The rapid proliferation of industrial robots, often termed the ‘China robot’ phenomenon, represents a pivotal shift in the nation’s manufacturing landscape. As a critical component of advanced manufacturing, the application of industrial robots is a key indicator of a country’s technological innovation and high-end manufacturing prowess. The Chinese government has placed significant emphasis on this development, with strategic documents like “Made in China 2025” and successive robotics industry plans underscoring its importance. Data from the International Federation of Robotics (IFR) vividly illustrates this surge: from 2009 to 2019, the operational stock of industrial robots in China skyrocketed from 37,300 units to 782,700, representing an average annual growth rate of 35.58%. Since 2016, China has consistently led the world in industrial robot installations, with installations reaching 243,300 units in 2021 alone, accounting for nearly half of the global total.
Concurrently, China’s Outward Foreign Direct Investment (OFDI) has experienced remarkable growth, transforming the country into a net capital exporter. In this context, a critical question arises: How does the application of industrial robots—the embodiment of automation and smart manufacturing—influence the internationalization strategies of Chinese firms, particularly their decisions and intensity regarding OFDI? This article investigates this nexus, analyzing how the ‘China robot’ wave shapes the “going global” trajectory of Chinese manufacturing enterprises.
1. Theoretical Framework and Hypotheses
The decision to engage in OFDI is traditionally explained by the eclectic paradigm (OLI framework), which posits that firms require Ownership, Location, and Internalization advantages. The application of industrial robots can fundamentally enhance a firm’s ownership-specific advantages, thereby influencing its OFDI capabilities and choices.
1.1 Core Effects and Location Choice
Industrial robot adoption confers two primary ownership advantages. First, it creates significant scale economy advantages. While the initial fixed costs for robots are high, they drastically reduce marginal production costs. Robots enable uninterrupted, stable operation at consistent quality levels, allowing firms to expand output rapidly while lowering the average cost per unit. Second, robots foster technological and innovation advantages. As embodiments of advanced production technology, their integration improves product quality, process efficiency, and management systems. This “capital-embodied” technological progress can spur innovation, creating a cycle of improvement that strengthens a firm’s monopolistic advantages in products and processes.
These enhanced advantages should, in theory, increase both the probability and the scale of a firm’s OFDI. However, the effect is likely not uniform across all destinations. High-income countries typically present higher technological and market entry barriers. The ownership advantages forged through ‘China robot’ adoption may be precisely the key to overcoming these barriers. Furthermore, investing in high-income countries offers potential for reverse technology spillovers, aligning with strategic asset-seeking motives. Conversely, a primary motive for investing in low- and middle-income countries has been the exploitation of low labor costs. Robot application, by saving on domestic labor costs, could potentially reduce the incentive for such efficiency-seeking OFDI. Therefore, we hypothesize:
H1: Industrial robot application promotes Chinese firms’ OFDI decision and intensity.
H2: This promoting effect is more pronounced for OFDI directed toward high-income countries than toward low- and middle-income countries.
1.2 Mechanism Analysis
The promotion effect is channeled through two primary mechanisms.
Mechanism 1: Scale Economy Effect. Robot application reduces costs and expands output. The cost reduction operates through:
1. Production Cost: Direct substitution of labor lowers wage bills and increases precision, reducing material waste.
2. Management Cost: Automation flattens organizational structures and improves information flow, enhancing managerial efficiency.
3. Output Scale: The ability to scale production rapidly, coupled with improved quality, helps capture larger market share.
The resultant scale economies lower average costs and boost productivity, directly strengthening a firm’s competitive edge for international expansion. We can model a firm’s cost function as:
$$ C(Q, R) = F_R + (c_l(R) + c_m(R)) \cdot Q $$
where $C$ is total cost, $Q$ is output, $R$ is the degree of robot application, $F_R$ is the fixed cost of robots, $c_l(R)$ is the marginal labor cost (decreasing in $R$), and $c_m(R)$ is the marginal material cost (decreasing in $R$). The average cost $AC = C/Q$ decreases with $Q$ and $R$, illustrating the scale economy effect.
Mechanism 2: Innovation Incentive Effect. Robot adoption stimulates innovation inputs and outputs.
1. Inputs: The enlarged output base helps amortize R&D costs per unit, reducing innovation risk. Capital-biased technological change increases the demand for high-skilled R&D personnel. The “creative destruction” brought by robots pressures firms to increase R&D expenditure.
2. Outputs: Robots facilitate efficient configuration and monitoring of R&D tasks, improving innovation success rates. Training to adapt human capital to automation further boosts innovative output.
Enhanced innovation capability solidifies a firm’s technology-based ownership advantage, facilitating OFDI, particularly in technology-intensive sectors or for strategic asset-seeking motives. This can be represented as:
$$ I = f(RD_{input}(R), A(R)) $$
$$ RD_{input}(R) = g(L_{RD}(R), K_{RD}(R)) $$
where innovation output $I$ is a function of R&D inputs $RD_{input}$ (itself a function of skilled labor $L_{RD}$ and capital $K_{RD}$, both increasing in robot intensity $R$) and absorptive capacity $A$, which is also enhanced by $R$.
