The Origins of the Swedish Wage Bargaining Model

Abstract

This paper revisits the development of the canonical Swedish wage bargaining model, from the 1930s to the 1950s. The question at the core of the debate is: how did Sweden achieve “good” wage bargaining institutions -- good, in the sense of facilitating investment, employment, and controlled inflation? The conventional account focuses on the actions of employers and trade unions in export industry, and a cross-class alliance between the two. This paper questions this account. The paper builds on archival sources in the Swedish Labour Movement's Archives and Library in Stockholm: the minutes from the main trade union confederation's yearly wage policy discussions, in preparation for bargaining rounds. In total, some 1,500 pages of wage policy discussion. I find that the export sector cross-class alliance played a very small role, and that macro-corporatist concerns, that the labour movement had to take responsibility of all of society and pursue a planned wage policy, was much more important. This has theoretical implications for the analysis of wage bargaining institutions in general and the Swedish model in particular.

Introduction

In the twentieth century, Sweden had one of the most successful labor movements in the world. A reformist trade union movement and Social Democratic party in alliance reached a high degree of influence, building a generous welfare state and an unusually equal society.¹ Figuring out the “success concept” of the movement has attracted much interest among researchers. The capacity for compromise, between various societal interests and actors, tends to play a central role. The historian Geoff Eley writes in his overview of the history of the Left, that: “Business acknowledged the legitimacy of social democratic government, with its commitment to high wages, social welfare, and full employment, while socialists conceded the sanctity of private property, guaranteed private control of capital markets, and restrained union militancy in the interests of social peace.”² The economists Moene and Wallerstein formulate it more generally: “Social democracy is, in essence, a series of political and economic compromises.”³ The compromises can be between business and labor, and between economic growth and equality. Wage bargaining institutions are, along with the welfare state and economic policy, at the core of this Social Democratic institution-building.

This paper investigates precisely the evolution of wage bargaining institutions in Sweden from the 1930s to the late 1950s. The arrangement of wage bargaining is crucial in any society, in its effects on income distribution and inflation, and its implications for the strategies of companies. The conventional understanding of the Swedish success story is that one particular wage policy was crucial: export-sector led wage restraint was key to “how Social Democracy worked,” and was at the center of the postwar Swedish model.⁴ Trade unions and employers held back wages to facilitate investment and growth in high-productivity sectors, and this wage restraint was combined with equalization between different sorts of workers. According to the received wisdom, in an analysis proposed especially by the political scientist Peter Swenson, this wage restraint was driven by an outsize influence of the export sectors, and a class alliance between workers and employers in those sectors, such as engineering. The macroeconomic logic at the heart of this argument is that export-sector employers cannot compensate for high wage increases by pushing the costs onto consumers and thereby increasing inflation—they are assumed to be price-takers in the international markets—and therefore their wage leadership prevents a wage-inflation-spiral.⁵ Thus, export-sector led wage policy, created by an alliance between unions and employers in the export sectors, was, in the conventional account of the success of Swedish Social democracy, a crucial compromise.⁶

The contribution of this paper is to provide a new analysis and explanation of the Swedish postwar wage bargaining system. There are two reasons why it is important to revisit the development of Swedish wage bargaining post-1938. One is the empirical: The centrality of wage bargaining in the analysis of Swedish Social Democracy and the postwar success story makes it important to understand. The second is theoretical. The impact of Swenson's study is closely related to a theoretical shift in comparative political economy. “Swenson upended the conventional wisdom by revealing that specific segments of business were prime movers in the push for centralization,” say the leading institutionalists Kathleen Thelen and James Mahoney, and thereby inspired the shift of focus to the employers, and away from the class conflict focus of the power resources school.⁷ As Svensson says in his analysis of the 1938 Saltsjöbaden Agreement, the conflict between internationally competing industry and the sheltered sector is “primary to the opposition between labour and capital.”⁸ The present paper shows that, at least in case of the centralization of Swedish wage bargaining, the focus on the employer side as the driving force, and the focus on cross-class coalitions, has been overstated in the new consensus.

