The Impact of News Sentiment on Exchange Rate Volatility: A Study Based on AI Sentiment Analysis

This paper explores how news sentiment impacts the nominal effective exchange rates (NEER) of the US dollar and the Japanese yen using AI sentiment analysis and behavioral economics theories. Using extracted data of news regarding the US dollar and Japanese yen between 2014 and 2024 and their respective sentiment scores, this study develops a regression model with controlling for macroeconomic variables. Our results show that positive sentiment about US news significantly enhances the dollar's appreciation while negative sentiment about Japanese news has a stronger impact on the yen's appreciation as a safe haven to increase yen appreciation. This result suggests that although risk aversion dominates expectations, the market tends to react more quickly to changes in sentiment. Moreover, the negative sentiment of financial news has a particularly strong market transmission effect. This study provides an alternative view of exchange rate risk management as well as new avenues for sentiment analysis within multilingual processing and model optimization, with potential implications for future exchange rate modeling and policy decisions on financial markets.