Nordic Economic Policy Review 2025

Trade unions appear to affect the labour market quite differently in Scandinavia (Norway) than in the United States. In the U.S., their impact on the labour market when they establish a power base is highly contested, and many causal studies identify only minor effects. Senior and key workers appear to leave unionised firms, and innovative activity drops. In Norway, the benefits associated with unions are clear in certain sectors but less so in others. In manufacturing sectors, where collective bargaining agreements are common, unions raise wages and productivity but have a greater impact on productivity. Across all sectors, unions raise wages, particularly for low-paid workers and help reduce inequality in concentrated markets, thus ameliorating a market failure caused by the employers’ market power. Unions lead to fewer calls on social security benefits in the long run. They stimulate product innovation but make less of a contribution to labour-saving technologies. Unions appear less interested in promoting new technology in service sectors.

 

Keywords: Unions; market power; technological development; welfare

Acknowledgement: I would like to thank the editors, Erling Barth, Elin Svarstad and Anniken Hagelund for their invaluable discussions and helpful suggestions, which have made a distinct improvement to my work.

Trade unions appear to affect the labour market quite differently in Scandinavia (Norway) than in the United States. In the U.S., their impact on the labour market when they establish a power base is highly contested, and many causal studies identify only minor effects. Senior and key workers appear to leave unionised firms, innovation rates drop due to firms relocating innovative activities to states less favourable to unionisation, and the market evaluates these firms negatively over time. In Norway, unions and employers have and exercise market power in the labour market. The interplay of these players’ market power shapes the Norwegian labour market, creates winners and losers and stimulates technological development differentially. While the benefits are clear in certain sectors, they appear less so in others. In manufacturing sectors, where collective bargaining agreements are common, unions raise wages and productivity but have a greater impact on productivity. Across all sectors, unions raise wages, particularly for low-paid workers and help reduce inequality in concentrated markets, thus ameliorating a market failure caused by the employers’ market power. The impact of trade unions in the long run appears more positive than in the short run, for example, they lead to less call for social security benefits by workers of all ages. Firms respond to union wage demands partly by raising prices (if they have the power to do so), but unions also stimulate product innovation, so product quality can also improve. Unions are less keen on labour-saving technologies and appear less interested in promoting new technology in service sectors. Nevertheless, although unions do not always work for the greater good, in my view, their positive contributions outweigh the negative. Thus, if lower union density implies reduced union power, policymakers should worry.

The utilisation of market power is common in the labour market. For example, workers can collectively organise the supply of labour by being in unions. Depending on their collective strength, this may raise wages above the competitive level, which is an example of utilising their market power. However, employers also have the potential to utilise product and factor market power to charge the highest possible product prices while pushing for the lowest possible factor prices. Wages – the price of labour – is just one of those factor prices.

In this paper, I will argue that it is the interplay between these players’ market power that shapes the labour market, creates winners and losers and stimulates technological development differentially. While the benefits are clear in certain sectors, they are less so in others. My arguments are based on a literature review of recent research into how unions affect the labour market, for firms and workers, in particular drawing inferences from causal evidence from Norway and the United States. I will also highlight policy recommendations and topics for further research.

In most Western economies, unions bargain with employers in one way or another (OECD, 2017, 2019; Bhuller et al., 2022), but there is considerable heterogeneity in how this bargaining is conducted (by whom, on what level (local, sectoral, or national)), and it changes over time. Still, few would dispute that unions and collective bargaining are important for labour market outcomes (OECD, 2018). There is also considerable variation in what employers and unions bargain over. While wages might be considered the most important topic, non-wage amenities such as training, pensions, retirement, working hours, employment and the working environment are common topics in many countries. As indicated by Kauhanen et al. (2023) and Bhuller et al. (2022), this heterogeneity of national systems and sectors implies the need for a certain degree of care when drawing inferences based on results that stem mainly from a single country, which the reader should bear in mind regarding inferences I mainly draw from Norway.

The structure of the remainder of the paper is as follows: In Section 2, I briefly describe how imperfect competition in labour and product markets influences market power. In Section 3, I describe the institutional aspects of bargaining in Norway, while Section 4 discusses the traditional view of unions and bargaining. Sections 2–4 set the stage for my analysis. Section 5 describes Norwegian public policy that supports unions, and this serves as the foundation for my causal analyses of various topics such as new technology, wages, productivity, and market power. In Section 6, I examine the general relationship between unions, productivity, and wages. However, this average relationship might differ for sub-groups of workers and firms. Thus, Section 7 addresses these heterogeneous effects and responses by workers and firms, highlighting issues related to market power and inequality. Section 8 addresses unions and innovation, while Section 9 focuses on the policy implications of the impact unions have on both firms and workers.

Market power is the ability of a participant in a market to influence the prices of what is traded in that market so that they deviate from competitive prices (perfect competition). Typically, market power is associated with a welfare loss. The idea that firms have market power and influence prices in product markets has been widely acknowledged for centuries (e.g., Smith, 1776: 71-72; Cournot, 1897). However, they can also wield market power in factor markets, such as the labour market (where the factor of production is labour, and the price of labour is the wages). If the firm is the sole determinant of the price (single buyer) in the market, this would be an example of monopsony.

Smith also addressed employer cartels, thus indirectly addressing monopsony, and imperfect competition in the labour market was explicitly analysed by Robinson (1933). The classic example of monopsony was the mid-western mining town, where all businesses were owned by the company, which also employed everyone. In more moderate cases, it has since been recognised that this extreme form of geographically based market power can be related to preferences, transportation costs, and distances (Bhaskar et al., 2002; Manning, 2003, 2011). Monopsony can also arise due to labour market frictions, such as information flows and search costs (Manning, 2003).

In a perfectly competitive labour market, a company offering lower wages than the competitive level would lose all its workers. Similarly, if it offered wages higher than the competitive level, everybody would want to work for it. Thus, in the competitive case, each firm’s labour supply would be represented by a horizontal line, reflecting the competitive level, in a diagram where wages are on the y-axis and employment on the x-axis. However, when firms possess some degree of monopsony power, their labour supply increases slightly when it puts wages up slightly. In the same diagram, this would be represented by an upward-sloping labour supply curve.

A continuum of wage levels designed to maximise profit arises in models of an economy characterised by labour market frictions, reflecting the fact that some firms choose low rates of labour turnover while others prefer to be small and have higher turnover rates. However, the wage level will be lower than the productive value. Numerous studies have been conducted based on models inspired by the dynamic monopsony or equilibrium search framework. Recently, employer market power has been recognised in a series of studies focusing on market concentration, for example, Azar et al. (2022) and Thoresson (2024).

Is monopsony in the labour market relevant in Scandinavia in general and Norway specifically? Several studies over the years suggest that it is. For instance, decades ago, research identified labour market frictions that influence worker turnover and wage-setting at the company level, indicating the presence of monopsony (Dale-Olsen, 2006). These frictions have also been linked to gender wage differentials (Barth and Dale-Olsen, 2009) and differences between occupations (Falch, 2010). More recently, Dodini et al. (2024a) used Norwegian register data to examine job task concentration, where mass layoffs serve as shocks to these task markets. They found that workers in more concentrated markets experience more negative outcomes after such shocks compared to those in less concentrated ones.

However, it is not just employers who can utilise market power in the labour market. Whenever workers are organised in unions and bargain about wages, they are utilising market power if they are strong enough to raise wages above the competitive level. This is called the monopoly face of unions (to which I will return later).

As pointed out at the start of this section, market power is often associated with welfare loss. If employers utilise monopsony powers, wages are set too low. When unions bargain about wages, they might counteract this by pushing wages upwards. However, in theory, this depends on the union’s preference for wages relative to employment (Manning, 2003:358-360). Many Anglo-American studies assume an employer’s right-to-manage perspective, i.e., unions bargain for wages and the employer determines employment conditional on the wage level. In these cases, the union has no preference for employment in their utility function. If the union has a stronger preference for wages than employment, Manning’s model yields an unambiguous prediction – the negative wage impacts of stronger employer monopsony power are counteracted by union bargaining.

