中国老年人口长期护理需求预测——基于多状态分段常数Markov分析

Forecasting Demand for Long-term Care: Based on Multistate Piecewise Constant Markov Process

文章利用中国老年健康影响因素跟踪调查数据,追踪微观个体观测期前后健康变化,运用多状态马尔科夫(Markov)链构建状态转移概率矩阵,避免回归方法中变量选择的多样性和主观性;引入转移强度为分段常数、按年龄队列矩阵相乘的Markov方法预测长期护理人口数量,解决Markov时齐性假设与健康状态年龄非齐性问题;基于精算理论推导并预测护理需求期望时间。结果发现,健康状态存在明显的性别和年龄差异,相同条件下女性具有生存优势,男性具有健康优势,二者叠加使未来10年女性失能老年人数量达到男性的2倍左右;失能持续时间随年龄增长先增后缓慢递减,男性老年人护理需求平均时间小于女性,年老体弱的女性是未来长期护理需求的主要人群。

Using the samples from the latest two waves of Chinese Longitudinal Healthy and Longevity Survey, this paper constructs transition matrices based multistate Markov model, which avoids the malpractice of subjective in selecting variables in regression. Using transition intensity piecewise constant, the paper forecasts the size of long-term care demand of the elderly after ten years in Markov. The piecewise constant can solve Markov time homogeneity hypothesis. With the former yielding the statuses expected time are be studied. The results show that the health statuses transition has significant difference in age and sex. The female elderly have the relative advantages in survival, and the male elderly have the relative advantages in keeping in healthier status. Also women who need long-term care services are two times of men in the same status to 2026. Moreover, women spend more time in long-term care status than men. Therefore, the frail older women become main demanders of long-term care services.