Explaining postwar wage compression

Abstract

One of the main problems confronting labor unions during wage bargaining is how to deal with the conflicting demands of different groups of workers over the division of labor market earnings. This article explains how their internal organizational blueprint determines how they deal with this and criticizes the scholarly preoccupation with union density and wage bargaining centralization as explanatory variables for cross-national and temporal variation in wage inequality. It does so based on a critical analysis of collective bargaining in the Netherlands and the United Kingdom during the first four decades of the postwar period.

Keywords:

• Collective bargaining
• income inequality
• wage rates
• The Netherlands
• The United Kingdom

The question of how labor unions affect the distribution of labor market earnings continues to attract much scholarly attention. The literature on organized labor’s effect on wage inequality is large and spans several decades. Yet much of it, and especially the most recent, has paid insufficient attention to the very different ways in which labor union movements deal with the different and often conflicting demands of their members regarding the wage distribution. In much of the literature on wage inequality unions are viewed as simple utility maximizers whose policies are a direct extension of the interests of a majority of their members. This has resulted in a strong tendency to assume labor union support for wage compression and other policies that affect the distribution of income among different groups of workers (e.g. Baccaro, 2011; Forth & Millward, 2001; Hyman, 2001; Marks, 1989; Metcalf, Hansen, & Charlwood, 2001; Stephens & Wallerstein, 1991). Illustrative of this tendency is the frequent use of simple median voter models to explain both labor union support for redistribution and the importance given to union density and wage bargaining centralization as explanatory variables for cross-national and inter-temporal variation in wage inequality (e.g. Blau & Kahn, 1996; Iversen & Soskice, 2009; Pontusson, 2006; Pontusson & Rueda, 2000; Pontusson, Rueda, & Way, 2002; Scheve & Stasavage, 2009; Visser & Checchi, 2011; Wallerstein, 1999).

This article critically examines the scholarly preoccupation with union density and wage bargaining centralization. It emphasizes that union support for within-class redistribution cannot be taken for granted, and depends on the internal organizational structure of labor unions. It argues that the way in which workers organize into labor unions is consequently as important for the distribution of labor market earnings as the degree to which they do so. The article also criticizes the popular view that centralized bargaining reduces wage inequality because it strengthens the relative bargaining power of unions representing low-paid workers (Iversen, 1999; Iversen & Soskice, 2009; Wallerstein, 1999; Pontusson & Rueda, 2000; Pontusson et al., 2002). It shows that, to the contrary, long-term centralization of wage bargaining is only possible when labor unions representing higher paid workers are willing to support wage compression. The analysis suggests that centralized wage bargaining, rather than being an independent explanatory variable of wage inequality, like union support for wage compression depends on organized labor’s internal organizational structure.

An important goal of the article is to identify for what reasons unions support or oppose redistribution of income among different groups of workers. The issue of labor union support for wage compression in particular raises important questions that have received insufficient attention from scholars. How, for example, are we to explain the many instances of union support for wage compression that only benefit workers in the lowest decile of the earnings distribution and do so at the expense of all other workers? Such instances cannot be explained by arguing that union policies reflect majoritarian interests. Building on previous work that has sought to show how normative considerations shape the behavior of collective actors, this article emphasizes that the exclusive focus of wage compression policies on the lowest paid during much of the postwar period shows that unions are not merely utility maximizers, but are guided by norms and values that emphasize fairness and worker solidarity (Elster, 1989; Hyman & Brough, 1975). How these norms work out though, this article shows, crucially depends on the organizational blueprint of labor unions.

Equally important are questions relating to the willingness of higher paid workers to accept wage bargaining outcomes and policies that reduce wage differentials. How do union leaders who for instance seek to reduce occupational and skill differentials persuade their higher paid members to support this? Why do these members not simply break off and form separate organizations? The conventional literature on wage inequality often either ignores these questions or simply assumes that higher paid workers support wage compression out of solidaristic considerations.¹ Following neo-corporatist studies on wage restraint from the 1980s, this article engages these questions by arguing that unions who are able to compress wages in the long-term generally have a representational monopoly which rules out that discord can find effective organizational channels (e.g. Keller, 1981; Regini, 1983). It shows that union support for wage compression policies can invoke much resentment among higher earning union members. Yet when higher paid workers organize in the same unions as lower paid workers, they are generally unable to put up effective resistance to it.

With an emphasis on the importance of labor’s internal organizational blueprint for union wage bargaining behavior, the analysis also advances another way of looking at the relationship between wage compression and wage restraint. Working on the assumption that unions support wage compression, there has been a strong emphasis in the political economy literature on the usefulness of egalitarian wage outcomes in persuading unions to moderate their wage demands (e.g. Flanagan, Soskice, & Ulman, 1983; Iversen, 1999; Pontusson, 2011; Swenson, 1989). This article shows that wage compression may just as well work to undermine labor union support for wage restraint policies. It does so by describing several instances in which the resistance of unions representing high-paid workers to wage compression resulted in the demise of incomes policies aimed at achieving wage restraint – and the centralized wage bargaining procedures that accompanied these policies.

I present my argument through a comparative-historical analysis of postwar wage bargaining in the Netherlands and the United Kingdom. I do so for two reasons. First, the British and Dutch labor union movements arguably differed more from each other in terms of their internal organizational features than any other two union movements in the postwar period. Second, a comparison of these two countries allows me to capture both cross-national differences and changes in union attitudes toward wage compression over time. Both are of crucial importance to illustrate my claims. In the first part of the analysis, I contrast the strong opposition of British occupationally organized unions to any attempt to improve the relative wages of Britain’s lowest paid workers with the unequivocal support of Dutch unions for wage compression in the lower half of the wage distribution during the first decades of the postwar period. I illustrate the consequences of this for the long-term survival of wage restraint policies and wage bargaining centralization in both countries. I then show how changes in the internal structure of the Dutch labor union movement over time enabled it to successfully press for wage compression in the upper half of the wage distribution as well. I show how higher paid workers were unable to put up effective resistance to this because the sectoral basis of labor relations in the Netherlands worked against them.

Union structure and the distribution of labor market earnings

The question of how unions solve the vast potential for conflict between different categories of workers over the distribution of labor market earnings is of huge importance to our understanding of the determinants of wage inequality. It is also a question that has received far too little systematic attention in the literature. Comparative-historical analyses that describe how unions solve internal conflicts between higher and lower paid workers are scarce and pay little attention to how their organizational blueprints matter to this. Most recent accounts of organized labor’s effect on the distribution of wages neglect the issue altogether and, following a long-established tradition in the field of labor economics (e.g. Booth, 1995; Checchi, Visser, & van de Werfhorst, 2010; Freeman, 1980) explain union policies through theoretical models that view collective behavior as an aggregated form of individual means-tested rationality. In this view, labor unions are utility maximizing agents whose policies are a direct extension of the interests of a majority of their members. In the past decade, many prominent writings have used simple median voter models to explain their assumption of labor union support for intra-class redistribution and emphasis on union density as an explanatory variable for wage inequality (Kaufman, 2007; Oskarsson, 2005; Pontusson, 2006; Pontusson & Rueda, 2000; Wallerstein, 1999). The logic of the median voter model is also commonly invoked to explain why centralized wage bargaining, by strengthening the bargaining power of lower paid workers, is supposed to lead to a more equal wage distribution (Iversen, 1999; Oliver, 2008, p. 1554; Pontusson et al., 2002; Wallerstein, 1999).