Finally, it is also worth noting that market power in different markets can easily be related. Large firms like Apple and Tesla, for example, not only influence the prices of their products but can also impact the prices of intermediate factors used in their production processes. This relationship is supported by empirical evidence from studies such as Abowd and Lemieux (1993), Dobbelaere and Kyota (2018), and Soares (2020), which demonstrate that market power tends to be correlated across different markets. In such a setting, the union-bargaining model in Manning (2003) becomes too simple because the amount of labour a firm chooses will – depending on how high the negotiated wage is –either be bound by the product demand curve or by the labour supply curve. This will be determined by a threshold for wages (Dodini et al., 2023b). When the wages negotiated are below this threshold will higher bargaining power offset company monopsony power and yield higher wages and higher employment. When wages are set above this threshold, product market considerations cause negative employment effects which thus yield ambiguous utility impacts. Later, we will explore how product and labour market power is related in Norway (Dodini et al., 2022, 2023b).

Since my review includes many studies utilising data from Norway, a summary of the Norwegian institutional context should be valuable for an understanding of the results. Norway has not adopted the Ghent system under which unemployment benefits are ensured by union membership. Union representation is still prevalent, although it has been declining slightly in recent decades. In the public sector, aggregate union density hovers around 80%. In the private sector it varies above 40%. Norway has a highly coordinated system of wage bargaining (Barth et al., 2014; Bhuller et al., 2022), though it has gradually shifted towards decentralisation, similar to trends observed in many other countries. Bhuller et al. (2022) distinguish between two types of coordination in wage bargaining: horizontal and vertical. The former is typically the result of organisation by trade, i.e. workers with the same trade are employed in many firms. When these types of unions work together, it implies a higher degree of centralisation. Vertical coordination is when many different types of workers unionise within a firm or a plant. Corporate unions, which unionise workers across all firms and establishments within the same corporation, imply higher vertical integration. With this differentiation as a starting point, Bhuller et al. (2022:34-Figure 2) show that Norway has moved from a case of very high horizontal coordination and centralised sectoral vertical coordination in 1980 to only high horizontal coordination and some sectoral vertical coordination in 2018. 

Norway conducts a form of pattern bargaining in which export-exposed industries bargain first, and the results from those rounds of bargaining set a wage norm for bargaining in other industries and sectors (Calmfors, 2025). In addition, subsequent local bargaining supplements these sectoral rounds. Barth et al. (2014) note that close to 80% of all employees work in plants with local bargaining, and 80% of these also bargain about non-wage amenities such as training, productivity targets, downsizing, reorganisation, etc. However, one important contrast to Central-European pattern bargaining countries is that collective agreements on the sectoral level in Norway are generally only binding for workplaces where local unions have been strong enough to establish a collective agreement, i.e., union density at the workplace level is important for the implementation of bargaining outcomes (in certain industries mandatory extensions exist). In practice, this narrows down to even smaller units, since a workplace might comprise several bargaining areas (e.g., one differentiation is between blue- and white-collar workers, but even narrower differentiations might exist). If the firm is a member of an employers’ association, it will have a formal threshold for when workers are able to demand a trade union agreement. For the largest employers’ organisation, the Confederations of Norwegian Enterprises (NHO), the threshold is a union density within the bargaining area of the workplace of at least ten per cent. Other employer organisations endorse higher thresholds. Usually, however, the unions require much higher union density before they demand an agreement. Kostøl (2024) reports a range of 10–50%. For firms that are not members of employers’ organisations, the unions will attempt to establish an agreement either voluntarily or as a result of industrial action. Norway does not have a legally set minimum wage, but a collective agreement may include minimum wages binding for employees covered by it and, in some cases, for non-members as well (through mandatory extensions) (Kauhanen, 2025).

To understand the impact of unions on the labour market, a natural starting point is what has been called the two faces of unions (Freeman and Medoff, 1984) – the monopoly effect and the voice effect. We met the monopoly face of unions in the first paragraph of Section 1, i.e., the ability to reap extra-market returns. By leveraging their bargaining power, unions may negotiate wages that are higher than the competitive market wage or the wage an employer would otherwise offer. The bargaining outcome is influenced by both the strength of the union and the issues being negotiated. In the classical right-to-manage model of Pencavel (1984), unions only care about wages, and the employer sets employment. As wages are pushed upwards by unions, labour demand declines, and employment drops. Additionally, since investments are “sunk costs”, unions gain leverage in wage negotiations – the so-called hold-up problem – which can reduce the return on investments and subsequently lower the incentive for further investments and innovation (Grout, 1984). Multi-country empirical analyses support the notion of sunk costs, where this mechanism is stronger in more sunk-cost-intensive sectors (Cardullo et al., 2015). In cases where unions are concerned with both wages and employment, known as efficient bargaining, McDonald and Solow (1981) show that they internalise the potential negative employment effects when they negotiate high wages. In both these cases, the union face is a rent-seeking device.

The union “voice” effect tries to capture the idea that while each worker has preferences and concerns at work, for each of them to contact management on every issue would be highly inefficient. At the same time, it appears unwise, at least from a motivational point of view, to ignore what employees think. They might also have important ideas and notions regarding production. Instead, unions synthesise these ideas and notions into one powerful, coherent thought, which can lead to efficiency gains. The unions provide a “voice” to workers’ ideas and opinions.

Unions may also have opinions about or preferences for specific new technology, for instance, regarding labour-saving technologies (Dowrick and Spencer, 1994; Lommerud and Straume, 2004). Unions usually care about their members’ jobs; thus, labour-saving technology is not necessarily their preferred form of innovation. That said, if such technologies are to be implemented, trade unions can reduce the costs associated with them. For example, Bryson et al. (2013) found that local unions helped mitigate the negative effects of process innovation on workers, such as job anxiety and reduced job satisfaction. Where unions are established is not random either. Workers in the U.S. appear to target organising in young, productive firms, where potential rents are higher (Dinlersoz et al., 2017), and similarly in France, Breda (2015) finds that unions organise in firms with high profitability.

I end this section by briefly discussing some theoretical implications of bargaining regimes. The recent move towards the decentralisation of wage bargaining in Scandinavia (see, e.g. Dahl et al., 2013; Kauhanen, 2023 and Willén, 2021 for analyses from Denmark, Finland, and Sweden, respectively) is outside the scope of my study. However, the degree of centralisation in wage bargaining is important for understanding some of the recent empirical developments on company wage setting and innovation. Almost 40 years ago, Calmfors and Driffill (1988) published their seminal study on the relationship between union wage demands and the structure of collective bargaining. They showed that the relationship between wages and bargaining level is hump-shaped, where the wage demand in intermediate or industry-level systems is much higher than that theoretically predicted in two other extremes (completely decentralised or fully centralised systems). If firms are price-takers in the product market, then it does not matter whether wages are set centrally or in a fully decentralised manner. In both cases, the union members bear the consequences of their nominal wage demands, and they pay the price in the form of lower employment. However, if firms have market power in the product market, which in many cases is a reasonable assumption, then firm-level and national centralised wage-setting differs. In this case, the firm faces a downward-sloping demand curve, i.e., an increase in product prices reduces product demand somewhat, but in contrast to the competitive case, the demand for the firm’s product does not disappear completely. Under local, company-level wage-setting, firms can pass part of the wage increase demanded by unions onto product prices, and unions take this price-shifting into account when negotiating wages (Moene et al., 1993). This results in higher real wages compared to centralised wage setting but at the cost of lower employment. This idea of firms adjusting prices in response to union wage demands is thus well established and is revisited in a recent study from Norway (Dodini et al., 2023b).

