men more often than women still get higher pay.Magnusson's (2009) results mirror this, when she finds that even when occupations are rated as enjoying the same prestige, the raters are prepared to give higher salaries to those employees working in a male-typed occupation.
Hence, reasons for unequal wages in an organization can lie in salary policies, which, even though seemingly gender neutral, can have gendered consequences due to the gender segregation of tasks and positions.Unequal wages can also be a result of managers exercising conscious or unconscious pay discrimination.Pay audits are undertaken to discover if any of this has taken place and to come to terms with it.
However, Acker (2006) also points out the organizational incitement to pay gendered wages: women's lower salary levels are the precondition for keeping costs low, both in private and public sector organizations.She regards that as a major reason why gender salary gaps continue to exist.According to Acker, incentives for change from both outside and inside the organization are needed for lasting change to take place.
The requirement of gender pay audits is an incentive from the outside but, according to Acker (2006), it is not likely that it alone will result in gender equal pay.To eliminate pay gaps in an organization also active support from insiders is needed.These insiders, for example the pay auditors, would need to have an understanding of the organizational inequality regime.To the degree that pay differences are due to gender essentialism, the auditors, if they are embodying the same essentialism, have difficulties in detecting them.If policies are regarded as gender neutral, their gendered consequences easily go unnoticed.Moreover, the auditors are pressed not to reveal the inequality regimes in the organization, and this entices them to not attribute pay differences to the existence of such regimes.By disregarding inequality regimes that are the source of the pay differences, the auditors warrant and perpetuate them.This study is about the linguistic means that municipal gender pay auditors use to accomplish that.

Problems with Gender Pay Audits in Organizations
Legislation and organizational strategies for equal pay seldom address organizations as inequality regimes (Acker, 2006) in their entirety, hence the gender pay audit becomes separated from consideration of these other potential organizational gender equality measures (Salminen-Karlsson & Fogelberg Eriksson, 2021).Singling out pay audits as the only positive duty of the employer in respect to pay inequalities has therefore been suggested to be too narrow a legislative strategy (Fredman, 2008;Svenaeus, 2017).
Gendered norms and hierarchies of the organization permeate the gender pay audit process.For example the pivotal phase of job evaluation has been found to suffer from gender bias, in such a way that those jobs that men typically have are determined to be of higher value than the jobs women typically have.Because job evaluation is a tool used to compare the "value", and thus remuneration, of different jobs, gender-bias at this stage can distort the whole pay audit (Alksnis et al., 2008).In addition, the job evaluation does not always feed through to the wagesetting practices in the organization (Koskinen Sandberg, 2017).Indeed, these patterns have been identified in earlier Swedish studies on systems and processes of job evaluation (Cedersund & Kullberg, 1996;Rosenberg, 2004), and it is clear that, despite being intended as neutral and systematic tools for avoiding unjust gender-based disparities in pay, gender pay audits instead risk contributing to the reproduction of gender stereotypes and gender inequalities.
The formulations in the Swedish Discrimination Act mention female-dominated occupations, which are to be compared with occupations that are not female dominated, i.e. the female-dominated occupations are presented as the special case and the non-femaledominated occupations as the norm.It is important to note that employers are allowed to justify gender differences in pay as being fair.The differences can be considered fair if they align with the pay criteria and pay policy of the organization, taking into account, for example, the nature and requirement of the job, individual performance, or the market (Armstrong, 2020).

