20211201-北京大学国家发展研究院-CCER讨论稿_Misappropriation_of_R_D_Subsidies_Estimating_Treatment_Effects_with_One-sided_Noncompliance_69页_2mb

E2021007: Misappropriation of R&D Subsidies: Estimating Treatment Effects with One-sided Noncompliance

This study investigates the misappropriation of R&D subsidies in China, examining its impact on policy effectiveness and exploring methods to measure the compliance rate. Using data from 2001 to 2011, the research addresses whether funds allocated to R&D are actually used for innovation.

The introduction highlights the critical role of public R&D funding, noting two major concerns: crowding out private investment and potential moral hazard behavior leading to fund misuse. Despite this, the literature often ignores moral hazard. China's rapid economic growth and increased R&D funding have raised concerns about fund management, with media reports suggesting that a significant portion of public research funds is misappropriated. This paper aims to quantify this concern and analyze its consequences, using a combination of theoretical modeling and empirical analysis with firm-level data.

Key Findings

1. Prevalence of Misappropriation: About 42% of firms receiving R&D subsidies misused the funds for non-research purposes, accounting for 53% of the total subsidy amounts. This confirms anecdotal evidence of substantial misuse.
2. Trends Over Time: Misappropriation declined substantially from 81% in 2001 to 18% in 2011, largely due to stricter sanctions and monitoring following the 2006 Mid-to-Long-term Science & Technology Development Plan.
3. Firm Characteristics: Firms exhibit a binary choice: either almost fully misappropriate or do not misuse the funds at all. High-tech industries show lower misappropriation rates, while state-owned enterprises are more likely to misappropriate.

Methodological Approach
To analyze the effectiveness of R&D subsideu policies, the study employs a theoretical framework and estimates two key treatment effects:

• Intention-to-Treat (ITT) Effect: Measures the average effect of a granted subsidy, accounting for noncompliance. The ITT shows how effective the policy is when fund misuse occurs.
• Complier Average Causal Effect (CACE): Measures the effect of a subsidy in an ideal scenario without misuse. C7 computes the policy's efficacy, showing how effective it could be if all funds were properly allocated.

Econometric Results

• Treatment Effects: The ITT effect indicates a moderate level of partial crowding out, confirming that R&D subsidies increase total R&D expenditures by less than the subsidy amount. In contrast, the CACE shows strong evidence of additionality, suggesting that the policy could have stimulated significantly more R&D without misuse.
• Treatment Heterogeneity: Effects varied by time period, subsidy size, industry, and ownership. The policy design improved after 2006, but Chinese R&D policies still show a gap between effectiveness and efficacy, driven primarily by noncompliance.
• Robustness: Extensive testing using entropy balancing and controlling for various confounders showed consistent results across alternative specifications, including both short-term and long-term effects.

Conclusion
This research demonstrates that misappropriation significantly reduces the effectiveness of China's R&D subsidy policies. Accounting for noncompliance allows for a more accurate evaluation, revealing that the policy's lost potential is driven by both policy design flaws and insufficient monitoring. The theoretical and empirical contribution provides a rigorous strategy to assess the impact of public good R&D policy, potentially serving as a model for research in other countries with similar challenges.