Sectors or countries? Typologies and levels of analysis in comparative industrial relations

Introduction

Industrial relations studies have traditionally focused on the sector as the most relevant level for the definition of product and labour markets, technologies and work rules, actors and regulations. However, in comparative industrial relations studies attention to the national level has often come to the cost of neglecting the sector. Such neglect is particularly problematic in Western Europe, where the sector has been traditionally fundamental for the governance of labour markets, at the level both of the country and the European Union (EU): most trade unions and employers’ associations are organized and negotiate at the sector level (Marginson, 2005), and the sectoral governance of labour markets through social dialogue is considered a key feature of the proclaimed ‘European social model’ (Kittel, 2002).

Over the past decade, attention has been drawn to the apparent deepening of differences between sectors. For instance, in the case of Germany – the most paradigmatic organized system of industrial relations – it has been questioned whether industrial relations in some expanding sectors (such as telecommunications and other services) correspond to the German national model at all (Batt et al., 2010; Doellgast, 2009; Royle, 2004). Sectoral shifts in employment, in turn, may lead these sector differences to transform the German system as a whole (Lehndorff et al., 2009). More generally, comparative research has also started to look at the sector with more attention (Katz and Darbishire, 2000; Marginson and Sisson, 2004; Marginson et al., 2008). However, such insights are still rather rare and based on comparisons of a maximum of two sectors. A consolidated tradition of systematic international sector comparisons is still awaited, despite some important single-sector studies (Doellgast et al., 2009; Dølvik, 2001; Nergaard et al., 2009; Regini et al., 1999).

This article provides some analytical steps towards the development of an international analysis of industrial relations at sector level. Data on the most important industrial relations institutions are compared across nine sectors and 27 EU countries. To start with, this allows us to demonstrate that industrial relations tend to vary more by sector than by country.

Comparative sector industrial relations

The comparative study of national industrial relations systems dates back over a century. Yet despite the context of increasing internationalization and economic differentiation, the study of sectoral industrial relations at the international level is in its infancy. If we look at the most popular textbooks in comparative industrial relations (Bamber et al., 2010; Bean, 1994; Ferner and Hyman, 1994, 1998; Morley et al., 2006), we do not even find ‘sector’ in the index, and while there are frequent mentions of various sectors, there is no systematic analysis of them. The focus on the national level is particularly visible in existing typologies of industrial relations that tend to adopt geographical configurations (the most recent and authoritative of these is the Industrial Relations in Europe Report for 2008, European Commission, 2009).

Indeed, national borders are very important for understanding industrial relations, as these systems have been generally institutionalized either by national states directly, or within clearly distinct national economic and political spaces. For instance, Crouch (1993) has traced the impact of national state traditions on European industrial relations since the nineteenth century and even earlier; Traxler et al. (2001) have tested empirically the relevance of national ‘path dependencies’ and the resilience of national industrial relations institutions; and Barbier (2008) has highlighted the role of national cultures for the framing of social and employment issues. In addition, industrial relations as a form of regulation of employment is closely interrelated to the welfare state, which is an eminently national institution with clearly demarcated national models (Esping-Andersen, 1990).

However, there is a growing awareness that variation within countries may be as large, or even larger, as between countries. Industrial relations theory has always been aware that one particularly important source of diversity is sectors, given that each sector is shaped by specific product and labour markets, resulting in different workforces, different work practices and different economic contexts. Yet until recently, international comparisons of sectors have been rather rare.

Interest in international comparisons of sectors was a by-product of the corporatist school in the 1970s and 1980s (Cawson, 1985). That approach culminated in a major attempt at mapping sectors and countries by a team led by Hollingsworth et al. (1994). Their pioneering study focused already on questions such as globalization effects on sector governance, hypothesizing that ‘at the same time that it reduces the significance of nations as units of analysis, globalisation may increase the importance of sectors’ (1994: 12). The theoretical reasons behind this hypothesis were the increasing role of socially constructed exchanges at sector level, and the agency by firms and public policy. However, the extremely ambitious and time-expensive research design of that study (a qualitative comparison of case studies across nine sectors and eight countries) could never be completed to a degree that would provide a test to the hypothesis. Only 22 of the 72 (= 9*8) designed cases could be completed, and in three sectors only one country was included: not enough to compare sector and country variation. Moreover, that study, focusing on governance in general, paid little attention to industrial relations, on the assumption that these were already well-studied (Hollingsworth and Streeck, 1994). This however was not (and is still not) true at the sector level. The partial conclusion of the scholars was accordingly the enduring relevance of national differences, especially on industrial relations, and of obstacles to convergence (Hollingsworth and Streeck, 1994).

