The strategic implementation of “Made in China 2025” has ushered in a new phase of accelerated development for the manufacturing sector, a cornerstone of the national economy. Within this landscape, the robotics industry stands as a critical barometer for gauging a nation’s manufacturing innovation capability and competitive strength. It converges with new energy, new materials, and novel technologies, emerging as a pivotal force driving technological and industrial revolutions. Recognizing this, the Chinese government has released the “Robot Industry Development Plan (2016-2020),” providing targeted policy support. However, significant gaps persist between the domestic robotics industry and its counterparts in other industrial powers. The Plan itself highlighted weaknesses, including inadequate core competitiveness, an imperfect industrial structure, challenges in ensuring the quality of high-end products, and a prevalence of numerous small, scattered, and weak enterprises overly reliant on demographic dividends. Total Factor Productivity (TFP) is intrinsically linked to these shortcomings. Therefore, enhancing the TFP of China’s robotics industry is not only vital for the sector’s own advancement but also plays a crucial role in realizing the “Made in China 2025” strategy and accelerating the growth of the real economy.
While domestic academic interest in China’s robotics sector is growing, research remains in its nascent stages, predominantly focused on theoretical guidance—covering target planning, roadmap design, policy impacts, and technological innovation. For instance, studies have analyzed the industry’s developmental drivers and objectives from technological, demand, and social perspectives, examined its evolution and trajectory through patent mapping, and proposed strategies involving global market integration and value chain upgrading. Policy analyses emphasize the need for more targeted measures and improved alignment between policy and industrial needs. Innovation research, employing patent data and efficiency evaluation methods, indicates a steady development trend but notes inefficiencies in translating technological achievements into productivity. A common thread is the macro-level approach to understanding the industry’s significance and challenges. However, compared to other high-tech sectors, there is a notable scarcity of empirical research quantitatively analyzing the production efficiency and, more specifically, the Total Factor Productivity of China’s robotics industry. Filling this gap is essential for a deeper, data-driven understanding of the industry’s health and drivers.
This investigation, therefore, seeks to address the following core questions: What is the current level of Total Factor Productivity within China’s robotics industry? What factors constrain or drive its development? To what extent do these elements influence the industry? Adopting a micro-level perspective, this analysis utilizes panel data from 33 listed robotics enterprises in China, spanning five consecutive periods from January 1, 2015, to December 31, 2019. By employing the DEA-Malmquist and BCC models, it delves into the measurement and decomposition of TFP and rigorously examines its key determinants, aiming to reveal the underlying dynamics behind the industry’s recent growth and provide theoretical support for enhancing its quality.
Methodological Framework and Variable Selection
Model Construction
To measure the TFP of robotics enterprises, the Malmquist productivity index, grounded in Data Envelopment Analysis (DEA), is employed. This index, introduced by researchers like Banker, Charnes, and Cooper, measures productivity change over time. Assuming a distance function $$d_0^t$$ for inputs $$x^t$$ and outputs $$y^t$$ in period $$t$$, the Malmquist TFP index between period $$t$$ and $$t+1$$ is defined as:
$$M_0(x^{t+1}, y^{t+1}, x^t, y^t) = \left[ \frac{d_0^{t+1}(x^{t+1}, y^{t+1})}{d_0^t(x^t, y^t)} \times \frac{d_0^t(x^{t+1}, y^{t+1})}{d_0^{t+1}(x^{t+1}, y^{t+1})} \times \frac{d_0^t(x^t, y^t)}{d_0^{t+1}(x^t, y^t)} \right]^{1/2}$$
This can be decomposed into two primary components:
$$M_0 = \text{EFFCH} \times \text{TECHCH}$$
Where EFFCH (Efficiency Change) measures the change in technical efficiency (the ability to maximize output from given inputs, or minimize inputs for a given output) from period $$t$$ to $$t+1$$. TECHCH (Technical Change) captures the shift in the production frontier, representing technological progress or regression. Furthermore, under variable returns to scale (VRS) assumption, the efficiency change (EFFCH) can be further decomposed:
$$\text{EFFCH} = \text{PECH} \times \text{SECH}$$
Here, PECH (Pure Technical Efficiency Change) reflects changes in management and organizational practices, while SECH (Scale Efficiency Change) indicates whether the firm is moving closer to or farther from its optimal operational scale. A value greater than 1 for any of these indices indicates improvement, while a value less than 1 denotes decline.