Equally important, the structure of bargaining also has implications for technological development and innovations. As demonstrated by Moene and Wallerstein (1997), Haucap and Wey (2004) and Barth et al. (2014), under certain assumptions, centralised bargaining tends to provide stronger incentives for job creation and innovation than local bargaining. The empirical evidence presented by Haucap and Wey (2004) and Barth et al. (2014) also supports these implications. The underlying mechanism in these models is similar and can be explained as follows. Think of a stylised productivity path for a firm from its inception until it is shut down. When the firm is established, it operates with cutting-edge technology, reaching peak productivity. Over time, as the firm ages, its relative productivity declines as it moves away from the technological frontier. When productivity drops below the competitive wage level, the value of production is less than the wage costs, and the firm is forced to close. Under local bargaining, if the bargaining power of unions is strong enough, wages are set as a share of productivity, which corresponds to the firm’s productivity level over time. As is the case in competitive wage settings, the firm closes when productivity drops below the market clearing wage, but since union wage demands make investments less profitable, labour demand is reduced at all age levels, which consequently also reduces the market clearing wage. Under centralised industry-wide wage setting, however, the union sets the wage based on the average productivity in the market across all age levels. This average will be considerably higher than the least productive firm in the industry, e.g. operating under competitive wage setting or local bargaining. This average will also be considerably less than wages paid under local bargaining at the most productive firm. The profits accruing to surviving firms under centralised industry bargaining can thus be higher than under local company bargaining, while the least productive firm will, however, be forced to shut down sooner. Stronger unions and local bargaining thus contribute to longer life for firms, but reduced productivity when a firm is forced to close (when the marked-clearing wage becomes higher than the value of production), and fewer incentives for job creation. Stronger unions under central bargaining thus contribute to shorter life for the firms, but increased productivity when a firm is forced to close, and higher incentives for job creation. This means that centralised wage bargaining subsidises the most productive firms at the price of killing off the least productive ones. Similarly, in the innovation race mentioned in Haucap and Wey (2004), the average wage in the market under centralised bargaining (set for both winners and losers in the innovation race) is much lower than the bargained wage under local bargaining for the innovation winners. Thus, being the winner of the innovation race is much more profitable under centralised bargaining than local bargaining, and this stimulates innovation.

While the discussion above is based on homogenous labour, the implications are similar for heterogenous labour and union preferences for wage equality (wage compression). Wage compression makes the most productive firms more profitable due to lower wages for high-paid employees, while the least productive firms are hurt by paying higher wages to low-paid workers. Consequently, being at the technological frontier is more profitable under centralised wage bargaining, stimulating technological development and innovation. Or, in the words of Barth et al. (2014:1), “wage compression fuels capitalist investments in the process of creative destruction, increasing the average productivity and the average wage for a constant employment level”.

Finally, does this mean that centralised bargaining always provides stronger incentives for job creation and innovation than local bargaining? Not necessarily. These studies highlight stylised mechanisms that lead to this outcome for bargaining regimes, but they also ignore relevant issues, e.g. related to the interplay with implementation and mobility costs. Thus, at least at the micro-level, it is easy to imagine the possibility that local union bargaining, to a greater extent than central bargaining, also entails local issues such as process innovation, reorganisation, work environment/performance plans and training, i.e., aspects that could twist this conclusion around. An indication that this is possible is found in Cardullo et al. (2020). They point to the case of two-tiered bargaining systems, in which local bargaining on top of a centrally set wage links pay to local productivity, which might mitigate the previously described hold-up problem.

Measured in membership or aggregate density, trade unions have been in decline for decades (OECD, 2017; Schnabel, 2020; Kjellberg, 2025), albeit less so in Norway than in many other countries. This could make them weaker (bargaining power) and reduce workers’ labour market power. 

This decline in the demand for union services is probably influenced by several factors, as discussed quite extensively by Barth et al. (2025). The fact that union coverage has not decreased to a similar extent suggests that the free-rider problem is at least one of the explanations. Why pay for union dues when the benefits of collective bargaining are available regardless of membership status? The free-rider issue appears to be further aggravated by the introduction of mandatory extensions. Fear of social dumping (see Hayter and Visser, 2021 for a discussion) and downward wage pressure following labour immigration after the EU enlargement in 2004 encouraged Norwegian unions to exert influence on politicians to introduce mandatory extensions of collective agreements in several industries (see Kauhanen (2025) for a survey of the Nordic countries). This increase in the prevalence of mandatory extensions of collective agreements in Norway from 2005 to 2011 causally reduced union uptake by 2.7% (Flaarønning, 2024). However, the politicians have also implemented schemes that support collective organisation. 

In Norway, both employers’ organisations and trade unions are supported by the government. By allowing tax deductions for membership fees, the government influences the net cost of joining a trade union or employers’ organisation. However, there is a major difference in how these subsidies are determined. While the tax deductions for membership of employers’ organisations have remained basically unchanged for decades, they changed quite dramatically and non-linearly for union membership during the period from 2001 to 2012. These changes were partly due to changes in political leadership, from a liberal-democratic coalition to a government led by Labour.

It is these changes in net union dues that several authors have recently utilised to derive the causal impacts of unionisation on several different outcomes. The tax changes act as a natural experiment. While the deduction is identical for each worker, the relative importance of the subsidy differs depending on what the union dues are for each worker’s particular occupational group. Since some unions are expensive and some cheap, a subsidy of, for example, NOK 500 has different importance, and this difference varies over time as the subsidy changes. Since nobody will receive money or will be paid for joining a union, the tax subsidy is limited by the size of the dues for cheap unions. Thus, the tax treatment varies in strength in a non-linear way.

When fixing the first observed union fee (and controlling for variation between unions), any subsequent variation in net union dues is driven by changes in the subsidy. Over time, this creates a varying-treatment design environment, which makes it possible to identify causal effects when utilising net union dues as an instrumental variable in regressions of union density on different outcomes.

Studies endorsing this approach address how unionisation causally affects wages (Barth et al., 2020; Svarstad, 2023), prevalence of low-pay (Svarstad, 2023), earnings (Dodini et al. 2022, 2023a, 2023b), technological changes influencing routine and non-routine workers differently (Kostøl and Svarstad, 2023), productivity (Barth et al., 2020), wage inequality (Dodini et al., 2022; Dodini et al., 2024b), career outcomes (Dodini et al., 2023a), employment (Dodini et al., 2022, 2023b) and price-taker behaviour (Dodini et al., 2023b). Apart from Dodini et al. (2022, 2023a), these studies primarily focus on private-sector unionisation. Public-sector unionisation is very high, and non-members probably have strong preferences regarding membership. Thus, there is less room for public tax policies to influence uptake.

In Figure 1, which is based on figures presented in Barth et al. (2020, 2025), I depict the income deductions and the tax subsidy for tax purposes during the years 2001 to 2012. The left-hand axis measures the deductions and subsidies in nominal NOK. The figure also shows the subsidy measured relative to the average private-sector union dues. This is shown on the right-hand side axis as a rate. We see from the figure that the tax subsidy started at roughly NOK 250 in 2001 and grew to NOK 1,000 in 2012. Measured relative to average union dues, this implies growth in the subsidy from under 10% to nearly 25%. Furthermore, and this is not shown, for industries where the average union dues are lower, the increase was even larger. The bottom line is that in the period 2001–2012, the subsidies for union membership in Norway increased substantially. The question is whether this increase had any impact on union uptake.

Figure 1. Income deduction, tax subsidy and subsidy share of union membership fee. 2001-2012

Note: Based on Figure 1 in Barth et al. (2025). The subsidy share is noted on the right-hand axis and expresses a rate. The subsidy share expresses the tax subsidy relative to the average union membership fee for union members in private-sector industries. The income deduction and the tax subsidy are noted on the left-hand axis, expressed in nominal NOK.

