In the context of rapid advancements in automation technologies such as artificial intelligence, the application of industrial robots has become a pivotal force in modern manufacturing. China, as a global manufacturing hub, has witnessed a significant surge in the adoption of industrial robots, yet its robot density remains below that of developed nations. This raises a critical question: what drives the acceleration of automation in Chinese industrial firms? One plausible factor is the rising labor costs, particularly influenced by minimum wage policies. This study investigates whether increases in minimum wage, as an exogenous shock to labor costs, propel Chinese industrial firms toward automation, specifically through the import of industrial robots. We leverage firm-level microdata and employ econometric models to explore this causal relationship, offering insights into the dynamics between labor market regulations and technological adoption.
The proliferation of industrial robots globally has transformed production processes, with China emerging as the largest market for robot sales since 2013. However, China’s robot density—measured as the number of industrial robots per 10,000 manufacturing workers—was only 97 in 2017, lagging behind countries like South Korea (710) and Germany (322). This gap underscores the potential for growth in automation. Historically, China’s industrial expansion relied on abundant low-cost labor, but demographic shifts and escalating wages have prompted firms to seek labor-saving technologies. Minimum wage adjustments, mandated by provincial governments, serve as a key driver of wage inflation, compelling firms to reconsider their production strategies. This research delves into how such policy-induced cost pressures influence the decision to adopt robotics, thereby contributing to the broader discourse on industrial upgrading in China.
Automation, epitomized by China robot integration, is not merely a technological shift but an economic imperative. The rise of China robot imports reflects a strategic response to competitive pressures. From 2005 to 2018, installations of industrial robots in China soared from around 5,000 to over 133,000 units, capturing one-third of the global market share. This trend aligns with theoretical expectations from Hicksian factor substitution: as labor becomes costlier, firms substitute capital for labor. Minimum wage hikes, by elevating the wage floor, directly increase production costs, especially for low-skilled workers who are more susceptible to replacement by machines. Thus, we hypothesize that rising minimum wages accelerate automation in Chinese industrial firms, primarily through cost-effect channels. This study tests this hypothesis using robust empirical methods, addressing endogeneity concerns and exploring heterogeneous effects across firm types and regions.
Literature Review and Theoretical Framework
The literature on minimum wage effects spans labor economics and industrial organization. Early studies focus on employment impacts, with findings suggesting that minimum wage increases can reduce low-skilled job opportunities. For instance, research in the U.S. and China indicates that higher minimum wages raise average wages but may dampen employment growth. In the context of automation, Acemoglu and Restrepo (2020) demonstrate that robot adoption displaces workers, particularly in routine tasks. However, fewer studies examine the reverse causality: how labor costs drive automation. This gap is notable for China, where policy-driven wage changes are frequent. Theoretical underpinnings draw from the task-based model, where minimum wage hikes raise the cost of labor-intensive tasks, incentivizing firms to invest in automation technologies like China robot systems.
In China, minimum wage policies have evolved since the 2004 regulation, leading to varied regional adjustments. Studies show that these adjustments affect firm performance, survival, and productivity. For example, higher minimum wages can force low-productivity firms to exit, improving resource allocation. When it comes to automation, Cheng et al. (2019) link robot adoption in China to labor cost pressures, but they do not isolate the role of minimum wage. Our study extends this by using minimum wage as an exogenous instrument to identify causal effects. We argue that minimum wage increases primarily affect low-skilled wages, making automation more attractive. This aligns with Lordan and Neumark (2018), who find that minimum wages spur automation in jobs with high substitutability. Thus, our theoretical framework posits that minimum wage rises accelerate automation through cost-push mechanisms, with heterogeneity based on firm characteristics.
Data and Methodology
We utilize multiple data sources to construct a comprehensive dataset for analysis. First, China robot import data are derived from customs records, which detail firm-level imports of industrial robots. This measure is reliable because, prior to 2013, most robots used in China were imported, and over 80% of imports are for direct production use. Second, minimum wage data are manually collected from municipal statistical yearbooks, reflecting the legally mandated wage floors across cities. Third, firm-level data come from the Annual Survey of Industrial Firms in China, covering financial and operational variables from 2000 to 2013. After merging and cleaning, our sample includes over 830,000 firms with 2.64 million observations, of which 0.32% imported industrial robots. Table 1 summarizes the descriptive statistics of key variables.