To investigate the factors influencing TFP and its components, econometric models are constructed based on panel data. The core models are specified as follows:
$$\ln(\text{TFP}_{mt}) = \alpha + \beta_1 \text{Pro}_{mt} + \beta_2 \text{Caps}_{mt} + \beta_3 \text{Capf}_{mt} + \beta_4 \text{Ope}_{mt} + \beta_5 \text{Mac}_{mt} + \epsilon_{mt}$$
$$\ln(\text{EFFCH}_{mt}) = \alpha + \beta_1 \text{Pro}_{mt} + \beta_2 \text{Caps}_{mt} + \beta_3 \text{Capf}_{mt} + \beta_4 \text{Ope}_{mt} + \beta_5 \text{Mac}_{mt} + \epsilon_{mt}$$
$$\ln(\text{TECHCH}_{mt}) = \alpha + \beta_1 \text{Pro}_{mt} + \beta_2 \text{Caps}_{mt} + \beta_3 \text{Capf}_{mt} + \beta_4 \text{Ope}_{mt} + \beta_5 \text{Mac}_{mt} + \epsilon_{mt}$$
Where $$m$$ denotes an individual robotics enterprise, and $$t$$ represents the time period. The dependent variables are the natural logarithms of the calculated TFP, Efficiency Change (EFFCH), and Technical Change (TECHCH) indices.
Variable Selection and Data
The selection of input and output variables for the DEA models, as well as the influencing factors, is based on established literature and data availability from the CNRDS database, which provides detailed and reliable financial statements for listed companies.
DEA Input-Output Variables: The inputs represent the capital and labor invested by the robotics enterprises. The outputs reflect their core competitiveness in terms of profitability and performance.
| Category | Variable | Explanation | Proxy Metric |
|---|---|---|---|
| Inputs | Capital | Total scale of assets employed | Total Assets |
| Working Capital | Funds for daily operations | Operating Working Capital | |
| Labor | Compensation for human resources | Employee Compensation | |
| Outputs | Profitability & Scale | Revenue generation capacity | Total Operating Revenue |
| Innovation & Net Gain | Ultimate net profit performance | Net Profit |
Influencing Factors (Independent Variables): These variables are hypothesized to affect the TFP of China’s robotics firms.
| Factor | Symbol | Explanation | Measurement |
|---|---|---|---|
| Profitability | Pro | The ability to generate profit from operations. A higher value indicates stronger profitability. | Net Profit Margin (Net Profit / Operating Revenue) |
| Capital Structure | Caps | The proportion of debt in the capital mix. A lower value suggests a more reliable, less leveraged structure. | Asset-Liability Ratio (Total Liabilities / Total Assets) |
| Cash Flow Operation | Capf | The quality and efficiency of cash flow generated from core operations. | Cash Operation Index (Operating Cash Flow / Net Profit + Depreciation & Amortization) |
| Operational Scale | Ope | The efficiency and scale of working capital utilization in generating revenue. | Working Capital Turnover Ratio (Operating Revenue / Average Working Capital) |
| Management Capability | Mac | Reflects the cost of administration relative to total operations. A higher ratio implies lower management efficiency. | Management Expense Ratio (Management Expenses / Total Operating Cost) |
The study focuses on 33 publicly listed companies in China identified as core players in the robotics industry, as referenced in the “Robot Industry Development Plan.” Annual data for the five-year period from 2015 to 2019 is extracted from the CNRDS database. The DEAP 2.1 software is used to compute the Malmquist indices and BCC efficiency scores, while statistical software is employed for the subsequent regression analysis of influencing factors.
Empirical Analysis of Total Factor Productivity
Temporal Evolution of TFP
Applying the DEA-Malmquist model to the panel data yields the annual changes in TFP and its decomposed components for the Chinese robotics industry from 2015 to 2019.
| Factor | 2015-2016 | 2016-2017 | 2017-2018 | 2018-2019 | Mean |
|---|---|---|---|---|---|
| Pure Technical Efficiency Change (PECH) | 0.972 | 1.119 | 1.000 | 1.037 | 1.031 |
| Scale Efficiency Change (SECH) | 0.981 | 0.987 | 1.005 | 0.942 | 0.979 |
| Efficiency Change (EFFCH) | 0.954 | 1.105 | 1.005 | 0.976 | 1.008 |
| Technical Change (TECHCH) | 0.994 | 0.928 | 1.041 | 1.068 | 1.006 |
| Total Factor Productivity Change (TFP) | 0.948 | 1.025 | 1.046 | 1.043 | 1.015 |
The results indicate that during the five-year period following the implementation of the national plan, the TFP of China’s robotics industry experienced a modest average annual growth of 1.5%. This growth was primarily driven by a 0.8% increase in technical efficiency (EFFCH) and a 0.6% increase in technological progress (TECHCH). The decomposition reveals that the improvement in efficiency was solely due to a 3.1% annual growth in pure technical efficiency (PECH), which was partially offset by a 2.1% annual decline in scale efficiency (SECH). This suggests that better management and organizational practices (pure technical efficiency) were the dominant force behind efficiency gains, while diseconomies of scale or suboptimal scaling acted as a drag.
