Trade union growth

Summary: Dutch trade unions have seen a number of periods of fast membership growth. It has been suggested that such periods are the result of disruptive protests during periods of uncertainty. An analysis of the historical context of periods of trade union growth in the Netherlands suggests that strikes often played a role.

Dutch trade unions enjoy broad support among workers,¹ but the share of employees who are members of a trade union is decreasing. This hasn’t always been the case, of course. In this article I’ll explore trade union growth.

The chart below shows the share of employees who are union members in the Netherlands.

Of course, if you squeeze a long period of time into the x-axis of a chart, ups and downs will automatically look a bit spiky. Even so, there have been periods that saw a substantial rise in union density over just a few years.

This pattern seems to be pretty widespread. Various authors have argued that unions don’t grow gradually, but in spurts. For example, Dan Clawson, discussing the labour movement in the US, wrote:²

Historically, labor has not grown slowly, a little bit each year. Most of the time unions are losing ground; once in a while labor takes off. From the mid-1930s on, in the stretch of little more than a decade, the number of union members increased fourfold.

The chart below shows wat he means:

And the chart below shows a similar pattern for the ADGB, the general union federation in pre-WWII Germany, and its precursors.

These charts suggest that sharp increases in union membership were often quickly followed by a decline, although the level after the peak was often higher than before.

Richard Freeman has analysed trade union membership in 18 countries, including the Netherlands. Many countries saw ‘growth spurts’ during the first world war and in the period from the mid-1930s to 1945. He argues that such growth spurts are difficult to predict:³

Another lesson is that any such resurgence of unionism will come suddenly, probably surprising the current crop of experts and labour historians as much as the depression spurt surprised […] observers of the period.

Most growth spurts discussed by Freeman happened during the first half of the 20th century. To explore whether there have been instances of rapid trade union growth after the second world war, I checked Jelle Visser’s ICTWSS database for instances where union density rose by at least 10 percentage points over 5 years, either for the entire workforce or for women workers or public sector workers.

I’m not sure all of these examples are in fact ‘growth spurts’, but some do seem to qualify.

Explanations of sudden membership growth

As far as I know, there are broadly two types of explanations for ups and downs in union membership. One uses time-series analysis to find associations between union membership and economic factors; the other looks for disruptive events that precede periods of rapid trade union growth. The two approaches aren’t mutually exclusive.⁴

I’ll discuss an example of a time-series analysis below. First, I’ll discuss the view that periods of rapid union growth originate in disruptive events. Interpretations in this category often focus on developments in the US in the 1930s. This period saw a wave of strikes for union recognition, for example at textile mills, rubber plants, mines and car factories. Many of these actions were sit-down strikes, which are basically factory occupations.

To contain labour unrest, the federal government created a mechanism for union recognition through elections. Sit-down strikes continued, including the historical confrontation at the GM factory in Flint, Michigan. However, labour relations became increasingly institutionalised and eventually sit-down strikes were banned.

In their book Poor people’s movements, Francis Fox Piven and Richard Cloward argue that disruptive protests during a period of uncertainty resulted in concessions that allowed unions to grow:⁵

For the most part strikes, demonstrations and sit-downs spread during the mid-1930s despite existing unions rather than because of them. Since these disorders occurred at a time of widespread political instability, threatened political leaders were forced to respond with placating concessions. One of these concessions was protection by government of the right to organize. Afterwards, union membership rose, largely because government supported unionization.

When Piven and Cloward say that protests happened ‘despite existing unions’, they probably refer primarily to the craft unions affiliated to the AFL. The strike wave of the 1930s coincided with the rise of a new type of industrial unions, such as the United Automobile Workers (UAW). Piven and Cloward argue that strikes like the one in Flint were the combined result of UAW organising and more or less spontaneous protests.

Some examples from other countries seem to fit the pattern Piven and Cloward describe. For example, in Italy, union density rose from about 30% to about 50% after a new law on workers’ rights was introduced in 1970. The new legislation had been triggered by the autunno caldo or hot autumn of 1969. This period saw widespread protests and spontaneous strikes, most prominently at the FIAT factories in Turin.

In Germany, union membership exploded after employers’ organisations had recognised trade unions at the end of the first world war. At an earlier stage, unions had announced they would not strike during wartime. However, local, more or less spontaneous protests grew into a nationwide strike wave, involving millions of workers.⁶ After the November revolution of 1918, the employers’ organisations reached an agreement with the trade unions.

