GENDER – THE MISSING DIMENSION IN COMPARATIVE INDUSTRIAL RELATIONS?

GENDER – THE MISSING DIMENSION IN

COMPARATIVE INDUSTRIAL RELATIONS?

An operationalization and test of Rubery’s and Fagan’s (1995)

gender approach based on a sample of 15 European countries.

Master Thesis

Master of Science / International Business and Management June 2010

Supervisors:

Dr. N. A. Lillie and Mr. A. Visscher Faculty of Economics and Business

Nettelbosje 2 9747 AE Groningen

The Netherlands

Nadja Dörflinger Nieuwe Ebbingestraat 109a

9712 NJ Groningen The Netherlands nadja_doerflinger@web.de

Abstract Page 2

ABSTRACT

This thesis deals with the diverging field of comparative industrial relations and aims on the further development and strengthening of Rubery‟s and Fagan‟s (1995) gender-based approach. It proposes a systematic, conceptual framework claiming that features of national IR systems impact on gender variables, such as earning differentials and segregation by occupation. Their model is transformed into a quantitative approach allowing for hypotheses-testing. Thus, new insights are yielded as a further, statistical perspective is added. Data of around the last ten years is retrieved from diverse sources for the EU-15 and examined with multiple regression analyses. The results suggest large variances in the gender wage gaps and the extent of gender segregation by occupation between the sample countries. The three independent variables national minimum wage, collective bargaining coverage rate and the use of part-time work proved to be statistically significant in the gender wage gap analyses. Nationally defined minimum wages as well as high collective bargaining coverage rates are likely to decrease wage gaps between men and women, whereas an extensive use of part-time work tends to widen them. This shows that patterns of wage determination and work organization influence equality in labor markets. Gender segregation by occupation comes along with earnings differentials. Besides, the analyses show that the indicators used to estimate the extent of segregation strongly depend on the female labor force participation rate. This is likely to bias the results. Overall, significant evidence could be found that characteristics of national IR systems impact gender differentials and equality within a country. Thus, both Rubery‟s and Fagan‟s (1995) model and the role of gender as determinant in comparative industrial relations could be strengthened as it also proved to be valid in the context of a quantitative analysis. Moreover, the results show that combining IR research with gender-related aspects adds value.

Index of contents Page 3

INDEX OF CONTENTS

2 COMPARATIVE INDUSTRIAL RELATIONS ... 13

2.1 The study of comparative industrial relations ... 13

2.2 Dunlop‟s institutional approach (1958) ... 14

2.2.1 Explanation of the model ... 14

2.2.2 Sources of critique ... 16

2.2.3 Common refinements ... 16

2.2.4 Evaluation of the institutional approach ... 17

2.3 Poole‟s strategic choice approach (1986) ... 18

2.3.1 The strategic choice model ... 18

2.3.2 Evaluation of the strategic choice model ... 19

2.4 Kuruvilla„s and Erickson‟s change and transformation model (2002) ... 19

2.4.1 The change and transformation model ... 19

2.4.2 Evaluation of the change and transformation model ... 20

2.5 Black‟s cultural approach (2005) ... 21

2.5.1 The cultural model ... 21

2.5.2 Evaluation of the cultural model ... 22

Index of contents Page 4

2.6.1 Corporatism ... 22

2.6.2 Varieties of capitalism (VoC) ... 23

2.6.3 Development theory ... 23

2.7 Discussion ... 23

3 GENDER – A FURTHER DIMENSION IN CIR? ... 25

3.1 Link to feminist theory ... 25

3.2 Possible theoretical explanations for inequalities ... 25

3.3 Rubery‟s and Fagan‟s gender approach (1995) ... 27

3.3.1 Background ... 27

3.3.2 Dimension 1: gender pay equity ... 27

3.3.3 Dimension 2: gender segregation ... 29

3.3.4 Evaluation ... 30

3.4 Diverse gender frameworks ... 31

3.5 Discussion ... 33

4 SOME HYPOTHESES ... 35

4.1 Gender pay equality ... 35

4.2 Gender segregation ... 36

5 METHODOLOGY ... 38

5.1 Research approach ... 38

5.2 Operationalization ... 39

5.3 Sample ... 41

5.4 Modeling and model specifications ... 42

5.5 Data collection ... 44

5.6 Summary ... 45

6 RESULTS ... 46

6.1 Selected general results ... 46

Index of contents Page 5

6.2.1 Results ... 47

6.2.2 Interpretation ... 50

6.3 The second dimension: Gender segregation by occupation ... 52

6.3.1 Results ... 52

6.3.2 Interpretation ... 54

6.4 Reliability and validity ... 57

7 DISCUSSION ... 59

8 OUTLOOK ... 63

APPENDIX ... 65

Index of tables Page 6

INDEX OF TABLES

Table 1: Functional and numerical flexibility according to Kuruvilla and Erickson (2002, p. 178 ff) .... 20

Table 2: Comparison of different CIR approaches ... 24

Table 3: Evaluation of Rubery's and Fagan's (1995) gender-based CIR approach ... 31

Table 4: Gender- or women-related publications in IR journals ... 33

Table 5: Hypotheses linked to the wage determination system ... 35

Table 6: Hypothesis linked to unionism ... 36

Table 7: Hypothesis linked to the organization of work ... 36

Table 8: Hypotheses concerning gender segregation ... 37

Table 9: Steps in the deductive research process according to Gill & Johnson (2002, p. 34ff) ... 38

Table 10: Sources and characteristics of the dependent variables ... 44

Table 11: Sources and characteristics of the independent variables... 45

Table 12: General characteristics of the data of the dependent variables ... 46

Table 13: Regression analysis for the GWG based on average values (R2=0.736) ... 47

Table 14: Regression analysis solely based on the significant independent variables (R2=0.705) ... 48

Table 15: GWG analysis over time ... 48

Table 16: Co-relations between the GWG and wage determination systems (R2=0.517) ... 49

Table 17: Bivariate GWG-correlations ... 49

Table 18: Determinants in Central Europe based on the average GWG (R2=89.8%) ... 50

Table 19: Regression analysis for occupational gender segregation based on average values ... 52

Table 20: Relation between the ID/MSS/IP and the female labor force participation rate (2000) ... 53

Table 21: Correlations between GS and the GWG/use of part-time work (based on average values) ... 54

Table 22: Analysis including the female labor force participation rate as control variable ... 55

Index of figures/appendices Page 7

INDEX OF FIGURES

Figure 1: Examples for determinants in IR systems ... 9

Figure 2: Interrelation of gender and IR variables ... 10

Figure 3: Problem statement ... 10

Figure 4: Illustration of Dunlop's industrial relations system ... 15

Figure 5: Poole's multidimensional strategic choice model (Poole, 1986, p.15 with modifications) ... 18