Median voter models have long been a popular instrument in the social science literature. Yet their frequent use to explain labor union support for wage compression is a more recent phenomenon.² Applied in this way, labor unions are believed to support wage compression because in situations in which the mean wage is higher than the median wage, which is typically the case, a majority of union members are supposed to benefit from this. As ‘utility maximizers’ unions should therefore, for instance, prefer flat-rate wage increases over percentage increases (Booth, 1995). This explanation of labor union behavior is filled with empirical and theoretical problems. One obvious problem with it is that the median union member is by no means necessary a below-average earner.³ And even if the median union member was to gain from wage compression, there is still the problem that unions clearly also organize many above-average earners. The view of unions as mere utility maximizers evidently fails to make clear why higher paid union members should accept wage compression policies – especially when they organize separately, and thus cannot be forced to support wage compression by a majority of lower paid workers. Nor can the idea of unions as utility maximizers explain why during much of the postwar period, as we will see, wage compression policies were primarily focused on improving matters for a small group of low-paid workers that represented a mere fraction of the (unionized and non-unionized) labor force.

Labor unions, then, are not mere utility maximizers. Different types of unions have on different occasions supported the introduction of minimum wages that benefited a fraction of union members at the expensive of all other workers, wage policies that combined percentage increases with relatively higher wage increases for the lowest percentile of wage earners, and social security policies with wage limits that restricted the scope for redistribution to the lower half of the wage distribution (Oude Nijhuis, 2009). Such activities show that union policies are not merely shaped by the interests of a majority of union members. Unions can evidently also be guided by normative orientations that emphasize fairness and broad worker solidarity. Yet the influence of such norms on unions cannot be taken for granted. Whether and how such norms shape labor union policies depends crucially on their organizational blueprint. Of most importance to this, I argue, is whether unions seek to organize workers of various occupations or limit their membership to certain (often privileged) occupations.

The way in which norms like fairness and solidarity interact with union structure can be found in the organizational logic of these two types of union organization. Unions that seek to organize workers regardless of skill and occupation may do so because they emphasize the importance of labor unity against capital. Alternatively they may do so because either a socialist or confessional ideology that emphasizes broad worker solidarity guides them.⁴ Whatever the reasons though, such unions will inevitably have to deal with conflicting demands from higher and lower paid members regarding the intra-labor division of wages. One successful way of reconciling these different demands is by emphasizing their organizational logic, which is based on class unity and solidarity among its membership. By contrast, the leaders of unions that limit their membership to certain occupations do not have to deal with conflicting interests; nor does their organizational logic emphasize class unity and broad worker solidarity. On the contrary, and as pointed out by various labor historians, the organizational logic of such unions is grounded in the occupational defense of the skilled wage differential (e.g. Bain, 1970; Gross, 1982). As a result, they can be expected to oppose any measure that reduces the wage gap between their members and lower paid workers.

An excellent way to demonstrate how their organizational blueprints shape union views of occupational wage relativities is by comparing the postwar wage bargaining behavior of the Dutch and British union movements. During much of this period, they provided near-complete examples of union movements that, respectively, seek to organize workers of all skills and those in which higher paid, more skilled workers organize separately from less privileged workers. In the United Kingdom, nearly all skilled manual workers as well as large groups of white-collar workers have always organized on an occupational basis. While the importance of occupational unionism was at its peak during the first half of the postwar period, it has, as we will see, remained a powerful economic force in subsequent years (Ebbinghaus & Waddington, 2000). In the Netherlands, by contrast, occupational divisions have been of little importance during most of the postwar period. By as early as the 1920s, all manual workers there had united under the banner of industrial unionism (Visser, 2000). From then on, occupational unionism only persisted among white-collar workers. Yet it would not continue to persist indefinitely. From the 1930s on, the major union federations worked actively, and often forcefully, to disband these remnants of occupational unionism in the Netherlands. This process was completed by the late 1960s. By then, all members of the three main union federations in the Netherlands, which together represented some 85% of all organized workers, were organized on an industrial basis – a percentage that remained more or less steady in subsequent decades.⁵ Together, the three federations completely dominated the national wage bargaining machinery – thus reinforcing the dominance of industrial or ‘open’ unionism in the Netherlands.

The following analysis of how these very different organizational blueprints shaped British and Dutch union views of wage relativities opens with an account of their wage bargaining behavior in the immediate postwar period. The analysis shows how the industrial nature of union organization in the Netherlands enabled unions there to push for steady improvements in the relative wages for the low paid. It contrasts this with the situation in the United Kingdom, where occupationally organized unions successfully resisted any attempt to reduce wage differentials – thus making it impossible to improve matters for the lowest paid. Illustrating that union support for wage compression does not have to reflect majoritarian interests, the analysis shows that the redistributive agenda of Dutch unions was in this period exclusively aimed at improving matters for the lowest paid. The vast majority of both organized and unorganized workers in the Netherlands paid the price for this solidaristic wage stance. In the United Kingdom, the motivations for the resistance of occupationally organized unions to wage compression were various, but always reflected an absence of solidarity with lower skilled workers.

With an emphasis on the occupational defense of wage differentials in the United Kingdom, the following analysis criticizes the view that centralized wage bargaining reduces wage inequality because it strengthens the relative bargaining power of unions representing low-paid workers. It shows that, to the contrary, long-term centralization of wage bargaining is only possible when unions that represent large groups of higher paid workers support wage compression. It does so by pointing out that while unions representing low-paid workers may indeed demand ‘their fair share of the agreement’ (Iversen and Soskice, 2009, p. 448. Also Iversen, 1999; Pontusson et al., 2002), under coordinated bargaining, unions representing high-paid workers are sure to do the same. Under such circumstances, coordinated wage bargaining can only empower unions representing low-paid workers if bargaining at the central level were to resemble negotiations within a single union – a state of affairs which can only come about when unions agree to surrender a large part of their power to a central union federation. Such a state of affairs is highly unlikely to come about under conditions of occupational unionism under which the economically most powerful unions jealously defended their autonomy in order to preserve the privileged position of their skilled, high-paid members.⁶ Such unions may, as we will see, agree to centralized bargaining when economic conditions demand this. Yet they will no longer support centralized wage bargaining when this results in a reduction in the occupational wage differential.

The analysis similarly criticizes the wage restraint literature for its failure to appreciate the importance of the preoccupation of occupationally organized labor unions with wage differentials for the failure of postwar British attempts to set up durable incomes policies. Because of the strong tendency to assume labor union support for wage compression and based on a superficial reading of British sources, it has for instance been common for scholars to argue that temporary British union support for wage restraint was at least partly grounded in ‘the prospect which it offered at bargaining for a fairer distribution of incomes’ (Flanders, 1970. See also Hyman & Brough, 1975). Others have argued that British labor unions at least took ‘advantage’ of the situation ‘to raise wages for low-pay workers faster than for the rest’ (Swenson, 1989, p. 28. See also Flanagan et al., 1983, pp. 673–675; Mares, 2006, p. 177). The view that attempts to improve the relative wages of the lowest paid facilitated broad union support for wage restraint in the United Kingdom is incorrect. The following analysis shows that, to the contrary, these attempts strongly worked to undermine it.

Britain: occupational unionism and the importance of wage relativities

Part of the reason why scholars have been able to link temporary British labor union support for wage restraint policies to wage compression is that such policies did often produce a (temporary) reduction in wage differentials in the United Kingdom. During the early 1940s for example, when the war effort forced unions to support a policy of voluntary wage restraint, there does indeed seem to have been a relative increase in the wages of the country’s lowest paid workers. To protect these workers from the consequences of wage restraint in a time of rapid inflation, several industries occasionally adopted flat-rate wage increases.⁷ Some industries at times also opted for percentage wage increases with numbers that varied depending on the wage level (MRC, MSS.292/110/1: War additions or bonuses, 1 December 1940). It is hard to see whether or to what degree these policies were standard practice during the War. What is clear, however, is that they did affect the wage distribution, especially in the manual category (Turner & Jackson, 1969).

