Effects of manufacturing input servitization on labor income share and income distribution

The study investigates whether and how manufacturing input servitization (MIS)—the increasing use of service inputs within manufacturing—affects labor income share (LIS) and broader income distribution. As economies develop, both servitization and LIS tend to rise, with manufacturing’s LIS surpassing the all-industry average at higher per capita GDP levels. The motivation stems from rising income inequality and the need to rebalance functional income distribution by strengthening labor’s share. The paper posits that integrating services reshapes firms’ input, output, and distribution structures, potentially elevating the role of human capital and labor productivity and thereby raising LIS. The primary research question is how MIS impacts LIS and income distribution, including the mechanisms via labor productivity and human capital, and whether effects vary across industries and development stages. The authors combine theoretical mechanisms with empirical evidence using multinational, multi-industry panel data to inform policy for inclusive growth.

Prior work documents a global decline in labor income share (e.g., Karabarbounis and Neiman, 2014) and links LIS to returns on capital, technological change, and bargaining power (Bentolila and Saint-Paul, 2003; Acemoglu, 2003; Spector, 2004). Servitization is conceptualized as input servitization (greater service inputs) and output servitization (service offerings alongside products) (Sousa and da Silveira, 2019). Studies highlight servitization’s roles in deepening specialization (Francois, 1990), boosting performance (Song et al., 2022), raising productivity and global value chain position (Zhang, 2022), and increasing technological complexity (Zhang et al., 2023). The literature suggests servitization elevates human capital’s importance (Hayakawa et al., 2009), alters factor input structure, and affects LIS (Gang and Guang, 2010). Gaps remain: limited transnational macro-level empirical work on servitization–LIS links, insufficient analysis of heterogeneity drivers, and underexplored mechanisms connecting MIS to LIS and to income inequality. This study addresses these gaps by combining theory and cross-country, multi-industry evidence.

Theoretical framework: Manufacturing output is decomposed into intermediate inputs (physical and service) and value added. MIS is defined as the share of service inputs in total intermediate inputs, reflecting manufacturers’ reliance on services. LIS is the share of value added accruing to labor. The hypotheses are: H1, MIS positively impacts LIS; H2, impacts vary across industries and economies; H3, higher LIS helps reduce income inequality; H4 and H5, MIS affects LIS via labor productivity and human capital.

Empirical model: The benchmark specification is an OLS regression of log LIS on log MIS with controls and fixed effects at the economy, industry, and year levels; additional models include industry–time fixed effects and clustered standard errors at the economy level. To address endogeneity (potential reverse causality and omitted variables), the study employs 2SLS using three instruments: a one-period lag of log MIS, economy-level MIS in the current year, and the current-year growth rate of service value added at the economy level. Weak-instrument diagnostics (Kleibergen–Paap and Cragg–Donald statistics) are reported.

Variables: Dependent variable: LIS (labor income/value added). Key independent variable: MIS (service input share in intermediate inputs), constructed from input–output tables using direct and complete consumption coefficients. Controls at industry level: capital stock, employment, value-added rate, return on capital. Economy-level controls: GDP per capita, foreign direct investment (inward+outward), R&D expenditure share of GDP, trade barriers. Mediators: labor productivity (output per employee) and human capital (labor income per employee).

Data: World Input-Output Database (WIOD 2016) input–output tables and socio-economic accounts, OECD (2021), and Asian Development Bank (2022). Coverage: 43 economies, 18 manufacturing sub-sectors, 2000–2014 (ISIC Rev. 4). All variables are in logarithms; data adjusted to exclude exchange rate and price effects per WIOD indices.

Robustness: Multiple specifications with varying fixed effects; bilateral shrinkage and censoring of extremes; split-sample analysis pre- and post-2008 financial crisis; random sampling with replacement. Group analyses by technology intensity (low, medium, high) and by development status (developing vs developed). Extended analysis links MIS to the poverty gap (defined relative to half the median household income) with additional controls: R&D spending and gross fixed capital formation (OECD).

