How are coordinated market economies coordinated? evidence from Sweden

Abstract

Central actors in coordinated economies have a dense network of associations for coordinating their actions. However, we lack knowledge of how such relationships are constituted and how the mechanisms of such coordination work. In the present article, we concentrate on the non-market mechanisms of power and trust. We want to know: Are all actors in the industrial relations system connected to each other or are they divided into contending but coherent groups? And in that case, do certain actors playing the role of ‘brokers’ connect these groups? A social network analysis on survey data from Sweden shows that the state and the peak-organisations still hold powerful positions. Simultaneously, when concentrating on reciprocal ties, the organisational borders are in particular bridged over by public sector organisations. However, actors with power are not highly trusted. Instead, other state actors – especially the Labour Court–play an important role as a broker that all parties trust.

Notes

The authors would like to thank the participants of the ECPR workshop ‘Changing Industrial Relations in Contemporary Capitalism’ for useful comments. Financial support by the National Institute for Working Life (ALI) and the Swedish Council for Working Life and Social Research (FAS) is also gratefully acknowledged.

1. In earlier articles (Öberg and Svensson 2002; Svensson and Öberg 2005) we used two additional measures of power: estimated influence and estimated importance of having other actors as allies. An actor ranked high as an associate has the ability to facilitate the creation of a needed coalition that would not otherwise be formed.

2. The share of individual respondents is just below 50 per cent (369 individuals). Considering the demanding type of questionnaire, and the fact that this is an elite survey, the response rate was quite satisfactory (see Öberg and Svensson 2002, Table 1 and footnote 4). The missing actor is the Swedish Food Workers' Union (Livsmedelsarbetarförbundet). In the process of creating a sample some demarcation problems appeared. The selection was based upon size in combination with historical importance. This choice excluded one quite large union (slightly more than 60,000 members) organising managerial staff–the Swedish Association for Managerial and Professional Staff (Ledarna). This would have been problematic if this group was in fact central in the labour market network. In order to control for this type of problem in the survey, we asked respondents to mention actors of some importance other than those explicitly given in the questionnaire, with whom they have contacts. Only one actor explicitly mentioned this particular union. Other actors are also mentioned, but none appeared frequently. Our sample includes the important actors and does not exclude anyone of real relevance.

3. The network analysis was carried out using Ucinet, a program package developed by Martin Everett and Stephen Borgatti (Copyright (c) 1999–2000 Analytic Technologies, Inc). We use ‘indegree’ as a measure of an actor's centrality. This is a measure of directed links towards a certain actor. A normalised measure means that the value of degree is divided by the maximum possible degree expressed as a percentage. For further discussion of actor centrality and how to characterise networks, see, for example, Wasserman and Faust (1998) and Scott (2000). The use of network analysis puts heavy demands on data quality. One collective actor is missing from the data. However, this is an actor of minor importance with regard to the research questions and of little consequence for the interpretation. The picture regarding individual dropouts or individual mistakes is mixed. All large organisations, especially the peak organisations, and important authorities are quite well represented. Generally, more than 50 per cent of the individual representatives answered. The lowest response rates are to be found among big companies and political parties (see Öberg and Svensson 2002: Table 1). However, there are some specific actors with low or very low response rates that could have had an influence on the results. In cases where a collective actor is represented by only one or a few positions, individual differences within the organisations can have a significant effect (for example, the Left Party with one out of six, or the Industrial Workers Union where one out of three responded). This forces us to take a cautious approach. Apart from being non-symmetric, the network is also valued. As the calculations involve dichotomisation of data, we can make use of this property in order to find a valid division. However, it should be pointed out that different dichotomisations lead to networks of different kinds, or at least of different complexity. On the basis of the variation of answers within the different questions (Öberg and Svensson 2002) we maintain that the dichotomisation also reflects an actual variation and underestimates differences.

4. We elaborate more on the relation between power and trust in Svensson and Öberg (2005).

5. Density in the network of honest intent is .48, keeping one's word .62 and listening to arguments .31 (Svensson and Öberg 2005).