Katz and Darbishire (2000) re-opened the issue with their comparison of metalworking and telecommunications in seven countries, and their theory of ‘converging divergences’: they argued that owing to internationalization, industrial relations systems converge within sectors, but diverge between them. Because countries may increasingly specialize in some sectors rather than others, this means that in countries with different specializations the dominant industrial relations arrangements (especially collective bargaining) may diverge – even if within each sector they converge. A broader theoretical foundation for this argument is provided by the ‘varieties of capitalism’ (VoC) approach (Hall and Soskice, 2001). According to this, firms vary by the sector they operate in, and states develop complementary institutions, including industrial relations systems, that provide support to some sectors rather than others. In such a situation, globalization fosters ‘institutional arbitrage’ and increases country specialization in those sectors where there is an institutional comparative advantage, such as machinery production in Germany or biotechnology in the USA. The VoC approach has stimulated major advances in industrial relations research, but has been criticized for exaggerating complementarity and coherence within ‘models’ (or ‘types’) of capitalism and, again, underestimating the remaining intra-country differences. Therefore, the various combinations of governance and regulation by country and sector are still an open problem, as for instance the idea of ‘recombinant governance’ suggests (Crouch, 2005). In particular, the degree to which the sectors in which a country does not specialize (such as services in Germany) will follow the dominant national practice, stemming from other sectors, or the dominant international practice, stemming from the same sector in other countries, is still to be explored and understood. For instance, Doellgast (2009) has recently questioned whether the German telecommunication sector governance corresponds to its ‘coordinated’ national model. Additionally, it has been frequently observed that few countries actually conform to the ‘ideal types’ described by the VoC school, and we do not know whether other, more ‘mixed’, countries maintain institutions and models of a predominantly national character, and whether they maintain them across all sectors.

This research problem confronts, more broadly, the enduring debate between national path dependency factors versus international convergence factors. While in the 1950s and 1960s convergence arguments were based on organizational and technological change, more recently they are based on globalization. This debate is relevant to controversies which centre upon typologies of industrial relations systems. Existing typologies have focused on countries and national models, but it may be asked whether they are able to account for inter-sector variation. For instance, with regard to the most recent European typology (European Commission, 2009), five ‘models or clusters’ of industrial relations were identified in the EU, each with clear geographic concentration: ‘organized corporatism’ in Nordic countries, ‘social partnership’ in Central Western Europe, a ‘state-centred’ model in Southern Europe, a ‘liberal’ model in North-Western Europe, and a residual, less clearly defined ‘mixed’ or ‘transitional’ model in post-communist Central Eastern Europe.

While a degree of generalization and simplification is inherent in any taxonomic effort, an important empirical question is how these clusters or models vary by sector. In particular, given that studies of corporatism have focused on the differential ‘encompassingness’ of associations and collective bargaining, that is, their capacity for combining different sectors (Crouch, 1993), it is likely that very important cross-sector variation affects countries whose industrial relations systems demonstrate little ‘encompassingness’, such as for instance UK and most Southern and Eastern EU member states. In this regard, a sector analysis is needed in order to respond to criticisms expressed about classic typologies of industrial relations ‘models’: namely that they distort because, in a teleological manner, they overestimate national similarity, coherence and rationality, while underestimating cross-border influences and internal variation, change and contestation (Meardi, 2004). Both terms ‘model’ and ‘system’ imply internal coherence, integration and rationality, and the term ‘model’ also has a normative connotation as something which should be followed and imitated. For this reason, in the current research when suggesting different typologies we limit ourselves to the more descriptive and less value-laden terms of ‘type’ or ‘cluster’, in a descriptive rather than explanatory fashion. In the next section’s explanation of industrial relations indicators, where discussing how to combine different dimensions, the terms ‘systems’ and ‘regimes’ are used interchangeably to describe theoretical configurations of industrial relations characteristics, and the terms ‘types’ and ‘clusters’ are used in the following, empirical sections when describing real cases.

