Belonging to Unions Across the World

Macro and Micro Dynamics  ·  Interactive Analysis  ·  Manuscript Revision, Feb 19 2026

Using multilevel models on 46,301 workers in 75 countries (World Values Survey 2010–2020), this study reveals that democracy and economic integration do not uniformly raise or lower union membership. Instead, they act as institutional switching mechanisms: democracy narrows class gaps while amplifying ideological divides; economic integration does the exact reverse.

75 Countries 46,301 Workers WVS Waves 6 & 7 Multilevel LPM IRT-GRM Indices
0
Countries Analyzed
0
Workers in Sample
0%
Average Union Rate
0%
Range (Haiti → Sweden)

The Central Argument

Why do some workers join labor unions while others do not? This study demonstrates that macro-level political and economic contexts do not operate as independent additive forces — they act as institutional gatekeepers that determine which individual characteristics matter, and for whom.

Core proposition: "Conflict may follow capital, but membership follows institutions." Democracy and union rights determine whether grievances, solidaristic orientations, and ideological commitments translate into formal union participation.

The study integrates world-systems theory (Arrighi, 1994; Silver, 2003) with institutionalist accounts of associational rights (Kraus, 2007; Wang, 2017; Kučera & Sarna, 2006) and applies a multilevel framework to data from both the Global North and Global South — the first study to do so jointly.

Analytical Framework — 2 × 2 Variable Structure

Four core variables are examined, with every macro × micro interaction tested:

Macro Level (country)Micro Level (individual)
Economy 🌐 Economic Integration
Trade/GDP + FDI/GDP stock accumulated 1980→survey, 1% annual depreciation (World Bank 2025)
🏭 Class Position
IRT-GRM composite: education (X025R) + subjective class (X045) + income decile (X047_WVS)
Politics 🗳️ Democracy Level
IRT stock: V-Dem liberal democracy + Freedom House (rights + liberties) + Kucera & Sari LRI (1980→survey)
⬆️ Ideology
IRT-GRM composite: E036 (ownership) + E037 (gov. responsibility) + E039 (competition) — higher = more left

Four Hypotheses — Click to Expand

H1  ·  Class Position × Democracy → Flattening Class Differences

As democracy increases, the union membership gap between workers in secure and precarious class positions shrinks. Democratic institutions and enforceable labor rights lower the organizational and legal barriers that advantage higher-class workers, enabling precarious workers to "catch up."

Empirical result: In highly authoritarian contexts, a ~6 pp class gap exists (Upper ≈21%, Lower ≈15%). In the most democratic contexts, all three class groups converge to ≈22–23%, effectively eliminating class-based stratification. The marginal effect of class declines from ~4 pp at the lowest democracy level to near zero at the highest. ✓ Confirmed — H1 supported

H2  ·  Ideology × Democracy → Rising Role of Left-Wing Militancy

As democracy deepens, left-leaning (militant) individuals become increasingly more likely to be union members than non-left individuals. Under authoritarianism, political freedoms are restricted and the ideological gap is small or absent; as democratic space expands, ideological commitments translate more readily into union membership.

Empirical result: Under authoritarianism, right-, center-, and left-leaning individuals all exhibit ≈15–16% membership. In the most democratic contexts, left-leaning individuals reach ≈23% while right-leaning remain ≈18% — a ~5 pp divergence. The marginal effect of left ideology grows from ≈0 to ≈3.5 pp across the democracy spectrum. ✓ Confirmed — H2 supported

H3  ·  Class Position × Economic Integration → Strengthening Class Divides

As economic integration increases, the gap in union membership between workers in secure/protected and precarious class positions widens. Integration reorganizes labor markets through trade exposure, FDI, subcontracting, and informality — costs that fall disproportionately on precarious workers.