The use of the term ‘spontaneous’ here needs some clarification. The term ‘spontaneous strike’ is often used more or less as an equivalent of wildcat strike, that is, a strike not led by a union. However, most strikes probably aren’t entirely spontaneous; there’s often some form of informal organisation behind them.⁷

Union growth in the Netherlands

The chart above shows percentage change in trade union membership in the Netherlands.⁸ Instances of fast growth occurred in 1919, around 1930, and immediately after the second world war.⁹ Below, I’ll explore whether these periods were associated with economic phenomena. Subsequently, I’ll discuss whether they were preceded by disruptive events.

Economic factors

Annette van den Berg has done a time-series analysis of trade union growth in the Netherlands.¹⁰ The chart below shows how the variables in her model contributed to the predicted change in union membership before the second world war.

The model suggests that growth in 1919 was associated with rising wages, whereas strikes occurring in that same year contributed to the fall in membership in 1920. In other words, it appears that unions were militant and successful in 1919, but faced a backlash in the subsequent year.

Van den Berg’s model doesn’t contain a prediction for 1946, but her analysis does suggest that membership growth immediately after the second world war was associated with strike activity.¹¹

Growth in 1929 remains a bit of a mystery.¹²

Growth in 1931 appears to be associated with unemployment. Van den Berg’s model assumes that unemployment has an initial positive effect on union membership, which is subsequently reversed by a larger negative effect. This makes sense given the way unemployment insurance worked at the time.

Normally, you’d expect that unemployment has a negative effect on union membership, since it weakens the position of workers and makes it easier for employers to oppose organising. However, the Netherlands had the so-called Ghent system at the time, which means that trade unions managed government-subsidised unemployment funds.

This explains why rising unemployment initially would give workers an incentive to join a union. However, unemployed workers would only receive payments for a period of 6 or 13 weeks.¹³ This might help explain why the positive effect of unemployment on membership was only temporary, and followed by a larger (lagged) negative effect.

Strikes

If you assume that union growth is related to disruptive events, then wildcat strikes may be more relevant than strikes that are led by unions. The chart below distinguishes between three types of strikes: wildcat strikes, strikes led by unions, and strikes where the role of unions is unknown.¹⁴

There have been peaks in strike activity in 1913, 1919, and 1946. The latter two coincide with periods of strong membership growth (labour unrest in 1913 could be relevant to the introduction of the Ghent system in 1914, as will be discussed below). Simple correlation analysis suggests that union-led strikes might be less relevant for membership growth than other types of strikes.¹⁵

The timing is interesting. Piven and Cloward suggest that disruptive protests lead to concessions, which in turn facilitate membership growth. More specifically, they refer to changes in labour law, which may take months to be implemented and take effect. You’d expect some time to pass between labour unrest and membership growth.

However, the data suggests that in the specific cases discussed here, membership growth more or less coincided with labour unrest. In other words, it appears that people started joining unions before any institutional changes might be implemented.

So why do people join unions during labour unrest? A reason might be that unions compensate their members for the wages they lose during a strike. However, this consideration seems less relevant in the case of wildcat strikes, where unions will often be hesitant to open their strike funds.

On the other hand, it has often been observed that protests arise when people perceive a situation as unjust, but at the same time are optimistic that it can be changed.¹⁶ Such a combination of indignation and optimism may well be a motivation to join a union as well.

Historical context

Below, I’ll look into the historical context of periods of rapid trade union growth in the Netherlands. I’ll start with the strikes of 1903, which were a defining moment in the history of the Dutch trade union movement.

1903

In January 1903, a strike of dock workers in Amsterdam triggered a solidarity strike of railway workers. Within days, the spontaneous action grew into a national railway strike. This came as a complete surprise, not least to the railway workers’ union NV.¹⁷ The chairman of this union had said earlier that it would take at least 25 years before railway workers would be ready to go on strike.¹⁸

The strike was succesful: all demands were met, including recognition of the NV.