The analysis builds on the central trade union wage bargaining papers from 1939 to 1959. They show that industry wage leadership was non-existent in the 1930s, and that coherent, coordinated trade union wage policy only arose with the centralization of wage bargaining in the 1950s, which in the literature has been considered as a discretely different event. I show that the responsible wage policy, which in the previous literature has been attributed to a cross-class alliance of export sector employers, rather should be understood as an outcome of the postwar integration of wage policy into a planned economic policy. In the concluding part, I discuss the theoretical implications of re-interpreting the Swedish model.

The evolution of wage bargaining in Sweden: previous literature

The key question in the debate on Social Democracy and wage leadership—cf. the references to Eley or Moene and Wallerstein above—is: How are wages restrained to facilitate investment?⁹ Trade unions are assumed to pursue their own interests, and the risk is then that they push up wages too much that profits are hurt, and then too is investment. One way of achieving wage restraint is to simply weaken the unions so that they cannot raise wages—but that is not compatible with the Social Democratic model. Instead, Social Democratic wage restraint—or more generally, responsible wage policy—is assumed to have come from one of two mechanisms. Either a centralized wage setting where those encompassing organizations who set wages—centralized unions, centralized employers—are also the ones who would suffer (higher unemployment or inflation) if wages are set too high. Since costs cannot be shuffled to outside groups, unions in this centralized model behave responsibly.¹⁰ The other way is not to centralize but to coordinate wage bargaining, under the leadership of the export sector. Since export-sector employers cannot increase prices to compensate for higher wages, without losing market shares and thereby causing unemployment, unions in this sector are more restrained than those in other sectors. Competing wage claims may be inflationary and cause wage-price spirals, but if all claims are subordinated via coordination to those of export industry, then the problem is solved.¹¹

The received wisdom on the evolution of the Swedish wage bargaining system is built on such logic.¹² Here, a responsible wage bargaining system evolved in the 1930s, as result of a cross-class compromise of employers and unions in export-oriented industry. The leading scholar of this interpretation is Peter Swenson, but the analysis is shared by several other scholars. In Swenson's analyses of the Swedish labor market in the 1930s, the key contradiction in the trade union movement was between export sector unions, who wanted peaceful labor relations and wage restraint to facilitate competition, and militant home-market sector unions (especially in construction), who, unconstrained by concerns for international competitiveness, frequently went on large strikes for higher wages. The 1938 Saltsjöbaden Agreement between the unions (LO) and the employers (SAF) is famous for initiating a period of labor peace and constructive class compromise.¹³ In Swenson's analysis, this was deep down a cross-class alliance between export-sector unions and employers, with the restrictions on conflicts in the agreement serving to stifle the construction unions. Swenson argues, building on Åmark, that “both sides agreed that export industry should play the role of the ‘wage leader,’ while building wages should be held back.”¹⁴ Swenson's analysis has become the consensus position in research on Swedish wage bargaining history, as shown by the anthology published for the seventieth anniversary of the 1938 Saltsjöbaden Agreement, in which Svensson and Fregert elaborate upon Swenson's analysis of wage leadership as created in 1938.¹⁵ In this narrative, post-1938 establishment of export-sector wage leadership is key to understanding the subsequent Social Democratic success story: A cross-class alliance was at the heart of it.

However, neither Swenson nor any other scholar has shown any “smoking gun” when it comes to the construction of export industry wage leadership. Swenson does show that there were expressions of discontent from the metal workers with the wages awarded to construction workers, but this does not mean that such discontent was official LO policy.¹⁶ Furthermore Swenson shows that in the 1920s and especially the 1930s, there was a discussion on both the employers’ side and the union side about the wage relationship between the home market sectors and the export sectors.¹⁷ Swenson also convincingly demonstrates that the well-known strength of the Swedish trade union movement does not imply that employers have been weak: on the contrary, they have been extraordinarily well-organized, and have forced unions’ hands several times. Specifically, Swenson has shown that, in the 1930s, SAF used the cross-sector weapon of lockouts to nudge the LO into a centralizing direction. He asserts that lockouts “gave organized capital in Sweden the ability to hammer unions into a shape that made them useful as partners in centralized regulation of labor markets.”¹⁸ This gives all the agency to the employers and none to the unions. But Swenson does not show that wage negotiations in the 1930s or after Saltsjöbaden were coordinated with the establishment of wage leadership by the export industries. He shows only that after the construction conflict in 1933–1934 there was a one-off reduction in the home market industries’ wage advantage. In other words, the creation of the wage leadership system—when, how and by whom—is still an open question.

