What Explains Differences in Minimum Wage Growth Between EU Member States?

1 Introduction

Twenty-two of the 27 EU member states have national minimum wages. This is also true for the UK, which is included in this study. The remaining five EU member states have collectively negotiated minimum wages. Differences in the nature of these two types of minimum wage regimes have been widely noted (Boeri 2012; Fernández-Macías and Vacas-Soriano 2016; Garnero, Kampelmann, and Rycx 2015b). However, there is also considerable variety within the group of EU member states with national minimum wages. This concerns variation in the way minimum wages are adjusted. For example, Schulten (2014) distinguishes between four types of adjustment regimes. The OECD-AIAS-ICTWSS database identifies differences in setting the minimum wage according to 12 variables (OECD and AIAS 2021). Eurofound (2023a) captures the different national adjustment mechanisms for the minimum wage by using 16 variables for actors and decision criteria. The variety within EU member states also concerns the scope of minimum wage adjustments. In particular, member states that were part of the EU before 2004 display much more muted minimum wage growth than those that acceded to the EU in 2004 or later.

This article asks whether there is a connection between the institutional variation in minimum wage adjustment and the differences in minimum wage growth. It investigates differences in minimum wage growth between member states in general, and differences in growth rates between the group of member states that belonged to the EU before 2004 and the group of those that acceded in 2004 or later in particular. It tests whether different minimum wage growth rates are the result of a catch-up process. This seems especially pertinent in the context of the accession of new EU member states, which enjoyed accelerated economic growth by joining the internal market. The article also examines whether macroeconomic growth imperatives, in particular the degree to which GDP growth depends on net exports or on consumption, affect minimum wage policy as a result of muted or accelerated wage growth. Finally, it tests hypotheses on how actors such as government and social partners may affect minimum wage policy, for example, by introducing electoral, policy or redistributive considerations into minimum wage adjustments.

The variation in minimum wage adjustment regimes across EU member states is not surprising in light of overall variation in political economies inside the EU. EU member states are marked by persistent differences in the areas of labour markets, industrial relations and government regulation, despite the interdependencies created by EU integration, most notably in the context of the internal market and the currency union. The patterns of institutional continuity and change resulting from forces of convergence on the one hand and idiosyncratic historical trajectories and distinctive interests and power distributions among actors on the other hand have been analysed and documented in various fields of comparative political economy, including industrial relations (e.g. Ferner and Hyman 1998; Streeck 1998), the literature on varieties of capitalism (e.g. Hall and Soskice 2001), and welfare state, European integration and transition studies (e.g. Bohle and Greskovit 2012; Cernat 2006; Höpner and Schäfer 2010; Nölke and Vliegenthart 2009; Scharpf 2002). Minimum wage policies do not feature prominently in any of these fields.

National minimum wages are a widely researched topic in labour economics, where the main focus is the employment effects of the minimum wage (see for example Caliendo et al. 2018; Card and Krueger 1994; Dolton, Bondibene, and Stops 2015; Dube 2019; Neumark and Wascher 2008). Economic research also increasingly focuses on effects of the minimum wage on other parameters than employment (Schmitt 2015). This includes, for example, effects on prices (Harasztosi and Lindner 2019; Link 2022), on profitability and investments (Bossler et al. 2020) and on in-work poverty (Bruckmeier and Bruttel 2021). In political science and economic sociology, research on the minimum wage is focused mainly on the politics of its introduction and its interaction with industrial relations (see for example Bosch et al. 2021; Grimshaw and Bosch 2013; Mabbett 2016; Marx and Starke 2017; Meyer 2016). Less researched are the nature and impact of the factors in setting the minimum wage once it is introduced. The exceptions are studies by Boeri (2012) and by Arpaia et al. (2017) who both examine different aspects of adjustment regimes in their impact on minimum wage growth.

With continuing efforts towards EU-wide coordination of minimum wage policy and the adoption of the EU directive on adequate minimum wages in the European Union in October 2022, a better understanding of the determinants of minimum wage growth in EU member states is helpful for effective policy formulation. If adjustment differentials in minimum wages were a mere result of different economic development speeds, there would not be a strong case for policy intervention to begin with, as minimum wage levels would become harmonized by economic convergence over time. If minimum wage policy was inherently linked to growth strategies, any initiatives for harmonizing minimum wages that ignored this link would hamper economic growth. Finally, if actors were found to determine minimum wage growth, policy proposals would have to acknowledge their motives and intentions and assess the compatibility of these with the objective of harmonisation.

The article is structured as follows: It first presents the development of minimum wages across EU member states. It then develops hypotheses to explain the differences in minimum wage growth based on economic development rates, growth models and actor interests. After describing the data and the method, I examine the hypotheses descriptively and in multivariate analyses and discuss the results. The concluding section summarizes the findings.

2 Different Minimum Wage Trajectories in Old and New EU Member States

The study comprises 21 EU member states with national minimum wages. Missing are Cyprus and Malta.^[1] Included is the United Kingdom which was an EU member state throughout the observation period. The time period covered spans the years 2000–2020. It ends with the onset of the Covid pandemic. The outcome variable of interest is the growth of the gross nominal minimum wage. Its value determines the extent to which minimum wage recipients receive (hourly/monthly) wage gains before taxes and transfers. Overall, the average hourly (monthly) minimum wage growth per year across the sample was 6.9 per cent (6.4 per cent) per year (cf. Table 1). It was higher than the average wage growth of 5 per cent per year. Average inflation was 2.8 per cent per year. GDP grew at 2.4 per cent and employment at 0.5 per cent per year. The average Kaitz index, which is the ratio of the minimum wage to the median wage of full-time employees, was just under 48 per cent, meaning that minimum wages amounted to approximately 48 per cent of the median wages across the sample.