A further topic of debate, with important policy implications, is the possibility of not merely international convergence between national states, but the development of ‘supranational’ industrial relations models. Nowhere has this issue received as much interest as in the EU. The development of an EU-level industrial relations system has been the object of extensive debates that can only be summarized very sketchily here. From a neo-functionalist perspective, there are reasons to expect that increased economic integration and joint regulation across EU countries will foster the development of industrial relations regulation at the same level. From an institutionalist perspective, however, important scepticism has been raised, for instance by Streeck (1998), because of the persistence of major differences and even incompatibilities between national institutions. It is apparent that such an issue is particularly relevant at the sector level, given that the impact of the European single market and of EU regulations varies by sector, and that a specific effort has been made to develop sector-level European social dialogue, but in a rather uneven way (Keller and Sörries, 1998; Marginson, 2005; Pochet et al., 2009). For instance, the steel sector has been ‘Europeanized’ since the European Coal and Steel Community came into effect in 1952, whereas some service sectors, such as hairdressing, have remained localized with little or no transnational influences until recently. One important effort at investigating Europeanization at the sector level has been made by Marginson and Sisson (2004), who highlighted in particular the impact of multinational companies in fostering a complex double movement: towards decentralization within national sectors, but also, indirectly and more tentatively, towards cross-border co-ordination. Their research, however, focused on only two sectors (banking and manufacturing) in only four countries. Studies covering a broader range of countries and sectors are still lacking.

The questions originally asked by Hollingsworth et al. (1994) but not answered, on the international relevance of the sector level, deserve further analysis given the technological and economic changes of the last 20 years. While new products and new conglomerate companies may have blurred traditional sector demarcations, many economic and regulatory changes have been sector-specific, for instance in the evolution of the European single market. We attempt to answer those questions by exploring industrial relations variation at the sector level more systematically. We hypothesize that the hitherto neglected classifications and processes at the sector level may be as important as the national ones, not only for the current reality of industrial relations, but also for potential transnational developments. This argument is rooted in an emerging recognition of the relevance of the industry level for methodology and theory of labour economics (Sako, 2008), the understanding of intra-national variation in industrial relations (Arrowsmith, 2010), and more generally in international economics and business (Dicken, 2007).

We investigate and compare industrial relations system variation across sectors and across countries, asking whether industrial relations systems vary more by country or by sector; and subsequently, which sectors and which countries are more diverse, and which ones are more internally similar. This in turn relates to the debates referred to above concerning convergence and divergence in comparative industrial relations: do economically internationalized or Europeanized sectors display stronger cross-country similarities than those characterized by national markets and regulations? And do certain countries display higher internal diversity by sector than others, and how does this relate to theories and typologies of national industrial relations systems? While we do not dispose of longitudinal data that would allow us to detect trends of convergence or divergence across time, we can expect that even a ‘snapshot’ of the mid-2000s can highlight links between sector industrial relations, national models and sectoral internationalization patterns.

Second, we analyse the details of the differences across sectors and countries, looking at the configurations, in terms of industrial relations regimes, of the various sectors and various countries. Again, this is related to internationalization debates: which industrial relations characteristics, in terms of associational density, collective bargaining and consultation, prevail in the internationalized sectors as compared with the others? By analysing the data for nine sectors in 27 countries (or at least in all the countries where there are social partner organizations), we are able to test whether the types of industrial relations systems apparent at the sector level hold any resemblance to the established national typologies. In this process, we are able to propose a tentative classification of industrial relations models in the EU at the sector level, and to show how far they depart from the national-level ones.

This analytical contribution can be used to achieve a more fine-grained and nuanced description of European industrial relations than the traditional ones offer. This is important from a firm-centric perspective, as adopted by the VoC approach, as it shows whether firms in non-dominant sectors will operate in the same industrial relations regimes as those firms in dominant sectors. This is also important from a gender perspective, as sectors vary deeply (and much more deeply than countries) in the gender composition of their workforce, and the traditional industrial relations research focus on male-dominated sectors such as manufacturing may result in neglect of the specificities of female-dominated sectors. For instance, Hassel (2007) has discussed at length how exactly the dominance of the sector in German industrial relations, combining with labour market gender segregation, results in specific gender representation gaps and undermines the idea of an overarching German ‘model’.