Empirical approach

Which materials should be used to trace the evolution of wage bargaining institutions in Sweden after the 1938 Saltsjöbaden Agreement? In the beginning of this period, collective agreements typically lasted a year.¹⁹ The agreements ended in early spring, and negotiations were held in the autumn and winter. The LO's discussion of the upcoming bargaining round was at the autumn meeting of the Representation (Representantskap). There, the LO stated their aims in the upcoming bargaining round, and gave a bargaining commission the right to bargain with the employers. The Representation was a twice-yearly conference that was the highest decision-making board between the congresses (which were held every fifth year), and gathered representatives of all the LO's member unions. The representatives, who took part in the meetings, were typically high representatives from the member unions, but could also be rank-and-file members. The minutes of the meeting were verbatim records of the discussions, which make them very useful for historical research. The discussions were expressly secret and each meeting began with an agreement that the discussions must not leak out of the meeting room. This makes the minutes especially reliable. They were not doctored with a critical audience in mind, but were as forthright as possible, so they could work as internal evidence of the policies that the LO had decided upon.

Herein, I use the Representation minutes, in the LO archive at the Labour Movement's Archive and Library in Huddinge outside of Stockholm.²⁰ If wages were coordinated in the form of wage leadership by the export industry, we will find this in the aims and instructions formulated at the Representation, as well as the evaluations of the previous wage rounds. This source should be especially good for picking up any internal conflict in the LO, such as the sectoral differences discussed in the previous literature.

I have studied the Representation minutes from the year after the Saltsjöbaden Agreement, 1939 to 1959. The study begins in 1939, as Swenson and Åmark have claimed that the 1938 Saltsjöbaden Agreement was central for the LO's wage policy.²¹ It ends in 1959 when wage coordination was well under way. This end point is also motivated by the fact that the Representation minutes become much less informative after 1960. From 1960 on, the minutes are no longer verbatim. However, since the classical Swedish model investigated here had already been established,²² the changing character of the minutes is not a problem for the present study, although the Representation minutes are not sufficient for studying wage policy in the 1960s. The minutes will be referenced by month, year, and page.

Studying the evolution of wage bargaining based on primary materials from only one of the labor market parties is in its nature one-sided. Nevertheless, that is how the present article proceeds. This entails some limitations. But on the core issue of establishing how the wage leadership system was established in Sweden, the materials used here are sufficient. The Representation provided the mandate for the negotiators of the LO every year, and the negotiators reported back to the Representation. Thus, any type of bargaining coordination or centralization implemented will be visible in these archival materials. However, while we do get the reports of negotiators and the LO leadership when they have discussed or negotiated with employers, it is difficult to capture a broader building of cross-class trust. I will get back to this issue in the conclusion.

The transformation of wage bargaining: The rise of the Swedish Model

The road to a coordinated wage policy was much slower and indirect than the previous literature has suggested. Already the wage policy discussions of the LO during the years immediately after the 1938 Saltsjöbaden Agreement show, contrary to the conventional understanding, no trace of export-sector wage leadership.²³ Wage bargaining was predominantly local and the LO had neither the formal nor the informal capacity to coordinate wage policy. In 1938, only sixteen of the thirty-two LO unions had any national collective agreements while the other sixteen stuck to the traditional model of local agreements. Only 37 percent of LO members were covered by a national agreement. Of course with local agreements, national coordination with the other unions is more or less impossible. Throughout the 1920s and 1930s, a critical minority within the LO had grown, arguing for a more centralized, more ambitious, and more “solidaristic” or “socialist” wage policy. At this point, however, these ambitions foundered on a leadership pessimistic about the possibility that wage policy could change the wage structure. In practice, wage policy in the 1930s followed the principle of “everyone get what you can.”²⁴ With the exception for years of wage freezes and extreme restraint during World War II and just after, this was true for the 1940s, too. In practice, bargaining oscillated between complete decentralization and central inflation indexing; there was no proactive trade union wage policy at all. The many statements of intention—for a more ambitious, more coordinated wage policy—from various actors in the LO and SAF was at this point only arguments, not descriptions of what actually occurred. Ullenhag sums up LO's solidaristic wage policy debates in the 1930s and 1940s as “active debating—little concretization.” The same is true for any principled wage policy.²⁵

The LO simply did not have the power or capacity or will to pursue a systematic wage policy. Table 1 shows the growing importance of the trade union confederation, the LO, for wage policy over time. At the beginning of the period, the joint wage policy discussions were short and most of the member unions did not bother to voice an opinion. In the late 1940s and early 1950s, the salience of a joint wage policy grew.