Research Context: Swedish Municipalities and Their Conditions for Producing Pay Audit Reports
In Sweden, the local governance consists of 290 municipalities.Municipalities vary greatly in size, from 325 to 43,000 employees.Together, municipalities employ 26% of all employees in Sweden, and the vast majority of these (79%) are women (SCB [Statistics Sweden], 2020).The largest groups of employees are female dominated and reflect the municipal responsibilities within social, child and elderly care and schooling.The male-dominated municipal areas, often related to the technical infrastructure, are not as staff intensive as the female-dominated ones.In the Swedish context, the gender pay gap in municipalities is relatively small compared to other sectors: The overall gender pay gap (when part time work is recalculated to correspond to full time salaries) in 2020 was 9,2%, while it was 1,8% in municipalities (Medlingsinstitutet, 2021).
Previously, the fact that alternative employment in the private sector has been more readily available in male-dominated areas of work has potentially inflated the salaries paid by municipalities in these areas.In the last few years, however, the opportunities for alternative employment have increased in female-dominated occupational groups, with the privatization of schools and care institutions.Hence, 'the market'-the need to match the salaries of employers in other sectors-is an external factor that operates actively in salary setting and is mirrored in the pay audit process.Generally, the "market" is regarded as a justifiable reason for pay differences (Armstrong, 2020;Fransson, 2007;Svenaeus, 2017).However, the market is highly gendered since the Swedish labour market is gender segregated, with most men working in the private sector while the public sector is female-dominated.
The gender pay audit is normally conducted by the central HR unit of a municipality.This unit also formulates the policies and principles for the municipal pay structure, in accordance with a mainstream gender neutral HRM-understanding of wage determination based on education, experience, job demands, performance and market-resulting in individual and differentiated pay for employees.Usually the managers are the actual salary setters in the organization.The different areas of municipal responsibilities dispose of varying operational budgets, and this reflects diverse gendered conditions for wage-setting managers.Furthermore, collective agreements with various trade unions influence municipalities' salary setting and individual salaries are also negotiated with trade unions.In conducting the pay audit, the employer and trade union representatives should cooperate, according to the law.This is realized in different ways in different organizations-sometimes union representatives take part in all phases of the work, sometimes a ready-made report is presented to the unions.The HR unit conducts or at least leads the audit and writes the report, but has no decisive influence over the salaries.

Methodology
This study is a sequel to a previous study where we used observations and interviews to follow gender pay audit processes in five municipalities (Salminen-Karlsson & Fogelberg Eriksson, 2021).We have chosen to study municipalities, in spite of their small gender salary gaps, because, as public employers, they are likely to follow the law and actually perform the pay audits and because they have diverse occupational groups to perform the audit on.These features make municipalities a preferred choice, as our interest is not in the salary gaps per se, but in the way they are motivated.In this study we apply critical discourse analysis (CDA) to forty gender pay audit reports that document the results of gender pay audit processes.

Critical Discourse Analysis as a Research Approach
Critical discourse analysis, CDA, has its origins in Norman Fairclough's (1989) book Language and Power.CDA is interested in how power relations, in particular "social wrongs", are created and maintained through language in the social order created by the combination of a number of social practices.CDA has commonly been used to analyse policy texts at different societal levels.
A CDA analysis should include the text itself, the discursive practices in which the text is produced and interpreted, and the social practice that relates to the discursive practices (Fairclough, 1989).A CDA analysis is interested in relationships between dominant societal discourses and other discourses that may be marginal or oppositional.It is common to find paradoxes and internal contradictions between discourses when examining policy texts.
The textual analysis in CDA can be performed in different ways: the researcher can count word occurrences, concentrate on word classes like conjunctions or adjectives, look for metaphors, etc. CDA can work with key concepts in a text and see how they relate to larger societal discourses.The choice of vocabulary can be analysed to see how the social world is represented by these particular words, and what kind of meaning they constitute.The style of writing also conveys meaning (Carvalho, 2008) by establishing a certain relationship to the reader.The analysis can look for internal paradoxes and anomalies, intertextual loans or other aspects that seem relevant.Concepts and phrases that appear repetitively in organizational texts are also of interest for CDA.Holmes (2005) also mentions summarizing as a powerful linguistic device, as it, although seemingly neutral, always includes an interpretation of what is summarized.The summary becomes the official standpoint of the organization that the authors represent.As a text is seen as part of a social reality, it is also interesting to analyse what is not in the text, despite being, or potentially being a relevant feature in the social context (Fairclough, 1989).
The point of departure in CDA is to perceive texts as instances of discursive manipulation of social reality, in pursuit of a certain goal (Carvalho, 2008).This manipulation does not need to be malevolent, or even conscious.Rather, authors of texts are normally immersed in a discourse which itself has a certain relationship to social reality, and the discourse is mirrored in their texts.What is interesting for a CDA analyst is what discourses texts draw on, what they want to achieve and whether they succeed.
Lately, Fairclough and Fairclough (2018) have used the concepts of normative and explanatory critique to clarify the ways in which CDA contributes to the critique of the existing societal system.Normative critique is the more immediate critique made on the basis of discourse analysis, the traditional CDA, criticizing the manipulative means in the text and their aims.In contrast, explanatory critique focuses on wider societal discourses, structures and wrongs, which restrain and direct the actions of the text producing institutions and individuals.