Table 1 shows countries in ascending order of minimum wage growth. The sample countries line up along the axis of old and new membership of the EU. While the 10 countries in the sample that were members of the EU before 2004 (old EU member states) form the upper half of the table, with lower growth rates, the countries that became members of the EU in 2004 or after (new EU member states) form the lower half of the table, with higher growth rates. On average, the hourly (monthly) minimum wage in old EU member states grew by 2.9 per cent (2.8 per cent) per year. The average hourly (monthly) minimum wage in new EU member states grew by 10.3 per cent (10.1 per cent) per year.

There are two outliers in the sample. Greece is the only country that displays episodes of negative minimum wage growth during the observation period, a result of the reduction in the nominal minimum wage of 22 per cent in 2012 and its subsequent freeze as part of the memorandum of understanding for receiving financial assistance from the EU and the IMF in the wake of the Eurozone crisis (ECB 2012). Romania, on the other end, had the strongest minimum wage growth on average, with single increases of more than 100 per cent at the beginning of the observation period, when the country was not yet a member of the EU.

The result of the different magnitudes of minimum wage growth in old and new EU member states was a converging trend of the ratio of minimum wages to median wages (cf. Figure 1). While the average Kaitz index in old EU member states shows comparatively little variability during the 20-year observation period, new EU member states have closed the gap to old EU member states with an overall increase in their average Kaitz index. The Kaitz index of old EU member states was approximately 12 percentage points higher than that in new EU member states in 2000. This difference had decreased to 2.5 percentage points in 2020. During the period 2013 to 2017, the average Kaitz index in new EU member states even exceeded the one in old EU member states. In the following, I will discuss and then test three hypotheses that each provides stand-alone explanations for the differences in minimum wage growth trajectories between EU member states: the dynamic catch-up process of new EU member states, the imperatives of growth models and the influence of actors.

Figure 1:

Kaitz index 2000–2020. Note: The Kaitz index comprises data on full time employees only. Source: OECD; own calculations.

3 Explanatory Approach A: A Dynamic Catch-Up Process

With the accession of new member states to the EU and to the single market, their economies grew on average at a higher rate than in old EU member states and their GDP per capita also grew at a higher rate than the EU average (Alcidi 2019). As a result, the minimum wage may have grown in line with fast economic growth. Minimum wage adjustments may thus be correlated with the overall state of economic conditions. Countries with above-average incomes may display relatively lower minimum wage adjustments as the result of the plateauing of economic development. Countries with below-average incomes may display relatively higher minimum wage adjustments as the result of a catch-up dynamic. We should therefore expect higher minimum wage growth in countries with lower GDP per capita and higher GDP growth, and lower minimum wage growth in countries with higher GDP per capita and lower GDP growth. If GDP growth can significantly explain a large part of the difference in minimum wage growth between old and new EU member states, this hypothesis cannot be rejected.

4 Explanatory Approach B: An Effect of Growth Models on the Minimum Wage

The growth model approach to comparative political economy (Baccaro and Pontusson 2016; Hassel, Palier, and Avlijas 2020) characterizes the different ways that advanced capitalist economies generate growth. The post-WWII welfare states of Western Europe all relied on wage-led demand for their economies to grow. Real wage growth led to rising household incomes translating to higher consumer demand, which in turn allowed firms to expand production. With the oil price shock and stagflation in the 1970s and with capital mobility and international competition increasing throughout the following decades, national incomes shifted from labour to capital across OECD countries (IMF 2017). With the decline in the wage share, the wage-led growth model reached its limits.

Stockhammer and Onaran (2012) distinguish between wage-led, profit-led, debt-led and export-led growth. Thus, consumer demand, if not supported through real wage growth, can also be supported through credit or government transfers (Crouch 2009). Alternative drivers of demand are investment, government consumption and net exports (Baccaro and Pontusson 2016). Baccaro and Pontusson show empirically that countries have chosen different sets of these available growth strategies, with the UK, for example, opting for a model relying on household consumption, whereas Germany opted for a model dependent on exports (see also Höpner 2019). Other countries, like Italy and Sweden, are less one-sided and more balanced in terms of their growth strategy. For purposes of this article, the authors’ observations on the connection of growth models, wages and household incomes are pertinent.

If growth models impact wage developments according to growth priorities, it seems plausible that their influence also extends to minimum wage adjustments. In the case of consumption-led growth, minimum wages may constitute an important support for consumer demand, especially if a sizable portion of the labour force is working at the minimum wage. If the propensity to consume is higher for low-income households than for high-income households, a proposition that empirical studies have to some extent supported (Aitken, Dolton, and Wadsworth 2014; Arpaia et al. 2017), a higher minimum wage may also have a relatively higher impact on consumption than a proportional increase in income for high earners. Strong minimum wage growth may thus support a consumption-led growth model.