All the above-mentioned papers indirectly answer this question in their analytical approach (what might be called an instrument variable approach (IV approach/IV analysis)) since they study what happens to different outcome variables following changes in union density induced by tax changes. Barth et al. (2020) and Kostøl and Svarstad (2023) provide individual membership analyses, but they are primarily focused on IV analyses. However, both Barth et al. (2024a) and Dodini et al. (2023a) primarily study the impact of the policy change. Their orientation differs slightly when estimating the price elasticities in that Dodini et al. (2023a) primarily study the net income cost of union uptake, while Barth et al. (2025) measure the impact purely from the subsidy. This partly reflects the main concerns of the studies. Barth et al. (2025) focus on private-sector segments of the labour market where unions have low representation in the first place and how public policies can stimulate union membership uptake in these segments, while Dodini et al. (2023a) study the overall career effects of unionisation across public and private sectors. In addition, Dodini et al. (2023a) conducted a separate survey (in addition to their administrative data analyses) to reveal union members’ sensitivity to dues. Still, the similarities are striking. Let me briefly recapitulate the main findings. 

The subsidy elasticity and the net union cost elasticity are highly significant on average, i.e., subsidies and net union costs matter for workers, but they vary greatly between groups of workers. On average, across all workers, Barth et al. (2025) find a subsidy elasticity of 0.29 (meaning that a 10% increase in the subsidy would boost union membership by 2.9%), but this elasticity is much higher for young workers than older ones. In general, there is much higher sensitivity to subsidisation of union membership among workers employed in segments of the labour market with low union representation (e.g., immigrants, temporary workers, young people). Similarly, Dodini et al. (2023a) find that the net cost elasticity diminishes from -0.067 for ages 25–29 to -0.0361 for ages 60–64. Their survey reveals a similar pattern of diminishing willingness to pay for unionisation based on age.

Did the tax policy significantly influence unionisation rates? At the private-sector aggregate level, Barth et al. (2025) show that if the tax legislation/tax subsidies had been unchanged from 2001 to 2012 (kept at 2001 levels), the counterfactual development would have ended in a further drop in aggregate unionisation of 5%. In segments of the labour market with low rates of unionisation, the drop would have been even steeper.

The U.S. studies present a rather bleak or negative picture of how unions affect the labour market. U.S. studies usually apply a regression-discontinuity (RD) approach, leveraging the fact that unions require a majority vote (more than 50%, known as the ‘Wagner Act’ model). This method allows researchers to compare outcomes between firms just below and just above this threshold. Why is the picture negative? As seen in the seminal contribution of DiNardo and Lee (2004), in the short run, the impact of unionisation on employment, output, productivity and even wages appear to be negligible. However, the equity value of newly unionised firms drops after 15–18 months, and cumulative abnormal returns correlate more negatively as the vote share in support of unionisation increases (Lee and Mas, 2012). This suggests that financial markets anticipate stronger negative impacts as union bargaining power grows. Even worse, Frandsen (2021) shows that it is not random who leaves a firm after a close union ballot. After close ballot results in favour of unionisation, he observed that earnings, employment, and company survival decreased. However, this was primarily due to older and higher-paid workers leaving. When he studied workers who remained at the firm, the impact of unionisation on wages was relatively small. Thus, the benefits of unionisation in the U.S. economy are far from obvious.

The Norwegian studies present a much more positive view of unions. Most of them use an instrument-variable design, utilising the tax policy changes described in Section 5 (if not explicitly noted). Barth et al. (2020) focus on manufacturing firms, as data covering this sector allows them to measure total factor productivity (TFP) by a control function approach (Ackerberg et al., 2015; Gandhi et al., 2020). Their analysis reveals a strong positive impact of union density on TFP, where a 1% increase in union density increases TFP by 1.7%. Importantly, when they account for potential confounders—such as variations in workers and jobs, differences in technology and technical changes, or the influence of endogenous transitory inputs—their findings remain largely unchanged. Moreover, they find that not only does productivity rise with greater union density, but wages do as well. Barth and colleagues find that a 1% increase in union density implies 1–1.5% higher wages, conditional on fixed worker productivity (known as fixed effects). Thus, as union density grows, both productivity and wages rise, but productivity more so. One-fifth of the wage effect reported above can be attributed to rent sharing, i.e., the division of the surplus value of what is produced between the firm and its workers.

If unionisation improves companies’ productivity and wages, what about other outcomes for companies? Dodini et al. (2023b) explore the impact of unionisation on a wide range of company outcomes in addition to wages and productivity. Their key findings, primarily for manufacturing firms, are that higher union density increases employment, capital, sales, and profits. An increase of 1% in union density increases employment by 1%. In Section 7, I will return to the impact of unionisation on market power and labour markdown.

By ignoring the productivity issue (which limits which sectors that can be analysed due to available data), Dodini et al. (2022) could extend the analysis of Barth et al. (2020) to cover the whole Norwegian labour market. Surprisingly, when they study the impact of union density on wages, they find nearly the same impact across all sectors and industries as Barth et al. found in manufacturing. Raising union density by 1% causes earnings to grow by 1.9%.

These studies utilise the Norwegian tax reform as a natural experiment in IV analyses and implicitly study three groups of workers: i) those who would never join a union regardless of the extent to which union dues are subsidised, ii) those who would always be members regardless of to what extent union dues are subsidised, and iii) the compliers, i.e., those who respond when union dues are subsidised. It is the union membership changes in group iii) that cause the impact on outcomes. By applying survey data, Barth et al. (2020) observe that when the compliers increase union density, three organisational changes also occur: i) a formalisation of local bargaining, ii) a formalisation of sectoral bargaining, and iii) the establishment of local consultative committees. Thus, the establishment of a trade union agreement is clearly important, as this is expressed by i) and ii). However, the establishment of trade union agreements appears mostly associated with union density levels below 25%. The establishment of local consultative committees is most strongly associated with union density levels between 25–50%. Barth et al. also studied whether the tax changes had differential impacts (in strength) across the union density distribution, and while these did vary, it was strongest when density was around 25–50%. Thus, the establishment of local consultative committees occurs most frequently where union density has been most strongly affected by the union tax subsidisation, and, from this perspective, it might be considered the key driver.

An indication that this is the correct interpretation is found in Svarstad and Kostøl (2023), which studies how the productivity-enhancing effects of greater union density vary depending on whether a trade union agreement has already been established. They also attempt to derive causal estimates. However, they have chosen to study the impact on outcomes from changes in union density but not those induced by tax changes. Instead, they assume that workers’ unionisation is related to their parents’ unionisation, potentially for cultural reasons. Thus, they study the impacts of changed union density in companies when it is induced by changes to the unionisation of co-workers’ parents. Their key finding is that greater union density primarily has a productivity-enhancing effect when a trade union agreement has been established. For companies not covered, increased union density appears detrimental to productivity. However, one problem with this study is that the union density and the establishment of a trade union agreement are clearly related (see Section 3), and the analysis ignores that relationship. It might also be argued that the relationship between the unionisation of the parents of co-workers and union density now could reflect circumstances that also affect productivity (for example, networks, occupations, task proficiency, and ability).

While most of the Norwegian studies that causally identify the impact of unions on outcomes utilise the exogenous variation in union dues induced by tax, exceptions exist. Another methodological approach is to utilise information on distinct pre-determined unionisation (rate, levels) and then relate these to the potentially differential impact of common shocks, i.e., the ‘shift-share approach’. Uncommon shocks could easily be associated with selection and endogenous responses; thus, by avoiding these, the shift-share approach yields estimates that are interpretable as causal. In one such shift-share design, using regional unionisation information from the end of World War 1, Dale-Olsen (2021) identifies that when local unionisation increases by 1%, total factor productivity increases by 0.175%, while regional wages grow by 0.1%. Thus, the impact is weaker than the company-level analysis of Barth et al., but increased local unionisation still causes productivity to grow more than labour costs. Employment appears unchanged on average, but entry-level companies grow bigger.

Unions affect the labour market unequally for different workers and firms. For example, Barth et al. (2020) found that wages increase for both low and high-productivity firms but at a faster pace for high-productivity ones. Several recent studies have explored other forms of heterogeneity.