CDA in this Study
We have broadly followed Cummings et al. (2018) suggestion as how CDA can be operationalized into practical steps: Step 1. a) Selection of a research topic related to a social question, b) including a study of past discourses.
The topic of our research project, of which this particular study is a part, is gendered salaries in the public sector, which is related to the social wrong of economic gender inequality in society.After studying the interactional processes of conducting gender pay audits, limited by structural conditions (Salminen-Karlsson & Fogelberg Eriksson, 2021), we wanted to study by which textual means gender pay gaps are made acceptable.To add to our previous knowledge of the prevailing discourses, we conducted a search on the concepts "gender pay difference" and "gender pay gap" in the Swedish Retriever database, which comprises the most important Swedish daily newspapers.
Step 2: Selection and analyses of texts, including a) selection of texts, b) description of how the texts were created, c) analysing the texts at the level of individual words and phrases, how the words and phrases relate to each other in the text and the priority given to different themes.
On basis of our previous research, in following pay audit processes in a few organizations, our understanding was that the staff conducting pay audit processes often used other explanations than gender to justify differences in pay between women and men (Salminen-Karlsson & Fogelberg Eriksson, 2021).We chose to study the resulting audit reports to see how these explanations are communicated to other stakeholders, to show them that unjustified gender wage differences do not take place in their organization or that the differences are minor.
Our approach was to read carefully each gender pay audit.For example, we took notes every time we could see contradictions, between two statements in the text, or between an explanation and the actual data, when that was provided.We took notes about particular words but also grammatical features-such as using a plural form when a group consisted of only one individual-that appeared often when men's higher salaries were justified.We notified when women's lower salaries were attributed to their missing some important qualifications, or when women were absent altogether in the text.We made tables on which words were used of men's and women's qualifications.Even after we had started on the next step, we recurrently went back to the texts to check the basis of our interpretations.
Step 3: Considering whether the question can be addressed in the current social order, or whether it can only be addressed by changing the social order.
From a detailed examination of expressions, we continued to identifying overarching discourses in the texts.While we could be critical of the inconsistencies we had discovered, we also saw that they fitted in and were made possible by the overarching discourses on gender and pay in Swedensuch as technical areas having more value than care areas-and the societal practices related to them.In our final reflections, found in the conclusion section of this article, we discuss to what extent the process of doing gender pay audits and writing gender pay audit reports in this societal context, can or cannot narrow gender pay gaps.
Step 4: Reflections on possible semiotic solution.
The conclusion section ends with a reflection on whether the understanding of gender pay gaps could be discursively changed in gender pay audits.
We are aware that doing CDA on documents in one language and presenting the results in another language adds some issues to the study.The translations from the audit reports are ours.Luckily, the language in these reports is generally quite standardized and simple, which makes it easier to follow the original without losing nuances.