H3: Industrial robot application promotes OFDI through the channels of scale economies and innovation incentives.
2. Empirical Methodology and Data
To test these hypotheses, we employ firm-level data from Chinese A-share listed manufacturing companies (2009-2019) matched with industry-level robot stock data from the IFR.
2.1 Model Specification
We examine both the extensive margin (decision) and intensive margin (strength) of OFDI.
For the OFDI Decision (Extensive Margin):
$$ Prob(OFDI\_D_{ijpt}=1) = \Phi(\alpha_0 + \alpha_1 \ln(CHF_{it}) + \gamma X_{it} + \lambda_j + \lambda_p + \lambda_t + \epsilon_{it}) $$
where $OFDI\_D_{ijpt}$ is a dummy variable equal to 1 if firm $i$ in industry $j$, province $p$, and year $t$ undertook OFDI. $\ln(CHF_{it})$ is the log of firm-level robot penetration. $X_{it}$ is a vector of firm-level controls. $\lambda_j$, $\lambda_p$, $\lambda_t$ are industry, province, and year fixed effects, respectively. $\Phi$ is the cumulative distribution function of the standard normal distribution. This is estimated using a Probit model (IV-Probit for robustness).
For the OFDI Intensity (Intensive Margin):
$$ OFDI\_I_{ijpt} = \max(0, \beta_0 + \beta_1 \ln(CHF_{it}) + \gamma X_{it} + \lambda_j + \lambda_p + \lambda_t + u_{it}) $$
where $OFDI\_I_{ijpt}$ is measured as the log of (1 + number of new overseas subsidiaries established in year t). Given the censored nature of this variable (many zeros), a Tobit model (IV-Tobit) is used.
2.2 Key Variables and Instrumental Variable
Core Explanatory Variable: Robot Penetration ($CHF_{it}$). Following established methodology, firm-level robot penetration is constructed by weighting industry-level robot stock per worker by the firm’s share of industry employment in a base year (2009):
$$ CHF_{it} = \left( \frac{PWP_{i,2009}}{ManuPWP_{2009}} \right) \times \left( \frac{MRCN_{jt}}{LCN_{j,2009}} \right) $$
where $PWP_{i,2009}$ is employment of firm $i$ in 2009, $ManuPWP_{2009}$ is the median employment in manufacturing in 2009, $MRCN_{jt}$ is the stock of robots in industry $j$ at time $t$ in China, and $LCN_{j,2009}$ is employment in industry $j$ in China in 2009.
Instrumental Variable. To address potential endogeneity, we employ a Bartik-style instrument based on U.S. robot adoption:
$$ USCHF_{it} = \left( \frac{PWP_{i,2009}}{ManuPWP_{2009}} \right) \times \left( \frac{MRUS_{jt}}{LUS_{j,1990}} \right) $$
where $MRUS_{jt}$ and $LUS_{j,1990}$ are robot stock and employment for industry $j$ in the U.S. The U.S. robot adoption trend is correlated with global technological trends (relevance condition) but unlikely to directly affect OFDI decisions of Chinese firms (exclusion condition).
Control Variables ($X_{it}$): Firm age, size (log revenue), capital intensity, leverage ratio, return on assets (ROA), fixed asset ratio, state-ownership dummy, and proportion of executives with overseas experience.
3. Empirical Findings
3.1 Baseline Results: The Aggregate ‘China Robot’ Effect
The baseline IV regression results confirm H1. Industrial robot application has a statistically significant positive effect on both the likelihood and the intensity of OFDI for Chinese manufacturing firms.
| Variable | OFDI Decision (IV-Probit) | OFDI Intensity (IV-Tobit) |
|---|---|---|
| ln(CHF) | 0.140*** (0.045) | 0.130*** (0.050) |
| Controls | Yes | Yes |
| Industry/Province/Year FE | Yes | Yes |
| Wald Test (p-value) | 34.06 [0.00] | 24.89 [0.00] |
| First-Stage F-statistic | 1741.76 | 1741.76 |
| Observations | 8,892 | 8,892 |
Note: Robust standard errors clustered at the firm level in parentheses. *** p<0.01.
3.2 Location Choice: High-Income vs. Middle-Low-Income Destinations
Splitting the sample based on the World Bank’s income classification of host countries provides strong support for H2. The positive impact of ‘China robot’ application is entirely driven by investments flowing to high-income countries.
| Variable | High-Income Destinations | Middle-Low-Income Destinations | ||
|---|---|---|---|---|
| OFDI Decision | OFDI Intensity | OFDI Decision | OFDI Intensity | |
| ln(CHF) | 0.146*** (0.051) | 0.178*** (0.061) | 0.142 (0.087) | 0.173 (0.106) |
| Observations | 7,650 | 7,654 | 5,930 | 6,697 |
The coefficients for OFDI to middle-low-income countries are statistically insignificant. This indicates that robot adoption helps Chinese firms overcome barriers to enter advanced economies but does not motivate (and may even deter) cost-oriented investments in developing regions.