Table 1. The quantitative extent of the wage policy discussions at the representation

Note. Source: own calculations from the Representation minutes in the LO archive, ARBARK 2964/A/2/A.

The bumpy road to a coherent and coordinated wage policy

The wage leadership model can be broken down into two constituent parts. One, the consideration of sectoral differences, and the allocation of a special role in wage bargaining or one sector (exporting industry). Two, the coordination of wage policy, and the implementation of sector-specific policies. Both aspects were amply discussed in trade union movement in the 1940s.

The topic of competition between unions and the need for more coordination on that point was a recurring discussion. In 1941, Oscar Westerlund from the Metal Workers, in a dissenting point of view, supported centralization of bargaining, saying that the last couple of years had been chaotic, and that more planning was needed. He clarified his point:

A bargaining round for one union has repercussions also for other unions. We remember that the public servants (Sw. statstjänarna) went ahead and reached an agreement which then became a precedent for the other groups. At any time it might happen, for example, that the Metal Workers went first.²⁶

Westerlund had a point, but the prevailing preference of many unions for a mind-your-own-business-kind of wage policy stopped any coordination such as that he preferred. It is also interesting to see that the group competition that he alludes to is between public servants and workers, not between export-sector and home-sector workers as we might expect from the previous literature. Again and again, the multidimensional distributional conflicts of wage policy crop up in the debates. In 1949, Engstrand from the Wood Industry Workers (Träindustriarbetareförbundet) pointed out that the government, in this time of massive intervention in agricultural markets, in practice had given a pay raise to the farmers with the last added subsidy. Thereby, the government had admitted exceptions to its anti-inflationary policy, and so it should also be possible to pursue wage increases for the low paid of the LO.²⁷ Again, in relation to the literature, it is interesting that the construction workers are not the criticized groups; rather, LO wage coordination is seen in a much wider context, including government support to farmers.²⁸

In 1948, similar complaints of group competition were made, but with a new dimension: they were then explicitly put in terms of exporting sectors versus home-market sectors. This is the first time in my material where the distinction between export and home-market industries is used in a discussion of future wage policy. Henry Hansson from the Factory Workers (Grov-och fabriksarbetareförbundet) complained about the lack of coordination in the last bargaining round (1946–1947) where some home-market unions such as the Typographers in the end got away with good wage increases, since they were not threatened by foreign competition. Hansson suggested that “the leadership should before the next meeting put forward a proposal which makes it possible to follow a unitary line, to avoid pure anarchy.” Representatives of the export sectors argued for a coordinated wage policy to make their sectors more attractive for workers in terms of the wages offered.²⁹

Again and again, the discontent with the laissez faire of LO wage policy was formulated. The recurring complaints indicate the stagnation of actual policy development. By 1949, a prolongation agreement of the previous year had provided wage increases across the board, but wage drift, i.e., workplace-level pay increases outside of what had been formally negotiated, was unequally distributed. Gunnar Lantz from the amalgamated United Unions³⁰ was unhappy with this issue and argued for a more proactive union line in the 1949–1950 bargaining round:

Is not there another line? It's said that either we have full prolongation or we let the bunch loose, and then we know where it's going. It's sometimes said that the Swedish union movement is stable and strong. But is not it a declaration of weakness to argue that there is nothing between absolute restraint, prolongation, and each union fighting for itself on its own front?³¹

To understand why there was so little progress in terms of regulation of inter-sector differences, we must move to the second issue outlined above: the practical coordination of wage policy. The first discussant in my material to raise the connection between the two aspects was Oskar Karlén from the Wood Industry Workers’ union in 1947. He then made a wide-ranging and visionary statement:

we should be able to agree to apply the method, that the low-paid come first in the bargaining round. We have changed the LO's statutes to make such a wage policy possible, but apparently we are not yet mature enough to apply the statutes in this respect. /…/ we run the risk that things will work out as before: each union takes what it can possible, without any regard for the common concern to restrain prices. /…/ I would ask the Secretariat to look more closely on the proposal and re-formulate it so that it becomes possible for the Secretariat to keep the unions’ bargaining rounds together and steer them.³²