The Analysed Texts
For this article, we have created a database of forty gender pay audit reports, from forty different Swedish municipalities.Thirty-six reports were received as responses to email enquiries to a random selection of fifty out of the 290 Swedish municipalities.The other four reports were collected during our previous empirical study.The reports date from 2017 to 2020.The general gender salary gaps vary normally between 10 and 0%.However, in ten municipalities the general salary gap is to the advantage of women, between 1 and 4%.
The reports vary considerably in length, from five to approximately 350 pages.In some cases, separate appendices with figures supplement the pay audit report, but in other cases the (relatively short) reports only contain text that describes the process and result of the pay audit, and, in some cases, action plans.Two of the reports only contain tables of figures, without explanatory text.
Gender pay audit reports count as official records and hence need to be publicly accessible by law.Even so, the municipalities are anonymized in this paper, and have random numbers as identifiers.

Results -Justifications of Gender Differences in Pay
Most of the pay audit reports follow a routine pattern.They begin by citing or explaining the current legislation, as a motivation for the audit.They have a section on methods, where the computer system used is mentioned, and sometimes, though not always, an explanation of how the legal requirement for collaboration with the trade unions has been fulfilled.A section on the salary policy and salary practices of the organization also arises from the legal requirements.Often, summary results are presented with general figures across the whole municipality.Where reports include comparisons between individuals and groups, this analysis comprises the bulk of the report.Unless the report concludes that there are no unjustified salary differences, it ends with an "action plan" or how the gaps will be eliminated or narrowed.
In the following, we present our analyses of the reports in four sections.The first focuses on the introductory part of the reports, the second on comparisons between individual women and men doing similar work, and the third on caretakers as an example of a male-dominated group that is compared with a number of female-dominated groups doing work of equal value.The fourth analysis section shows how men's higher salaries are motivated by market factors, both in comparisons between individuals and in comparisons between male and female-dominated groups.

The Introductory Texts
There are some striking similarities in the introductory texts.A few texts, from municipalities using the same pay audit system provider, are partly identical, which means that they are not actually formulated by the municipality itself but generated by the pay audit system.
The reports in general find very few gender salary gaps.The introductory texts signal an expectation that this will be the case.Especially when salary policies are described, there are often hints that gender pay gaps will not be found, such as in this statement, found in three of the reports (in two of the reports it is actually used twice): The salary policy of [19, also 25 and 26] municipality is gender neutral and thereby vouches that possible gender-related differences in salary and other benefits will not occur.There are also statements that stress the infallibility of the analysis and the word "ensure" is quite often used.In the text from the system provider, cited by several municipalities: This document summarises the result of an extensive task with the aim to ensure that all employees, regardless of gender, are salaried on the basis of equal factors.
Besides using the demanding word "ensure", this quotation also points out that the work has been extensive.It also precedes the analysis by using the system provider's terminology, i.e. "factors" instead of the word "criteria".This delimits the gender salary policy to those factors that are assessed by the system.

Explaining Salary Differences between Individuals
When the reports explain why some men have higher salaries than their female colleagues doing similar work, two explanations emerge more often than others: experience and special tasks.
"Experience" is by far the most common.It is used both when men's higher salaries are explained and when women's higher salaries are explained, by those municipalities that choose to do so.More experienced employees are seen as being worth a higher salary.
One of the municipalities [14] explains clearly what they regard as experience: "Because experience cannot be statistically determined, age and length of employment are generally used as a basis for comparison."Similar ideas are stated in different ways in several other reports.Several other municipalities, however, define experience, as applied in their job evaluation schemes, in a way that looks more elaborate: "Work related, relevant experience that is required to be able to perform a task independently and with satisfactory quality [22]."Even for those municipalities, however, the number of years seems to be most important, and, for example, shorter employment history with more varied work tasks is never mentioned (cf.Salminen-Karlsson & Fogelberg Eriksson, 2021).Hence, in practical pay audit work, "experience" boils down to being older than those one is being compared with and/or having been in the organization/in the job for a longer time, and normally does not refer to any particular qualifications.The graphs that come out from the computer systems also invite to this interpretation because they plot salaries against age and length of employment.Hence, experience becomes a comfortable explanation.It can easily be used when the men are older or have been employed longer, but it is sometimes used even when women, according to the graphs, have a longer background in the municipality while men have higher salaries. 1 In [21] the salaries of the 95 male caretakers, which are much higher than the salaries of the 5 female caretakers, have several explanations: The salary difference is due to differences in experience, the nature of the tasks and individual and differentiated salary.
Here, three common explanations are put together.The nature of the tasks is expressed in different ways in different municipalities, but the idea conveyed is that men have more specific tasks demanding more competence.In the groups doing similar work, men's tasks could be "extra qualified", "specific", "special", "extended", "heavy", "strategic", "supervisory" and "managerial".
A formulation, used by some municipalities is "the nature and weight of the task".Here, it is not only indicated that men do something of a different nature than the women who, supposedly, all do similar work, but also that this task lies heavy on their shoulders.
The presence of individually differentiated salaries is presented by some municipalities as an argument for gender not being the reason for women's lower salaries.While this can be seen as a natural consequence of the overall gender blind salary discourse of HR staff, it goes directly against the rationale of the law, which is to make sure that individual and differentiated salary setting does not result in gendered salaries.