3.3 Mechanism Test Results
To test H3, we employ interaction models. The results confirm that both scale economies and innovation incentives serve as significant transmission channels, particularly for OFDI to high-income countries.
| Channel | Proxy Variable | Interaction with ln(CHF): OFDI Decision | Interaction with ln(CHF): OFDI Intensity |
|---|---|---|---|
| Scale Economy | Production Cost (Cost/Revenue) | -0.236* (0.138) | -0.292* (0.161) |
| Management Cost (Mgmt Exp/Revenue) | -0.437*** (0.156) | -0.642*** (0.170) | |
| Output Scale (Log Output) | 0.006* (0.003) | 0.008** (0.004) | |
| Innovation Incentive | R&D Personnel Intensity | 0.858* (0.486) | 1.096** (0.552) |
| R&D Expenditure Intensity | 0.018*** (0.006) | 0.021*** (0.007) | |
| Innovation Output (Intangible Assets/TA) | 2.294** (1.028) | 2.759** (1.199) |
Note: The table reports the coefficient for the interaction term ln(CHF) × Channel Variable. All models include full controls and fixed effects. *, **, *** denote significance at 10%, 5%, and 1% levels.
The negative interaction for cost variables implies that the positive effect of robots on OFDI is stronger for firms that achieve greater cost reduction. The positive interactions for output scale and innovation variables indicate that the effect is stronger for firms that realize greater scale expansion or innovation enhancement from robot use.
3.4 Heterogeneous Effects
The impact of the ‘China robot’ revolution on OFDI is not uniform across all firms. The promoting effect on investment in high-income countries is concentrated in specific subsets of enterprises, as summarized below:
| Dimension of Heterogeneity | Subgroup with Stronger Effect | Potential Explanation |
|---|---|---|
| Ownership | Non-State-Owned Enterprises (SOEs) | SOEs face greater constraints in labor substitution due to social stability mandates; non-SOEs have more flexibility to adopt robots and realize cost savings. |
| Geographic Location | Coastal Region Firms | Higher openness, better access to technology and capital, and a industrial structure more suited to automation (more technology-intensive firms). |
| External Finance Dependence | Low External Finance Dependence Firms | Robot adoption requires significant upfront investment; firms with stronger internal funds can better bear the short-term cost and realize long-term benefits. |
| Factor Intensity | Technology-Intensive Firms | Production processes are more amenable to automation; higher absorptive capacity allows them to better leverage the technology spillovers from robots for innovation. |
3.5 Analysis by Investment Motive
Categorizing OFDI projects by their primary motive reveals nuanced effects of robot application.
| OFDI Motive (Host: High-Income Countries) | OFDI Decision | OFDI Intensity | Interpretation |
|---|---|---|---|
| Business & Service (e.g., sales, distribution) | 0.197*** (0.052) | 0.401*** (0.099) | Robots boost domestic production scale and quality, creating a larger exportable surplus and strengthening the need for overseas sales networks. |
| Research & Development | 0.241* (0.139) | 0.550** (0.273) | The innovation advantage gained from robots provides both the capability and the strategic need to seek complementary advanced technologies in global innovation hubs. |
| Local Production | 0.120 (0.102) | 0.199 (0.167) | Effect is insignificant for high-income hosts. The labor-cost saving from robots in China reduces the efficiency-seeking motive to produce in other high-wage countries. |
When expanding the sample to include all host countries, the effect on local production OFDI becomes positive and significant. This suggests that ‘China robot’ adoption might actually promote local production in middle-low-income countries, possibly to circumvent trade barriers or to combine automated production from home with final assembly abroad, alleviating concerns about widespread “reshoring” due to automation.
4. Conclusion and Implications
This analysis confirms the significant and multifaceted role of industrial robot application—the core of the ‘China robot’ narrative—in shaping the outward investment patterns of Chinese manufacturing firms. The evidence strongly supports that robot adoption promotes OFDI, primarily by enabling and incentivizing investment into high-income countries. This effect operates through the concrete channels of achieving scale economies and stimulating innovation. The benefits are most pronounced for non-state-owned, coastal, financially robust, and technology-intensive enterprises. In terms of strategic intent, robots particularly facilitate market-seeking (business/service) and strategic asset-seeking (R&D) internationalization.
The policy implications are clear. To harness the ‘China robot’ trend for high-quality opening-up:
1. Promote Deep Integration: Encourage the deep integration of robots across production, management, and logistics, coupled with training for high-skilled technicians, to fully realize scale and innovation benefits.
2. Deepen SOE Reform: Reform state-owned enterprises to enhance their flexibility and adaptability in adopting advanced technologies like robots, unlocking their potential as leading forces in high-quality OFDI.
3. Implement Differentiated Support Policies: Design tailored support policies for firms with different OFDI motives (e.g., market-seeking vs. technology-seeking) to align national strategic objectives with firm-level competitiveness built on automation.
The rise of the ‘China robot’ is more than a story of domestic productivity; it is a key driver reshaping the country’s position in global investment networks, steering its firms toward more innovation-oriented and upstream engagements in the world economy.