Empirical result: In low-integration contexts, class differences are minimal (19.7%–22.6%, a 2.9 pp gap). At the highest integration levels, the gap nearly doubles: upper-class workers maintain ≈20.6% while lower-class workers drop to ≈16.5% — a 4.1 pp gap. The marginal effect of class rises from 1.7 pp to 2.5 pp across the integration spectrum. ✓ Confirmed — H3 supported

H4  ·  Ideology × Economic Integration → Weakening Ideological Gaps

With greater economic integration, the difference in union membership between left- and non-left individuals decreases. As integration shifts bargaining constraints outward — toward global value chains, mobile investment, and international markets — ideological commitments matter less than employment security and organizational access.

Empirical result: In low integration contexts, left-wing workers achieve 22.5% vs. 20.1% for right-wing workers (2.4 pp gap). At high integration, this compresses: left-wing 19.1% vs. right-wing 17.4% (1.7 pp gap) — a 29% reduction in ideological differentiation. The marginal effect of ideology declines from 1.6 pp to 1.2 pp (a 25% reduction). Note: the overall decline is driven by all groups falling, not right-wing rising. ✓ Confirmed — H4 supported

The Switching Mechanism — Summary

🗳️

Democracy acts as a political liberalizer

→ Flattens class barriers: Precarious workers gain access. Class gap narrows to zero.

→ Heightens ideological divides: Left-wing commitments "activate" into membership. Ideology gap grows from 0 to ~5 pp.

Source: Baccaro et al. 2019; Kučera & Sarna 2006; Wang 2017

🌐

Economic integration acts as a market constraint

→ Sharpens class inequalities: Precarity penalty intensifies. Class gap nearly doubles.

→ Mutes ideological differences: Left–right gaps compress as structural insecurity becomes broadly binding. Marginal effect of ideology shrinks by 25%.

Source: Scruggs & Lange 2002; Checchi & Visser 2005; Vachon et al. 2016

Worker Profile Simulator

Configure a hypothetical worker's individual characteristics and macro context. The simulator calculates the predicted probability of union membership by combining effects from both the democracy and integration interaction models, based on the paper's reported findings.

Note: Predictions are derived from the paper's reported marginal effects and predicted probabilities (Figures 2–6). They represent point estimates and are intended for illustration, not precise forecasting.

⚙ Worker Profile

Class Position
Political Ideology
Employment Type

🌍 Macro Context

Democracy Stock
Authoritarian (−2.0)Democratic (+1.5)
Democracy Stock: 0.0
Economic Integration Stock
Low (−1.2)High (+1.5)
Integration Stock: 0.0
Country Presets

📊 Predicted Outcome

0% 32% 65%
21.4%
Predicted probability of union membership

Contribution Breakdown
📍 Base Average 21.4%
🏭 Class Position Effect
⬆️ Ideology Effect
💼 Employment Type
🌍 Context Effect

Interpretation

What-If Comparisons

Holding your worker profile constant, see how predicted probability changes under different macro contexts.

Democracy as an Institutional Switching Mechanism

Democracy opens political channels through two distinct mechanisms: it reduces class-based barriers (enabling precarious workers to catch up) and amplifies ideological effects (enabling left-wing commitments to convert into membership). These two processes operate simultaneously but reshape different cleavages.

The democracy stock index combines V-Dem's liberal democracy scores, Freedom House's political rights and civil liberties measures, and Kucera & Sari's (2019) Labor Rights Index — accumulated from 1980 to the survey year with 1% annual depreciation. This "stock" approach captures how institutional democracy accumulates and shapes organizational outcomes over time (Gerring et al., 2012).

Figure 1 — Coefficient Plot: Democracy, Class & Ideology Models

Estimated coefficients from three multilevel LPM specifications. The dependent variable is binary union membership. Horizontal bars represent 95% confidence intervals. A coefficient that does not cross zero is statistically significant at p < 0.05.

Reading the plot: Points = coefficient estimates (marginal effects on probability of union membership). Error bars = 95% CIs. A negative Democracy×Class coefficient means democracy reduces the effect of class position, i.e., flattens class barriers.

Interactive Predicted Probabilities

How does class position matter differently across the democracy spectrum?