Elites were shocked by the strike and its outcome, and spoke of a ‘coup d’état’ that would likely spread to other sectors.¹⁹ That fear may have been justified. February saw a increase in strike activity. Most of these strikes were spontaneous.²⁰

In response to the railway strike, the government prepared new legislation, including a ban on strikes by public servants and railway workers. Unions called a general strike to protest against these measures, but this strike was a failure. In retaliation, thousands of railway workers were laid off and left without income. Initiatives to support the victims of the strike continued until the 1930s.²¹ Meanwhile, a Parliamentary committee was set up to investigate the labour conditions of railway workers, which resulted in some improvements.

As a result of the first strike, the membership of the railway workers’ union NV grew from 4 thousand to 12 thousand.²² My great-grandfather, who was stationed at the Utrecht union headquarters at the time, «almost couldn’t find enough paper to sign up the new members». After the second strike though, the union was ‘smashed to pieces’. Members left for fear of retaliation, and by 1904, only about 1,200 were left.²³ The syndicalist NAS followed a similar pattern: initially their membership rose from 10.5 thousand to 17 thousand, but after the second strike it dropped to just 8 thousand.²⁴

However, some unions emerged stronger from the turmoil. This applied to the Christian trade unions, which had rejected the strikes; but it also applied to the diamond workers’ union ANDB, which had participated in the general strike.²⁵ The ANDB was a ‘modern’ union modelled after trade unions in England. It emphasised that members must follow the instructions of the union leadership, also in matters relating to strikes. The ANDB would be the main force behind the creation of the union federation NVV in 1906.

1919

During the first world war, the Netherlands saw growing labour unrest. As the number of strikes increased, membership growth accelerated. While the regular unions were often not involved in the strikes, they did engage in other forms of protest and the NVV did at some point consider organising a general strike. There was some concern among elites that a revolution might take place, as happened in Russia (1917) and Germany (1918). The mayor of Rotterdam even called a meeting with leaders of the workers’ movement to discuss a peaceful transition of power.²⁶

Anxious to show that there was no need for a revolution, the government quickly introduced a range of social measures, including the 8-hour working day; the suffrage for women; subsidies for affordable housing; and provisions for workers with disabilities and retired workers.²⁷ Unions had fought for some of these measures for years.

The mood changed in 1920, when economic growth came to an end. Employers and the government launched an offensive to reverse some of the concessions of the past years. Union membership declined.

1931

Van den Berg’s model suggests that unemployment caused a rise in union membership in 1931, but subsequently slowed down growth. This pattern can be explained by the so-called Ghent system, which had been introduced in 1914. One might wonder why it took so long for this system to have an impact on union membership. A look at unemployment figures can explain this: unemployment was high in 1914, but much higher in 1931.

Arguably, if you want to explain union growth in 1931, you’ll have to explain what led the government to introduce the Ghent system in 1914. Strike activity increased from 1909 until the start of the war. While many strikes were controlled by the modern unions and funded from the strike funds they had started to build, some strikes had a more spontaneous and chaotic character.²⁸ Still, it doesn’t appear that strikes were a real cause of panic for employers and the government.

The outbreak of the first world war had perhaps a bigger impact. The Netherlands remained neutral, but the outbreak of the war in 1914 created a brief panic, with people hiding coins and ‘storming’ grocery stores.²⁹ The government needed the help of the trade unions to maintain a ‘spirit of national unity’ and to maintain peace on the workfloor.³⁰

Meanwhile, unemployment was rising. Some unions had set up unemployment funds, but these were inadequate to meet with the rise in unemployment. In 1914, within weeks after the German invasion of Belgium, the national government asked municipalities to subsidise the unemployment funds of the unions, an arrangement that would later be expanded.³¹

1946

During the second world war, the Netherlands was occupied by Germany. A nazi named Hendrik Jan Woudenberg was put in charge of the NVV union federation. There were some notable instances of labour unrest during the war, but in most cases the established unions weren’t actively involved.

The first major strike was the February strike of 1941, which was a general strike against the persecution of Jews and against anti-Jewish measures (organisers also presented the strike as an opportunity to oust Woudenberg from the NVV). In April and May 1943, hundreds of thousands of workers participated in spontaneous strikes when Dutch soldiers were sent to Germany to do forced labour in the German war industry.³² Meanwhile, an underground union organisation was created, which was critical of the NVV. This organisation would later become the EVC.

