基于Post-LASSO方法的就医需求多控制变量选择

分析省级层面就医需求的政策变量和交互要素,并控制地区和时间效应的异质性,为精确估计医疗改革效应和医疗机构区域合理布局提供科学依据.以就医需求和就医供给的代理变量、区域特征控制变量建立指标体系,采用Post-double-selection-LASSO方法选择潜在变量及其函数形式.一阶差分、全控制变量和各省标准差集聚三个模型的比较结果显示,标准差集聚模型较好地控制时间趋势和初始差异,证实复杂就医需求的影响因素包括医院规模、医疗业务、医疗价格、区域特征,以及医疗收入和人口密度的交互作用.由此,应考虑不同工具变量的传递路径、不同区域特征及其初始差异和交互要素对就医需求的影响,以实现医疗资源空间均衡化.

Selection multiple control variables of medical treatment demands based on Post-LASSO methods

To analyze the policy variables and interaction factors of provincial level medical treatment demands, and control heterogeneity of area and time effect, to provide a scientific basis for the estimation of medical reform effects and regional rational distribution of resources. With control variables of medical supply and demand variables, regional characteristics to establish the index system, post-double-selection-LASSO method is used to select potential variables and function forms. The contrasted results of first difference, all control variables and standard deviation cluster model shows that, the standard deviation cluster model is better control the time trend and initial differences, especially confirmed the complex medical treatment demands factors including hospital scale, medical services, medical price, regional characteristics, interaction and medical income and population density etc. It should considered transfer path of different instrumental variables, different regional characteristics and initial differences, interaction factors on medical behavior, in order to achieve the equalization of medical resources in space.