中国省际高技能人才迁移的时空演化机制

高技能人才迁移是推进新型城镇化建设的重要议题,也是影响地区创新产出和高质量发展的关键要素。针对人才迁移数据中蕴含的零膨胀和网络自相关特性,本文将特征向量空间滤波（ESF）技术和“两阶段”Hurdle模型结合,构建空间Hurdle引力模型,结合2000—2015年中国省际高技能人才迁移面板数据,研究人才迁移的时空演化格局和驱动机制。研究结论显示：① 2000—2015年人才迁移的跨省迁移比例先升后降;人才迁移表征出集聚格局,维系了其空间分布的不均衡性;随时间推移,人才迁移格局呈现分散趋势,人才空间分布集聚性下降;人才迁移和空间分布均呈现出持续显著的网络与空间自相关性特征。② 引力因素（人口规模、空间距离）、地区经济和科技发展水平（工资、科教投入）、自然舒适度（平均温差、空气质量）、城市舒适度（医疗及教育公共服务、城市绿化）以及其他因素（社会网络、生活成本、人口密度）共同驱动了跨世纪以来中国省际人才迁移过程。③ 人才迁移可被看作一个“两阶段”过程,影响其迁移概率和迁移规模的因素呈现一定差异。④ 经济增速、科教投入、自然舒适度和基础公共服务对人才迁移的影响随时间增强,而工资和城市绿化的影响随时间减弱。本文的研究结论为地区人才治理及实现地区均衡发展提供政策参考。

Spatio-temporal evolution mechanism of China's internal skilled migration

The migration of skilled individuals has become an important issue in promoting new-type urbanization in China and a key factor affecting China's regional innovation output and high-quality development. Considering the issues of zero inflation and network autocorrelation in skilled migration data, this paper combines the eigenvector space filtering (ESF) technique and the "two-stage" hurdle model into a comprehensive united framework to construct a longitudinal spatial hurdle gravity model. It then has been employed in the case study exploring the spatiotemporal patterns and influencing factors of interprovincial skilled migration in China during 2000-2015. The results are listed as follows: First, from 2000 to 2015, the mobility proportion of the skilled migration increased first and then decreased. The agglomeration pattern of skilled migration maintains the imbalance of its spatial distribution. With the elapse of time, the migration of skilled presents a dispersing trend and drives the decline of its spatial distribution and agglomeration. Talent migration presents a persistent and significant network autocorrelation, and its distribution presents a persistent and significant spatial autocorrelation. Second, China's interprovincial skilled migration during 2000-2015 was driven by gravity factors (population scales at origin and destination, distance), regional economic and scientific and technological development (average wage, spending on science & technology and education), natural amenities (average temperature difference, air quality), urban amenities (public health and education services, urban greening), and other factors (social networks, the cost of living, and population density). Third, the migration of skilled people can be regarded as a "two-stage" process, where factors affecting its migration probability and migration scale are different. Such differences are mostly reflected in factors of amenities versus economy. Fourth, the impact of economic growth, investment in science and education, natural amenities, and basic public services on skilled migration has increased over time, while the impact of wages and urban greening has weakened over time. The conclusion of this paper provides policy references for regional talent management and the balance of regional development.