基于区间型金融时间序列数据的宏观经济预测研究

Research on Macroeconomic Forecasting Based on Interval-valued Financial Time Series Data

基于均方误差准则给出了构建区间数据模型的变量选择方法,并利用股票市场、基金市场、期货市场以及货币市场的区间型金融时间序列数据对宏观经济进行区间预测分析,给出了有别于传统点值数据模型的宏观经济区间预测方法。实证结果表明,区间型金融数据中的深证成分指数、上证基金指数、期货市场交易金额、狭义货币供给量对宏观经济区间预测模型拟合误差较小。通过变量选择得到了基于区间金融时间序列数据的宏观经济区间预测模型,并利用单一模型结构和组合模型结构给出我国2020-2023年的宏观经济变化区间,预测表明我国宏观经济将延续总体平稳、稳中趋缓的发展态势。

This paper provides a variable selection method for constructing interval data model based on the mean square error criterion,and gives interval forecasts and analysis of macro economy by using interval-valued financial time series data collected from stock markets,fund markets,futures markets and money markets,which is different from the traditional point-value data forecasting model of macro economy. The empirical results show that,the interval-valued financial data,such as Shenzhen Component Index,Shanghai Composite Fund Index,trading amount in the futures market and monetary aggregates M1,have little fitting error for the macroeconomic interval forecasting model. Macroeconomic interval forecasting models based on interval-valued financial time series data are obtained through variable selection methods,and future variation intervals of China’s macro economy during 2020-2023 are predicted and analyzed by the single forecasting models and combination forecasting models,which shows that China will continue to develop steadily on the whole but with a slowing development trend.