Figure 6: Wage determination in EU-countries in 1995 (Rubery & Fagan, 1995, p.214) ... 28

Figure 7: Operationalization of the dimension gender wage equality ... 39

Figure 8: Operationalization of the dimension gender segregation ... 41

Figure 9: Relation of the two dependent variables ... 47

Figure 10: Gender segregation and the female labor force participation rate ... 53

INDEX OF APPENDICES

Appendix 2: Scatterplot analyses of the dependent variables ... 66

Abbreviations Page 8

ABBREVIATIONS

CB Collective bargaining

CIR Comparative industrial relations

CME Coordinated market economy

EC European Commission

EU European Union

GS Gender segregation

GWG Gender wage gap

HRM Human resource management

ID Index of dissimilarity

IP Standardized or Karmel and MacLachlan-index

IR Industrial relations

IND Individualism

LME Liberal market economy

MAS Masculinity

MNC Multinational corporation

MSS Moir and Selby-Smith segregation indicator

OECD Organization for Economic Co-operation and Development

OLS Ordinary least squares

PDI Power distance

UAI Uncertainty avoidance

UN United Nations

VoC Varieties of capitalism

* Significance at a 0.1 level

** Significance at a 0.05 level

*** Significance at a 0.01 level

Introduction Page 9

1 INTRODUCTION

“Industrial relations requires a theoretical core in order to relate isolated facts, to point new types of inquiries, and to make research more additive.” (Dunlop, 1958, p. 6)

Dunlop‟s quotation stresses the need to strengthen the theoretical background in the field of industrial relations (IR). His institutional approach is one of the most popular theories to analyze IR, also focusing on national differences. He identifies institutions, contexts and an underlying ideology to be the main parts of an IR system. This has been a starting point for the development of refined theories. Well-known mainstream approaches stress for example the influence of human behavior and decisions or of transformations on national IR systems. Moreover, there are various context factors, such as technology or the state of the economy. Not commonly accepted, but controversially discussed alternative models claim that for example culture or gender variables are interrelated with an IR system. This indicates that the field of comparative industrial relations (CIR) is shaped by divergence, as there is no common agreement regarding the factors causing differences in national systems. This is illustrated in figure 1.

Figure 1: Examples for determinants in IR systems1

Since the end-1980s, the question if gender is a missing dimension in IR theory and CIR in particular has been raised. Mainly female researchers have argued that IR is gender-biased and advocate for establishing a gender perspective in that field. This is due to reciprocities between gender variables and IR. Some brief examples are shown in figure 2. On the one hand, an IR system is likely to influence work-related gender patterns. Discrimination, e.g. in the context of earning differentials between men and women might be caused by the shape of national frameworks. On the other hand, gender aspects also influence IR systems, e.g. in terms of gendered labor market participation. Women and men could have different demands concerning flexible work, parental leave or workplace security. Moreover, role models could influence the work of mixed teams. Those examples show that there might be interrelations between gender and IR. This has also been incorporated in EU policies: Equality between men and women was officially put on the agenda in 1975 (Rubery, 1995). Since 1997, the topic has even gained in importance due to the EU summit in Luxembourg. The members accorded in adding the

1 _{Please note that this figure does not contain a complete list of determinants and context factors that have been discussed in IR}

theory. It just stresses the variety and the resulting divergence.

Introduction Page 10 strengthening of equal opportunities as a fourth pillar in their employment guidelines (Rubery, 2002). As a result, the so-called gender mainstreaming2 has been requested. Nonetheless, although there have been raising female labor force participation rates and laws to promote equality, differences between men and women still prevail in European labor markets.

Figure 2: Interrelation of gender and IR variables

Such developments have also been incorporated in research. There are publications on gender issues and IR, but studies with an explicit focus on CIR are scarce. Rubery and Fagan (1995) claim that the shape of a national IR system influences gender aspects in a country. Their approach considers variations in gender variables as an output factor of IR. The study proposes a systematic framework to examine which IR variables affect earning differentials and gender segregation. Their concept has never been explicitly tested in a quantitative way, even though a methodological mix could increase its analytical and explanatory power (cf. Locke & Thelen, 1995). Generally, there is a lack of quantitative gender-related IR research. Consequently, this thesis will fill the research gap by transforming Rubery‟s and Fagan‟s (1995) model into a statistical approach. The derived problem statement is shown in figure 3.

Figure 3: Problem statement

2

The United Nations (UN) define gender mainstreaming as “… the process of assessing the implications for women and men of

any planned action, including legislation, policies or programs, in all areas and at all levels. It is a strategy for making women‟s as well a men‟s concerns and experiences an integral dimension of the design, implementation, monitoring and evaluation of policies and programs in all political, economic and societal spheres so that women and men benefit equally and inequality is not perpetuated.” (United Nations, http://europeandcis.undp.org/gender/gendermainstreaming, 2010/04/20).

Introduction Page 11 Thus, new insights concerning statistical co-relations (based on hypotheses-testing) between gender and IR will result. This contributes to find out if gender serves as a valuable dimension and helps to explain the prevalence of national differences. Therefore, the following research question in applied:

Do characteristics of national industrial relations systems affect both the gender wage gap and gender segregation (and if yes, which ones)?

To answer the research question, two sub-questions are formulated. Breaking down the research question allows for a more structured answering process. Moreover, the sub-questions are linked to the structure of the thesis. An extensive literature review aims on systemizing and explaining different CIR-models. The most common approaches are explained, analyzed and compared. Based on those findings, hypotheses will be derived. Therefore, the first sub-question (S1) is:

S1. What is the current state in the field of comparative industrial relations and particularly regarding the inclusion of gender as a further dimension?

The answer of S1 will provide an overview about CIR in general and the status of gender as a possible further dimension. The second sub-question (S2) deals with the methodology. Rubery‟s and Fagan‟s (1995) basic gender approach is developed in the context of a conceptual paper. Therefore, it must be operationalized to enable hypotheses-testing; this is reflected in the second sub-question:

S2. In which way can Rubery‟s and Fagan‟s (1995) gender-based approach be operationalized in order to allow for hypotheses-testing?

To transform the gender approach into a quantitative model, it must be figured out which IR-related variables can be associated with the gender issues proposed by Rubery and Fagan (1995). Moreover, suitable statistical tests have to be chosen; all this is covered by S2. To answer those questions, a deductive approach is applied (cf. Thomas, 2004). IR research has mostly been based on inductive models (cf. Strauss & Whitfield, 2008). Edwards (2005) notes that this has one main advantage: IR is an interdisciplinary field; inductive approaches do not exclude disciplines which might provide suitable explanations. But there is also a weakness: a lack of explicit theory-building. There is a tendency to respond, but not to create something new (Edwards, 2005). Nevertheless, researchers (e.g. Onwuegbuzie & Leech, 2005) demand a method mix to increase the analytical and explanatory power of theories as triangulation is encouraged (cf. Strauss & Whitfield, 2008).

