DCC-MIDAS-sentiment modelThe preceding empirical results fundamentally exhibit the long-term correlation between stock and bond markets. Based on theframework, we extend the specification by allowing long-term volatility and correlation to be affected by the sentiment index. Themodel combines daily stock and bond returns with the monthly sentiment index and decomposes the total dynamic correlation into longandshort-term components.We subsequently estimate the DCC-MIDAS-sentiment parameters by incorporating monthly investor sentiment into the long-termvolatility and correlation component. Table 2 presents the DCC-MIDAS estimation results of dynamic correlation influenced byinvestor sentiment.Panel A of Table 2 shows the results from estimating the GARCH-MIDAS model for stock returns and bond returns when weincorporate investor sentiment in the MIDAS equation. Then, we discuss the estimated θ parameters individually for the two markets.The sign of θ indicates the response of long-run volatility to investor sentiment. The coefficient θv is negative and significant at the 5%level for bond returns, indicating that sentiment has a significantly negative influence on long-term bond market volatility. Our findingconfirms the negative relationship between sentiment and bond market volatility proposed by Asgharian et al. (2016). However, theestimated coefficient θ of stock market is insignificantly negative, which means that the effect of sentiment on the long-term stockvolatility is rather ambiguous. The DCC estimation results are shown in Panel B of Table 2. The estimated parameter θc reflects howstock-bond correlation responds to sentiment shocks. Here, the coefficient θc is positive and statistically significant at the 5% level.