羊群行为视角的系统重要性地方政府识别研究

Study on the Recognition of Systematic Importance of Local Government from the Perspective of Herd Behavior

守住不发生地方政府系统性金融风险底线,重点要监控好系统重要性地方政府,而其核心在于领头示范引导性较强,复杂关联中心度很高.理论分析表明地方政府债务风险承担具有时空正关联作用,该作用导致羊群效应.据此,构建动态空间面板杜宾模型,验证该命题,随后用CH模型与CCK模型识别,利用参数与非参数的分位数回归估计,稳健测度出羊群效应,并利用神经网络模型分位数回归估计法,非线性识别了系统重要性地方政府.研究表明:地方政府债务风险承担显著具有空间模仿与时间跟随的羊群效应微观演化特征;羊群效应显著较大;神经网络分位数回归模型可以较好地拟合空间网络溢出传染引领效应,能够识别出短期与长期的系统重要性地方政府,从而对地方政府债务风险的宏观审慎监管提供参考.

To keep the bottom line of non-occurrence of systemic financial risks of local governments, we should focus on monitoring the systemically important local governments, and the core lies in the leading demonstration, strong guidance and high complex correlation centrality. Theoretical analysis shows that local government debt risk-taking has a spatiotemporal positive correlation, which leads to herding. Therefore, the dynamic spatial panel Durbin model is constructed to verify the proposition. Then, ch model and CCK model are used to identify, and the quantile regression estimation of parametric and nonparametric is used to measure the herding effect, and the quantile regression estimation method of neural network model is used to identify the systemically important local government nonlinearly. The results show that: the local government debt risk-taking has the micro evolution characteristics of spatial imitation and time following herding effect;the herding effect is significant;the quantile regression model of neural network can better fit the leading effect of spatial network spillover contagion, and can identify the short term The long-term and long-term systemic importance of local government, so as to provide a reference for the macro prudential supervision of local government debt risk.