Literature review

In line with the increased prevalence and voice of the labour movement, the past years have seen an increase in scholarly attention on this topic in political science. As public support for unions in the United States has reached its highest level since the 1960s (Bucci Reference Bucci2023), several authors have looked at how strikes affect public opinion. Zooming in on the 2018 teacher walkout in the United States, Hertel-Fernandez, Naidu and Reich (Reference Hertel-Fernandez, Naidu and Reich2021) find that parents of affected children have higher support for unionisation and express greater interest in labour action to improve working conditions. Unions can change behaviour as well as opinion: Lyon and Schaffner (Reference Lyon and Schaffner2021) find that state-wide union prevalence and social ties to unions can lead non-members to engage more in collective forms of political participation. However, effects may be conditional on circumstantial factors, such as the sector of employment: focusing on the 2019 grocery store strike, Hertel-Fernandez (Reference Hertel-Fernandez2025) finds only a limited effect of strike exposure on changes in attitudes and behaviours related to the labour movement. Under the right circumstances, labour strikes can affect public policy as well as public opinion and behaviour. Focusing on Latin America, Kim, Kim and Villegas (Reference Kim, Kim and Villegas2020) find that the confluence of labour strikes and mass protests can motivate governments to expand social spending.

Of particular interest for this paper due to its focus on globalisation as a cause of strike activity is work by Abouharb and Fordham (Reference Abouharb and Fordham2020). Focusing on America, they find that states that experience greater uncertainty from exposure to trade also experience more strikes than states where the effect is less ambiguous. This is because uncertainty muddles the ability of workers and firms to reach a state of equilibrium (ie a deal between negotiating parties) without costly action. When both sides are uncertain about the prospective gains and willingness to bear costs of the other, and knowing each side has an incentive to misrepresent their true negotiation parameters, resolution without escalation becomes less likely. Uncertainty therefore drives union activity. Building on work by Tsebelis and Lange (Reference Tsebelis, Lange and Sanford1995), Piazza (Reference Piazza2005) relatedly argues that the effect of globalisation on strike activity is mediated by pre-existing union density. In places where globalisation creates uncertainty between workers and employers about union strength – either because strong unions weaken or weak unions strengthen – strike incidence increases as neither organised labour nor firms pre-empt strikes by agreeing to a bargain. The ‘moderate union strength’ hypothesis is similarly endorsed by Lindvall (Reference Lindvall2013), though he focuses on neither globalisation nor partisanship. While important contributions to the study of globalisation and union organising, these studies hold constant circumstantial political factors, thus potentially obscuring important heterogeneous effects.

By contrast, Hamann, Johnston and Kelly (Reference Hamann, Johnston and Kelly2012) do look at the political determinants of union activity, though without considering the role played by the position of the state in the global economy. They find that general strikes against the government are less likely when unions are incorporated into the policy-making process, when social-democratic parties are dominant, and when governments hold a minority in parliament. Although general strikes, being aimed directly at governments, are overtly political, total strike activity is also likely affected by politics, as labour-friendly administrations can pursue policies that address labour’s demands. Schmidt (Reference Schmidt and Weber2001) and Cooper, Ellem and Wright (Reference Cooper, Ellem, Wright, Head and Crowley2015) both find that labour movements have a window of opportunity to affect policy directly when left-wing governments are in power but are often made outsiders under conservative rule. Strikes become necessary under the latter condition, but under the former, unions have an inside track to affect policy. A similar argument comes from Brandl and Traxler (Reference Brandl and Traxler2010), who analyse labour action as a strategic choice of union leadership that faces divergent pressures from rank-and-file members, employers, and political elites. Ceteris paribus, unions prefer to achieve their aims without militant action but can only do so when left-wing governments are in power. Furthermore, they look at globalisation, arguing that economic openness could dampen union activity due to the territorial limits of state regulation in an open economy. However, they do not explicitly model the interaction between government partisanship and openness, though such an interactive effect is implied by their various hypotheses. By contrast, Kwon and Pontusson (Reference Kwon and Pontusson2010) examine how union density and globalisation interact to lead to partisan convergence or divergence in social spending. They find that globalisation leads to government convergence in socio-economic policy, as also argued in this paper, but only when union strength is weak. It should be noted here that union strength itself could be a consequence of exposure to globalisation.

Pilati (Reference Pilati2025) similarly focuses on political factors. Drawing on a rich dataset of labour and social action occurring in Italy, they show that militant union activity increases as legislation affecting workers’ rights, wages, working conditions, and precarious employment are being discussed by the government. They also find that ‘the political orientation of elites may influence how unions and non-worker actors respond, as their reactions may differ depending on whether the government is right-wing or left-wing, due to unions’ historical political ties with certain parties’ (Pilati, Reference Pilati2025: 15). While the aforementioned studies look at either general strikes (Hamann, Johnston and Kelly Reference Hamann, Johnston and Kelly2012; Lindvall Reference Lindvall2013) or workplace economic strikes (Abouharb and Fordham Reference Abouharb and Fordham2020; Brand and Traxler Reference Brandl and Traxler2010; Piazza Reference Piazza2005), Höltmann, Borbáth and Hutter (Reference Höltmann, Borbáth and Hutter2025) examine under which circumstances unions pursue one or the other. They find an inverted relationship between union strength and corporatism and the prevalence of general strikes and broader protest. This suggests that unions are more likely to stick to workplace activity when national institutions make this the best way for them to improve their societal support.