However, Karlén's proposal was dismissed, with arguments such as that the LO Secretariat would not have the necessary expertise to coordinate all the collective agreements from the member unions. This was probably more of a pretext: of course the LO could have built up such expertise if they wanted to. But the evasive argument is indicative of the deep-rooted opposition in the LO to more coordination. In the 1941 debate, where Westerlund complained about the lack of coordination, a principal discussion of more coordination in the future met some very intense criticism, called “Soviet wage policy” (the Foundry workers) and threats of unions leaving the confederation (the Typographers) if the program would be realized.

We must trace the evolution of the opposite view that wage policy should indeed be coordinated. A recurrent theme was that now that the trade union movement had reached a position of power, it must act responsibly. In 1942, the LO chairman August Lindberg stated:

We cannot pursue a wage policy independent of society's economic policy. Our bargaining rounds grow ever less a matter between us and the employers, because the bargaining affects social policy overall.³³

This is, of course, completely in line with the macroeconomic discussion in section two: Calmfors-Driffill would say precisely that a centralized trade union could be responsible, because it must consider the implications of its actions for all of society. In this spirit, in 1948, the Representation held a wide-ranging discussion on the theme of wage policy in a full employment society with economic planning. By now, Sweden had had a Social Democrat-led government, closely affiliated to the trade union movement, for sixteen years, and economic planning was du jour, a development reinforced by the Second World War. “It appears,” the LO-affiliated lawyer Arnold Sölvén argued, “that wage policy must be subordinated to general economic planning.”³⁴ Sölvén argued that no social movement could be above the law, and that this was true also for collective bargaining, but that to avoid direct state intervention in bargaining rounds the LO would need to coordinate its wage policies more closely. In the discussion, several representatives expressed support for a more activist LO wage policy. Representatives of the Factory Workers and Telegraph and Telephone Workers argued for a more coordinated wage policy, including the creation of a wage policy council or research institute, to encourage the development of wage policy. The Sawmill Workers agreed that the LO should view wages in a more macroeconomic way, and Gustav Vahlberg from the Secretariat (previously the Metal Workers) stressed the need for planning:

Allow me to be a bit brash and say that the trade union movement, which is the movement in the country which basically carries Social Democracy and a Social Democratic government, and time after time calls for planning, and time after time calls for solidarity between social groups, when we inspect ourselves from the outside, we do not show a lot of what we tell others they should manifest. The last couple of bargaining rounds show very strikingly that we are lacking in internal solidarity.³⁵

This is crucial. In the late 1940s the LO still had no systematic wage policy. But they were part of the perhaps most powerful labor movement in the world, in their own country, and the discussion of the new possibilities of wage bargaining intensified. One can almost see when reading the Representation protocols from the late 1930s to the late 1950s how over time the old “mind your own business” mentality of the constituent unions, reflective of the multitude of unions existing in early twentieth century Sweden, faded away. It was replaced by a much more centralized, “responsible” mentality, where a union leader should not only care for his (at this time, it was “his”) own members, but for society as a whole—and the Social Democratic project.

A wage bargaining model for a Social Democratic full employment society

In the late 1940s and early 1950s, there was a qualitative shift in the LO wage debates. A new element was that economists and experts were becoming central. In the uncoordinated wage bargaining rounds of the 1930s and 1940s, there was no need for macroeconomics: the policy was basically that each union should attempt to get as much as possible, without regard for the economy as a whole. But in the late 1940s, the discussions turned decisively to wage policy as a responsible part of overall economic and social policy, and then macroeconomic projections were needed.³⁶

In 1949, the Representation discussed a proposal to start an LO research institute. Two interesting arguments for LO research were presented: one, that in a full employment society the trade unions were part of the overall economic policy and therefore must have the expertise to take part in planning economic policy and public commissions. Two, that “a rational and more centralized wage policy makes considerable demands on the LO's capacities for an effective planning and design of wage rounds.”³⁷ The research institute proposal started a major debate at the Representation and had overwhelming support. The only critic argued in a markedly antimodernist way that psychology was more important than economics in bargaining rounds, and that the influence of economists would lead to a more conservative wage and strike policy.