Explaining Salary Differences between Groups Doing Work of Equal Value
There are very few male-dominated occupational groups in municipalities.One such group is caretakers, which is male dominated in almost all of the thirty-two pay audit reports in which they are mentioned.It is a group with low educational requirements but quite diversified tasks: Most often caretakers take care of buildings (maintenance, repairs, keys), parks and yards, equipment of different kinds and transports as well as sorting mail.In smaller municipalities they often work alone at a site, such as a sports arena or a school.In the pay audit groupings, in some municipalities, caretakers are a group of their own, in other municipalities there are additional work titles such as janitors, carpenters or garden workers, in the same group.
Twenty-two of the reports present comparative analyses between caretakers and jobs of equal value.The salaries of caretakers are compared with female-dominated groups that also, mostly, do not have high educational requirements.There is some variation, but commonly such groups are library assistants, day-care assistants, municipal childminders in their own homes, assistant nurses, other caring jobs, cleaners, school kitchen staff and receptionists.Among the female-dominated jobs there are some (such as assistant nurses and day-care assistants) that actually require a certain level of education.This is interesting in itself: since work evaluation schemes normally place significant emphasis on education, it could be expected that these occupational groups would be evaluated on a higher scale than caretakers.
It can also be seen that, almost without exception, caretakers have higher salaries than employees in female-dominated groups which have been evaluated as doing work of equal value, i.e. there are gender salary gaps.These gaps are presented as acceptable and justified by a limited number of linguistic means.It is common to describe (at least some) caretakers' tasks as extended, common, demanding, or particular compared to the tasks of the comparable female dominated groups.Such words are not used when female-dominated groups are described.This circumvents the fact that the work evaluation has found the jobs to be similar.While men's tasks are often described as special and demanding, however, an explicit comparison is normally not made.This lack of comparison appears in regard to education, too.When some employees in female-dominated occupational groups requiring a certain level of education (such as day-care assistants) do not have this education, it is used as an explanation for why the group has a lower salary level than caretakers (who often do not have particular educational qualifications).
In the group day-care assistants there are differences in the educational background.Some have further training, while others do not have relevant education for the occupation.Day-care assistants were part of the investment in 2019 year's salary revision, where educated employees got an allocation supplement.[. . .]The difference between caretakers and day-care assistants has been almost halved.[28] The statement describes day-care assistants as lacking qualifications and conceals the fact that even those day-care assistants with the highest salaries (presumably the educated ones) still earn less than the caretakers with the highest salaries, and the day-care assistants with the lowest salaries (presumably those lacking the relevant education) would need a 22% salary raise to earn as much as the caretakers with the lowest pay (whose education is not mentioned).
Normally, caretakers' higher salaries are explained away with such rhetorical devices.Even when a gender pay gap is acknowledged, it is done in a non-committal way.As an example, [31] finds that they "cannot exclude the possibility that the salary difference between caretakers and home care workers has an indirect relationship to gender", because the difference "cannot altogether be explained by justified arguments".Hence, the gender pay gap is expressed in very cautious terms rather as a possibility than a fact.The report continues by explaining that: The difference can partly be explained by caretakers having longer experience than homecare workers, shortened working hours 2 and that some of the homecare workers have restricted responsibility.
Since there is the possibility of a gender pay gap, however, the report recommends an extra allocation of ten euro/month/home care worker "to be distributed individually and differentiated and has to mirror achieved goals and results".
What is typical here is the reference to home care workers' allegedly restricted responsibility.The responsibility of the caretakers, to which home carers' responsibility is compared, is left undescribed, and the reader is just left with the impression that caretakers, in contrast to home carers, have a fair amount of responsibility.
It is even more important for the municipalities to present acceptable explanations when it comes to gender pay gaps between groups than when it comes to individual salary differences, since remedying the situation would be more costly.Explanations, which in closer scrutiny do not really hold, can nevertheless be presented as acceptable, as both the writers and the readers of the reports are part of the same discourse where male-dominated occupations by default warrant higher pay.