← Authoritarian (−2.0)Democratic (+1.5) →
Democracy Stock: −2.0
H1 confirmed: Authoritarian contexts produce a stark class hierarchy — Upper ≈21% vs. Lower ≈15% (~6 pp gap). Democratic contexts compress this completely: all three groups converge to ≈22–23%, eliminating class-based stratification. The marginal effect of class declines from ~4 pp at the lowest democracy level to near zero at the highest (Figure 2, Panel a).

Substantive Interpretation

Why does democracy flatten class barriers?

Democratic institutions and union rights expand the political channel for collective organization. As democratic guarantees and union rights grow (measured by fewer LRI violations per ITUC 2025), the political channel becomes more salient. Workers in precarious positions — who face higher barriers in authoritarian settings — gain access to organizational channels previously unavailable to them.

This is consistent with the institutionalist argument (Avdagic & Baccaro, 2014) that legal protection and extension mechanisms, rather than spontaneous militancy, open membership to previously excluded workers.

Why does democracy amplify ideology?

Under political repression, left-leaning orientations have little practical impact on behavior — the organizational channel for translating political commitments into union membership is closed. As democratic space expands, these orientations become "actionable": left-wing workers can join unions without the same risk of repression or co-optation.

The result is that democracy simultaneously broadens access (by class) and differentiates it (by ideology) — precisely the "switching" pattern the study predicts.

Economic Integration as an Institutional Switching Mechanism

Economic integration does not simply "move" capital; it reorganizes labor markets through trade exposure, FDI, subcontracting, and competitive upgrading. These pressures segment employment into more protected jobs versus a larger pool of precarious work — sharpening class divides. Simultaneously, integration depoliticizes workplace relations by shifting bargaining constraints toward global value chains and mobile investment, narrowing ideological differences.

The economic integration stock combines trade share and FDI inflows as percentages of GDP, accumulated from 1980 with 1% annual depreciation. This captures the long-run embeddedness in global capitalism rather than year-to-year fluctuations.

Figure 4 — Coefficient Plot: Integration, Class & Ideology Models

Note: Economic integration itself is not statistically significant in any specification (coefficient consistently negative but CIs cross zero). Its effects operate primarily through interactions with class position and ideology.

Interactive Predicted Probabilities

How does the "precarity penalty" intensify with economic integration?

← Low Integration (−1.2)High Integration (+1.5) →
Integration Stock: −1.2
H3 confirmed: Low integration → minimal class gap (19.7%–22.6%, spread of 2.9 pp). High integration → pronounced gap (16.5%–20.6%, spread of 4.1 pp) — nearly double. Marginal effect of class position grows from 1.7 pp to 2.5 pp across the integration spectrum. Lower-class workers experience the steepest absolute decline (3.2 pp) vs. upper-class (2.0 pp).

Precarity Penalty — Side-by-Side Comparison

Low Integration Context (−1.2)

Lower Class19.7%
19.7%
Middle Class21.0%
21.0%
Upper Class22.6%
22.6%

Class gap: 2.9 percentage points

High Integration Context (+1.5)

Lower Class16.5% (↓ 3.2 pp)
16.5%
Middle Class18.6% (↓ 2.4 pp)
18.6%
Upper Class20.6% (↓ 2.0 pp)
20.6%

Class gap: 4.1 percentage points (+41% wider)

Union Membership Rate Across Countries (Figure A1)

Union membership rates range from 62% in Sweden to approximately 1% in Haiti, with an overall average of 21.4% across the 75-country sample. Filter by region or search for specific countries.

Region:

Country-Level Patterns

High Membership — Different Mechanisms

Nordic democracies (Sweden ≈62%, Finland ≈50%) show high rates reflecting robust democratic institutions, strong Ghent-system union-managed unemployment insurance, and long-standing labor rights. These are genuine union memberships enabling collective bargaining.

Caution: Tajikistan (≈60%) shows similarly high rates but these reflect state-controlled organizations that primarily serve regime interests rather than workers' collective bargaining rights. The paper acknowledges it does not assess "union authenticity" — a limitation for authoritarian regime cases.