Let us begin with Svarstad (2024), which explores whether unions are successful in achieving one of their core objectives: raising the wages of the lowest paid. Her findings are quite striking and clear: as union density grows, the less likely a worker is to be low paid, even when controlling for all fixed productivity differentials within a job. She also observes that the effect of local bargaining power on the probability of low pay is stronger for immigrants than natives. Thus, her findings support the notion that unions improve the welfare of groups with a weak position in the labour market. In contrast to Svarstad (2024), Dodini et al. (2024b) argue that unionisation primarily benefits natives and that the positive wage effects diminish for Western immigrants and disappear for non-Western immigrants. However, they only account for fixed productivity differentials within a firm, i.e., different firms might comprise very different workers when it comes to productivity. Thus, the different results of Svarstad and Dodini et al. can, at least partly, be attributed to different selection processes to firms for natives and immigrants. 

Next, remember my discussion of market power in Section 2, where the more concentrated a labour market segment is, the greater the employer’s market power. Similarly, it is not difficult to imagine weak labour market groups facing monopsonistic employers (with more labour market power). This is explored by Dodini et al. (2022), who study how the union-wage relationship is sensitive to the market concentration of blue-collar and white-collar workers. While white-collar workers receive a significant union wage premium in competitive markets, this premium is negligible for blue-collar workers. However, as market concentration increases, the more of the additional rent the union extracts, the more it benefits blue-collar workers. Thus, unions contribute to greater inequality in competitive markets but narrow the inequality gap in markets characterised by monopsonistic competition. Equally important, these changes in wages also manifest in changes along the intensive margin (working hours). In competitive markets, an increase in wages tends to reduce hours worked, but in monopsonistic markets, higher wages cause an increase in the hours worked. However, no such significant pattern is observed concerning employment. Dodini and co-authors also explore the sensitivity of the union-wage relationship to concentration across the earnings distribution. While their findings align with previous descriptions, it is worth noting that local labour market inequality in the upper half of the earnings distribution remains unaffected. Thus, the inequality-reducing effects of unions appear to be concentrated at the lower end of the earnings distribution. Once again, these results support and extrapolate on Svarstad (2024). Weak labour market groups are more likely to be concentrated at the lower end of the earnings distribution. 

Next, one might wonder whether the benefits of increased union density are limited to pay or are detectable for other job characteristics and amenities and if these gains are short-lived or not. From the Norwegian administrative register, Dodini et al. (2023a) construct four measures expressing i) monetary compensation, ii) job protection (unemployment benefit), iii) promotion facilitation (probability of advancement), iv) working environment (sick leave) and v) public transfers. They analyse the impact of union density on these measures for public-sector and private-sector workers both in the short run and over a worker’s life cycle. In the short term, they find that young workers primarily benefit from higher monetary compensation and reduced reliance on public transfers, with little impact on job protection, promotion, or working environment benefits. As workers age, the importance of monetary compensation diminishes, while other dimensions, such as job protection and the working environment, become more significant. Over the life cycle, increased union density is associated with a clear reduction in the need for unemployment benefit, more sick leave (indicating greater job security), and fewer public transfers. One interpretation of these findings is that increased union density is associated with improved future employment probabilities and, thus, less need for public support (note that very few people receive sick pay while unemployed). Interestingly, the long-term gains from increased unionisation appear more aligned with workers’ responses in surveys. Even more important is the observation that the long-run use of the social security system declines for all ages.

When it comes to attitudes to sick leave, this causal study presents a contrast to the more correlation-oriented study of Mastekaasa (2013), which basically identified attitudes to sick leave as individually motivated and not related to the firm. However, he also observed that workers joining unions started to be more absent, while workers ending their union membership stopped being absent. This indicates that job security is higher for unionised workers.

For unions, it is of great importance that their workers stay attractive in the labour market. As technology progresses, workers will have to adapt and reskill to avoid becoming obsolete. This makes training (aka re-education) an important non-wage amenity. Although Kostøl (2024) is not able to derive causal estimates, correlations point to how unions shape participation in further education. In population-wide register data from 2004–2019, he finds that increased union density increases a worker’s individual propensity to participate in tertiary vocational education. In unionised establishments, these workers receive higher salaries while participating in education, but post-training wage premiums become lower. At the same time, the turnover probability of workers drops. Kostøl links these results to the notion that frictions are necessary for employers to sponsor training or investments in workers’ skills. Few employers will voluntarily sponsor training in general skills if the worker might move to a competitor afterwards.

Unionisation also changes firms’ utilisation and disposition of labour. How do they adapt to unions? Kostøl and Svarstad (2023) asked whether the changes in union density driven by the tax reform altered pay differently for routine and non-routine workers and, if so, do unions have any impact on relative labour demand depending on relative wages. While the wages of routine workers are 12% lower than those of non-routine workers, Kostøl and Svarstad find that a 10% growth in union density increases the wages of non-routine workers by 1.7%, while the corresponding figure for routine workers is 3.4%. In other words, unions compress the wage differentials between routine and non-routine workers. However, such a compression should also increase the relative demand for non-routine workers, as they have become relatively cheaper. As such, union demands may potentially accelerate technological change by increasing the demand for non-routine workers. However, unions may also force firms off of their optimal relative labour demand curve, and this is what Kostøl and Svarstad find. In general, unions increase the relative demand for private-sector routine workers depending on relative wages. Since it might be anticipated that these private sector firms would have used machines to replace routine workers, this implies that these workers either do new tasks and have been reskilled or that unions have decelerated technological development.

However, there is considerable variation across industries. In the manufacturing sector, for instance, Kostøl and Svarstad find that unions contribute to reduced demand for routine workers, which means they contribute to accelerated technological change. Therefore, while unions influence technological developments, their impact can be either positive or negative depending on the industry context. As Kostøl and Svarstad (2023: 11-12) write: “Within construction, unions are found to increase the conditional relative demand for routine workers. In other words, by influencing internal relations, unions are found to counteract the positive effect on technological change that we establish through the wage channel. Within manufacturing industries, however, unions are found to reduce the conditional relative demand for routine workers, thereby reinforcing the estimated positive influence of unions on technological change”.

In addition, since market power in different markets is found to be correlated, and unions might target high-rent firms and industries, it is an open question how firms react to union wage demands. As the discussion in Section 6 indicates, in many of the Norwegian studies where changes in unionisation are caused by tax reforms, productivity increases as union density increases, i.e. unionisation has productivity-enhancing effects. Could these effects be the sole product of firms utilising market power (and thus only reflect union wage demands being shifted over onto the price to consumers)? Barth et al. (2020) provided an analysis of compliers, that is, those that react to the tax changes, and studied how these varied between different groups or characteristics (whereof workforce size groups are particularly relevant for here). On average, and for most groups, the compliers comprised 5-10 % of the workforce. When they looked on the compliers across size groups, they found that strongest union uptake effect was found in medium-sized firms (25–100 employees) with union density around 20–40%, a union density level where unions are strong enough to establish formalised arenas for co-operation with employers. They found minor differences concerning the level of technology, but they did not explore export status, a well-known dimension related to market power. Still, it is hard to identify that it is these firms that primarily wield product market power, which is usually strongly related to aspects such as size and productivity. Their productivity analysis was based on a method less sensitive to market power in the product market (see Gandhi et al. (2020) for a discussion of gross versus value-added product functions, whereof the latter appears to be a better choice when firms have market power).