Accepting Market as an Argument
Most of the justifications analysed thus far relate to internal organizational factors.In contrast, market forces, as they are commonly called, are an external reality that has a strong influence on municipal salary setting.Municipalities need to attract competencies which are in demand among many employers.
In the audit reports, "market" is regarded as gender neutral, as in this one from [24] and [31] (exactly the same wording): Because the market influences salary setting in a similar way for women and men in jobs exposed to competition, market influence on salaries is regarded as a justified argument for salary setting in the analysis.This means that a salary gap between a male group of electricians and a female group of librarians is not due to gender.The male electricians just have a higher paying alternative job market.
"Market" is mainly used when justifying salary differences between male and female-dominated occupational groups.It is not similarly relevant when explaining individual salary differences in a single occupational group, as the pull from market forces outside the municipality should be approximately the same for individuals doing similar work.That said, several pay audit reports also use market forces as an explanation when individuals are compared.
When comparing individuals, the market is sometimes used as an explanation when the work the group is doing is not actually similar.Such groups can be, for example, "project leader" or "head of division", where the higher salaries of leaders of male-dominated projects or divisions can be said to be caused by market forces.For example, [09] uses one single phrase to explain why individual men's salaries are higher than individual women's in groups of administrators, foremen, architects, project leaders and specialist nurses, and also when women have higher salaries than men: There are no unjustified salary differences.The salaries can be explained by experience, achievement and/or market factors.
Male-dominated occupations are often in technical areas.The market argument is deployed strongly as an explanation: namely that there is a demand for technically-educated people in the private sector, and thus municipalities have to pay high salaries to attract and retain employees in these areas.The following statement from [27] is fairly typical: The salary difference in favour of the male-dominated group is explained by market factors.An analysis according to the legal requirements does not show any unjust salary differences.
According to the salary database of Statistics Sweden, the salary levels of system technicians lie higher than the salary levels for the roles of refugee coordinator.
Market factors are used as an explanation, and it is pointed out that the analysis is made according to the legal requirements and the result is legally acceptable, in spite of the salary gap.Instead of really doing an analysis of what allegedly are market factors-in a particular municipality, for a particular occupational group-national or regional statistics as such are used as a reference point.
A few municipalities, such as [23], have observed that with the increasing privatization of education and care, some female-dominated occupations also now have market alternatives.Those municipalities are still a small minority, however.When the report from [23] describes the situation, both male and female-dominated groups are described as having a specialist competence: The employer's need for front-edge competence/special competence forces the municipality to adjust the salary level in some instances, with the aim to recruit staff with the desired competence.This is particularly clear in cases where the employer competes with the private sector regarding individuals with a competence which is difficult to find and to recruit.
Historically, the technical, male-dominated occupations have referred to the market as an explanation for salary differences.For the present, the market has changed and a number of female-dominated occupations are also difficult to recruit.
When explaining salary differences between some technically-educated male-dominated occupations and female occupations with the same work evaluation, the municipality uses a fairly standard formulation about the male-dominated groups being "difficult to recruit" and having "specialist competence" or even "specific specialist competence".The legal requirement of only inspecting occasions where men have higher salaries contributes to the reports repeating and reproducing men's special competence, as it is only their higher salaries that need to be explained.
While most of the comparisons between groups follow the same pattern in most reports, there are exceptions.Thirteen municipalities actually find unjustified salary differences between groups.The municipality [14] sticks out because the market is not used at all in explanations of salary differences.Instead, gender differences in salaries are acknowledged and quite a lot of thought is devoted to actually analysing the salary gap in quantitative terms, resulting in a precise amount of money needed to level the salaries.Making such precise statements can possibly convince a politician reading the report that the suggestion for desired compensation to level the salaries is well founded.