Yet it would be plainly mistaken to argue that this compression received unequivocal labor union support in the United Kingdom. And it would be even more mistaken to argue that it facilitated labor union support for wage restraint. On the contrary, in the long-term it strongly worked to undermine labor union support for such restraint. There is much primary evidence that suggests that Britain’s occupational unions – especially the craft unions that represented skilled manual workers – strongly resented the decline in wage differentials that they had to accept during the War. They made this perfectly clear when they announced that they would seek to restore lost differentials in immediate postwar years.⁸ They also made this clear when the incumbent Labour government in 1947 approached them with a request to support a new incomes policy. Acknowledging the ‘serious economic position’ in which the United Kingdom found itself, they agreed to a new voluntary incomes policy based on coordinated wage bargaining. Yet they made it clear that they would only do so if the government were to

recognize the need to safeguard those wage differentials which are an essential element in the wages structure of many important industries, and are required to sustain those standards of craftsmanship, training and experience that contribute directly to industrial efficiency and higher productivity. (TUC, 1948, p. 290)

Both the Labour government and the TUC’s main executive body, the General Council, took this warning to heart. In private dealings between the two, General Council members repeatedly emphasized that it was simply not possible to do something extra for the low paid as this would conflict with maintaining existing differentials (MRC, MSS.292/110.44/1b: Meeting Ministers and Special Committee of the TUC, 11 February 1948, p. 9). Labour representatives were evenly frank on the issue. In an attempt to obtain TUC support for wage restraint in late 1948, Foreign Secretary and ex-union leader Ernest Bevin even went so far as to suggest that ‘the Government might be assisting the General Council by making no reference to the low-paid since this would raise the question of differentials’ (MRC, MSS.292/110.44/2b: Special Committee on the economic situation, 14 November 1949, p. 6). This does not mean that Bevin was wholly unconcerned with the plight of the many British workers ‘whose wages are at present below a reasonable subsistence level’ (MRC, MSS.292/110.44/1b: Meeting Ministers and a Special Committee of the TUC, 11 February 1948, p. 9). It simply means that he realized that his government had no chance of obtaining renewed labor union support for wage restraint if it continued to insist on doing something ‘extra’ for Britain’s lowest paid workers.

The absence of specific measures for the lowest paid facilitated the coming about of a new voluntary incomes policy in 1948. Yet it could not guarantee its survival for long. An important reason for this was that many occupational unions were not satisfied with freezing the existing wage gap, but were actively seeking to undo previous reductions in wage differentials. This eventually set off a series of competing wage claims that focused primarily on restoring wage relativities. TUC representatives were well aware of this, and noted in 1950 that those unions that were ‘continuing with their claims for wage increases … negotiated on behalf of persons entitled to differentials’ (MRC, MSS.292/110.33/2a: Minutes Special Committee on the economic situation, 26 April 1950). They also noted that this ‘drift’ had come about because of the ‘discontent felt by skilled and semi-skilled workers at the steady narrowing in differentials which has taken place in recent years’ (MRC, MSS.292/110.44/2a: Wages policy – statement to affiliated unions, 20 June 1950, p. 2. See also MRC, MSS202/110.44/2b: Wages policy, 1 June 1950). While acknowledging that other factors also played a role in the breakdown of the incomes policy, TUC representatives clearly viewed this drift as most important. As noted by a TUC committee on the matter, the competitive wage increases that led to the incomes policy’s implosion were the inevitable consequence of previous reductions in wage differentials, and this meant that ‘the consequential adjustments to maintain such differentials must be faced’ (MRC, MSS.292/110.44/2a: Minutes Special Committee on the economic situation, 9 June 1950).

Under these circumstances, it obviously proved impossible to maintain a system of centralized wage bargaining in the United Kingdom. Following on the breakdown of the voluntary incomes policy in 1950, Britain’s occupationally organized unions aggressively set out to restore lost differentials. They were quite successful in this as the manual wage differential continued to widen in the United Kingdom during the 1950s and 1960s, and eventually reached pre-war levels during the latter decade (Pond, 1981).

The Netherlands: industrial unionism and broad worker solidarity

The events that took place in the United Kingdom in the immediate postwar period stood in stark contrast to those in the Netherlands where it did prove possible to improve the situation for the low paid by reducing wage differentials on a permanent basis. Like in the United Kingdom, the first major reduction in wage differentials in the Netherlands occurred during the War.⁹ In contrast to in the United Kingdom, this did not provoke a backlash for a restoration of wage differentials in the immediate postwar period though. Instead, the socialist Dutch Union of Labor Associations (Nederlands Verbond van Vakverenigingen, henceforth NVV), the largest of the big three, in 1945 announced that its support for wage restraint depended on the introduction of a ‘social minimum’ that had to be increased ‘not just in absolute but also in relative terms.’ It did so by arguing that under the current circumstances of austerity it was ‘reasonable that this sacrifice will be made by the better situated to prevent hardship for the most deprived’ (IISG, Archief NVV, doos 21, Notulen Hoofdbesturenvergadering, 19 July 1945, p. 22). Both this demand and its motivation were fully supported by the other main union federations in the Netherlands, the Catholic Workers’ Movement (Katholieke Arbeidersbeweging, henceforth KAB) and the protestant Christian National Labor Union (Christelijk Nationaal Vakverbond, henceforth CNV). They were also accepted by their employer counterparts, who in exchange obtained a system of centralized wage bargaining that proved to be remarkably successful in producing long-term wage restraint.¹⁰

The result was an initial sharp and then continual rise in the relative wages of the nation’s poorest workers. With a brief exception from 1954 to 1956, all wage agreements in the first 15 years of the postwar period contained specific measures to improve the relative wages of the low paid or broader redistributive measures that often benefited the low paid in particular.¹¹ During the 1960s such measures were somewhat less common, but this only seems to have slowed down the overall trend toward more compression of in particular the lowest half of the wage distribution. The decade also witnessed the introduction of a statutory national minimum wage, which came about mainly because of fears that the breakdown of the ‘guided wage policy’ that had characterized wage negotiations in the Netherlands for so long would undermine attempts to maintain a generous social minimum (Hueting et al., 1982, pp. 316–328). From the late 1960s on, the statutory minimum wage became the vehicle for a further sharp increase in the relative wages of the nation’s poorest workers. This resulted in a social minimum wage that was set at an unprecedented rate and was more generous in relative terms than any of its counterparts abroad (O’Riodan, 1981, p. 5).

There is no doubt that this massive increase in the relative wages of the poorest workers could come about in the Netherlands because there were no craft unions defending manual wage differentials there. To be sure, the union federations did sometimes report that there were ‘tensions’ within industries that forced the industrial unions to limit their demands for improvements in the relative wages of the lowest paid. But these rarely forced unions to refrain from doing something extra for the lowest paid altogether.¹² And while there must have been workers who resented the strong decline in differentials that occurred mainly in the lower half of the wage distribution, and voiced their opposition to it, this never led to major internal conflicts within the labor union movement. There certainly never was an organized movement seeking to restore skilled manual differentials that were eroded by the practice of combining percentage with minimum wage increases – a movement that, as we have seen, did come about in the United Kingdom.

Yet this does not mean that wage compression was wholly uncontroversial in the Netherlands. On the contrary, in the first decades of the postwar period there was only a broad consensus on the need to improve matters for the nation’s lowest paid workers. This consensus did not extend to measures that would have resulted in a more equal division of wages across the whole wage spectrum, such as flat-rate wage increases. Nor did it extend to measures that were specifically designed to curtail the relative wages of top earners, such as percentage increases combined with maximum ceilings on pay rises. During the first decades of the postwar period, such measures were much less common than wage compression policies that only focused on improving matters for the lowest paid workers by combining flat-rate wage increases with minimum pay rates. The reason for this is that they could often count on strong opposition, coming from organized employers and the three main union federations.

