集中连片特困地区多维贫困测量及优化调控研究——以河南省秦巴山区为例

The project takes the Qinling-Daba contiguous poor areas in Henan Province as the study area. On the basis of introducing the Alkire-Foster multidimensional poverty measurement method, poverty dynamics and reduction performance is analyzed, and the evaluation index system is constructed with the combination of objective data and subjective wishes, and then the breadth, depth, and characteristics of spatial heterogeneity of multidimensional poverty is described. And also individual effects and background effects leading to poverty are identified in the family and township levels, while the dynamic feedback mechanism between multidimensional poverty and its affecting factors is explored. Compared to previous studies, the poverty threshold is mainly determined based on the subjective judgment of people living in the study area. Also, non-parametric test methods are taken to assess the robustness of multidimensional poverty index for different weights program. After the presentation of the multi-dimensional spatial distribution of poverty using thematic maps outputting from GIS software, the spatial correlation and agglomeration state of Multidimensional Poverty Index in the study area is tested by global and local spatial autocorrelation test. In order to distinguish the individual effects and background effects of the poverty factors in family and township levels, the Graded Linear Model is established. Finally, a relative complex system dynamics model is built to examine the restraint relationship of mutual multidimensional poverty and the poverty factors with the aim of providing a scientific basis for the establishment of regional poverty targeting mechanism.

秦巴山区作为全国14个连片特困地区之一，已成为扶贫攻坚的主战场。随着以人均收入衡量贫困的局限性日益凸显，多维贫困的度量已成为研究热点。本项目以河南省秦巴山连片特困地区为研究区域，引入Alkire-Foster多维贫困度量方法，评价研究区域贫困动态和扶贫绩效，构建客观数据和主观评价相结合的指标体系，分析多维贫困的广度、深度和空间异质特征，识别家庭和乡镇两个层次因素的致贫效应，探讨多维贫困和致贫因素之间的动态反馈机制。相比已有研究，本研究注重建立参与式的指标体系和阈值标准，以增加贫困度量的社会基础；通过非参数检验评判指标权重的稳健性实现区域外推；利用空间自相关分析检验多维贫困的空间集聚情况，加强空间异质性研究；从家庭和乡镇两个层次识别致贫因素，从而区分贫困的个体和背景效应；建立系统动力学模型模拟多维贫困和致贫因素之间的相互制约关系，刻画其动态作用机制，以期为区域扶贫瞄准机制的建立提供科学依据。

围绕河南省秦巴山连片特困地区多维贫困问题，利用数据包络分析方法评价了研究区扶贫绩效，发现全要素扶贫绩效在2000-2016年呈现明显波动性特征；改善地区技术效率是提升该区域全要素扶贫绩效的重要途径。引入Alkire-Foster多维贫困度量方法，测度了研究区多维贫困特征，发现不管是在什么k值水平，多维贫困程度都是较低的；超过50%的维度被剥夺的农户只有10.12%，而且k值临界水平越高多维贫困程度越低。空间自相关分析表明，研究区镇域经济发展水平具有明显的空间相关性和空间集聚特征，低低集聚区主要分布在卢氏县的绝大部分乡镇。利用中国健康与营养调查数据的研究表明，河南省2009-2015年多维贫困指数呈现明显下降趋势，降幅接近50%，但农村仍是多维贫困发生的重点区域。利用Stayer-Mover模型的验证分析表明，多维贫困在中国不是一个贫困陷阱，但对于在不同状态之间迁移的家庭，应该制定有针对性的短期政策或长期政策帮助其脱贫。分层线性模型模拟得出，不同社区的家庭进入多维贫困的概率存在显著差异，其中46%的差异可以由社区层面的变量来解释，其余54%的差异可由家庭层面的变量解释；良好的社区发展条件有助于降低多维贫困的发生率。最后，从提高就业能力，健全医疗体系，提升教育质量等方面提出了扶贫开发的对策建议。