Does government investment push up manufacturing labor costs? Evidence from China

The paper investigates why labor costs in China’s manufacturing sector have risen rapidly, eroding the country’s historical low-cost advantage. China’s earlier development strategy emphasized high investment with low wages and benefits to accelerate industrialization. However, from 2000 to 2012, China’s hourly manufacturing labor costs increased fivefold, outpacing other major economies and narrowing its competitive edge relative to emerging markets. Rising labor costs affect firm competitiveness and can trigger relocation of labor-intensive industries. The study posits that, in China’s context of strong local government involvement and reliance on infrastructure-led growth (e.g., the 2008 CNY4 trillion stimulus), government investment materially shapes labor market dynamics. It asks whether and how government investment increases manufacturing labor costs, examining both unit labor costs (which incorporate productivity) and nominal labor costs (labor compensation). The authors argue that government investment, particularly in infrastructure, can increase labor demand, alter transaction costs and production scale, and crowd out private sector activity, thereby affecting labor costs. The study contributes by focusing on government investment as a determinant of labor costs, a factor underexplored in prior literature that emphasized demographics, social insurance, and minimum wage policies.

The review covers two strands: (1) economic and social effects of changing labor costs, and (2) determinants of labor costs. On effects, studies show higher unit labor costs reduce firms’ export entry and increase exit (Decramer et al., 2016; Altomonte et al., 2013), with policy evidence from France indicating a 10% ULC increase reduces exports by ~2% (Malgouyres & Mayer, 2018). Comparisons of Greece and Portugal suggest austerity-reduced ULCs supported export growth (Doulos et al., 2020). Rising labor costs can induce innovation, with stronger effects in labor-intensive firms and under certain market structures (Cui & Lu, 2018; Jianqiang Li et al., 2020; Chih-Hai Yang, 2023), but may erode competitiveness (Ghizdeanu et al., 2007) and, in China, hamper employment growth with environmental regulation’s innovation compensation insufficient to offset compliance costs (Huang et al., 2021). On determinants, wages reflect cost of living, education/training (human capital), and leisure preferences (Thurow, 1974), while factor price dynamics change with development stages. Empirical work examines immigration’s mixed effects on labor costs (Duszczyk et al., 2013; Zvaigzne et al., 2015; Orrenius et al., 2020), tax and social security contributions (Lehmann et al., 2013; Alvaredo et al., 2017; Adam et al., 2019; Holzner et al., 2022), and minimum wage policies (Harasztosi & Lindner, 2016). In China, minimum wage rules, urbanization, and aging are important (Fan et al., 2013; Gallagher et al., 2013; Chen et al., 2014; Zhang & Han, 2013). The gap identified is the lack of focus on government economic behavior—particularly government investment—as a driver of labor costs in China’s government-led development model. The paper addresses this by linking government investment, infrastructure, and labor cost dynamics.

Design: A provincial panel data analysis is used to estimate the impact of government investment on manufacturing labor costs with fixed effects to control for time-invariant regional heterogeneity and year shocks. Model: LC_it = a + β ln(GI_it) + X_it + μ_i + δ_t + ε_it, where i indexes provinces and t years; LC_it is labor cost; ln(GI_it) is (log) government investment; X_it includes controls; μ_i are region fixed effects; δ_t are year fixed effects. Key variables: - Dependent variables: (1) Unit labor cost (ULC) = hourly labor compensation / hourly labor output. Hourly labor compensation is derived by dividing estimated annual average compensation by annual working hours; annual working hours are 49 weeks times average weekly hours. Annual compensation is estimated using urban manufacturing wages adjusted by 0.981 (to reflect all manufacturing workers) and multiplied by 1.27 to include non-wage costs (social security, etc.). Hourly output is manufacturing value added divided by annual working hours of manufacturing employment. (2) Nominal labor cost: average labor compensation/wages of employed manufacturing workers. - Core explanatory variable: Government investment (GI). Measured primarily as fixed asset investment in the state-owned economy; log-transformed in baseline. Robustness uses “budgeted funds for fixed asset investment” as an alternative proxy. - Controls: lagged GDP per capita (economic development), industrialization degree (secondary industry value added/GDP), labor availability (share of population aged 15–64), number of employed persons in secondary industry, and regional (east/central/west/northeast) and year fixed effects. Data: Main sources are the China Statistical Yearbook and provincial statistical yearbooks. The study uses a panel of 31 provinces/autonomous regions/municipalities. Figures describing trends span 2000–2018, while model estimation relies on data primarily from 2000–2016, yielding 527 observations in regressions. Variables include GDP, industrial value added, manufacturing employment and wages, fixed asset investment in the state-owned economy, working-age population, urban population, and employment in secondary industry. Estimation: Fixed effects (FE) panel regressions are the main approach; Feasible Generalized Least Squares (FGLS) is applied in robustness checks. Robustness includes alternative GI proxies, regional subsamples (east/central/west), and subsamples by government investment scale (25th, 50th, 75th quantiles). Mechanism analysis regresses infrastructure investment on government investment (overall and by region) to test the infrastructure channel.

