统计与数据科学学院知识库

当今世界的经济格局呈现全球化、多极化与区域化的发展趋势，同时也给我国经济发展带来更多不确定性的风险与挑战。经济韧性的正是一种帮助我国抵御、复苏与适应面对不确定风险与挑战的道路，也是中国式现代化的重要动力，对推动全球经济复苏也起着重要作用，如何提高经济韧性已成为国内外学者的研究热点。
我国经济韧性仍是一个值得关注的问题，当今贸易保护主义、全球经济增速放缓明显，经济高质量发展离不开强有力的经济韧性支撑。因此，为探究我国经济韧性发展现状、运行机制与提升路径，本文以2008—2021年我国31省、市、自治区经济韧性为研究对象，首先依据凡登定律构建因果模型，在反事实框架下,采用动态空间面板模型SEM-RE对我国经济韧性进行测度与分析，并对我国经济韧性动态演变、地区差异与特征分类方面进行分析；然后从因果反馈机制与系统稳定性机制两个角度，将我国经济韧性系统视为网络结构，对我国经济韧性网络系统运行机制进行分析，因果反馈机制方面，构建DPSIR框架，并使用PLS-SEM模型对我国经济韧性网络系统进行因果检验，在检验成立的基础上，使用ENA模型对经济韧性系统整体稳定性、利用率与冗余率进行分析，最后使用Lasso-ELM、Lasso-PSO-ELM等方法对我国经济韧性提升路径进行动态模拟。得到一系列研究结论与建议。
（1）经济韧性测度方面，整体来看，我经济韧性呈现逐年递增、稳健向好趋势，具体来看，我国各省市经济韧性呈现增速快、地区差异大的特点。研究期内，我国东部沿海地区经济韧性水平优于内陆及西部地区，南方地区经济韧性优于北方地区， 2021年我国31省市经济韧性水平均大于1，表明我国实际产出高于反事实产出，经济韧性水平良好。
经济韧性分析方面，动态演变角度，我国经济韧性核密度曲线波峰呈现下降趋势，即我国经济韧性发展水平差异进一步加大；从核密度曲线的移动趋势来看，经济韧性核密度曲线向右移动，表明经济韧性水平逐年提高；地区差异角度，总体基尼系数的减少表明我国经济韧性的地区差异性逐年向好，我国经济韧性组内差异大小为东部地区>西部地区>中部地区>东北地区；特征分类角度，采用Jenks自然断点法，将我国各省市经济韧性分为不同的类别，第一类为含甘肃、西藏等23省市；第二类为山东、广东、福建、浙江、天津5省市；第三类为江苏、北京、上海3省市。
（2）经济韧性网络系统机制分析方面，因果反馈机制角度，采用PLS-SEM模型对我国经济韧性框架各元素因果关系进行检验，DPSIR框架下各个元素因果关系均在显著性水平α=0.01时通过检验，从具体取值大小来看，R→D响应层面对驱动层面正向作用最大，而D→P驱动层面对压力层面负面作用最大，表现为经济系统驱动越大时，经济系统所面临的压力也就越大。系统稳定性运行机制方面，我国经济韧性系统优势度A=6.239>冗余R=5.013，表明我国经济系统利用大于冗余；从可持续发展角度来看，经济系统的利用率与冗余率都不宜过高，系统最佳利用率A/C值为0.4455，我国经济韧性系统利用率A/C=0.554略大于此值。
（3）本文将韧性理论与机器学习理论相结合，对我国经济韧性的提升路径进行模拟，探究我国经济韧性的最佳提升路径。相比于Lasso-ELM，粒子群优化算法改良后的Lasso-PSO-ELM模型对我国经济韧性动态模拟的结果与真实值吻合度得到了一定提高，且动态模拟评价指标更加优良。基于上述所构建的Lasso-PSO-ELM模型，对我国2022年各省市经济韧性进行情景假设与动态模拟，结果显示高经济韧性情景假设下，我国各省市经济韧性模拟值最高。

 The current global economic landscape is showing a trend of globalization, multipolarity, and regionalization, which also brings more uncertain risks and challenges to China's economic development. Economic resilience is just a way to help China resist, recover and adapt to uncertain risks and challenges. It is also an important driving force for Chinese path to modernization and plays an important role in promoting global economic recovery. How to improve economic resilience has become a research hotspot for scholars at home and abroad.
