Scholarlink

<p>Government debt management and quantification are increasingly critical for fiscal sustainability assessment, particularly in emerging economies undergoing rapid structural transformation. This study develops a Bailout-Oriented Implicit Debt (BOID) framework for China, integrating gradient boosting classifiers with system GMM econometric methods to quantify Direct Implicit Liabilities (DIL) and Contingent Implicit Liabilities (CIL) at the provincial level. Using panel data from 30 Chinese provinces spanning 2013–2022, we estimate national implicit debt at approximately 35–42 trillion RMB (28.5−34.2% of GDP) in 2022, with Local Government Financing Vehicles (LGFVs) comprising around 84% of the total. Macro-fiscal stress testing reveals that a combined adverse scenario—GDP slowdown, land revenue collapse, and interest rate increase—could push implicit debt to 61.8 trillion RMB (47.3% of GDP), with provinces breaching a 50% debt-to-GDP threshold quadrupling from 3 to 12. The analysis reveals significant provincial heterogeneity, with coastal regions exhibiting more sustainable debt patterns than inland areas. Employing dynamic panel models, we identify a critical institutional quality threshold (IQ <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\approx \)</EquationSource> </InlineEquation> 7.65 on a 10-point scale): below this threshold, explicit debt expansion suppresses implicit debt accumulation through transparent financing channels; above it, the two debt types become complementary. Resource allocation efficiency and investment levels serve as significant mediating variables, though paradoxically exhibiting negative effects on high-quality development due to structural adjustment costs and diminishing marginal returns. The model demonstrates robust predictive accuracy (cross-validation RMSE = 0.087) and strong convergence with IMF extended debt calculations and major rating agency assessments. The Fiscal Value-at-Risk metric and stress-testing module translate these estimates into decision-ready risk measures, providing an empirical foundation for differentiated policy responses and a transferable methodological framework for implicit debt quantification in transition economies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

China’s provincial implicit debt: estimation, determinants, and fiscal risk using machine learning and system GMM

  • 期刊论文

  • 发表日期：

    2026年4月6日

查看全文
  • Xintao Zhao,
  • Wenxiu Hu,
  • Bingqian Zhao

摘要

Government debt management and quantification are increasingly critical for fiscal sustainability assessment, particularly in emerging economies undergoing rapid structural transformation. This study develops a Bailout-Oriented Implicit Debt (BOID) framework for China, integrating gradient boosting classifiers with system GMM econometric methods to quantify Direct Implicit Liabilities (DIL) and Contingent Implicit Liabilities (CIL) at the provincial level. Using panel data from 30 Chinese provinces spanning 2013–2022, we estimate national implicit debt at approximately 35–42 trillion RMB (28.5−34.2% of GDP) in 2022, with Local Government Financing Vehicles (LGFVs) comprising around 84% of the total. Macro-fiscal stress testing reveals that a combined adverse scenario—GDP slowdown, land revenue collapse, and interest rate increase—could push implicit debt to 61.8 trillion RMB (47.3% of GDP), with provinces breaching a 50% debt-to-GDP threshold quadrupling from 3 to 12. The analysis reveals significant provincial heterogeneity, with coastal regions exhibiting more sustainable debt patterns than inland areas. Employing dynamic panel models, we identify a critical institutional quality threshold (IQ \(\approx \) 7.65 on a 10-point scale): below this threshold, explicit debt expansion suppresses implicit debt accumulation through transparent financing channels; above it, the two debt types become complementary. Resource allocation efficiency and investment levels serve as significant mediating variables, though paradoxically exhibiting negative effects on high-quality development due to structural adjustment costs and diminishing marginal returns. The model demonstrates robust predictive accuracy (cross-validation RMSE = 0.087) and strong convergence with IMF extended debt calculations and major rating agency assessments. The Fiscal Value-at-Risk metric and stress-testing module translate these estimates into decision-ready risk measures, providing an empirical foundation for differentiated policy responses and a transferable methodological framework for implicit debt quantification in transition economies.