• Benchmark regressions show a significant positive association between MIS and LIS across specifications. Representative coefficients on log MIS: 0.195 (SE 0.017), 0.082 (SE 0.010), 0.197 (SE 0.053), 0.112 (SE 0.037), all statistically significant, supporting H1.
• Robustness tests (1% bilateral shrinkage/censoring, split 2000–2008 vs 2008–2014, random sampling) consistently find positive and significant MIS–LIS relationships. The MIS coefficient is slightly larger post-2008, suggesting strengthened effects after the financial crisis.
• Endogeneity checks using 2SLS with instruments (lagged MIS, economy-level MIS, service value-added growth) preserve sign and significance. Kleibergen–Paap LM statistics strongly reject underidentification; weak-instrument tests indicate no weak-IV problem, validating identification.
• Mediation (H4, H5): MIS raises labor productivity (coef ≈ 0.090, p<0.01) and human capital (coef ≈ 0.197, p<0.01). Productivity positively affects LIS (≈ 0.140, p<0.01) and human capital positively affects LIS (≈ 0.331, p<0.01). Estimated indirect effects are significant: via productivity ≈ 0.013 and via human capital ≈ 0.065; direct effects remain positive but smaller (≈ 0.070 and ≈ 0.017), confirming partial mediation.
• Heterogeneity (H2): By technology intensity, MIS impact on LIS is largest in medium-technology industries (≈ 0.108***), followed by high-tech (≈ 0.064**) and low-tech (≈ 0.032**). By development status, effects are stronger in developing economies (≈ 0.142***) than in developed ones (≈ 0.061***).
• Extended analysis (H3): Higher MIS is associated with a lower poverty gap. In preferred specification with controls, the coefficient on log MIS is about −0.159 (p<0.10), while R&D (≈ −0.094, p<0.05) and gross fixed capital formation (≈ −0.168, p<0.01) also reduce the poverty gap, implying that increases in manufacturing LIS contribute to reduced income inequality.

The evidence directly addresses the research question by showing that integrating services into manufacturing inputs increases the labor income share. This occurs both directly and through enhanced labor productivity and upgraded human capital, substantiating the proposed theoretical mechanisms. The heterogeneity results suggest that the complementarities between services and production tasks, typical in medium-technology industries, translate more effectively into wage gains and a higher LIS than in highly automated high-tech sectors or simpler low-tech sectors. The stronger effects in developing economies indicate greater scope for labor-intensive service complementarities and human capital upgrading to shift functional income distribution toward labor. The extended analysis links higher LIS in manufacturing with a narrower poverty gap, suggesting that servitization-driven improvements in LIS can mitigate income inequality. Collectively, these results underscore servitization as a viable policy lever for inclusive growth, highlighting the importance of complementary investments in skills and innovation to realize distributional benefits.

This study demonstrates that manufacturing input servitization is positively associated with the labor income share across countries and industries. The relationship is robust to multiple specifications and identification strategies and is partially mediated by labor productivity and human capital. Effects are strongest in medium-technology industries and in developing economies. An extended analysis suggests that higher LIS in manufacturing contributes to reducing income inequality. Key contributions include: (1) macro cross-country evidence on MIS–LIS links; (2) identification of productivity and human capital as mediators; and (3) documentation of heterogeneous impacts across industry technology levels and development stages. Future research should extend beyond manufacturing to other sectors, employ firm-level microdata, explore additional mediators/moderators (e.g., labor institutions, automation, bargaining power), and examine broader macro outcomes and varied country contexts.

• Sectoral scope: Focused on manufacturing; findings may not generalize to services or other sectors without further study.
• Data constraints: 2000–2014 WIOD/OECD coverage and industry aggregation may mask within-industry heterogeneity and recent structural changes.
• Measurement: MIS constructed from input–output tables may not capture all forms of service integration or quality changes; LIS measures rely on macro factor income allocations.
• Identification: Despite 2SLS and strong IV diagnostics, residual endogeneity or omitted variables at finer levels (e.g., firm practices, labor institutions) may persist.
• Mechanisms: While productivity and human capital are validated mediators, other channels (e.g., bargaining power, organizational change, digitalization intensity) warrant investigation.