THE SWEDISH INDUSTRIAL AGREEMENT

Henry Ohlsson, Department of Economics, University of Gothenburg

Abstract[14]

The Swedish collective bargaining system has changed a lot the last two decades. The changes do not, however, mean that the Swedish system has converged to those in other European countries. One distinct new feature is the Agreement on Industrial Development and Wage Formation between the social partners in the manufacturing industry. The initial reasons for these changes are the employers wish to decentralize and to find more flexible pay systems together with increased skill requirements. This in turn, may depend on capital–skill complementarity in technical progress and increased international trade and specialization. The employers did not get exactly what they wanted. The actual outcome is very much dependent on the high union density in Sweden. There is still conflict of interest but the rules of the game have been agreed on. There are 60 different bargaining areas within the framework of the Industrial Agreement. The bargains have finished in time. They have delivered high real wage increases for the employees and competitive labor costs for the employers but also increased wage dispersion.

Keywords: collective bargaining, wage growth, wage dispersion, structural change

JEL classification: E24, J31, J50

Introduction

There are considerable cross–country differences in industrial relations. The systems for collective bargaining are almost genuine national characteristics. The starting points for this paper are three questions concerning collective bargaining in Sweden:

To what extent and in what sense does the Swedish system of collective bargaining become more similar/dissimilar to these in other industrialized countries?

The Swedish collective bargaining system has changed a lot the last two decades. The changes do not, however, mean that the Swedish system has converged to those in other European countries. The Swedish system is still different but in some new ways. One distinct new feature is the Agreement on Industrial Development and Wage Formation between the social partners in the manufacturing industry from 1997. Although this agreement only affects ca 10 percent of those employed directly, it has had a strong impact on collective bargaining on the whole Swedish labor market. The doctrine of freedom from government intervention in the labor market stays intact.

How can we account for these tendencies of change?

The initial reasons for these changes are the employers wish to decentralize and to find more flexible pay systems. Increased skill requirements are also important for this process. The increased skill requirements may depend on capital–skill complementarity in technical progress and increased international trade and specialization. The employers did not get exactly what they wanted. The actual outcome is very much dependent on the high union density.

To what extent do these tendencies of change affect the distribution of power, income, wealth, and political leverage?

There is still conflict of interest but the rules of the game have been agreed on. There are 60 different bargaining areas within the framework of the Industrial Agreement. The bargains have finished in time. They have delivered high real wage increases for the employees and competitive labor costs for the employers but also increased wage dispersion.

The paper is organized as follows: The next section gives a background by very briefly presenting some main facts about collective bargaining in Sweden. The contents of the Industrial Agreement are presented in the following section. Then follows a section that reports the outcome for wages since the agreement started to affect bargaining.

Background

For many decades the Basic Agreement was the cornerstone in the Swedish centralized collective bargaining system. The Basic Agreement was made in 1938 in Saltsjöbaden. It was a central agreement between SAF (The Swedish Employers’ Confederation) and LO (The Confederation of Trade Unions).[15]

Centralized bargaining between SAF and LO did for decades set the standards for the rest of the Swedish labor market. LO pursued a solidaristic wage policy that resulted in wage compression. PTK (The Negotiation Cartel for Salaried Employees in the Private Business Sector) joined forces with LO.

There was a big labor market conflict in 1980. This marked the beginning of the end of the centralized collective bargaining system. As is clear from Figure 1, the 1980 conflict was much more serious than any other conflict since 1965. More than 90 percent of the workdays lost were because of lockouts. Before 1980, workdays were lost mostly because of legal and illegal strikes. To put things in perspective, the average number of workdays lost 1980 corresponded to ca 6 days per worker affected.

(Illustrasjon mangler)

The first sector specific agreement came in 1983. The agreement was between VI (The Association of Swedish Engineering Industries) and Metall (The Swedish Metal Workers’ Union). At this time, VI worried more about the processes and criteria for pay determination than industrial peace. There were many pay disputes during the 1980s. As is clear from Figure 1, a considerable number of workdays were lost during the second half of the 1980s.

