Will falling domestic labor compensation share really be improved when global trade slowdown?

Since the 1980s, IT advances and trade liberalization reduced trade costs, enabling vertical specialization and task trade, and fostering global value chains (GVCs). While GVC participation is associated with broader markets, higher efficiency, and rising national income, the distribution of these gains is uneven. Prior work suggests GVCs may contribute to declines in domestic labor compensation shares as gains accrue to a few multinational firms. This paper asks whether slowing global trade (declining GVC participation) improves domestic labor compensation shares and through which mechanisms. The study advances prior research by: (1) using structural decomposition analysis (SDA) to isolate the causal impact of GVC participation on labor share from technology and demand-structure effects; (2) introducing the availability of overseas labor to explain why Stolper-Samuelson predictions may fail under fragmentation; and (3) distinguishing forward (upstream) versus backward (downstream) GVC participation to identify participation modes that better stabilize domestic labor shares. The paper formulates H1: GVC participation negatively affects domestic labor compensation share; H2a: capital-biased technological progress mediates this effect; and H2b: availability of overseas labor mediates this effect.

Early empirical work (e.g., Timmer et al., 2014) and subsequent studies recognized that embedding in GVCs can depress domestic labor shares, with mechanisms including adoption of more capital-intensive technologies and weakened labor bargaining power. Evidence spans developing and developed economies: offshoring tends to raise high-skill returns and reduce low-skill wages (e.g., Hummels et al., 2014; Borghi and Crinò, 2013). The literature also highlights two mechanisms: (1) capital-biased technological change associated with GVC integration raises the capital share (Goldberg and Pavcnik, 2007; Feenstra and Hanson, 1996, 1997); and (2) increased availability of overseas labor via fragmentation raises the elasticity of labor demand and suppresses domestic wages (Milberg, 2004; World Bank, 2020). From these, the paper states: H1: GVC participation reduces domestic labor compensation share; H2a: capital-biased technological progress mediates this effect; H2b: overseas labor availability mediates this effect.

Measures and identification: (1) GVC participation is measured using the KWW decomposition (Koopman et al., 2014), separating domestic and foreign value added in exports. Backward participation (foreign value added in domestic exports divided by exports) and forward participation (domestic value added absorbed by other countries' exports divided by exports) are computed; total GVCParticipation equals the sum of forward and backward participation. (2) Availability of overseas labor is measured via the GVC income method (Timmer et al., 2013), decomposing domestic employment into domestic jobs retained (DL), foreign labor hired offshore (FL) for domestic final demand, and related flows, enabling calculation of average labor price and the stock of offshore labor embedded in production. (3) Structural decomposition analysis (SDA), following Solow (1958) and Acemoglu & Guerrieri (2008), decomposes the labor share into: intra-industry distribution (technology/markups), intermediate input substitution (linked to GVCParticipation), network multiplier effects (forward linkages), and final demand composition (consumption structure). This identifies relevant controls and isolates GVC participation's role. Econometric models: The benchmark panel regression (country-year, 51 economies, 2000–2018) regresses labor share on GVCParticipation (or GVCPar_forward / GVCPar_backward for structure) and controls, with fixed effects. Controls include: technological sophistication (VTech, adapted TSI using value-added exports and IP payments), consumption structure (service share in final demand), capital per capita (KL), and high-skill labor share (Highskill). Mechanisms are tested via moderated regressions: (a) capital-biased technical change (TB) as both outcome of GVCParticipation and moderator in labor-share regressions; (b) availability of overseas labor (FL) as both outcome of GVCParticipation and moderator. Robustness checks include excluding crisis years (2008–2009) and replacing the dependent variable with the Gini coefficient. Endogeneity is addressed using lagged GVCParticipation and two-stage least squares with gravity-model trade flows and 1980 trade dependence as instruments. Heterogeneity analyses test time (pre/post-2008), development level (Advanced dummy), and position (Manufacturing linkages dummy) effects. Data sources: OECD ICIO/ TiVA, WIOD, ILOSTAT, and World Bank; final balanced sample covers 51 economies after data screening.