Selection of industrial relations indicators

We map sectoral industrial relations on the basis of seven dimensions: i) union density, ii) union fragmentation, iii) employers’ association density, iv) fragmentation of employers’ associations, v) collective bargaining coverage, vi) centralization of collective bargaining and vii) the involvement of social partners in socio-economic policy-making. Categories i) to iv) relate to the actors, while v) to vii) relate to the relations between them. All these dimensions are extensively discussed as relevant in the comparative industrial relations literature. In short, the rationales for each dimension are the following.

Involvement in policy-making

Industrial relations structures overlap and interact with state (legal) regulations at the national level, but the degree of involvement of the industrial relations actors in public policy may also vary by sector. This dimension is relevant for the issue of the quality of industrial relations, on which we face a shortage of sectoral data: where industrial relations actors are consulted by government or government agencies, we can expect more potential for ‘political exchange’ (Crouch and Pizzorno, 1978) and therefore for ‘positive-sum’ games and integrative bargaining. The relevant indicator is constructed on the basis of i) whether unions are consulted by public authorities; ii) whether employers’ associations are consulted by authorities; and iii) whether tripartite bodies exist. The relevant indicator is a composite index, summing i) to iii). Thus the variable on involvement in policy-making ranges from observations 0 to 4, with 4 expressing the highest involvement.

Unfortunately, no cross-national sectoral data are available on other dimensions, such as the role of the law, workplace representation systems and strikes. We can assume that the former two do not formally vary by sector, but may vary in practice, while the latter notoriously varies significantly. In comparison to the European Commission’s national typology, three of our indicators are identical (union and employer association density; collective bargaining coverage), and two are comparable: centralization and policy involvement (approximately corresponding to European Commission’s ‘corporatism’). The main differences are our inclusion of fragmentation, a dimension as close as possible to European Commission’s ‘union concentration’ and ‘sector organization’ variables, and our exclusion of workplace representation, because of the absence of data at the sector level. Employee representation systems are dependent on national laws, so their exclusion might lead to an underestimation of national differences. However, it is reasonable to expect that the actual coverage of workplace representation varies by sector similarly to union density, being lowest in sectors dominated by small private companies, such as hairdressing.

Our analysis is based on nine sectors, for which data are available from comparative representativeness studies by the European Industrial Relations Observatory. The selection of the nine sectors maximizes variety, with no pretension of representativeness: it covers both manufacturing and services, and both internationalized and less internationalized sectors: steel, sugar, tanning and leather, civil aviation, railway infrastructure, sea and coastal water transport, hospitals, hairdressing and other beauty treatment, and telecommunications. By concentrating on sectors we suppose that sectors exhibit differences in their ‘nature’, according to such factors as technologies, product segments, degree of internationalization, size and ‘relevance’. The notion, demarcation and definition of sectors are always disputed (Wolfe, 1955); but even though clear-cut definitions of sectors are difficult, our notion and conceptualization of sectors follows the Statistical Classification of Economic Activities in the European Community (NACE) classification, which is based on the existence of sufficient differences in product segments. (For details of the sector definitions, see Figure 1.) However, one limitation of the study is that product segments (NACE demarcations) do not perfectly overlap with industrial relations ‘segments’. Some product segments cover different industrial relations types but also some industrial relations types cover different product segments. Nevertheless we suppose that the sample selection in this work comprises enough differences in industrial relations and product market segments between our sectors in order to be able to be able to contrast them.

OPEN IN VIEWER

Figure 1.

Variation of sectoral industrial relations, by country and sector.

OPEN IN VIEWER

Table 1.