In the case of export-led growth, the inverse logic applies, as restraint in minimum wage adjustments may support export-led growth. Export industries do not normally belong to the low-wage sector of the economy and therefore do not usually display a high share of minimum wage workers. By keeping consumer prices low, a low wage level in the services sector, where minimum wages are most common, may facilitate wage restraint further up the pay scale in export industries, however (Hassel 2014; Palier and Thelen 2010). In this sense, low minimum wage growth may support an export-led growth model.

For growth models to explain the relatively stronger growth of the minimum wage in new EU member states, they would have to display a clear tendency towards a consumption-dependent growth model. In contrast, the lower growth of the minimum wage in old EU member states would have to be correlated with a clear tendency towards an export-dependent growth model.

5 Explanatory Approach C: Actors and Their Interest in the Minimum Wage

Minimum wage adjustments are policy decisions that require discretionary action by actors (sometimes indirectly by establishing an adjustment rule). Hence, the size of adjustments may primarily vary with the type of actor responsible for it. The difference in adjustment rates between EU member states and between the groups of old and new EU member states may thus be explained by the dominance of different types of actors in the adjustment process.

Several authors have developed typologies to describe minimum wage adjustment mechanisms (Arpaia et al. 2017; Boeri 2012; Eurofound 2016, 2023b; Garnero, Kampelmann, and Rycx 2015a; Schulten 2014). The prime purpose of these typologies is to reduce the diversity of real types and arrive at classifications of minimum wage regimes. In all of these typologies, sets of actors are used as criteria to identify regimes. Boeri (2012) and Arpaia et al. (2017) have drawn up typologies specifically to examine the connection between actors and minimum wage outcomes. They suggest that minimum wage adjustments vary systematically with the involved actors and the adjustment processes. According to the results of Boeri (2012), trade union involvement has a positive effect on minimum wage adjustments in comparison to adjustments without the participation of trade unions as collectively negotiated minimum wages are higher on average than minimum wages administered by government. According to Arpaia et al. (2017), who examine 17 EU member states with national minimum wages, rules-based adjustment regimes display lower minimum wage adjustments than non-rules-based adjustment regimes. According to the authors, rules lower the likelihood that adjustments follow criteria other than economic fundamentals, especially by offering policy-makers the chance to engage in opportunistic behaviour in the face of elections.

To formulate stylized hypotheses that can be tested empirically, I distinguish four actor constellations in minimum wage adjustments: adjustments that are solely set by government, those that are set by or in consensus with the social partners, those that are administered by expert commissions and those that follow indexation rules.

Governments may follow their own agenda in adjusting the minimum wage. In particular, governments may use minimum wage raises as a tool for re-election. Arpaia et al. (2017) point out that their empirical findings suggest that the presence of elections leads to more redistributive minimum wage adjustments, i.e. higher adjustments than the adjustments in periods without elections (ibid.: 17). We would thus expect minimum wage growth to be higher in election years. To explain the divergent minimum wage growth paths in old and new EU member states, governments in new EU member states would have to be more responsive to elections than those in old EU member states.

In contrast, the economic policy orientation of the government may have a substantial influence on minimum wage growth. Right-leaning governments may favour supply-side policies and a lower minimum wage and left-leaning governments may favour a Keynesian demand-side approach with higher minimum wages. As a result, we would expect minimum wage raises to be higher the more left-leaning the economic policy orientation of a government is. Again, to explain the different minimum wage policies in old and new EU member states, economic policy in new EU member states would have to be systematically more left-leaning.

From a political economy point of view, adjusting the minimum wage is a distributional conflict between labour and capital. It is in workers’ interest to receive higher wages, and it is in businesses’ interest to limit labour costs. It has been shown empirically that trade unions prompt a compression of the wage distribution (Ahlquist 2017; Baccaro 2011; Freeman and Medoff 1984). Pushing up the minimum wage is one lever to achieve this (Freeman 1996). For employers, the obvious interest is to keep labour costs down and therefore to keep minimum wage adjustments to a minimum. We assume that these conflicting interests are represented by trade unions and employers’ associations and that both promote their own interest when they are involved in adjusting the minimum wage.

The power distribution between trade unions and employers are considered to depend on their organizational strength and other institutional power resources (e.g. labour law). Union density and coverage of collective agreements indicate the organizational and institutional strength of trade unions (see for example Lehndorf, Dribbusch, and Schulten 2018). We should thus expect minimum wage adjustments involving social partners to rise with increasing union density and collective bargaining coverage. The higher adjustments in new EU member states could thus be caused by high union strength. For this to be true, trade unions’ organizational strength would have to be higher in new EU member states than in old EU member states, an empirically questionable proposition to start with, and social partners would have to be involved in minimum wage adjustments frequently enough to matter.

Expert commissions are a more technocratic way of resolving the distributional issue of adjusting the minimum wage. Expert commissions are made up of independent academic members and partially, such as in the case of the Low Pay Commission in the UK, of members with business and trade union backgrounds. Their mandate from government is to make recommendations for adjusting the minimum wage based on empirical evaluations and model-based projections.

Indexation in turn automates the adjustment process to a large extent and limits discretionary interventions by linking the minimum wage to macroeconomic parameters. It usually follows average wage growth and/or inflation to allow minimum wage workers to partake in overall economic growth and to keep their real wages stable. As a result, both types of adjustment will not produce minimum wage adjustments that diverge substantially from economic fundamentals.

