基于GARCH-DCS模型的我国碳金融市场收益预测研究

Research on Return Forecasting of China's Carbon Finance Market Based on GARCH-DCS Model

动态条件得分(Dynamic Conditional Score,DCS)建模思想是一种充分利用分布信息的时变参数建模方法,为碳金融市场资产收益率波动建模提供了新的思路。本文基于动态条件得分建模思想来构建GARCH-DCS模型,同时选取我国碳金融市场中具有代表性的湖北和广东两个试点市场作为实证研究对象,并依据平均加权连续排名概率得分等得分规则对GARCH-DCS模型和GARCH模型的一维/多维收益预测效果进行对比分析。一维收益预测实证结果显示,GARCHDCS-SST模型比GARCH(1,1)-SST模型可以更有效地预测两个试点市场收益;多维收益预测实证结果表明,GARCH-DCS-MVT模型预测效果优于DCC(1,1)-MVT模型。

The idea of dynamic conditional score modeling is a time-varying parameter modeling method that makes full use of distribution information,and provides a new idea for modeling the volatility of asset returns in carbon financial markets.This paper builds the GARCH-DCS model based on the idea of dynamic conditional score modeling,and selects two representative pilot markets in our country's carbon finance market,Hubei and Guangdong as the empirical research objects,and based on the scoring rules such as the average weighted continuous ranking probability score,the one-dimensional/multi-dimensional return forecasting effect of GARCH-DCS model and GARCH model is compared and analyzed.The em-pirical results of one-dimensional income prediction show that GARCH-DCS-SST model can more effec-tively predict the income of the two pilot markets than GARCH(1,1)-SST model;The empirical results of multi-dimensional income prediction show that GARCH-DCS-MVT model is superior to DCC(1,1)-MVT model.