The government in exile was concerned that the established unions no longer represented the Dutch workforce.³³ After the liberation, labour unrest grew. In 1945 and 1946, the number of strikes increased; some for decent living conditions, others for the removal of bosses who had collaborated with the nazis. Large numbers of workers participated in solidarity strikes, supporting workers in other industries or fellow workers across the border.³⁴

Meanwhile, the established unions and the employers’ organisations had laid the foundation for a new system of labour relations. Basically, unions gave up any claims for worker participation at the company level, in exchange for «a generous representation of the workers’ movement in the highest economic body that will assist the government».³⁵ Under the new system, wages were set at the national level and only minor increases were allowed. Usually, strikes weren’t supported by the recognised unions.

By 1945, the new union organisation EVC had seen rapid growth to over 160 thousand members. However, it was kept outside the new system of labour relations, partly because it wouldn’t renounce strikes. After a few years, it dwindled away. Meanwhile, employers had little reason to oppose membership in the recognised trade unions, which at the time had little involvement with the workplace and didn’t support strikes or demands for large wage increases. These unions saw substantial growth in 1946.

Discussion

Trade unions tend to grow during short bursts, rather than gradually. This phenomenon has been discussed in the US, but it also applies in other countries, including the Netherlands. Piven and Cloward argue that union growth is the result of concessions made by elites, in response to disruptive protests at a time of uncertainty.

In the Netherlands, peaks in strike activity in 1919 and 1946 coincided with rapid union growth. A disruptive strike in 1903 also triggered a rise in membership. Membership growth in 1931 is a bit more complicated. It seems likely that the Ghent system of unemployment insurance played a role. The introduction of this system in 1914 was perhaps partly motivated by a brief sense of panic at the outbreak of the first world war.

Before the second world war, periods of rapid growth were sometimes followed by a backlash from employers and the government, causing a subsequent decline in membership. Still, the level of union membership after the peak would generally be higher than it had been before the peak. This raises the question: what made membership gains increasingly sustainable?

It seems that there are two phases in membership effects related to disruptive protests. Initially, people probably join unions for more or less the same reason why they go on strike: a combination of indignation and optimism, and perhaps the perception that it’s possible to win higher wages. Subsequently, protests may lead to concessions that help consolidate the membership (or even allow further growth, as appears to have been the case in the US and Italy). These phases may affect different unions in different ways.³⁶ It would be interesting to further analyse the mechanisms that help consolidate union membership.

Another interesting question is whether unions shape the outcomes of disruptive events. Piven and Cloward see unionisation in the 1930s as an important victory for workers, that brought them higher wages and more job security for years to come. Still, they argue that the concessions won by social movements are circumscribed by historical circumstances: if workers had demanded something different than organising rights, they’d still have gotten organising rights, because that’s what elites were willing to give them at that point in time.³⁷

However, none of the examples discussed here suggest that unions got outcomes they didn’t want in the first place. The railway workers’ union wanted union recognition in 1903; Dutch unions had been advocating the Ghent system before it was introduced 1914 (although not unanimously); the 8-hour working day and voting rights had been union goals long before they became reality at the end of the first world war; and in the 1940s, a place at the table was what the unions wanted.

Of course, it’s impossible to tell what would have happened if unions had demanded something different. With that caveat, it does appear that these outcomes were in part determined by the demands that unions had made. This suggests that the strategic choices of unions matter, and that these choices can have a long-lasting impact on the position of workers.

Another question is whether growth spurts can happen again in the future. The events that triggered growth usually came as a complete surprise, not least to the unions themselves – that’s also why they could be perceived as so disruptive. This suggests that it’s probably impossible to predict whether they will happen again in the future, and if so, when.

Appendix: membership data

Trade union membership data for the Netherlands is available from Statistics Netherlands (CBS). The earliest data is from 1901; an estimate for the entire union movement is available from 1910. Data for the second world war, when the Netherlands was occupied by Germany, is missing.

Unfortunately CBS doesn’t discuss the sources and methodology behind the early union membership data, but it appears to be based on data published or provided by unions. The CBS data seems largely consistent with other sources (e.g. Harmsen and Reinalda, Van den Berg).

Of course, if you want to explain sudden changes in union membership, timing is important. It appears that most data is per 1 January (cf. Harmsen and Reinalda 1975: 427 note g and 431 note c). Van den Berg has subtracted one year of the date, so that it becomes end-of-year data. For example, 1 January 1920 is recoded as 31 December 1919. This makes sense if you want to explain membership changes by something that happened during the year. I’ve applied the same procedure.