Country clusters of national and sectoral industrial relations

	Type 1:	Type 2:	Type 3:	Type 4:	Type 5:
National clusters of countries a	Organized corporatism	Social partnership	State-centred	Liberal	Mixed
	DK, FI, SE	AT, BE, DE, LU, NL, SI	EL, ES, FR, IT, PT	IE, CY, MT, UK	BG, CZ, EE, LV, LT, HU, PL, RO, SK
Sectoral clusters of countries	‘Dense’	‘Political’	‘Lean’	‘Fragile’	‘Empty’
STEEL	BE, DK, FI, DE, PL, RO, SK, SI, SE	AT, CZ, IT, FR, NL, ES	EL, LV, LU, MT*, PT, UK	BG, HU, LT
SUGAR	BE, BG, FI, IT, RO, SI, SE	AT, DK, FR, DE, HU, NL, SK, ES	EL*, LV*, PT	LT, PL*	CZ, UK
TANNING	FI, EL, SI, SE	AT, FR, NL, RO, ES	BE, BG*, DE, IT, PT	LT*, UK	CZ, EE, HU*, LV, MT*, PL, SK*
AIR	BE, DK, FI, IT, MT, NL, SI, SE	AT, CZ, FR, DE, LT, RO	BG, CY, EL, LU, PL, PT, SK*, ES*, UK*	HU, IE	EE, LV
RAIL	AT, CZ, DK, FI, DE, EL, IT, PL, RO, SK	FR, LU, NL, SI, SE, UK	BE*, EE, HU, IE, LV, LT, PT*, ES	BG
SEA	BE, DK, FI, EL, IT, NL, RO, SE*	FR, DE	LV, LT*	BG, CY, EE, IE, SI*, ES*, UK	PT*
HOSP	AT, DK, EE*, FI, EL, IT, IE, LV, RO, SE, UK	BE, FR, LU, NL, SK	BG*, MT*, PT, SI, ES*	CY, DE, HU, PL	LT*
HAIR	DK, FI, NL	AT, IT, SI	BE, FR*, PT*, ES*	DE, LU*, SE, UK*	BG, CY, CZ, EE, EL*, HU, IE*, LV, LT, PL, RO, SK
TELECOM	DK, FI, FR, MT, SE	AT, EE, IT, NL, SK, ES	BE, CY, EL, HU*, LV*, LT, RO	DE*, PL, SI, UK	BG, CZ, LU, PT

Notes: For abbreviations of sectors, see Figure 1.

a

National types correspond to the European Commission (2009) typology/clusters. Bold fonts for countries in the sector typology/clusters indicate correspondence to the national typology of countries.

*

= countries on the margin to be categorized into another cluster/type.

Missing data: Steel: IE; Sugar: CY, EE, IE, LU, MT; Tanning and leather: IE, LU; Railway infrastructure: CY, MT; Sea and coastal water transport: HU, MT, PL; Hospitals: CZ; Hairdressing and other beauty treatment: MT; Telecommunication: IE. Excluded cases because of very low number of employees: Steel: CY, EE; Tanning and leather: CY, DK; Sea and coastal water transport: AT, CZ, LU.

Variety of sectors

The analysis is carried out by examining the standard deviations of the seven key industrial relations dimensions mentioned above, across both countries and sectors. As different variables are of a different nature, operationalization and have different scales, the standard deviation of each variable is normalized to the range of 0 to 1, with 0 expressing the lowest (or no) standard deviation and 1 the highest. As a consequence of normalization, the variations of all dimensions are equally weighted (thus the difference between the lowest and the highest observation for all variables is the same). The average of the normalized standard deviations of the seven variables expresses the average variation of country and sector industrial relations. The higher the score, the higher is the variation.

Figure 1 shows a combined ranking of countries and sectors according to their sectoral diversity in industrial relations systems. It can be seen that both countries and sectors differ substantially in their degree of cross-sectoral similarity of industrial relations systems.

In terms of country differences, few countries exhibit rather homogeneous industrial relations across the nine sectors: these are one very small country (Malta), the corporatist countries Austria, Denmark and Finland, and the ‘statist’ country France. In particular, these countries are marked by uniformly organization density, collective bargaining coverage and collective bargaining centralization across sectors, within each of these countries; the degree of organization fragmentation is also uniform, across all sectors, within each of these countries. In the remaining 22 countries, the differences among sectors are strong. The most varied countries are from Central and Eastern Europe: in particular, Poland and the Czech Republic. If we consider the well-established typologies of national industrial relations (European Commission, 2009), we find that just one cluster, the Nordic, displays strong cross-sectoral similarity, with the addition of France: only in these countries does it appear legitimate to speak of national models. In the others, it is better to talk of mixed and varied systems, and particularly so in the Central and Eastern European countries, which the literature has still been unable to classify and define (Bluhm, 2006; European Commission, 2009; Kohl and Platzer, 2004). The nine most heterogeneous countries are all among those that joined the EU in 2004 or 2007. Also, there is no effect of country size: we find both small and large countries at either end of the continuum.