The opposition of the three federations to these measures strongly conflicts with the conventional view of labor unions as utility maximizers. It is, by contrast, in full agreement with the view that labor union support for wage compression depends on its organizational blueprint. After all, while lacking craft unions that only represented skilled manual workers, the Netherlands did have white-collar unions that exclusively catered for specific (read: privileged) white-collar occupations. These occupationally organized white-collar unions sometimes organized independently, but most did so within one of the three main union federations (Harmsen & Reinalda, 1982). Their presence had a strong impact on the union federations’ policies toward pay relativities. It did so because it made it impossible for them to ignore the interests of the high-paid members of their white-collar affiliates. While not by definition opposed to efforts to improve the relative wages of the lowest paid – perhaps also because this mostly affected relativities in the lower half or ‘manual’ part of the wage distribution –, these white-collar unions strongly opposed any wage settlement that included flat-rate wage increases or maximum ceilings on pay rises. They did so for good reason, as such outcomes were in clear conflict with the interests of their members.¹³

That the main union federations did not ignore the stance of their occupational white-collar members regarding pay relativities can be seen from the start of the postwar period. When these federations met with their employer counterparts to negotiate the first postwar wage settlement in 1945, they immediately agreed on a substantial reduction of pay differentials in the manual category. They did so by creating a classification system that distinguished only three manual categories: skilled, semi-skilled and unskilled workers. The first of these were allowed to earn, respectively, 20 and 10 % more than the latter. It is hard to say by how much this would have affected existing wage differentials, but there is wide agreement that this amounted to a considerable degree of wage compression (Mulder, 1956, p. 21; Smulders, 1952, p. 96). Because of acute shortages in the demand for skilled labor, this agreement, and the considerable amount of compression it entailed, soon proved impossible to maintain. Yet the classification system itself was maintained and defended by the union federations as a way to do improve matters for the nation’s lowest paid workers. Crucially, and on the insistence of both employers and the white-collar unions, it continued to apply only to manual workers though (Smulders, 1952, p. 96).

The limited nature of union support for wage compression at the time also shows in the statements of union leaders. On several occasions, they motivated union demands with a redistributive element by denying that they had a broader redistributive agenda beyond the goal of improving matters for the lowest paid (IISG, Archief NVV, doos 21, Verslagen looncongres, 31 October 1945, pp. 7, 19; IISG, Archief NVV, doos 21, Notulen Hoofdbesturenvergadering, 19 July 1945, p. 19; Ibidem, 27 August 1946, p. 106 and 9 January 1946, p. 42). In line with these remarks, the three union federations frequently resisted government attempts to introduce maximum ceilings on the growth of wages and not once agreed to push for flat-rate wage increases.¹⁴ The only occasion on which some union federations pushed for flat-rate wage increases came in the late 1940s. While defended by the Christian-democratic KAB and CNV because the government had announced that it would only allow price compensation ‘which affected all workers in the same degree’, the socialist NVV strongly resisted this (IISG, Archief NVV, doos 312: Notulen vergadering Raad van Vakcentralen, 17 November 1949, p. 4). The NVV motivated this by referring to productivity considerations. Yet its stance undoubtedly also had much to do with the strong opposition of one of its most powerful members, the occupational white-collar union Mercurius, to flat-rate increases. The crisis eventually led to a strong ‘crisis of trust’ within the NVV, as several of its manual-oriented industrial union members had initially insisted on more overtly redistributive wage demands (IISG, Archief NVV, doos 11, Notulen 4e vergadering Verbondsbestuur, 14 November 1949, p. 41; IISG, Archief NVV, doos 11, Notulen 30e vergadering Verbondsbestuur, 30 October 1950, p. 323).

Despite additional support from employers, the occupational white-collar unions did not manage to preserve existing wage relativities for their members completely. Yet their strong opposition to flat-rate wage increases and maximum ceilings on pay rises certainly limited the scope for redistribution in the Netherlands. In fact, during the mid-1950s, they even managed to increase the wage gap between manual and white-collar workers (NVV, 1957). Broad support for wage compression was in the first decades of the postwar period thus only guaranteed if it focused on the limited goal of improving matters for the nation’s lowest paid workers. This would change radically in following years. By the late 1960s, all of the three main union federations were to adopt a much more ambitious egalitarian agenda. This included pushing aggressively for flat-rate increases and maximum ceilings. As we will see, both the coming about of this agenda and the ability of the three union federations to implement it depended crucially on changes in their organizational blueprint that had taken place in preceding years.

Union structure and the response of high-paid workers to wage compression

The preceding section illustrated how the respective dominance of industrial and occupational unionism among manual workers in the Netherlands and the United Kingdom mattered to attempts to improve the relative wages of the lowest paid in both countries. In addition, it showed how the presence of occupational unionism among white-collar workers in the Netherlands limited the scope for redistribution there. Another way of demonstrating the importance of labor’s organizational blueprint is by capturing how changes in union structure affect labor union attitudes toward wage relativities over time. The following section does so by demonstrating how the gradual disappearance of occupational unionism among Dutch white-collar workers, a process that was more or less completed by the late 1960s (Hueting et al., 1982) further increased the scope for redistribution in the Netherlands. It also shows how the persistence of occupational unionism among both manual and white-collar workers made sure that both attempts to improve the relative wages of the lowest paid and more far-reaching wage compression policies remained impossible to implement – or at least to sustain – in the United Kingdom.

The analysis also sheds light on a problem that is of central importance to our understanding of labor union support for wage compression: how do union leaders who represent both higher and lower paid workers and who seek to compress wages between workers with different occupational and skill levels persuade their higher paid members to support this? Why do these higher paid members not put up effective resistance against this, for example by breaking off and forming separate organizations? As we will see, when wage compression along the whole wage spectrum became an explicit goal of the main union federations in the Netherlands, many above-average earners strongly resented this. Yet they never managed to put up effective resistance to it. To understand this, it is important to appreciate the strong representational monopoly that the three major employer federations always had in the postwar period over the wage bargaining machinery in the Netherlands. This monopoly, I argue, prevented the creation of a viable occupationally organized union movement there.

An equally important question is how Britain’s occupationally organized unions managed to preserve wage differentials in the face of strong and persistent pressure for wage compression from other unions and consecutive Labour (and sometimes even Conservative) governments. Even combined, craft unions representing skilled manual workers and occupationally organized white-collar unions never represented more than 35% of the unionized workforce in the United Kingdom in postwar years. Moreover, and mainly because of the decline of craft unionism, their share of the unionized membership declined rapidly from the 1960s on. Yet, initially at least, and as we will see, this did not affect their ability to defend the occupational wage differential. To understand this, it is crucial to realize that this type of unionism does not draw on numbers, but on the skilled profile of their members. This, the following analysis shows, granted them a degree of economic power that far exceeded what we would expect them to have if we were only to focus on their numeric strength.

