基于多变量动态Probit模型的中国经济景气预测  

Prediction of China’s Economic Prosperity Based on Multivariate Dynamic Probit Model

作　　者：桂文林[1] 程慧 Gui Wenlin;Cheng Hui(School of Economics,Jinan University,Guangzhou 510632,China)

机构地区：[1]暨南大学经济学院,广州510632

出　　处：《统计与决策》2020年第11期16-22,共7页Statistics & Decision

基　　金：国家哲学社会科学基金项目(16BJY014)。

摘　　要：文章基于改进的动态probit模型对中国经济景气进行预测。首先对月度数据进行季节调整和H-P滤波处理以分离出循环成分,然后根据McFadden R^2最大化原则构建出最优模型,并对各种probit模型进行样本内分析和样本外预测。样本内分析结果表明,动态probit模型的预测效果优于静态probit模型,不论是模型预测评估还是转折点预测,一般动态probit模型和动态自回归probit模型的预测能力都显著高于静态probit模型或自回归probit模型;样本外预测结果表明,动态probit模型的预测效果优于静态probit模型,四种probit模型都能够提前发出预警衰退信号,并对经济景气状况给出合理的信息。Based on the improved dynamic probit model, this paper forecasts the economic boom of China. First, the monthly data are given seasonal adjustment and H-P filtering to isolate the circulation components. Then the optimal model is constructed according to the maximization principle of [McFadden R^2], and various probit models are used to make in-sample analyses and out-of-sample prediction. The in-sample analysis shows that the prediction effect of dynamic model is better than that of static model, and that whether it is model assessment or turning point prediction, both the general dynamic probit model and the dynamic autoregressive probit model have significantly higher predictive power than the static or autoregressive probit model. The out-of-sample prediction results show that the prediction effect of the dynamic probit model is better than that of the static probit model, and all the four probit models can give early warning of recession, and reasonable information about economic situation.

关 键 词：样本内估计 动态probit模型 景气预测 样本外预测 

分 类 号：F224[经济管理—国民经济] F061.6