Examining the year-on-year trends, TFP initially declined in 2015-2016 before showing consistent improvement in the subsequent three periods. The drivers of this growth were not synchronized. The years of highest TFP growth (2017-2019) were characterized by strong technological progress (TECHCH > 1.04), whereas the efficiency change (EFFCH) was more volatile. This asynchrony implies that the primary drivers of TFP growth in China’s robotics industry shift over time, and there is often a lag between improvements in technical efficiency and technological progress. The persistent negative growth in scale efficiency, despite overall TFP growth, signals that many enterprises may still be engaged in blind expansion or suffer from misallocation of production factors, indicating a developmental stage that remains partly reliant on factor-driven, extensive growth patterns rather than intensive, efficiency-driven growth.
Regional Disparities in Efficiency
To understand spatial patterns, the 33 sample firms are grouped into four major economic regions: the Yangtze River Delta (YRD), the Pearl River Delta (PRD), Northeast China, and Central & Western China. Using the BCC model under the assumption of Variable Returns to Scale (VRS), we calculate the average technical efficiency (TE), pure technical efficiency (PTE), and scale efficiency (SE) scores for each region from 2015 to 2019. A score of 1 indicates full efficiency.
Technical Efficiency (TE) by Region: TE measures the overall ability to produce maximum output from a given set of inputs under the VRS assumption.
| Region | 2015 | 2016 | 2017 | 2018 | 2019 | Mean |
|---|---|---|---|---|---|---|
| Northeast China | 0.810 | 0.658 | 0.758 | 0.747 | 0.728 | 0.740 |
| Yangtze River Delta | 0.739 | 0.734 | 0.780 | 0.834 | 0.859 | 0.789 |
| Central & Western China | 0.789 | 0.794 | 0.866 | 0.758 | 0.705 | 0.782 |
| Pearl River Delta | 0.726 | 0.689 | 0.762 | 0.782 | 0.705 | 0.733 |
| National Mean | 0.762 | 0.726 | 0.794 | 0.790 | 0.770 | 0.768 |
The national average technical efficiency score of 0.768 suggests that, on average, robotics firms in China could increase their output by 23.2% using the same level of inputs if they operated on the production frontier. Regionally, the Yangtze River Delta (0.789) and Central & Western China (0.782) performed relatively better, while Northeast China (0.740) and the Pearl River Delta (0.733) lagged. This indicates that the earlier industrial advantages of the Northeast and PRD may have diminished, potentially turning into comparative disadvantages. The YRD’s steady improvement and leading position by 2019 suggest a more rational allocation of inputs. The performance of Central & Western China underscores the positive impact of targeted policy support for high-tech industries in less developed regions, granting them a “latecomer advantage.”
Pure Technical Efficiency (PTE) by Region: PTE isolates the efficiency of management and production technology, excluding scale effects.
| Region | 2015 | 2016 | 2017 | 2018 | 2019 | Mean |
|---|---|---|---|---|---|---|
| Northeast China | 0.848 | 0.730 | 0.874 | 0.897 | 0.858 | 0.841 |
| Yangtze River Delta | 0.787 | 0.789 | 0.842 | 0.882 | 0.828 | 0.826 |
| Central & Western China | 0.855 | 0.865 | 0.895 | 0.790 | 0.806 | 0.842 |
| Pearl River Delta | 0.756 | 0.720 | 0.850 | 0.869 | 0.837 | 0.806 |
| National Mean | 0.809 | 0.784 | 0.862 | 0.860 | 0.888 | 0.841 |
Pure technical efficiency scores are generally higher than overall technical efficiency scores, with a national mean of 0.841. Interestingly, Central & Western China and Northeast China lead in PTE, surpassing the more economically advanced YRD and PRD. This suggests that the management practices, technological application, and organizational forms adopted by robotics firms in these regions are well-suited to their local economic contexts. The PRD’s consistently lower PTE, despite its economic prowess, indicates potential inefficiencies in managing input factors, possibly due to rapid expansion or external environmental challenges.