In terms of sectors, one large manufacturing industry (steel) and those services with large providers display relatively similar industrial relations across the EU27. For instance, steel, an ‘old’ organized sector, displays a rather high organization density across the EU, including the new member states. By contrast, sectors such as civil aviation, tanning and sugar have very different industrial relations characteristics country by country. In a very ‘localized’ sector, hairdressing, union and employer density is around zero in many countries, but high in others, such as the Netherlands, Denmark and Finland. It appears that sectors frequently dominated by very few large companies, and especially with large dominating large multinationals and strong EU-level regulations (steel, telecommunications), as well as service sectors with transnational provision (transport) display more homogeneous industrial relations than local services (hairdressing) and smaller manufacturing industries (tanning, sugar). Hospitals are surprisingly homogenous across countries despite being a service sector.

The overall comparison of degrees of industrial relations similarity shows that 15 of the EU27 countries are more internally varied that any sector but hairdressing and tanning. Broadly speaking, sector variation (within countries) emerges as at least as deep as national variation (within sectors). Considering that our manufacturing sector selection concentrates on well-established, rather than more modern sectors (pharmaceuticals and rubber and plastic), across the economy sector variation is likely to be even greater, given that modern sectors tend to be shaped by new specific and internationally uniform technologies rather than national traditions. This result clearly questions the cogency of classifications of industrial relations built exclusively at the national level.

Sectoral types of industrial relations in the EU27

We proceed to test the distribution of sectors, according to the industrial relations typology proposed by the European Commission (2009), which, as noted above, detected five distinct regional groups: Nordic (organized capitalism), Central European (social partnership), Southern European (state-centred’), North-Western Europe (liberal) and Central-Eastern European (mixed). The relevance of European Commission typology is that it is at the same time empirical/descriptive (cluster analysis) and theoretical, as it is closely linked to theoretical classifications of varieties of capitalism, welfare states and employment regimes. Although only three of our indicators (union density, employer density, collective bargaining coverage) are identical to those used by European Commission, our sector-level indicators can be treated as reasonable proxies for the same dimensions. By using the k-means clustering method, we can distribute our cases (nine sectors in 27 countries) alongside the same demarcation lines (by the same cluster means) as the European Commission. This method enables us to partition the sector industrial relations scores into the clusters described by European Commission: each sector belongs to the cluster with the nearest means (calculated as k-means) of the seven industrial relations indicators. For the indicators that do not directly correspond to the European Commission’s analysis, we use as cluster references the average of the national-level characteristics of the countries that belong, according to the European Commission, to that cluster.

Table 1 shows the result of the cluster analysis and the distribution first of the EU27 countries according to the European Commission (2009), and then the distribution of 219 sector/country cases (24 cases are excluded because of the lack or extremely small size of data). It has to be kept in mind that cluster analysis does not identify ‘perfectly homogenous’ types: each cluster includes cases that are most similar (in the sense of nearest scores), not identical.

We now examine in more detail (see Table 2) the distribution of cases against country and sector clusters. The largest number (65) is found in the ‘organized corporatism’ type, which at the national level is only represented by three Nordic countries. But this result should not be overestimated: our nine sectors are not representative for the whole economy, and organized sectors are over-represented.

OPEN IN VIEWER

Table 2.