The above stylized hypotheses propose that social partners and governments introduce redistributive, electoral or policy considerations into minimum wage adjustments. They suggest that expert commissions follow macroeconomic parameters more closely than social partners and governments, while indexation mechanisms follow them exactly. As a result, we can expect average adjustments governed by social partners, governments and, to a lesser extent, expert commissions to be higher than those regulated by indexation. Accordingly, if EU member states display a higher number of adjustments managed by government or social partners than other EU member states, higher minimum wage growth can be expected.

6 Data and Method

The data combine information on minimum wage levels and macroeconomic parameters with information on actors, elections and the political orientation of governments across countries and over time (cross-sectional time-series data). The information on hourly minimum wage levels stems from the WSI minimum wage database of the Hans Böckler Foundation (WSI 2023). Information on monthly minimum wage levels comes from Eurostat. Information on macroeconomic parameters is taken from Eurostat, from the Organisation for Economic Co-operation and Development (OECD) and from the AMECO macroeconomic database of the European Commission’s Directorate General for Economic and Financial Affairs. Information on the elections and policy orientation of governments comes from the Database of Political Institutions (DPI) of the Inter-American Development Bank (Cruz, Keefer, and Scartascini 2021). Information on actors and processes stems from Eurofound’s EurWORK database on wages, working time and collective disputes (Eurofound 2023a). The EurWORK database provides country-level data on wages, working time and collective disputes, including on minimum wage setting, and relies on input from Eurofound’s network of national correspondents. It covers all 27 EU member states, the UK and Norway. Unlike the information on minimum wage setting from the OECD-AIAS-ICTWSS database, which covers only the year 2019, the EurWORK database provides yearly data for the entire observation period. It offers variables to identify the four types of actor-centred adjustment for which hypotheses were formulated. It also allows to identify a fifth type of adjustment, namely to adjust the minimum wage by setting targets for its level over time. The data thus comprises five types of adjustment which are described in Table 2. Countries are uniquely assigned to each of the five types for each year. There is variation across time if actor constellations change within countries over time or if adjustments deviate for individual years from the regular adjustment pattern. The classification of each country in each year on the basis of the EurWORK database is documented in Table A.1 of the appendix. The sample comprises 426 observations (country/year cells).

It is important to keep in mind that the typology of adjustment settings and the classification of member states into the adjustment types reduce the complexity of reality. Herein lies an advantage, but it also creates limitations. Where institutional settings in reality often comprise multiple elements (see for a detailed overview Eurofound 2024), this study identifies the dominant one in order to ensure that every country/year cell has a unique value. This discounts elements of the real cases. The case of France is an example: Its minimum wage follows an indexation rule, but the government has the option to add to the indexation result a so-called coup de pouce as it sees fit. For this study, France was classified as an indexation type in years in which it followed the indexation rule, and as a government type in years with a coup de pouce. However, there is also the Groupe d’Experts, an expert body, and the Commission Nationale de la Négociation Collective, a social partner committee, both of which have an advisory role in the adjustment process. As advisory bodies without a guaranteed say in the adjustment of the minimum wage per se, they were omitted in classifying France for this study. Their real role and impact, which may be indirect but distinct, is hence not reflected in the results of this study.

The following linear model describes the relationship between the outcome variable and the various explanatory and control variables:

∆NMW denotes nominal minimum wage growth for country i in year t. β ₁ to β ₅ stand for the coefficients of interest. They indicate the influence of variables that are arranged in five groups. δ _t designates year fixed effects, i.e. the difference of year t relative to the base year 2000 that is identical across all countries; it captures for example the effects of the financial crisis of 2008. μ _it is the error term. Variable GDP _it represents GDP growth for country i in year t. GrM _it represents the growth model variables, which are the contributions of net exports and private consumption to GDP growth respectively.^[2] The variable group Actors _it represents the setting types distinguished by actors indexation, social partners, government, expert commission and target. Pol _it contains the variables for the economic policy orientation of the government and for the occurrence of elections on the national level. Econ _it contains the control variables consumer price inflation (CPI), average wage growth, employment growth, Kaitz index and the square number of the Kaitz index value. The latter will detect potential non-linear effects of the Kaitz index. The control variables represent parameters that constitute probable correlations with minimum wage growth. Their selection follows Arpaia et al. (2017). The Kaitz index is added in order to test whether minimum wage growth is connected to relative levels of minimum wages to median wages.

An ordinary least squares (OLS) estimation was employed with standard errors clustered by country to account for the fact that observations within the same country are not fully independent of each other but serially correlated. The database contains 63 missings (17 per cent), mostly for the variables economic policy orientation and Kaitz index, which were imputed for the multivariate analysis by using multiple imputation (m = 20) by chained equations in Stata for the two groups of old and new EU member states separately (cf. White, Royston, and Wood 2011).