Introduction Page 12 differs compared to typical forms of work. Thus, wage determination and work organization patterns contribute to (pay) equality between both sexes in working life. Occupational gender segregation comes along with higher earning differentials between both sexes. This is due to the fact that wages in „female‟ occupations and branches tend to be lower and less regulated compared to male working regimes. Yet, problems concerning the use of the indicators to measure the extent of occupational gender segregation could be discovered due to their dependence on the female labor force participation rate. Therefore, their accurateness is likely to be biased. On the whole, the tests prove that gender variables are influenced by a national IR system, which strengthens Rubery‟s and Fagan‟s (1995) model and the role of gender in IR (theory) in general. It shows that combining feminist and IR research adds value, too. The outcome of the thesis will be relevant in both a professional and an academic sense. Professionally, it is an interesting piece of work for HR departments of MNCs. It might be used as a guideline for hiring policies and helps to explain national IR frameworks. This might be valuable for location decisions, for instance. It is also interesting for unions in Europe, as many of them have included gender aspects in their agendas and promote equality (cf. Dickens, 1998). Finally, especially globally active and working women may benefit from the information offered concerning their job location decisions. Academically, it contributes to the further development of the field CIR and particularly to the refinement of the gender model as example for an alternative approach. Perhaps, it might contribute to the establishment of an integrated, holistic approach in CIR as this is still missing. It also reflects the growing importance of gender studies or the studies of minorities in general and votes for the inclusion of such variables in business and management. Lastly, the gender approach can be used as a measure for equality in a society or more specifically in IR systems. Equality is reflected in laws and possibly also in value systems of countries. It is currently demanded by many societies, intergovernmental organizations and by MNCs‟ codes of conduct. Therefore, it is interesting to check how equal IR systems are and which implications result. Besides, the thesis might also add value in the field of economics, as distributional aspects within societies are considered.

Literature review Page 13

2 COMPARATIVE INDUSTRIAL RELATIONS

2.1 The study of comparative industrial relations

The field of CIR aims on explaining national differences in a structured, theoretical way (Stauss, 1986). Locke et al. (1995) state that a “description without the support of an analytical model or theoretical

argument has little value for advancing understanding or providing the information (…) needed”

(Locke et al., 1995, p.158). This need of theory-building is also stressed in CIR definitions, e.g.: “Comparative industrial relations is a systematic method of investigation relating to two or more

countries which has analytic rather than descriptive implications” (Bean, 1994, p.4). Therefore, the

essential determinants shaping national systems must be identified and specified related to theory-building. This also guarantees effective cross-national comparisons. For Poole (1986), the main aim of comparative theory is to evaluate the implications of context factors on differences in national IR systems. Based on this requirement, Poole (1986, p. 8ff) defines four crucial principles:

(1) Focus on structures and processes within a society

Reasons for divergence are originally related to the context factors. One important example is society; it is likely that causes for divergence can be explained with related structures and processes.

(2) Multidisciplinary perspective

A comparative approach must allow for a multidisciplinary perspective. Consequently, a suitable model should contain insights of various social sciences. The need for the resulting holistic perspective is also stressed by other scholars, such as Bean (1994) or Arts (1990).

(3) Explanations

Purely descriptive methods and categories within the comparative approach should be avoided as the core of a model are its explanatory conditions. Thus, explaining instead of describing is demanded.

(4) Historical and contemporary perspective

The timeframe of the analysis should include both historical and current developments. This helps to understand origins for certain patterns, related developments and tendencies.

Literature review Page 14 because of the diversity of cross-national IR frameworks. Although institutions might have the same overall tasks, they could perform in different functions. Bean (1994) uses „participation‟ as example. In France, profit sharing is associated with it whereas in other national contexts, it is understood in the sense of employee involvement. Third, Henley (1978) highlights the problem of culturally-biased researchers, as objectivity in their studies could suffer. Fourth, attempts to focus on theory-building in IR and CIR suggest that methodological problems tend to occur, as not all IR variables can accurately be operationalized and measured (Bean, 1994).

The field of CIR was strongly influenced by the creation of Dunlop‟s (1958) institutional approach. His multidimensional model became a standard in analyzing and comparing national systems. In the next section, his approach as well as further developments will be explained, analyzed and discussed.

2.2 Dunlop’s institutional approach (1958)

2.2.1 Explanation of the model

Dunlop‟s (1958) institutional approach originates from observations and analyses of the building and coal mining sector in several countries. The establishment of the model was benefited by a few factors: First, it was supported by the importance of pluralist3 thinking concerning work after World War II (Nicholls, 2003). Second, the prevailing dynamism in the American society especially regarding working life (e.g. migration, American dream) led to a demand for new models explaining employee relations (Nicholls, 2003). Third, the opening of the previously political field of IR to other disciplines, such as social science or business also benefited Dunlop‟s theory (Ackers & Wilkinson, 2008).

Dunlop‟s (1958) framework is shaped by the assumption that differences in national IR systems are reflected in institutional settings. To explore the reasons for the cross-national variety, he derives assumptions from systems theory4 and applies them to IR. This leads to the following definition: “An

industrial-relations system at any one time in its development is regarded as compromised of certain actors, certain contexts, an ideology which binds the industrial relations system together and a body of rules created to govern the actors at the work place and work community” (Dunlop, 1958, p.7). This is

illustrated in figure 4. IR is seen as a unique sub-system; he distinguishes it from a purely economic model and opens his theory to sociology as well.

Dunlop (1958) focuses on the three different actors in an IR system, namely government (e.g. special government agencies), employers (e.g. in terms of employer associations) and workers (e.g. organized as unions). He specifies it by stating that “the actors in given contexts establish rules for the work place

and the work community, including those governing the contacts among the actors in an

3

Pluralism recognizes that there are different interests between employers and employees. Thus, a conflict of interests is likely to occur (Daniels, 2006).

4

Literature review Page 15 relations system. This network or web of rules5 consists of procedures for establishing rules, and the procedure for deciding their application to particular situations” (Dunlop, 1958, p.13). The idea of an

IR system establishing a web of rules describes the interwoven relations and power distribution between the actors. Those are institutions in contrast to persons – they are therefore not gendered and human aspects like preferences or behavior are neglected. The model considers an abstract, macro-level without including micro-level variables which would enable a more detailed overview on how the institutions work. The actors administer and possibly change patterns over time. The expression and enforcement of rules is diverse, e.g. through laws or collective bargaining agreements. The variance of rules in national systems is explained with the help of different contexts.