However, it is not possible to reject completely the notion that export firms respond to union wage demand by shifting the costs onto export prices. In a recent study, Dodini et al. (2023b) examined manufacturing export firms, utilising data on product export prices and transportation costs to calculate product-market markups. They found that the tax-induced increase in unionisation led to an increase in companies’ markups and reduced labour markdown, with no loss in profit despite rising wages. How is this possible? It can partly be explained by larger firms being able to charge higher prices, which are ultimately borne by consumers in export markets. This concept, where companies transfer wage demands onto prices, has been recognised for decades, as discussed in Section 4, and Norwegian exporters do have product market power, as we see from their response to new trade union agreements (Dale-Olsen, 2024). As the sole explanation, however, I find this less likely. Additionally, as Dodini et al. (2023b) point out, it is plausible that unionisation could lead to an increase in product quality among exporters, either through entirely new products or an improved product portfolio. As we will see in the next section, unionisation is usually found to have a profound impact on product innovation, and in my view, this is the more likely explanation, possibly combined with a price response. 

The regional analysis by Dale-Olsen (2021) revealed that while unionisation increases productivity and wages, the benefits only accrue to companies and jobs that survive. Low-productivity firms are forced to shut down due to the higher labour costs, resulting in layoffs, whereas new entrants into the market are less affected. Thus, unions contribute to the process of creative destruction. However, this process is costly. Although the inflow into disability pensions following layoffs is limited, and most workers find new employment (often after a brief period of unemployment in line with the basic concept underpinning the Scandinavian model), there is a noticeable increase in transitions into retirement.

Innovation causes increased productivity and economic growth and can address societal problems (OECD, 2015). As Aghion et al. (2014) point out, innovation drives productivity growth by either reducing costs, improving product quality, or both. One way to reduce costs is to adopt labour-saving technologies, which naturally makes innovation a key concern for trade unions.

Theoretically, the implications for how unions affect firms and workers with respect to innovation are threefold. One strand of the literature focuses on the previously discussed hold-up effect, where the sunk-cost nature of R&D provides unions with rent-extracting powers, which then causes under-investment. The other strand of literature acknowledges the hold-up effect but shifts the focus to how market and bargaining structures can act as counterforces (Ulph and Ulph, 2001). Within this perspective, the incentives for innovation may be stronger under centralised bargaining (Haucap and Wey, 2004) or local bargaining (Braun, 2011). The third strand focuses on how the impact of unions differ when innovations can be counteracted or shaped (Lommerud and Straume, 2012; Berton et al., 2021; Bryson and Dale-Olsen, 2022). In these cases, unions tend to promote product innovation while actively seeking to avoid labour-saving innovation.

Empirically, the evidence of how unions affect innovation is equally mixed. To an extent, this probably reflects the lack of causal studies. Most empirical studies of unions and innovation are based on correlations. During the last decades, the two survey articles by Menezes-Filho and Van Reenen (2003) and Doucouliagos and Laroche (2013) found typically negative correlations in Anglo-Saxon countries and those with weaker labour regulation but non-negative relationships between unionisation and innovation in European countries. Surprisingly, however, Addison et al. (2017) observed that sectoral bargaining in Germany was associated with greater innovation, while Berton et al. (2021) observed that product innovation increased with union strength in Italy for 2010–18, and only a very strong or a very weak union had a negative effect on process innovation.

In a comparison of Norway and the UK in 2010/2011, Bryson and Dale-Olsen (2022) observe support for the notion that local bargaining is more conducive to innovation, whether product and process innovations occur jointly or product innovation happens on its own. This confirms that unions tend to favour product innovation over labour-saving innovation, but when both types occur, the benefits from greater product demand can offset the utility loss from labour-saving technologies. Furthermore, acknowledging the possibility of selection effects, it is notable that during the COVID-19 pandemic, Norwegian firms with a collective agreement were much more likely to introduce new technology (not just online meeting platforms like Zoom) than those without such agreements (Barth et al., 2024). Non-unionised firms were similarly much more likely to postpone planned investments. Why? While this study does not explicitly explain why, they do attempt to test different explanations. The authors find that it was not driven by unionised firms being more productive than non-unionised ones, nor is it associated with size, industry, financial problems or the scale of disruption. What explanations does that leave? We know that unions ameliorate stress among workers when organisational changes occur, implying reduced implementation costs (Bryson et al., 2013). One interpretation could, therefore, be that unionisation in Norway implies a certain flexibility and adaptability not found among non-union firms, thereby potentially reducing and ameliorating negative implementation costs, a trait that proves valuable during unexpected shocks.

Among U.S. studies, the work by Bradley et al. (2021) stands out for using a regression discontinuity (RD) approach to examine the causal relationship between unionisation and innovation. As described earlier, bargaining following unionisation in the U.S. comes after a local workplace ballot that requires a majority vote. Thus, Bradley and co-authors compare firms just above and just below this threshold (the Wagner threshold) with the weaknesses this might entail (close ballots reflect weaker unions). Applying this RD approach, Bradley et al. find that unionisation causes reduced patent quality and quantity. Why this is so is not clear. From Section 6, we remember that Frandsen (2021) observed that when collective agreements were established following a ballot, the older and higher-paid workers (potentially the most senior and productive) left the firm. Thus, one explanation could be that the most innovative workers exit the organisation. Additionally, Bradley et al. show that firms are shifting their innovation activities from states where unions have been successful in organising to states where unions are weaker.

In the previous sections, I have discussed how unions affect the labour market for workers and firms. From these discussions, several policy implications can be drawn.

First, we should be slightly worried about the decline of unionisation in Scandinavia, while unions are much less important for the U.S. labour market. This indicates that these economies are quite different. Previous studies indicate strong support for unionisation in the form of the Scandinavian model, and even the more recent studies in sections 6 and 7 from Norway clearly identify positive effects associated with unionisation. Not only does productivity increase with union density, but wages do as well, and productivity increases even more. A key policy question is what causes these positive effects of increased union density. It might be tempting to attribute this just to the establishment of a collective agreement (remember the discussion in Section 3 on when unions can demand a trade union agreement in Norway). However, Section 6 shows that although this is related to both the establishment of trade union agreements (formalisation of bargaining) and the establishment of local consultative committees, it is the latter that appears the most important. Therefore, it is not enough just to enter a collective agreement; formalised routines for consultation and co-operation are also needed. 

Second, from a policy point of view, one must question the external validity of studies conducted in the U.S. and in Norway. Both these countries can potentially be considered as extremes along a wide range of societal dimensions (for example, trust and coordination of wage setting (www.oecd.org/en/data/indicators/trust-in-government.html)). This heterogeneity is one reason why these countries are presented separately. As the Norwegian studies are dominated by manufacturing sectors, are their findings relevant in other ones? We find that the establishment of local consultative committees is important in Norway, but what does this mean for countries where such committees are mandatory by law? Similarly, since the U.S. studies draw inferences primarily from union ballots with close results, are their findings relevant for strongly unionised vs non-unionised firms? Clearly, a certain care needs to be taken when drawing inferences from other countries, industries, and kinds of companies, but these studies indicate gains and losses worth taking into consideration. For example, consider the importance of the local consultative committee. In a Norwegian context, this reflects a formalised arena for co-operation between management and unions, locally embedded and manifested through local strength. In countries where such committees are mandatory by law, it varies how active locally and how important such committees are for day-to-day operations.

Third, we can draw policy implications for specific groups of workers and firms. In Section 7, we saw that unions in Norway raise the wages of low-paid workers and combat poverty for the employed. Unions also help reduce the wage/earnings gap between immigrants and natives within a job, which helps make society more equal, although they appear less successful at determining what kinds of jobs workers have over the life cycle. However, although this has not been empirically studied, one might expect that when unions contribute to higher wages and more productive firms, this comes at a cost in the form of reduced employment opportunities for workers who have very low productivity. Thus, public agencies should ensure that such workers receive enough training to make them employable. This could help offset the productivity slowdown of the last decade (Goldin et al., 2024).