7 Descriptive Results

The potential explanatory factors developed in the hypotheses are first examined on the basis of descriptive data. Table 3 lists relevant indicators separately for old and new EU member states. Differences in GDP per capita and GDP growth between the two groups of countries are large. New EU member states show an average GDP per capita that is half that of old EU member states and a GDP growth rate that is almost twice as high, lending plausibility to the catch-up hypothesis. Union strength as measured by union density is higher in old EU member states, and considerably so when measured by collective bargaining coverage. It thus seems unlikely that union strength is behind the strong minimum wage growth in new EU member states. It also seems unlikely that a more left-leaning economic policy orientation is a decisive factor. Although economic policy is on average slightly more left-leaning in new EU member states, the difference from old EU member states is minor. The indicators for export- and consumption-dependency of economic growth partly support the growth model hypotheses. The respective shares in GDP of net exports and of private consumption show a higher reliance on private consumption of new EU member states and a higher reliance on net exports of old EU member states. However, the contribution of net exports to GDP growth is higher on average in new EU member states than in old EU member states.

Table 4 displays separate minimum wage, GDP and wage growth rates as well as CPI in new EU member states for country/year cells without EU membership, with EU membership outside the currency union and with Eurozone membership. GDP and minimum wage growth rates are highest in the years before becoming EU member. The catch-up hypothesis would suggest higher GDP growth rates after rather than before accession to the EU. However, the accession process in itself may have already generated significant additional GDP growth. Growth rates are roughly halved after accession to the EU, and further reduced in case of Eurozone membership. GDP and minimum wage growth rates are on different levels but move in parallel. Diminishing growth rates for both minimum wages and GDP suggest a plateauing effect in both.

Table 5 shows minimum wage growth rates by actor types in descending order, separately for old and new EU member states. Following the hypotheses, the variables for government and social partners in particular should be associated with higher growth rates than the variable for indexation. Additionally, minimum wages should be more frequently adjusted by social partners and governments in new EU member states than in old EU member states. With respect to new EU member states, these expectations are met. Adjustments driven by government and social partners display considerably higher growth rates than those driven by indexation. Both actors are more frequently responsible for minimum wage policy than their peers in old EU member states, where indexation is the most common adjustment regime. Old EU member states fit expectations less clearly, with government-driven adjustments displaying the lowest growth rates of all, contrary to the hypotheses.

The descriptive evidence thus provides preliminary support for all three hypotheses formulated earlier. The catch-up and the growth model hypotheses seem to hold up, and the relationship between actors and minimum wage growth in new EU member states can provide an explanation for their higher minimum wage growth compared to old EU member states.

8 Results from Multivariate Analyses

In order to control for endogeneity, the discussion will focus on the results from a multivariate analysis with a forward outcome variable (cf. Table 6). The direction of causality is unclear when outcome and independent variables are in the same year (cf. Table A.2 in the appendix). For example, it is not clear from a correlation between contemporaneous variables whether CPI influences minimum wage growth or vice versa. If the independent variables display correlations with minimum wage growth in the following year, the proposition is strengthened that the direction of influence is indeed from the independent to the outcome variable, as it is not logically possible that minimum wage growth in t + 1 affects any of the independent variables in t. The two year horizon also better reflects the reality of minimum wage adjustments where actors regularly decide on minimum wage levels for the following year on the basis of macroeconomic data of the current year. Table 6 shows the effects of the independent variables on minimum wage growth in the following year, t + 1. Using the linear model described above, effects on forward hourly and monthly minimum wage growth are shown for the entire sample (columns 1 and 2), for the sub-sample of old EU member states (columns 3 and 4) and for the sub-sample of new EU member states (columns 5 and 6).

For the entire sample and the sub-sample of old EU member states, minimum wage growth is statistically different and higher than in the base case of indexation if social partners are responsible for minimum wage adjustment. Adjustments by government and by experts are higher, but not statistically different from those of indexation across all samples. Adjustments by setting targets are significantly higher in the sub-sample of old EU member states. Measured in the same period (cf. Table A.2), government adjustment displayed a significant and large positive effect in the full sample and for new EU member states. This suggests that actors’ influence on minimum wage adjustments in t and t + 1 is not identical. The deviations in patterns may reflect the fact that adjustments by actors are implemented on different timelines, so that government adjustments come into force predominantly in t and not in t + 1.

With respect to macro-economic variables, consumer price inflation displays a correlation with minimum wage growth that is significantly different from zero in the full sample and in the sub-sample of new EU member states (cf. Table 6). Due to the lagged effect, these effect sizes can be considered to measure the causal effect of CPI on minimum wage adjustments. Hence, for every percentage point increase in CPI in t the minimum wage is adjusted by a little more (hourly minimum wage) or a little less (monthly minimum wage) than one percentage point in t + 1, thus keeping the real value of the minimum wage stable. The effect is present for new EU member states but not for old EU member states. Minimum wage adjustments in the latter group do not respond to CPI. For wage growth, statistical significance is limited to the full sample and non-existent for the sub-samples of old and new EU member states. Measured in the same period (cf. Table A.2), the coefficients for wage growth are larger and significant across all samples. This contrast between effects in t and t + 1 suggests that the direction of causality is mainly from minimum wage to wage growth, meaning that minimum wage growth contributes to overall wage growth, but overall wage growth is only of limited influence to minimum wage growth. For employment growth, there is a positive sign and weak significance in the sub-sample of old EU member states. This suggests that positive employment growth in the previous period tends to affect monthly minimum wage growth positively in old EU member states. The contribution of domestic consumption to GDP growth shows a weakly significant effect on minimum wage growth in the case of the full sample and for the monthly minimum wage for old EU member states. This suggests that an increase of consumption in GDP leads to higher minimum wage adjustments in the following year.