Industrial relations characteristics of national types and of sectors

Industrial relations indicators:	UD a	UF b	ED a	EF b	CBC a	CENT b	INV b
Characteristics of national types: c
Organized corporatism	74.7	1.4	58.0	0.6	86.8	2.5	2.0
Social partnership	35.4	1.0	72.7	0.4	82.8	2.5	2.5
State-centred	20.2	1.2	65.8	1.2	75.4	2.0	1.0
Liberal	33.9	0.5	47.5	0.5	35.3	0.0	0
Mixed	22.8	0.9	28.4	0.7	34.5	0	1.0
Characteristics of sectors (medians):
STEEL	59.0	1.1	80.4	0.0	99.5	2.0	2.0
SUGAR	55.0	0.7	100.0	0.0	100.0	2.0	1.0
TANNING	24.0	0.7	46.8	0.7	95.0	3.0	1.0
AIR	61.2	1.8	63.6	1.1	95.0	1.0	1.0
RAIL	72.0	1.1	94.0	0.7	100.0	3.0	1.0
SEA	80.0	1.1	88.0	0.0	75.0	3.0	2.0
HOSP	59.5	1.6	81.8	1.1	95.0	3.0	2.0
HAIR	3.0	1.1	32.0	0.9	24.0	2.0	0
TELECOM	47.9	1.1	52.0	0.7	82.5	2.0	1.0

Notes: UD = union density (%), UF = union fragmentation, ED = employers’ association density (%), EF = fragmentation of employers’ associations, CBC = collective bargaining coverage (%), CENT = centralization of collective bargaining, INV = involvement of social partners in socio-economic policy-making.

a

Industrial relations indicators and scores correspond to European Commission (2009).

b

Scores are calculated as the average of the national-level characteristics of countries that belong to the national cluster according to European Commission (2009).

Calculation of sector scores: scores are calculated as the average of countries.

c

National types correspond to the European Commission (2009) typology.

In the ‘organized corporatism’ type we find the Nordic countries in nearly all sectors. But in addition, in all sectors but hairdressing and other beauty treatment we find a large number of countries from different national traditions. This includes some new member states, especially in steel (Poland, Romania, Slovakia), where the imprint of former strong centralization is still visible. For instance, steel is quite exceptional in Poland for the achievements of national-level sectoral social dialogue. Hospitals and railways are also found most frequently in this cluster, including for instance UK hospitals, which, with their strong centralization, state ownership and high unionization and collective bargaining coverage, differ sharply from the British liberal model. The sectors in this cluster have ‘dense’ industrial relations, in the sense of involving strong and numerous actors, at many levels, with extensive levels of engagement in collective bargaining and involvement in policy-making with the state. A typical case for this cluster is the Finnish hospital sector: very high density and collective bargaining coverage, but also considerable fragmentation.

The ‘social partnership’ cluster is more heterogeneous. Of the six countries that are deemed to represent this cluster, sectoral industrial relations actually correspond to this in a systematic manner only in Austria and the Netherlands (in six out of nine sectors). Germany, by comparison, displays ‘social partnership’ industrial relations only in three sectors (sugar, civil aviation and sea transport). By contrast, we find in this cluster very frequently France (seven times), quite frequently Spain (four), and occasionally a variety of new member states. This kind of industrial relations, with high collective bargaining centralization and coverage despite relatively low union density, actually corresponds to two different situations: cases where collective bargaining coverage relies on strong employers’ organizations (as in sugar in terms of sectors, and Austria in terms of countries), and those where it relies on state intervention, as measured by the variable ‘involvement’ (as in civil aviation, railways and telecommunications in terms of sectors, and France as a country). It is therefore disputable whether the term ‘social partnership’ is actually pertinent. Possibly, the inclusion of the employee participation dimension could allow a redefinition of the profile of this cluster more precisely.

The ‘state-centred’ type is quite numerous, but we must notice that a number of cases (marked by an asterisk) are relatively distant from the average values, or centre, of the cluster. Of the Southern European countries that should belong to this cluster, only Portugal is quite consistently found in it. Unsurprisingly, the sectors that most frequently present industrial relations of this kind are those once dominated by national providers: railways, civil aviation, telecommunications.

The ‘liberal’ cluster is a rather residual one, with the smallest number of cases (29) and no clear concentration in any sector or country. Besides the UK, the most frequently represented countries are new member states such as Lithuania, Poland and Bulgaria, as the characteristics are not very far from those of the ‘mixed’, or ‘Central-Eastern European’ cluster. Yet we find here also German telecommunications and German hospitals, confirming researchers’ findings about the distinctive disorganization of these sectors in that country (Doellgast et al., 2009; Grimshaw et al., 2007).