Fourth, by contributing to rent sharing and wage compression, union wage demands force low-productivity firms out of business and their workers into new employment relationships or, to a minor degree, into retirement. They also make employers re-appreciate the role of workers versus machines. However, they also enhance the skills of the workforce by reducing worker turnover and ensuring that workers participate in training. This area is clearly underexplored. We have few convincing papers documenting the creative-destruction effect of unions in the labour market in general, on outcomes such as company survival and technological change, and the interplay between unionisation and collective agreements. The link between unions and creative destruction is not very well established, particularly concerning how this mechanism works across different industries. The overall conclusion here, albeit tentative, is that unions contribute to creative destruction and shape new technology indirectly.

Fifth, there is a disturbing lack of studies addressing the impact of unionisation on public sectors alone. Although most workers are employed in the private sector in Scandinavia, one-third of the employment is in the public sector (32% of all workers in Norway in 2020).

Sixth, in Section 8 unionisation appears to be important for both the implementation and development of technology. If the introduction of new technology is associated with detrimental effects on workers, unions might ameliorate these effects and help the directly affected workers. This is reminiscent of what we saw in Section 7 on welfare spending. Unionised firms also seem to be flexible and adaptable, and from this perspective, they contribute to the greater good of society. However, the number of causal studies is quite limited, and more research is needed.

Seventh, unionisation comes at a price. In some cases, the negative outcomes are less worrisome. One such example is that inequality is enforced in competitive markets. However, this negative outcome of unionisation in more competitive markets appears minor, while the gain in concentrated markets appears greater, which indicates that unions ameliorate a market failure caused by employers’ market power. In the long run, unionisation has even more positive effects. For example, the long-run use of the social security system declines for all ages. From a policy point of view, this is extremely important, as it implies a reduction in welfare payments by the government due to unionisation, primarily caused by unions work to increase job protection and improving work environment quality. Although a back-of-the-envelope calculation, Dodini et al. (2023a) estimate that the marginal union member saves the government NOK 120,000 ten years after joining and that they enjoy higher wages and provide the government with additional tax revenues.

Unfortunately, some of the negative findings entail more worrying aspects. While unions contribute to product innovation and might improve the quality of products (and/or shift their wage demands onto prices), they contribute less so when it comes to labour-saving technologies. They might help ameliorate the negative effects associated with these, but they appear more likely to shift innovation over to products. Unions appear to have a much more positive impact in manufacturing than in the service industries, where they may even put the brakes on technological change. Further research is needed to understand this divide. For example, could it be that there is less room for product innovation in the service industries than in manufacturing, and the less favourable impacts of unions in service industries reflect their avoidance of labour-saving technologies? Or is this just a reflection of the structures and boundaries of firms in the service sector, where they have less to gain from a formalisation of routines for consultation and cooperation?

Finally, unions need to acknowledge that they should contribute to a rewarding and challenging work environment for all workers, not only for the weakest groups because it is not in the interest of the unions that the best people move to (non-unionised or weakly unionised) competitors. In such cases, unions have a detrimental effect on innovation and technological development. Furthermore, unions are not homogenous and can have conflicting interests. Some of the positive impacts of unions rest on the ability to encompass many different groups of workers and solve conflicts internally.

It is precisely on this subject that research is sorely lacking. We have little information on how unions contribute to the flow of resources and knowledge between firms as workers receive wage offers and move to more attractive jobs. By compressing wages in some sectors or firms, unions make poaching easier for competing industries and companies. Union wage policies also distort the price signal embedded in the wage. The causal study referred to in Section 8 gives a bleak picture of this in the U.S., but in the Scandinavian countries, which are more favourably inclined towards redistribution, equality and co-operation between unions and employers, the outcomes could be more positive.

Abowd, J. A. & Lemieux, T. (1993). The effects of product market competition on collective bargaining outcomes: The case of foreign competition in Canada. Quarterly Journal of Economics, 108, 983–1014.

Ackerberg, D., Caves, K., & Frazer, G. (2015). Identification properties of recent production function estimators. Econometrica, 83, 2411-51.

Addison, J. T., Teixeira, P., Evers, K., & Bellman, L. (2017). Collective bargaining and innovation in Germany: Cooperative industrial relations? Industrial Relations, 56, 73-121.

Aghion, P., Akcigit, U., & Howitt, P. (2014). What do we learn from Schumpeterian growth theory? In P. Aghion & S. Durlauf (Eds.), Handbook of Economic Growth,2,515-563. Amsterdam: Elsevier.

Azar, J., Marinescu, I., & Steinbaum, M. (2022). Labor market concentration. Journal of Human Resources, 57, S167-S199.

Barth, E. & Olsen, H. D. (2009). Monopsonistic discrimination and the gender wage gap. Labour Economics, 16, 589–597.

Barth, E., Moene, K., & Willumsen, F. (2014). The Scandinavian model – an interpretation. Journal of Public Economics,117, 60–72.

Barth, E., Bryson, A., & Dale-Olsen, H. (2020). Union density effects on productivity and wages. Economic Journal, 130, 1898-1936.

Barth, E., Bryson, A., & Dale-Olsen, H. (2024). Creative disruption - Technological innovation, labour demand, and the pandemic. IZA DP No. 15762.

Barth, E., Bryson, A., & Dale-Olsen, H. (2025). Turning non-members into members: Do public subsidies to union membership matter? Journal of Economic Behaviour & Organization, 229, 106855.

Bashkar, V., Manning, A., & To, T. (2002). Oligopsony and monopsonistic competition in labor markets. Journal of Economic Perspectives, 16, 155-174.

Berton, F., Dughera, S., & Ricci, A. (2021). Are unions detrimental to innovation? Theory and evidence. IZA Discussion Paper DP No. 14102.

Bhuller, M., Moene, K. O., Mogstad, M., & Vestad, O. L. (2022). Facts and fantasies about wage setting and collective bargaining. Journal of Economic Perspectives, 36, 29-52.

Bradley, D., Kim, I., & Tian, X. (2016). Do unions affect innovation? Management Science, 63, 2251-2271.

Braun, S. (2011). Unionisation structures, productivity and firm performance: New insights from a heterogeneous firm model. Labour Economics, 18, 120-129.

Breda, T. (2015). Firms’ rents, workers’ bargaining power and the union wage premium. Economic Journal, 125, 1616–52.

Bryson, A., Barth, E., & Dale-Olsen, H. (2013). Organizational change, worker well-being and the moderating role of trade unions. Industrial & Labor Relations Review, 66, 989-1011.

Bryson, A. & Dale-Olsen, H. (2021). Union effects on product and technological innovations. Research in Labor Economics, 49, 41–65.

Calmfors, L. & Driffill, J. (1988). Bargaining structure, corporatism, and macroeconomic performance. Economic Policy,6, 13-62.

Calmfors, L. (2025). Coordination of wage setting across sectors. Nordic Economic Policy Review.

Cardullo, G., Conti, M., & Sulis, G. (2015). Sunk capital, unions, and the hold-up problem: Theory and evidence from cross-country sectoral data. European Economic Review, 76, 253-274.

Cardullo, G., Conti, M., & Sulis, G. (2020). A model of unions, two-tier bargaining, and capital investment. Labour Economics, 67, 101936.

Cournot, A. A. (1897). Researches into the Mathematical Principles of the Theory of Wealth. (Originally published 1838, Recherches sur les Principes Mathématiques de la Théorie des Richesses). Translated by N. Bacon. New York: Macmillan Company.

Dahl, C. M., Le Maire, D., & Munch, J. R. (2013). Wage dispersion and decentralization of wage bargaining. Journal of Labor Economics, 31, 501-533.

Dale-Olsen, H. (2006). Wages, fringe benefits, and worker turnover. Labour Economics, 13, 87-106.

Dale-Olsen, H. (2021). Do unions contribute to creative destruction? PLOS ONE.

Dale-Olsen, H. (2024). The impact of new free trade agreements on incumbent firms and workers. Paper presented at EEA-ESEM 2023 in Barcelona.

DiNardo, J. & Lee, D. S. (2004). Economic impacts of new unionization on private sector employers: 1984–2001. Quarterly Journal of Economics, 119, 1383–1441.

