地方政府债务、数字经济与城市经济韧性

王文军，傅晨辉

（陕西师范大学 国际商学院，陕西西安 710119）

摘要：当前中国地方政府债务持续扩张，已成为地方财政风险防控领域的重要议题，加强地方政府债务管理对提升城市经济韧性至关重要。基于2015—2022年中国281个地级及以上城市数据，使用堆叠处理下的双重机器学习模型探讨地方政府债务与城市经济韧性之间的因果效应，并将数字经济纳入分析框架，旨在考察二者的协同效应。研究结果发现，地方政府债务扩张会削弱城市经济韧性。这一结论在一系列稳健性检验下仍然成立。交互效应分析结果表明，数字经济与地方政府债务的交互效应对经济韧性起促进作用，且这种促进作用呈现阶段性特征；当数字经济发展到特定阶段后，其交互效应足以扭转债务扩张的负面影响，有效增强城市经济韧性。因此，地方政府应实现债务规模合理扩张，重视对数字经济等新兴产业的财政投入，并考虑地区差异制定差异化政策，以提升城市经济韧性。

基金项目：陕西省社会科学基金年度项目“数智技术助力陕西省生态韧性提升的驱动机制与实现路径研究”（2025D037）；榆林市科技计划项目“碳达峰碳中和目标下的榆林市产业转型发展研究”（2023-CXY-90）

关键词：地方政府债务；数字经济；城市经济韧性；双重机器学习；交互效应

作者简介：王文军，陕西师范大学国际商学院教授；傅晨辉，陕西师范大学国际商学院硕士研究生。

引用格式：王文军,傅晨辉.地方政府债务、数字经济与城市经济韧性[J].首都经济贸易大学学报,2026,28(1):63-74.

Local Government Debt, Digital Economy, and Urban Economic Resilience

WANG Wenjun, FU Chenhui

(Shaanxi Normal University, Xi'an 710119)

Abstract: Amid the critical juncture of profound global economic restructuring and domestic economic transformation and upgrading, urban economic resilience has emerged as the core pillar for withstanding external shocks and achieving high-quality development. Meanwhile, the fiscal risks stemming from the expansion of local government debt have become a major bottleneck constraining the enhancement of urban economic resilience. The rise of the digital economy offers a new pathway to resolve this dilemma.

Based on panel data from 281 prefecture-level and above cities in China from 2015 to 2022, this study employs a double machine learning model under stacked processing to systematically investigate the causal effects of local government debt on urban economic resilience. It further incorporates the digital economy to analyze their synergistic moderation mechanism. The findings reveal that local government debt expansion significantly weakens urban economic resilience. This conclusion remains robust after robustness tests, including instrumental variable treatment, sample adjustment, and variable substitution. Interaction effect analysis indicates that the digital economy can significantly mitigate the negative impact of local government debt on economic resilience. On one hand, the digital economy leverages big data analytics to precisely identify high-resilience industries, optimize debt capital allocation, and alleviate resource misallocation. On the other hand, blockchain technology enables transparent management throughout the debt lifecycle, while large language models can predict debt default transmission pathways.

Threshold effect analysis confirms the nonlinear nature of the digital economy's synergistic effects. Results from a panel threshold model using digital economy level as the threshold variable reveal a dual threshold relationship: When the digital economy is low, local government debt exerts a greater negative impact on urban economic resilience. When the digital economy surpasses the first threshold, the negative impact coefficient of debt decreases, and the buffering effect of the digital economy begins to emerge. After the digital economy surpasses the second threshold, the debt's impact on resilience shifts from negative to positive.

The conclusions offer new theoretical insights and empirical evidence for enhancing urban economic resilience and promoting coordinated regional development from the perspective of synergistic debt management and digital empowerment. Accordingly, it recommends: (1) establishing a dynamic management system for local government debt that links debt scale to urban economic resilience targets; (2) strengthening synergistic empowerment between digital economy and debt policies and leveraging digital technology to enhance debt management efficiency and capital utilization; (3) implementing differentiated regional strategies by tailoring policies to each locality's stage of digital economic development and debt-bearing capacity; and (4) providing crucial empirical evidence and decision-making references for balancing debt risks with resilience enhancement to achieve high-quality urban economic development.

Keywords: local government debt; digital economy; urban economic resilience; double machine learning; interaction effect

下载全文