IMF-中国产业政策_量化及其对错配的影响（英）-2025.8_30页_1mb

Summary of "Industrial Policy in China: Quantification and Impact on Misallocation"

Core Content

This IMF Working Paper examines the size and impact of industrial policies (IP) in China, focusing on their fiscal cost and effect on factor misallocation and productivity. The authors use data from financial reports of listed firms and the land registry to quantify IP instruments and estimate their effects on resource allocation efficiency and aggregate total factor productivity (TFP).

Main Industrial Policy Instruments and Their Fiscal Costs

The paper identifies four main types of industrial policy instruments and quantifies their fiscal costs:

1. Cash Subsidies

   • Directly reported in the financial statements of listed firms.
   • Subsidy rates have declined from 2.4% in 2013 to 2.0% in 2023.
   • Top sectors receiving subsidies include semiconductors, high-tech manufacturing, and automobiles, while consumer goods, services, real estate, and energy receive less support.
   • Private firms (POEs) receive higher cash subsidies than state-owned enterprises (SOEs), though SOEs still benefit from lower interest rates and higher cash subsidy rates when controlling for sector.

2. Tax Benefits

   • Measured as the difference between the statutory tax rate (25%) and the effective tax rate at the sector level.
   • Tax benefit rates increased from 4.4% in 2013 to 6.3% in 2023.
   • The sectoral ranking of tax benefits is highly correlated with that of cash subsidies.
   • Private firms receive larger tax benefits than SOEs, suggesting that IP extends beyond SOE support.

3. Subsidized Credit

   • Measured as the difference in effective interest rates across sectors, after controlling for financial variables.
   • Manufacturing firms benefit from lower effective interest rates (0.4 percentage points below other sectors).
   • Central SOEs receive significantly lower interest rates than POEs.
   • Larger firms and those with lower intangible asset shares also benefit from lower rates.
   • The effect is robust to various control variables and specifications.

4. Subsidized Land

   • Estimated using land registry data, comparing government-sold land prices to non-manufacturing land prices within a 1 km radius.
   • Manufacturing land is substantially cheaper (at least 2/3 discount) than land in other sectors.
   • The land price gap is not mainly due to land development differences.
   • No significant price difference between SOEs and POEs for industrial land.

Total Size of Industrial Policy

• The total fiscal cost of IP in China is estimated at 4.4% of GDP in 2023.
• Cash subsidies account for the largest share (2.0% of GDP), followed by tax benefits (1.5%), land subsidies (0.5%), and subsidized credit (0.4%).
• The composition of IP has changed over time, with tax benefits increasing post-pandemic and other instruments decreasing slightly.
• The total size of IP has remained broadly stable, though the proportion of tax benefits has grown.

Impact on Factor Misallocation and Productivity

• The paper uses a structural model to estimate the impact of IP on factor misallocation and aggregate productivity.
• Subsidies lead to excess production relative to a no-distortions benchmark.
• Trade and regulatory barriers tend to limit production, possibly by increasing market power of incumbents.
• IP is estimated to reduce domestic aggregate TFP by about 1.2%, compared to a no-IP baseline.
• This channel could lower GDP by up to 2%.
• Industrial champions (market-leading firms) benefit from higher productivity and policies favoring their production relative to average firms in the sector.

Key Findings

• IP is not limited to SOEs; private firms are also heavily supported, especially in tax benefits and subsidies.
• Factor misallocation from IP has a significant negative impact on aggregate productivity.
• The fiscal cost of IP is substantial, amounting to 4.4% of GDP in 2023.
• Tax benefits have become a more important component of IP in recent years.
• Subsidized land and credit are also significant, but less so than cash subsidies and tax benefits.
• The paper abstracts from potential benefits of IP, such as correcting market failures or knowledge spillovers, focusing instead on fiscal costs and misallocation.
• It contributes to the literature by using a sector-level definition of IP and providing a comprehensive estimate of its size across the entire economy.

Policy Implications

• The IMF recommends scaling back industrial policy and increasing transparency to reduce misallocation and improve productivity.
• The results align with previous studies that highlight the role of IP in distorting market outcomes in China.
• The paper complements recent work on IP policy counts and factor misallocation, offering a quantitative assessment of the economic impact of IP.

Methodology and Data

• The empirical approach uses Hsieh and Klenow (2009) model for measuring productivity and factor misallocation.
• Data sources include WIND financial statements, Orbis firm data, and land registry data from the Ministry of Natural Resources.
• The analysis covers listed firms and non-listed firms, with controls for firm type, leverage, and other financial variables.
• The "ring method" is used to estimate land subsidies, analyzing price differences between manufacturing and non-manufacturing land sales within the same year and geographical proximity.

Conclusion

• Industrial policies in China have substantial fiscal costs and distort resource allocation.
• The overall impact on aggregate TFP is negative, with factor misallocation reducing productivity by 1.2%.
• Industrial champions are supported by both productivity advantages and policy interventions.
• The paper suggests that reducing IP could improve economic efficiency and productivity in China.