The effect of the Kaitz index is relatively large, negative and significant for the full sample and the sub-sample of new EU member states. This means that a higher Kaitz index leads to lower minimum wage growth in the following year. The very small positive, but significant effect of the square number of the Kaitz index suggests that the negative effect of the Kaitz index is slightly non-linear and becomes somewhat smaller with higher Kaitz index values, resulting in a leveling off effect where the decline in minimum wage increases becomes ever smaller. Effects of the political orientation of government and of elections on minimum wage adjustment are nonsignificant. This is true for the full sample and for both sub-samples.

In sum, the analysis so far has shown several factors that are statistically significant for minimum wage growth, of which adjustment by social partners, CPI and the level of the Kaitz index are the most distinct ones. Of these, it is CPI that contributes most clearly to explaining the difference in minimum wage growth between old and new EU member states as it affects minimum wage growth significantly in new EU member states but not in old EU member states. However, the constants in the two sub-samples also differ considerably. This reflects the difference in minimum wage growth that exists between the two sub-samples independent of the explanatory variables in the estimation. Although this suggests that a part of the higher minimum wage growth in new EU member states is beyond the reach of the hypotheses tested here, a further breakdown will allow to understand better the dynamics behind minimum wage growth in new EU member states.

In order to see whether and how the variables and the constant vary during the time before and after accession to the EU and depending on whether member states are part of the Eurozone or not, forward minimum wage growth in the group of new EU member states is estimated with the linear model described above, supplemented with interaction terms of all independent variables with dummy variables for no EU membership, for EU membership without common currency and for Eurozone membership (cf. Table 7). Indexation before accession to the EU is the reference case. Coefficients in columns (3) and (5) have to be interpreted relative to column (1), and likewise columns (4) and (6) relative to column (2). There is clear variation in adjustment patterns between statuses. First of all, note that EU and Eurozone membership display large nonsignificant negative coefficients that have to be viewed relative to the constants, meaning that minimum wage growth became overall much smaller with EU and Eurozone membership, a fact that was already visible in Table 4. Before accession to the EU, adjustment by government and social partners displayed nonsignificant positive effect sizes. These turn negative with EU and Eurozone membership, noticeably reducing the difference to indexation, meaning that adjustments by these actors also declined after entering the EU and the Eurozone. Only adjustments by target were still positively different from indexation during EU and Eurozone membership, albeit not statistically significant.

CPI had a large and significant impact on minimum wage growth before accession to the EU. A percentage point of additional inflation led to between 2.8 (hourly) and 2.1 (monthly) percentage points growth of the minimum wage. After accession to the EU, this is more or less completely reversed. In case of Eurozone membership, negative effect sizes increased slightly further, so that the effect size of CPI is approximately zero or even negative after new EU members became part of the Eurozone. In contrast, the weakly significant positive coefficients for wage growth suggest that average wage growth became a significant influence for minimum wage growth after EU and Eurozone membership. The reference for indexation thus seems to have changed from CPI to wage growth with the change of status from not being a member of the EU to being a member, and even more so with membership of the Eurozone. The negative effect of a high Kaitz index on minimum wage growth is limited to the period before accession to the EU. Finally, the orientation of economic policy seems to have become a salient factor with Eurozone membership, which would mean that a more left-wing economic policy of the government led to higher minimum wage growth. However, statistical significance is weak and only observable in hourly minimum wage growth.

9 Discussion of Results

The results of the multivariate analyses allow us to evaluate the explanatory power of the hypotheses formulated earlier. It is useful to distinguish between parameters that explain differences in minimum wage growth between individual countries and those parameters that provide explanations for the difference in minimum wage growth between old and new EU member states as groups. The catch-up hypothesis tries to explain the latter. The growth model and the actor-centred hypotheses provide explanations for differences between the two groups of member states, but they also provide explanations for growth differentials between countries or adjustment settings respectively, irrespective of their status as old or new EU member state.

Starting with the actor-centred hypotheses, the multivariate analysis shows that adjustments by social partners are significantly higher than those by indexation. This is true for the entire sample and for the sub-sample of old EU member states. There is a positive coefficient with similar size in the sub-sample of new EU member states, but with no statistical significance. In general, this means that social partner adjustment favoured the interests of labour in the distributional conflict of adjusting the minimum wage more than indexation. The fact that the effect differs between old and new EU member states in statistical significance but not much in size only partly aligns with the hypothesis from Section 5 that social partner adjustments rise with increasing union density and higher coverage of collective agreements. Both are clearly higher in old EU member states compared to new EU member states (cf. Table 3), but without translating into larger effect sizes for old EU members. The cause for social partners’ influence may thus not only lie in labour’s power resources but also in capital’s support for a minimum wage. It may be that consensual social partner adjustment was the adjustment type of choice predominantly in contexts where employers were interested in effects of the minimum wage that go beyond strictly distributional aspects between capital and labour, for example the prevention of cut-throat competition between businesses.