Skills over numbers: the economic strength of occupational unionism

When labor unions oppose wage settlements that involve wage compression, they may also be reluctant to accept redistributive government policies that result in a reduction in wage differentials. This makes it somewhat surprising that in much of the literature on labor market development unions are treated as key allies of left parties in their pursuit of a redistributive agenda (e.g. Bradley et al., 2003; Iversen & Soskice, 2009; Pontusson & Rueda, 2010; Simoni, 2013). The British experience shows that this does not have to be the case. When, for instance, a British Labour government in the mid-1960s first contemplated the introduction of a statutory national minimum wage to improve matters for the lowest paid, the TUC strongly opposed this. It continued to do so for the next twenty years. Existing explanations for the long-standing opposition of the TUC to this generally center on union suspicions of state motives or fears that increased state intervention might affect their future ability to attract workers (Blackburn, 2007, p. 189; Gospel & Palmer, 1994; Howell, 2005, p. 181; Metcalf, 1999, p. 172; Minkin, 1991, p. 429; Terry, 2000, p. 157). Yet empirical evidence suggests that the occupational defense of wage differentials was as least as important. For instance, when the TUC first approached its affiliates on the issue during the late 1960s, craft unions like the Iron and Steel Trades Confederation simply motivated their opposition against it by arguing that it would ‘retard the progress … of those workers whose earnings were above the statutory minimum.’ This, the union warned, would not be accepted by its members. To support this, it mentioned that when steel employers had some years ago suggested concentrating wage increases on the low paid, its members had strongly opposed this. Occupational white-collar unions like the Civil and Public Services Association even went so far as to state that their membership was ‘unwilling to sacrifice … any contraction of incremental scales in order to help those at the lower levels, essentially because many of those at the top felt low paid themselves’ (MRC, MSS.292B/115.4/3: Low pay: further appraisal in the light of union comments, 10 June 1970). Under these circumstances the TUC felt that the only effect of the introduction of a statutory national minimum wage would be to increase inflation.¹⁵ The Labour government agreed and withdrew its statutory wage proposals when the TUC publicly voiced its opposition to it.

Those unions that opposed the statutory national minimum wage because of its redistributive consequences naturally also resisted granting higher than average wage increases to the lowest paid. In addition, they resisted measures aimed at compressing the whole range of the wage distribution. This became clear in the early 1970s, when consecutive Conservative and Labour governments once more pressed unions to accept coordinated wage bargaining in order to limit wage growth. It is worth investigating this period – which was characterized by a degree of wage bargaining centralization that was higher than anywhere else in Europe at the time – for at least two reasons. First, it confirms that long-term centralization of wage bargaining is impossible to achieve in the presence of strong occupational unionism. Second, it confirms that the ability of these unions to preserve wage differentials resulted not from their numerical strength, but from their economic power.

The five years ranging from 1973 to 1977 have been the one period in the United Kingdom where it proved possible to reach national wage settlements with a clear redistributive element. Known as the years of the ‘social contract,’ wage settlements in each of these years contained (various combinations of) minimum pay rates, flat-rate increases, or maximum ceilings. To explain the creation of the social contract it is imperative to realize that the United Kingdom at the time went through one of the worst economic crises in its history. The crisis was so severe that the British government in 1976 had to approach the International Monetary Fund for a loan to reach its debt obligations. Under these circumstances, it proved possible to suspend the normal wage bargaining machinery for a couple of years. After serious threats from the incumbent Labour government that it might otherwise impose such a policy by statutory means, a majority of TUC members gave its official support to a national wage policy aimed at achieving wage restraint in 1974. That this policy included strong redistributive elements can be attributed to the numeric power of the general union movement. That these hardly affected existing differentials can in turn be attributed to the economic power of Britain’s occupationally organized unions.

Perhaps the most remarkable feature of the social contract is that it no lasting effect on the wage structure. Nearly five years of wage settlements with minimum pay rates, flat-rate wage increases and maximum ceilings on pay rates neither compressed the manual wage structure nor substantially affected the manual–non-manual wage differential. On the contrary, and as illustrated by Table 1, in the years immediately before and after the social contract, the wage gap between workers in the lowest and highest deciles of the manual distribution had only increased. Moreover, by 1982 top earners actually received a larger share of the manual distribution than they did almost a hundred years ago. To contemporaries, both the long-term stability of the manual wage differential and lack of impact of the social contract were ‘startling’ (Mayhew, 1981, p. 92). One of the reasons for this lay in the strong contrast with the situation in most other western European countries, where occupational and skill differentials did decrease substantially in the postwar period (Forster & Vleminckx, 2004; Hibbs, 1991; OECD, 1996). In nearly all of these countries this process accelerated between the late 1960s and early 1980s, although to a higher degree in some countries than in others. For reasons that will be outlined below, the reduction in occupational and skill differentials would be particularly severe in the Netherlands.

Table 1. Dispersion of weekly earnings of full-time male manual workers in the United Kingdom, 1886–1980.

Year	% median lowest decile	% median lowest quartile	% median upper quartile	% median highest decile
1886	68.6	82.8	121.7	143.1
1906	66.5	79.5	126.7	156.8
1938	67.7	82.1	121.7	145.2
1960	70.6	82.6	121.7	145.2
1970	67.3	81.1	122.3	147.5
1982	68.3	81.8	123.5	152.6

Sources: Pond, (1981, pp. 1–10); Pond and Winyard, (1984, p. 15).

So how can we explain the lack of impact of the infamous social contact? We cannot do so by emphasizing the numerical strength of Britain’s occupational unions. After all, craft unions representing skilled manual workers and occupational white-collar unions by the 1970s represented less than a third of the unionized workforce (Ebbinghaus & Waddington, 2000, p. 739). As a result, they were strongly outvoted at annual meetings of the TUC Congress during the first years of the social contract.¹⁶ Yet, beyond limiting the overall scale of compression by voicing their strong opposition to it, they managed to maintain existing differentials in two ways: first, by negating national bargaining outcomes through substantial wage drift; and second, by simply withdrawing their support for the social contract when dissatisfaction with centralized bargaining reached a boiling point, and subsequently making up for any lost differentials through decentralized bargaining.

Wage drift certainly explains much of the ability of occupationally organized unions to maintain wage differentials. While a phenomenon that occurs in all countries with centralized wage settlements, it arguably was much more severe in the United Kingdom where actual wage increases in one year were nearly 26% higher than the nationally agreed rate (Mayhew, 1981, p. 80). Such rates obviously left much room for a restoration of wage differentials. There is no doubt that wage drift was so substantial under the social contract because occupationally organized unions aggressively pressed employers to maintain wage differentials (Brown, 1976). One reason why they were able to pressure employers into doing so is that the skill profile of their membership gave them substantial bargaining leverage. What also helped was that the social contract was only loosely policed. Although formally enforced by law, there was hardly any oversight (Dean, 1981, p. 60). Under such circumstances, it is not surprising that national agreed rates were not maintained on a sectoral or firm level. A good example of this is the fate of what was arguably the most controversial wage settlement under the social contract, the so-called ‘six pound pay policy.’ Enacted in 1975, this policy was strongly resisted by the craft unions. That they were defeated on this in the General Council hardly seems to have mattered in the end, as the policy was so loosely policed that in many sectors it simply meant that the lowest paid workers received 6 pounds while higher paid workers received as much as 16 pounds (Ashenfelter & Layard, 1983, p. 130).

It is hard to say to what degree wage drift enabled the occupationally organized union movement to maintain wage differentials exactly. What is clear, however, is that any lost relativities were restored when the social contract eventually broke down. This breakdown neatly illustrates a major problem with the view that wage bargaining centralization reduces wage inequality because it empowers lower paid workers. The problem here is that unions representing higher paid workers are likely to withdraw their support for centralized bargaining when this is the case. Craft union representatives to the TUC General Council had made this painfully clear at the start of the social contract, by warning that ‘if the flat-rate system was adopted it would finish up in a wage explosion’ (MRC, MSS.292D/20/8: General Council minutes, 21 May 1975, p. 101). And indeed, this is what eventually happened. Mirroring events in the 1940s, after a few years of redistributive wage settlements the annual TUC Congress in 1976 first adopted a contradictory resolution that called for ‘the recognition and payment of satisfactory differentials’ while ‘assist[ing] workers on low pay’ (TUC, 1976, p. 637). And like in the 1940s, the return to percentage increases in the following year did not prevent occupationally organized unions from withdrawing their support for centralized bargaining and seeking to restore lost differentials. Contemporaries viewed this as the major reason for the wage explosions that eventually came about during the infamous 1978–1979 ‘winter of discontent,’ describing it as the outcome of a ‘harsh internecine conflict within the working class itself’ (Taylor, 1982, p. 207).