Scale Efficiency (SE) by Region: SE measures how close a firm is to its optimal production scale.
| Region | 2015 | 2016 | 2017 | 2018 | 2019 | Mean |
|---|---|---|---|---|---|---|
| Northeast China | 0.939 | 0.888 | 0.851 | 0.836 | 0.870 | 0.877 |
| Yangtze River Delta | 0.939 | 0.928 | 0.920 | 0.939 | 0.915 | 0.928 |
| Central & Western China | 0.925 | 0.914 | 0.962 | 0.956 | 0.922 | 0.936 |
| Pearl River Delta | 0.951 | 0.950 | 0.886 | 0.898 | 0.890 | 0.915 |
| National Mean | 0.938 | 0.921 | 0.911 | 0.917 | 0.899 | 0.917 |
Scale efficiency is the highest among the three efficiency measures, with a national average of 0.917. The YRD and Central & Western China show the best scale efficiency, while the Northeast exhibits the lowest. The lower scale efficiency in the Northeast may be attributed to legacy factors from its old industrial base, such as state-owned enterprises and policies encouraging industrial revitalization, which might have led to over-expansion or scale mismatches. The PRD’s scale efficiency has declined, potentially due to foreign capital influx driving rapid scale increases without corresponding productivity alignment. The overall declining trend in national scale efficiency signals that the industry is in an expansion phase, but one that is not consistently moving firms toward their most productive scale.
The regional analysis clearly shows that the development stage and constraints vary significantly across China. For the robotics industry in China to advance, improving technical efficiency is paramount, but this requires targeted efforts to enhance both pure technical efficiency and scale efficiency based on regional contexts.
Analysis of Factors Influencing TFP
Building on the efficiency measurements, we now examine how the hypothesized firm-level factors influence TFP and its components. The results of the panel data regression models are summarized below.
| Variable | Model (1): ln(TFP) | Model (2): ln(EFFCH) | Model (3): ln(TECHCH) | Robustness Check |
|---|---|---|---|---|
| Profitability (Pro) | 0.005* (1.930) | 0.007 (0.831) | 0.006* (1.601) | 0.003 (0.138) |
| Capital Structure (Caps) | -0.261 (-1.320) | 0.025 (0.093) | -0.057 (-0.171) | -0.235 (-0.869) |
| Cash Flow Operation (Capf) | 0.064** (2.076) | 0.067* (1.538) | -0.043 (-0.836) | 0.018 (0.363) |
| Operational Scale (Ope) | 0.055*** (3.948) | 0.014 (0.491) | 0.039** (2.183) | 0.032* (1.935) |
| Management Capability (Mac) | -0.002 (-0.254) | -0.006 (-0.388) | -0.007 (-0.856) | -0.004 (-0.913) |
| Constant | 0.049 (0.299) | -0.072 (-0.331) | 0.111 (0.429) | 0.232 (1.151) |
| R-squared | 0.686 | 0.706 | 0.758 | 0.704 |
| F-statistic | 3.633 | 6.726 | 8.723 | 7.389 |
Note: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. t-statistics are in parentheses.
Impacts on Total Factor Productivity (TFP): Analyzing Model (1), the operational scale (Ope) has a strongly positive and statistically significant impact (1% level) on TFP. This indicates that larger, more efficient working capital operations are a key driver for the robotics industry in China. Cash flow operation quality (Capf) also shows a significant positive effect (5% level), highlighting that efficient cash flow management directly enhances productivity. Profitability (Pro) has a weakly positive influence (10% level). Conversely, capital structure (Caps) and management capability (Mac) exhibit negative coefficients, suggesting that higher leverage and lower management efficiency (higher Mac ratio) hinder TFP growth, though these effects are not strongly significant in the base model. The robustness check column, using an alternative specification, largely confirms the direction of these relationships, particularly the strong positive role of operational scale.
Decomposed Effects on Efficiency Change (EFFCH) and Technical Change (TECHCH):
- Profitability (Pro) & Operational Scale (Ope): These factors positively influence both EFFCH and TECHCH, with Ope having a significant effect on TECHCH. This dual-channel positive impact explains their overall beneficial role for TFP. It suggests that profitable and efficiently scaled firms in China’s robotics sector are better at both catching up to the efficiency frontier and contributing to its outward shift through innovation.