Adjustments by government were shown to be significantly higher than indexation for the full sample when measured in the same period (cf. Table A.2), but not in the following period (cf. Table 6). This difference may be caused by the timeline for implementation, with government adjustments more frequently implemented in period t rather than in period t + 1. Lower coefficients in old EU member states compared to new EU member states point to different patterns in minimum wage adjustment by government in both sub-samples. In old EU member states, governments were frequently involved in deflationary minimum wage policies. Minimum wages fell under conservative leadership in the years 2000–2004 in Spain and then again under conservative leadership in the years of the Eurozone crisis in both Spain (2011–2013) and Portugal (2012 and 2013). The relative subdued effect of government adjustment is further cemented by the fact that responsibility for minimum wage adjustments switched from government to social partners in Spain after 2004 and in Portugal after 2005 and in 2015 when social-democratic governments took over from conservative governments. Thus, there are deflationary minimum wage policies tied to conservative rule behind many cases of government adjustment in old EU member states, while social-democratic governments every so often handed over responsibility for adjustments to social partners.

It was hypothesized that government adjustment is motivated by elections and/or the orientation of the government’s economic policy. The multivariate analysis provides only scant support for either proposition. An explanation for why the economic policy orientation is not correlated with minimum wage growth may be that both left-leaning and right-leaning governments pursued minimum wage growth during the observation period. In contrast to Spain and Portugal in the earlier part of the observation period, right-leaning governments in new EU member states Hungary and Poland and in old EU member state UK promoted minimum wage increases in the later part of the observation period. In all three countries, policies for increasing the minimum wage were embedded in otherwise right-leaning economic policies. If this was indeed responsible for the null result of economic policy, it would align with findings from other studies. On the basis of analysing party manifestos of European political parties of the last 25 years, Cova (2023) finds that right-wing populist parties pay similar attention to minimum wages than left-wing parties.

Concerning elections, the main analysis assumes that elections in the same period as the minimum wage adjustment are those that are relevant for actors if they choose to include them in their adjustment calculus. However, this assumption may not be correct. It may also be that elections affect adjustment decisions early in advance of their occurrence. This may be the case if governments try to cater to the electorate ahead of elections by increasing the minimum wage. In contrast, elections may affect minimum wage adjustments only in post-election periods. This may be the case if minimum wage increases are part of an election campaign and are implemented with a time lag after the elections. Such a time lag is assumed by Arpaia et al. (2017) who split their sample according to whether minimum wage growth takes place in a year that follows legislative elections or not. Repeating the main analysis with a forward and a lagged election variable does not change the results, however. Neither elections that take place in the period prior nor elections that follow the period of minimum wage adjustments display a significant correlation with minimum wage growth. This is true for the full sample and both sub-samples (cf. Tables A.3 and A.4 in appendix). This means that on average minimum wage growth did not change systematically in size due to present, past or upcoming elections in the observation period, despite the fact that minimum wages are an often broached issue in election campaigns.

The growth model hypothesis suggested that a disproportionate dependence of GDP growth on either domestic consumption or net exports will affect minimum wage adjustments positively or negatively respectively. The descriptive analysis provides some support for this proposition. However, the multivariate analysis does this to a much lesser extent. In particular, there is no indication at all that a high net export dependence exerts a deflationary effect on minimum wage growth. With respect to domestic consumption, there is selective evidence for a positive effect of GDP growth from consumption in old EU member states, where a relatively stronger dependence on domestic consumption in period t contributed to higher minimum wage growth in period t + 1. This suggests that a higher reliance on domestic consumption for GDP growth has an independent effect and affects minimum wage growth in old EU member states.

The catch-up hypothesis proposed that stronger minimum wage growth in new EU member states is caused by stronger economic growth due to the accession to the EU and its single market. GDP growth was considered to be the crucial influence on minimum wage growth. The multivariate analysis showed no significant correlation between GDP growth and minimum wage growth. However, there is clear evidence that the accession to the EU and the adoption of the Euro changed the adjustment pattern of minimum wages in new EU member states. Minimum wage growth was highest during the accession process and before accession to the EU itself. Inflation was the relevant causal factor for minimum wage growth during this period, with minimum wage growth multiplying inflation by a factor of up to 2.8, irrespective of the actors responsible for adjustment. The ratio of minimum wages to median wages also grew rapidly during this period. The multivariate analysis in Table 7 showed that a high level of the Kaitz index had a negative effect on minimum wage growth in the following year during that period, with diminishing increases of the minimum wage for higher values of the Kaitz index. After accession to the EU, the disproportionate influence of inflation on minimum wage growth stopped. For new EU member states in the Eurozone, average wage growth became the relevant guidance for an altogether lower minimum wage growth. Although the catch-up hypothesis must be rejected as initially stated, its underlying assumption is confirmed by the fact that minimum wage growth in new EU member states was highest during the accession process when GDP growth was stronger and inflation higher than after the accession to the EU. The overall larger constant in the sub-sample of new EU member states also shows, however, that additional factors are at play behind the higher minimum wage growth of new EU members. A potential source may lie in the rebalancing of labour markets of accession countries as a result of the outflow of workers to old EU members before, during and after their accession to the EU.^[3] The elevated minimum wage growth represented by the large constant may be a response to decreasing labour surpluses or even labour shortages in new EU members due to an intensification of outward labour migration after the fall of the iron curtain and in the wake of EU accession (see for example Atoyan et al. 2016). At the same time, minimum wage increases may have been the most readily available social policy tool to support incomes in new EU member states. In most old EU member states, welfare systems provide a minimum level of subsistence and outweigh the minimum wage as a poverty-reducing instrument. Ronchi (2018) shows that the budgetary welfare effort for social protection policies was markedly lower in new EU member states than in old EU member states during the observation period of his study (2000–2014). Although minimum wages do not reach beyond job holders and are thus severely limited in addressing common causes of poverty such as unemployment, new EU member states may have used them as a substitute for more encompassing and targeted benefit systems that require much larger administrative and budgetary commitments than those needed for implementing a minimum wage.