When numbers matter: the representational monopoly of industrial unionism

The events that took place in the United Kingdom during the 1960s and 1970s stood in stark contrast to the Netherlands, where the three main union federations had by the mid-1960s succeeded in completing their transition toward full industrial unity, and had done so by disbanding their last white-collar affiliations and reallocating their member over various sectoral organized affiliates. This enabled them to adopt a much broader redistributive agenda. As a result, flat-rate wage increases and maximum ceilings became the norm in the Netherlands from the late 1960s on to the early 1980s when the rise in unemployment enabled employers to put a halt to this. The consequences of this for the wage distribution were immense.¹⁷ By the early 1980s the Netherlands no longer mainly distinguished itself from the United Kingdom through a much more compressed lower half of the wage distribution; it had also managed to compress the wages of above-average earners in a much more robust manner.¹⁸ While receiving unequivocal support from the union members of the three federations, many higher paid workers strongly resented this. And many of these were union members. Contemporary observers therefore predicted that the redistributive agenda of the big three would inevitably lead to a surge in independent, occupationally organized, unionism (Windmuller, 1969, p. 170). Yet this surge never took place. Although the events of the 1970s do seem to have benefited independent unions somewhat and even led to the creation of a separate union federation for higher paid white-collar workers, there never was a massive move toward occupational unionism in the Netherlands.¹⁹ Tellingly, while aware that there might be a relationship between their support for wage compression and the popularity of independent unionism among white-collar workers at the time (NVV, 1970, 1975, p. 75), the big three never felt the need to stop pushing for flat-rate wage increases and maximum ceilings on pay rises.

They had an excellent reason for this. After all, they had only recently completed a long process of breaking up their occupationally organized white-collar union affiliates and reclassifying the members of these unions as members of their industrial union affiliates. This process had met with much resistance, forcing the NVV to expel at least one of its affiliates during the 1960s (Hueting et al., 1982, pp. 196, 207; Reinalda, 1985, pp. 178–179). Despite some fears about this at the time, this forced reorganization of the labor union movement had hardly led to a surge in independent unionism. How can we explain this? Much of this seems to have to do with the self-reinforcing nature of industrial unionism. Long before adopting a broad redistributive agenda, the three union federations had obtained such a large share of the total union membership that they had become labor’s exclusive voice on all issues related to national wage bargaining, general economic policy, and social security implementation. This large share itself, as well as the industrial nature of wage bargaining and social security implementation, made them the natural choice of most workers who considered joining a labor union. Well aware of this, the big three in the late 1960s boasted that the industrial nature of labor relations was now so strongly entrenched that the labor union movement itself could hardly influence this (IISG, Archief NVV, Doos 742, Concept-rapport, 23 October 1967, p. 84). Naturally, this did not stop them from jealously protecting the industrial nature of labor relations in the Netherlands, through activities for which they received full support from employers (Hueting et al., 1982, pp. 305–310).

Perhaps the latter tells us most about the robustness of industrial unionism. There is no doubt that employers strongly disliked the broad redistributive agenda adopted by the three union federations in the late 1960s. Their resistance to this seriously frustrated wage negotiations during the 1970s, resulting in lengthy negotiations, several failures to reach central wage agreements (thus pushing wage bargaining to the industrial level), and much bickering over their interpretation. In sum, during the 1970s, and to a large extent due to conflicts over wage relativities, the relationship between the industrial union movement and employers was more discordant than ever before – and probably ever since. Yet crucially, this never stopped employers from supporting the industrial nature of labor unionism in the Netherlands. To do so would have meant to break with a long history of employer support for the creation of a sectoral organized social security system, a neo-corporatist system of decision-making that was based on cooperation with the main employer federations, and occasional non-recognition of independent unions. Employers greatly valued the relatively harmonious system of labor relations that followed from this, and continued to do so when the relationship with the industrial labor movement became more dissonant during the 1970s. As a result, they never gave their full support to workers who wanted to break with the sectoral nature of labor unionism. For instance, when associations for higher personnel approached them for financial assistance to set up independent occupational unions, employers opposed this because ‘existing [industrial] unions never received support either’ (VNO-NCW, AB 74-131: Vergadering Algemeen Bestuur VNO, 10 September 1974, p. 13). For administrative reasons, they also supported union demands for integration of manual and white-collar agreements (IISG, Archief NVV, doos 769, Concept-notitie inzake de integratie van cao’s, 23 February 1967). And while resisting such integration for the highest paid managerial workers, they at the same time resisted pressure from these workers to bargain separately from industrial unions. They motivated this by emphasizing their appreciation of the ‘harmonious nature of negotiations’ in the Netherlands. In doing so, they acknowledged the need to ‘adjust higher and lower wages’ (VNO-NCW, AB 74-115: Vervolgnota werkgevers en hoger personeel, 1974, pp. 1–4). As a result, they indirectly came to support the redistributive agenda of the industrial union movement.

The consequences of union structure

The preceding analysis of British and Dutch labor union wage bargaining behavior confirms the crucial role of labor’s internal organizational blueprint in shaping union views of occupational wage relativities. It did so by capturing how this affected the willingness and ability of labor unions to pursue wage compression policies during roughly the first four decades of the postwar period. Such policies were to become much less common in subsequent decades. Even in the Netherlands, the union movement dropped its demands for across-the-board wage compression in the 1980s when concerns over inequality slowly gave way to distress over soaring unemployment and inactivity rates. This greatly facilitated the coming about of a new peak agreement in 1982, paving the way for a new period of prolonged wage restraint that lasted until far in the 1990s (Visser & Hemerijck, 1997). During these years, the industrial union movement no longer pushed for further compression, but focused on making sure that the decentralization of wage bargaining and government measures that for instance undermined the relative value of the statutory national minimum wage did not work to increase wage inequality in future years. Its successful ability to preserve the country’s exceptionally dense wage structure and prevent low pay from increasing testifies of its enduring ability to display intra-class solidarity. In the remainder of this article, I will explore the consequences of my findings for the scholarly preoccupation with the level of wage-setting and labor union density as determinants of wage inequality. In addition, I will focus on the scholarly tendency to assume labor union support for redistributive government intervention in the labor market and, consequently, to assume that higher levels of union organization increase the scope for such intervention.

Does wage bargaining centralization reduce wage inequality?

According to a large and prominent literature, the level of wage setting stands out as a major explanation for cross-national and inter-temporal variation in wage inequality. This claim is based on several macro-economic studies which have found that more centralized levels of wage bargaining are generally associated with lower levels of wage inequality. These findings have widely been interpreted as implying that ‘more centralized bargaining institutions strongly encourage wage equality’ by suppressing occupational, regional, and firm-specific wage differentials (Golden & Londregan, 2006, p. 208). Exactly why the presence of centralized wage bargaining should reduce wage differentials, especially among workers of different skills and occupations, remains unclear though (Oliver, 2008, p. 1554). The preceding analysis already criticized the most often used explanation for this, which is that centralized bargaining strengthens the relative bargaining position of low-paid unions. It did so by emphasizing that coordinated wage bargaining can only empower low-paid unions if bargaining at the central level was to resemble negotiations within a single union – a state of affairs that, for reasons explained above, is unlikely to come about under conditions of occupational unionism. The analysis showed that Britain’s occupationally organized unions were both able and willing to withdraw their support for peak bargaining when this interfered with their goal of maintaining occupational wage relativities. It also showed that they subsequently found it quite easy to make up for any lost differentials.