The complaint was not heeded; the economists perceptibly gained ground in the 1949 debates. At the autumn meeting of the Representation, the wage policy discussion was introduced both in a statement by the Finance Minister (Per Edvin Sköld) and, crucially, one by the LO economist Rudolf Meidner. Not coincidentally, this was also the year when a programmatic statement was made in favor of a completely new order of wage bargaining, with calculations of the possible scope for total wage increases, and a centralized allocation of who should get what. The daring proposal, a milestone in LO wage policy, came from Knut Larsson from the Wood Industry Workers:

there is some scope for wage adjustments. These should then in the first place come to low-paid groups of workers /…/ Representation should commission the Secretariat to review the member unions’ proposals for wage adjustments. /…/ The Secretariat would then reconcile the proposals, so that the estimated total result of the wage rounds would fall within the framework of what is possible without disturbing the macroeconomic balance.³⁸

This represents the most developed sketch so far of a truly coordinated wage policy and includes the calculation of a “room” or framework for total wage increases; centralized calculation and decision-making about distributing the “room” among the workers; and the transfer of bargaining authority to the center.

Such an elaborate proposal was possible in the new context of macroeconomic expertise. In 1949, for the first time the LO wage discussion extended to the potential size of the total wage increases. Implicitly, this was about maximizing wage-earners’ gains without harming the economy. Naturally, the calculations differed: The Finance Minister argued that the scope was between 300 and 500 million kronor, while the LO economist Meidner claimed 600–700 million.³⁹ But the important part here is simply the new technocratic influence over policy, that it was considered relevant at all to calculate a total “room” for wage increases. Meidner admitted that it was controversial for a functionary like him to take a stand in wage policy debates, and this was not unnoticed. One unionist claimed that “our economists look something like the dervishes of our time. I am aware that we need these people to help us, but something seems to have happened lately. Something must have been added to their food.”⁴⁰

In the 1949 bargaining round, the LO economists very concretely introduced a more macroeconomic understanding of wage policy. For the first time in the bargaining rounds investigated here, they also presented an inventory of the wages for all the LO groups. This gave rise to a long discussion about which groups should be considered low-paid and therefore prioritized in the bargaining round, and how this solidaristic aim could be accommodated in a macroeconomic framework. In this discussion, the representatives of quite a few unions—the lithographers, brewery workers, transport workers, wood industry workers, paper workers, etc. —debated precisely the issue of which groups were low paid and how this could be determined. As a representative of Wood remarked, “probably no-one among us can explain what solidaristic wage policy means.”⁴¹ With hindsight one can say that defining “solidaristic wage policy” was precisely what they were up to in debating the descriptions of various workers claimed to be low-paid by different member unions, with their different demands, and how they could be made to add up to the estimated room for total LO wage increases.

The greater capacity for wage policy coordination—and solidarity—must be seen in the macroeconomic context, and in the context of post-war enthusiasm for economic planning. Mohlne from the Factory Workers was certain that “the development will forcefully cause greater planning in wage policy,” and Näslund from the Wood Workers was equally confident, proclaiming that “the government should always intervene with union policy, at least when it's a Social Democratic government,” and that the LO and the government should always consult each other. The general tenor was that the age of free wage policy was over, and that somehow under the conditions of full employment a coordinated and differentiated wage policy should be combined with stabilization.⁴² Lantz argued that the proposals put forward by each of the member unions added up to about the same as the wage room found by Sköld, which, Lantz argued, refuted the view that the Swedish trade union movement would not be mature enough to supply a differentiated wage policy in a tough economic situation.⁴³

Similarly, the wage debate in the following year, 1950, started from a calculation of the room for total wage increases—700 or 800 million kronor, according to Finance Minister Sköld—and how this should be distributed.⁴⁴ The wider political context invoked by Sköld was the need to keep the farmers as coalition partner happy, and the diatribes of the Communists against company profits. Again, we must see that contrary to the assumptions in the existing literature of the home market worker vs export market worker as the crucial conflict, the conflicts of the wage bargaining rounds were much more many dimensional than that, including the classic worker vs. capital conflict as well as farmers and other groups. At the August 1950 meeting, the LO chairman advocated a return to the war-time practice of extra taxes in sectors with high profits. This issue stuck. The year after, the high inflation induced by the Korea War continued, and so did the high profits in certain export sectors. Sköld maintained that “we must attack the profit boom” – but with tax policy rather than wage policy.⁴⁵