The Swedish economy became more and more overheated in the end of the 1980s and the beginning of the 1990s. In 1991, the Rehnberg group pressured the social partners to an agreement. In a sense, this bargaining was extremely centralized. But it was far too late to avoid a severe economic crisis. Swedish unemployment exploded in 1992. Economic growth was negative 1991, 1992 and 1993. Being a small open economy with a high export share, international competitiveness is crucial for Sweden. Figure 2 reports a measure of this for the last three decades. Periods of excessive nominal wage increases (for example mid 1970s and end 1980s) were followed by devaluations (end 1970s, 1982, and 1992–93).

(Illustrasjon mangler)

There was new environment for collective bargaining in the beginning of the 1990s apart for the effects of the crisis. Sweden had experienced a trend of higher exposure to international competition during several decades. The country had a history of high nominal wage increases, high inflation, devaluations, and no real wage increases.

There was no longer any wage leadership from SAF–LO. SAF stopped to be a bargaining party in 1990. Since then collective agreements are sector–specific. LO’s relative strength compared to other unions also decreased. Two important reasons for this the increasing share of white–collar workers in the private sector and the fact that public employment corresponded to a third of total employment.

The bargaining in 1995 illustrates that decentralization can create coordination problems. VI and Metall had agreed in principle, only some minor details remained. The exhausted bargainers took the weekend off. On Saturday the social partners in the paper and pulp industry agreed on a contract giving nominal annual wage increases of 6 percent. Metall forced VI to start all over again.

But the last five years have meant decreasing relative unit labor costs for Sweden, see Figure 2. This has meant increased international competitiveness and increased employment within the manufacturing industry.

Figure 3 shows that nominal wage growth has been slow during the last years both overall and in manufacturing.

(Illustrasjon mangler)

But because the very low inflation rates during the last years, real wages have increased by on average 2 ½ percent per year. Figure 4 shows the development the last years. Real wages have not increased so much since the 1960s.

(Illustrasjon mangler)

Sweden has gone from a severe economic crisis to a very promising economic development. What happened?

The industrial agreement

There are many reasons why the economic development in Sweden has been good. Some have to do with what has happened internationally. There are also domestic reasons, a successful economic policy is one. But changes in wage formation have also played an important role.

The trade unions within the manufacturing industry took the initiative to the discussions on how the social partners should interact during the summer 1996. 12 employers’ organizations and 8 trade unions signed the agreement of industrial development and wage formation in March 1997. The agreement has two parts, a negotiating agreement and a part where the social partners express common views on the conditions for industrial enterprise.

The agreement does not remove conflict of interest but it sets rules of the game. It is a procedural framework for industry level collective bargaining within manufacturing. These procedures are used when doing collective bargaining within 60 bargaining areas in the manufacturing industry. The engineering industry area (VI and Metall/SIF/CF) affects most employees.[16]

The agreement introduces three new bodies. The Industry Committee consists of members from the social partners. The committee supervises the bargaining in the different areas. It can force parties to temporarily halt their industrial actions.

There are 10 appointed impartial chairpersons. The chairs support the negotiations. The agreement stipulates a timetable for the negotiations. The intention is that negotiations should be concluded before the previous contract expires. Negotiations should start at least three months before the expiry date. All demands shall be put forward when the negotiations start, no new demands are allowed during the negotiations. If there is no agreement one month before the expiry date, the impartial chair shall intervene on her own initiative. Should the parties wish to extend the current agreement, this requires the consent of the impartial chair. A party who intents to give notice of industrial action is obliged to inform the impartial chair. The chair can put forward own proposals, and she can delay industrial action for 14 days. These parts of the agreement replace the Mediation Law.

In addition the agreement says that if a union starts an industrial action there will be no retroactive compensation. The partners are tied to comply with agreements made. Disputes can be settled by arbitration rather than in the Labor Court. If there is illegal industrial action, local negotiations are terminated immediately.