Low Membership in the Global South

Many developing countries in the sample show rates below 10% — reflecting the combined effect of weaker democratic institutions and high economic integration, precisely the conditions the study shows create the "precarity penalty" and suppress ideological mobilization.

Key theoretical insight: Anticipated surges in Southern unionization following capital relocation (Arrighi, 1994) did not materialize. The study explains why: capital moves conflict, but institutions — specifically democratic protections — determine whether conflict consolidates into formal membership.

Empirical Strategy

The study employs hierarchical linear probability models (LPM) with random intercepts to account for individuals nested within countries. Although the dependent variable (union membership) is binary, the LPM is used to ensure interpretability of coefficients as marginal effects on the probability of union membership — particularly valuable in multilevel contexts with between-group and within-group variation (Pugh et al., 2022; Yaskewich, 2021; Hu et al., 2022).

The dataset comprises 46,301 individuals from 75 countries, restricted to part-time and full-time workers (the unemployed and self-employed are generally not legally eligible to join trade unions). Three models are estimated for each macro context: (1) base model, (2) interaction with class position, (3) interaction with ideology.

Outcome Variable: Union Membership

Union Membership is a binary indicator (0/1) of whether a worker is currently a member of a labor union, derived from WVS question A101. In the sample, the union membership rate ranges from 62% in Sweden to 1.06% in Haiti, with an average of 21.4%.

The study does not distinguish between active and inactive membership, nor does it assess union authenticity — acknowledged as a limitation for countries where unions are state-controlled.

Ideology Index (IRT-GRM)

Constructed using a Graded Response Model (GRM) from three WVS survey items (all on 1–10 scales, higher = more left-leaning):

  • E036 — Private vs. state ownership of business (1=private, 10=government)
  • E037 — Government responsibility (1=individual, 10=government)
  • E039 — Competition (1=good, 10=harmful)

Individual-level latent trait scores extracted via empirical Bayes means. Standard errors clustered at the country level. Discrimination parameters: E036 α=0.49***, E037 α=0.53***, E039 α=1.80*** (Appendix C).

Class Position Index (IRT-GRM)

Constructed from three WVS items using IRT-GRM. Higher values = higher socioeconomic position:

  • X025R — Educational attainment: Lower (below upper secondary) / Medium (upper & post-secondary) / Higher (tertiary)
  • X045 — Subjective social class: 1 (lower class) to 5 (upper class)
  • X047_WVS — Income rank: decile position within country

Discrimination parameters: Subjective Class α=2.76***, Income α=1.57***, Education α=0.93*** (Appendix C).

Democracy Stock Index (IRT)

To treat political regimes as historically informed rather than contemporary snapshot variables (Gerring et al., 2012), a stock of democracy is constructed: each year's democracy score is accumulated from 1980 to the WVS survey year with a 1% annual depreciation rate. Starting in 1980 captures the post-"third wave" democratization period (Huntington, 1992).

Three measures are combined via IRT to capture complementary institutional dimensions:

  • V-Dem Liberal Democracy Index — constitutional protections, judicial independence, constraints on executive power
  • Freedom House Political Rights & Civil Liberties — procedural democratic processes, freedom of expression, assembly
  • Kucera & Sari (2019) LRI — union-specific freedoms, collective bargaining protections, labor organizing violations

Discrimination parameters: Freedom House Civil Liberties Stock α=50.86***, Political Rights Stock α=55.86***, LRI Stock α=84.89***, V-Dem Stock α=7.37*** (Appendix C).

Economic Integration Stock Index (IRT)

Economic integration draws on two indicators from the World Development Indicators (World Bank, 2025):

  • Trade share in GDP — (exports + imports) / GDP
  • FDI inflows as % of GDP

As with the democracy stock, these are accumulated from 1980 to the survey year with a 1% annual depreciation rate. The IRT approach extracts country-level latent trait scores capturing the multidimensional nature of economic embeddedness in global capitalism.

Note: Venezuela's trade data cover 1980–2014; Ethiopia's cover 2011–2020; Uzbekistan's cover 1997–2022 (footnote 4 in the original paper).