Previous contributions have separately studied the consequences of government partisanship and globalisation in isolation for strike activity and labour organising. However, the extent to which these things might interact is unclear based on the existing literature. While this interaction is not tested in the studies discussed above, they provide a solid foundation on which to build hypotheses regarding this interaction. In the following section, I theorise how economic globalisation, especially trade penetration, might interact with government partisanship in changing union activity and labour strikes.

Theory and hypotheses

Because left-wing parties are more closely aligned with the labour movement and unions, one could reasonably expect strike activity to be lower when left-wing parties are in power. Left-wing governments face fewer politically motivated general strikes on average (Hamann, Johnston and Kelly Reference Hamann, Johnston and Kelly2012). They also have more leeway to pursue their preferred policies and can achieve strong macroeconomic performance when labour unions are strong (Alvarez, Garrett and Lange Reference Alvarez, Garrett and Lange1991). Governments controlled by left-wing parties will pursue labour market and industrial policies broadly in line with the preferences of their constituents, historically workers and unions, including active labour market policies, health and safety regulations, and social protections (Martin and Swank Reference Martin and Swank2004; Nelson Reference Nelson2013). Therefore, there will be a lower incentive to engage in strike activity: militancy is pursued only when peaceful means of achieving unions’ aims are expected to yield worse payoffs (Brandl and Traxler Reference Brandl and Traxler2010).

As noted by Alvarez, Garret and Lange (Reference Alvarez, Garrett and Lange1991: 542), ‘unions’ expectations about the types of policies governments may pursue are likely to have significant consequences for the behaviour of labour movements’, and ‘policies preferred by leftist governments will provide incentives for unions voluntarily to restrain their market militancy’ (ibid: 540). Even when conflicts do emerge, labour unions have an inside track with left-wing parties through their recruitment pipeline: leadership positions in labour movements and unions are a comparatively common background for left-wing politicians (Ellis, Horowitz and Stam Reference Ellis, Horowitz and Stam2015). This means conflicts could be resolved through informal channels, rather than strikes (Schmidt Reference Schmidt and Weber2001). All things equal, it is therefore reasonable to assume that governments dominated by left-wing parties will see decreased strike activity, not just as it regards politically motivated general strikes (Hamann, Johnston and Kelly Reference Hamann, Johnston and Kelly2012; Lindvall Reference Lindvall2013; Pilati Reference Pilati2025) but economic strike activity as well, due to the types of policies they typically pursue. These observations are broadly in line with seminal work on power resource theory, which states that left-wing governing parties are able and willing to mobilise government resources to support organised labour in its demands vis-à-vis employers (Korpi and Shalev Reference Korpi and Shalev1979).

However, the ability of left-wing governments to respond to the demands of labour is constrained by external factors. In the context of increasing economic globalisation, concerns over firms relocating production abroad and competing on international markets may keep governments from effectively addressing workers’ concerns. While the so-called ‘race to the bottom’ argument has received mixed empirical support (Mosely Reference Mosley2017a, Reference Mosley2017c; Mosley and Uno Reference Mosley and Uno2007; Urpelainen Reference Urpelainen2010), it is also true that the possibility of offshore production has improved the bargaining position of firms and investors over states and workers (Levy Reference Levy2005; Owen Reference Owen2017; Rickard Reference Rickard2022). Furthermore, even without active government regulation, a strong labour movement in and of itself might lead firms to pick up shop and relocate. In the context of global value chains and ‘arm-length production’, this issue has been especially salient (Mosley Reference Mosley2017b). If the marginal utility of moving or outsourcing production to a location with low wages, poor rights protections, and acquiescent labour exceeds the marginal utility of not moving while paying high wages and adhering to local labour rights, firms have a strong incentive to do so in a competitive, open economy.