Resisting the Market as an Argument
One of the forty pay audit reports, from [05] takes a totally different, holistic approach to the problematic influence of the market.Instead of justifying the salary gaps between male-dominated and female-dominated areas with statistics, special competencies, or other such issues, this report states the impossibility of closing the salary gap as a single public sector employer.This is done in the introductory text: In the [05] municipality, the salary levels of the surrounding world influence in different ways.Partly they are about certain occupations which have an alternative labour market with totally different salary levels from those of the municipal reality.As we in Sweden to a high extent have a gender-segregated labour market, this means that those occupations that exist both in the municipality and in the private market are maledominated, such as engineers and technicians.That results in the salaries in those occupations being set in a totally different way from other comparable occupations.[. . .]This is something we do not accept, but we cannot, either, do anything about it, if we should have these occupations in our activities.We work continuously to still limit the effects to as great extent as possible.
When then performing the comparisons between the three male-dominated groups that have higher salaries than a number of female-dominated groups, the report starts by stating that "in all [the three] cases they are groups with a clear alternative labour market, which means that the market salary is higher than generally in the municipal sector".The report then gives the average salaries of each male and female-dominated group used in the comparison, without any further explanation.This way of presenting the data suggests that the men's only special qualification is to have chosen an occupation that is valued in the private market.