| Variable | Description | Mean | Std. Dev. |
|---|---|---|---|
| mw | City minimum wage (yuan) | 695.61 | 312.92 |
| ln_gdp | Log city GDP per capita | 10.32 | 0.96 |
| ln_cw | Log city average wage | 10.07 | 0.56 |
| cpi | Consumer price index (0-1) | 1.03 | 1.95 |
| ln_fdi | Log foreign direct investment | 10.71 | 1.85 |
| old | Aging rate (proportion over 65) | 0.09 | 0.02 |
| hhi | Market competition index (0-1) | 0.11 | 0.16 |
| ln_P | Log relative price of robots | 0.004 | 0.02 |
| ln_lp | Log labor productivity | 12.44 | 1.26 |
| ln_sub | Log government subsidies | 0.58 | 1.82 |
| ln_emp | Log firm employment | 4.94 | 1.11 |
| ln_dw | Log wage growth | 0.19 | 0.55 |
| ln_kl | Log capital-labor ratio | 10.63 | 1.42 |
| age | Firm age (years) | 9.41 | 9.38 |
| lrl | Profit rate (0-1) | 0.04 | 0.11 |
| deb | Financing constraints (0-1) | 0.01 | 0.03 |
| exp | Export dummy (0-1) | 0.35 | 0.48 |
To address the issue of zero robot imports (i.e., firms that do not adopt robots), we employ a Heckman two-stage selection model. This approach corrects for sample selection bias by first modeling the decision to import robots and then the intensity of imports. The first-stage probit model estimates the probability of robot adoption, while the second-stage linear model estimates the log of robot import quantity or value. The equations are specified as follows:
First stage (selection equation):
$$ \text{Pr}(r_{ijct} = 1) = \Phi(\alpha_0 + \alpha_1 \ln mw_{ct} + X_{ijct}\beta + Z_{ct}\gamma + \epsilon_{ijct}) $$
Second stage (outcome equation):
$$ \ln rob_{ijct} = \phi_0 + \phi_1 \ln mw_{ct} + X_{ijct}\varphi + Z_{ct}\psi + \theta \lambda_i + \mu_{ijct} $$
Here, \( r_{ijct} \) is a dummy variable equal to 1 if firm \( i \) in industry \( j \), city \( c \), and year \( t \) imports robots; \( \ln rob_{ijct} \) is the log of robot import quantity or value; \( \ln mw_{ct} \) is the log of city minimum wage; \( X_{ijct} \) and \( Z_{ct} \) are vectors of firm-level and city-level control variables; \( \lambda_i \) is the inverse Mills ratio from the first stage; and \( \Phi \) denotes the cumulative distribution function of the standard normal distribution. Control variables include firm size, capital intensity, profitability, financing constraints, productivity, age, export status, subsidies, and regional factors like GDP per capita and aging rate. Industry and year fixed effects are included to capture unobserved heterogeneity.
Empirical Results
Our baseline results, presented in Table 2, confirm that rising minimum wages significantly increase the likelihood and scale of China robot imports. The first-stage probit estimates show a positive coefficient for minimum wage, indicating that a 10% increase in minimum wage raises the probability of robot adoption by approximately 0.16 percentage points. The second-stage estimates reveal that minimum wage hikes also boost the volume of robot imports: a 10% increase leads to a 0.05% rise in import quantity and a 0.01% rise in import value. The inverse Mills ratio is statistically significant, validating the Heckman model’s correction for selection bias. These findings support Hypothesis 1, that minimum wage acceleration drives automation.
| Variable | Stage 1: Probit | Stage 2: Quantity (FE) | Stage 2: Value (FE) |
|---|---|---|---|
| ln_mw | 0.1616** (0.0690) | 0.0210* (0.0104) | 0.0046* (0.0024) |
| ln_P | -1.6157** (0.6748) | 0.0021 (0.0212) | 0.0043 (0.0071) |
| ln_dw | 0.0510*** (0.0127) | 0.0024** (0.0010) | 0.0005* (0.0003) |
| ln_emp | 0.3991*** (0.0073) | 0.0143*** (0.0047) | 0.0024*** (0.0008) |
| ln_kl | 0.3172*** (0.0084) | 0.0071*** (0.0023) | 0.0011** (0.0004) |
| lrl | 0.3565*** (0.0826) | 0.0276** (0.0103) | 0.0058** (0.0024) |
| deb | -4.6283*** (0.7252) | -0.0427* (0.0217) | -0.0024 (0.0037) |
| ln_lp | -0.0422*** (0.0076) | 0.0011** (0.0005) | 0.0002** (0.0001) |
| age | -0.2321*** (0.0095) | -0.0071*** (0.0025) | -0.0011** (0.0004) |
| exp | 0.2634*** (0.0185) | 0.0106*** (0.0038) | 0.0017** (0.0007) |
| ln_sub | -0.0158*** (0.0033) | -0.0011*** (0.0004) | -0.0002** (0.0001) |
| Inverse Mills Ratio | 4.1317*** (0.1456) | 0.6107*** (0.0504) | – |
| Controls | Yes | Yes | Yes |
| Fixed Effects | Yes | Yes | Yes |
| Observations | 2,331,351 | 2,375,022 | 2,375,022 |
| R² | 0.3460 | 0.9487 | 0.7891 |
Robustness checks reinforce these results. We control for firm compliance with minimum wage laws by including an obedience index and by restricting the sample to foreign-owned and state-owned enterprises, which typically adhere more strictly to regulations. The results remain consistent, as shown in Table 3. Additionally, we split the sample around the 2004 Minimum Wage Regulation, finding that the automation-accelerating effect is stronger post-2004, aligning with policy enforcement (Table 4). This temporal analysis suggests that sustained minimum wage hikes have long-term impacts on automation trends.