An important problem with existing claims that centralized wage bargaining suppresses wage differentials is that they are almost exclusively based on quantitative studies.²⁰ The findings of this study suggest that wage bargaining institutions interact with wage inequality in a way that is much less straightforward than these studies assert. They also suggest that union structure plays an important intermediary role in this. Both the British and Dutch cases suggest that centralized wage bargaining is most likely to emerge under economic conditions that call for wage moderation. Under these circumstances employers and governments may demand, and unions may be willing to accept, wage bargaining coordination at the central level in order to reduce the inflationary consequences of wage competition between unions. The wage restraint policies that result from this will likely affect the purchasing power of the lowest paid the most, fueling demands for countermeasures that will inevitably have redistributive consequences. Wage bargaining centralization is consequently most likely to be successful under conditions of industrial unionism, as this type of unionism makes organized labor opposition to such measures – and inter-union conflict in general – much less likely. When the industrial union movement adopts a broader redistributive agenda it may actually be more difficult to reach wage agreements at the central level though, as organized employers will strongly resist this. The increasingly radical redistributive agenda of the Dutch union federations, as we have seen, clearly undermined centralized bargaining during the 1970s. In Sweden too, the collapse of central wage bargaining during the 1980s has often been attributed to the increasingly broad redistributive agenda of the blue-collar Landsorganisationen I Sverige (LO) (Swenson & Pontusson, 2000, p. 85).

Does union density explain wage inequality?

An almost equally popular claim of recent macro-comparative studies on wage inequality is that high levels of labor union density tend to have a leveling effect on the broad range of the wage distribution. Much of the theoretical reasoning behind this claim is equally unconvincing. While sometimes partly based on the view that unions generally favor standardization of wage rates across and within firms, a more commonly invoked argument is that unions also seek to reduce differentials among different categories of workers.²¹ The latter explanation inevitably boils down to flawed median voter thinking. The previous analysis criticized this line of thinking in various ways, and emphasized that union support for wage compression depends on its organizational blueprint. In addition, it is possible to cast doubt on the view that unions generally push for standardization of wage rates across and within firms. Here too, it may matter a great deal whether unions organize and bargain along broad-sectoral or occupational lines. In the former case, the notion of ‘equal pay for equal work’ will for reasons of both equity and survival likely have strong appeal among union negotiators. Equally important, it will be relatively easy to implement. In the latter case, however, numerous craft and occupationally organized white-collar unions are likely to bargain at the plant-level, making pay standardization both less salient for union negotiators and more difficult to achieve. In sum, a closer look at these two explanations for the importance of union density suggests that the effect of unionization on the wage distribution depends greatly on the internal organizational blueprint of unions.

That labor’s organizational blueprint matters greatly to the way in which union organization affects wage inequality can be illustrated in another way as well. In recent years, scholars have frequently pointed to union decline as an explanation for recent increases in wage inequality in some countries (Machin, 2008; Mosher, 2007). Yet whether such decline has an actual effect on the wage distribution obviously depends greatly on whether it results in lower collective bargaining rates. In the Netherlands, for instance, the sharp decrease in union density levels since the 1980s has hardly affected the wage distribution because statutory extension there has made sure that collective bargaining rates remained relatively stable at about 85%. This has enabled the unions to preserve the relative wages of most low-paid workers, despite the steep decline in the level of the statutory national minimum wage since the mid-1980s. At the same time, they even managed to slightly reduce the income of top earners further. In the United Kingdom, by contrast, a roughly equally sharp decrease in union density levels has been associated with a much higher increase in wage inequality.²² There is no doubt that this increase would have been much less severe if the United Kingdom would have had similar statutory extension laws. Yet crucially, such laws are unlikely to come into force under conditions of occupational unionism. The untidy and competitive inter-union environment that this creates not only makes such laws difficult to implement; it also makes sure that a large part of the union movement may have little interest in them.²³

Do strong unions increase the scope for redistribution?

The tendency to focus on the degree to which – and to neglect the way in which – workers organize in labor unions is also apparent in the broader literature on labor market development. Above, I already noted the strong tendency in the literature on labor market development to view unions as key allies of left parties in their pursuit of a redistributive agenda. While sometimes based on simple median voter thinking, this tendency also reflects the longstanding preoccupation of much of this literature with worker strength and unity against capital (Oude Nijhuis, 2009). The preceding analysis criticized this tendency by pointing out that when unions oppose wage settlements that involve wage compression, they will also oppose statutory intervention that results in a reduction in wage differentials. It then illustrated how concerns over its possible adverse effect on existing differentials motivated British occupationally organized unions to oppose the introduction of a statutory national minimum wage during the 1960s. These unions continued to do so in following decades – thereby resisting the TUC General Council which, supported by a majority of members, grew more appreciative of statutory intervention in the mid-1980s. They among others did so by warning that they would seek to reverse any resulting reductions in wage differentials that their members might experience. Such warnings were voiced by the powerful Amalgamated Engineering and Electrical Union in as late as 1995 (Dorey, 2001, p. 222). There is no doubt that this threat still had a potent effect on policy-makers. When the statutory national minimum wage finally came into force in 1999, it was set at a much lower level than promised by Labour when it was still in opposition. This drew many protests from unions representing lower paid workers. In response, the government argued that the level had to be determined in a way that took account of wider economic circumstances such as the income distribution – or in other words, wage differentials (Metcalf, 1999, p. 173).

Conclusion

One of the main problems confronting labor unions during wage bargaining is how to deal with the conflicting demands of different groups of workers over the division of labor market earnings. These conflicting demands constitute a considerable internal challenge for unions that require them to manage conflicts within the working class. The preceding analysis illustrated that unions do not simply deal with such conflicts by acting as utility maximizers. Instead, unions deal with such conflicts in quite different ways, with union policies toward wage relativities reflecting their internal organizational blueprint. This blueprint not only shapes union views of what constitutes a ‘fair’ division of labor market earnings; it also matters greatly to their ability to embark on policies that aim to alter the wage distribution. The preceding analysis illustrated this in two major ways. It first showed how the respective absence and presence of craft unionism explains the success and failure of attempts to improve the relative wages of the lowest paid in the Netherlands and the United Kingdom. It then illustrated how the disappearance of occupational unionism among white-collar workers in the Netherlands enabled the union movement there to adopt a much broader redistributive agenda.

These findings have consequences for our understanding of labor union wage bargaining behavior and the wider literature on wage inequality and labor market development that far exceed the scope of this study. Although the main preoccupation of this article has been with explaining labor union wage bargaining behavior, it has on several occasions pointed to the consequences of its findings for the wage inequality literature – thereby strongly criticizing the tendency to argue that high union density levels and centralized wage bargaining ‘lead’ to a more equal wage distribution. In addition, the analysis described at length how labor’s internal organizational blueprint matters not just to unions’ willingness and ability to pursue wage compression policies, but also to the scope for redistributive government in the labor market, incomes policies aimed at achieving wage restraint, and centralized wage bargaining. To further explore how union organization interacts with these labor market policies and institutions, it is crucial to fully appreciate the difficult role of unions as ‘conflict mediators’ between different categories of workers.

