Abstract：This paper uses text analysis technology to construct media sentiment index based on news texts related to exchange rate, and studies the impact of media sentiment on the level and volatility of RMB exchange rate.Based on EGARCH model, we find that media sentiment can significantly reduce the fluctuation of RMB exchange rate.We also find that the interaction term between media sentiment and exchange rate expectation has a significant negative effect on both the level and volatility of the RMB exchange rate, indicating that media sentiment has a negative moderating effect in the influence of exchange rate expectations on the RMB exchange rate, which is conducive to the stability of the foreign exchange market.We expand our study through the quantile regression to explore the different impact under various RMB exchange rate levels, finding that, when the RMB exchange rate is in the depreciation range, media sentiment has a significant negative impact on both the level and volatility of the RMB exchange rate, and the inhibitory effect increases with the degree of depreciation.By classifying media sentiment into positive and negative categories, we find that positive media sentiment can lead to appreciation of the RMB exchange rate, while negative media sentiment causes a significant depreciation of the RMB exchange rate.The findings of this paper are useful for clarifying the role of media sentiment in exchange rate decisions and safeguarding the stability of the foreign exchange market.

Keyword：Media sentiment; RMB exchange rate; Text analysis; EGARCH model; Quantile regression;