- Main effect on unit labor cost (ULC): In fixed-effect regressions (Table 2, column 4), a 1% increase in government investment is associated with a 0.013 unit increase in manufacturing ULC. The paper also states in the abstract and conclusion that a 1% increase in government investment raises ULC by 0.0013 units. - Effect on nominal labor cost: Using wage/compensation as the dependent variable (Table 3), government investment increases nominal labor costs; the reported coefficient indicates an increase of 1.443 units per unit of government investment. The paper also elsewhere reports that a 1% increase in government investment raises nominal labor cost by 12.82%. - Controls: Higher GDP per capita is positively associated with ULC, with diminishing returns (negative GDP squared). Higher industrialization correlates with lower ULC (suggesting productivity growth outpaces remuneration), while higher employment in the secondary sector raises ULC; a larger working-age population share reduces ULC. - Robustness: Results remain when replacing the GI measure with “budgeted funds for fixed asset investment” and using FGLS (Table 4). Effects are consistent for both ULC and nominal labor cost. - Regional heterogeneity (Table 5): GI’s impact on ULC is strongest in the eastern region (coefficient 0.065), smaller in the central (0.015) and western (0.014) regions. - Investment scale heterogeneity (Table 6): Across GI scale quantiles, positive effects persist with coefficients 0.064 (25%), 0.024 (50%), 0.023 (75%), indicating diminishing marginal effects as GI scale increases. - Mechanism via infrastructure (Table 7): Government investment positively drives infrastructure investment overall (coefficient ~1.314) and significantly in eastern (2.772) and western (2.144) regions; the central region’s effect is not significant. The study argues this infrastructure expansion both directly increases labor demand and indirectly reduces transaction costs, enabling output expansion that further raises labor costs.

The findings support the hypothesis that government investment—particularly infrastructure-led spending by local governments—raises manufacturing labor costs in China. Three channels are emphasized: (1) direct labor demand expansion from large-scale infrastructure projects, tightening labor markets and pushing up wages; (2) reduced transaction costs from improved infrastructure, which facilitates firm expansion and increases labor demand; (3) potential crowding out of private sector resources (labor, financing) by government projects, influencing labor market dynamics and firm investment in R&D. Regional heterogeneity aligns with differing development levels and infrastructure endowments: the eastern region, with advanced economies and stronger market responses, shows the largest ULC sensitivity; the western region, with scarcer infrastructure, also exhibits strong infrastructure response to GI. Diminishing marginal effects by investment scale suggest that initial increases in GI have stronger impacts on labor costs than later increments. Overall, the results indicate that government-led investment strategies, while supporting growth, can accelerate labor cost increases and erode cost competitiveness in labor-intensive manufacturing, reinforcing the need to balance public investment with market mechanisms.

The paper contributes by identifying government investment as a significant driver of rising labor costs in China’s manufacturing sector and by explicating the infrastructure channel. Using a provincial panel with fixed effects and multiple robustness checks, it finds that increases in government investment are associated with higher unit and nominal labor costs, with stronger effects in eastern regions and diminishing marginal effects at higher investment scales. Policy recommendations include: rationally controlling the scale of government investment to keep labor cost growth moderate; channeling government investment toward rural infrastructure and public goods to reduce urban crowding-out and balance labor supply-demand; crowding in private capital for major infrastructure to reduce fiscal burdens; and focusing public investment on foundational and strategic high-tech sectors to promote innovation and industrial upgrading. Future research should examine how government debt and investment affect firm innovation and labor productivity, and more broadly how macro policies shape micro firm behavior, industrial development, and market structure.

- Data constraints: Labor cost data are partly derived from secondary calculations (e.g., converting annual to hourly measures and applying adjustment factors), and access to some underlying data is limited. - Measurement choices: Government investment is proxied primarily by fixed asset investment in the state-owned economy, with an alternative proxy used in robustness; different proxies may capture distinct facets of public investment. - Temporal coverage inconsistencies: While descriptive figures cover up to 2018, the main regression sample reflects a subset of years with 527 observations across 31 provinces, which may affect generalizability across the full 2000–2018 period. - Mechanism heterogeneity: The infrastructure channel is not uniformly significant across regions (e.g., central region), indicating regional institutional and developmental differences that may limit generalization of the mechanism.