Figure 4: Illustration of Dunlop's industrial relations system

IR systems are embedded in certain contexts which Dunlop (1958) divides into three parts: technological features of work places, market constraints and power distribution in society. Technological features influence both employers and employees, e.g. in terms of the production organization or safety issues. Especially regarding safety issues, differences between the sexes tend to occur, e.g. concerning pregnant women. Market structures also vary between countries causing implications for IR systems. Dunlop (1958) points out differences between state-protected industries and those who have to cope with global competition. Power distribution refers to the negotiation powers of the three actors, e.g. regarding the establishment of rules. Particularly the dominance of a certain actor may influence a system‟s development, structure and the rule setting process (Dunlop, 1958). As the actors are not gendered, variables like role models or equality are not considered although they might impact on power distributional patterns. The IR system would be incomplete without an ideology. Dunlop defines it as “a set of ideas and beliefs commonly held by the actors that helps to bind or to

integrate the system together as an entity” (Dunlop, 1958, p.16). It influences the roles of the actors and

can provide a certain degree of stability. If the actors‟ ideologies are congruent, rules are more likely to be acceptable for every party. Gender aspects are not considered to be a part of the ideology.

5

Literature review Page 16 Later on, Dunlop‟s institutional approach has been associated with the theory of convergence. Kerr et al. (1961) confirm the importance of Dunlop‟s model. Nevertheless, according to their approach, technologies and other homogenizing factors will lead to convergence. In the long-run, diversity will remain on an individual level, but IR systems will tend towards a higher degree of uniformity.

2.2.2 Sources of critique

From the 1970s on, critique concerning Dunlop‟s model has evolved. To cite some examples: Bain and Clegg (1974) assume that IR systems are more open than in Dunlop‟s model. Clegg (1979) criticizes the missing explanations about power distribution within the society. Moreover, the link between it and evolving job rules is not clear. Adams (1991) points out that Dunlop‟s framework does not allow for hypothesis-testing. Bean (1994) highlights that it is impossible to analyze system changes based on Dunlop‟s model. Moreover, he suggests that different developmental stages also impact on national diversity. Those are just few examples, but critique has also evolved from different schools of thought. Hyman (1975) as a representative of the Marxist model emphasizes the dynamics of social relations; additionally, he demands a historic perspective. Dunlop‟s (1958) approach seems to be stable, but the three main actors are subjects of change. Hyman (1994) encourages studying that, e.g. based on developments like globalization. Bean (1994) also stresses that IR systems react to changes. Critique has evolved from scholars representing the strategic choice model, too. According to Kochan et al. (1984), actors are influenced by their environmental contexts when making choices. Thus, the role of individuals and decision-making processes within the system is highlighted. Streeck (1988) confirms the close link between Dunlop‟s and Kochan‟s models as choice is largely dependent on prevailing institutional regulations and agreements. Feminist scholars such as Forrest (1993) criticize the focus on institutions instead of persons in Dunlop‟s model (1958), as this excludes a gender perspective.

2.2.3 Common refinements

interest-Literature review Page 17 based approach. The first advocates for adding a historical dimension (cf. Thelen & Steinmo, 1992)6 whereas the later takes preferences of different groups of actors into consideration. Overall, those models have something in common: the actors are still non-gendered institutions and gender-related context factors are not taken into account. Thus, they do not assume that the roles of men and women differ in working life and neglect possible interrelations between gender and IR.

2.2.4 Evaluation of the institutional approach

Dunlop‟s (1958) framework is a holistic model to understand national systems and to explain sources of distinction. It integrates various factors shaping an IR system into one comprehensive framework. Therefore, it provides an entire structure for the field of IR – that did not exist prior to Dunlop‟s model (Nicholls, 2003). Nonetheless, there are weaknesses. The term „system‟ which is used to describe IR implies a certain order. In reality, there might be several sub-systems with different norms, values or ideological backgrounds. Furthermore, a „system‟ suggests a certain balance between the actors but problems like unemployment or labor disputes occur. Thus, it might be difficult to discover equilibrium (Nicholls, 2003). This also shows that the model does not deal with conflicts; there is rather a focus on cooperation even though the actors‟ interests are conflicting (Nicholls, 2003). Rules play a key role in Dunlop‟s model. It is assumed that they are understood and interpreted similarly by all actors, but complications are likely to evolve (Nicholls, 2003). Additionally, the model does not consider implications of an unequal power distribution between the actors, but this could influence stability. Besides, the role of change is neglected (Nicholls, 2003; Bamber et al., 2004). Huiskamp (2004) adds that the system itself is described too statically and demands a more dynamic view on IR. Related to that, system creation is not incorporated as well. A further limitation is mentioned by Locke et al. (1995); they criticize that Dunlop‟s study has an American bias, as the researchers who worked with the data were mainly American. Finally, Hansen (2002) criticizes that the model does not leave space for the inclusion of gender variables and labels it as gender-blind. The actors are not gendered, gender domination roles are not considered and the context factors do not include for example family obligations or relations. Dickens (1999) suggests modifying the institutional approach regarding equal opportunities with a focus on the state‟s legal regulation, the trade unions‟ social regulation and the employers‟ voluntary regulation or compliance. Hantrais and Ackers (2005) extend these thoughts and state that EU policy implications should be added as a fourth pillar to this framework.

Dunlop‟s institutional approach has largely influenced the field of IR. Prior to it, “fact-finding and

description rather than theoretical generalizations” (Winchester, 1983, p. 101) prevailed. It creates a

framework describing the system‟s structure, the role of actors, different determinants of the context as

6

This mindset can also be linked to the theory of path dependence (e.g. Teague, 2009 or Ebbinghaus, 2005). Its basic idea is for example expressed by Hyman (1982), who claims that “the present must be viewed as historically conditioned and

Literature review Page 18 well as the role of the underlying ideology. The model has also encouraged further research in (comparative) industrial relations. Meaningful selected approaches are explained in the next sections.

2.3 Poole’s strategic choice approach (1986)

2.3.1 The strategic choice model

This approach assumes that strategic choices of the actors influence institutions and practices in an IR system. Therefore, national differences are also reflected in the choices of the main actors. In contrast to the institutional approach, it views the actors as persons and considers features of human behavior and social action of the different parties. Anyway, the actors are not gendered. The model was created by Kochan et al. (1984), inspired by case study research about changes in the US IR in the late 1970s. It includes behavioral variables in Dunlop‟s (1958) model. Poole (1986) transforms it into a process-related framework, but the basic ideas remain the same as illustrated in figure 5.