The ‘mixed’ cluster contains a majority of cases from the new member states, with the addition of individual cases from the UK, Portugal, Ireland, Cyprus, Malta and Luxembourg. This type of industrial relations displays extreme values on nearly all dimensions, in the sense of very low organizational density and collective bargaining coverage. It is typical of tanning and leather in the new member states, where this sector, unlike in Western countries such as Italy and Portugal, is disorganized and often foreign-owned and concentrated on sub-processing. Even more, it is typical of hairdressing, where it combines all post-Soviet bloc new member states. Although defined by the European Commission as ‘mixed’, such characteristics would deserve the definition of ‘empty’, given the near-total lack of organized industrial relations.

This ‘mixed’ or ‘empty’ cluster is the only ‘sector industrial relations’ cluster that corresponds closely to the spatial distribution of the European Commission’s national typology. In Table 1, sector and national typologies are contrasted for all countries. ‘Anomalies’ are particularly evident in three sectors: steel, railways and hospitals. For instance, in steel unexpected countries such as Poland and Romania have features of ‘organized corporatism’. In railways, the same happens in Italy, Poland, Romania and Slovakia; and in hospitals also in the UK and Ireland. Spatial patterns are visible in a very localized sector such as hairdressing, but not in sectors characterized by supranational regulation (telecommunications), international labour markets (hospitals) or multinational companies (steel).

Conclusion

This article represents a critical-empirical exercise testing dominant typologies of industrial relations, by looking at the sector level. Any such exercise can be deemed to be ‘empiricist’, merely stating the infinite complexity of the real world. In fact, authors who propose typologies nearly always admit this complexity, with acknowledgments such as this: ‘obviously, as with any classification, the real world is messier than these typologies and the application to single countries is an approximation at best’ (European Commission, 2009: 50). In this sense, our description of the disconnection between sectoral industrial relations and presumed national ‘types’ is not a falsification of existent typologies: to falsify them, alternative theoretical typologies would be needed.

Yet our descriptive effort is not simply empiricist, as it opens up the possibility of theoretical reconsideration, which we can here only sketch. Such a theoretical perspective would focus on both causes and effects: the problem of why some sectors and countries display certain characteristics, and the implications of these characteristics.

There are strong theoretical reasons to argue that industrial relations vary most deeply by sector: from the theory of the firm, through industrial economics or labour market segmentation theory, to labour process theory. On one side, the technology and product market factors that affect industrial organization, and thereby the industrial relations, of sectors appear more significant than when first analysed by Hollingsworth and Streeck (1994). By contrast, the national variation factors they highlighted – rules of behaviour, and political and cultural resources – no longer appear as generally dominant as then. In particular, by deconstructing the geographical clusters defined by the European Commission, our study suggests that cultures are not as influential as was previously thought.

Sector differentiations do not replace national ones: our analysis shows that both levels are equally important, and that indeed in a small group of countries (the Nordics and Austria) national models are still very visible. But to understand industrial relations both levels are important: we cannot derive the kind of industrial relations that affect a company, or a group of employees, simply by the country in which they are located; we must also know in what sector they operate. Some countries that are often considered as ‘models’ of industrial relations types (the UK for voluntarism, Germany for social partnership) emerge as particularly differentiated by sector, to the point where the cogency of the concept of ‘model’ has to be questioned. It also emerges that the new post-Soviet bloc member states, while they do present extremely disorganized industrial relations in many sectors, also contain important exceptions, especially in steel and railways. Conversely, not all sectors are equally homogeneous: the more internationalized ones share more common features than the localized ones.

In terms of implications, a sector focus may provide a better understanding of why transnational coordination and regulations develop more in some sectors than others (Anner et al., 2006). In particular, it is striking that, although all nine sectors we selected have the same formal transnational institutions (a European Sector Social Dialogue Committee), the actual ‘productivity’ of these institutions in terms of agreements and process-oriented texts appears to be stronger in more homogeneous and more organized sectors (rail and sea transport).

So far, comparative industrial relations research has predominantly focused on the national level; but our analysis highlights that the sector requires a similar focus and explanatory effort. For such a focus, however, more systematic cross-country sector data will need to be collected, to make up for decades of developmental delay in comparison to ‘methodological nationalism’.