Control Variables

All models include the following controls:

  • Immigrant status (binary) — accounts for lower unionization among immigrant workers
  • Employment type (full-time vs. part-time) — full-time workers have significantly higher unionization
  • Sex (female=1) — modest negative effect on unionization probability
  • Age and Age² — captures the inverted U-shaped relationship (middle-aged workers most unionized)

Summary Statistics

VariableNMeanSDMinMax
Union Member (binary)43,9220.210.410.001.00
Democracy Stock Index46,3010.000.97−2.111.44
Global Integration Stock Index46,3010.000.85−1.923.28
Democracy Index (Survey Year)46,3010.000.97−1.541.41
Global Integration Index (Survey Year)46,3010.000.86−2.453.87
Class Position Index46,3010.000.84−2.012.25
Ideology Index46,3010.000.71−1.361.90
Immigrant44,1010.070.250.001.00
Full-Time Employed46,3010.790.410.001.00
Sex (Female=2)46,2691.440.501.002.00
Age46,07139.6912.531699
Age Squared46,0711,732.61,073.12569,801

Core Theoretical Contribution

Unionization as a macro-conditioned micro-outcome

The study reframes union membership from a simple function of individual characteristics or macro forces to a macro-conditioned micro-outcome. Individuals decide to join unions, but the institutional architecture of global capitalism determines which individual predispositions can be converted into durable organization. The same worker characteristics yield radically different membership probabilities depending on the institutional context in which they are embedded.

Refining the "conflict follows capital" argument

The paper sharpens Arrighi's (1994) world-systems claim that as capital relocates from North to South, the locus of labor conflict follows. While the study confirms that capital mobility repositions sites of potential conflict, it shows that the translation of conflict into durable union membership is contingent on political openings. This explains why anticipated surges in Southern unionization failed to materialize despite decades of production shifts — formal membership follows institutions, not just capital flows.

Beyond Global North / Global South: a unified framework

Existing research has been geographically fragmented — focusing either on OECD countries (Ebbinghaus et al., 2011; Keyes et al., 2025) or on Southern countries ignoring the North (Martin & Brady, 2007). This study addresses that gap by adopting a global multilevel framework examining 75 countries spanning all world regions, demonstrating that the same conditional mechanisms operate globally even as their expression varies by context.

Policy Implications

Institutional reform over market spontaneity

Reversing union decline depends less on the spontaneous dislocations generated by global capitalism than on deliberate political reform. The study shows that strong labor rights and democratic protections are the precondition through which grievances convert into durable membership. Relying on market forces to generate mobilization from below is insufficient.

Dual strategy for precarious workers

Because economic integration heightens the precarity penalty, unions must pursue a dual strategy: (1) expand organizational repertoires toward sectoral, supply-chain, and community-based models with low-cost membership designs; and (2) simultaneously articulate a broader agenda of democratic renewal and institutional reform to reduce the coercive costs of joining for the most vulnerable.

Limitations

  • Does not distinguish between active and inactive union membership (WVS question is binary)
  • Cannot assess union authenticity — state-controlled unions in countries like Tajikistan may inflate membership figures without genuine collective bargaining
  • Reverse causality: unionization may itself strengthen democratic institutions over time; the study mitigates this using stock variables with temporal lag but cannot fully resolve it
  • Constrained to the 75 countries covered by WVS Waves 6 & 7 (2010–2020)
  • Does not evaluate the role of past union experience as a determinant of current membership

Directions for Future Research

Substantive extensions

  • Unionization of vulnerable communities, especially immigrants
  • De-unionization patterns — what drives exit from union membership over time
  • Additional macroeconomic variables (e.g., financialization, austerity regimes)
  • Regionally disaggregated analyses to probe whether the gatekeeping mechanism varies across regions

Methodological extensions

  • Panel designs that track individual union membership changes over time
  • Distinguishing authentic from state-controlled union membership
  • Quasi-experimental designs to better identify causal effects of democratization on unionization
  • Extending the framework to other forms of collective action beyond formal union membership