• Benchmark effects: GVCParticipation significantly reduces domestic labor compensation share. For every 1% decrease in GVCParticipation, labor share increases by about 0.11%. In levels, GVCParticipation's coefficient is approximately −0.10 to −0.11 and statistically significant. Among SDA-identified channels, only GVCParticipation robustly predicts labor share declines (Table 4). - Structure of participation: Forward (upstream) participation does not significantly affect labor share (coefficients near zero, insignificant), whereas backward (downstream) participation significantly reduces labor share (e.g., coefficient ≈ −0.09, significant). - Robustness: Excluding 2008–2009 yields consistent negative effects for overall and downstream participation; forward participation remains insignificant. Using the Gini coefficient as the outcome, GVCParticipation increases inequality overall; forward participation reduces inequality, while backward participation increases it. - Endogeneity: Lagged GVCParticipation predicts lower labor share (≈ −0.18, significant). IV estimates using gravity-model trade flows and 1980 trade dependence yield a significant negative effect (≈ −0.16), with strong first-stage statistics (e.g., Kleibergen-Paap LM and Wald F significant) and valid overidentification (Hansen test). - Mechanisms: Capital-biased technical change (TB) rises with GVCParticipation (≈ +0.37% per 1% increase in GVCParticipation). The interaction term GVCParticipation × TB has a significant negative effect on labor share, confirming H2a. Availability of overseas labor (FL) rises with GVCParticipation (≈ +1.01% per 1% increase), and GVCParticipation × FL significantly reduces labor share, confirming H2b. Quantitatively, a 1% increase in TB and FL further reduces labor share by roughly 0.01 and 0.03 percentage points, respectively (as summarized in the conclusion). - Heterogeneity: The negative effect of GVCParticipation on labor share strengthens after the 2008 crisis (significant interaction with crisis). Effects are similar across development levels (interaction with Advanced not significant). Negative effects are more severe in manufacturing-linkage economies (significant interaction with Manufacturing). Overall, results support H1, H2a, and H2b and are consistent across multiple specifications and tests.

The findings demonstrate that higher GVC participation causally reduces domestic labor compensation shares, addressing the core question of whether slowing globalization can improve labor shares. A decline in GVC participation is associated with higher labor shares, implying that trade slowdowns/deglobalization may relieve downward pressure on labor compensation, though at the cost of efficiency and trade gains. Mechanisms clarify why: GVC integration promotes capital-biased technological progress and expands access to overseas labor, both of which shift factor income toward capital and depress domestic wages, especially in downstream, assembly-intensive roles. Forward participation (upstream, intermediate-good supplying, R&D- and resource-intensive) does not significantly erode labor shares and may balance efficiency with equity better than downstream modes. Policy relevance is high: governments can cushion labor-share declines without fully retreating from global integration by promoting upstream/forward linkages, fair domestic competition, services-led upgrading, skill formation, and redistributive measures. The evidence also reframes Stolper-Samuelson in a fragmented world: with cross-border task trade and offshore labor availability, labor-intensive exporting need not raise domestic wages, explaining observed paradoxes.

Using input-output based GVC measures, SDA decomposition, and panel regressions for 51 economies (2000–2018), the study shows: (1) GVC participation reduces domestic labor compensation shares; a 1% decline in GVCParticipation raises labor share by about 0.11%. (2) Capital-biased technical change and increased availability of overseas labor are key conditions mediating this effect; increases in each exacerbate labor-share declines. (3) Forward (upstream) participation better stabilizes labor shares than backward (downstream) participation, offering a path to balance efficiency and equity. Policy recommendations include promoting fair competition (e.g., anti-monopoly enforcement), developing services and export of services to absorb domestic labor and reduce capital bias, setting appropriate minimum wages and growth targets, and using taxes/transfers and human capital policies to protect workers affected by offshore competition. Internationally, fostering regional trade agreements and relational value chains can mitigate conflicts over upstream positions and support cooperative upgrading. Future research should examine industry-level heterogeneity and the nature (capital- vs labor-biased) of technology spillovers across development stages.

• Potential multicollinearity risk because GVC participation and embedded position variables are all constructed via value-added decompositions. - Theoretical underpinnings regarding the (in)validity of the Stolper-Samuelson theorem under fragmented production and limited labor mobility need further formal modeling to compare pre/post relaxation of labor-mobility assumptions. - Scope for deeper industry-level analyses of GVC participation’s effects on labor shares and investigation into whether technology spillovers are capital- or labor-biased across country contexts.