| Variable | Probit | Quantity (FE) | Value (FE) |
|---|---|---|---|
| ln_mw | 0.1718** (0.0692) | 0.0207* (0.0107) | 0.0045* (0.0025) |
| obedience | 0.0397* (0.0239) | -0.0007 (0.0025) | -0.0002 (0.0005) |
| Controls | Yes | Yes | Yes |
| Observations | 2,331,351 | 2,375,022 | 2,375,022 |
| Period | Probit | Quantity (FE) | Value (FE) |
|---|---|---|---|
| Pre-2004 | -0.2478 (0.1549) | -0.0174* (0.0094) | -0.0026 (0.0017) |
| Post-2004 | 0.2779*** (0.0987) | 0.0306* (0.0153) | 0.0071* (0.0038) |
To address endogeneity, we employ instrumental variable (IV) approaches. We use two instruments: the lagged minimum wage and the average minimum wage of other cities in the same province. Both satisfy relevance and exogeneity conditions. The 2SLS results, in Table 5, show that after accounting for endogeneity, the effect of minimum wage on China robot adoption becomes more pronounced, with coefficients larger than in baseline estimates. This indicates that OLS estimates may understate the true impact, reinforcing our hypothesis that minimum wage rises are a causal driver of automation.
| Instrument | Probit | Quantity (FE) | Value (FE) |
|---|---|---|---|
| Lagged mw | 0.2842** (0.1197) | 0.0931*** (0.0060) | 0.0182*** (0.0023) |
| Other cities’ mw | 0.9495*** (0.1613) | 0.1780*** (0.0095) | 0.0335*** (0.0035) |
| Both instruments | 0.3010** (0.1232) | 0.0945*** (0.0058) | 0.0175*** (0.0022) |
| F-test (first stage) | – | 45.01*** | 9.738*** |
Mechanism tests reveal that minimum wage increases primarily raise wages for low-skilled workers, who are more replaceable by China robot systems. Using urban household survey data, we estimate that a 1% increase in minimum wage boosts low-skilled wages by 0.13%, with no significant effect on high-skilled wages. This cost push motivates firms to automate routine tasks. The share of low-skilled workers in high-automation industries exceeds 80%, as depicted in Figure 1, underscoring the substitution risk. Thus, Hypothesis 2 is supported: minimum wage hikes accelerate automation via low-skilled labor cost inflation.
Heterogeneity analysis, summarized in Table 6, shows that the effect varies across firms. Industries with higher substitution risk (e.g., automotive, electronics) exhibit stronger responses to minimum wage increases. Labor-intensive firms are more likely to adopt robots than capital-intensive ones, confirming Hypothesis 3. Larger firms and those with higher labor cost shares show greater automation propensity. Regionally, firms in eastern and central China, where minimum wages are higher, are more responsive than those in western regions. These findings highlight the nuanced impact of labor cost pressures on China robot adoption.
| Group | Probit Coefficient | Interpretation |
|---|---|---|
| High-substitution-risk industries | 0.4521*** (0.1023) | Strong automation response |
| Low-substitution-risk industries | 0.0987 (0.0876) | Weak or insignificant effect |
| Labor-intensive firms | 0.3876*** (0.0954) | Higher automation probability |
| Capital-intensive firms | 0.1124* (0.0642) | Moderate effect |
| Large firms (employment > 500) | 0.5218*** (0.1205) | Significant acceleration |
| Small firms (employment ≤ 500) | 0.0893 (0.0789) | Limited impact |
| Eastern/Central regions | 0.3345*** (0.0887) | Pronounced automation trend |
| Western regions | 0.0672 (0.0921) | Less responsive |
Conclusion and Policy Implications
This study demonstrates that rising minimum wages accelerate the automation of Chinese industrial firms, as evidenced by increased import of China robot technologies. The cost effect, driven by higher low-skilled wages, compels firms to substitute labor with capital, particularly in sectors prone to automation. Our findings, robust to endogeneity and heterogeneity checks, underscore the role of labor market policies in shaping technological adoption. From a policy perspective, several implications emerge. First, as China robot reliance grows, policymakers should foster domestic innovation to reduce dependence on foreign robotics. Second, minimum wage adjustments should be tailored to regional economic conditions to avoid stifling firm competitiveness. Third, given the displacement of low-skilled workers, investment in vocational training is crucial to mitigate unemployment and inequality. Ultimately, understanding the interplay between minimum wage and automation can guide China’s industrial upgrading in the era of intelligent manufacturing.
In summary, the integration of China robot systems into industrial production is not merely a technological shift but a strategic adaptation to labor cost pressures. Minimum wage policies, while aimed at protecting workers, inadvertently spur automation, leading to complex outcomes for employment and productivity. Future research could explore long-term effects on skill demand and income distribution. For now, this study provides empirical evidence that in the quest for efficiency, rising minimum wages indeed accelerate the march toward automation in China’s industrial landscape.