Figure 5: Poole's multidimensional strategic choice model (Poole, 1986, p.15 with modifications)

Strategic choices are shaped by factors like political and social structures or the power distribution within a society. They are embedded in contexts containing subjective meanings, policies and societal structures. The decision making processes of the actors are not independent; intervening variables like institutional or organizational issues are likely to have an impact. Examples are relationship patterns, informal aspects of organizations or corporate strategies. None of those (social) context variables explicitly includes gender-related factors even though they could be meaningful. The results of choices are wide-ranging and depend on the nature of the decision. Strikes, wage scales or new HRM practices

En v iro n m e n ta l c o n d it io n s In te rv e n in g c o n d it io n s Pro x im a l c o n d it io n s C h o ic e s O u tc o m e s ‘Subjective’ meanings and policies (1) Socio-cultural values (2) Political ideologies (3) Economic policies (4) Public/legal policies Structures (1) Social (2) Economic (3) Political (4) Technological (5) Legal (6) Demographic Organizational structures and policies Industrial relations institutions

Industrial relations processes

The marshalling of power resources in interaction, accommodation processes

Strategic choices of the actors

Literature review Page 19 are only few possibilities (Poole, 1986). In the end, Poole (1986) argues that IR systems will diverge over time. Although countries become more industrialized, the cultural and social backgrounds will shape a variety of heterogeneous systems. Kochan et al. (1984) highlight the dynamic component of the model, as decisions tend to influence the relationships of the actors. This interest-based view is also combined with the notion of strategy. It stresses that the roles of the actors can be formed in many ways and implies a certain amount of structured long-term thinking of the parties involved in an IR system.

2.3.2 Evaluation of the strategic choice model

This framework tries to be more holistic than Dunlop‟s approach. Referring to CIR, it means that national differences are not only reflected in institutions, but also in the choices of the actors. It adds a focus on individual human actors. Consequently, a micro-level variable is included in the macro-level framework. It also inserts a dynamic component as choices may result in strategic adaption processes. Therefore, there might be an overlap with Wailes et al.‟s (2003) approach of combining the new institutionalism with an interest-based view. There are also shortcomings. Although strategic choices are stressed, the model does not explain how choices are made and which outcomes are likely. Thus, a closer look into behavioral science might be useful in order to increase the model‟s accurateness. Besides, it does not become clear how strategic choices differ in distinct national contexts. This might decrease the extent of possible generalizations. Furthermore, Poole‟s (1986) procedural model is rather descriptive. To increase the analytical power and stress the impact on cross-national diversity, (statistical) tests could be helpful. Finally and similar to Dunlop‟s model, the actors and context factors are not gendered. Thus, different roles and positions of men and women in the labor market are not considered. In this context, the possibly gendered character of decision making processes is neglected.

2.4 Kuruvilla‘s and Erickson’s change and transformation model (2002)

2.4.1 The change and transformation model

Various scholars advocate for including the role of change and transformation in CIR, e.g. Dunlop (1958) states that “an industrial relations system implies an inner unity and consistency, and a

significant change in one facet of the context or the ideology may be expected to displace an old equilibrium (...) and to create new positions within the system and new rules” (Dunlop, 1958, p. 388).

Based on the institutional model, Kuruvilla and Erickson (2002) develop a framework considering the interrelation of stability and flexibility with a focus on system transformations7. Literature suggests that IR systems are rather stable over time (e.g. Bean, 1994 or Katz et al., 1994). Labor peace and conflict avoidance were prioritized to create stable national IR frameworks after World War II. This was seen as

7

Literature review Page 20 a precondition for economic development. The rising pace of change has modified IR patterns. Thereby, industrial peace and stability have become less important than providing a competitive environment for companies. Thus, external forces put pressure on national IR systems, leading to transformations. There are various possible triggers of change, but gender-related factors (such as rising female labor force participation rates) are not examined. As a consequence, IR systems are likely to provide a higher level of flexibility which can be either functional or numerical (illustrated in table 1). Its main aim is to create more effective labor market patterns. The implications of changes are not considered for different groups of workers, such as men and women, but for the whole system. This will be stabilized after a transformation, but further changes tend to occur over time (Kuruvilla & Erickson, 2002).

Functional flexibility Numerical flexibility

Refers to … strategies which aim on modifying characteristics

of existing labor markets. Thus, both redeployment and motivation of the labor are stressed.

… actions which decrease the job security and the number of the (core)

workforce.

Examples Changes in work organization and processes, flexible

pay structures, improved worker involvement.

More extensive use of temporary labor, subcontracting, retrenchments.

Implication Implies a higher productivity. Implies lower overall costs.

Table 1: Functional and numerical flexibility according to Kuruvilla and Erickson (2002, p. 178 ff)

Most scholars agree to the importance of a dynamic feature in CIR (e.g. Tailby, 2003, Fagan, 2004 or Bean, 1994). The acceptance of Kuruvilla‟s and Erickson‟s (2002) macro-level approach has led to refinements. A common micro-level model is the trade union focused perspective. Both Hurd et al. (2003) and Behrens et al. (2003) indicate that trade unions endeavoring for change can be categorized in distinct strategies. Thus, unions react to external forces, which may also imply changes on both a micro- (union) and macro-level (IR framework). Gender-sensitive refinements have not been developed so far.

2.4.2 Evaluation of the change and transformation model

Literature review Page 21 raises the question if it is mainly applicable to premature or also to mature IR systems without major limitations. Thus, it might be interesting to test its eligibility for other samples.

2.5 Black’s cultural approach (2005)

2.5.1 The cultural model

Black‟s (2005) approach has been benefited by globalization and the growing importance of cultural theories. He argues that CIR models neglect the role of national culture as a determinant for the variety of IR systems. To build up a new model, Black (2005) uses the institutional framework of Meltz (1993) and the cultural model of Hofstede (1983, 1988, 2001)8. Gender aspects are reflected in the later, as the cultural dimensions masculinity/femininity (e.g. contains role models) and power distance (considers equality) are included. Black (2005) makes use of hypotheses-testing9 to find relations between IR and culture in a sample of OECD countries in the 1980s and 1990s. The statistically significant results confirm the assumptions; thus, he concludes that national culture is one of the main factors predicting the shape of a national IR system. He sets the results in the context of IR in general and argues that the whole field is culture-bound (Black, 2005). More specific gender-related conclusions are absent. The role of culture in IR is seen ambiguous in literature. Few scholars stress its role, e.g. Ahiauzu (1982) found out that culture influences the actors‟ behaviors; this results in culture-bound decisions. Andersen (1988) compared the work of British and Norwegian labor and their unions on oil-rigs in the North Sea. Although the technology and workplaces were identical, differences could be identified – those are justified with cultural patterns. Bean (1994) claims that cultural characteristics, norms and values influence an IR system. Possible examples are the Netherlands or Belgium – there is a vertical segregation between religious and non-religious groups, impacting on unions and employer organizations (Visser, 1992 and Vilrokx & Leemput, 1992). The role of culture in CIR models is not specified. Dunlop‟s (1958) institutional approach largely ignores it as a determinant, although the notion of ideology could contain hints for culture (within the common understandings). This differs in convergence theory (Kerr et al., 1960, 1973) as culture is mentioned in two contexts: First, it states that industrialism has affected national cultures. Second, it claims that even though IR systems converge, national cultures are responsible for keeping a certain level of diversity. The strategic choice model leaves space for culture, but its explicit role is not specified, too. It is likely that the context comprises variables like cultural as well as other social or economic aspects. (Kochan et al., 1984). Poole‟s (1986) related model includes a link between strategic choice and national diversity. Thus, early CIR models