Globalisation and free trade are therefore salient topics with those in the labour movement and could be associated with increased strike activity as workers protest their increased uncertainty and precarity (Abouharb and Fordham Reference Abouharb and Fordham2020; Piazza Reference Piazza2005). As noted above, in a closed economy left-wing parties can pursue policies in line with the interests of labour and side with them in tripartite decision making. Such policies help take away push factors for unions to engage in strike activity (Brandl and Traxler Reference Brandl and Traxler2010; Hamann, Johnston and Kelly Reference Hamann, Johnston and Kelly2012). However, as openness to the global economy increases, the ability of left-wing governments to pursue these policies dwindles. As trade penetration expands, firms can more easily relocate abroad, benefit from cheap labour, and export products back to home markets. Left-wing governments, who have an incentive to keep employment opportunities at home, are likely to give policy concessions to firms to promote job creation in an open economy (Shin Reference Shin2017). However, such concessions may run counter to the wider preferences of unionised labour. Indeed, as noted by Piazza (Reference Piazza2005: 295, emphasis added):

However under a situation with a left-wing government and low levels of trade penetration, unions need not go on strike to achieve favourable working conditions (Brandl and Traxler Reference Brandl and Traxler2010; Cooper, Ellem and Wright Reference Cooper, Ellem, Wright, Head and Crowley2015; Martin and Swank Reference Martin and Swank2004; Nelson Reference Nelson2013; Schmidt Reference Schmidt and Weber2001); at high levels of trade penetration, this will no longer be the case. To keep firms from relocating abroad, left-wing governments will need to provide them policy concessions that they otherwise would not be willing to give. This is borne out by empirical evidence in other studies of government partisanship and open economy politics. For example, Pinto (Reference Pinto2013) finds that left-wing governments are more likely to adopt policies that incentivise foreign direct investment, looking to bring high-productivity jobs into the country. Similarly, Shin (Reference Shin2017) finds that left-wing governments, while imposing higher tax rates on firms than right-wing governments, impose fewer restrictions on shifting profits abroad to incentivise firms to keep employment levels up. In the post-1990 period, which was also characterised by decreasing unionisation rates, the convergence of socio-economic policy from left- and right-wing governments is generally attributed to the dual pressures of globalisation and union decline (Kwon and Pontusson Reference Kwon and Pontusson2010; Rodrik Reference Rodrik1997). Subsequently, while economic strike activity is likely lower under left-wing than right-wing governments in a closed economy, in an open economy this need not apply due to the competitive pressures of global markets.

In addition to pursuing pro-labour union policies, left-wing parties typically also implement generous welfare state regimes and active labour market policies. While such policies attenuate the effect of free trade on populist backlash, they do not fully neutralise it (Swank and Betz Reference Swank and Betz2003). All things equal, workers prefer gainful employment over being on welfare due to the stigma associated with the latter (Handler and Hollingsworth Reference Handler and Hollingsworth1969). Another reason workers will prefer employment in industry union jobs over welfare or employment in precarious service jobs is the economic security and benefits associated with the former. While welfare and job transitioning can cushion the negative effects of job loss, they do not fully replace its economic benefits or social-psychological appeal, especially for male workers (Clark, Khoban and Zucker Reference Clark, Khoban and Zucker2025). Furthermore, changes in the left’s position on free trade and welfare become salient for workers as left-wing parties change their electoral tactics, as exemplified by the New Labour movement in the United Kingdom and the pivot of social-democratic parties to the ‘Third Way’ ideology more broadly (Kim Reference Kim2025).

Considering the above, this note tests for the conditional effect of globalisation on the relationship between government partisanship and labour strikes, which moves beyond the current literature by integrating these two causes of labour (dis)satisfaction into a single framework.

Data

This research note uses data from the Comparative Political Data Set (Armingeon et al. Reference Armingeon, Engler, Leemann and Weisstanner2024a, Reference Armingeon, Engler, Leemann and Weisstanner2024b), which covers 23 OECD countries between 1960 and 1990 and 36 OECD and EU countries between 1990 and 2022. The main dependent variable of interest is the number of working days lost due to strike activity per 1000 workers in a given country-year. The main independent variables of interest are total trade (imports and exports) as percentage of GDP (in percentage points) and the proportion of cabinet posts held by left-wing parties (in percentage points). Additional controls include real GDP growth in current US dollars, inflation, and unemployment. CPDS sources these variables from the ILO, Penn World Tables, the OECD, and AMECO, as well as calculations by the CPDS authors themselves. In addition, the empirical analysis includes country and year-fixed effects, and cluster standard errors at the country level. Models are estimated using ordinary least squares regression. A table with an overview of descriptive statistics is provided in the Online Appendix, section E.

All control variables are selected in line with best practices recommended by the methodological literature, selecting on prior common causes of the dependent and independent variables, while avoiding post-treatment bias (Wysocki, Lawson and Rhemtulla Reference Wysocki, Lawson and Rhemtulla2022). The main threat to inference would be if despite this selection, prior common causes of strike activity and left-wing government representation would continue to exist. In such a case, correlation would exist between the dependent and independent variables on account of their prior common cause instead of a direct causal relationship. Similarly, the inclusion in the covariates of ‘bad controls’, such as those that are affected by the independent variables, might bias the estimates (Agnoline, Colantone and Stanig Reference Agnolin, Colantone and Stanig2025). A further discussion of the selection of control variables is included in the Online Appendix. In addition to the included variables and further robustness extensions in the Online Appendix, country- and year-fixed effects account for baseline differences between states and non-linear time trends.