The Industry Committee and the impartial chairperson can request reports from the Economic Council of Industry. The council consists of four independent economists.[17]

Similar agreements have been made between the social partners in the public sector and between white–collar worker and employers in retail trade. The transport industry, much of the retail trade, and the construction industry are not covered by agreements.

Since 2000 there is a National Mediation Office responsible for mediating in labor disputes and promoting an efficient wage formation process.[18]

The outcome

The first agreements within the framework of the Industrial Agreement came in 1998. There were three year contracts in most bargaining areas. In 2001, bargaining according to the negotiating agreement was made in 59 bargaining areas, 565,000 employees were affected. In some areas bargaining is completely parallel for white–collar and blue–collar workers. The smallest area affected 300 employees, the biggest 280,000.

Almost all bargains were completed within a week after the expiry date. Completion in time appears to be very much appreciated among the union members. This suggests that at least some employees are liquidity constrained.

The Economic Council of Swedish Industry (2002) has studied wages in industry. We have used data from the Confederation of Swedish Enterprise 1997 – 2001. The data cover businesses with more than 10 employees. We can study ca 400,000 employed each year within the framework of the Industrial Agreement. The measurement period is September – October. We have not adjusted the data for changes in working time or for fringe benefits.

We study wage increases and wage dispersion. The measured wage increases may be affected by structural changes. This may be because the people employed changes – personal turnover. It can also be because the people employed acquire new knowledge – personal development.

The composition of the labor force in industry has changed during the studied period. The share of women was 25.85 in 1997 and 25.88 in 2001. Women earned 13.4 % less than men did during the period. The share of the category white–collar workers has increased a lot during the period. It was 37.77 in 1997 and 43.60 in 2001. White–collar workers earn 27.3 % more than blue–collar workers do. The average age increased from 40.33 years to 41.01 years.

Figure 5 shows the age distribution for industry employed in 1997 and 2001. In 1997 there was a pronounced peak at ages around 30. This peak has four years later moved and become even more pronounced around age 35. In addition, there was a less pronounced peak slightly above age 50 in 1997. This peak is now less pronounced. In general, there has been a decrease for the age groups age 20–30 and around age 50 while the groups aged 30–45 and 55–60 have increased.

The wages in industry have increased by on average 3.8 percent per year, see table 1.[19] The outcome is slightly above the agreements that gave on average 2.8 percent per year for the three year period 1998–2000 and 2.8 percent also 2001. Both types of structural changes may however, affect the actual wage increases.

(Illustrasjon mangler)

We can remove the effects of hirings and separations by instead calculating the average wage increase for individuals. In this case, we therefore compare wage levels different years for the same persons. For these the wage level 2001 was on average 17.5 percent higher than in 1997. This corresponds to an annual wage increase of 4.1 percent, slightly higher than the increase computed when using all observations.

The average age has, however, stayed relatively constant while the individuals who are included when we make the calculations for individuals have become four years older during the period. When employees age, they gain experience. This increased experience has been valued to an average annual wage increase of ca 0.3 percent (3.8 percent compared to 3.6 percent) if we base our calculations on all observations. If we instead calculate on an individual basis, it is ca 0.4 percent per year (4.1 percent compared to 3.7 percent).

Those who have started to work in the manufacturing industry have received lower wages, and those who have left have had higher wages, than those who have stayed on. The conclusion is that labor turnover has decreased annual wage growth by ca 0.1 percent (3.6 percent compared to 3.7 percent).

The increasing share of the category white–collar workers is almost exclusively the result of turnover. Blue–collar workers have left the manufacturing industry while white–collar workers have joined. This increasing share can explain an annual increase of the wage level by on average ca 0.5 percent if we use all observations (3.3 percent compared to 3.8 percent). If we take the age, gender, category, and county into account the remaining wage increases are 3.2 percent per year on average. If structural changes are measured in this, they, in other words, correspond to wage increases of 0.6 percent per year during the studied period.