8

Meltz‟s (1993) framework is based on Dunlop‟s (1958) institutional model, but he adds several context factors as well as a cultural variable (this identifies targets and values within a national society). Hofstede‟s cultural framework is based on the following definition of culture as ”the collective programming of the mind that distinguishes one group or category of people

from another” (Hofstede, 2001, p.13). With the help of surveys in the large multinational corporation IBM, he created five

dimensions which can be used to distinguish different cultures.

9

Literature review Page 22 seem to be more likely to deny the role of culture, whereas more recent ones at least take a possible influence into consideration. Gender issues and roles as a part of culture are widely neglected.

2.5.2 Evaluation of the cultural model

The main strengths of Black‟s model are empirically tested hypotheses in a longitudinal study. With the help of them, a link between IR systems and culture is proved. However, problems are likely to be associated with the approach: First, it is based on a sample of OECD countries which means that Asia, Latin America and Africa are underrepresented. As the sample includes developed countries, it is possibly more suitable to describe mature IR systems. Additionally, the sample is small and the data not up to date. It might be helpful to enlarge it and consider more recent data; this would also benefit the generalization of the results. Second, it could be interesting to check if the measurable IR data can solely be explained with culture. Thus, it could be examined if certain particularities originate in a country‟s history or if they are culture-bound. Beyond, the use of Hofstede‟s model also sets data limitations and raises the question why this model was chosen and not for example the approach of Trompenaars (1996). Finally, the role of gender in Black‟s model is marginal. It is incorporated through the use of Hofstede‟s model, but the IR variables are gender-free. Besides, Hofstede‟s dimensions only give a tendency if a society is more masculine or feminine or equal or unequal. It would be necessary to break this down and examine more precisely which gender variables really interrelate with IR systems. It could also be interesting to consider Hofstede‟s gendered masculinity index (cf. Hofstede, 1997, p.83ff), as there is specific data based on the survey answers of men and women. On the whole, Black‟s (2005) approach offers an interesting framework how to analyze an IR system from a different, cultural point of view. Nevertheless, it might need further refinements to strengthen its meaning in CIR.

2.6 Diverse frameworks related to various disciplines

There are also other frameworks to compare IR systems which are rooted in related disciplines. They rank national systems corresponding to different determinants (the importance of such models is stressed by Edwards, 1992). Thus, based on a macro-view, micro-level implications for IR can be derived. Exemplarily, three approaches are briefly pointed out: corporatism, the varieties of capitalism and the development approach. Those theories are frequently used in IR research, but none of them explicitly considers gender-related patterns and their relation to IR.

2.6.1 Corporatism

A commonly used approach to compare different countries and their IR systems is based on the level of corporatism (Bean, 1994). Wiarda (1997) defines it as an “approach for understanding new social,

economic and political phenomena (the role of the state, the formal incorporation of interest groups into government decision making, new areas of public policymaking, and so on) that other approaches fail to provide” (Wiarda, 1997, p.4). The conception has been applied to IR with a focus on the degree of

Literature review Page 23 influence on wage negotiation processes or the existence of works councils. Therefore, corporatist states with a high degree of centralization were more successful in dealing with problems like inflation or unemployment. Amongst others, Newell and Symons (1987) continued this work with a focus on institutions. They confirm that corporatist countries are better in handling macroeconomic shocks.

2.6.2 Varieties of capitalism (VoC)

The VoC-approach originates of the field of comparative political economy (Hall & Soskice, 2001). It basically argues that the shape of a national economic system influences the economic performance of a country. Thereby, the main actors are companies as their behavior mainly composes the national performance. They must deal with the other actors to increase the benefits. This might come along with coordination problems. The theory distinguishes two different modes: coordination through competitive markets or through strategic interaction processes. Such actions occur within all capitalist economies, but the balance between both modes is crucial. Thus, a continuum is created that distinguishes between two extremes: liberal market economies (LME) and coordinated market economies (CME). In CMEs, companies largely depend on institutional settings whereas in LMEs, the market primarily regulates the relations (Hall & Gingerich, 2009). This conception can also be assigned to IR (e.g. Rugraff, 2006 or Konzelmann, 2005). National IR systems are strongly influenced by the type of capitalism, e.g. CMEs tend to be more regulated in contrast to LMEs. Hamann and Kelly (2008) associate greater pay inequality with LMEs due to the lack of regulation. Thus, it is possible to compare national frameworks based on the degree of state intervention and regulation, for instance.10

2.6.3 Development theory

The theory argues that economic systems pass through developmental phases (Nicholls, 2003). Based on the UK, Nicholls (2003, p.36) identified the following stages: pre-capitalist, early capitalism, emerging monopoly capitalism, monopoly capitalism and post-industrialism. Thus, every period is shaped by diverse social relations; also at the workplace level. Such information is helpful in two ways: First, it can be used for sample selection processes. Depending on the purpose of the analysis, either similar or different countries can be clustered. Second, the developmental stages have implications for IR systems as well (Nicholls, 2003). Therefore, development theory helps to classify countries and related to that, IR systems. In IR research, the theory can for example be applied with a link to the history of a country, a system or the actors in the system (e.g. McIlroy & Campell, 1999).

2.7 Discussion

This chapter shows that there are various theories which aim on explaining national differences in IR systems. Although Dunlop‟s (1958) institutional approach can be viewed as common ground, they all

10

Literature review Page 24 focus on different determinants shaping national systems. The institutional approach highlights the three main actors, their embeddedness in certain contexts as well as an underlying ideology. Altogether, a web of rules within the IR system is created. The idea of IR being a system remained unchanged by most other CIR scholars. A comparison of CIR models is illustrated in table 2.