The practice of coordinated wage policy

In 1951, the LO's official policy was for the first time a coordinated bargaining round with differentiated/solidaristic pay increases. Lantz from the United Unions was jubilant that the Secretariat had come around to the position—a centralized, solidaristic wage policy—that he had espoused already in 1949. The policy built on a calculation of the room for wage increases, and the LO chairman Strand presented a model where the growth of the total wage sum would be 4 percent but that especially prioritized groups, such as farm workers and forestry workers, would get 9–10 percent. The LO economist Rehn emphasized the difficulty for the economists to calculate the room for wage increases, but did report that 4–5 percent real wage increases compared to 1950 would be possible without “a price increase, which would be too provocative.” Finally, then, the LO had arrived at a fully-fledged coordinated wage policy; the wage policy statement of the Representation gave the Secretariat the task of negotiating centrally with the SAF, and instructed the member unions to coordinate their demands.⁴⁶

With regard to the actual implementation of a solidaristic wage policy, very concrete detailed proposals were put forward. Representatives of metal, forestry, municipality, and other unions expressed support for the innovation, but submitted opinions—not without an eye to their own unions’ interests—on the way to determine which groups were to be considered low-paid and thus deserving of extra pay increases. One concrete proposal on how to allocate the wage increases built on the wage increase per agreement area since 1939, another on a calculated average wage per job in 1950. This proposal prompted a lively debate on the relative positions of saw-mill workers, carpenters, painters, plumbers, and clothing workers (the women in particular). The decision of the meeting followed the Secretariat's proposal for coordinated bargaining, with 5 percent average real wage growth, plus an extra tax on the profits of the forestry industry. The follow-up came with an extra meeting at the end of January 1952, when the Secretariat had to report that the employers were demanding central negotiations. The final, path-breaking central agreement gave larger pay increases for low-paid workers and for women, a proposal eliciting much satisfaction. Beyond some criticism from high-wage unions (the Miners, Construction Workers and Book Binders), the main criticisms were that the increases for the low-paid and the special pay rises for women did not go far enough.⁴⁷

The centralized wage bargaining model, with special considerations for various groups (“solidaristic wage policy”) had now been established. With some temporary steps back (for example 1953–1954) this model was practiced until the 1970s. Wage policy was now a part of macroeconomic policy, and a part of welfare policy. The 1954–1955 discussions began with a presentation by an LO economist on workers’ benefits. This investigation was intended to be part of the “long-term planning of wage policy,” in which not only wages but also benefits would be considered.⁴⁸

The bargaining round of 1955 is a good example of the further developments. It was marked by both optimism about the economy and the LO's own capacity, but also with a certain unhappiness with the fall back into uncoordinated bargaining in 1953–1954. The LO economist Meidner reported happily that the economy was between boom and over-heating, and that wages had increased more than productivity over the last year. But wage growth had been unequal. The LO chairman Strand argued for a truly centralized, solidaristic wage policy: the LO should strive for a fair share of increased productivity, and with a solidaristic distribution among the workers. A cunning politician, Strand would not of course have proffered this argument had he not known that it had support within the organization. Winroth from Forestry (Skogs- och flottningsarbetareförbundet) concurred, and emphasized that greater year-to-year continuity was needed, with an end to the oscillation between coordinated and “free” bargaining rounds. Furthermore, he argued that the LO should also coordinate with the Union for Professional Employees (TCO) and the farmers so that the latter would not be able to wait for an LO agreement and then leapfrog it. This is fascinating as it adds another dimension to the coordination problem: not only competition between sectors within the LO, or with the employers, or with farmers, but also with white-collar employees. In fact, such coordination, LO-TCO-SAF, was implemented in this year. A two-year central agreement with SAF was reached, with a solidaristic distribution of pay increases.⁴⁹

In 1956–1957, TCO refused to participate, but the LO-SAF central negotiations continued. Again, the solidaristic wage policy was formulated especially as real income growth (öre) for the low-paid and a set percentage for everyone else.⁵⁰ In the 1958–1959 bargaining round, when the previous agreement expired, the LO and SAF again pursued a two-year agreement. By now, LO power over the member unions’ wage policies, and two-year agreements were by now almost uncontroversial, taken for granted. Thus, the transformation that I have sought to reconstruct was over; the classical “Swedish model” of wage bargaining had been established.

