【摘要】 利用混合频率数据抽样方法(MIDAS)构建不同权重函数形式的混频预测模型,使用5种高频解释变量对中国季度GDP增长率进行短期预测,并根据预测结果分析各种加权方式基于每种高频解释变量的预测效果。研究结果表明,利用不同的高频解释变量和不同的加权方式预测结果差异明显,利用国房景气指数对于中国季度GDP进行预测时精确度优于使用其他4种变量。更多还原

【Abstract】 We use MIDAS model method to build mixed frequency prediction models with different weighting functions,and use five high frequency explaining variables to predict China ’s quarterly GDP growth rate in short term,and then analyze the prediction performance of every weighting method and every high frequency explaining variable based on the prediction results.The research results show that there are big differences among the prediction results using different high frequency explaining variables and different weighting methods,the prediction accuracy with Real Estate Climate Index is better than the other four variables.更多还原