10 Robustness of Results

Results from monthly minimum wage growth largely confirm the results emanating from hourly minimum wage growth. The signs and sizes of coefficients are in line with each other and display no bias potentially caused by the separate sources from which hourly and monthly data are obtained. Another source of bias may be the outliers Romania and Greece. For all samples, effect sizes of actors do not change considerably and significances do not change at all if these two outliers are excluded from the sample (cf. Table A.5 in appendix). Note, however, that the effect of inflation on minimum wage growth declines somewhat in the sub-sample of new EU member states if Romania is excluded. Another source of bias may lie in the fact that the sample is unbalanced because of Germany, which contributes only 6 years of observation compared to 21 for the other 20 countries. The exclusion of Germany from the sample leads to only very small changes in effect sizes and no changes in significances (cf. Table A.6 in appendix). Finally, the growth model variables may introduce bias due to multicollinearity as they are highly correlated with GDP growth. The exclusion of the two growth model variables from the main analysis, i.e. GDP growth from net exports and GDP growth from domestic consumption, does not alter the results, however (cf. Table A.7 in appendix). Effect sizes and significances remain unchanged. GDP growth in particular continues to have no correlation with minimum wage growth.

11 Conclusions

There is considerable variation in minimum wage growth in EU member states with national minimum wages. The largest growth difference is between member states that were part of the EU before the EU enlargement in 2004 and those that became members in 2004 or thereafter. This article examined the explanatory power of three hypotheses to explain the differences in minimum wage growth between individual countries in general and the difference between the group of old EU member states (before 2004) and new EU member states (2004 or after) in particular. The first hypothesis focused on the economic catch-up of new EU member states as the driving force for their stronger minimum wage growth. The second hypothesis suggested an effect on minimum wage growth of a country’s reliance on domestic consumption or net exports for GDP growth respectively. The third set of hypotheses looked at the likely impact of actors that are responsible for adjustment of the minimum wage, which vary between countries and points in time. The hypotheses were tested on a panel data set of 21 EU member states with national minimum wages over the period 2000 to 2020, comprising information on minimum wage levels, macroeconomic parameters and information on actors, elections and the orientation of economic policy.

Across all countries and throughout the observation period, minimum wage growth followed most closely consumer price inflation. Looking at new EU member states separately, minimum wage growth overcompensated inflation. This was especially the case during the period of accession to the EU, before EU membership itself. This is a central explanation for the difference in minimum wage growth between new and old EU member states. It is also evidence that the accession process was instrumental for the advancement of minimum wage levels in new EU member states. From the disproportionate growth of the minimum wage resulted a rise in the ratio of minimum to median wages in new EU member states. Especially during the accession period, but also during EU and Eurozone membership, wage inequality as measured by the Kaitz index dropped considerably in new EU member states. However, the analyses also showed that the increase in the Kaitz index had a negative feedback effect on minimum wage growth. A higher Kaitz index resulted in a deceleration of minimum wage growth and thus a levelling off effect in new EU member states.

There is essentially no evidence that the economic policy orientation of governments in terms of supply- or demand-side policies had an effect on average minimum wage growth. This may reflect the fact that policies to increase the minimum wage have become part of the policy toolbox of right-leaning and right-wing governments and that support of minimum wages has stopped being an identifier for left-leaning demand side policies. Elections, whether before, after or in the same period as minimum wage adjustments, did not affect the size of minimum wage growth. This was true for both old and new EU member states.

Adjustments of the minimum wage by social partners were shown to be higher than indexation, significantly so across all countries and in the sub-sample of old EU member states. Adjustments by governments were also higher than indexation, although this was not true across all estimations. Effect sizes, where significant, were higher in new EU member states and more muted in old EU member states. Adjustment by setting targets for minimum wage growth, which was in use only in few country/year cells during the observation period, was also connected to stronger growth than indexation. As a result, the analyses have shown that actors matter and that differences in institutional adjustment settings impact minimum wage growth, with adjustments based on social partner consensus most clearly surpassing the benchmark of adjusting the minimum wage by indexation.

With the convergence of relative, although not nominal, minimum wage levels across old and new EU member states and the lessening of the dynamic of the accession process, minimum wage growth in old and new EU member states will probably become more similar in the future. The EU directive on adequate minimum wages will very likely further this development. It stresses the importance of transparent adjustment processes and the need for ensuring adequate minimum wage levels. The directive envisages prominent roles for social partners and advisory bodies in the overall process of setting and adjusting minimum wages. It also prescribes clear criteria for adjusting minimum wages and refers to commonly used standards for measuring the adequacy of minimum wage levels. Especially the implementation of the latter, by raising minimum wages relative to average wages, may result in future adjustments across all EU member states that will not only keep real minimum wages stable but also affect the wage distribution to a larger extent than seen to date.