A recent study compared union density based on union-provided data with density estimates based on surveys (the study doesn’t include the Netherlands). While the estimates were often similar, there were substantial differences for specific periods in specific countries. Bias may play a role in surveys, but the authors also suggest unions may have an incentive to exaggerate their membership (Batut a.o. 2021).

In the Netherlands, there have been accusations in the late 1940s that the EVC provided inflated membership figures. An American observer noted that this was frowned upon «in a country, where it is expected that such data have the same accuracy as the reports of the Central Bank» (Windmuller 1970: 130). Apart from this specific instance, there don’t seem to have been serious questions about the reliability of membership data provided by the major unions in the Netherlands.

Sources

Batut, Cyprien, Ulysse Lojkine and Paolo Santini (2021), Which side are you on? A historical perspective on union membership and composition in four European countries. Working Paper, Paris School of Economics.

Beer, Paul de and Lisa Berntsen (2019). Vakbondslidmaatschap onder druk in Nederland, maar niet in België. Tijdschrift voor Arbeidsvraagstukken 35(3): 255-74.

Berg, Annette van den (1995). Trade union growth and decline in the Netherlands. Tinbergen Instituut.

Bouman, P.J. (1950). De april-mei-stakingen van 1943. Martinus Nijhoff / Rijksinstituut voor Oorlogsdocumentatie.

Braambeek, H.J. van (1936). Van lichten en schiften: Gedenkboek van de Nederlandsche Vereeniging van Spoor- en Tramwegpersoneel 1886-1936.

Clawson, Dan (2003). The next upsurge: Labor and the new social movements. ILR Press.

Cruchten, Jo van Rob Kuijpers and Sjaak van der Velden (2006). Werkstakingen 1900–2004. CBS, Sociaal-economische trends, 1e kwartaal 2006.

Freeman, Richard B (1998). Spurts in Union Growth: Defining Moments and Social Processes. In Michael D. Bordo, Claudia Goldin en Eugene N. White (eds), The Defining Moment: The Great Depression and the American Economy in the Twentieth Century. University of Chicago Press.

Harmsen, Ger and Bob Reinalda (1975), Voor de bevrijding van de arbeid: Beknopte geschiedenis van de Nederlandse vakbeweging. SUN.

Jong, F. de (1956), Om de plaats van de arbeid: Een geschiedkundig overzicht van ontstaan en ontwikkeling van het Nederlands Verbond van Vakverenigingen. Amsterdam: De Arbeiderspers.

Piven, Frances Fox and Richard A. Cloward (1979/1977). Poor people’s movements: Why they succeed, how they fall. Vintage.

Pizzolato, Nicola (2012). ‘I terroni in città’: Revisiting southern migrants’ militancy in Turin’s ‘hot autumn’. Contemporary European History 21: 619-34.

Treub, M.W.F. (1920). De economische toestand van Nederland gedurende den oorlog. In H. Brugmans (ed), Nederland in den oorlogschtijd. Amsterdam: Elsevier.

Velden, Sjaak van der (2000), Stakingen in Nederland: Arbeidersstrijd 1830-1995. IISG/NIWI.

-- (2004). Werknemers in actie: Twee eeuwen stakingen, bedrijfsbezettingen en andere acties in Nederland. Aksant.

-- (2005). Werknemers georganiseerd: Een geschiedenis van de vakbeweging bij het honderdjarig jubileum van de Federatie Nederlandse Vakbeweging (FNV). Aksant.

-- and Rosa Kösters (2019). Labour Conflicts in the Netherlands, 1372-2019. https://hdl.handle.net/10622/APNT4U. IISH Data Collection, V10.

Weinhauer, Klaus (2017), Labour Movements and Strikes, Social Conflict and Control, Protest and Repression (Germany). In Ute Daniel, Peter Gatrell, Oliver Janz, Heather Jones, Jennifer Keene, Alan Kramer, and Bill Nasson (eds), 1914-1918-online. International Encyclopedia of the First World War, issued by Freie Universität Berlin.

Windmuller, John P. (1970), Arbeidsverhoudingen in Nederland. Utrecht / Antwerpen: Het Spectrum.