- Capital Structure (Caps): It has a slight positive effect on EFFCH but a negative one on TECHCH. This implies that debt capital might be used to acquire existing technologies, helping firms improve their technical efficiency in the short term. However, it may crowd out investment in genuine R&D, thereby stifling technological progress. This trade-off necessitates careful optimization of capital structure.
- Cash Flow Operation (Capf): It significantly promotes EFFCH but weakly inhibits TECHCH. Efficient cash flow supports smoother operations and efficiency gains but may not be sufficiently directed towards long-term, risky technological development. To foster overall TFP growth, firms must allocate high-quality cash flows not just for operational efficiency but also for advancement in core technologies for China robots.
- Management Capability (Mac): The negative coefficients across all models, though not always highly significant, consistently indicate that poor management efficiency (high administrative costs) drags down both the firm’s ability to utilize existing technology efficiently (EFFCH) and its capacity for technological advancement (TECHCH). This underscores a critical internal bottleneck.
In summary, the growth of TFP in China’s robotics industry is currently most strongly propelled by expanding and optimizing operational scale and ensuring healthy cash flow. Profitability supports this growth. However, the industry must navigate the dual role of capital structure and strategically allocate resources to fuel technological progress, all while undertaking essential improvements in internal management practices.
Conclusions and Policy Implications
Key Findings
This empirical investigation into the Total Factor Productivity of China’s robotics industry yields several salient conclusions. First, the industry has experienced modest TFP growth since 2015, primarily driven by improvements in technical efficiency rather than groundbreaking technological progress. This efficiency gain is itself fueled by advances in pure technical efficiency (management and organization), which are being partially offset by declining scale efficiency. This pattern suggests the industry is still transitioning from a factor-driven, extensive growth model.
Second, significant regional disparities exist. The Yangtze River Delta and Central & Western China demonstrate relatively better performance in overall technical and scale efficiency, leveraging industrial agglomeration and policy-driven latecomer advantages, respectively. In contrast, the traditional industrial bases in Northeast China and the Pearl River Delta face challenges related to scale inefficiency and relatively weaker pure technical efficiency, indicating that their historical development models require adaptation.
Third, the drivers of TFP are multifaceted. Profitability and, most importantly, the efficiency and scale of operations are robust positive influencers. The quality of cash flow operation enhances efficiency. However, the capital structure presents a complex trade-off, potentially aiding efficiency but hindering technological progress. Furthermore, inadequate management capability emerges as a consistent drag on both efficiency and technological advancement.
Strategic Recommendations
To propel China’s robotics industry towards higher-quality, innovation-driven growth in line with national strategic goals, targeted actions are required at the enterprise, industrial, and governmental levels.
For Robotics Enterprises:
1. Undertake Management Model Reformation: Firms must critically assess and modernize their business and organizational models. Reducing administrative bloat and improving resource allocation processes are essential to convert inputs into outputs more effectively, thereby boosting pure technical efficiency.
2. Strategically Optimize Financial Structure: Enterprises should carefully balance debt and equity to avoid the innovation-inhibiting effects of over-leverage. High-quality cash flows should be strategically channeled into R&D and core technology development for next-generation China robots, not just operational maintenance.
3. Pursue Optimal Scaling: Growth should be deliberate and data-informed. Firms, especially in regions with low scale efficiency, need to find their optimal production scale to achieve economies of scale, moving away from blind expansion.
For Industrial Planning:
1. Implement Region-Specific Development Strategies: A one-size-fits-all approach is ineffective. The Northeast should focus on technological upgrading and management renewal to improve pure technical efficiency before further scale expansion. Central and Western China should capitalize on their efficiency by strategically scaling up operations. The Yangtze River Delta must lead the transition to intensive, quality-focused growth. The Pearl River Delta needs to integrate its external technological inflows with enhanced internal management and scale optimization.
For Government Policy:
1. Deepen and Differentiate Policy Support: Beyond broad subsidies, policies should be tailored to different market segments and firm sizes (e.g., differentiated subsidies, targeted cultivation of champions, specialized support for SMEs). Policies should evolve from being technology-push oriented to more market-pull and innovation-ecosystem focused.
2. Facilitate “Industry-University-Research-Government” Collaboration: The government should act as a convener and catalyst. Initiatives could include organizing learning exchanges for firms to advanced robotics hubs, funding joint R&D platforms between enterprises and research institutes, and creating forums for continuous dialogue to align policy with industrial needs. The ultimate goal is to build a self-reinforcing ecosystem that sustains the growth of TFP for China robots, ensuring the industry’s long-term global competitiveness and its pivotal role in modernizing the national manufacturing base.