The importance of labor’s organizational blueprint for this is perhaps best illustrated by the example of the skilled manual worker. Historically the organizational backbone of labor union movements in most advanced industrial countries, skilled manual workers can either organize in craft or in broad-industrial unions. In the United Kingdom, skilled manual workers have long preferred to organize in craft unions. Considering their location in the wage distribution, many scholars would probably argue that such workers stand to gain from flat-rate benefits that reduce the overall wage distribution. Yet because of their preoccupation with the manual wage differential, British unions representing skilled manual workers were among the staunchest opponents of flat-rate wage increases. In the Netherlands, by contrast, skilled manual workers have historically preferred to organize in industrial unions. These unions not only strongly supported flat-rate wage increases; they also repeatedly pushed for percentage increases combined with minimum pay rates and redistributive government intervention that benefited a small percentage of workers at the expense of all other workers. The latter, more than anything, illustrates that union wage bargaining behavior cannot be understood through models that emphasize majoritarian interests. Unions evidently also live in a moral economy that is shaped by normative orientations that emphasize fairness and solidarity. How such norms shape labor union policies crucially depends on their organizational blueprint though.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Dennie Oude Nijhuis is assistant professor in Economic and Social History at Leiden University and senior researcher at the International Institute for Social History (IISH) in Amsterdam. Oude Nijhuis specializes in the comparative political economy of labor markets and welfare states and the political economy of European integration. He is the author of Labor Divided in the Postwar European Welfare State (Cambridge University Press, 2013) and Christian Democracy, Labor and the Postwar Development of the Dutch Welfare State (Amsterdam University Press, 2018). His current main research project involves a comparative analysis of twentieth century labor market development in Belgium, Germany, the Netherlands, Sweden, the United Kingdom, and the United States.

Notes

1. Arguing that such workers have an insurance incentive to remain part of such unions, economists have tackled with these questions (e.g. Agell & Lommerud, 1992; Horn & Wolinsky, 1988).

2. Its use for this purpose can be seen as surprising given the noted failure of median voter-theory to explain variation in government redistribution (Moene & Wallerstein, 2001).

3. In fact, up to the 1970s it was a commonly held view that unions increase wage inequality because membership concentrated among workers in the upper half of the earnings distribution (e.g. Johnson, 1975; Millward, Forth, & Bryson, 2001; Rees, 1962; Rosen, 1970).

4. The history of the evolution of unionism in the Netherlands suggests that both played a role, with both socialist and confessional unions emphasizing broad worker solidarity, and socialist unions also emphasizing labor unity against employers (e.g. Hueting, de Jong, & Neij, 1982).

5. Some independent unions also organize various skill and income levels. The long-lasting struggle between white-collar unions and the largely industrially organized union federations over the organizational features of the Dutch union movement has been described in detail (e.g. Hueting et al., 1982).

6. Labor historians have long illustrated how this resistance prevented the coming about of a more centralized union federation in the United Kingdom in the pre-war period (e.g. Fulcher, 1980).

7. For a list of industries in which flat-rate increases were common during the first two years of the War, see MRC, MSS.292/110/1: TUC research department, War additions or cost of living sliding-scales introduced during the War, 18 November 1940; Ibidem, Wage increases since the War in industries not covered by cost-of-living sliding scales, 1 August 1941.

8. For events in the building and engineering industries, see MRC, MSS.292/110/2: Proposed increase in basic rates of wages and introduction of incentive pay system, 1 September 1947; Ibidem, Memorandum on the need for a new wages structure in the engineering industry, 1 July 1945.

9. According to the NVV, wages for unskilled, semi-skilled and skilled manual workers had increased by respectively 22, 16 and 15% between 1940 and 1942. It did not have exact numbers for later years. IISG, Archief NVV, doos 21, Notulen Hoofdbesturenvergadering, 19 July 1945.

10. There was a broad consensus on the need to raise the lowest wage levels (Mulder, 1956, p. 22).

11. Another exception was the year 1946 when there was a general wage freeze that did not include specific measures for the low paid (van Hulst, 1984, pp. 258–260).

12. In 1949, for example, the NVV initially wanted to push for a substantially higher minimum rate than the 2 guilders it eventually demanded. Its executive committee motivated this by arguing that a higher minimum rate would increase labor market tensions. Crucially, however, the minimum of 2 guilders meant that the lowest paid still received a much higher percentage increase than all other workers. IISG, Archief NVV, doos 11, Notulen vijfde vergadering Verbondsbestuur, 12 December 1949, p. 57.

13. Some white-collar unions also opposed increasing the relative wages of the low paid. In the late 1940s, for example, union federation efforts to create a minimum salary rate for office personnel were opposed by KAB-member St. Franciscus because of its effects on wage differentials (Reinalda, 1985, p. 6).

14. For instance, in 1949 and 1950 the three union federations opposed government proposals for maximum ceilings, wanting to set these at substantially higher levels than offered by the government. At the same time, they wanted minimum rates that were far more generous than proposed by the government. ISSG, Archief NVV, doos 312: Kort verslag hoofdbesturenvergadering van 9 November 1950, p. 1; Ibidem, Kort verslag NVV-looncommissie, 2 September 1950, p. 7.

15. On the TUC’s realization that craft unions would immediately seek to restore wage differentials and that this would lead to inflation as there is an overall limit on what firms can afford to pay without raising prices, see MRC, MSS.292B/115.4/4: LPWP, restructuring incomes and jobs, 12 November 1969; MRC, MSS.292B/116/1: LPWP 4/3, 2 December 1969.

16. For instance, during a vote on the so-called 6 lb policy, 19 unions voted in favor, 13 voted against and 6 abstained. Tellingly, nearly all general union representatives voted in favor and nearly all of craft and occupational white-collar representatives voted against (Dorey, 2001, p. 149).

17. To illustrate, between 1970 and 1980, the level of the net minimum wage increased from 71 to 85% of median wage. By comparison, someone earning four times the gross median wage in 1975 had enjoyed net earnings of 309% of the median wage in 1970 and only 245% in 1980 (Centraal Planbureau, 1981, p. 147).

18. In the mid-1980s, the 90–50 and 50–10 differentials were respectively 1.94 and 1.86 in the United Kingdom vs. 1.84 and 1.58 in the Netherlands (Alber, 2006, p. 206).

19. From 85% in the 1960s, the amount of organized workers represented by the ‘big three’ declined to 80% in the 1990s. Moreover, a part of the small increase in independent unionism can be attributed to ‘natural growth’ resulting from the relative growth of white-collar occupations (Ruysseveldt & Visser, 1996, p. 228).

20. An exception to this is the coming about of centralized bargaining in Sweden, which did reinforce the bargaining power of low paid unions. Yet it could arguably only do so on a long-term basis because employers there wanted to reduce inter-industry differentials and consequently pushed for centralized bargaining (Swenson, 1989).

21. Whereas most economists initially believed that labor unions increased wage inequality, arguing that union membership concentrated among members in the upper half of the wage distribution and that unions increased the wage gap between union and non-union members (an outcome that could not occur in countries that allowed for statutory extension of collective bargaining outcomes), during the 1980s a growing number of economists came to argue that the wage standardization effect of unions dominated (e.g. Freeman, 1980; Pontusson et al., 2002, p. 356). The tendency to argue that labor unions generally bring a redistributive agenda to the bargaining table that includes compression of occupational and skill differentials is a more recent phenomenon though.

22. Between 1980 and 2000, the 90–50 differential decreased from 1.86 to 1.67 in the Netherlands while it increased from 1.95 to 2.13 in the UK (Alber, 2006, p. 406).

23. Tellingly, when the issue of statutory extension was debated in the United Kingdom in the 1930s, the TUC displayed little interest. Of much importance to this was likely that its powerful craft members did not have to worry about employers undercutting wages by attracting unorganized workers – a problem that greatly worried the industrial union movements elsewhere. For events in the United Kingdom during the 1930s (Charles, 1973).