We have also computed the median wage. If the median wage deviates from the average wage this is a sign that the wage distribution is skewed. This is usually the case when studying wage distributions. The median wage does not increase as fast as the average wage during the studied period. The median wage has increased by on average 3.3 percent per year during the period. In 1997 the median wage was 94.5 percent of the average wage. Four years later, this number had decreased to 92.7 percent. This suggests that the wage distribution has become more skewed.

Figure 6 shows how the distribution of unadjusted wages 2001 differs from the corresponding distribution 1997. As is clear from the figure, wage dispersion has increased. The share with wages in the interval from 25 percent under the mean to 50 percent above the mean has decreased slightly while the shares below and above this interval has increased. The distribution has also become more skewed in the sense that the peak of the distribution has moved further away from the mean.

(Illustrasjon mangler)

Concluding remarks

This paper discusses three questions concerning collective bargaining in Sweden. To what extent and in what sense does the Swedish system of collective bargaining become more similar/dissimilar to these in other industrialized countries? The Swedish collective bargaining system has changed a lot the last two decades. The changes do not, however, mean that the Swedish system has converged to those in other European countries. The Swedish system is still different but in some new ways. One distinct new feature is the Agreement on Industrial Development and Wage Formation between the social partners in the manufacturing industry from 1997. Although this agreement only affects ca 10 percent of those employed directly, it has had a strong impact on collective bargaining on the whole Swedish labor market. The doctrine of freedom from government intervention in the labor market stays intact.

How can we account for these tendencies of change? The initial reasons for these changes are the employers wish to decentralize and to find more flexible pay systems. Increased skill requirements are also important for this process. The increased skill requirements may depend on capital–skill complementarity in technical progress and increased international trade and specialization. The employers did not get exactly what they wanted. The actual outcome is very much dependent on the high union density in Sweden.

To what extent do these tendencies of change affect the distribution of power, income, wealth, and political leverage? There is still conflict of interest but the rules of the game have been agreed on. There are 60 different bargaining areas within the framework of the Industrial Agreement. The bargains have finished in time. They have delivered high real wage increases for the employees and competitive labor costs for the employers but also increased wage dispersion.

References

Albåge, L.–G. and R. Larson (1998): Industriavtalet i 1998 års förhandlingar —händelseförlopp och resultat, rapport till Industrikommittén av de opartiska ordförandena, september. Economic Council of Swedish Industry (2002): Wage growth in Swedish industry 1998 – 2001, Report to the Industry committee.

Elvander, N. (2000): The industrial agreement: An analysis of its ideas and performance, ALMEGAS förlag, Sandviken.

Elvander, N. (2002): The labour market regimes in the Nordic countries: A comparative analysis, Scandinavian Political Studies, 25, 117–137.

OpO (2002): Industriavtalet i 2001 års avtalsrörelse — händelseförlopp och resultat, rapport till Industrikommittén av de opartiska ordförandena inom industrin, februari.

Sandgren, A. (2002): Verkstadsindustrin och industriavtalet, VI, Stockholm.

Sheldon, P. and L. Thornthwaite (1999): “Swedish engineering employers: The search for industrial peace in the absence of centralised collective bargaining”, Industrial Relations Journal, 30, 514–532.

[14] A previous version of this paper was presented at the conference “National regimes of collective bargaining in transformation: Nordic trends in a comparative perspective” arranged 5–6 September 2002 in Oslo by Fafo – Institute for Labour and Social Research and the Norwegian Power and Democracy Study. Some of the work with the paper was done when I enjoyed the hospitality of ERMES, Université Panthéon–Assas, Paris II.[]

15 The legal framework was set by the 1906 mediation law, the 1928 collective agreements law, and the1928 labor court law.[]

16 Metall organizes the blue–collar workers, SIF (The Swedish Union of Clerical and Technical Employees in Industry) the white–collar workers, and CF (The Swedish Association of Graduate Engineers) the engineers.[]

17 I have been a member of the Economic Council since the start 1997.[]

18 The National Mediation Office took over the tasks of the National Conciliator’s Office.[]

19 The outcome for particular years should be carefully interpreted. The annual measurement period is short. This makes the results for single years sensitive to the wage agreements’ timing of wage increases.