Discussion
The results of this paper show how municipalities as employers explain and justify gender pay gaps in their own organization, in the legally-required gender pay audit reports.The overarching message of this paper is that essentialist notions of gender characteristics and the resulting gender segregation are mirrored in texts which are supposed to analyse and change unequal gender regimes.
The task of the introductory texts of the pay audit reports was largely to convince the reader that the salary policy of the municipality was gender neutral.However, such statements were presented with no evidence that the consequences of implementing the policy may not result in gender unequal wages (cf.Koskinen Sandberg, 2017).
When analysing pay differences between male and female individuals doing similar work, gender segregation was not assumed to exist.However, a common explanation relied on presenting the work of the male individuals as more qualified or responsible.Likely, this was often the case, following Reskin's (1993) and Acker's (2006) observations of gender segregation of tasks.However, as the reports did not consider the inequality regimes more widely the salary gap was presented as acceptable.
In the sections where salary gaps between different occupational groups were presented, the municipalities needed to deal with occupational segregation.Often arguments were lacking, but the idea of the male dominated group's tasks being more qualified was sometimes written clearly and in some other cases indicated by referring to the nature of the female dominated group's work.The relatively weak argumentation begs the question whether the general essentialist idea (Acker, 2006;Magnusson, 2009) of men naturally being entitled to higher salaries was at play.
When the market argument was used, the societal gender order was the determining factor.In the Swedish society the salaries in the female dominated municipal sector must not exceed the salaries in the private sector where men mostly work (see Erikson, 2021).This means that several male dominated occupations have an alternative labour market with higher salaries.While this is clear evidence that an organization cannot create and sustain a gender regime irrespective of its surroundings, it does not mean that the organizational gender regime cannot stretch the limits of the societal gender order.However, almost no municipalities tried to do that.
The Discrimination Act envisions organizational pay policies and practices where women and men get equal pay, in the hypothetical situations of similar performance in similar jobs.However, in organizations the general HR discourse rather deals with pay differences which are justified with the help of such characteristics as education, experience, job demands, performance and the market.In the salary discourse on individual pay decided by such factors, gender equality as a concept does not hold a powerful position.
Normally it is just a few men and male-dominated groups that are the root of gender pay differences in a municipality.Because it is difficult to deal with the problem of the high salaries of these minorities, the most natural way to deal with the issue is to state that they are entitled to their higher salaries.This approach discursively strengthens the general gender ideology, in that it reinstates men as special and worthier than women (cf.Acker, 2006).
While our analysis keeps close to organizational discourses, we now relate the texts in the pay audit reports to two societal discourses: the higher economic value of technical jobs compared to caring jobs and employers being responsible for gender pay gaps.
Through a CDA lens, our analysis would imply a normative critique directed at the individuals conducting the audits and writing the reports.There are constraints on the room for action of the HR staff, however, both when it comes to salary-setting practices and to performing gender pay audits.A deeper understanding of the problem results in an explanatory critique.
In that critique, the gender pay gap appears between sectors, rather than among individuals and groups of the same employer.The municipal sector is the feminine sector in society, both when it comes to the gender composition of employees and the traditional gendered distribution of tasks.The municipal sector also includes a minority of male occupational groups.Generally, men in Sweden work in the private sector, doing the production work that is traditionally masculine.The male occupational groups in the municipal sector can therefore be seen as exceptions, or visitors, in the feminine sector.The general discourse on equal pay as the responsibility of employers, however, obliges municipalities to integrate those male groups in the salary structure of the feminine sector.This is quite difficult as their "ordinary" salary in the male-dominated sector is on the other side of the gender salary gap.Hence, the main problem is not that municipalities pay gendered salaries, but that municipalities constitute a sector on one side of the gender pay gap-the sector on the losing side.
Being responsible for gender pay gaps as employers is the position assigned to the municipalities in the in the Discrimination Act, and it is from that position that they act.When held responsible for issues over which they have no power of change, the municipalities try to prove discursively that they have no or minimal gender pay gaps, no matter what the salary comparisons between women and men display.Painting a positive picture of the situation is essential for all potential recipients of the report: the Equality Ombudsman can otherwise require actions concerning unjustified gender salary gaps, the politicians need to be assured that the municipality follows the legislation on gender-equal salaries and, in case the report is also accessible to employees, the salary situation must seem just.To introduce inequality regimes as an explanation for gender pay gaps is hardly doable in this context.
Hence, societal gender inequality, as the origin of gender pay gaps, is, paradoxically, reproduced by the legal requirement of pay audit reports.It might not, however, be totally impossible for some municipalities to move slightly from the position that they have been assigned.One example is the way [05] refuses to accept the position of being the one to be made responsible.By changing positions [05] also challenges, not only the discourse about employers' responsibility, but also the discourse on men being worthy of their higher salaries.While such moves do not profoundly change the discourses on men and their occupations being of higher value, they can abate their reproduction.

Concluding Remarks
Using CDA to analyse gender pay audit reports makes visible the paradox that a tool which is intended to forward gender equality, instead results in the reproduction and legitimization of gender pay differences.The study also shows how the reproduction of pay inequalities at the organizational level is made possible by the overall societal discourses on the value of women's and men's work.In addition to an empirical contribution, the study provides a contribution to policy setting by demonstrating the inefficiency of gender pay audits as a tool that all employers are obliged to use.Examining gender pay audits with CDA confirms earlier research by showing that gender pay audits as tools are not strong enough to come to terms with societal gender pay gaps.
Our results would be interesting to compare with other countries and their approaches to manage gender pay gaps in the public sector, as well as male dominated organizations in the private sector.

Notes
1.In our previous study we learnt that this can be done, for example by arguing that a man may not have a long employment history in that particular municipality, but may have had experience of similar tasks from previous jobs.This argumentation was not used for women.2. This refers most likely to home-care workers' shortened working hours, even if it does not look so in the text.