Institutional approach Strategic choice model Change/ transformation Culture

Authors, e.g. Dunlop, Kerr,

Wailes, Meltz

Kochan, Katz, McKersie, Poole

Kuruvilla/Erickson, Hurd et al., Behrens et al.

Black

Timeframe Late 1950s on Mid-1980s on Late 1990s on Early 2000s on

Core elements

IR as a system, web of rules

Inclusion of the actors‟ strategic choices

Interrelation of stability and flexibility

Culture influences actors‟ choices

Refinements Old, new and

neo-institutionalism

Poole‟s procedural framework

Micro-perspective by Hurd et al., Behrens et al.

-

Approach Mainstream Mainstream Mainstream Alternative

Table 2: Comparison of different CIR approaches

The approaches which were developed after the institutionalism include new variables: The strategic choice model adds a behavioral component, the change and transformation approach a dynamic element and the cultural model stresses the influence of cultural patterns. Those models can be subdivided into mainstream and alternative ones. The mainstream approaches are multidimensional; they are commonly accepted and have probably been refined. In contrast, the alternative ones and their meaning for the field of CIR are controversially discussed. Up to now, they have been shaped by small sample sizes, a lack of tests and seem to be in a premature stage of development. Thus, their applicability should be enhanced, e.g. with further refinements. Generally, the variety of approaches indicates the prevailing divergence in the field of CIR. Edwards (2005) criticizes a common aspect of the theories: the institutional focus in IR research, as it leads to a lack of theory-building. Up to now, there is no commonly accepted, holistic framework that combines the main characteristics of all models. Nevertheless, there are scholars who aim on developing such a framework (cf. Heery, 2008).

Although the separate approaches are different in many ways, there is one crucial similarity: they widely neglect the role of gender. There are few scholars which particularly advocate for the importance of gender in comparative analyses, such as Edwards (2005), who states: “Within IR itself, the integration of

a gendered perspective remains an issue, and one way forward in this area is comparative analysis, for example, of different conditions promoting or retarding gender-based solidarity” (Edwards, 2005,

Literature review Page 25

3 GENDER – A FURTHER DIMENSION IN CIR?

3.1 Link to feminist theory

Feminist theories have evolved from the end of the 19th century on due to the unfavorable conditions of women in working live. In the preceding industrialization, women mostly worked in rural areas, they were badly paid and had nearly no chance to achieve similar careers to men. Recently, women make up about 50% of the workforce, but their position in employment relationships still differs compared to men (Nicholls, 2003).11 Feminist theories investigate reasons for the unequal treatment of women and search for ways to achieve equality. In general, models can be divided into liberal and radical (Nicholls, 2003): Liberal feminism takes the prevailing institutions for granted; thus, alterations should be implemented through reforms and policies (Nicholls, 2003). In contrast, radical feminism demands extensive changes within the structure of a society as women‟s suppression results from systematic inequalities in capitalist societies. Thus, political reforms do not enable the necessary changes (Nicholls, 2003). There is also a postmodern/poststructuralist feminist theory, which explains inequality with the help of everyday meanings (Nicholls, 2003; cf. Nentwich, 2006, for further information).

The different role of women in working life is also reflected in IR. Dickens (1989) notes that “the

predominant focus of industrial relations academic study, and of related disciplines, has tended to neglect the fact and nature of „women‟s work‟, giving the impression that industrial relations academics were either gender-blind or, like industrial relations practitioners, held the view that „if it‟s only women, it doesn‟t matter that much‟” (Dickens, 1989, p.167). This perspective has several implications

(Nicholls, 2003): First, theories which are not gender-biased should be developed to correct or adjust for the current ones. Second, as around half of the workforce is made up by women, it is valuable to understand their specific roles and positions in working life. Third, the amount of women in the total workforce can also change policies concerning the organization of work (e.g. regarding parental leave or job safety). The developments since the end of the 19th century have altered employment patterns with respect to gender roles. This has led to demands of mainly female scholars to include gender in the current IR theory. Possible explanations for inequalities between men and women as well as the role of gender in IR are considered in the next paragraphs.

3.2 Possible theoretical explanations for inequalities

There are few theories which attempt to explain sex inequalities in working life based on structural and economic factors. Those are often linked to discrimination; Reynolds et al. (1998) mention that it “occurs whenever someone‟s opportunities are not based on his or her individual capabilities but are

limited because of membership in a group” (Reynolds et al., 1998, p.171). Three approaches are briefly

11

Literature review Page 26 discussed: human capital theory, segmented labor market theory and reserve army of labor theory (Howe, 2003). Those approaches are commonly used in literature to explain gender differences in labor markets (e.g. by Rubery & Fagan, 1994 or Howe, 2003).

Human capital theory

This theory is based on the assumption that labor markets are free from discrimination. Earning differentials can be seen as a result of different levels of productivity (Reynolds et al., 1998; Howe, 2003). As it is difficult to measure the individual productivity of a worker, variables like education or professional experience are used as proxies. Representatives of the human capital theory argue that women are hindered to invest in their individual human capital, e.g. because of their family orientation; this leads to a decreased productivity. The theory has been criticized by various scholars, as it neglects the existence of discrimination (e.g. Rubery et al., 2005). Nevertheless, it is widely used to measure inequalities between different groups of workers in labor markets (cf. Smith, 1994).

Segmented labor market theory

Another explanation for gender differentials is provided by segmented labor market theory12. It states that a labor market is divided into two sectors (Howe, 2003): The primary sector offers stable employment patterns, relatively high wages and excellent working conditions. In contrast, the secondary sector is shaped by low wages, a high fluctuation, bad working conditions and little chance of promotion (Piore, 1975). Interaction between both sectors is limited (Smith, 1994). This conception can also be applied to gender roles in the labor market. Barron and Norris (1976) found out that women are rather concentrated in the secondary labor market. Atkinson‟s (1984) model of core and peripheral workers can be associated to this mindset, too. Regarding gender, the model implies that women are more likely to be found in the peripheral workforce than men and are therefore disadvantaged (Howe, 2003).13

Reserve army of labor theory

The theory originates from a Marxist mindset using the reserve army as a metaphor for partially employed workers (Rubery & Fagan, 1994; Howe, 2003). Those are regularly attracted and rejected by the labor market (e.g. seasonal workers). Women tend to be a part of the reserve army; they had been involved in the workforce in the world wars and left the labor market afterwards (Dickens, 1989). Therefore, the theory concludes that labor shortages tend to lead to an encouragement of women into the market – when the situation changes, such practices may be withdrawn again. Thus, an amount of workers will be forced to leave the labor market in the context of downsizing (Howe, 2003). As women are more likely to be partially employed, a different treatment might result with women being more likely to be forced to leave the labor market again.