The resilience of China's economy is still a matter of concern. Today, with trade protectionism and a significant slowdown in global economic growth, high-quality economic development cannot be achieved without strong economic resilience support. Therefore, in order to explore the current development status, operating mechanism, and improvement path of China's economic resilience, this article takes the economic resilience of 31 provinces, cities, and autonomous regions in China from 2008 to 2021 as the research object. Firstly, a causal model is constructed based on Van den's law. Under the counterfactual framework, the dynamic spatial panel model SEM-RE is used to measure and analyze China's economic resilience, and the dynamic evolution, regional differences, and feature classification of China's economic resilience are analyzed; Then, from the perspectives of causal feedback mechanism and system stability mechanism, the economic resilience system in China is viewed as a network structure, and the operational mechanism of the economic resilience network system in China is analyzed. In terms of causal feedback mechanism, the DPSIR framework is constructed, and the PLS-SEM model is used to conduct causal testing on the economic resilience network system in China. On the basis of the test, the ENA model is used to evaluate the overall stability and effectiveness of the economic resilience system Analyze the utilization rate and redundancy rate, and finally use methods such as Lasso ELM and Lasso PSO ELM to dynamically simulate the path of China's economic resilience improvement. Obtain a series of research conclusions and recommendations. 
(1) In terms of measuring economic resilience, overall, China's economic resilience has shown a steady and positive trend of increasing year by year. Specifically, the economic resilience of various provinces and cities in China has shown characteristics of rapid growth and significant regional differences. During the research period, the economic resilience level of the eastern coastal regions in China was better than that of the inland and western regions, while the economic resilience level of the southern regions was better than that of the northern regions. In 2021, the economic resilience level of all 31 provinces and cities in China was greater than 1, indicating that China's actual output was higher than the counterfactual output, and the economic resilience level was good.
In terms of economic resilience analysis, from the perspective of dynamic evolution, the peak of China's economic resilience kernel density curve shows a downward trend, indicating that the difference in China's economic resilience development level is further increasing; From the moving trend of the nuclear density curve, it can be seen that the economic resilience nuclear density curve is moving to the right, indicating that the level of economic resilience is increasing year by year; From the perspective of regional differences, the decrease in the overall Gini coefficient indicates that the regional differences in China's economic resilience have been improving year by year. The size of the differences within China's economic resilience group is as follows: Eastern region>Western region>Central region>Northeast region; From the perspective of feature classification, Jenks natural breakpoint method is used to divide the economic resilience of China's provinces and cities into different categories. The first category includes 23 provinces and cities such as Gansu and Xizang; The second category includes five provinces and cities: Shandong, Guangdong, Fujian, Zhejiang, and Tianjin; The third category includes three provinces and cities: Jiangsu, Beijing, and Shanghai. 
(2) In terms of analyzing the mechanism of the economic resilience network system, from the perspective of causal feedback mechanism, the PLS-SEM model is used to test the causal relationships of various elements in China's economic resilience framework. Under the DPSIR framework, the causal relationships of each element are all at a significant level α= By testing at 0.01, it can be seen from the specific values that the R → D response layer has the greatest positive effect on the driving layer, while the D → P driving layer has the greatest negative effect on the pressure layer. This is manifested as the greater the driving force of the economic system, the greater the pressure it faces. In terms of system stability operation mechanism, the advantage degree of China's economic resilience system A=6.239>redundancy R=5.013, indicating that China's economic system utilization is greater than redundancy; From the perspective of sustainable development, the utilization rate and redundancy rate of the economic system should not be too high. The optimal utilization rate A/C value of the system is 0.4455, while the utilization rate A/C of China's economic resilience system is slightly higher than this value at 0.554.
(3) This article combines resilience theory with machine learning theory to simulate the improvement path of China's economic resilience and explore the best path to enhance China's economic resilience. Compared to Lasso ELM, the improved Lasso PSO-ELM model based on particle swarm optimization algorithm has improved the consistency between the results of dynamic simulation of China's economic resilience and the actual values, and the evaluation indicators of dynamic simulation are better. Based on the Lasso PSO-ELM model constructed above, scenario assumptions and dynamic simulations were conducted on the economic resilience of various provinces and cities in China in 2022. The results showed that under the high economic resilience scenario assumption, the simulated values of economic resilience in various provinces and cities in China were the highest.