Dinlersoz, E., Greenwood, J., & Hyatt, H. R. (2017). What businesses attract unions? Unionization over the life-cycle of U.S. establishments. Industrial & Labor Relations Review, 70, 733-766.

Dobbelaere, S. & Kiyota, K. (2018). Labor market imperfections, markups, and productivity in multinationals and exporters. Labour Economics, 53, 198-212.

Doucouliagos, C. & Laroche, P. (2013). Unions and innovation: New insights from cross-country evidence. Industrial Relations, 52, 467-491.

Dowrick, S. & Spencer, B. J. (1994). Union attitudes to labor-saving innovation: When are unions Luddites? Journal of Labor Economics, 12, 316-344.

Dodini, S., Salvanes, K. G., & Willén, A. (2022). The dynamics of power in labor markets: Monopolistic unions versus monopsonistic employers. IZA Discussion Paper DP No. 15635.

Dodini, S., Salvanes, K. G., Willén, A., & Zhu, L. (2023a). The career effects of union membership. NHH Discussion Paper SAM 12/2023.

Dodini, S., Stansbury, A., & Willén, A. (2023b). How do firms respond to unions? IZA Discussion Paper DP No. 16697.

Dodini, S., Lovenheim, M., Salvanes, K. G., & Willén, A. (2024a). Monopsony, job tasks, and labor market concentration. Economic Journal, 134, 1914-1949.

Dodini, S., Willén, A., & Zhu, L. (2024b). The role of labor unions in immigrant integration. NHH Discussion Paper No. 24.

Falch, T. (2010). The elasticity of labor supply at the establishment level. Journal of Labor Economics, 28, 237-266.

Fjellberg, A. (2025). Changes in union density in the Nordic countries. Nordic Economic Policy Review.

Flaarønning, T. (2024). Union decline through extension of collective agreements. British Journal of Industrial Relations,63.

Frandsen, B. (2021). The surprising impacts of unionization: Evidence from matched employer-employee data. Journal of Labor Economics, 39, 861-894.

Freeman, R. & Medoff, J. (1984). What Do Unions Do? New York: Basic Books.

Gandhi, A., Navarro, S., & Rivers, D. (2020). On the identification of gross output production functions. Journal of Political Economy, 128, 2973-3016.

Goldin, I., Koutroumpis, P., Lafond, F., & Winkler, J. (2024). Why is productivity slowing down? Journal of Economic Literature, 62, 196–268.

Gomez, R., Gunderson, M., & Meltz, N. (2002). Comparing youth and adult desire for unionization in Canada. British Journal of Industrial Relations, 40, 519-542.

Grout, P. A. (1984). Investment and wages in the absence of binding contracts: A Nash bargaining approach. Econometrica, 52, 449–460.

Hayter, S. & Visser, J. (2021). Collective agreements: Extending labour protection. International Labour Review, 160, 169-195.

Haucap, J. & Wey, C. (2004). Unionisation structures and innovation incentives. Economic Journal, 114, C149–C165.

Heyes, J. & Stuart, M. (1998). Bargaining for skills: Trade unions and training at the workplace. British Journal of Industrial Relations, 36, 459-467.

Kauhanen, A., Maczulskij, T. & Riukula, K. (2023). Heterogeneous impacts of the decentralization of collective bargaining. IZA Discussion Paper DP No. 13867.

Kauhanen, A. (2023). The effects of the decentralization of collective bargaining on wages and wage dispersion: Evidence from the Finnish forest and IT industries. British Journal of Industrial Relations, 62, 319-334.

Kauhanen, A. (2025). Universally binding contracts and their effects on wages. Nordic Economic Policy Review.

Kostøl, F. B. & Svarstad, E. (2023). Trade unions and the process of technological change. Labour Economics, 84, 102386.

Kostøl, F. (2024). Do unions increase participation in further education? British Journal of Industrial Relations, 64.

Lee, D. S. & Mas, A. (2012). Long-run impacts of unions on firms: New evidence from financial markets, 1961–1999. Quarterly Journal of Economics, 127, 333–378.

Lommerud, K. E. & Straume, O. R. (2012). Employment protection versus flexicurity: On technology adoption in unionised firms. Scandinavian Journal of Economics, 114, 177–199.

Manning, A. (2003). Monopsony in Motion: Imperfect Competition in Labor Markets. Princeton, NJ: Princeton University Press.

Manning, A. (2011). Imperfect competition in the labor market. In D. Card & O. Ashenfelter (Eds.), Handbook of Labor Economics (Vol. 4B, pp. 973-1041). Amsterdam: Elsevier.

Mastekaasa, A. (2013). Unionization and certified sickness absence: Norwegian evidence. ILR Review, 66, 117-141.

McDonald, I. M. & Solow, R. M. (1981). Wage bargaining and employment. American Economic Review, 71, 896-908.

Menezes-Filho, N. & Van Reenen, J. (2003). Unions and innovation: A survey of the theory and empirical evidence. In J. Addison & C. Schnabel (Eds.), International Handbook of Trade Unions (pp. 293-334). Edward Elgar.

Moene, K. O., Wallerstein, M., & Hoel, M. (1993). Bargaining structure and economic performance. In R. J. Flanagan, K. O. Moene & M. Wallerstein (Eds.), Trade Union Behaviour, Pay Bargaining and Economic Performance (pp. 63-154). Oxford: Oxford University Press.

Moene, K. O. & Wallerstein, M. (1997). Pay inequality. Journal of Labor Economics, 15, 403-430.

OECD. (2015). The innovation imperative: Contributing to growth, productivity, and well-being. Paris: OECD.

OECD. (2017). Collective bargaining in a changing world of work. OECD Employment Outlook 2017, 125–173. Paris: OECD.

OECD. (2018). Ch.3 The role of collective bargaining systems for good labour market performance. Employment Outlook 2018, 73–122. Paris: OECD.

OECD. (2019). Negotiating Our Way Up: Collective Bargaining in a Changing World of Work. OECD Publishing. https://doi.org/10.1787/1fd2da34-en

OECD. (2021). OECD Skills Outlook 2021: Learning for Life. Paris: OECD Publishing.

Pencavel, J. (1984). The trade-off between wages and employment in trade unions’ objectives. Quarterly Journal of Economics, 99, 215-231.

Robinson, J. (1933). The Economics of Imperfect Competition. London: Macmillan.

Soares, A. C. (2020). Price-cost margins and bargaining power in the European Union. Empirical Economics, 59, 2093-2123.

Schnabel, C. (2020). Union membership and collective bargaining: Trends and determinants. In K. F. Zimmermann (Ed.), Handbook of Labor, Human Resources and Population Economics (pp. 1-37). Cham: Springer.

Smith, A. (1776). The Wealth of Nations. Hertfordshire: Wordsworth Editions Limited.

Svarstad, E. & Kostøl, F. B. (2021). Unions, collective agreements, and productivity: A firm-level analysis using Norwegian matched employer-employee panel data. British Journal of Industrial Relations, 60, 864-894.

Svarstad, E. (2023). Do unions care about low-paid workers? Evidence from Norway. Industrial Relations, 63, 417-441.

Thoresson, A. (2024). Employer concentration and wages for specialized workers. American Economic Journal: Applied Economics, 16, 447-479.

Ulph, A. & Ulph, D. (2001). Strategic innovation with complete and incomplete labour market contracts. Scandinavian Journal of Economics, 103, 265–282.

Vandaele, K. (2018). How can trade unions in Europe connect with young workers? In J. O’Reilly, J. Leschke, R. Ortlieb, K. Seeleib-Kaiser & P. Villa (Eds.), Youth Labour in Transition. Oxford University Press.

Willén, A. (2021). Decentralization of wage determination: Evidence from a national teacher reform. Journal of Public Economics, 198, 104388.

Willman, P., Bryson, A., & Forth, J. (2019). UK trade unions and the problem of collective action. British Journal of Industrial Relations, 58, 447-470.