TL;DR
This paper uses a novel three-stage Data Envelopment Analysis (DEA) model to evaluate the innovation efficiency of Chinese Information Technology (IT) companies, accounting for environmental and random factors. The findings reveal that while IT innovation efficiency has improved, it remains suboptimal, primarily due to low technical efficiency, suggesting a need for better management and scale optimization.
What Problem Does This Paper Address?
The rapid growth of China's economy, particularly the IT sector, has brought significant economic and social benefits. However, this growth also raises concerns about resource consumption and environmental impact, especially in the context of achieving "double carbon" goals. While IT is inherently greener than many industries, understanding and improving its innovation efficiency is crucial for sustainable development. Existing studies often overlook the influence of external environmental factors and random noise on innovation efficiency. This paper aims to address this gap by providing a more accurate assessment of IT companies' innovation efficiency by disentangling these external influences.
Methods
The core methodology is a three-stage Data Envelopment Analysis (DEA) model.
- Stage 1: DEA Model (BCC Model): This stage uses the Banker, Charnes, and Cooper (BCC) model, which accounts for variable returns to scale, to calculate initial technical efficiency (TE), pure technical efficiency (PTE), and scale efficiency (SE) for each company. The relationship TE = PTE × SE is fundamental here.
- Stage 2: Stochastic Frontier Analysis (SFA) Regression Model: This stage addresses the limitations of Stage 1 by using SFA to model the slack variables (input redundancies) from Stage 1. These slacks are regressed against environmental variables. This allows for the separation of management inefficiency from environmental factors and random errors. The model then adjusts the input values to a common external environment and luck level.
- Stage 3: Adjusted DEA Model (BCC Model): The adjusted input values from Stage 2 are fed back into the BCC model to re-evaluate TE, PTE, and SE. This final efficiency score is considered a more accurate reflection of the enterprise's true innovation efficiency, free from external influences.
This three-stage approach is an advancement over traditional DEA by systematically accounting for and removing the impact of environmental and random factors, leading to a more robust efficiency measure.
Datasets
The study utilizes data from 80 software and information technology service companies in China for the years 2017 and 2018. The data was collected from the CSMAR database. Environmental variables were sourced from the National Bureau of Statistics, Wind financial database, and provincial/city statistical yearbooks. Negative indicators were standardized to a [0.1, 1] range using a specific formula.
Key Insights and Ideas
- Environmental Factors Matter: The study strongly confirms that external environmental variables significantly impact IT companies' innovation efficiency. Ignoring these factors can lead to misleading efficiency assessments.
- Technical Efficiency is the Bottleneck: While both technical and scale efficiencies contribute to overall innovation efficiency, pure technical efficiency is identified as the primary constraint for most Chinese IT firms. This suggests issues with management practices and the optimal utilization of resources.
- Scale is Mostly Increasing, but Not Optimal: The adjusted analysis shows that most companies operate under increasing returns to scale. This implies that their current innovation investment scale is not yet optimal, and there's room for growth by increasing investment to match their potential.
- Regional Disparities Exist: Significant regional differences in innovation efficiency are observed, with Eastern China showing a concentration of both quantity and quality of innovation, leading to a polarization effect compared to Central and Western regions.
Key Results
- Overall Efficiency Improvement: After accounting for environmental and random factors, the average innovation efficiency of the sample companies increased from 0.728 in 2017 to 0.755 in 2018.
- Technical Efficiency Gains: Pure technical efficiency (PTE) saw a notable improvement, rising from 0.783 to 0.802, while scale efficiency (SE) also increased from 0.932 to 0.964.
- Low Overall Efficiency: Despite improvements, the overall innovation efficiency is still considered not high, with a significant portion of companies falling into "Bad" or "Worse" categories before adjustment. Even after adjustment, many companies still have room for improvement.
- Company Age and Quality Impact: Company age and personnel quality were found to have varying correlations with input slacks, suggesting that while experience and skilled personnel are beneficial, their optimal utilization requires careful management.
- Regional Dominance: Eastern China leads in innovation quantity and quality, while Central and Western China show potential but lag in overall development and industrial cluster effects.
Limitations and Future Work
- Data Limitations: The study acknowledges limitations due to the lack of comprehensive statistical data for many periods, preventing a longer-term analysis of efficiency changes. The subdivision of main business into specific industry fields was also not possible, limiting the granularity of analysis.
- Indicator Verification: The authors suggest that the indicators used, particularly environmental variables, could benefit from further verification.
- Future Research: Future work could benefit from more complete and accurate statistical data to conduct longer-term studies and a more detailed analysis of innovation efficiency across different sub-fields within the IT industry. The authors also suggest that with improved statistical data, a more comprehensive and accurate overview of enterprise innovation can be achieved, leading to more targeted recommendations.
Summary updated