Conclusions

The present investigation has shown that the received wisdom on the origins of the Swedish coordinated wage bargaining model is misleading. Industry wage leadership did not develop with any cross-class alliance of employers and unions in export industry in the 1930s, as argued by the previous literature.⁵¹ A coordinated wage policy, with an emphasis on responsibility of the economy as a whole including competitiveness developed in the 1950s and concerns for the exporting sector was a part of this project, but not a dominating one. The LO's wage policy, once it was freed of the parochial mind set of “everyone get what you can” of the 1930s and 1940s, did turn in a responsible direction. But not in the way suggested by previous research, neither in its shape (it wasn't export sector wage leadership, but centralized bargaining) nor in its drivers.

Incentives for responsible wage policies on behalf of trade unions can arise both in coordinated/wage leadership systems, and in centralized systems.⁵² While previous research on Swedish wage leadership, in line with the “export-sector bias”⁵³ of comparative political economy, has seen cost internalization in the shape of wage leadership from the 1930s on, it arose differently and later—in the 1950s, in the context of bargaining centralization. It was not about export-sector dominance over home-market unions such as construction; instead, it was the unions’ participation in the Social Democratic dirigisme of the 1940s and 1950s.

The Swedish coordinated wage bargaining model should be understood as part of the Social Democratic project of reining in capitalism; responsible trade union wage policy in these years was from the beginning more “macro-corporatist” than sectoral.⁵⁴ The reason why the LO started to emphasize responsibility for the economy as a whole was not that they were bludgeoned into it by employers in engineering.⁵⁵ Instead, it grew out of the sense that they were, with the Social Democrats continuously in government after 1932, were building a Social Democratic society. As the LO's chairman Arne Geijer emphasized in the 1960s when he explained that the LO's wage policy had to take responsibility for the wider society: “society is more or less ourselves and our families.”⁵⁶ In a situation with almost universal union membership and the dominance of blue-collar work, this statement made sense. The LO in the 1950s or 1960s could not pursue a wage policy without regard to the economy's overall performance; the “get what you can” wage policy of the 1930s was not possible anymore.

Instead, the new instruments of macroeconomics were put to use to calculate the possible scope for wage policy, to get as much as possible without harming the economy. Here, the investigation of this paper ties in with the rather rich literature on the development of the so-called Rehn-Meidner model of coordinated wage policy and macroeconomic policy.⁵⁷ This model was a Swedish labor movement alternative to Keynesianism. It entailed a combination of solidaristic and expansive wage policy, a relatively austere economic policy to counter inflation, and a degree of economic planning to steer investments and consumption. This was indeed a fascinating policy model of some originality. However, the Rehn-Meidner literature typically focuses on the postwar situation and the need to develop a policy mix for an economic situation of full employment and rapid growth. Here, the purpose has rather been to trace the development of wage policy in a longer time frame. In this way, the research here is complementary to the stream of research on the Rehn-Meidner model.

Finally, let us consider the evolution of wages and income distribution. In the postwar period, the LO's wage policy gave wage increases even above the high pace of productivity growth, which meant that the wage share of the national income grew.⁵⁸ Wage policy was redistributive both between different types of workers, as noted in the literature,⁵⁹ and between the classes, redistributing from capital to labor. Wage policy could redistribute incomes from profits to wages and employee benefits without harming the economy, as consumer demand was boosted by rising wages, and the economies with which Sweden competed, also boosted their wage shares in the same period.⁶⁰ Swenson's argument that within-class conflict matters as much as between-class conflict, or more, is important and brilliantly presented but must not be overstated: The construction of the Swedish wage bargaining model was not only about within-class conflict and cross-class alliances. Neither on the institutional level, nor on that of the real economy. A peaceful labor market with collaborative relations between unions and employers did evolve after 1938, but this does not mean that the underlying conflict between labor and capital vanished, just that it was handled in a collaborative, positive-sum way.⁶¹ The history-writing on the Swedish model has in the last decades turned to a perspective of within-class conflict and between-class collaboration. This is certainly an important aspect of the Swedish labor market, but not the whole story.