12 _{In economics, the theory is also known as „dual labor market hypothesis‟ (Smith, 1994).}

13 _{Segmented labor market theory is often associated with the job crowding hypothesis in economics. It claims that wages are}

Literature review Page 27 Those theories aim on explaining gender roles in labor markets. According to Rubery and Fagan (1995), such differing gender issues might also be influenced by a national industrial relations system. Therefore, the gender-based approach (which is explained in the next section) was created.

3.3 Rubery’s and Fagan’s gender approach (1995)

3.3.1 Background

Rubery‟s and Fagan‟s gender approach14 (1995) was developed against the background of the creation of the Single European Market in 1992. This led to a revival of discussions about convergence or divergence in European IR systems. Both scholars criticize the lack of gender-sensitiveness in those discussions, as the existing research mainly focuses on male working regimes. Little or no attention is paid to female workers or „female‟ branches, such as the service sector. In order to create a systematic framework for a gender-sensitive CIR approach, they consider two dimensions: gender pay equity (and associated with that, the gender wage gap (GWG)) and gender segregation (GS). Labor markets in the EU-1215 show various kinds of segregation, e.g. by occupation, industry, firm or employment contract. The extent differs depending on the country and its specific IR system. Gender pay equality refers to the fact that women are particularly exposed to low pay. The roots of that are diverse; segregation or traditional role models could be meaningful (Vianello & Siemienska, 1990). Both dimensions are analyzed based on the EU-12 as sample and linked to IR-related variables, such as indicators of wage determination systems.

3.3.2 Dimension 1: gender pay equity

Rubery and Fagan (1995) link gender pay equity to wage determination systems and explain possible implications: First, the impact of collective and legal regulatory systems is assessed. Women are more likely to be concentrated in low-wage jobs and in sectors that are complicated to organize. The legal system can produce relief with the help of minimum wages, as they guarantee a common national or industry-wide wage floor. Legal regulation probably protects women in a better way than collective regulation. The later could reflect different wage levels per industry and those maybe correspond to the gender composition of the workforce. Therefore, the existence of a (national) minimum wage is likely to decrease the gender wage gap. Nevertheless, the level of the minimum wage is crucial as it only provides a base floor. Moreover, its implementation and enforcement must be monitored, e.g. by the state or trade unions (Rubery & Fagan, 1995). Second, collective bargaining (CB) coverage rates are of interest when analyzing gender pay. The gender wage gap is likely to be reduced with higher coverage rates, as common floors are defined. But men are more likely to be covered by CB due to several reasons: Women make more use of part-time work than men; in general, part-timers are less likely to be

14

Rubery and Fagan developed the gender approach based on their work for the Equal Opportunities Unit of the European Commission. They researched effects of wage determination, the GWG and GS (Rubery, Fagan, 1995).

15

Literature review Page 28 covered. Women also tend to be more concentrated in smaller firms with less or no union engagement. In this context, the implementation of regulation standards may be more difficult. Extension agreements are beneficial, too, as the quota of covered workers rises (Rubery & Fagan, 1995). Third and associated to the last paragraph, the quality of collective agreements differs. Some agreements might only provide a wage floor, whereas others regulate additional, workplace-related issues. Coverage rates do not include the extent or quality of regulation – this fact might lead to biased estimations. A corresponding classification of EU-countries is illustrated in figure 6. Based on those considerations, Rubery and Fagan (1995) discuss the implications on gender pay equality in the EU-12.

Figure 6: Wage determination in EU-countries in 1995 (Rubery & Fagan, 1995, p.214)16

In those days, the UK and Ireland had the least amount of regulation. This has changed, as both countries introduced minimum wages. Collective regulation was weak, which led to various wage standards on both company and industry level. Furthermore, regulation in „female‟ sectors (e.g. clerical work, services) tended to be poorer in both countries. The five countries with a general minimum wage but weak or uneven CB have distinct systems. Regulation in Luxembourg differs extremely between sectors; Portugal‟s system is patchy; the Greek system is shaped by firm-specific and regional differences, for instance. France and Spain have faced many changes in recent years. Yet, the effectiveness is still uneven between sectors in France. In Spain, the content of the agreements differs. Thus, although a wage floor is guaranteed, the wage distribution between men and women is not equal. The absence of strong CB patterns tends to widen the GWG. The other five countries offer higher levels of regulation for both male and female working regimes. The three countries without a minimum wage offer protection with the help of meaningful CB. In Italy, collective agreements are automatically extended to workers in the same industry. In Denmark, high union density and membership quotas as well as a voluntary compliance of many companies lead to a relatively stable system. In Germany, firms are largely organized in employer associations which actively support compliance. Besides, the existence of works councils benefits gender wage equality as well. Belgium and the Netherlands offer both strong CB and a minimum wage. Moreover, the use of extension agreements pushes the CB

16

Literature review Page 29 coverage rates in all sectors (but initially, they tend to be high) (Rubery & Fagan, 1995). Thus, those findings stress the importance of the wage determination system and the link to gender pay equality. Both minimum wages and strong CB patterns tend to reduce inequalities. Moreover, the analysis indicates that the European systems differ to a large extent. Thus, the national context does not only shape the IR system, but also gender wage equality.

3.3.3 Dimension 2: gender segregation

The last paragraph has shown that regulation varies between sectors. Such differences could be one root of GS. Segregation can occur in various ways, e.g. by sector/industry, firm, employment contract or occupation. Thus, Rubery and Fagan (1995) analyze the four predominant forms of GS and possible co-relations with the national IR frameworks in the EU-12. The main results are briefly explained:

To examine segregation by sector, male- and female-dominated industries as well as mixed ones are analyzed. Four main conclusions are drawn: First, the results show that regulation differs in female and male sectors, as more efforts are made in predominantly male sectors (Rubery & Fagan, 1995). Second, collective agreements in female sectors proved to be less comprehensive and detailed. Consequently, the quality differs with female sectors being poorly regulated, e.g. regarding working time or pay premiums for weekend work. Third, large inter-industry differentials exist; this might also widen the GWG. This is due to the fact that similar jobs are paid differently depending on the industry. The bargaining system might be one reason, as collective regulation at a local level tends to increase inequalities; the same is valid for weak regulation in general. Nevertheless, there are also contradicting examples: Even though Germany is highly regulated, there are large inter-sector pay differentials. Fourth, differences between the structures in the private and the public sector are identified. Wages in the public sector tend to be structured, regulated and integrated, whereas the ones in the private sector are more likely to be unregulated and fragmented. Thus, pay differentials in the public sector tend to be narrower than in the private sector (Rubery & Fagan, 1995).